You are on page 1of 359

Modeling Complex Systems

Volume 52 of the Nebraska Symposium on Motivation

University of Nebraska Press Lincoln and London

Volume 52 of the Nebraska Symposium on Motivation
Richard A. Dienstbier Bill Shuart Will Spaulding Jeffrey Poland Presenters Bill Shuart Will Spaulding Jeffrey Poland Richard W. J. Neufeld Wolfgang Tschacher Zeno Kupper Susanne P. Lajoie Mark A. Musen Eduardo Salas Kevin C. Stagl C. Shawn Burke Gerald F. Goodwin Michael J. Mahoney

Modeling Complex Systems
Series Editor Volume Editors

Madonna Rehabilitation Hospital University of Nebraska–Lincoln Rhode Island School of Design / Brown University University of Western Ontario University of Bern University of Bern McGill University Stanford University University of Central Florida University of Central Florida University of Central Florida U.S. Army Research Institute University of North Texas

Modeling Complex Systems is Volume 52 in the series CURRENT THEORY AND RESEARCH IN MOTIVATION © 2007 by the University of Nebraska Press All rights reserved Manufactured in the United States of America International Standard Book Number ISBN: 978-0-8032-1387-6 (Clothbound)

The Library of Congress has cataloged this serial publication as follows: Nebraska Symposium on Motivation. Nebraska Symposium on Motivation. [Papers] v. [1]–1953– Lincoln, University of Nebraska Press. v. illus., diagrs. 22cm. annual. Vol. 1 issued by the symposium under its earlier name: Current Theory and Research in Motivation. Symposia sponsored by the Dept. of Psychology of the University of Nebraska. 1. Motivation (Psychology) BF683.N4 159.4082 53-11655 Library of Congress

Preface

The volume editors for this 52nd volume of the Nebraska Symposium on Motivation are Bill Shuart, Will Spaulding, and Jeffrey Poland. The volume editors coordinated the symposium that led to this volume, including selecting and inviting the contributors and coordinating all aspects of editing. My thanks to our contributors for excellent presentations and chapters. This symposium series is supported by funds provided by the chancellor of the University of Nebraska–Lincoln, Harvey Perlman, and by funds donated in memory of Professor Harry K. Wolfe to the University of Nebraska Foundation by the late Professor Cora L. Friedline. We are extremely grateful for the chancellor’s generous support of the symposium series and for the University of Nebraska Foundation’s support via the Friedline bequest. This symposium volume, like those in the recent past, is dedicated to the memory of Professor Wolfe, who brought psychology to the University of Nebraska. Richard A. Dienstbier Series Editor

and Gerald F. Kevin C. Mahoney 245 275 287 293 307 Contributors Subject Index Author Index . Stagl. Musen 185 Eduardo Salas. and Jeffrey Poland Richard W. Neufeld Wolfgang Tschacher and Zeno Kupper Susanne P.Contents ix Bill Shuart. Shawn Burke. Lajoie Introduction 1 85 123 Composition and Uses of Formal Clinical Cognitive Science A Dynamics-Oriented Approach to Psychopathology Developing Computer-Based Learning Environments Based on Complex Performance Models Technology for Building Intelligent Systems: From Psychology to Engineering Fostering Team Effectiveness in Organizations: Toward an Integrative Theoretical Framework Constructive Complexity and Human Change Processes Editors’ Postscript: Modeling Complex Processes in a Rehabilitation Application 145 Mark A. C. J. Goodwin Michael J. Will Spaulding.

.

and so patterning and biological utility are what we see. pp. 1954. Will Spaulding. 64–65) I have tried. and to make biological sense. . Rhode Island School of Design / Brown University Capturing the complexity of human behavior has been a recurring theme in the Nebraska Symposium on Motivation: We expect behavior to be patterned or integrated. 314) When fundamental psychologists do make excursions into the human motivational world . it is organized. (Koch. Such matters are threateningly complex.Introduction Bill Shuart. behavior or experience relevant to man in his most characteristically human performances: man as he creates or loves or plays or responds to the aesthetic surfaces of the human and natural environment. let alone analysis. . 1956. University of Nebraska–Lincoln. to show that it is possible to formulate a meaningful theory of complex motivation by analyzing the . More remote still is the chance that anyone will select for illustration. it is rare that they survey the requirements for theory or pre-theory by intensive descriptive analysis of behavior related to such motives as produced by concrete human beings. first. And of course what we see is actually there—behavior in general is not chaotic. p. and Jeffrey Poland Madonna Rehabilitation Hospital. (Nissen.

pp. The complexity of the processes captured in the contributors’ models enhances the models’ applicability to the complexities of clinical practice. in this volume. I have sought. 1962. industry. One special feature of all the contributions from this particularly distinguished group of theorist-practitioners is an emphasis on practical applications of the conceptual frameworks in which they work. I suppose the moral is: be not afraid of complexity. it is diverse. pp. motivation. a good deal of complexity can be generated.x modeling complex systems sorts of variables involved. To consolidate the relevance of application . . and education. concept formation. to show that we possess. then let us find the means to cope with it. we shall advance in our attempts at conceptualization and classification only as we are willing to embrace the limits of symbol systems to capture human uniqueness and the ultimate ineffability of complex system dynamics. 49–50) Although disorder may be experienced and expressed in highly patterned processes of human activity. 1966. (Burke. . now. that of translational research. individually unique. many sound and useful concepts and techniques to translate these complexities into productive experimental research. emotion. learning. in a rudimentary mathematical way. they all explicitly or implicitly incorporate complex frameworks of dynamic. If motivation is indeed complex. 42–43) Through the representation of a few very simple psychological concepts. This reflects an important idea in the zeitgeist of the contemporary scientific community. 265–266) The contributions to this volume of the Symposium describe contemporary approaches to the modeling of complex psychological and behavioral processes. and systemic. Let us now proceed to generate complexity from simplicity. together with their interactions. Translational research is a process of translating the principles and truths that emerge from basic science into practical applications. Although the contributions reflect a range of theoretical and epistemic perspectives. second. system-like relations involving perception. behavior. and the social context in which behavior occurs. pp. (Vinacke. (Mahoney. intention. . ranging from molecular to molar phenomena.

not just practical application. Dr. In the first contribution Richard W. describing a practical model for the complex processes of rehabilitation. while the tradition of mathematical modeling in psychology has a long and honored past. At a more general level. this volume ends with a volume editors’ postscript. as manifest in rehabilitation services currently evolving in Nebraska. Decrying a continuing overreliance in much psychological research on statistical analyses associated with Fisher and Pearson. but continuity with theory and basic science. Neufeld characterizes his approach as a novel form of construct validity. Dr. one based on the inherent mathematical properties of the cognitive processes he studies. their discussion invites us into the heart of psychology’s theoretical legacy. It is noteworthy that. Neufeld discusses the advantages of formal mathematical theory for illuminating relations between variables as they interact in experimental science. This converges with the historic role of the Nebraska Symposium as a prominent (and now the oldest sustained) forum for psychological theory. In the next contribution. Tschacher and Kupper then apply their perspective and methods to the complex realm of psychopathology. Translational research demands. He applies these advantages to the clinical practice of assessing cognitive impairments. computers and analytic software) supports the kind of sophisticated modeling in the hospital or clinic that was impractical in earlier decades. Neufeld asserts that formal mathematical modeling of cognitive processes will. ultimately. In this sense his contribution is a sophisticated exemplar of the use of complex modeling to achieve traditional theoretical goals of experimental and clinical psychology.xi Introduction and translational research. All the contributions in this volume emphasize the theoretical basis of application and the necessity of logical and conceptual continuity in understanding complex processes. Wolfgang Tschacher and Zeno Kupper provide a synthesis of dynamic systems theory and current cognitive science. Drs. the increasing availability of powerful computational tools (e. lead to greater theoretical clarity about normal and abnormal cognition and better clinical-assessment techniques. They present data sets and analyses from recent . as articulated by such historical figures as Lee Cronbach and Paul Meehl. Inspired by the historic role of Gestalt psychology in the evolution of cognitive science.g. J..

She also emphasizes the importance of other variables. in a tribute to the Nebraska Symposium’s historic focus. Dr. and social. emotional. A second exemplar of complex modeling to achieve traditional goals is provided by Suzanne P.. and how to model the content and/or process. he envisions a future role for psychology in the development of artificially intelligent systems to manage our already enormous and rapidly expanding knowledge base. Dr. motivational. self-monitoring).. they show how complex. whom or what should serve as the model. His technology is the technology of computer engineering. in developing effective models. Finally. their contribution culminates with a characterization of motivation as identical to the ongoing action of complex human cognitive processes operating to order and simplify a complex world. Lajoie highlights the importance of discerning experts’ relevant “dimensions of expertise” (e. e. Returning to theoretical principles. they identify unique patterns or dimensions of intrasubject characteristics that have complex but meaningful interrelations. Lajoie. Nevertheless. She then translates these principles into design considerations for effective educational technology. and she describes strategies for determining what to model. Musen’s theory base is not psychology or neuroscience but artificial intelligence. as expressed in a specific context. Musen discusses past and current efforts to develop computer applications to support decision making and data representation in health care. The next contribution extends application of complex modeling from education to knowledge management. It is noteworthy in this regard that psychology has drawn from engi- . Learning how experts go about problem solving and decision making through “cognitivetask analysis” is an important aspect in the process of developing an effective tutoring system.g.g. Dr. Using a time-series analysis. Mark A.xii modeling complex systems research with people diagnosed with schizophrenia and demonstrate the importance of tracking individuals with multiple measurements over time in order to detect oscillations or trajectories in rehabilitation and recovery that would be missed in the typical cross-sectional approach. dynamic formulations can be translated into useful clinical instruments and methods. Lajoie uses theoretically grounded performance modeling in the development of computerbased “intelligent” tutoring systems designed to help learners master the complexities of real-world endeavors.

In the next contribution. an integrative framework and family of theories. In their review. consequently. shared mental models. Kevin C. Stagl. Mahoney’s contribution is a nuanced interlacing of several kinds of “models. Dr. and external factors. Stagl. The authors conclude that both influences are important and that. from radar-inspired signal detection models of perception to band-filter models of attention to computer models of executive cognition. Dr. Michael J. Salas. Salas.” including verbal metaphor. particularly in naturalistic settings. Mahoney discuses various perspectives on “complexity” theory and its precursors in philosophy and science. consonant with the “general systems” framework that has influenced many areas in the social sciences during the past several decades. cited artificial intelligence as a promising model for human cognition. placing them in the context of the history of ideas. and Gerald F. The first provides . Stagl. to be a key commonality among these models. Shawn Burke. The impact of complex models for knowledge management may be in psychology’s future rather than its past. He describes and elaborates on constructivism. including current theoretical frameworks such as dynamic systems theory. They advance “the science of teams” by providing a detailed review of representative models of team performance in organizations and other naturalistic settings generated over the past quarter century. and Goodwin find the invocation of input-process-output (ipo) models. in his 1994 Nebraska Symposium on Motivation contribution. This new model is distinctive in the importance that it attaches to individual team members’ cognition as an important moderating variable as well as group decision making. There is greater diversity among models with respect to emphasis on internal team processes versus greater attention to the influence of external. C. Burke. Drs. Burke. contextual factors. Herbert Simon. narrative. Goodwin scrutinize complex processes associated with small groups of people brought together for common purposes. Eduardo Salas. and Goodwin then turn to a description and elaboration of a new and unique multilevel integrative framework for understanding team functioning. Mahoney’s contribution includes two appendices. more sophisticated modeling techniques are needed to successfully deal with the resulting dynamic complexity. complexity studies. and chaos theory.xiii Introduction neering as much as vice versa. photography. Drs. and poetry.

We found a richness of such ideas. Walker (1964). Leeper (1965). Newcomb (1953). Complexity. We selected contributions that. to identify ideas that anticipate or shape the approaches to complexity that we find in contemporary work. Systems. molar and molecular. These frameworks are most generally associated with general systems theory. psychotherapy. in addition to showing the nascent ideas about complexity and systems theory discussed in this volume’s contributions. Processes and mechanisms are molar rather than molecular to the degree that they represent the integrated interaction of multiple components. The historical Nebraska Symposium contributions reviewed here follow . and other educational pursuits.xiv modeling complex systems a synopsis of important aspects of human change from the perspective of constructivism. so many that only a few can be highlighted here. and Barker (1960). The second provides rich and provocative perspectives for incorporation in the practice of counseling. One common characteristic of systems theories is an organizational scheme that orders specific mechanisms and processes according to their respective complexity. Psychology itself reflects this type of ordering. and the Nebraska Symposium on Motivation: A Brief History of Ideas The perspectives reflected in this volume are exemplars of an evolving set of conceptual frameworks that influenced thinking in many areas of science during the second half of the 20th century. serve to define the poles of these schemes. Two terms from classic learning theory. as this volume’s editors we saw a useful purpose in reviewing the more than 300 individual contributions that constitute the previous volumes. and addressing complexity is one of their key common features. Having recently celebrated a half century of the Nebraska Symposium on Motivation. coaching. ranging as it does from theories of neuronal activity to neuropsychology to the psychology of social cognition and interpersonal behavior. The remainder of our introduction to this volume is a review of five especially resonant contributions from volumes past: Heider (1960). have clear relevance to practical application and translational research and especially to our own particular interests in physical medicine and psychiatric rehabilitation.

g. self-organizing systems) that. and at the social/interpersonal level (Newcomb. in an effort to preserve the style and tone of the original presentation. Klein’s comments highlight important aspects of the view of rehabilitation and the notion of participation as an important goal of rehabilitation (World Health Organization. traditional Gestalt investigators. evolved from earlier. In addition to describing Heider’s unique viewpoint. 1965). 1964). However. in many cases. Interestingly. Heider’s work has been more influential on certain theorists than the relative paucity of scientific citations would indicate. and the Psychological Environment (1959).xv Introduction a molecular to molar rather than a chronological sequence. An exemplary illustration is Fritz Heider’s “The Gestalt theory of motivation” (1960). then editor of the journal Psychological Issues (itself representative of the rise of interest in ego functioning in the psychoanalytic literature the 1940s and 1950s). Event Structure. we hope to provide enough of the language in sufficient detail for readers to draw their own conclusions about the relation of the historic ideas to those of the present volume.. George S. addressing “spontaneous” organizational processes at the level of a human community. 1953). Also. Klein. Other contributions to be reviewed address system organizational processes at the cognitive level (Walker. Fritz Heider (1960) Many Symposium contributions have included Gestalt psychology frameworks offering concepts and models that resonate with contemporary perspectives (e. The reader is encouraged to consult the full chapters in areas of particular interest. We have included extended excerpts from the original contributions. provides a brief preface to a set of selected papers published as Fritz Heider’s monograph On Perception. while Barker is at the other end of the continuum. Heider (1960) and Barker (1960) integrate classic Gestalt principles with subsequent theories. but Heider addresses comparatively molecular expressions of key processes. the emotional level (Leeper. these passages cannot fully convey the logic or the eloquence of the source materials. The . contributions from common theoretical perspectives can differ with respect to the molecular-molar dimension. 2001).

Throughout the papers the composition of a “unit”— whether spatial. as it were. and other variables that will be discussed in the final section of this introduction. The general macrostructures which he describes may apply to their subjective counterparts in ego organization.” therefore.” The defining properties of a “unit. each intimately associated with the behavior settings available to the individual. Heider’s influence has been detectable in perception theory. . as is common in psychological theories—but from the outside inward. . impact on some of the most important theorists of our time. come in for extended and penetrating analysis. . According to Klein (1959): “Fritz Heider’s work . More recently. for example. and Egon Brunswik. . . if relatively unobtrusive. his writings cannot be called ‘popular’” (p. Klein continues: Heider’s emphasis on the “macrophysics” of things (in contrast to the reductionist emphasis on microphysics). the important distinction he develops between those parts of the environment which mediate (“medium”) and those which are mediated (“thing”). Kurt Lewin. has had over the years a significant. Heider has made [an attempt] to penetrate the essential nature of the concept of structure. The result. .xvi modeling complex systems ideas presented have direct relevance to current models of rehabilitation and crucial aspects of participation and quality of life. or causal—is of central importance to Heider’s distinction between “thing” and “medium. then. by specifying the architectural rules of the extrapersonal world of physical object and event units. notably Kurt Koffka. . is an extraordinarily fresh confrontation of the external structures which are assumed but never specified in psychoanalytic notions of reality testing and adaptation. . temporal. his analysis of how we may distinguish behavioral events attributable to the structure of the environment and those attributable to the structure of the perceptual system—all of these merit close study. vii) . . that is. in the work of James Gibson. the quality of his or her social networks. Still. A unique feature of Heider’s approach is his attempt to fathom environmental structure not from the response side— from the inside outward. (p. . . v).

Kohler then related this observation to a similar tendency found in physical systems. symmetry. . advanced by Wertheimer. We may notice this irregularity in another person’s face when we look at his mirror image. It embraces such properties as regularity. . The thought model is one of a complex process with many part events which interact in such a way that a certain endstate is reached. In this definition. slight irregularities in the shape of visual forms are usually not noticed. Ernst Mach has pointed out. The model is based on the work of investigators such as Wertheimer. Kohler gives the following example: Faces of people usually appear to us symmetrical. . all this fits very well together. the process attains an equilibrium and nothing more will happen. Since isomorphism is assumed. when it is reached. or as Koffka says: “Psychological organization will always be as ‘good’ as the prevailing conditions allow. and Koffka. . Of course the same tendency is then assumed to rule the process in the physiological brain field. On the other hand. who applied it to visual processing: This principle states that the perceived figure tends to be as good as the stimulus pattern will allow. simplicity and others. in spite of the fact that they are rarely objectively symmetrical. for instance. . Heider begins by emphasizing the Gestalt concept of good figure. but ordinarily we don’t see it. Köhler.” For instance. and which has characteristics the other possible states do not have. ‘good’ is undefined. as long as this end-state is not reached something will happen. . Heider highlights a number of aspects of this perspective that bring to mind such concepts as patterning and perceptual organizing processes that seem quite consistent with concepts used today. Heider terms the initial model the classic Gestalt theory. Kohler gives many examples in his book on Physical Gestalten [1924]. as.xvii Introduction In his Symposium contribution Heider provides an overview of four “thought models or schemata” characterizing the Gestalt tradition at the time. Kohler has called attention to the fact that a tendency towards simplicity can also be observed in physical systems. Describing this model. an end-state which is in some way distinguished. Let us recapitulate: Wertheimer observed the tendency toward good form with percepts. .

Essentially . . . 1938. belong not merely to the phenomena themselves. 1925. For instance. 303): The essential characteristic of regulation is an invariance of direction. (Heider. 1938. . . which implies direction. Instinctive activity then becomes an objective mode of behavior analogous to such phenomena as rhythm. we see that . However. the thought model of a system tending towards a standard state is applied to directed action. . Koffka does so in his book on The Growth of the Mind. . which first appeared in 1921. including all reactions made to the environment. pp. but likewise to the behavior taken as a whole. 1960. Tendency to closure is therefore only another name for tendency toward simplicity. . and this model had its origin in the principle of good figure. The word “standard” points here to the fact that the final status is independent of the initial configuration. or goodness of figure. . p. the end-state will come about regardless of what the beginning state of the system is: thus one can talk about a tendency. 145–146) Heider follows his discussion of Köhler’s ideas with a perspective from Kurt Koffka: It is not surprising that Gestalt psychologists have applied this same thought model to behavior. This is what Koffka says (Koffka. melody. and he describes it as follows (Kohler. 325). . . Whatever initial configuration may obtain in those systems when we begin to observe them—if we observe long enough their inner displacements or transformations will always be found to bring them nearer to a standard status. a reaching of the same endstate by different routes. a closed figure being a better figure than an open one.” It has to be distinguished from the state of equilibrium (Kohler. when we try to find out how it is carried out.xviii modeling complex systems Furthermore. . p. . p. and figure. . . . He [Köhler] calls the state toward which the processes in the organism are directed a “standard state. I should mention that Koffka uses the term “closure” for the distinguished end-state. The characteristics of closure . 103): .

This behavioral field is conceived of as having similarities with the visual field. is the state of the person who has reached the goal in his life space. But . pp.e. . The Gestalt psychologist would characterize this relation as one of isomorphism. and the way the organism effects changes in it. one can conceive of it as representing the environment of the person as the person himself experiences it—and it is in some way related to the brain field. . I can only say that in a first approximation. 1960. First. the state of the person being at the goal. is not entirely correct.e. The second step requires a more extensive consideration. not only a part of the organism.xix Introduction two steps are necessary for the transition from the phenomena of the visual field to action. . The idea of the feedback or circular process . in some way comparable to the simple figure. . .. Now. . 147–149) He later continues: In action. . This distinguished end-state. . . Thus we have to substitute for the visual field what Koffka called the behavioral field and Lewin the life space. which. we have to understand how the tension in the behavioral field makes the person reach the goal in reality. which are “encapsulated. in his physical environment? We assume that this behavioral field changes in the direction of a distinguished state.. It also is a system containing a great number of part processes which interact. So far we have only considered processes which are “inside” the organism in some way. i. and tends to arrange itself in such a way that a distinguished end-state is reached. How is it possible that they produce effects outside the organism. we have to consider the objective environment. . This is the first step we have to accept when we apply the principle of good form to activity: namely. . but the whole organism is involved. (Heider. this life space or behavioral field is a concept which involves many difficulties and unsuspected depths and snares. and secondly. however. maybe a state of minimal tension. to physical processes going on in the brain. i. we have to take into account not merely perceptual appearances but a space in which behavior occurs. it exhibits forces and tensions.” as Brunswik says. the step from perceptual to behavioral field. of structural similarity.

of Newcomb’s (1953) discussion of processes of com- . they will be realized by the person either in such a mental reorganization as wishful thinking. However. The main idea is that certain of these configurations are preferred. He notes the move from the perceptual sphere to the behavioral realm in Koffka’s work and what Lewin terms the life space. as for instance. and on the other hand unit relations of belonging. and thus the tension in this field is removed. . together constituting what Heider terms Lewin’s spatialized psychology. which changes the relation between organism and objective environment in such a way that the goal is reached. The entities can be persons—the own person or other persons—and other entities. What distinguishes the circular process of Gestalt theory from simple feedback is the interpolation of the behavioral field with its tendency towards a distinguished state [emphasis added]. this tension is communicated to the executive system. I remind you. via perception this objective state is communicated to the behavioral field. or in an actual change through action. situations. Heider’s discussion of Lewin’s concepts is thought provoking. 150) Heider goes on to describe two models advanced by Kurt Lewin—the person model and the environment model. one has to keep in mind that this circular process is not a simple feedback process. (p. things.xx modeling complex systems can be applied also in this case: as long as the person has not yet reached the goal. there is a tension in the behavioral field. describing what he calls his balance theory. and the reader is encouraged to review those concepts in the source material. and that. for instance. . . or groups. which he feels answers some questions left inadequately treated by the classic Gestalt theories he has summarized: This theory of balance deals mainly with configurations consisting of a number of entities between which exist certain relations. if circumstances allow. I have used the term “feedback” to characterize this process. Heider then offers a relatively brief summary of his own recent theorizing. In recent times a number of theories have been proposed which are similar to the one just outlined. The relations considered are mainly of two sorts: on the one hand attitudes of liking or disliking.

This model implies a number of different entities with certain properties and standing in certain relations. . . . nor do the forces affect the structure in a specifiable way. and simplicity. . balance. consonance. The dynamic factors arise out of definable structural characteristics and the forces toward the standard state tend to change the structure in definite directions. structure is not intimately connected with the conditions of tendencies. . . In Lewin’s environment model . Thus we see that in these [Lewin’s] models the dynamic factors are not very closely linked with structure. . . Neither do the properties of the structure imply forces. and the relation of these attitudes to each other enter as significant factors. . . The properties of these configurations which determine their meaning and their fate are whole-qualities. are.xxi Introduction munication [see below]. nor with their effects. . Consonance or simplicity of the structure cannot be derived from the properties of the parts . [The] difference between Lewinian theory and balance theory [is] in regard to the role structure plays. . the attitudes toward the parts of the configuration. namely. the idea of a ‘good’ figure. implied in that idea with which Gestalt theory started and which always was central to it. Structure helps us to derive the direction toward means from direction toward goal. of course. We therefore have returned to the model we considered first. . That is. and determine the attitude toward the whole configuration. . Thus. . then we find that the state of balance depends on the attitudes of p toward o or x. but is derived from the properties of the structure [emphasis added]. If we study the p-o-x system1 which is composed of the own person (p). which make up a constellation of factors tending toward a standard state. . In the balance model the dynamic factors are intimately connected with the structure. . we are able to specify more exactly the conditions of goal selection. at least in some cases. . These conceptions. . symmetry. but it does not help us to derive the direction to the goal. and an impersonal entity (x). . The goal is not taken to be an unanalyzed entity which in some way acquired valence. another person (o).

entitled “Psychological complexity as a basis for a theory of motivation and choice. 48–49) He follows precedent in setting the temporal length of a psychological event at 0. 1960. and all the parts of the structure are relevant to this difference. not only the relation between two parts. (Heider.” “motive. . 1964. as with Heider’s view. what are the determinants of the next event? A great many subareas of psychology are devoted to an effort to discover and quantify the determinants of choice behavior. What is the mechanism that terminates an event? 2. pp. . Then he asks: When an event is terminated. . . Walker (1964) Edward Walker’s contribution to the 1964 Symposium. . . though the terminology employed by these scholars and the associated research traditions from which they come are quite different.” is generally congruent with Heider’s (1960) views. Walker (1964) provides a concrete example of model building as well as an interesting perspective on the concept of complexity and mechanisms that contribute to the organization and self-regulation of behavior at the level of the organism. Walker states at the outset: This paper is an attempt to state what I believe to be the most basic questions of behavior theory.” “util- . learning. What are the determinants of the next event? 3. and motivation are all seen as quite interdependent and quite closely related to environmental stimulus context. and to test its clarifying contributions with respect to some critical problems of behavior theory. to elaborate the concept of psychological complexity as a potentially unifying concept. cognition. in the balance model structure in a state of equilibrium is definably different from one in a state of disequilibrium. 1. 167–170) Edward L. The three basic questions of behavior theory are .” “subjective probability. . adaptation. Also. pp. “Habit.5 seconds. [Unlike Lewin’s models].xxii modeling complex systems . person and goal. What is the fate of an event after it is terminated? (Walker. mechanisms of perception.

Psychological complexity is a characteristic of the event itself. (pp. pp. (Walker. (pp. but all. under the heading “Neural Net or Neural Process Complexity”: .” “attitude. are reducible to names of intervening variables or theoretical constructs.” Walker asserts: The major distinction that must be made is between “stimulus complexity” on the one hand and “psychological complexity” on the other. The first is a characteristic of the external stimulus. he turns to an elaboration of his concept of psychological complexity: Psychological complexity is a characteristic of the event itself and is thus a characteristic of the interaction of the organism with the distal stimulus when the event in question is initiated by a stimulus. 51–52) Under the subheading “Psychological Complexity Theory. Thus it is possible for two organisms to react with equal psychological complexity to stimuli with very different distal stimulus complexity values. Thus. but this third basic question or problem is beyond the scope of the present paper. It will be the argument of this paper that the concept of psychological complexity can be used to account for the termination of psychological events. more or less independent of the individual organism. the concept of psychological complexity can be useful in answering the first two of the three basic questions.” along with many other concepts in psychology. The same organism may also react with different degrees of psychological complexity at different times to the same stimulus. Psychological complexity and neural process complexity will be assumed to be completely isomorphic. the organism’s response. Psychological complexity can also be used to account for many of the phenomena associated with the trace of a past event.xxiii Introduction ity.” “set. 52–53) In the context of a basic definition of terms. and the choice of the next event over a wide range of traditional concepts of determiners of choice. ultimately. each related to different sets of operations. 54–55) Walker offers a brief description of the (assumed) underlying nervous system basis of the experience of optimal complexity. determiners of choice behavior. 1964.” and “trait.

Since neural net . is. psychological complexity far above optimum results in discoordinated tetany. If a psychological event is more complex than the optimum.” A lower limit is a level of complexity that is below the threshold of consciousness. the organism will behave in such a manner as to reduce the complexity of the event. p. The sequence is inevitable and the fall in neural net complexity is an automatic result of observable and fairly well understood neurophysiological characteristics.” “focal. In motor activity.e. We shall refer to the relative complexity of a four-dimensional neural process as the relative complexity of the relevant neural net. a pattern of neural events. If a psychological event is less complex than the optimum. . . It is assumed that variation can occur in the relative complexities of two neural processes. “complex” nets are relatively “large. Walker turns to the concept of psychological complexity itself. Under the heading “Optimal Complexity. The simplest and most straight forward psychological definition of optimal complexity is the following: Optimal complexity is that degree of psychological complexity the organism will seek to maintain.” “long. a fourth dimension.” “short. an input level far above optimum produces “mental dazzle. then the organism will behave in such a manner as to increase the complexity of the event.” “diffuse. 1964. 55). In perception. we shall assume a complete isomorphism between neural net complexity and psychological complexity (Walker. origin within the central nervous system). Optimal complexity can be bracketed by other values of psychological complexity. Furthermore. and there is a lower value which constitutes the threshold for action.” and “central” (i. Such events are spatially three dimensional and occur over a finite period of time.” and “peripheral.xxiv modeling complex systems Underlying any psychological event.” he asserts: The key concept of the theory I am attempting to fabricate is the concept of optimal complexity. .” Having offered a neurologically oriented substrate. He offers as relative characteristics of simple versus complex neural net in terms of the four dimensions noted above: “simple” nets consist of processes that are relatively “small.. The optimum is a “normal” percept or a smoothly coordinated movement. of course.

literally forces choice of that event among available next events which will be nearest optimum. (p. curiosity and other collative2 variables. 1964. 60) . motivation. psychological complexity will rise to and exceed the threshold of consciousness. It is assumed as a working hypothesis that many of the major determinants of choice behavior such as reinforcement. an organism will choose as a next event that activity which is nearest optimal psychological complexity. pp. he asserts: Repeated activation of a neural net will result in a progressive decrease in the psychological complexity of the event involved. 59) With respect to the second major question (What are the determinants of the choice of the next event?). habit. and others ultimately can be reduced to a single concept—psychological complexity. (Walker. The termination of one event occasioned by the automatic reduction in its psychological complexity below the optimum level. [The reader] will recall that the first of the three basic questions of behavior theory was: What is the mechanism which terminates an event? The answer is: Whether the stimulus for an event is continued or not. 56–58) With respect to repeated activation of an event. he continues: The principle of optimal complexity incorporates the dynamism that the organism will seek such a level. will fall below optimal complexity. subjective probability and utility. will rise to and exceed optimal complexity. a psychological event undergoes a sharp and automatic drop in complexity during a period of approximately one-half second after its initiation. and will usually be followed by another psychological event.xxv Introduction complexity and psychological complexity are assumed to be isomorphic. will then drop below the threshold of consciousness automatically. With a sufficiently complex stimulus. Therefore: Among available alternatives. psychological complexity may be said to rise and fall as well. (p.

1964.xxvi modeling complex systems Walker offers a brief recapitulation: The theory can be stated in an abbreviated form. The psychological complexity of alternative behaviors or events will be a function of four variables. Find a more complex stimulus in the environment to which he had not attended previously. React by locomoting. an individual might commonly respond in one or more ways (slightly adapting Walker’s text): 1. the psychological complexity of the event rises abruptly and falls more slowly. or he might fall to daydreaming. This may result when the exter- . (Walker. The automatic reduction in the psychological complexity of an event insures that it will drop promptly below the optimum to be replaced by that one of the available alternative events which is nearest optimum. p. 1964. Search the environment or his own repertory for more complex events. All of these devices would serve to increase the complexity level of the sequences of events which are occurring. p. getting up and moving about. All would serve to move the sequence nearer an optimal level of complexity. During the course of a psychological event that has a duration of approximately 500 [msec]. Seek arousing stimuli. Seek to differentiate previously unexplored potential complexity in his environment or in old thought sequences and problems. 61) At the other extreme: Situations in which the psychological complexity levels are above optimum are usually situations in which the sensory inputs into the nervous system are providing more information than the organism can process. They are: (1) the stimulus complexity of the initiation stimulus. (2) the time since this particular event has occurred previously. 2. (Walker. In the case of all available alternatives below optimum. 60) Walker addresses the issue of what behavior(s) might be expected when no near-optimum event is present. 5. 4. 3. (3) the number of times that event had occurred before. (4) the arousal properties of the stimulus or event.

he applies his theory of optimal psychological complexity to the often-observed (but less frequently reported) decremental variations in “conditioned responses” following “learning” experiments that have been taken by many to be unexplainable anomalies. If the overload is of external origin. The reason that we see few “learning” curves of the postulated type is that most experimenters know in advance what a learning curve is supposed to look like. The curve that rises to a steady maximum and remains there indefinitely is likely to be rare. let me take the position that the appropriate “learning” curve shape in running studies. 4. or when the motivational or emotional system is in a highly aroused state. 2. 3. the organism may attend repeatedly to the same stimulus in an effort. The organism may shift attention or narrow attention to a limited portion of the stimulus input. After noting a few such “anomalous” observations: For the sake of the argument I am certain to get. is one that rises and falls to zero or to a steady level below the maximum performance. he notes common reactions: 1. if they obtain a curve which does not fit their conception of what one should look like. For example. and selective learning studies. (pp. If either are difficult or impossible. to produce a reduction in the psychological complexity of the situation through repeated activation of the relevant event. they stop training when the “asymptote” is reached. conditioning studies. 61–62) In such circumstances. or. when the problem one is working [on] is beyond immediate solution. An associated result of repeated activation is to organize a very complex stimulus into a smaller number of “chunks. usually successful. He buttresses his concept by applying it to existing experimental data that are not easily explainable by any other existing theory.xxvii Introduction nal environment is too complex. the organism may locomote to a less complex circumstance.” (p. As a result of this knowledge. they find a great many other ways to respond . 62) Walker’s contribution concludes with a survey and critique of relevant research and theoretical distinctions to provide support for and elucidation of the conceptual framework adduced therein.

98) Robert Ward Leeper (1965) Robert Leeper’s contribution.” simply “energizing” or “arousing” factors that are then guided by “higher-order” functions (e. and decision-making in terms of a single construct certainly is one we should applaud. “Some needed developments in the motivational theory of emotions” (1965). his fellow presenter. motivation. I can attest that this process is carried out in good faith and under the assumption that in so doing. It predicts that most experimental situations will produce a drop in performance if training is continued. is perhaps the critical feature necessary to achieve a general integration. rather than “lower.xxviii modeling complex systems . other than to publish their sin against respectability. Walker’s attempt to develop a system that can deal with behavior dynamically. perception and/ . in addition to suggesting caveats to the former’s views.e. And. and the drop in performance that often occurs under continued reinforcement. It is also becoming increasingly recognized that the fundamental behavioral operation of an organism is selection or choice. They change the parameters of the study. Sooner or later they manage a situation in which they obtain the “right” answer. is focused on urging greater attention to understanding emotions as motivational factors. . extinction. Thus psychological complexity theory handles the usual learning curve. I can attest to this because I am one of the sinners. by whatever means. one is behaving like a sound. .. interdependent inseparability of psychological events. p. 1964. visualizing such disparate concepts as habit. i. This process is known as the establishment of experimental(er) control. They change the design. concludes with an important observation: There are several features of Walker’s approach with which I am in strong agreement. . (Logan.g. They throw away their data.. 85–86) In later comments concerning Walker’s presentation. A language that avoids the artificial separation of stimuli and responses more nearly captures the unified. pp. They restructure the apparatus. (Walker. Frank Logan. 1964. . continuously over time. . and careful experimentalist. rigorous.

This is the more surprising in view of the fact that most of this group are more or less closely related to one another both personally and as regards their general theoretical outlooks and interests. . E. C. . Leeper highlights the ultimately inseparability of processes of perception. 1950a. he urged. he said. it took me a long time to recognize that Lewin’s ideas might be thought of as a perceptual theory of emotion. Kurt Lewin similarly had been discussing many problems of motivation in terms of factors in the organism’s “psychological environment. Though proposing a less drastic statement on this point. David Krech (1949. are so definitely organic unities that no single aspect of such a system can be changed without changing the other aspects as well—we have been dealing in myths in believing that we could vary some one of these aspects while keeping the other aspects constant.” they did not make much sense for me. It seems. . . we ought merely to conceive of “Dynamic Systems. as though each of these persons had to grope to the concept on his own. In my own previous writing. motivation. . 1948) had been suggesting some perceptual factors in motivation in his view that motivation is partly a matter of reward expectations and punishment expectations. my original paper on a motivational theory of emotion (1948) was extended to some extent into the perceptual-motivational theory which has been elaborated in the present paper. therefore. . peculiarly. and emotion: Still earlier. I make this suggestion with somewhat more confidence because I remember that. Instead. in his usual impassioned style. had reasoned that it is unrealistic to conceive of psychological phenomena in terms of separate processes of perception. . . motivation and learning. Tolman (1932. when I first read Krech’s papers on “Dynamic Systems. . In fact. in my own case. One odd fact about these various earlier discussions of a perceptual or conceptual interpretation of motivation is that their authors have made practically no references to the related ideas of the other papers. Maybe this sort of thing will continue to be the case. 1951).” . 1950b.xxix Introduction or cognition). . even though possibly helped in ways that he did not recognize by his predecessors or colleagues. And. . .” These. If a perceptual theory of motivation is to become more common.

individual behavior in a learning situation. is never specific enough to predict within a specific area those details in which we are often most interested. second. he also notes that the breadth of such an overarching model would not lend itself to making predictions or heighten understanding of specific processes or variables within a particular subdomain of psychology in general (e. 113. is. 139) The relevance of his contribution to the topic of self-organized systems is seen in his analysis of the dynamics or organization of communicative behavior in an interpersonal context. are perceptual or representational processes. it is from the relatively limited theories that the relatively inclusive theories must in the long run emerge.g. 115) Theodore Newcomb (1953) Theodore Newcomb’s contribution. . Newcomb makes clear that he does not believe that a psychology of motivation in social situations should be fundamentally different or discrepant from a general psychology of motivation. 1953. to the very first volume of the Symposium. it should be subsumed by a broader model of motivation that describes human motivation in any situation. that emotions are motives. . 111–112) Leeper summarizes some major themes: The suggestion that comes from a number of sources. If emotional habits are perceptual habits. Even though perceptual habits are hard to change in some cases . first of all. behavior in a social situation): A general theory. these same possibilities should exist for them. . 1965. was “Motivation in social behavior” (1953).. Indeed.xxx modeling complex systems perhaps each psychologist will have to figure it out for himself. The suggestion that comes is that emotions and other motives do not exist or operate in any less complex sense than this. whether of motivation or of evolution of species. (Newcomb. . (Leeper. that emotional processes. therefore. along with all other motives. and then. . Rather. p. . it seems that all perceptual habits can be modified by learning and that sometimes such modifications can occur suddenly and dramatically. pp. However. (pp.

Human social behavior is thus to be studied in terms of the conditions and consequences of varying communicative acts. motivational phenomena are intimately interlinked with perceptual phenomena. of course. is a . it is often necessary to distinguish two aspects of person-object relationships. For present purposes. one does not need to make this distinction. while still bearing in mind that both aspects are involved. only as they are inferred from observable behavior. Orientations are known. consisting of discriminative stimuli. Insofar as such behavior involves reciprocal stimulation and response (or anticipations thereof) it is traditionally referred to as “interaction.e. one person (A) transmits information to another person (B) about something (X). one must observe discriminable units of behavior. Thus in the simplest possible communicative act. however.. persons as recipients of transmitted information and objects (including persons) as referents of transmitted information. therefore.” except that it connotes “existing directedness” and not merely a predisposition or a readiness. The relationship between orientations and communicative acts. which may be labeled the cathetic and the cognitive. of course). they are studied not as things-in-themselves.” But one cannot observe interaction-in-general.xxxi Introduction Newcomb goes on to define terms and delimit his focus to communicative behavior between individuals: The properties of objects may be studied either objectively or phenomenally—preferably in both ways—but in any case. as we have all learned in recent years. defined as any observable behavior by which information. but as related to persons. and in such instances the term “orientation” is a useful one. it is assumed that the discriminative stimuli have an object as referent. as we shall see. The term is similar to the concept of “attitude. to use as such an interaction unit the communicative act. I propose. Often. is transmitted from a human source to a human recipient. Thus the characteristic way in which social psychologists study motivation is in terms of person-object relationships (the term “object” includes other persons. And problems of motivation in social behavior are to be studied in terms of orientations toward the two kinds of objects necessarily involved in communicative acts—i. Since.

. and conversely.xxxii modeling complex systems circular one. Newcomb moves to a section titled “Communicative Behavior as Varying with Orientation toward Persons and toward Objects.” In this section he explicates the systemic relations between communicative acts and the orientations of individuals in a communication setting: I can hardly imagine anything that would surprise you less than to hear that communicative acts are learned in ways that seem to have something to do with rewards and punishments. Al- . . represent an exception . Secondly. A. . A. not to be oriented to them would mean to have no cognitive content regarding them and to have no “hypotheses” (in the Postman-Bruner sense) as to their potentialities for reward or punishment. . First. . and communication. . The further assumption that co-orientation has adaptive value stems from what I believe to be the fact that neither kind of orientation occurs singly and independently of the other. . of course. occurs in the total absence of an orientation toward B. We may start with the assumption that orientations both toward persons as potential co-communicators and toward other objects have adaptive value. the orientation of the communicator. . in connection with communicative acts. 140–141) Following a second section summarizing relevant findings concerning group membership. (“Autistic” verbalization. so that it will be necessary to consider each of them. . only to indicate in the most general kind of way what seems to be the nature of the learning conditions of communicative behavior. of the kind Piaget reports in young children. (Newcomb. occurs in an objectless vacuum. 1953.). .e.. orientations. however. if ever. rarely. toward B. would. The very fact that B is a potential recipient requires some kind of orientation toward him. I shall stop. pp. relating himself simultaneously both to objects and to persons as actually or potentially related to those objects. These conditions have to do with what I have already referred to as the individual’s necessity for co-orientation—i. the potential recipient of his communication. if ever. in turn. a potential recipient of his communication. . as varying with the other [emphasis added]. toward almost any conceivable X rarely. the orientation of any communicator.

. . . of A’s and B’s orientations toward X. the interdependent orientations among them are A’s toward B and toward X. communicative acts are instigated by the anticipation of increased similarity or congruence (or. The elements in this system are. by the threat of decreased similarity or congruence). I shall make one further set of assumptions about the system. as a result of learning. A. he goes on to articulate an important aspect of his “A-B-X” system. a model employed by the observer. . . .” it is characterized not by the absence but by the balance of forces. Since. there is included in this orientation toward B some assumption—however accurate or inaccurate— about B’s orientation toward the object of communication. The implications of this model are: (1) that while at any given moment the system may be conceived of as being “at rest. These assumptions are that (under the stated conditions) communication tends to result in increased similarity. and that.. 149). moreover. 1953. .xxxiii Introduction most invariably. For some purposes the system is best treated as an objective one—i. and B’s toward A and toward X. The systemic perspective of Newcomb’s contribution and his initial observations of the relation between explanations/models at the level of subareas of psychology concludes with a view to the future: I should like to suggest (with a good deal of tentativeness) that something along the lines of the framework of co-orientation which I have roughly sketched out may find a place in . and (2) that a change in any part of the system may lead to changes in any of the others.e. there are relationships of interdependence among several distinguishable orientations. . B and X (a source. or congruence. alternatively. a recipient and an object of communication). according to these assumptions. it is convenient to regard them as together constituting a system. pp. which he considers a “strain toward congruence” (p. . 147–149) After positing adaptive advantages of his concept of congruence. From this elaboration of what is perhaps only too obvious. I want to deduce a single point—that there is a necessary interdependence between co-orientation (which itself involves an interdependence of orientations) and communicative acts. minimally. (Newcomb.

Barker underscores the necessity of taking individual and environment into account as a unit in any thorough analysis of behavior. and far from least. . they are inextricable from the eternal triangle of self. Many. (Newcomb. when it is mature enough to include these interdependent orientations. If so. 1953. an adequate theory of motivated social behavior will have contributed something to a general theory. and. Last. . Theorists from McDougall and Freud to Murphy and Rogers have properly accorded to the self a central place. Barker’s conceptual framework is like that of Heider (1960) and Kurt Lewin’s concepts of field and life space (Lewin. Not only are self-orientations part and parcel of other-orientations. 159) Roger Barker (1960) Roger Barker’s contribution. . as well as helping to establish it. 1938). . . will have borrowed from a theory of motivation in social behavior. as in the case of other concepts of peculiar relevance to social motivation. other persons. an adequate general theory would take fuller account than it does today of self-orientation. . and the common environment. at least. are capable of plural orientation.xxxiv modeling complex systems general motivation theory. higher-order patterns or change. in ways other behavior cannot. of how behavior directedness varies with multiple orientations. it is my belief that more extrapolations from a general theory will not suffice. p. Here. others being held experimentally or hypothetically constant. though not always. Like Heider and Newcomb. “Ecology and motivation” (1960). have all of them seen that place in its full social context. includes an account of alteration of individual state(s) as a function of external. hence. his contribution advances themes consistent with the conceptual frameworks advanced at the 2004 Symposium. I suspect that the study of social behavior can provide evidence. in my judgment. A general theory of motivation. and the actual direction of behavior at any given moment often cannot be accounted for in terms of any single object-orientation. I would insist. among the higher forms of animal life. This provides an intellectual context in which to consider the importance of taking both in- .

(1) the ecological sector of objects and physical stimuli (preperceptual). p. At a practical level. This vein extends from the environment to the environment. through proximal stimuli at the receptor surfaces of a person. The three major sectors of this unit are . and (3) the behavioral sector which occurs. Barker begins by incorporating from the work of Egon Brunswik an emphasis on the critical importance of including in accounts of perception and behavior the environment in which an individual acts and perceives. to his proximal reactions. 1) Barker. This construct is of importance because it underscores the importance of including assessment and modification of an environment. along with Brunswik. it is “the basic psychological entity. which provides an important window on our understanding of a range of psychological phenomena as a “system” and is. (Barker. regards the entire span of the E-E (environment-environment) unit as the fundamental unit of analysis with respect to psychology. for example. . through his central processes. a very useful unit of analysis for psychology. and through his peripheral effector systems. this resonates with major themes in rehabilitation.xxxv Introduction dividual and environment into account. (2) the organism or intrapersonal sector. at times. from distal objects in the ecological environment. in the ecological environment. . as a vital part of the rehabilitation process. in addition to clinical treatment. Barker outlines features of his concept of psychological ecology. Barker describes a framework that explicitly relates systems concepts to adaptive processes at the social/community level. and it finally terminates in the focus of the total unit: the person’s achievement with respect to the nonpsychological world of things. In Barker’s words: Brunswik (1955) described psychological schools and theories in terms of their positions upon a macro-unit he considered to be the true vein of psychological ore. 2001). or means behavior. The relevance of Barker’s concepts for the issues addressed in the present volume is that. including the central concept of behavior settings. through the person’s peripheral receptor mechanisms. again. namely. the World Health Organization’s recent emphasis on the construct participation as the ultimate aim of rehabilitation efforts (see World Health Organization. 1960.” . like Newcomb (1953).

These considerations almost always make personal motives the whole story of the energetics of behavior. 1960. it is indivisible. particularly at the junction point between the ecological and the intrapersonal sectors of the unit. in fact. But a unit is a unit. one obvious way to understand this unit is as a multisectored system. the interaction of sectors of the E-E unit is considered with respect to emergent social aspects in Barker’s system: For a psychology defined in terms of E-E units. (Barker. one theme that recurs in other examples of systems approaches is that of the close relation between the processes of change. then. (Barker. When it is a psychological unit. As can be seen in the material reproduced below. and the behavior are all involved. and motivation. 3) The E-E unit. and especially with respect to motivation. p. a defining aspect of all exemplars of models informed by general systems theory. the environment.xxxvi modeling complex systems He takes issue with some of Brunswik’s conclusions. and energetics must occur in all of the parts. is to be taken as the ultimate unit of analysis. and place them within the organism. A second major theme is the influence of changes or properties in one part of the system on the qualitative status of other parts of the system. In a later section he provides further elucidation of the relation between the “entity” and the “environment” elements of his model: Ecology is concerned with relations between entity and environment. This. conceptual. p. Either the E-E unit is false. 1960. But before this statement has any useful meaning. is the theme of my paper. and explanatory account of the whole course of events can be achieved. 4) Barker goes on to lay the groundwork for discussion of his theory of behavior settings by introducing some concepts that provide the context for his central theses. perception. Finally. . the usual considerations of motivation are not adequate. advancing the hope that taking the entire E-E span into account can lead to more than a probabilistic framework for psychological explanations: I hold the hope that a more detailed. the organism. or motivation theory is too limited.

Indeed. The facts of learning demonstrate. the explanations. . Here. Where does each entity end and its environment begin? . . become incommensurate. as almost all learning theory recognizes. 1960. by which the environment influences the organism and its behavior. . yet the linkage remains.xxxvii Introduction entity and environment must be defined. . Another distinction that is crucial for the definition of environment is that between inside and outside. The test is this: As we move from any discriminable thing to more remote. To clarify this problem. pp. rather than abstracting only elements of it for psychological examination. . almost the only way. however. . . He offers the study of psychological principles of learning as an example: The field of learning is interesting in this connection. . Every entity has a discriminable boundary. 7–8) Barker asserts the desirability of taking the entire E-E continuum as the crucial unit of analysis. surrounding parts. . This is the predominant way. the behaving organism that endows the environment with behavior-controlling properties. the guiding and coercing powers of the environment have been shown to . it is necessary to revert to the levels of phenomena in science. so far as we know them. Learning is usually interpreted as the process. par excellence. (Barker. which have been devised to account for occurrences on one level are inadequate to explain occurrences on a different level. . . is the basis for delimiting an entity from its environment. what is within the boundary constitutes the entity’s inside. . that even here the organism is the locus of driving forces without which learning does not occur. I have emphasized that the essential distinction between levels is this: The laws. and what is without constitutes its outside. a point is reached at which the governing laws. . a culture is presumed to shape the personality and behavior of the individuals born into it. The environment of an entity is made up of those parts of the outside regions with which the entity is coupled by laws on a different level from those which govern the entity itself. This point marks the boundary of the entity and the beginning of the environment. . paradoxically. within the context of learning it is. for ecological problems. yet the levels are coupled systems. . . .

However. it defines the range of an organism’s power to transform its connections with the ecological environment. This is important information. at least. . (Barker. an approach to a satisfactory understanding) by exploring earlier ideas of Fritz Heider’s. . This relation between the demands of a given behavior setting and the impact on the behavior and life of the individuals populating these settings seems very compatible with more recent concepts. pp. and the intrapersonal and the behavior sectors. along with these.. Barker continues. a detailed presentation will not be offered here. and these depend more upon the organism than upon the environment. there is a relation between the peopling of these settings and both the number of responsibilities and the adequacy of performance that can be expected to occur. .xxxviii modeling complex systems depend upon what activities the organism has previously had with it. economics. in a section entitled “Theory of Behavior Settings”: Field studies in which I and my associates have been engaged. which Brunswik took as the realm of psychology. 1960. Barker makes a gradual transition to his concept of behavior settings. of the behavior of children in their natural habitats. chemistry. Indeed. 11–12) He then attempts to formulate an account of how these incommensurable system elements might be related (or. It will be clear now where ecology enters the environment-environment unit. one would highlight a particularly important element of his concepts concerning behavior settings: Barker highlights how behavior settings are regions in a community that offer certain opportunities and. learning studies have demonstrated that almost every discriminable part of the ecological environment can be coupled with almost every kind of behavior. governed by psychological laws. require certain responsibilities. etc. Psychological ecology deals with the relations between the nonpsychological sectors of this unit. governed by the laws of geometry. and it implies that parts of the environment are almost equipotential. Furthermore. Because these ideas are readily available to the interested reader. . Implications for the analysis of complex settings and behavior are evident.

” she does not refer to any individual’s behavior. people stand in the relationship of media to behavior settings. . The same is true of a newspaper item which reports. however. . laymen mention them in conversation and in writing as frequently as they do individual persons. arranged in a characteristic spatial pattern. “There is a baseball game in progress on the playground across the street. The wide ramifications of these simple ideas suggest that they may have a basic significance for psychology. [some] behavior settings: Streets and sidewalks Kane’s Grocery Clifford’s Drug Store Gwyn Café Pearl Café Midwest State Bank Of special relevance in the present connection. and particularly for the psychology of motivation. doors. “The annual fete held in the St.xxxix Introduction have brought us to the hypothesis that under certain precisely defined and frequently occurring conditions. into data upon differences in the behavior of individuals in settings of different sizes. imposing certain absolute constraints on them. people stand in the relationship of things to behavior settings.” These are behavior settings. . 2. chairs. Ambrose Church garden was a great success. When a mother writes. ad infinitum. A behavior setting has a circumjacent soma of physical objects: of walls. typewriters. . . . fences. They are highly visible behavior phenomena. It is first necessary to describe behavior settings. dishes. but to the behavior of children en masse. and into data concerning the behavioral consequences of physical disability. . . and that under certain other less common conditions. This hypothesis brings some order into data upon American-English differences in the behavior of children and adults. Behavior settings involve ongoing patterns of extraindividual behavior whose identity and functioning are independent of the participation of particular persons. . Here are . at a particular temporal and physical locus. are the following characteristics of settings: 1.

pp. One frequently occurring means of balancing the forces and maintaining the homeostatic level of a behavior setting is compensating for a deficiency in the number or docility of the parts of the medium by an increase in the amount of energy applied to each of them. Barker discusses “People: Media of Behavior Settings”: Six features of the relationship between people and behavior settings must now be mentioned. These multiple balanced forces assure that the level of a setting is more stable than most of its parts or conditions singly. with any conceptualizations being not far removed from the surface appearance of settings.” (Barker. they normally persist. and vice versa. . . which gives behavior settings.xl modeling complex systems 3. this is enough to make a beginning in tracing the connections along the Brunswikian unit which has its origins in this part of the ecological environment. and some originate within the individuals who populate the setting . it is this. . often for years. those available have to work longer and/or “harder. characteristic level. Behavior settings are homeostatic systems. for example. but its laws of operation are not the laws of individual psychology . . Most of what we know about behavior settings is simple description. People are part of the inside manifold of behavior settings. some are intrinsic to the setting itself. 1960. under certain conditions. at a relatively stable. or the workmen. . When the media of a setting. For our purposes. . . . A behavior setting is a behavior entity. Behavior settings exhibit a stability-within-change. the tools. the machinery. a persisting functional level which is due to a balance of many influences. . the self-regulatory characteristic of behavior settings is crucial and must be considered further. the position of things which impose their own patterns on the people within them. 1. However. are in short supply. 15–21) In the last section of his contribution. Forces operate in every setting. who have the position of media. . Some of these issue from the larger community. indeed. .

people are the sine qua non. . This is true. . the seat of motivating influences. there is one important exception. These five features of the relation between people and behavior settings emphasize the position of people as the media of behavior settings. actually differ greatly in population. . . . . the setting will be modified. leads inevitably to the recognition that there must be a satisfactory awareness of the . In reality they are controls built into the structure and the dynamics of the setting. The opportunities within them are matched by the obligations they contain. Each quasi-stationary level of a setting has its optimal population requirements. and (2) the ecological environment appears to be. Different behavior settings on the same level of functioning. We sometimes call them self-discipline. . 21–22) He concludes: Behavior settings with less than optimal people for their homeostatic levels are self-disciplining settings. . . 1960. In particular. 3. . falls below the minimal number required by its homeostatic level. . However.xli Introduction 2. I would like to close with two remarks: (1) Brunswik’s environment-environment unit appears to be subject to more than empirical probabilistic laws. 5. 48–49) Barker’s contribution—as is his body of scholarly work in general—is novel and interesting and would seem to have continuing applications today. 6. . its population. . Of all the equipment and paraphernalia of a setting. . especially. 4. Of all the attributes of settings. people are among the most immediately malleable and adjustable. and therefore with the same optimal population requirements. pp. with the greatest degree of independence possible. When the number of people in a setting. . into the ecological environment. the growing acknowledgment that community reintegration and quality of life are vitally important ends of rehabilitation efforts and that rehabilitation cannot really be considered a successful endeavor unless an individual is supported to the point of maximal participation in the life of the community. (Barker. (pp.

Joe Brown for his superlative work as copy editor of this volume of the Nebraska Symposium on Motivation. and presentation of chapter materials contributed immeasurably to the clarity and coherence of the final product. Mr. (1966). stimulus aspects sometimes thought to provoke increased engagement. In M. Barker’s (and his students’) techniques for identifying and cataloging community venues can serve as a guide in expanding rehabilitation practice to include such analyses. 49–66). Lincoln: University of Nebraska Press. E. J. Nebraska symposium on motivation (Vol. novelty. clear and patient counsel. Nebraska symposium on motivation (Vol.). In this regard. In International encyclopedia of unified science (Vol. organization. and unflagging good humor throughout were notably catalytic in successfully bringing the 52nd edition of the symposium together in its present form. 1–49). Jones (Ed. and/or increase arousal. 14. in addition to Barker’s work and the models of Brunswik and Heider (already referenced). In D. discussed below. additional useful resources include Gibson (1979) and Wicker (1984). . As will be evident. pp. References Barker. The ecological approach to visual perception. Compare Theodore Newcomb’s “A-B-X” model. interest. Notes The editors would like to acknowledge and express a special thanks to Mr. 2. Burke. Pt. 656–750). Ecology and motivation. pp. R. Levine (Ed. Chicago: University of Chicago Press. Gibson. and stimulus change. Lincoln: University of Nebraska Press. Boston: Houghton Mifflin. pp. (1960). the conceptual breadth of material and diversity of perspectives reflected in this volume are considerable. J. Brown’s consistently resourceful and creative suggestions concerning the substance. Brunswik. (1955). Linear models for Pavlovian conditioning. Mr. 2.xlii modeling complex systems environment—the behavior settings and social/interpersonal environment to which an individual will be returning—before an optimal rehabilitation treatment plan can be developed and delivered. All volume editors should be so fortunate! 1.). 1. Brown’s timely and precise questions and observations. C. J. R. (1979). Variables such as curiosity. The conceptual framework of psychology. 8.

Vol.). Lincoln: University of Nebraska Press. Erlangen: Philosophische Akademie. Spaulding (Ed. R. Jones (Ed. Motivation in social behavior. 12. Lewin. On perception. A. R. 1–21). Nebraska symposium on motivation (Vol. Psychological Issues. Nebraska symposium on motivation (Vol. psychological fields. Behavior as “intrinsically” regulated: Work notes towards a pre-theory of phenomena called “motivational. Walker’s paper. (1951). H. pp. 18. (1938). Die physischen Gestalten in Ruhe und im stationären Zustand: Eine naturphilosophische Untersuchung von Wolfgang Köhler. S. W. K. 8. The bottleneck of attention: Connecting thought with motivation. Krech. D. (1965). In D. Levine (Ed. Nebraska symposium on motivation (Vol. pp.). Simon. (1938). Krech. (1960). Logan. New York: Harcourt. Lincoln: University of Nebraska Press. A note to the reader. Dynamic systems as open neurological systems. 42–87). 281–321). (1994).). F. 1. 2. pp. (1964). pp. 1–123. 96–98). The nature of the drive as innate determinant of behavioral organization. (1950a). K. W. The growth of the mind. 1(3). Psychological Issues. Leeper. 4. Jones (Ed. (1924). Nissen. pp.). R. 1(3). The place of value in a world of facts. Integrative view of motivation. (1948). Lincoln: University of Nebraska Press. Lincoln: University of Nebraska Press. No. (Eds. A motivational theory of motivation to replace “emotion as disorganized response. Notes toward a psychological theory. Psychological Review. Levine (Ed. 5–21. In W. and hypothetical constructs.). Jones (Ed. Jones (Ed.” Psychological Review. Koch. Köhler. Krech. (1959).). Vol. Dynamic systems. (1956). W. Journal of Personality. (Original work published 1921) Köhler. 4.xliii Introduction Heider. (1954). 57. W. and the psychological environment. D. event structure. Lincoln: University of Nebraska Press. Brace. Nebraska symposium on motivation (Vol. (1949). v–vii. D. Lincoln: University of Nebraska Press. 345–361. Pittsburgh: University of Pittsburgh Press. Current theory and research in motivation (Nebraska Symposium on Motivation. Nebraska symposium on motivation (Vol. The Gestalt theory of motivation. Koffka. Cognition and motivation in psychological theory. Krech. (1959). 139–161). Some needed developments in the motivational theory of emotions. H. G. ———. In D. New York: Liveright. Durham nc: Duke University Press. Dennis et al. 283–290. In M.” In M. T. Lincoln: University of Nebraska Press.). In M. In W. Serial No. D. 66–87. (1950b). Current trends in psychological theory (pp. S. 111–139). 57. Comments on Edward L. 4). Vol. 145–172).). D. 25–122). Newcomb. 41. Psychological Review. F. The conceptual representation and the measurement of psychological forces (Contributions to Psychological Theory. 55. . 1. cognition. (1925). 13. W. (1953). and emotion (Nebraska Symposium on Motivation. R. pp. Leeper. Klein. R. In M.

Lincoln: University of Nebraska Press. Purposive behavior in animals and men. E.xliv modeling complex systems Tolman. E. (1962). Walker. (1984). Psychological complexity as a basis for a theory of motivation and choice. 1–46). (1948). pp. (2001). E. Wicker. International classification of functioning. W. Jones (Ed. C. pp. 47–95). Levine (Ed. . E.). (1964). Geneva. Vinacke. Psychological Review. Nebraska symposium on motivation (Vol. Nebraska symposium on motivation (Vol. 12. (1932). 10. Motivation as a complex problem. In M. C. 189–208. 55.). L. An introduction to ecological psychology. A. In D. New York: Century. disability and health. W. R. Tolman. Cognitive maps in rats and men. World Health Organization. Lincoln: University of Nebraska Press. Cambridge: Cambridge University Press.

of which a prominent modus operandi was. Theoretical predic- . 470).” which occurs “when men study words and not matter” (Bacon. and emanating from within our discipline.” and an absence of salient direction for potentially profitable future thrusts of investigation.Composition and Uses of Formal Clinical Cognitive Science Richard W. p. 200 (bk. Neufeld University of Western Ontario “In every special doctrine of nature only so much science proper can be found as there is [applied] mathematics in it. Paul Meehl (1978) bemoaned the slow progress of “soft psychology” (which presumably included psychological clinical science). 1)). He indicted the dominant research paradigm (Kuhn. 1970. and for that matter remains. 1937. 1962).and Pearsonian-based statistical methods. Fisherian. More contemporary. who decried “the distemper of learning. These quotes are extracted from commentaries on activity done in the name of science by their authors’ contemporaries— activity possibly characterized by considerable redundancy of effort to resolve the day’s purportedly key issues.” So wrote Immanuel Kant about the status of natural philosophy. contentious debate devoid of clear resolution. J. This declamation seemingly reflected a certain frustration not unlike that possibly experienced almost two centuries earlier by Francis Bacon. as he saw it in 1786 (Kant. a lack of precise place keeping regarding the status of important segments of “the body of knowledge.

nevertheless. It constitutes a new form of construct validity. not because they were toppled by uncorroborated risky predictions. providing additional incentive for the particular sphere of application. It might. whereby the ever-popular anova and related tables of results were. Novel developments include proposals for monitoring cognitive aspects of an individual’s response to treatment over time and. Steiger & Fouladi. (Most readers will have their favorite accounts of such instances from their particular fields of expertise. Cohen. Benefits and selected (practical) liabilities are described. but because the discipline became inured to them and their proponents grew weary of marketing them. Following on that is an exemplary case in point from the domain of clinical cognitive science. a clear delineation of what does and what does not constitute formal theory is presented. pp. and to their combination. general. Meehl noted that “theories. 1991). It is maintained that the implementation of formal theory in psychological clinical science can go far to redress current quagmires and accelerate progress. in effect. clinical cognitive science. 17. using . Emphasis of course is on the field with which I am most familiar. 1988.” at least in fields of soft psychology. 16. as nature is said to abhor an absence of association. typically fell by the wayside. likewise.) I would contend that the foregoing lamentation retains considerable currency when it comes to contemporary psychological clinical science. To begin. be added that their proponents need not worry because it is well-known that old theoretical ideas frequently reemerge in different guises and with modernized labels (Staddon.2 modeling complex systems tions were derided as somewhat anemic. One such by-product emanates from a personal “labor of love. summary statements of such power (cf. dynamic assessment of treatment-program efficacy with respect to cognitive functioning. to psychological stress. typically embracing simply the presence of nonzero relations among ad hoc measured variables. However. This observation provides much in the way of motivation for the developments presented here. the merits of formal theory disclosed in that domain of investigation are. spin-offs that impinge directly on prominent issues in this area of study. arguably. 1997). so-called support rested on statistical power. are expounded on. Some surprising noteworthy spin-offs from excursions into formal modeling.” entailing the appropriation of stochastic modeling to the study of cognitive efficiency as it relates to schizophrenia.

3 Formal Clinical Cognitive Science mathematical heuristics. revealing the shortcomings of typical efforts at redress. mathematical derivation. however. predictions of empirical observations are dictated by syllogistic verbal reasoning.g. computer language. the Journal of Abnormal Psychology. is presented in Neufeld. The problem is translated into simple but comprehensive measure-theoretic terms. credited with motivating the investigation at hand (further elaboration of informal and other systems. or at least guiding question(s). It is at this point that the formal subsystem makes its entry. no particular governance by a set of rules constraining the format by which the verbal arguments are posed or the way in which one statement follows from another (e. or symbolic logic). A survey of contemporary issues of American Psychological Association clinical-science journals (e. Mea- . the study is then executed and results obtained. Braithwaite (1968). The most prominent type of theoretical system in psychological clinical science. namely. Summarily speaking. From there formal theory is shown to recast the entire issue in its own terms and in so doing render it as more or less obsolete. B. can profitably be used as a point of departure in explicating the nature and role of formal theoretical systems. The sequence of qualitative arguments culminates on transition to prediction(s) of results. the Journal of Consulting and Clinical Psychology). Mixed deductive systems consist of formal and nonformal subsystems.. There is. Within the nonformal subsystem. 1989). Another bears on an issue that I have engaged more or less since its appearance in the literature. possibly accompanied by visual aids and most certainly by antedating empirical findings. and any number of psychiatry research journals. still-cogent taxonomy of logical-deductive systems in science provided by R. reveals the following predominant structure. Demarcations of Formal Theory Described here are the features of formal theoretical systems that for all practical purposes distinguish them from nonformal and quasiformal systems.. Braithwaite’s mixed deductive system. Drawn on is the rigorous. the so-called psychometric-artifact problem in the study of differential cognitive deficit. in selected domains of applied psychological science.g.

Statistical treatment is part and parcel of a formal subsystem because its constituent operations rest on closed-form formulas that. Meehl. Formal theory can also be distinguished from prominent higherlevel statistical methods. & Viken. the entire logical-deductive enterprise is rendered informal. In other words.. Conversely. provide the data array submitted to statistical analysis. as will be demonstrated below. for example. verbal schemata are. as a “mathematical abstraction of an empirical process. The nature of authentically formal theory can be thrown into sharper relief by considering procedures that may appear to qualify as the former.. a stochastic model. Whereas statistical computations constitute mathematical methods for assessing arrays of values rendered by empirical investigations. Apart from a rare application of synthesizing quantitative models of possible agents of surveyed outcomes.g.4 modeling complex systems sures of dependent variables. as a rule. restatements of what is already known (e. For example. formal theoretical models are concerned with processes that make the data what they are (although. guided. Whereas the typical statistical apparatus (often elegant in its own right and with a virtual industry of software support) is substantively generic—its application transcending theoretical-content domains—formal theoretical models are content specific. expedited by associated software. in the explanatory chain of events. e. are a case in point. formal theoretical approaches replace nonformal reasoning in the above mixed deductive design with a formal logical-deductive subsystem. formal theories are far from mute about ways of analyzing the data whose genesis they address). pending closer scrutiny. whose development is governed by probabilistic laws” (p. emerge from theorem-proof continuity (nowadays increasingly accompanied by numerical simulation). Note that so-called qualitative research methodology replaces this formal data-analytic arm of the logical-deductive system with nonformal. Doob (1953) has described one such type of theoretical formulation. potentially useful in their own right to be sure.g. thus creating an entirely formal logicaldeductive structure. McFall & Townsend. 1998. such models weigh in with quantitative constraints ab initio. by the nonformal theoretical apparatus (see. Townsend. Integrative and evaluative reviews. In this way. 1995). 1978). verbal operations. v). by and large. if not “prescribed” as such. including “structural-equation causal mod- . McFall.

1988. With respect to sem. cf. Moreover. the hypothesized structure of interlatent-variable linkage and the relative magnitudes and/or signs of its associated path coefficients themselves are posited nonformally. & Baker. genetic endowment. Fiore. 2003.g. confirmatory factor analysis forecasts factor loadings and coefficients expressing interfactor relations. including those involving higher-order factors. 1982. Smith. the data configuration corresponding to the conjectured distribution is not in and of itself the product of formal-system reasoning. 2003). Concern is with an empirical covariance structure. MacCallum & Ashby. The chief subject of prediction for taxometric methods. Ruscio & Ruscio. definitions. Formal theoretical systems as well entail parameter estimation and testing for goodness of fit between predictions and observations. Selected parameters. and assumptions.5 Formal Clinical Cognitive Science eling” (sem. Waller & Meehl. Their parameters. exemplified below). 1989). specific symptoms or syndromes). e. are substantively significant. Maxeig. ordinarily. Suffice it to say that similar observations attend typical applications of hierarchical linear modeling. Once more. (It is surmised that criterion values for contemporary indexes of overall statistical fit. rather than derivations spawned by specific axioms. or nonformal argumentation. can be estimated. or other target characteristic). as imbued by their explanatory roles in the theoretical model in which they participate (Braithwaite. In a similar vein. Jorenby.. however. and L-mode) indicative of dichotomous versus continuous latent distributions of target variables (e. Maxslope. 1992. taxometric procedures (Meehl & Golden. is certain configurations of data summaries (calculated via methods prominently known as Maxcov. and hierarchical linear modeling (Bryk & Raudenbush. 2004.1 for all intents and purposes meaning the path coefficients of conjectured relations among the posited latent variables. By these characteristics. constitute the formal subsystem of a mixed deductive system (elaborated on in Neufeld.g. in turn.. taxometric methods and sem fall into the category of statistical analysis and. however. the subject of prediction comprises its parameters. Mambac. their . Kline. 1998). 1986).) Note that. Tomarken & Baker. see Piasecki. 1968. such as base rates of a taxon (a discrete group with the symptom. a sort of overarching prediction. in that way. but one whose features are defined according to antedating empirical observations. have been met.

the history of science makes it clear that systematic retreat from decidedly formal inferential methods inevitably detracts from progress (e. This treatment contrasts an essentially generic interpretation of parameter values. empirical covariance structures). Assets of Formal Theory explanatory insights A major advantage bequeathed by formal developments is the unique insight afforded into the studied phenomena.g. even superior. separated by distance r. Interplay of critical variables is disclosed by inspecting the structure of derived predictions and the steps to obtaining them. 1985). 2004a. they are vulnerable to the onslaught of frailties in reasoning and inefficiencies accompanying dependence on essentially verbal contentions. Link. As such. The former. 1984). Miles & Huberman. 1982. The law defines the force of attraction between two particles. The constant varies only with the units of . a notable case in point. I will return momentarily to the nonformal system widely known as qualitative research methodology because of its growing popularity. 1991. 1968. defects that formal procedures are designed to confront (Hintzman. especially in the social sciences. with masses m1 and m2. Newton’s universal law of gravity: m1m2/r2. 1985). Nonformal strategies give the appearance of flexibility in dealing with concepts and measures whose implementation may strain formal systems. or other model properties. despite their purported checks and balances (e.g. across even vastly divergent content areas and a focus on highly similar aspects of data (e. accommodate such challenges handily. Kline. 1994). Kline. Braithwaite.g.. for instance. Examples in the hard sciences of course abound.. alternative to systems of science implanted with formal methods. Indeed.g. 2004b).. Qualitative research increasingly has been touted as a legitimate.6 modeling complex systems variation generates selective changes in empirical data sui generis to the localized problem (e. Consider. Townsend & Wenger. but because they simply sidestep the constraints on precision of expression demanded by formal systems or subsystems (Staddon. not because they somehow transcend the explanatory merits of formal systems.. however.

In fact. for one. k is an increasing function of task-relevant competence (associated here with practice). One such model expresses the expected (mean) latency for proofreading lines of prose (e. An illustrative instance entails the stochastic dynamic modeling (see. is decipherable from the nature of the formal assertion and from the way in which the law serves in broader theoretical contexts. can be appreciated contextually by considering its operation in a formal deductive system. reflecting diminished speed). As opposed to ad hoc interpretation. thus. McFall et al. for the summary task load m can. Busemeyer. The function of each term..7 Formal Clinical Cognitive Science measurement.. 2002).. Glass & Singer. and m indicates task load. In psychological clinical science. making for their “interaction”: the effects of stress can be mitigated by practice. e. r conveys stress-related impairment of performance speed (r goes up with increasing stress. The expression is mr/(k 1) (Neufeld. quantified as the mean number of processed stimulus elements (chunks of letters scanned per line of prose). The expected latency for a given level of stress. the form of the stress-practice interaction is one of “superadditivity” of the interaction’s second-order difference: [(mean latency under higher stress and lower practice) (mean latency under . 1995) of the effects of stress (ambient noise) on cognitive efficiency (Weinstein. The actual physical mechanism of gravitational pull may be enigmatic. 1980). statistical interactions.. be predicted as a function of substantively significant parameters. in turn. Similarly. and the effects of this interplay on performance latency are “scaled” according to the prevailing task load m. 1996. the speed of light. Moreover. for example. and of the law itself. something of the nature and function of Einstein’s limiting constant. Such exposition. elucidation of obtained data configurations can be bolstered by stipulating relations between observed and inferred variables in terms analogous to the above. 1974). to be taken seriously. practice becomes more important as stress goes up. can be understood from formal depiction of the psychological processes under study and associated scales of measurement (e. In this formula. and amount of practice. elevating the mean. but the mathematical definition is not.g. The expression also discloses the fabric of interplay between stress and practice.g.g. any proposed physical mechanism would have to comply with this mathematical statement. discloses that any meaningfulness of a larger value rests on a solution to the imaginary number ( 1)½. 1972).

but can also reduce computational steps. endowing it. elaborated on below). Hastings.3 . in the second instance. the probability of the theory being tenable. 1984). for example. ease computer memory demands. It is enticing to depend on numerical simulations as surrogates for analytic derivations. & Vollick. 2000. accordingly. & Peacock. Partly for this reason.2 self-diagnostic. predicts rain in the Northwest during November. a strong prediction stipulates the precise amount to fall within a given time period in a specific location (Meehl. approaches 1. In Bayesian terms. including that located in psychological clinical science (Neufeld et al. Formal theory arguably makes for desirably strong prediction. 1991).8 modeling complex systems lower stress and lower practice)] [(mean latency under higher stress higher practice) (mean latency under lower stress and higher practice)] > 0 (Townsend. At the opposite extreme. also approaches 1. 2002). in the first instance. analytic expressions can also serve the practical cause of computational economy. It is readily shown that. 1978). theorists in longer-established sciences will opt to take analytic solutions to the limit. 2002).0 and. A weak meteorological forecast. and increase speed when simulations are inevitable. Carter. closed-form. Boksman. before invoking numerical algorithms. Note that each of the above parameters is also entrenched in its own quantitative infrastructure. according to their detailed structures. given that the predicted evidence occurs. with further analytic and substantive meaning (Evans. given the theory. Predictions emanating from strong theory stand to be formidably precise. In like fashion. This is why formal theories die while nonformal theories seemingly defy all laws of natural selection (Staddon. Beyond their revelatory function. apart from viability of the proposed theory. Jetté. bold conjecture is tantamount to a predicted observation being negligible. given contemporary computational technology. self-indicting properties Desiderata of useful theory are Popperian bold conjecture and falsifiability.0 (Neufeld.. distinct falsifiability is identified with an essentially zero probability of the tendered theory being defensible on failure of its specified prediction. however. the probability of the evidence. Analytic derivations. not only enhance understanding.

as discussed above.. they necessarily ease the problem of empirical measurement. Townsend. 2004a. cf. Owing to the precision of the formal deductive system. as discussed above). therefore. Displaced. In fact. Townsend & Fific.. 1995. notably parameter values. tests for goodness of fit and comparative goodness of fit). 2002). Neufeld & Williamson. As with longer-established sciences (e. theory is potentially strong enough to mandate certain empirical ramifications. Variation in model properties.. superadditivity of means. is the often ad hoc process of selecting off-the-shelf measures. The latter entail the contour of data across gradations of independent variables and their combinations (e. Townsend & Wenger. Such patterns can. 2004). those properties apply to indexes substantively tied to the theory at hand. expedite quantitative expressions of their relative superiority (e. however. serial processing. Nevertheless. moreover. even with very weak assumptions. in that.9 Formal Clinical Cognitive Science Formal theories provide specific predictions that can compete directly in terms of differential conformity to observed values. across individuals can also be profitably accommodated through Bayesian techniques (Batchelder. competing models cannot produce each other’s configurations (e. 1978).. formal deductive systems stand the selection of measures on its head. 1998. Chechile. In addition.g. 1990). Riefer & Batchelder. Not least among the endowments of formal theory is its inherent challenge to preferred interpretations of addressed phenomena. conjecturally connected to dependent variables (McFall et al. Explicitness of expression helps make salient current boundaries of . maximumlikelihood status).. in turn. 2004a).. physics examples.g.” e. to be sure remains applicable (e. If formal models mandate their own diagnostics.g. Individual differences in model expression can be accommodated within the context of model diagnostics. Assessment of the generated measures. Meehl. Included is variation in the structure of cognitive transactions (e. be mutually exclusive. Townsend & Wenger.g.g. 1988.. models can prescribe their own qualitative signatures of validity. Neufeld et al. “empirical equivalence. 1998. for the usual statistical properties of fallible estimates (e.g... and selected combinations of these structures. especially befitting clinical science and assessment.g. 1996). variations on parallel processing.g. Predictions set in quantitative terms.. theory and measurement become intertwined. bias.

10 modeling complex systems knowledge—flaws and limitations in extant formulations. In particular, empirical equivalence, on current measures, of nonetheless opposing positions is disclosed. However, so too are alternatives that can break the encountered experimental-mimicking logjams (Casti, 1989; Townsend, 1990).

absorption of theory-exogenous variance
With formal logical-deductive systems, random empirical variance can be explained by the theory itself. Quantitative theory, in other words, encompasses apparent observational indeterminacy, instead of relegating it to “error” or “noise variance” (cf. anova linear model, e.g., Kirk, 1994; “local independence,” Bollen, 2002). An additional classification of theoretical systems, cogent apropos this avowed asset, is that of Busemeyer and Townsend (1993; depicted in Table 1). Systems are categorized as dynamic versus static and as stochastic versus deterministic. First, static stochastic models (in the lower-left quadrant) predict frequencies of response values (e.g., categories, ratings, and so on), if not the times for producing them. An example is Tversky’s elimination-by-aspects model of preference and choice (Tversky, 1972a, 1972b; see Batsell, Polking, Cramer, & Miller, 2003). Comparative frequencies of choosing items from those available are considered to follow a multinomial distribution (e.g., Evans et al., 2000). Accordingly, probabilities of selection are the main focus. The probabilities are governed by parameters of the elimination-by-aspects model, which are subjective utilities of the item attributes (e.g., the likelihood of winning a lottery vs. its payoff; Rappaport, 1983; for applications in stress-and-coping research, cf. Kukde & Neufeld, 1994; Morrison, Neufeld, & Lefebvre, 1988). Note that an earmark of stochastic models is their prediction of probabilities, or distributional frequencies of events, versus predictions of an all-or-none nature. Turning to the upper-right quadrant, we see that deterministic dynamic theory is exemplified most notably by nonlinear dynamic systems models, popularly known as chaos theory (e.g., Gregson & Pressing, 2000; Koçak, 1989). To illustrate, variables such as psychological stress and coping can be depicted as being in constant flux, reciprocally influencing and being influenced by one another. The

11 Formal Clinical Cognitive Science Table 1. Schema of Theoretical Systems, with Selected Examples
Static Deterministic Verbal theory; Nondynamic flow diagrams Some decision-and-choice models (e.g., selected subjective-utility models) Dynamic Deterministic nonlinear dynamic systems (“chaos-theoretic systems”) Dynamic extensions of decision-and-choice models; classic stochastic models; nonlinear dynamic-systems models, with a stochastic element

Stochastic

Note: Adapted from Busemeyer & Townsend (1993).

system is represented by differential equations, each indicating the momentary change in a variable. The changes themselves are defined in these equations essentially by the variable’s own extant state and those of the other variables making up the network (Neufeld, 1999b). Stochastic dynamic theories are exemplified by time-related extensions of decision-and-choice models. The elimination-by-aspects model (discussed above), for example, has been reformulated by Marley and his colleagues (e.g., Marley, 1989; Marley & Colonius, 1992) so as to add temporal predictions about selection responses to the original model’s predicted frequencies of selection from among the competing items. Busemeyer and Townsend (1993; also Roe, Busemeyer, & Townsend, 2001) have presented an empirically compelling model of decision and choice, providing for speed of choice, type of choice, and their interconnections. Variability of expressed preferences and their latencies are, once again, accommodated, in that the crux of prediction involves frequency distributions. Stochastic dynamic models represent possibly the most widely used version of formal theory in psychology. Generally speaking, they depict probability distributions of events as a function of time (see, e.g., Johnson, Kotz, & Balakrishnan, 1994, 1995; Luce, 1986; Townsend & Ashby, 1983). Part of the model fabric, then, is variability of event latencies across occasions of observation (e.g., experimental trials). The provision for variability is graphically illustrated by considering the model-prescribed variance of time taken to complete a relatively simple cognitive process.

12 modeling complex systems The process comprises k stages, each stage having a specified frequency distribution, as a function of time. Each stage is executed at the rate of v stages per unit of time (whereby the mean stagewise completion is 1/v time units). Such a process may correspond to the encoding of k stimulus features (e.g., curves, lines, and intersections of an alphanumeric item) into a cognitive format facilitating further processing, such as selected manipulations involving the item in short-term memory (e.g., Sternberg, 1975). By the present model, known as the Erlang distribution (Evans et al., 2000), the mean time across trials for process completion is k /v. The across-trial variance of completion, in turn, is k /v2. Note that this statement of dispersion is squarely embedded in the substantively significant parameters of the theoretical model of process completion. Further provision for stochastic properties is available by allowing for individual differences in the values of the parameters k and v. This extension (a mixture model) “acknowledges” interparticipant variation in the above variance and related model expressions (e.g., Neufeld et al., 2002). Latency variance for a given value of k and v maps onto the familiar within-participant variance in (m)anova, and variation across participants in k and v maps onto between-participant variance (see, e.g., Parzen, 1962, p. 200; also “A Stochastic-Modeling Translation of the Statistical-Property Issue” below). Other contemporary exemplars of dynamic stochastic models include selected extensions of multidimensional scaling (see, e.g., Schiffman, Reynolds, & Young, 1981). These extensions quantitatively relate categorizations, or ratings of multidimensional stimuli, to their cognitive-process underpinnings, where the variable of time plays a central role (e.g., Carter & Neufeld, 1999, in press; Nosofsky, 1992; Takane & Sergent, 1983). Before leaving this subsection, it should be noted that nonlinear systems theory can, in principle, be augmented to incorporate a stochastic component. A specifically random element can be added to the otherwise deterministic differential equations describing intervariable influences across time (e.g., Brown & Holmes, 2001; Huber, Braun, & Krieg, 1999). This interlaced stochastic thread perturbs the momentary change in each variable and the consequent influence of those changes on the coupled network of variables. Interestingly, the temporal trajectories of such a “stochasticized” system can deviate dramatically from those of its strictly deterministic counterpart.

13 Formal Clinical Cognitive Science Suffice it to say that, overall, unaccounted-for variance in observed values may never be eliminated (cf. Gilden, 2001). However, such “model-exogenous variance” stands to be substantially diminished with increasing comprehensiveness of theory construction.

aesthetic appeal
P. A. M. Dirac, a Nobel laureate in physics, once declared: “It is more important for our equations to be beautiful than to have them fit the experiment” (quoted in Freedman, 1993). Given the historical success of beautiful equations, their apparent failure was held to be a temporary aberration. Perhaps, in their presented form, the equations were an inadequate approximation of empirical events. Or, possibly, the experimental observations themselves were flawed. Either of these shortcomings could be redressed. Beautiful equations were intrinsically self-vindicating, and the rest was detail. At the very least, if equations were to stand up to empirical challenge, they must spring forth from among the ranks of the beautiful, whereas equations that are not beautiful are inevitably doomed to experimental defeat (cf. Smolin, 2006). Formal theory thus potentiates the scientific merit of equational elegance. There is no inherent reason why psychological clinical science should be barred from this asset, which is, arguably, a proprietorship of formal developments. What, then, makes an equation aesthetically appealing? One criterion is succinctness. The simplicity of beautiful formal expressions belies their profundity, which becomes apparent when their copious and far-reaching implications are unveiled (Polkinghorne, 2003). One example, emanating from the field of nonlinear dynamics, is Mandelbrot’s formal representation of fractals. Although the representation is parsimonious, a feature of its constituent variables is that they take on a multitude of intriguing trajectories. The proliferation of exotic patterns has been widely disseminated, thanks to modern computer-graphics technology (e.g., Gleick, 1988; Townsend, 1994). A more modest example is available from the Erlang distribution of cognitive-process completion (described above). Here, the probability of completing the stated process by a given time interval, t, is

14 modeling complex systems
(k’ – 1)! – (k’, vt). (k’ – 1)!

(1)

Note that one can safely bypass the definition of each term in the equation and its equivalents (below) and still appreciate the pithiness of this expression. The somewhat involved operations, summarily contained above, include
1–
00

vt

x k’–1 e –x dx/(k’ – 1)! =

00

j=k’

j (vt)/j! e–vt.

(2)

These formulas, in turn, can be used to create visual aids that facilitate insight into the interplay of model parameters (for a treatise on theory-related intuitions afforded by visual imagery, see Clark & Paivio, 1989). Three-dimensional response surfaces convey variation in the present probability across alternate combinations of k , v, and the time window, t. Two examples are provided in Figure 1, for values of k ranging from 1 to 10 and values of v ranging from 0.01 to 3.0. The lower response surface is constructed for t = 4 and the upper one for t = 7. This figure presents but two instances of a virtually infinite number of possible graphic portrayals of the formally stated process. This is why, if a picture is worth a thousand words, a formula is worth a thousand pictures.

liberating qualities
Formal logical-deductive systems are constrained by observationally and quantitatively principled initial assumptions, definitions, and axioms.4 Building on this foundation, rigorous derivations render deductions about events of principal interest. In this way, precisely framed empirical implications can, in effect, be evaluated for their potential importance, in advance of actual empirical testing. Fruits of these labors, then, include, not only the resulting insights, but also a potential increase in the overall economy of inquiry. Important questions can be posed before launching out empirically. Will the forecasted yield of information justify the associated investment? Is that which may be found actually worth finding? Does the theory posit results that, if corroborated, could significantly affect explanation, measurement, and, indeed, future predictions? The point to be made is that the theorist is emancipated to safely

15 Formal Clinical Cognitive Science

Figure 1. Probability of completing k stages in four time units (lower surface) and seven time units (upper surface) with each stage transacted at rate v (see Equation [2]; k ranges from 1 to 10, and v ranges from .01 to 3.0.

wade into the staked out substantive territory, with the above and related issues in hand, because her deliberations are constrained mathematically. These constraints represent an antidote to the hazards of armchair philosophy and psychology’s ensuing reaction against exploration free of constant empirical vigilance (cf. instrumentalist vs. realist distinctions in scientific theorizing; e.g., Casti, 1989). As for the empirical enterprise, formal theory can increase the latitude of investigation in clinical science. It can do so by incorporating abnormalities associated with psychopathology into hypothesis formation and testing. Such possibilities are exemplified in the cognitive-science application presented below. In the enterprise of clinical science, certain hypotheses regarding pathognomonic disturbances may defy often-used methodological options, and formal theory can supply essential alternatives. To elaborate, one tack to take when testing hypotheses about cognitive debilities, for example, involves attempting to duplicate performance deviations in normals using experimental manipulations that are designed to mimic their etiology in patients. Formal theory can complement this strategy and, moreover, pick up where it leaves off. The approach is to modify quantitative mod-

16 modeling complex systems els of performance in ways implied by the hypothesized agent of deviation and then to evaluate the accuracy of the resulting predictions (elaborated on below). Such methodological endowments come into play especially where tendered sources of disturbance are beyond the reach of direct manipulations. Selected agents of dysfunction, although model hospitable, may be fundamentally alien to normal functioning or demand unethical extremes of experimental induction.

barriers: attitudes, aptitudes, and investigator role
Ideally, formal theory at minimum should be a ready-to-hand arrow in an investigator’s hypothesis-development quiver. It is recognized, however, that there are certain practical barriers to this ideal. One such barrier quite simply involves prevailing attitudes toward formal theory. Mathematical models are regarded with suspicion in some circles unless their implications are accompanied by readily appreciated parallels. In clinical science, such parallels often take the form of behavioral neurological mechanisms that clearly align with the mathematical developments.5 Interestingly, this position is counter to that of longer-established disciplines, where accepted mathematical necessity predates identification of a corresponding physical mechanism (e.g., Newton’s universal law of gravity, discussed above). One price paid for suspended belief in quantitative formulations is the forfeiture of potential unification of coexisting explanatory systems and new angles on synthesizing current evidence (e.g., Herrnstein, 1979). Another practical impediment concerns formal theory’s demands for hands-on tasking by someone closely involved with the subject matter. This requirement contrasts with practices where assignments can be delegated to an associate or a research assistant (as might be done, e.g., in data entry or statistical analysis). Formal modeling, furthermore, being customized to the immediate theoretical problem, is not a matter of output from routinely applied software. This is not to say, of course, that advances in computational power and computer-algebra programs such as maple and mathematica do not substantially aid the theorist’s job.

and applications to individuals for clinical assessment or other purposes . thus contributing to contemporary technology for measuring evoked neurophysiological responses. Positioning of stochastic models of cognitive performance within this framework is pinpointed. Application to Clinical Cognitive Science: The Case of Stimulus-Encoding Dysfunction in Schizophrenia This section leads off with a presentation of the philosophical blueprint guiding the present application. (c) monitoring changes in cognitive functioning with the progression of treatment. and (f) facilitating links of cognitive-performance indexes with clinically significant quantitative formulations on stress and coping. philosophical orientation Associations among cognitive science. (e) complementing neuroanatomical regions of interest with times of interest. however. need not be abandoned while cultivating the formal alternative as an increasingly friendly ally.17 Formal Clinical Cognitive Science On balance. (d) importing the informational infrastructure endowed by the model’s cognitive-science roots. it can be said that such hurdles to formal theorizing may be daunting at first blush. Current practices. at both the individual and the group levels. potential payoffs of the approach are expounded. Its exposition culminates in the appropriation of group-level inferences about cognitive functioning to the individual participant. This appropriation is mediated by the use of Bayesian-based procedures. Their promised value launches an exposition of the parent Bayesian framework dominating the current development. Aided by the present instantiation. (b) customization of the process model’s performance-latency distribution to the individual. entailing selected stimulusencoding dysfunction in schizophrenia. They include (a) precision in estimating performance-model parameters for the individual. clinical cognitive science. Eliminating the present hurdles can reap the substantial dividends on outlay in time and labor that have been enumerated above.

the paradigm may take the form of a conventional visual. parallel. Deliberations include viability of the experimental paradigm and relevance of its response parameters.g.or memory-search task (Sternberg.. Extant models may require elaboration or modification. In the first instance. a model of cognitive performance addressed to the targeted territory of mentation is chosen. Its arrows denote a smooth. Alterations can be directed toward the architectural structure of the modeled processing system and/or toward model parameters. Selection from among contenders is based on congruence between the model’s features (e. The formal account of cognitive processing is then adjusted so as to accommodate performance deviations of disordered individuals. memory. and collateral processes (e. Aptitudes. and so on). for example. if not seamless.g. befitting the currently driving issues. its parameters or other properties) and concepts indigenous to the substantive research question (e.. It may even be necessary to develop a model ab initio (see “Barriers: Attitudes. may be those identified with visual. efficiency of capacity allocation. . Transitions from basic cognitive science among normals to cognitive assessment among patients. Addressed cognitive faculties. and response parameters may include performance latency and/or accuracy. depending on points of contact between the available armamentarium and the clinical-science queries that are pressing. are depicted in Figure 2. processing architecture [serial. transition among these investigative domains. 1975. and Investigator Role” above).18 modeling complex systems Figure 2. Townsend & Ashby. cognitive capacity. stimulus encoding and response mechanisms)..g.6 Several candidate models may present themselves. 1983). or hybrid].

the same cannot be said for the values of model parameters (Neufeld et al. et al. Parameters refers to variables characterizing the processes. cf. in ascertaining and registering the existence of a presented alphanumeric item in a memorized set of items. and those remaining intact signify functions that are spared. beyond that attending response parameters per se. 1985. a “prior distribution. and response operations may be conducted serially. 2002). For the sake of parsimony. members of diagnostic groups (Batchelder.. in parallel. memory-search. Vollick. component processes composed of encoding. parallel vs. 1984. Specifically. In general. Berger. as well as performance-latency distributions. Analogous to the procedure followed in a medical-laboratory scenario. Busemeyer & Stout.g. Through the implementation of Bayes’s theorem (discussed below). such as their size (operationalized as the number of subprocesses making up the process) and capacity (operationalized as the speed with which these subprocesses are transacted)—the parameters denoted above as k’ and v. 2002.” in Bayesian statistical terms. in press. where a modest blood specimen is .. performance-model parameter estimates. 1983). these models become poised to accommodate unique aspects of individual performance.19 Formal Clinical Cognitive Science Architectural structure refers to the arrangement of constituent processes involved in task transaction. the normal model is minimally perturbed. By certain extensions of their stochastic element. in press). It is apparent from the foregoing that the models most closely aligned with the present purposes are those that are stochastic and dynamic (see “Absorption of Theory-Exogenous Variance” above). Fific. 1998. Necessitated adjustments purportedly indicate disorder-affected functions. stand to be individualized with considerable precision. pending realization of the target configuration. Townsend & Ashby. Townsend. however. The distribution of individual parameter values becomes. Structure refers as well to the arrangement of constituent stages (subprocesses) of the said processes (e. in the case of encoding subprocesses. Neufeld. To illustrate. serial input of the physical item features for template matching to those of the memorized item set. processing architecture tends be preserved.. Townsend. respectively. & Neufeld. mixture models are fashioned to incorporate unequal values of performance-model parameters across units of observation—in this case. or in some combination of the two. in effect.

In the case of mixture models. The arrows in Figure 2 are unidirectional. potentially reciprocates benefits back to basic cognitive science.g. v) can be customized to individuals by amalgamating each one’s performance sample with the parameter-distribution prior.g. ascertained for the associated diagnostic group... parametrically homogeneous subgroups can be formed. thus. whose own success or failure adjudges theory robustness (Carter & Neufeld. inasmuch as what has been altered is a distribution.g. With the individual estimates in hand. those that readily accommodate performance extremes are preferred to those that are strained or fail in this regard (Neufeld. there is. There is. 1982. 1980).20 modeling complex systems referred to the broader body of hematological knowledge.. be used to establish the likely epoch of occurrence of the process for the constructed group (e. 2000) and. movement of the deviant group’s Bayesian-prior distribution of parameter values. When it comes to measuring evoked neurophysiological responses (as in fmri and erp). 2001). elaborated on below). conveying the thrust of the current argument. Nor are the layers of investigation mutually exclusive (see the leftmost column of Figure 2). provision for individual differences in the expression of these effects is retained. This widening of application represents a unique opportunity for generalization testing (Busemeyer & Wang. Successful titration of stochastic models can lead to parallel manipulations at the connectionist level of analysis. 2002. The resultant specification of cognitive-process-latency distributions (e. In this way. Successful capturing of deviant performance among clinical samples speaks to model robustness. Roe et al. instead. deviation-accommodating adjustments no longer take place directly at the level of the task-performance process model. Neufeld & Mothersill. reciprocity among the above investigative domains. that of cognitively encoding a presented item into a task-facilitative format) can. estimated “time .. effects of disorder are registered as features of the prior. with the introduction of Bayesian-prior distributions of parameter values. in press. and. in principle. performance-model parameters (e. nevertheless. Note that. Neufeld et al.. As described above. a modest cognitive-performance sample is now embellished with information conferred by the prior distribution of parameter values emanating from the participant’s group. k . Neurophysiological levels of investigation also stand to benefit.

given A. A is often a hypothesized event. This statement appears simple enough. and B becomes a participant’s performance sample comprising N empirical latencies {t1. and Pr(B) is the unconditional probability of B. Ai’ becomes a parameter value of a test-performance process model. until its wide-reaching ramifications are unraveled. . .. in so doing.. (4) where Ai is one of a set of hypothesized events. (5) . The latencies refer to stimulus encoding because cognitive translation of presented items into a format that facilitates collateral functions (e. in calibrating the space-time coordinates of measurement. . 2002). In its present implementation. t2. Ai. (3) where A and B are events. tN}. belies its profundity obviously qualifies this theorem as one of statistics’s aesthetically exquisite equations. Pr(B|A) is the conditional probability of B. and B refers to experimental observation. provides a comprehensive backdrop for explicating the potential assets enumerated at the beginning of this section: G w( |{*}) = g=1 ({Pr(g|{*})}[w( |{*}. complementing neuroanatomical regions of interest. . nevertheless.21 Formal Clinical Cognitive Science windows of interest” may. be prescribed. or the probability of A preceding consideration of B. thus. g)]). search for a matching memory-held item) tends to be elongated in schizophrenia (Neufeld et al.7 or {*} for short. The following formula adapts these equations to the present setting of application and. group and individual differences in the encoding process: bayes’s theorem in the current context Recall that Bayes’s theorem states: Pr(A|B) = [Pr(A)Pr(B|A)]/Pr(B). Perhaps a more familiar general expression of Bayes’s theorem is Pr(Ai |B) = [Pr(Ai )Pr(B|Ai )]/[ iPr(Ai)Pr(B|Ai)]. Pr(A) is the prior probability of A. The parsimony of expression that.g. As in the present context.

Encoding of presented items is identified as the source of observed abnormalities in performance latency. that takes Equations (3) and (4) beyond a theoretically barren actuarial status to substantive significance. The precise role played by the group-specific priors in Equation (5) is pinpointed in its expansion (presented in Appendix A). A variant of a memory-search paradigm (Sternberg. . with their distribution being perturbed according to group membership. and w( |{*}. Note that the equation is partitioned by elongated braces { } and elongated square brackets [ ] according to the respective terms bearing on the main developments (outlined below). . given the set of empirical latencies {*}. . tN}. of the parameter value . The several potential benefits of this Bayesian layout. 1975) eliciting performance deviations among schizophrenic patients is described (Highgate-Maynard & Neufeld. given {*} as well as the prior distribution of for group g. are then detailed. Pertinent developments in quantitative cognitive science are exerted to discern responsible parameters. it is through variation in the prior distributions of that group-related performance deviations become registered in Equation (5). Its model parameters are deemed to vary randomly across participants. moreover. respectively). The study reported in Highgate-Maynard and Neufeld (1986) extended typical memory-search methodology by incorporating meth- . . Pr(g|{*}) is the probability that the participant belongs to diagnostic group g. The theoretical design of the encoding process is delineated. which composes the subject of prediction. It is the presence of cognition-relevant parameters in this equation. [w( |{*}) is the probability. g) is the probability (density) of . 1986). enumerated above. The latter are representative of deviations occurring with other memory-search and related paradigms. experimental paradigm and process of interest What then corresponds to in the present application? The answer requires consideration of the cognitive task on which the model parameters bear and whose performance supplies the latency sample {t1. given {*}.22 modeling complex systems In this equation. As stated above. t2. or probability density (in the case of discrete and continuous parameters.

Half the participants in each group were presented with similarly sized drawings of probe items. were those where the mean of the probe item. “fixed-set procedure”). 1977). Memory-set size. meant that the probe item’s size did not resemble that of a memory-set item. ranging from 1 through 4. . Size attributes of items in the memory set were spaced such that the means of their normative-size ratings (Paivio. 1975) were at least 2 standard deviations of their Thurstonian discriminal-difference size dispersions apart (practically. Scanning memory-held items for the presence of the probe’s size properties was considered to be comparatively taxing. for an examination of dual-coding and related predictions among schizophrenia participants. Wright. the probe item’s mean real-life size rating was at least 1 standard deviation of the Thurstonian discriminal-difference dispersion from the normative mean rating of each item in the memory set. at least 2 standard deviations of the distribution of difference scores between their normative-size ratings. were. were similar in number for each group. Presence versus absence of the probe’s size properties in a member of the memory set (positive vs. 1971. Encoding demands were. It required comparison of the probe and memory-held items’ overall volume. whose correct response was a “yes” button press. deemed to be greater in the latter instance because the initial verbal presentation of the probe would require referral to the imagery system for access to the demanded spatial-size properties (Paivio. Negative trials. 1984). Positive trials. whereas the other half were presented with names. and that of a memory-set member. Here. 1987. Highgate-Maynard & Neufeld. as set against subjective criteria of similarity in this property (Hockley & Murdock. ensuring familiarity with task requirements and the nature of correct responding. Responses were designated as correct or incorrect on the basis of whether or not they conformed to the following criteria for positive and negative trials. identical. accordingly. Individuals with schizophrenia and controls indicated “as quickly and accurately as possible” whether or not the real-life size—“overall volume”—of a presented item (the probe item was either an object or an animal) was similar to that of a member of a previously memorized set of items (the memory set.23 Formal Clinical Cognitive Science ods emanating from Paivio’s dual-coding theory (Paivio. Practice trials. negative trials) was similarly scrambled. see George & Neufeld. for all intents and purposes. 1986). 1986). 1986. varied randomly across trials. whose correct response was a “no” button press.

Considering the current emphasis on latency. response processes. Individuals with schizophrenia taking part in this study included paranoid and nonparanoid subgroups. Specific processes thought to be tapped by this task.g. evinced on the present and related paradigms.. and third. see also Neufeld. and neuroconnectionist assays of temporal and substantive aspects of item categorization (Carter & Neufeld. Neufeld et al. Results from Highgate-Maynard and Neufeld’s (1986) study represent a case in point of delayed completion of the encoding process among schizophrenia participants. amid integrity of remaining processes. Neufeld.. (2002). 1993). (2002. 2002. then. & Highgate. These include studies analyzing expressly the architectures and parameters of item encoding (Vollick. 1994. namely. in press). encoding—is detailed in Neufeld et al. potential significance . including those surrounding provision for potentially confounding clinical and demographic variables and ascertainment of the viability of the paradigm for each diagnostic group of participants (e. Vollick. The clinical significance of this delineation has been described in Nicholson and Neufeld (1993). second. Comparison groups involved nonschizophrenic psychiatric controls as well as nonpatient controls. estimated encoding durations for correct responses. Vollick. Convergent support for this form of deficit. Finally. first.. et al. in press.24 modeling complex systems Further specifics. paranoid subgroups have also tended to be more elongated in probe-item encoding than nonparanoids. the cognitive translation of the probe item into a format abetting comparisons with the memory set or the memory set’s relevant properties. include. Finally. Neufeld & Williamson. are detailed in Highgate-Maynard and Neufeld (1986) and summarized in Neufeld et al. 1999). has been presented in other sources (Neufeld. 1991. 1996). studies combining memory-trace theory and multidimensional scaling in assessing deviant similarity judgments (Carter & Neufeld. it should be emphasized that error rates were comparable across groups and did not contribute to latency in any confounding way. Vollick & Neufeld. applicability of normed item properties to each one). Convergent support for the modeled composition of the deficit (elaborated on below) arises from investigations of divergent paradigms and forms of analysis. trimming from the examined latency data those times estimated for processes residual to the one of current focus—namely. the comparisons proper. 2004). The present developments are expedited by focusing on the paranoid-schizophrenia participants and the nonpatient controls.

we may know perfectly well the group to which the present individual belongs. Nicholson & Neufeld. has been discussed in other sources (Neufeld et al. to the target expression and spotlights its associated computations (consider remnants left by cancellation of terms.g.0). Equation (5). The encoding process is highlighted because. This constraint reduces the parent equation. it is one that is diagnostic of group-performance differences. therefore. It comes into play as well when attempting to evaluate dynamic aspects of treatment-regimen efficacy in terms of estimated temporal changes in group density. We may isolate on the latter by setting Pr(g) = 1. of the several processes putatively involved in task transaction in toto. process-latency model The process-latency model has already been alluded to in the earlier discussion of the nature of stochastic dynamic models (see. The rationale . g)]. strictly with the expression [w( |{*}. where g is the relevant group. Our concern is. given the performance sample {Pr(g|{*})}. Thus. This provision is brought into play (see below) when we wish to use performance samples taken over time to profile an individual’s progress in terms of proximity to variously symptomatic groups (cf. e. 1993. we first evaluate the probabilities of group membership. “Absorption of Theory-Exogenous Variance”). g)].25 Formal Clinical Cognitive Science as a source of symptomatology. Neufeld & Williamson. 1996). parameter estimation at the diagnosticgroup level of analysis Equation (5) above is relatively complex (see the expansion in Appendix A). 1993). notably thought-form disorder (delusions and thematic hallucinations). and then cipher the groupwise posterior probability (density) of . but it comprehensively allows for the possibility that we may not know the group of origin for the individual supplying the current cognitive-performance sample. which is invoked to portray the stimulus-encoding process. conditional on group membership [w( |{*}. in the expanded version of Equation (5) presented in Appendix A. It consists of the Erlang distribution.. when Pr(g) is set equal to 1.0.. Meanwhile.

(6) which renders a mean latency of k /v and a variance across trials of k /v2.g. Erlang distribution. The stages proceed at the rate v per unit of time (seconds. constituent operations of covertly accessing size properties of the probe item). These influences are summarized according to the probability density of process completion at time t: f(t) = (vt)k 1 /(k 1)! ve vt. Ambient trial-to-trial influences naturally infiltrate processing time and perturb this pristine expression. . In the deterministic case.26 modeling complex systems Figure 3. the process would invariably terminate after an interval of k /v. for the adoption of the Erlang and other distributions used here and the qualifying considerations surrounding the present applications have been detailed in Neufeld et al. with various pairs of its parameters. The model postulates k stages of encoding (e.. (2002). in this instantiation).

27 Formal Clinical Cognitive Science The values of parameters k and v are not invariant. Haynes. 1996). I consider the distribution of each. k arguably conveys the speed-augmenting effects of task-related competence or performance skill (Neufeld.. 2000). the present mixture model provides for individual differences according to the group’s Bayesianprior distribution of parameter values that randomly enter into the above process-latency model. Figure 3 displays the f(t) for four combinations of parameter values. but posited to vary across group members. are involved in the process-latency model. its distribution is identified with the taskwise capacity pool. proportional to the relative frequency of its population occurrence. and r and k correspond to influences on this pool of stress and performer competence. Equation (5) may be readily extended to accommodate the present as representing this pair of parameter values (Neufeld et al. Construct validity for the mixing-distribution parameter interpretations. Its probability-density function. parameter-value mixing distributions It is now apparent that two parameters. in turn. 2001). 1978. As v formally conveys “processing capacity” (e. 1996. In the case of schizophrenia. Sufficient for the purposes at hand is the observation that the distribution of v parsimoniously is regarded as common across diagnostic groups and conditions of encoding load (supporting analyses are enumerated in Neufeld & Williamson. therefore. cf. particularly that of k.. processing capacity is . k and v. 1994. Introduced is a new format of construct validity endowed by the parameters’ formal moorings (cf. Vollick & Neufeld. is (7) (rv)k 1/ (k)re rv.g. The parameter v is held to be gamma distributed. for four hypothetical individuals. In contrast. Note that is the gamma function. 2002). its gamma distribution having the parameters r and k. Townsend & Ashby. where k is an integer. The parameter r tenably expresses stress-induced impairment in processing speed. Wenger & Townsend. is examined under “Construct Validity of Distribution Parameters: Analytic Considerations” below. 2004). (k) is (k 1)!. which is a continuous-variable analogue to the factorial. Thus.

The common distribution of v is presented in Figure 4. Neufeld et al. Townsend. The estimates of r and k. respectively.03735 and 2.g. incorrect responses were excluded. that of k varies with diagnostic group and encoding-load condition. 1990. its distribution parameters remain unaltered. obtained essentially through fitting theoretical to empirical moments (e. this distribution is shifted upward in a similar fashion. Its mean—the modeled average value of k —is moved to the right by a constant amount for each source. In this sense. with elevation in the encoding load. these factors represent exogenous and endogenous sources of encoding-load increase. were 0. This perturbation of the mixing distribution for k therefore signifies an additive . and with paranoid-schizophrenia diagnostic status. as is its variance. Specifically.. 1993).5044. 1984). The single parameter of this distribution is designated m. Gamma distribution of v.. Note that. Unlike the distribution of v. respectively.. spared. 2000). as were positive trials (Neufeld & Gardner. in order to circumvent certain interpretative hurdles apropos the present purposes.28 modeling complex systems Figure 4. The distribution of k is considered to be Poisson (Evans et al.

under the low-encoding-load condition. In keeping with this stance. as n approaches the limiting value of infinity and np approaches the limiting value of m. m) = mk /k !e m . (8) The estimated value of m was . The mean and variance of the Poisson distribution. depending on whether exogenous (encoding condition) and/or endogenous (paranoid schizophrenia) related increases in propensity toward greater subprocesses are present. The base value is increased by the amount h for each group under the higher-encoding-load. the Poisson and binomial distributions converge. verbal-probe format and by the amount g with paranoidschizophrenia diagnostic status under both higher and lower encoding demands. The term n is tantamount to a dormant “subprocess pool. is denoted m . to which is added h and/or g. The present interpretation designates p as the source of variation in m (Neufeld et al. are both m. Clearly. Reference to the binomial distribution’s counterpart to m allows a more concrete understanding of this parameter. This counterpart. approximates m as follows. for controls under low-encoding demands. that of h was 19. It is the value m = np that governs the probability of a specific value of k . The Poisson “intensity parameter. The value of n is m + h + g + x. in turn. The distribution’s base value for the nonpatient controls. The value of m suggests that the controls required hardly any processing to encode the rudimentary limned drawings of the probe items. is taken to indicate tendency toward encoding-subprocess recruitment on engagement of the probe item.0. The term p. then.00001.” m. The general structure of the mixture-distribution model is sum- . That of p approximates m/n. m) for the present values of m are plotted in Figure 5. where the base value of m. 2002). This probability is Pr(k . where x is merely some very large number. m increases with each term.7390.. Plots (smoothed) of the Pr(k . is m .29 Formal Clinical Cognitive Science increase in encoding subprocesses attributable to each factor of the diagnostic groups by encoding-load layout. and that of g was 70.” and p may be viewed as a “disinhibition” term. portrays the probability of a latent subprocess being marshaled to the encoding operation. The quantity n indicates the pool of encoding subprocesses potentially transacting the encoding process for any participant. denoted np.

are brought into play. k }. on which v and k bear in the individual case (exemplified in Figure 3 above). a sample was extracted from . estimation of these parameters stands to become substantially more precise (computational formulas are presented in Neufeld et al. marized in Figure 6. 2002). Poisson distribution of k (smoothed) for differing values of m. estimation of the base distribution’s parameters for an individual can. is the mixture’s “base distribution.30 modeling complex systems Figure 5. To illustrate the point. The Erlang distribution of t. In each instance. The gamma and Poisson mixing distributions of v and k . all that is needed is a modest sample of the participant’s own process latencies {*}. be substantially sharpened. individualized parameter estimation With appropriate application of [w( |{*}. under each condition of encoding load.” With this structure in hand.. representative performance samples from each diagnostic group. respectively. correspond to those displayed in Figures 4 and 5 above. in principle. g)] to the case of {v.

whereas that of the prior is 8. A graphic example of tailoring a parameter distribution to the individual is exemplified for a tentative case. . Similar observations attend the individualized estimation of v. Although unlikely according to Table 3. relative to the prior distribution aligned with the paranoid schizophrenia/low encoding load combination. Adjusting these latencies to highlight the encoding process has been described in Neufeld et al. In fact. Design of mixture model. one of the 10 participants in each of the four factorial combinations (2 diagnostic groups 2 encoding-load conditions). If so. consider the performance sample supplied by the control participant under the low-encoding condition {1. (2002).8110. The sets of trimmed values are presented near the top of Table 2. . Each sample size was four. The Bayesian posterior distribution of k (smoothed) is presented in Figure 7. under negative trials. indicated by the diminished width of the posterior distribution. under the low encoding load. this sample still could have been produced by a member of the paranoid-schizophrenia group. Note as well the increased precision of estimation. .037. alongside the prior distribution corresponding to m = 70. the standard deviation of this posterior distribution of k is 1. the customized distribution of k would have moved markedly to the left.}.31 Formal Clinical Cognitive Science Figure 6. This relocation reflects the joint probability of comparatively lower latencies making up this encoding-process performance sample.00001. one for each memory-set size. Specifically.36. .

9.03735. m = 70. m = 89. 3.29397(10-40) .73901.5044.80080(10-12) .238} Paranoid schizophrenia: High encoding loada Low encoding loadb .037.5044.498} . 7. k = 2. .177.00002 .371. 3.390} Paranoid Schizophrenia Low Encoding Load {2. 0.03735.00737 . c Parameters of prior distributions: r = .554.00684 . 1.204.70547(10-9) .999776(10-11) Control: High encoding loadc Low encoding loadd .32930(10-7) Pr{*}→ a b Parameters of prior distributions: r = .00001.03735. 2. 6.73901.03735. m = . 4. k = 2.54107(10-11) .75783(10-13) . Parameters of prior distributions: r = .00. k = 2.150. k = 2. m = 19. Unconditional Joint Densities of Encoding-Latency Performance Samples under Alternate Bayesian Priors Representative Individual-Participant-Latency Sets Control High Encoding Load {1.65324(10-12) Low Encoding Load {1.00001.476.Table 2.137. 2. d Parameters of prior distributions: r = .23614(10-32) .207} . 2.411.00083 .17176(10-12) .191.5044.1486(10-12) Classification High Encoding Load {3.5044. 0.184.

k = 2.5044.9511 .03735. k = 2.554.73901.188326 .371. 1.00007 . . 9.23972(10-11) Classificationa High Encoding Load {3. b Parameters of prior distributions: r = . Posterior Probability of Group Membership.99987 Pr{g} → 0 a Base rates for group membership = .05 for paranoid schizophrenia/high encoding load. 2. 2. 7.027 .411.5044.20 for paranoid schizophrenia/low encoding load.03735.037. m = 70.204.200186(10-12) Low Encoding Load {1. Given Representative Individual Latency Sets Representative Individual-Participant-Latency Sets Control High Encoding Load {1. 3.022 .03735. e Parameters of prior distributions: r = . .730536(10-29) . .184.00001.498} . d Parameters of prior distributions: r = . and . m = .Table 3.191.137. .5 for nonpatient controls/high encoding load. m = 89. 4.25 for nonpatient controls/low encoding load.476. m = 19.00001.81167 Paranoid Schizophrenia Low Encoding Load {2.00.03735.5044.5044. 2.73901. c Parameters of prior distributions: r = .207} .238} Paranoid schizophrenia: High encoding loadb Low encoding loadc . k = 2.52528(10-6) .23473(10-20) .27083(10-7) .390} . k = 2.00012 Control: High encoding loadd Low encoding loade . .99993 .177. 3.150. 6.

037.00001). essentially. .00001 is. . 6. in Figure 10. Figures 8 and 9 show the posterior latency distributions constrained by {*}.554. . . m = 70. Observe that. the distribution for m = 0. g). Posterior distribution of k . and prior distribution of k . The selected performance sample is that for the paranoid-schizophrenia low-encoding condition {2. not visible because the stochastic mechanisms involved result in its hugging the y axis. g) with those of the alternate mixture distributions f(t|g). Pr(k |{1. and Figures 10 and 11 give their unconstrained mixture-distribution counterparts. The customizing effects of a performance sample can be appreciated by contrasting f(t|{*}. m = 70.}. . posterior probability of group membership I now consider the case where a performance sample has been obtained from an individual and is to be used to evaluate that per- . sans reference to the performance sample {*}.39}.00001). Pr(k .191.34 modeling complex systems Figure 7.177. individualized latency distributions These slenderized posterior parameter-mixing distributions can be used to create posterior latency distributions f(t|{*}.

g). 2. 3. ostensibly. monitored so that base rates for a testee’s group membership. In effect. . Pr(g). . transitions in prevalence of the candidate groups are. . what is being asked is: In Bayesian terms.411. This scenario may apply to treatment evaluation over time: Is the individual being edged closer to normal functioning. .35 Formal Clinical Cognitive Science Figure 8.498}. are available. son’s “proximity” to variously symptomatic groups. It is assumed as well that the mixing distributions for these . in contrast to a member of a symptomatic or more severely symptomatic group. g = 1. Posterior latency distributions f(t|{2. Allowing that the performance is sampled at selected points in time. as assessed by methods rooted in contemporary cognitive science? Estimates are available in terms of the posterior probability of a representative of the target group—a group with less or no symptom severity—generating the sample.184. what are the compatibilities of the obtained process-latency specimen with each group’s pair of prior parameter distributions? The expression of Equation (5) above enclosed in elongated braces { } is now brought forth. In other words. G.177. 2. it is assumed that the representativeness of each symptom group along the way is known. 2.

therefore.411. Posterior latency distributions f(t|{2.36 modeling complex systems Figure 9. . Finally. The classificatory system is. the assumption of initial-prior stability is invoked. is a finite. available (Karabatsos. in fact. not available for the designated time points. This feature underwrites the intended description of the presenting individual. if they do not. 2. Instead of monitoring an individual’s progress by profiling his or her likelihoods of belonging to symptom groups of known base rates. The adopted approach entails an “odds-ratio evaluation strategy”—an odds-ratio evaluation strategy for assessing the appropriateness for the current performance sample {*} of the G respective priors. the primary issue may entail the base rates themselves. it is possible that diagnostic-group base rates are. regarded as comprehensive. indeed. the present computations imply that the G classifications are mutually exclusive and exhaustive. 2006). 2. groups. which. established at the outset. g). nevertheless.177. are. that new priors have been formulated for the ensuing points of performance sampling.184. other Bayesian strategies. discrete posterior-probability mixture. with unique statistical merits.498}. It is deemed to facilitate the flow of the present developments. remain valid or. In the present application. 3. In contrast to the scenario painted above.

This second case is taken up in its turn below. Here. to paranoidschizophrenia status/low encoding demands. to finally nonpatient status/low encoding demands. the focal problem may be one of evaluating the collective effectiveness of treatment by ascertaining movement in size of symptom-group membership. be broached through the elongated-braced expression { } of Equation (5) above. through nonpatient status/high encoding demands. the present classifications formed by factorially combining diagnostic status and encoding-load condition are appropriated. Apropos the encoding manipulation as a surrogate for symptom . Mixture-model latency distribution f(t|g). These combinations again provide the Bayesian mixing-distribution priors of k and v as well as the representative performance samples listed in Table 2 above. To expedite the exposition. We conveniently allow for a progressive decrease in symptomatology with movement across groups formed from combinations of paranoid-schizophrenia status/high encoding demands. once more. A question of this nature can.37 Formal Clinical Cognitive Science Figure 10.

to be sure. at a time of interest specified by the treatment regimen. 1993). severity. as arrayed in the first row of data in Table 2 above. To proceed. We posit a scenario where encoding-latency samples have been obtained from four participants.38 modeling complex systems Figure 11. The values have been obtained. as a stand-in for the elevated number that may well go along with increased symptomatology (Nicholson & Neufeld. say. It nevertheless conveys an escalation in recruited subprocesses. In all events. We wish to estimate probabilities of symptom-group membership for each participant. given the obtained performance data. responding to increased encoding demands with an increased number of encoding stages is hardly indicative of disorder. Mixture-model latency distribution f(t|g). this pretense for expedience is inconsequential to the ensuing line of argument and in no way detracts from its analytic points. it is necessary to know the values of the uncondi- . These groupwise posterior probabilities are endowed by the prior parameter-mixing distributions aligned with the candidate groups.

and the desired information consisted of the alternate likelihoods of the individual’s membership. given group g.39 Formal Clinical Cognitive Science tional joint probability density of each latency sample with respect to each group ucd({*}|g). The numbers assumed by ucd({*}|g) will change with g because the parameter-mixing distributions are group specific. clearly the far-right column presents the most encouraging profile and the second-from-the-left column the least so. are 0. This result is attributable to both ucd({*}|g) and Pr(g) being highest in connection with the nonpatient controls/high encoding load amalgam. 0. The sole exception occurs for the sample from the paranoid schizophrenia/high encoding load combination. Allowing that all individuals were members of the same group at the outset. and 0. all values of v and k considered. as conveyed by the Bayesian expansion of the present mixture model. enumerated as the Bayesian posterior probability of group membership Pr(g|{*}). in turn. such multivariate profiles can be charted across successive times of interest. mandates that the clinical scientist/practitioner is blind as to the actual origins of the latency samples. The ucd({*}|g) is the joint density of the obtained sample of latencies. Base rates assigned to the respective groups Pr(g). proceeding from the left to the right of Table 2 (from paranoid schizophrenia/high encoding load to nonpatient controls/low encoding load).05. Each individual is profiled in terms of distance from the respective groups. be statistically mediated to psychometric and other predictors. As intimated above. The point of the exercise. the highest values tend to be aligned with the groups from which the samples originated. Table entries can. provide pivotal information regarding participant status. 0. The posterior probabilities of group membership are entered in Table 3 above. however. then. base rates for the variously symptomatic groups were known. Not surprisingly. may. maximum-likelihood estimation of symptom-group base rates In the first case presented above. given his . at least as quantified by focal cognitive functioning.25.20. The resulting individual configurations of treatment (non)response. The values of ujd({*}|g) for the current performance samples are presented in Table 2 above.5.

For reasons of mathematical tractability. deleting the third row and third column of ujd({*}|g) values in the table. 1998). . The strategy for estimation is straightforward enough. it seems desirable to bring to bear our available bank of information on cognitive functioning. . This joint probability comes down to the product of the posterior probabilities of those categorizations. 2. The instantiation of this application once again makes use of entries in Table 2 above..g. . given the latency samples {*} linked to the categorized individuals. The procedure constitutes a somewhat unusual but valid combination of maximum-likelihood estimation of a meta-Bayesian parameter (see Appendix A). However. It may be argued that the desired course of action is simply to directly examine the array of diagnostic assignments that have been made at the selected times. however. . It may be prohibitive. e. In importing this information. the present case differs from the first case in that the base rates are now unknown. . only three of the four data sets listed in this table are considered. These densities of enclaves with differing symptomatology may be useful in assessing treatment efficacy across the designated times of interest. However. not only for a modest sample of individuals whose classifications are updated. The likelihood function is made up of the joint probability of the updated participant categorizations. 2. . . we may settle. The set corresponding to the nonpatient controls/high encoding load is excluded.40 modeling complex systems or her specimen of process latencies. Luke & Homan. we entertain a sample of individuals who have been allocated to one or another group on the basis of presenting symptomatology at the selected time of interest. and in the interests of illustrative neatness. be tacked onto sessions hosting symptom assessment at the successive times of interest. Batchelder. to diagnostically interview (or assess by some other means. It consists of constructing a likelihood function for the updated participant assignments and then maximizing the function with respect to Pr(g). . If so. The required sampling may. G. such as psychometric) a large enough sample to ensure reasonable fidelity of estimates (see. and it is this information that we seek. but also for a modest performance sample from each. The likelihood function and related computational details of maximum-likelihood estimation of Pr(g). feasibly. G. in . 1998. are presented in Appendix B. g = 1. g = 1. In the present case.

g. Nevertheless. Furthermore. base-rate estimates of the three groups. with an estimated relative frequency of nearly 0. matlab Optimization Toolbox). rather than three.. moreover. are 0. proceeding from left to right. one may resort to a numerical-search algorithm (e.41 Formal Clinical Cognitive Science practice.25.2399. 0. and those who are moderately symptomatic are sparsest. as described momentarily. and 0.0852. three individuals have been classified. that this exercise is nothing more than instructive. 0. given the associated performance samples {*}—as was done in the first case presented above. In larger-scale applications. sufficient to allow each individual to have been assigned to the group adjacent to the displayed performance sample. In other words. It is. as the base-rate probabilities sum to 1. however. see Appendix B). regarding the involvement of ujd({*}|g) and Pr(g) in arriving at the values of Pr(g|{*}). such a restriction is not necessary. The maximum-likelihood estimates of Pr(g) were obtained by solving for each value using closed-form solutions to two simultaneous equations (two. and 1.119736. in turn. the least symptomatic category is reasonably well represented. The most symptomatic category thus remains in the majority at this time of evaluation. The latter. become 0.0. The first amount resembles the second because the comparatively low ujd({*}|g) connected with the paranoid schizophrenia/high encoding load group for the individual so assigned is offset by that group’s . however. again proceeding from left to right in Table 2. to estimate the respective values of Pr(g). rather than pristine closed-form solutions. With these arrangements in place.6748. to evaluate the posterior probabilities of membership in the respective groups. even with the assistance of a computer-algebra program. It was found that increasing G beyond 3 became infeasible.00. where G > 3 and/ or the number of diagnostically updated participants is somewhat larger. the three sets of performance samples in the provisionally reduced Table 2 now represent those putatively obtained at the period of updated individual classification. Note. For the present purposes. the classified individuals have putatively supplied the very process-latency samples atop the columns of their respective groups. It is tempting to insert these base rates into the expression enclosed by { } in Equation (5) above and.119786. one to each group.

by and large. It is evident from these developments that tenability of the proposed analytic strategies makes for new and potentially useful avenues for formulating clinically significant information. and transference of the previous meaning of g in the presently estimated Pr(g). to practice effects. symptom criteria for designations need. Potential redress may require the substitution of different task items.00 for the nonpatient controls/low encoding load assignment evinces the fact that the ujd({*}|g) associated with this specific group exceeds the next highest value by a factor of .2766325061(1021). . as conveyed by cognitive-functioning aspects of symptomatology. Dislodgement of the established priors from their host symptom groups may occur because of a shift in the nature of task performance. The value of 1. This assumption is intrinsically bound up with stability of conditions under which the current sample of task performance is obtained. assumptional constraints are relatively salient. continuing tenability of the distributions themselves. and m. transitions in symptomatology eventuate in one or the other of the G classes addressed at the outset. Furthermore. k. Available as well are broader inferences about treatment efficacy. the computations once again imply that the G classifications are mutually exclusive and exhaustive. The assumption is also yoked to continuation of the initial classificatory criteria leading to participant allocation. the developments remain entrenched in a formal cognitive-process model. qualifying assumptions The above estimates of Pr(g) assume that the parameter-mixing distributions initially prescribed for the symptom-based classifications extend to the time of estimation. Finally. To guard steadfastness of the G categories. owing. analogous to using parallel forms of psychometric tests. for instance.42 modeling complex systems having the highest value for Pr(g). to be consistent. In other words. Finally. It is additionally assumed that differential dropout across the G groups and other sources of distortion in their representativeness are prevented. Available are inferences about the status of specific patients. These requirements come down to retention of their earlier values by parameters of the mixing distributions r.

. In moving from right to left in Figure 12.. & Penny. the kernel of mri measurement. on the one hand. Quantitative cognitive science is.. Rather. & Heeger. Friston.. They can be integrated into posterior cumulative probability functions F(t|{*}) whose values for the respective curves in Figure 9 are shown in Figure 12. thus. Consider the stochastic dynamic trajectories presented in Figure 9 above. on the other.. e. meanwhile. 1996). 2004. poised to complement these developments. Reiman. Moreover. 2004). Lane. Harrison. the time window for measurement does as well.43 Formal Clinical Cognitive Science when events are processes: potential contributions to fmri Recent years have seen an explosion of advances in fmri technology (e. for example. see. mapping deviations in neuronal-activation patterns corresponding to symptom-significant mentation (e. & D’Esposito. however. imputation of cognitive functions now stands to be anchored in a verisimilitudinous performance model. in principle. Van Petten. similar considerations attend other measures of neurophysiological functioning. Similarly impressive advances in the statistical treatment of fmri signals have taken place.g. as follows. Glover. be consulted to synchronize times of interest in bold-response measurement. although not emphasized here. Lanius et al. The amounts displayed in Figure 12 may.g. Engel. set in motion (cf. The interplay of such considerations now has a formal backdrop against which to estimate times of interest. & Bendettini.g. complementing regions of interest. Boynton. in calibrating mri measurement.g. has been put forth (e. Considerable dynamic modeling of the blood-oxygen-level-dependent (bold) response. Note that. Fokas & Marinakis. or “meg”. with probability of target-process occurrence. Song. 2004. avoiding the often-encoun- . Frackowiak et al. 2000). the ultimate subject of interest is not that of external-stimulus events per se. Huettel. it is the cognitive processes that are instigated by these events and that parallel the activation profiles that are.. 1998. 2006). The areas under these curves are proportional to relative frequencies of process completion. Aguirre.. the probability of process expiry decreases. Boksman et al. Zarahn. & McCarthy. Clinical cognitive scientists have capitalized on these successes. 2005. Implicated are stochastic dynamic models. such as magnetoencephalography. 2003. in documenting neuronal-activation concomitants.

even the simplest cognitive task recruits multiple stochastic processes (Smith.44 modeling complex systems Figure 12. along with associated latency trajectories. 1995. Separability of the respective processes then comes into play. 1984). The trajectories in these figures have been constructed for an individual performance sample/prior distribution combination. be extended to accommodate subsets of individuals whose sets of performance samples are similar and to whom the same priors apply. The resulting amalgams should render subsets with relatively homogeneous parameter profiles (see Figure 7 above). Amalgamation of parametrically homogeneous participants guards against an amalgam of systematic individual differences in the expression of cognitive functions under study. They could. Cumulative probability distributions corresponding to probability densities of Figure 12 above. Townsend. multiple processes Typically. tered circular reference to the very neurocircuitry whose functional significance is under study. however. Aggregation across participants is often necessary to attenuate noise of mri signals or for selected statistical treatment. The processes may occur .

As applied to stress negotiation. cognitive debility and stress susceptibility The adaptive significance of cognitive functions may be appreciated by deciphering how processing operations interface with environmental exigencies. Wenger & Townsend. This . in the form of information-processing demands. harboring implications for prediction and intervention. (b) stress-reducing benefits. in the form of minimization of threat (e. It is stated here without elaboration that the greatest likelihood of the second process surviving. delineating information-processing requirements for selecting advantageous. is defined by max([1 F2(t)] [1 F1(t)]) = max(F1(t) F2(t)). 1999). 2000). and (c) sources of vulnerability stemming from information-processing frailties. 1973. Their finishing times are. ultimately. is labeled decisional control (Averill. cf.45 Formal Clinical Cognitive Science in parallel. occurs at the modal value of f1(t). meaning that they commence simultaneously. A prominent form of coping with stress. alluded to above.g. The synthesis should. physical danger/discomfort. Let the probabilitydensity function of the initial one be f1(t).. according to their respective stochastic distributions (Townsend & Nozawa. Then the greatest likelihood that it has been accomplished. with the first having elapsed. This endeavor is potentiated through the application of mathematical combinatorics. Consideration of cognitive psychopathology invites the following extension: examination of quantitatively dissected cognitive debilities in relation to likewise quantified environmental pressures on which they bear. while the second has not yet elapsed. or max(f1(t)). Let the probability-distribution function for one of a pair of processes be F1(t) and that for the second be F2(t). 1995. Products of the analysis include the specification of (a) cognitive costs. and auxiliary procedures. to environmental prototypes. instead. Lees & Neufeld. social sanction). the above approach implies. threat-minimizing options when stressor situations are multifaceted. among other analyses. The processes may. provide a more precise picture of compromised negotiation of environmental demands. nevertheless. proceed serially. Measurement epochs addressed to the second process would be synchronized accordingly. staggered.

In line with this definition. 1999a. A factor potentially exacerbating the above risk is the toll taken on encoding speed by stress itself (Neufeld & McCarty. Townsend. construct validity for the purported psychological meaning of scores from a test. 1992). includes all evidence bearing on the measure. Neufeld. and response processes. 1988. including option-cue encoding. Recall that a stress-induced movement of the distribution of v to the left is identified with an increase in its mixing-distribution parameter r (Figure 4 above. visual and memory search.. 1999b). Apropos the formal depiction of encoding. 75). 1955. through the application of mathematical combinatorics and related methods to the essential structure of this form of coping. an inventory. Neufeld. 1983. Embretson. & Jetté. including those comprising retarded completion of encoding processes (Neufeld. discloses it to be a prototypical expression of selection and choice (Morrison et al. for a mathematical critique of selected data-analytic strategies for adducing psychometric construct validity. are its demands on cognition. 1959. this effect translates into a general reduction in the parameter v of the base distribution of latencies (Figure 3 above). 1994. Cronbach & Meehl. 2000). Neufeld & Carter. Neufeld.46 modeling complex systems type of coping basically takes the form of situating oneself in a multifaceted stressing context so as to engage the situation’s most innocuous option. see Neufeld & Gardner. Unveiled. Construct Validity of Distribution Parameters: Analytic Considerations Construct validity is a form of psychometric validity entailing a corpus of evidence brought to bear on the espoused interpretation of a measure. An overriding principle governing evidence for a measure’s . 1991. Haynes and O’Brien (2000) have summarily stated that it “comprises the evidence and rationales indicating the degree to which data from an assessment instrument measures the targeted construct. in press). Dissection of decisional control. Nicholson & Neufeld. 1994. or some other mode of assessment emphasizes the measurement tool’s portfolio of empirical correlates (Campbell & Fiske. in turn. Unveiled alongside is the undermining potential of debilities in the required cognitive functions. 1990). and includes all types of validity” (p.

and m. In the present treatment. The light thrown on the nature of parameters by the mathematical properties they possess arguably embodies a new form of construct validity (cf.47 Formal Clinical Cognitive Science construct validity is that the measure act in accordance with theory (Wiggins. have their own parameters. but. Within the present analytic context. Construct validity for their imputed meaning. In contrast. arguably from a commensurate level of analysis. Its binomial-based dissection cast a certain light on the potential composition of this parameter (cf. 1973. Distributions of the above parameters. in the case of a gammadistributed v. as follows. The behavioral significance of these properties hinges on the concept of maintenance of cognitive performance. Because they pertain to the mixing distributions for v and k . of performance requires that the value of k exceed a certain threshold. k. Cronbach & Meehl. r and k. increasing r and m prolongs performance times. called for is support for the constructs that they purportedly express. Finally. in turn. The parameter k is endowed with additional mathematical properties that effectively illustrate the type of construct validity imbued by analytic derivations. 406). in the case of a Poisson-distributed k . 1983). Its increase moves the stage-dispatching capacity parameter v to the left. and its increase shifts the distribution of v to the right. The parameter r was identified with stress effects on processing speed. actual failure of the processing system is not precipitated when a finite range of values is violated. in this case. “measures” consist of model parameters. p. Recall that the parameter m was aligned with “encoding load” and conveyed exogenous (task demands) and endogenous (paranoid schizophrenia) sources of increase. emanates from their mathematical properties. over and against its breakdown. 1955).” Embretson. the parameter k ostensibly corresponds to process-transaction competence. as opposed to that of k. as translated into a corresponding distribution of latencies. The parameters of the appropriated process-latency model included v and k . Accordingly. integrity. Their interpretations stemmed from the assumed structure of the encoding process. versus collapsing. “construct representation. the parameters r. and m are a step removed from the empirical touchstones of an individual’s latency data. . in turn. paralleling a critical level of performer skill. which are part and parcel of the stochastic dynamic models in which they participate.

As n increases. upward. v. its exponentiation of t correspondingly magnifies the consequences of elongated latencies for the computed moment. bypassing issues of sampling and measurement weaknesses. or essentially outstanding. 4) are somewhat familiar. resulting in E(Tn) being bounded to an even greater extent by finite values. the threshold value of k . as fallibly estimated from empirical data. The first four moments of a stochastic distribution (n = 1. and kurtosis. In effect. which. see the first section of Appendix C). At the same time. be driven to whatever amount is prescribed by the governing theoretical analysis. strengthening process-performance skill should influence the distribution of t downward. 2. including extremes. to economics. it should enable E(Tn) to withstand heightened values of n before transmogrifying to infinity. noncompletions). In model evaluation. to astrophysics. concerns consequences of variation in the processing system’s constituent parameters. where n is the order of the computed moment. variance. however. Consequently. The present theoretical deliberations. The order of the moment n can. The above integral indicates how values of t and the order of the moment n act in concert to affect E(Tn). including its transition to infinity (Appendix C). in this case. Observe that this approach to discerning the nature of a theoretical system’s parameters has much precedent in fields ranging from mathematical ecology. E(Tn) = . the integral 00 0 f(t)t n dt = E(T ) n does not converge (for further technical exposition. through its shifting the distribution of the process-completion rates. empirical moments beyond the second are seldom sampled because of instability. operate in the realm of population properties.48 modeling complex systems The current formal expression of performance breakdown takes the form of certain latency-distribution moments acquiring infinite values: more succinctly. the result signifies that the population of trials occurring under the prevailing value of k includes a critical mass of extremely long. Apropos the model parameter k. completions (in effect. 3. comprising or being involved in its mean. That is to say. therefore. skewness. such theoretical explorations retain a bona fide bearing on the interpretation of the parameter values. k acts to diminish densities f(t) for higher latencies. Indeed. Behaviorally.

Their largely verbal descriptions are presented below.g. Ross. 2004). Evans et al. Almost any empirical process-latency distribution can be well approximated by creating a finite probability mixture of one or more base distributions (defensibly expressing the relative percentage of trials best characterized by the respective base distributions. n. Johnson et al. although theoretical.g. The descriptions of these base distributions are available in various sources (e. with technical justifications allocated to the several sections of Appendix C. 1995. 1994. Insufficient k implies that appreciable densities of v.. Finally. Townsend & Fific. presentation here is restricted to theoretical analysis of the espoused substantive significance of k. for diverse process-latency base distributions. It transcends process-latency base distributions with diverse characteristics. These variations on analytic support are dealt with in succession. its interpretation is enhanced by pertinent Bayesian extensions. The above support for the construct validity of k as a capacityaffecting competence parameter is robust... e. Importantly. results of the present analyses regarding moment finiteness are preserved with such a mixture. . generating the critical magnitudes of f(t) at acutely high ranges. f(v).49 Formal Clinical Cognitive Science issuing in bounded moments is defined precisely in terms of n (see both below and Appendix C).. the avowed index of stress effects on processing capacity. is malleable with respect to empirical application. as assayed from multiple analytic standpoints. its effects are modified in expected ways by formally defined efficiency of capacity application by the operative base distribution. r. notably. critical values of k : convergence from multiple base distributions Finite moments of mixture-model latencies require that k exceed the order of the moment. its effects align with those of related parameters from similar mixing distributions. 1996). Note that the present treatment. and it is evinced in familiar base-distribution statistical properties. The current selection of base distributions is prominent in the stochastic-modeling literature and expresses alternate structures of a processing system. are perilously close to v = 0. its effects are distinguishable from those of parameters to which a different meaning is ascribed. 2000.

v constant across stages 00 (– 1) d MT ( )/d | where MT ( ) = 0 E(T n|k ) = 00. For example. hence E(T n) = 00. also by results for the general gamma distribution below E(T) = Var (T) = continued . (k – 1) r2m(2(k – 1) + m) (k – 1)2(k – 2) E(T n|k ) = 00.Table 4. MT ( ) = i v . k = 1. each with constant rate v n n =0 n rn (k + n – 1)! (k – n) (k – 1)! (k) . by results for the Erlang distribution above 00 (1 – e –m) = 00 -e-M f(v)[(em(M i ( T -M – 1) )-e |v]dv. Process-Latency Base Distributions and Their Moments with a Gamma-Distributed “Capacity” (Intensity) Parameter. (v + ) f(v) = (rv) k – 1rerv . mr . ] For the exponential distribution. v nth Moment k>n t= 00 Base Distribution k = n1 r k (k + n – 1)! ln(r + t) (k)(k – 1)! = 00 t= 0 a Distinguishing Features Comments Erlang k stages fixed. (k) and m is the parameter of the Poisson distribution of k .0 Compound Poisson k stages random.

the rate thus in effect for stage i is civ. and k 00 . (k) where > 0 00 .General gamma k i=1 Stagewise rate transition Ck cin!r –n (k – n) . . . 2. rate is continuous functions of time rn (n/ + 1) (k – n) . . ci > 0. i = 1. . where ci = (k – (i+1)). 00 = 00 i+1 () ir n ln(r + it)] (k)i Weibull Single stage. . as (k – n) = (0) a Expression of moment infinity as (0) or ln(00) merely rests on integration with respect to t first or v first. k = i=1 (k () r n! (k)i– n) (– 1) . as (k – n) = (0) Cik = j=1 j=i – ( ) k 1– ci –1 cj (– 1) i+1 n+1 k (– 1) i+1 n k i i=1 t= 0 k n k i k t= 00 Pure death general gamma process with linear death rate i=1 Rate for the ith stage is civ. respectively. (k) where ci is a scalar of v for the ith stage.

52 modeling complex systems The above relation between k and n occurs for the Erlang distribution (Figure 3 above). the latter distribution makes up the dual-parameter mixture model depicted in Figure 6 above. 1966. e. It extends to those with rate transitions across stages. Wenger & Townsend. 102. is constant for each stage in both the Erlang and the compound Poisson distributions.. see. if the nth-order moment is unavailable (infinite). known as the general gamma distribution (McGill & Gibbon. The selection of base distributions. so are moments of order exceeding n (for k as an integer. 2000). with a fixed-stage parameter k . A notable version of the general gamma is a pure death system with a linear death rate. This relation includes as well the extension to k as a Poisson-distributed random parameter. 1966. whereby moments about the origin become meaningless. 80–95). and their distinguishing properties. infinite. In the present context of base distributions whose parameter v is gamma distributed. v. 1996).g. 1994). to which the relation between k and n applies is summarized in Table 4 (further details are briefly presented in the second section of Appendix C. as in the case of n = k.g. p. The rate parameter.0. where k = 1. Also listed are the moments for k > n and k = n. the compound Poisson distribution is considered simply to be a base distribution whose Erlang stage parameter. Townsend & Ashby. This model has figured prominently in selected theorizing about the architecture of human information-processing mechanisms (see.. otherwise known as the compound Poisson distribution (e. the function (k n) appearing in the moment’s expression takes on a succession of negative. The above relation between k and n. 1990. Ross. Together with the mixture on v. Another well-known and frequently used distribution to which the present relation applies is one with a single stage of process completion but with a rate that changes continuously with time (Weibull distribution). additional elaboration being available in Neufeld. 1983. pp. transcends base distributions with a static rate parameter. k . e. Feller. where k is not an integer and k < n. which is a parametric version of an independent parallel-processing model with unlimited capacity (see Townsend. and positive values as k n decreases from 0. however. is Poisson premixed. Apropos the latter.g. At the heart of the matter is the attestation to k as a process- . note that. 1965). Harris. A specific case is the single-stage exponential distribution..

2001). can be immediately extracted from derivation (A. The fixed-parameter status of k streamlines the present exploration without curtailing the generality of available inferences. A value of k = n signifies that a finite E(Tn|{*}. . of course. is gamma mixed.53 Formal Clinical Cognitive Science transacting competence parameter. Insight is afforded according to the give-and-take between k and both empirical performance samples and collateral parameters. k) = if and only if Nk > n k. . The excursion into the psychometric analogy is followed once more by observations on the viability of the proposed interpretation of k. . k ) = n [r + ti ]n (Nk + k – n)(k + n – 1)! i=1 N (Nk + k)(k – 1)! . Isolating first on the term (Nk + k n). (2002). k ) is avoided if Nk + k exceeds n. i = 1. The current Bayesian extension once more incorporates a set of process latencies {t1. . . . . . it is possible to have k = n without E(Tn|{*}. Robustness of this nature bears a certain resemblance to empirical convergent validity. N. that the present ti < . Findings tenably apply to classes of individuals sufficiently homogeneous regarding k that this simplification is a reasonable approximation of the quantitative goings-on. A poignant exemplar involves the Erlangk . tN} = {*}. In this case. k ) does not characterize all members of the class of individuals to whom the current value of k applies. allowing. (9) This expression affords a certain perspective on the posited interpretation of k from the standpoint of rudimentary classic psychometric theory surrounding performance-sample size or test length. It appears as follows: E(T |{*}.v distribution whose rate parameter. which entails agreement among divergent methods of measuring the same trait (Haynes. according to invariance of its infinity-avoiding values across base distributions of heterogeneous makeup. bayesian extension Additional light can be thrown on the proffered interpretation of k by dissecting its properties within a Bayesian context. The required moment expression.10) of Neufeld et al. v. so conditionalized on k . an infinite value of n E(T |{*}. t2. and the nature of latency-distribution moments again occupies center stage. . 2.

N corresponds to the number of independent test administrations. . in principle. Bringing the above formulation to bear on convergent support for k as a process-transaction competence parameter. (Nk + k – n) E(Tn|{*}. corresponds to the number of parallel items composing a psychometric test. Inspection of Equation (9) reveals further interplay between k and N and between k and k . a value of Nk such that Nk + k = n fails to nullify the viability of the partial expectation E(Tn|{*}. Conversely. k ) is assured for the testee at hand. k ) < ). the following inference from Equation (9) is apparent.0. other things remaining constant. The psychometric analogy seemingly throws light on this conclusion as follows. be partitioned according to a probability mixture of two partial expectations. therefore. 0 < 1. combined with the extant value of k. Allowing Nk + k > n. consider the following ratio. along with k remaining con- . a finite value of E(Tn|{*}. Nk . = (Nk + k) (Nk + k – 1)(Nk + k – 2) . A performance sample size that is inadequate to ensure stability of finite latency values for the person at hand can be offset by a sufficiently large value of k for the class of individuals to which the present performer belongs.0. in effect renders equal to 0. where is the mixing parameter. The collective item sample is. N ti i=1 N Clearly. and an infinite moment ensues for the present individual. i=1 ti . characterized by a latency moment bounded by finite values. extracted from Equation (9): 1 (Nk + k – n) . With a sufficiently large empirical sample of finite latencies. specifically. . k . k ) can. k > n Nk . mle(v) being Nk . k) = . The argument can be carried further as follows. This result stems from the participant’s performance-based maximum-likelihood estimate of v. Consider that the Bayesian posterior moment E(Tn|{*}. both informative as to the nature of k. k ) = ) and (1 )(E(Tn|{*}. in concert with k. Second. A sufficiently large performance sample of finite process-completion latencies.54 modeling complex systems The specific member at hand is. Within this analogy. 0 1. (E(Tn|{*}. k ) evidently decreases as N and k increase. the stage parameter. nevertheless. mle(v) increases with N.

such eventual elevation can. The decrease is understandable in light of the influence of k on mle(v). if < 1. whose influence can override that of a progressive increase in task load k on E(Tn|{*}. as should be the case. 2002). eventuating in a movement upward of the Bayesian posterior moment. however. effects of the latter ultimately surpass those on mle(v). considering the imputed meaning of k as a capacity-enhancing competence parameter. the nonhomogeneous Poisson distribution specifies a continuous change in processing rate with time. there is a continuous decline in rate. be replaced with continuing decline if k is made increasingly large. f(t|v) = ((vt –1)t) k –1 (k – 1)! vt –1e–(vt )t. Finally. –1 . k ) with elevation in k can be nonmonotone. that. This result again should be observed if k represents a competence parameter. decreasing and then increasing. Likewise. Note. say.6) of Neufeld et al. Releasing these constraints in the study of k. therefore. Apropos the current thrust regarding k. Metaphorically. reflecting. interestingly. however.k-distributed v). the estimated process-completion capacity of the current participant is raised in a parallel fashion with elevation in k (Equation (A. affords certain insights stemming from the relation of its effects to those of . Equation (9) discloses that the interplay between k and k is comparatively complex. the rate continuously increases. as k goes up. As the rate parameter of the Erlang distribution is now replaced with vt 1. It is apparent that the Weibull distribution can be obtained from the nonhomogeneous Poisson by setting k = 1 and taking v to the power of . it signifies the exploitation of capacity dealt to the processing system by the “capacity-resource pool” (gammar. considered above.55 Formal Clinical Cognitive Science stant. but similar to the Weibull distribution (discussed above). The parameter is. Computer computations indicate that changes in E(Tn|{*}. interplay with capacity utilization Unlike the Erlang or general gamma distributions. k ). This dynamic rate is vt 1. a fatigue effect. If > 1. Evidently. 0. designated a capacity-application efficiency parameter. with respect to the number of stages involved. so does the process magnitude.

in press). Wenger & Townsend. . This index is aligned with E(Tn). intensified (for > 1. or the complement of the cumulative probability-distribution function. as set against as a capacity-application efficiency index. and t (formulas underlying Figure 13 are presented in the third section of Appendix C). & Jetté. These associations are apparent in Figure 13. indeed. or. by a factor of 1/ . The parameter k. (k – 1)! (k) rn/ On inspecting the structure of this moment.” as articulated in contemporary cognitive science (Townsend & Ashby. also Figure 12 above and related text). Similar to the escape of E(Tn) from finite values if k = n/ . It does so with respect to both E(Tn) and the allied expression of process-completion potential ln (F (t)). The current interpretation of k implies effects on 1n(F (t)) that should closely resemble or mimic those of . it consists of 1n(F (t)). is afforded by a rigorous and robust index of “process-completion potential. 2000 with respect to the clinical-science implementations. 1 F(t) (see the first section of Appendix C. respectively). performer competence as conveyed by k and capacity-application efficiency as conveyed by reciprocate with respect to moment retention of finite values. which plots the above index of process-completion potential as a function of k and for selected values of r.56 modeling complex systems The coefficient provides for 00 f (t|v)dt = 1. In this way. A complementary angle on the relation of k to capacity utilization. 0 Consequently. E(T n ) = = 00 00 0 0 (rv) k–1 –rv (vt ) k –1vt –1e –vt tn re dt dv (k) (k – 1)! (k – n/ ) (k + n/ ) . . 1983. The parameters k and should also play off one another in maintaining specific levels of 1n(F (t)).0. where F (t) is the process-latency distribution’s survivor function. see Neufeld. Townsend.0 and < 1. E(Tn) breaches finite bounds if = n/k. it is apparent that the burden placed on k for retention of finite values is now alleviated. k . behaves as it should as a capacity-supply competence index. Observe in passing that stress effects on moment magnitude are also modified according to in the term rn/ . then.0.

tenably.k distributed. parameters that take on a similar role in other distributions and affect performance indicators in a similar fashion to k indicate that the derived analytic properties are not idiosyncratic to a gamma mixing distribution.03735. convergence with allied mixingdistribution parameters The above developments surrounding construct validity for k specify that performer competence is a key contributor to capacity resources bearing on process transaction and that this shape parameter of gamma-distributed v is an index of such performer skill. This version of analytic construct validity again is analogous to empirical convergent validity. with properties like those of k. for k varying from 1 to 3 and varying from 1 to 5. k need not have a corner on conveying performer ability. 2001). where a trait evinces parallel values across divergent methods of measurement (e. In other words. Values of process-completion potential 1n(F (t)) for the nonhomogeneous Poisson distribution. indeed. t = 2. r = . k = 18. with rate parameter vt 1 and v gammar. express performer competence. .. Haynes. There is nothing about this formulation that excludes other capacity mixing-distribution parameters from assuming a role like that of k.g. The concept of capacity-infusing competence should transcend the shape parameter k of the gamma distribution as a particular quantitative instantiation.57 Formal Clinical Cognitive Science Figure 13.25. The shape parameter of related mixing distributions could.

whether they are mixing or base distributions. or Gammar. v exponentiated by .k Distributed. As the Weibull now serves as a mixing distribution. v Is Randomly Mixed Erlang. v Is Randomly Mixed Weibull (n/ + 1). then.58 modeling complex systems Table 5. An exhaustive analysis of the two distributions’ effects seems unnecessary. Moments of Weibull and Erlang Process-Latency Base Distributions. described above. is the Weibull distribution. Isolating on the far-right-hand gamma function. In the latter case. Base Distribution Mixing Distribution Weibull. these equivalent thresholds of k and B add to the constellation of analytic properties endorsing the construct validity of k. Convergently. Finally. in each case. (x). ranges from 0 to . The second section of Appendix C shows that. The respective moments are presented in Table 5. the Pascal distribu- . maintaining moments within finite bounds evidently requires identical values of k and B from gamma and Weibull mixing distributions. the gamma and Weibull are nested in the generalized gamma distribution (additionally allowing k of the generalized gamma to assume the status of k as a continuous variable). Selected process-latency base distributions now include the Erlang and the Weibull itself. the intensity (capacity) parameter v alone and. (1 – n/ B)) Vn (k + n) (1 – n/ B) (k – 1)! (n/ + 1) (k – n / ) (k) rn (k + n – 1)! (k – n) (k – 1)! (k) Gamma rn (n/ + 1) (k – n) (k) rn/ A notable example whose distributed variate. constitute the gamma. respectively.k distributed. these parameters are designated V and B. v . Sufficient to make the point are definitions of moment finiteness in relation to k and B. like gamma’s. v Is Randomly Mixed Vn (n/ + 1) (1 – n/B) Vn/ Weibull. similar observations attend the discrete-variable analogue of the mixture involving the Erlang base whose parameter v is gammar. with Capacity Parameters Either WeibullV.or Weibull-mixed random variate. with intensity parameter v and shape parameter . In contrast to the Erlang base distribution. respectively. whose functions are those of continuous time t. then.

which is the modal value of the gammar. where u is gamma distributed. as related to k for the class to which these participants belong (for discussions of empirical estimates of classcomposing homogeneity. necessarily. escape finite bounds if r = 1. The parameter p is. . & Benn. and the variance does so if r = 2. the grand mean taken across all participants is k r/(k 1). which is k /v.k distributed. The analysis affords another glimpse of k as indicative of the performance-ability level required to contain the statistical summaries of performance within finite bounds. the discrete-variable analogue of stage-completion capacity and. 1998. maps onto v of the Erlang distribution. . Neufeld. Finite values of these summary statistics demand that k > 1. k + 1. in that way. see. The Pascal-gamma mixture defines p as e u. e. Accordingly. With v being gammar. or k ½r/(k 1). . Clearly. 1968). for example. Similar observations apply to the individual whose standard deviation in latency equals the expected standard deviation. where p denotes the probability of successful stage completion on a trial (Patil & Joshi. Neufeld & McCarty. with parameters r and k.g. The first summary statistic is the mean process latency.. Each formally prescribed hypothetical participant is representative. Carter. Its parameters are k and p. the directions in which r and k affect the distribution of the base processes’ capacity parameter are now reversed. summary statistics of prototypical performance samples The final aspect of analytic construct validity considered here pertains to statistical summaries of latency distributions for representative individuals. 1994). . . inasmuch as the amount appropriated by the capacity parameter v of his or her Erlang distribution generates the mean or expected value of the summary statistic for the represented class.59 Formal Clinical Cognitive Science tion defines the probability of completing a process comprising k stages on the jth trial. (k 1)/r. j = k .k distribution. suggesting that moment finiteness may depend on critical values of r. The result shows that the value of v for an individual whose own mean latency equals that of the grand mean is. the mean and variance of process-completion trials. thus.0.

k has been released as a free parameter to express variations in cognitive (visual-scanning) performance across groups differing in stress proneness (Hamilton. 2002). Bamber. 1980). parameters linked to the detection and retention of informational sources (Riefer. as E((SD(T|v))2) = E(Var(T|v)) (E(SD(T|v)))2. Model predictions of empirical performance configurations failed to compete with those where r was. variation in k effectively expressed effects of practice. as well. v for an individual whose variance in process latencies is equal to that of the expected value—the average variance—is ((k 1)(k 2))<1/2>/r. complementing empirical support The above exposition of the analytic side of construct validity should dovetail with empirical support in its varied forms (enumerated in Haynes & O’Brien.60 modeling complex systems In like fashion. Concrete illustrations from the literature include those addressed to the understanding of parameters symbolizing determinants of retroactive interference with traces of memorized items (Chechile. 2000). As for the present parameters. Focus centered on support for . Knapp. 1987). A tactic of choice. & Manifold. 1974). 2002). freed to vary with stress susceptibility (Neufeld & Carter. in summary The method of illustration was used to motivate and exposit engagement with analytic aspects of construct validity as it related to interpretation of model parameters. is k /v2 and the expected or mean variance is k r2/ ((k 1)(k 2)). instead. They address. involves selective sensitivity to experimental manipulations or to variation in pertinent organismic.) Here. In other instances (Neufeld. given v. 2000). participant variables. (Note that the theoretically prescribed value of v for the expected standard deviation and that for the expected variance are not identical. Batchelder. The result follows because the latency variance. when it comes to formal-model parameters. unlike a competing parameter considered to indicate the number of stimulus features to be processed during task trials (Weinstein. finite values of the summary statistic require that k > 2.

tenably standing for diverse types of cognitive-task operations. The capacityresource pool was operationalized as a stochastic distribution of a process-model parameter conveying speed of process transaction. The stochastic distribution of capacity values was gamma. Next. Bayesian extensions lent additional support to educed inferences. with parameters r and k. Construct validity of the imputed substantive significance of the capacity-affecting parameter. were inspected for workings of k. was tendered as an indicator of process-completion skill or competence. Performance properties (summary-statistic functions of k) of these idealized participants again conformed to the claimed substantive significance of this parameter. k. Contrasting structures of base-distribution models were employed. Finally. Convergent support was also endowed by alternate-distribution parameters. in turn. whose effects on task performance should resemble those of k.61 Formal Clinical Cognitive Science espoused meaning of a parameter bearing on the capacity-resource pool of a cognitive process (e. whose distributions of process latencies typified those of their embedding class. given that they occupy a similar mathematically prescribed role. The shape parameter. Increasing values of these parameters shifted the capacity distribution downward and upward. Evidence constellated for the posited interpretation of k included substantively meaningful properties transcending specific processlatency (base) distributions. k. Consequences of varying k interfaced with the status of other processing-system parameters in ways to be expected with the interpretation accorded k. the compatibility of analytic results with empirical support for construct validity was described. took the form of its mathematically derived analytic properties. respectively.. Stochastic Modeling and the Psychometric-Artifact Issue in the Study of Differential Cognitive Deficit Stochastic dynamic modeling of cognitive performance can speak to certain long-standing measure-theoretic predicaments accompanying the study of differential cognitive deficit in schizophrenia and . performance of hypothetical participants.g. stimulus encoding).

Brenner. 1981). 2002). be differentially favored over psychometrically inferior measures in indicating statistically significant deficit or disproportionate deficit (expressed as an interaction with corresponding groupwise simple main effects). 1999.” which states that the apparent separation between pathological and control groups on a given measure conflates differences on the addressed cognitive faculty with the measure’s psychometric properties. Neufeld. purportedly. 2003). Advocated redress of the problem has entailed prematching measures’ psychometric properties using a standardization group composed of nonpatients who vary widely on the faculties being examined. Comparative effect-size estimates should. A measure can. other values (notably. Note that. incorporate variance in the measures’ observed scores and in their respective reliabilities. In addition. 1978. Lysaker.62 modeling complex systems other disorders (Carter & Neufeld. The issue is dissected here by translating it into simple but arguably sufficient measure-theoretic terms and then evaluating effects of typical psychometric-adjustment practices. thus. Neufeld. cf. conform to the above intermeasure differences in statistical properties. simplicity of factorial structure is recommended (Strauss. 1984. Neufeld et al. The issue and its endorsed solution continue to elicit considerable adherence (e. or the standardized population-group difference in means ( 1 )/ 2 . 2001. 1973.. Knight & Silverstein. 2001. Neufeld et al. is the common within-group standard devia- . for two groups. 2001). for some three decades. 1977. of course. Neufeld & Broga. sample size) being equal. spuriously indicate greater deficit on the faculty it putatively taps because it possesses greater psychometric precision than measures directed to other faculties (Chapman & Chapman. Matching should.g. the argument has. More psychometrically precise measures will have more statistical power for detecting control/ pathological-group separation and. and alternatives have been articulated (Carbotte. & O’Donnell. 1981.. therefore. At the forefront is the “psychometric-artifact problem. 1989.. In this expression. 2002). statistical power for detecting a difference on a selected measure is an increasing function of the squared noncentrality parameter n(1/4d2). been more or less as follows. 1984. Neufeld & Broga. Boiled down to essentials. The proposed solution of psychometric-property matching has also received substantial challenge. Wilt. Koyfman. where d refers to Cohen’s (1988) d.

where all between-group variance is conveyed by . e. and are functions of one another. within groups . The net contribution to the numerator of 1/4d2 from the latter source is denoted c . thus. the consequences for d2 of varying are complex. increasing —and. Clearly. composed of 2 some amalgam. and as is obvious from Equation (10). Thus. and. first. of those that represent between-group inequalities with respect to latent variables that also generate true-score variance across participants. is bypassed. But c and remain unknown. increasing measure reliability. The expression 1/4( 1 )2 conveys between-group effects and is. and.0. /( + e)—has adverse effects on power if ce < and favorable effects if ce > . tenably. > 0. unknowable from standardization-group variance. by Equation (10). evidently. where the term e represents measurement-error variance. Turning to the issue of psychometric adjustment using standardization-group data. it is apparent from Equation (10) that the nub of between-group inequality. interparticipant variation. makes 1/4d2 itself become (c + )/( + e). because the latter is mute about the structure of group differences composed of c and . and c = 0. second. the measure becomes less group discriminating as its standardization-group psychometric precision goes up. the effects of reducing are unveiled by differentiating (10) with respect to . 0 c 1. decreasing e invariably increases reliability as well as d2. be replaced with c + 2 . the result being (ce )/( + e)2. denoted here as . as would be expected. c and . and observedscore variance is equal to + e. Interestingly. be increased with reduction in . Finally. 1982). and n is the common group sample size. for mathematical tractability and the purpose of illustration. where the scalar c increases as sources of between-group separation on the measure overlap with those of within-group. 1/4d2 can. .63 Formal Clinical Cognitive Science tion. it is assumed that none of c. (10) Although not essential. within-standardization-group measure reliability is equal to /( + e). leaving psychometric matching to grope blindly when it comes to amounts by which e . In fact. duly replacing 2 with + e. That is to say. Of course. along the way. in fact. The expression 1/4( 1 )2 can. of those that are unique to group membership and constant among its members (corresponding to taxonic variables in Meehl & Golden.

in turn. 1981). 2001. Neufeld & Broga. and the comparative fit. given k and v: . a stochastic-modeling translation of the statistical-property issue The terms of Equation (10) can be mapped onto substantive stochastic-modeling expressions. Neufeld. Those of the dual-parameter mixture model are invoked in the present instance. it is shown that Equation (10) can be superposed over and against a stochastic-modeling formulation. 1984. 1989. ushered in is the substantial methodology of model selection. 16–17). Each translated term potentially becomes a model prediction of correspondingly partitioned data. Neufeld & Gardner. 2000.g. These instantiations need not be recapitulated here. but the resulting observations transcend the immediate case. The term e in Equation (10) signifies “psychometric measurement error. More generally. Myung. Forster. & Browne. chap.g. Statistical power considerations now pertain to rejection of a nontrivial competing model (cf. quantifies participantwise intertrial variance in latencies as the expected or mean conditional variance. More productively..8 Empirical examples instantiating the theoretical dissociation between control/pathological-group discriminability and classic psychometric properties have been amply documented elsewhere (Knight & Silverstein. The conversion metamorphoses the equation and arguably leaves behind as vestigial the statistical-property issue. 1990). the translation of Equation (10) breathes content into its otherwise substantively desiccated terms.64 modeling complex systems should be altered (say. 7. Along with accessing advances in model testing and refinement. Cohen. The upshot is that equalizing a pair of measures on and e has no problem whatsoever in leaving them grossly unequal in their group-discriminating power.. pp. of model predictions that go beyond conventional goodness-of-fit statistical-significance tests (e. Included are procedures for evaluating the empirical fit. through increased accuracy of instrumentation or preaggregating measurements. 2006). The current mixture model. see. 1988.” taking the form of random variation in participants’ empirical values across observational trials. Wagenmakers & Waldorp. e.

k Here.2. and m2 may be m + g. For example. for example. The differences in m1 and m2 are. m1 may be m of Figure 6. Equations (11)–(13a)/(13b) together embody twelve model predictions of empirical terms.2 . Among other considerations. (11) The model counterpart of is the variance in the conditional mean or expected latency. thus. The data of each diagnostic group. affiliated with c of Equation (10) above. given k and v: Var(E(T|k ∩ v)m 1. (k – 1)2(k – 2) (12) Both the expected conditional variance and the variance in the conditional expectations (discussed above) provide for within-group random variation in v and k among participants. m1. can be assembled into the expected conditional variance. Parameter estimates.2 . k is randomly distributed across participants within groups. differ only with respect to m. severity of model ( ( (m1r) (m2r) (k r) (k r) ) ) 2 (13a) 2 (13b) . of Figure 6). respectively.r.2r 2(k – 1 + m 1.k )= m1. through moment-fitting procedures. are five in number according to the prevailing model design. for simplicity and in keeping with the design depicted in Figure 6 above. therefore has a model representation as the following function of the between-group difference in means: 1/4 (k –1) (k –1) . The scaled value of . The actual amount prescribed by the mixture model is (r2m1.r. – The analogous value for k as fixed within groups is 1 2 1/4 (k –1) – (k –1) . c .65 Formal Clinical Cognitive Science E(Var(T|k ∩ v)m ).2)/((k – 1)(k – 2)). Model analogues of c and cast k as Poissonm mixed and fixed parameter. and group mean. The distributions themselves are separated according to the values of m1. obtained.2 signifies that the expected variance is taken across both control and pathological groups. variance in conditional expectations. under each experimental condition (encoding load.2) . It is apparent that each parameter is called on simultaneously to participate in empirical values whose sources range from within-participant variation to intergroup separation. In the first case. which. 1.

g. which is tantamount to a predicted second-order difference (see above) being equal to 0 (e.. ultimately rests on the rigor of the model diagnostics that are exerted. once more are engineered by the stochastically modeled system giving rise to Bayesian-posterior distributional moments themselves (illustrated in Neufeld et al. Figures 8–9 and 12 above). Also disclosed are assumptive underpinnings of the ad- . Overarching model designs can be constructed so as to mediate group-level findings to individual differences in their expression among constituent members. then.. 1994. Developments self-disclose implications for evaluation of the cognitive side of treatment-program efficacy.. 2002). Neufeld et al. and progress of an individual’s own response to treatment. 1996). in press). Van Zandt.g. illustrated in Figures 10 and 11 above).. in the case of dynamic stochastic modeling. Further tests are available from distribution properties. Vollick. with respect to cognitive functioning. Neufeld & Williamson. Augmenting significance tests for goodness of fit involving predicted distribution moments (crafted from Equations (11)–(13)) can also be constructed (Neufeld. Brokering the clinical ramifications of contemporary cognitive science makes for a potentially seamless transition between these two domains of investigation. Concluding Comments Placing clinical cognitive science on a decidedly formal platform can unearth productive routes of study and reveal otherwise hidden substantive nuances. Inferential validity regarding proposed dysfunction.66 modeling complex systems testing increases with the ratio of empirical observations to parameter-set size and the heterogeneity of conditions to which the model applies.. 2000). Bayesian extensions stand to further proliferate model predictions. These properties flow from the very design spawning the above distributional moments (e. 2000. For example. Posterior density and distribution functions.. 1993. density functions. Moreover. which. given an individual sample of latencies (e. are functions of time t (García Pérez. Predictions desirably include qualitative properties of empirical configurations. the present model prescribes additive effects on means of diagnostic groups and experimental conditions.g. et al.

it is to be hoped. corresponding to the relative frequency of members of g in the participant sample at large. one or another member of the set may be considered separately. a stronghold of formal clinical cognitive science and assessment methodology may be consolidated. tN}. given the parameter value . Separate consideration occurs. when constructing a posterior parameter-mixing distribution. including that of any specific value . all values of considered. along with lingering ambiguities awaiting resolution. The other comprises the discrete. and cjd({*}| ) is the conditional joint density of the obtained latency data set. . each with its own independent group-related mixing distribution. . given group g. Appendix A: Bayesian Platform for the Current Developments Expansion of {Pr(g|{*})} and [w( the following expression: G g=1 |{*}. Where becomes a set of parameters (e. corresponding to the base rates Pr(g). Formal clinical cognitive science uniquely convenes a quantitative infrastructure for assessing construct validity of meaning assigned to its parameters and.. v and k .” Note as well that the densities of . Pr(g) is the current base rate of group g. given group g. meaning the combined probability density of {t1. Note that the above expansion reveals two sets of Bayesian priors. ujd({*}|g) is the unconditional joint density of the obtained latency data set. It also engages measurement and statistical challenges to inferences about cognitive dysfunction. finite mixing probabilities of the groupwise priors being operative. following the acquisition of a . The latter is a “meta-prior” of this “compound Bayesian formulation. w( |g) is the probability (density) of . One comprises the groupwise mixing distribution of . given group g.g. . see the text).67 Formal Clinical Cognitive Science umbrated assessment technology. The current offering at minimum represents a beachhead from which. for example. In this expression. other properties. . g)] of Equation (5) results in ({ 1/[ G g=1 Pr(g) ujd({*}|g)] Pr(g) ujd({*}|g)} [ w( |g) cjd({*}| )/ ujd({*}|g) ]). t2. potentially. are determined by the group at hand’s prior distribution of .

G 1 G–1 (recalling that Pr(G 1) = 1 – Pr(g) ).1) This product can be show to be a reduction of the joint multinomial probability of the J observed classifications under the model Pr(g|{*}). g = 1. .1). setting each derivative to 0.1) with respect to each term in question. an expression that stands to be distributed as 2 with its associated degrees of freedom (cf. 2003). Riefer & Batchelder. by Equation (5) and observations in Appendix A. 1994. 2.. Specifically.g.0. (B. (2002). 1999).1) beguilingly invites a 2 test for empirical goodness of fit of the prevailing model of Pr(g|{*}). .68 modeling complex systems performance sample {*}. Computations entail aggregating over values of the set-aside parameter(s) according to its (their) group-prescribed prior probabilities (densities) while addressing specific values of the parameter that has been singled out (“marginals”). each probability being an expression of the multinomial likelihood of classification j (see. 2 ln(B. e. Note that Equation (B. which here can be shown to equal 2 ln(B. . is tantamount to 2 ln(multinomial-likelihood ratio). 1988). . Carter & Neufeld.1)/1. also according to its group-specified prior probabilities (densities). this expression is dubiously distributed as 2df=J (G 1) in the present case because the current data array suffers from the problem of “extreme sparseness of cell-wise observations” (see Delucchi. In fact. Maximum-likelihood estimates of the Pr(g). A numerical-search algorithm will be required to replace the analytic solutions as computations become unwieldy with increasing J and/or G. is J j=1 Pr(g|{*}) j = J j=1 Pr(g) ujd({*}|g) j / G g=1 Pr(g) ujd({*}|g) j . Appendix B: Estimation of Base Rates of Symptom Severities The joint probability of an array of J classifications with respect to G groups. as exemplified in Figure 7 above. Computational details undergirding the present developments have been presented in Neufeld et al. . and simultaneously solving the G 1 equations in the G 1 unknowns. are obtained by differentiatg=1 ing (B. García Pérez. Tollenaar & Mooijaart. 1993.

and its association with E(Tn). Wenger & Townsend. (C. n Equivalently. and their availability as finite values.1) 0 t where f(t )/F (t ) is the hazard function H(t ). This expression of process-completion potential.1) is readily seen to be simply 1n(F (t)) (see the discourses of Townsend & Ashby. 2000). including those of the mixtures discussed in the text.” The latter is defined as – f(t )/F(t )dt . Because F (t) = exp( – 0 (H(t )dt ). (– 1) nE(T n) = dm/d 00 MT ( )| =0 = 00 . MT( ) is the moment-generating function 0 f (t) e – tdt. to “process-completion potential. given noncompletion by – t . Here. where F(t) is the cumulative probability-distribution function (see Figure 12 above and related text) and 1 F(t) = F (t) is the survivor function. whose nth derivative with respect to = 0. E(Tn) can also be written as 00 encounters a singularity at n 0 (1 – F(t))tn–1dt = n 00 0 – F (t)tn–1dt. (C. t moments of selected base distributions mixed on a gamma-distributed “capacity” (intensity) parameter Mixture-model latency distributions considered in this section are composed of alternate process-latency base distributions whose “ca- . will be infinite if its integral does not converge: 00 0 f (t) tn dt = 00 .69 Formal Clinical Cognitive Science Appendix C: Finite Values of Distribution Moments values of distribution moments The nth moment of a distribution of latencies t. indicating the conditional rate of process completion at time t . is called on in developments presented in the text under “Interplay with Capacity Utilization” and the third section of Appendix C below. 1983. This version of E(Tn) relates distribution moments.

in turn. Geddes & Scott. Three of the base distributions. making for mathematical tractability (see. (k)(k – 1)! (r + t) k + k Then f(t)tndt = E(T n) = The next base distribution is the compound Poisson (e. whose probability-density function is (vt va) k 1 /( (k ))v exp( (vt va) ). with parameters r and k. and Weibull. Moments are computed for the case where k > n. The density function for the Weibull distribution results when k = 1. and where k = n. above and the related text). if k = n. if k > n. Erlang.g. n being the order of the moment.2) by setting equal to 1. is nested in both the Erlang and the Weibull distributions. r k (k + n – 1)! ln(r + t) (k)(k – 1)! ] t= 00 = 00 t= 0 . Feller.0 and a equal to 0..k.0 and a = 0. (C.2) when k = = 1.. in that both prior and base entail exponentials. A concise method of obtaining the moments of this second-order mixture capitalizes on its moment-generating function ( rn (k + n – 1)! (k – n) (k)(k – 1)! < 00 . The probability-density function of mixture-model latencies f(t) for the Erlang base distribution.0 and a = 0. is rktk –1 (k + k ) . Loosely speaking.g. in turn. the exponential. are nested in the generalized gamma distribution.2) The probability-density function for the Erlang distribution is obtained from (C. e. . 1989). 1996). The exponential distribution. each base distribution has a conjugate prior. as its density function is derived from (C.70 modeling complex systems pacity” (distributional intensity or scale) parameters are randomly distributed according to a gamma distribution whose own intensity or scale parameter is denoted r and whose shape parameter is denoted k. as well as Equation (8). of which the special case at hand is composed of an Erlang distribution whose stage (distributional-shape) parameter k is Poisson distributed with parameter m (see Figures 5 and 6. The Erlang rate parameter of this base distribution. is gamma distributed. Ross. 1966. with its capacity (also rate) parameter v being distributed as gammar.

yields the nth-order moment (for an especially lucid and still current exposition of the moment-generating function. . ci > 0. 1 where MT ( ) is the moment-generating function for each of the i k -stage intercompletion times. evaluated at = 0 and scaled by ( 1)n. unlike that of the Erlang. Note that T 1 MT ( )|(k = 0 ) = 0. presented above) = eme –m. E(Tn) = |k=n also for the general gamma base distribution (discussed below).71 Formal Clinical Cognitive Science MT( ). the nth-order derivative of the latter. 1951). 00 – e –m. MT ( )|(k ∩ v = M k ( )|v . MT ( ) = n 00 0 (rv) k–1 –rv+m(v/(v+ re (k) n n n )–1) dv – e –m . 2. and The nth-order moment of this second-order compound Poisson distribution will be infinite if k = n. 00| = 00 (1 – e –m)| k =n k =n = 00| . Proceeding accordingly. . . Thus. 00| k =0 k ! k =n (by results for the Erlang base distribution. or v/(v + ) for the Erlang distribution with rate parameter v. In the present implementation. i = 1. For this distribution. as follows: E(T n) = = k = 00 m k e –m E(T n|k )| k =0 k ! k =n k = 00 m k e –m. the rate parameter is stage specific. k . k =n By similar reasoning. As the k intercompletion times are identically and independently distributed for a given value of v. Turning to the general gamma base distribution. MT ( ) = 00 0 f (v)(MT ( )|v)dv m k e –m (M ( )|k ∩ v)dv T k! m k e –m (Mk ( ) – e –m )|v dv T k! 1 = 00 0 f (v) k = 00 k =0 = 00 0 f (v) [ k = 00 k =0 ] = 00 0 f (v) [(e m(MT( ) – 1) – e –m|v]dv. E(T ) = (– 1) d MT ( )/d | = 0 . . see Kenny & Keeping. the rate for stage i is civ. . 00 – e –m.

formulas underlying figure 13 1 For the nonhomogeneous Poisson process. k From here. in turn. the density function f(t) For the special case. 0 n (n/ + 1) (k – n) . f(v)E(T n|v)dv = r (k) which again is severed from finite values if k = n. ci v + where Cik is defined in Table 4 of the text: E(T n) = (– 1) nd n/d nMT ( )| = k i=1 =0 Cikci n! rn (k – n). from developments supplied by Morrison (1979). where ci = (k can be computed. if k > n k i=1 (– 1) i + 1 k i () ir kn! ln(r + it) (k)i n + 1 ] t= 00 = 00 t= 0 if k = n. (k) (i 1)). is readily available as 00 1 exp( (vt) ).72 modeling complex systems MT ( ) = = 00 0 f(v)MT ( )|vdv (rv) k–1 –rv re (k) k i=1 00 0 Cik civ dv. it is a simple matter to obtain E(T n) = 00 0 f(t) tn dt = ( k i=1 (– 1) i + 1 k rn n! i () (k – n) (k)i n < 00 .v . E(Tn). . The probability-density function for the Weibull distribution is v(vt) whereby E(Tn|v) = (n/ + 1)/vn.k distributed. as i k (– 1) +1 i=1 (r + it) k + 1 (ki ) r ik . with dynamic rate vt having been gammar.

Notes Sources of support for this research include operating grants from the Social Sciences and Humanities Research Council of Canada and the Workplace Safety and Insurance Board and operating grants and a New Emerging Teams grant from the Canadian Institutes of Health Research. Neuroconnectionist modeling. Consider. 1984). a special case being numerical solutions to differential equations composing nonlinear dynamic (“chaos-theoretic”) systems. Blind alleys are flagged and promising leads discerned. entailing computer simulations of task-transacting activations among neuronal units or unit modules (Rum- . 1989. sabbaticant host. Craig.73 Formal Clinical Cognitive Science f(t) = r k (k + k)t k – 1 . A parameter is a variable in a mathematical expression whose values can change without altering the structure of the expression in which it is embedded (cf. Kenneth D. Bandura. whether new formal proposals differ. p. private citizens. Even if they seem to rediscover preexisting ideas. 5. as contrasted with desired (but most often intractable) explicit solutions of the “initialvalue problem” (Koçak. numerical vs. They make clear what has been formulated. 2. and to Ruth and Douglas Barr. 435). for providing facilities during the writing of this manuscript. 1. Borowski & Borwein. precisely because they are explicit. The place of precedent is usually more fuzzy in the case of verbal statements. (k – 1 )! (k) (t + r ) k + k (j + k )rkt j . this integral is available as 1n(F (t)). in what respects. Informal theories escape because they are not forced to lay bare their assumptive moorings before commencing (cf. j! (k)(t + r ) j + k t and k –1 – F(t) = j= 0 The hazard function H(T) = f(t)/F (t) enters into the index of process-completion potential as follows: 0 H(t )dt . the Department of Psychology. Happily. 4. if so. if taken to task. as a representative example. usually appeal to allegedly important nuances of difference. 3. University of British Columbia. and. These explicit starting points can always be challenged. Formal statements foster place keeping regarding the current state of progress in the problem area. 1989). such as those above (see the first section of Appendix C). Appreciation is expressed to the University College of the Fraser Valley. symbolic integration. 1984. Staddon. subsequent proposals almost always escape indictment for ignoring antecedents and. especially considering the possibility of complex f(t) and F (t).

10. arguably has received a warmer welcome (including in the field of clinical cognitive science) because of its face validity as a “brain metaphor” (notwithstanding certain challenges to biological plausibility) and because of commercial software expediting the use of simulation algorithms virtually since their introduction (McClelland. 8. 1998). 80. pp.74 modeling complex systems melhart. & Miller. 8.. are deemed to express dynamic properties of operations occurring at other strata. 360–369. Useful . C. distribution functions. 91. Psychological Bulletin. C. Psychological Review. 1987). The performance sample is represented as a set of observations. Busemeyer & Townsend. Personal control over aversive stimuli and its relationship to stress. (1998). and Other Pieces (pp.. (2003). Polking. F.. The concept of target levels of inference. Essays. (1984). 169– 235). In Richard Foster Jones (Ed. or levels of analysis. The observations are deemed to be independent. The interstrata associations are isomorphic in the same way that a computer-machine language and a computer-programming language are coextensions of one another. (Original work published 1605) Bandura. Neuroimage. The Advancement of Learning. Bacon. Psychological Assessment. Although compelling in other respects to be sure. & D’Esposito. and similar properties). illustrated above. McClelland. (1998). Batchelder..). D.g.g. 1986. Added to considerations of response properties of interest are those of their distributional properties (e. References Aguirre. 508–511. Santor & Ramsay. 6. K. in press). 1993.. G. and associated boundaries of inference are underscored. R. M. Marr. (1973). (1937). Multinomial processing tree models and psychological assessment. & pdp Research Group. The variability of human.. and algebraic operations to which they are subjected do not depend on their ordering. as exemplified in Equation (2) above. Carter & Neufeld. as opposed to either an ordered set or a response vector. 286–303. New Atlantis. 443–445. 7. E. Targeting one or the other stratum then becomes a matter of the nature of the focal problem. and intertrial variances. Advancement of Learning. 331–344. Averill. Cramer. bold hemodynamic responses. A.g.. including means. R. either in its own right or via irt-informed measure composition (see. e. including the algorithmic (neuroconnectionist) and the implementational (neurophysiological. Rummelhart. moments of performance latencies. R. J. Batsell. Zarahn. M. methodology developing from item-response theory (irt) can be shown not to resolve the above dilemma. cf. R. W. J. H. Representing personal determinants in causal structures. e. 1982). & pdp Research Group.. New York: Odyssey. Model properties and response parameters entailing task-performance latency.

605–634. R. J. B. M. Malla. Journal of Neuroscience.. Drost. & Heeger. W.. E. 4207– 4221. 88.. & Stout. Densmore. 16. J. (1959). D. (2000). (1993). (1996). Importance of measurement theory.. (1978). Newbury Park ca: Sage. 247–263. (1989). Stochastics and Dynamics. & Neufeld. (1980). Boksman. C. . (2001). D. 432–459. O. D. R. R. & Raudenbush.. Brenner. Bollen. 633–654. 56. S. Borowski. 28–37... Linear systems analysis of fmri in human v1. Dictionary of Mathematics. & Neufeld. A. 171–189. Cognitive processing of multidimensional stimuli in schizophrenia: Formal modeling of judgment speed and content. Lysaker. A. 237–244. Wilt.. Statistical decision theory and Bayesian analysis (2nd ed. Berger. 313–316. M.. Psychometrically matched visual-processing tasks in schizophrenia spectrum disorders. H.. Boynton. B. R. J. P. R. K.. Théberge. W.0 Tesla fmri study of brain connectivity during word fluency in first episode schizophrenia. 47. 44.75 Formal Clinical Cognitive Science mathematical relationships embedded in Tversky’s elimination by aspects model. Williamson.. C. 100. Journal of Mathematical Psychology. Braithwaite. J. A. (1968). 75. & O’Donnell. 14. Convergent and discriminant validation by the multitrait-multimethod matrix. J. Bryk. H. Journal of Abnormal Psychology. A 4. G. J. & Borwein. Carbotte. G. 253–262. T. P. R. R. Converging operations or matched control tasks? Journal of Psychiatric Research. J. Hierarchical linear models for social and behavioral research: Applications and data analysis methods.. P. Psychological Bulletin. S. F. (2002). K. J. London: Cambridge University Press. J.. (1992). New York: Springer. Scientific explanation. Journal of Mathematical Psychology. Busemeyer. 14. and experimental design for testing the significance of interactions. (2005). 81–105. & Townsend. Latent variables in psychology and the social sciences. Y. Menon. Busemeyer. (1999). 112. (1985). Engel. Busemeyer. & Wang. Busemeyer. Glover. M.. J. E.. Model comparisons and model selections based on generalization test methodology. J. & Holmes. Modeling a simple choice task: Stochastic dynamics of mutually inhibitory neural groups. W. Brown. (2003). Carter. 538–544. 1–33. R.. A contribution of cognitive decision models to clinical assessment: Decomposing performance on the Bechara gambling task. D. Annual Review of Psychology. Pavlovsky. J. J. A. Schizophrenia Research. Takhar. J. W. 108. R. M... Campbell. Koyfman.. New York: HarperCollins. 1. A. H. Journal of Abnormal Psychology. A.). error theory. T. Psychological Bulletin. A. W... R.. Decision field theory: A dynamiccognitive approach to decision making in an uncertain environment. Psychological Review. & Fiske.. Psychological Assessment. (2002). 53.

52. D. & Benn. E. R. R. L. & Chapman. (2000). W. New York: Wiley. 44. J. Journal of Abnormal Psychology. V. 42. L. J. (1998). J. (2003). A new method for estimating model parameters for multinomial data. R. Statistical distributions (3rd ed. Problems in the measurement of cognitive deficit. K. Friston. Delucchi. M. L. J. Construct validity in psychological tests. Application of process models in assessment psychology: Potential assets and challenges. Construct validity: Construct representation versus nomothetic span. S. (1989). Hastings.. Hillsdale nj: Erlbaum. C. W.).). W. Frackowiak. Zeki.. 93. S. 295–320). J. (1973). 281–302. Journal of Mathematical Psychology. N. K. P. Friston. J.76 modeling complex systems Carter. (1966). 10. J.. & Penny. Dolan. R. R. Parameter estimation and goodness-of-fit test- .. J. C. Chechile. Clark. Keren & C. Cohen. (1998). J. S. 279–298. W. (2001).. Alternate realities: Mathematical models of nature and man.. In G. 116. The mathematics of the imaging techniques of meg. R.. (2004). (1993. Trieste. & Paivio. (1993). Evans. J.. Observational and theoretical terms in psychology: A cognitive perspective on scientific language. Fokas. (Eds. A handbook for data analysis in the behavioral sciences: Statistical issues (pp. 16. J. J. Z. Dynamic causal modeling. J.. Carter. & Marinakis. 2).. J. Embretson.. B. (1988). Psychological Bulletin. Neufeld. & Penny. Chechile. Chapman. Feller. New York: Wiley. (1953). W.). 110.. 1273–1302. Dirac Lecture. J. A. Casti. P. Commentary on two articles concerning generalized and specific cognitive deficits. Freedman.. M. 1671–1687. L. New York: Wiley. 432–471. J. S. Statistical power analysis for the behavioral sciences. R. Journal of Experimental Psychology. 203–222. P. Frith. Psychological Assessment. J. Cronbach. Psychological Bulletin. Doob. R. Human brain function (2nd ed. L. D.. (1983). 19. Cognitive processing of facial affect: Neuro-connectionist modeling of deviations in schizophrenia. ct. (1994). J. (2006). & Peacock. 500–512. An introduction to probability theory and its applications (Vol. (1989). & Meehl. Journal of Abnormal Psychology. & Neufeld. On the use and misuse of chi-square. L. (1955). A. Neuroimage. 179–197. International Centre for Theoretical Physics. M. W. 1989. A. Hillsdale nj: Erlbaum. pet and spect. A. International Journal of Bifurcation and Chaos.. D. New York: Wiley. K. Psychological Bulletin.. 79.. García Pérez. A. Trace susceptibility theory. (in press). 380–385. Ashburner. American Psychologist. Stochastic processes. November). Lewis (Eds. 31–39. K.). Chapman. (1987).. L. Some beautiful equations of mathematical physics. San Diego: Academic. & Chapman. Price. Harrison..

33–56. W. Spielberger (Eds. & Singer. Handbook of psychophysiology (Vol. (1986). Vol. Sunderland ma: Sinauer. Psychological Review. E. New York: Penguin. Consequences of deterministic and random dynamics for the course of affective disorders. Cacioppo. E. pp. L. Clinical applications of analogue behavioral observation: Dimensions of psychometric evaluation. (2000). S. 94. (1987). C. 193–207. (1972). Continuous univariate distributions (2nd ed.. New York: Cambridge University Press. George. W. 9–18.. J. In E. Recipes for classes of definite integrals involving exponentials and logarithms. J. New York: Springer.. New York: Academic. 46. Psychological Review. N. 42. 108.).). D. & Krieg. J. Principles of behavioral assessment: A functional approach to psychological assessment. K. S.. N. (1979). Journal of Abnormal Psychology. 53. (1966). A. Hockley. (1984). M. Imagery and verbal aspects of schizophrenic informational-performance. . British Journal of Clinical Psychology. 1). & McCarthy. Tassinary. J. O. 247–282.. 486–495. & Murdock.. García Pérez. 13–30). T. (2000). V. (2000). Washington dc: Hemisphere.. R. I. 39–56). 23. Exact finite-sample significance and confidence regions for goodness-of-fit statistics in one-way multinomials. Geddes. 73–85. W. G. M. E. New York: Wiley. (1980). Theory of probability. O. Murdoch (pp. Braun. Highgate-Maynard. Derivatives of matching. Gilden. N. Reading ma: Addison-Wesley. & Balakrishnan. Herrnstein. G. J. Berntson (Eds. An information processing analysis of environmental stress and life crises. D. & Scott. S. L. R. 47. (1989). B.. W. (1991). Haynes. 95. Watt (Eds. Psychological Assessment. L. B. T.. A. B.. A decision model for accuracy and response latency in recognition memory. 341–358.). R. W. Relating theory and data: Essays on human memory in honor of Bennet B. J. Stress and anxiety (Vol. A.). G. N. Hamilton. In I. G.. & Neufeld. R. 7. & Neufeld. & G. 192–201). pp. Hintzman. Cognitive emission of 1/f noise. British Journal of Mathematical and Statistical Psychology. Biological Psychiatry. D. Huber. Sarason & C. J. Johnson. Why are formal models useful in psychology? In W. H. Gregson.. Schizophrenic memorysearch performance involving nonverbal stimulus properties. Computers and mathematics (pp. Dynamic modelling. A. Haynes. M. & O’Brien. Gleick. (2001). Kaltofen & S. L. Huettel. Psychological Review. 13. 924–948). Kotz. T. Chaos: Making a new science. 256–262. New York: Plenum/ Klewer.. 86. In J. S. (1988). British Journal of Mathematical and Statistical Psychology. Hillsdale nj: Erlbaum. 67–73. Harris.. S. Urban stress. (2001).77 Formal Clinical Cognitive Science ing in multinomial models.. Functional magnetic resonance imaging. L. (1994). (1999). Lewandowsky (Eds. (2004). Glass. C. Hockley & S. D. & Pressing. Jr. Song.

& Ashby. Decision-theoretic aspects of stress arousal and coping propensity. Response times: Their role in inferring elementary mental organization. I. Journal of Personality and Social Psychology. New York: Oxford University Press.. J. 92. The “horse race” random utility model for choice probabilities and reaction times. Facial electromyographic measures distinguish covert coping from stress response to stimulus threat.. Mathematics of statistics (2nd ed. M. Link. R.. Luke. & Silverstein. 1–27. Personality and Individual Differences. New York: Van Nostrand. 36. (1988). Oxford: Oxford University Press. Experimental design: Procedures for the behavioral sciences (3rd ed).. M. E.. Kotz. (1998). (2006). W. MacCallum. (1985). Continuous univariate distributions (2nd ed. Correcting response measures for guessing and partial information. A. (1994).. 77. S. (1999). J.. R. Gati. D. Structural equation modeling. 110. (1989). A. (1995). (1970). 42. Williamson. 161. & Keeping. Knight. J. Chicago: University of Chicago Press. K. Karabatsos.). 36–44. (1989). (1992). Psychological Bulletin. New York: Guilford. W. Time and change: Using survival analysis in clinical assessment and treatment evaluation. R. Trans. Kirk. Kukde. Kenny. P. R. Kline. T. R. Hillsdale nj: Erlbaum. J. 185–208. Marley. Indianapolis: Bobbs-Merrill. 2). F. (2001).78 modeling complex systems Johnson. British Journal of Mathematical and Statistical Psychology. pp. B. C. (1951). 30. 50(2). 123–148. P. (1962). 15–30. W. Luce. and its competing risks interpretation. R. R. Lanius. W. R.. A process-oriented approach for averting confounds resulting from general performance deficiencies in schizophrenia. Vol. Monterey ca: Brooks/Cole. M. S.. (1982). G. 39–56). E. (2004). H. & Neufeld. M.. New York: Springer.. R. (1986). N. & Homan. S. Ellington. Neufeld.. J. Mathematics and the search for knowledge. Boksman. L. G. Kline.. 469–486. The structure of scientific revolutions. Kant. American Journal of Psychiatry. Psychological Assessment. Kuhn. (1986).. Metaphysical foundations of natural science (J. 10. & Neufeld. Relationships between linear systems theory and covariance structure modeling. & Balakrishnan. S. . (1994). 1–20. J. A. Journal of Mathematical Psychology. Koçak. The nature of traumatic memories: A 4-t fmri functional connectivity analysis. H.). 16. Journal of Mathematical Psychology. Lees. Densmore.. Differential and difference equations through computer experiments (2nd ed. & Colonius. 360–378. R. A. S. C. Marley. S. New York: Wiley. Bayesian nonparametric model selection and model testing. A. A.. A random utility family that includes many of the “classical” models and has closed form choice probabilities and choice reaction times. A. 211–228. S. M. Journal of Abnormal Psychology.. & Menon. D. N. M. 13–36. F. S. Journal of Mathematical Psychology. J.

Taxometric methods. R. 2. M. D. 10. London: University of Western Ontario. W. In R. A. California: Sage. San Francisco: Freeman.. T. J. I. E. J. R. Sir Ronald. Parallel distributed processing: Explorations in the microstructure of cognition: Vol. R. R. & pdp Research Group. 18.. Department of Psychology. J. (1982). (1986). (1998).. J. W. (1988). 19. J.). Neufeld. W. J. The cognitive bases of mental disorders (Annual review of psychopathology. 1. W. Neufeld. M. W. J. Journal of Mathematical Psychology. Meehl. McFall. 71–132). 172. Neufeld. 1. (1989). (1979). (1978). 316–330. Vol. & Viken. 44(1). J. pp. M. Journal of Consulting and Clinical Psychology. R. R. (1991). London: University of Western Ontario. W. R. (Eds.. G. Miles.. & Gibbon. Butcher (Eds. Qualitative data analysis. (1994). 806–843. & Townsend. Foundations of psychological assessment: Implications for cognitive assessment in clinical science. P. In P. M. M. 720). & Golden. R. Psychological stress and psychopathology (pp. Diathesis stress model or “just so” story? Behavioral and Brain Sciences. Advances in the investigation of psychological stress (pp. & Huberman. E. Kendall & J. 46. Neufeld.. Journal of Nervous and Mental Disease. N. J.. Theoretical stress and stress-proneness effects on information processing in light of mathematical models of stochastic processes (Research Bulletin No.). Meehl. McClelland.). R. Cambridge ma: mit Press. Foundations. An individual differences pure extinction process. M. Memory in paranoid schizophrenia. W.). . Neufeld (Ed. Handbook of research methods in clinical psychology (pp. A. & Browne. In R.). Neufeld (Ed. Re: The incorrect application of traditional test discriminating power formulations to diagnostic-group studies. L. In P.. Neufeld. Newbury Park ca: Sage. McGill. New York: Wiley. 127–181). 155–177.. R. S. (1982). New York: Wiley. Rummelhart. W. 240–270). 373–374. J. The economy of probabilistic stress: Interplay of controlling activity and threat reduction. Psychological Assessment. 1–18..79 Formal Clinical Cognitive Science Marr. J. New York: McGrawHill. Magaro (Ed. (1965). Department of Psychology. (1996). Journal of Mathematical Psychology. E. Stochastic models of information processing under stress (Research Bulletin No. P.. Neufeld. W. D. J. The general-gamma distribution and reaction times.. W. British Journal of Mathematical and Statistical Psychology. M. R. T. Townsend. W. R. 41. J. (1994). 565–566. Morrison. (2000). (1982). On decisional processes instigated by threat: Some possible implications for stress-related deviance. R. 734). Journal of Mathematical Psychology. B. C. (1984). Morrison. J. Special issue on model selection. Theoretical risks and tabular asterisks: Sir Karl. (1995). Methodological aspects of laboratory studies of stress. Vision. and the slow progress of soft psychology. J. & Lefebvre. D. Neufeld. Myung. Forster. McFall. 231–261). 307–315. J.

Stress as an irritant of psychopathology. R. Vollick. J. J. 31–56). W. Neufeld (Ed. R.. W. Journal of Mathematical Psychology. W. R. J. In R. R. Neufeld. Formally deciphering effects of stress on cognitive performance. Journal of Nervous and Mental Disease. Neufeld. 47. J. S. Oxford: Oxford University Press. In I. Quantitative schemata of decisional stress-control. J. Washington dc: American Psychological Association. and outcome: The Second Kansas Series in Clinical Psychology (pp. Data aggregation in evaluating psychological constructs: Multivariate and logical-deductive considerations. J. J.. J. W. Boksman. (2002). R. W. with application to schizophrenia.. (in press). (1999b. D. Stochastic modelling of stress effects on proofreading. C. W. 360. 34. Kingston on. treatment. Vollick. Snyder (Eds. & Mothersill. J.. J. 14.. J. George. L. R. Evaluation of information-sequential aspects of schizophrenic performance. R. D. (1980). Neufeld (Ed. W. In R. R. Department of Psychology. Carter. J. J. Neufeld.. & Broga.. Santa Cruz. 385–397. R. 193–226. C. 2: Methodological considerations. I.. K. W. L. (1990). Neufeld. & Carter. J. Boksman. R... D. Cromwell & R. Neufeld. J. & Jetté. (1977). Paper presented at the annual Summer Inter-Disciplinary Conference on Cognitive Science.). (1981).... Neufeld. London: University of Western Ontario. Quantitative response time technology for measuring cognitive-processing capacity in clinical studies. K. W. June). K. W.. (1994). J. Psychological Review. Psychological Assessment. M. W. Dynamic differentials of stress and coping.. W. D.. A formal analysis of stressor and stress-proneness effects on basic information processing. Neufeld.. & McCarty. (1999a). J. R. (2002. W. . Spielberger (Eds. & Gardner. Squamish bc. R. Stress and anxiety (Vol. Neufeld. 169. Carter. Advances in clinical cognitive science: Formal modeling and assessment of processes and symptoms. & Highgate. R. M. W. Advances in clinical cognitive science: Formal modeling and assessment of processes and symptoms. R. Fallacy of the reliability-discriminability principle in research on differential cognitive deficit (Research Bulletin No. 569–579. R. T.).). Schizophrenia: Origins. New York: Hemisphere.. Paper presented at the 32nd annual meeting of the Society for Mathematical Psychology.. (2000). W. Neufeld. R.). Queen’s University. Neufeld. Levy. I. 176–196). A mathematical process account of group and individual differences in memory-search facilitative stimulus encoding.80 modeling complex systems Neufeld. Neufeld. Neufeld. J. 276–296. J. Paper presented at the 33rd annual meeting of the Society for Mathematical Psychology. Jetté. In R. pp. 8. Stochastic modelling of stimulus encoding and memory search in paranoid schizophrenia: Clinical and theoretical implications. 106. (in press). & Vollick. L. Sarason & C. processes. (1993). J.. 279–298. & Broga. J. Washington dc: American Psychological Association. July). Townsend. R. J. British Journal of Mathematical and Statistical Psychology. Application of stochastic modelling to group and individual differences in cognitive functioning. T. & Jetté.

New York: Wiley. A. W. Busemeyer. Memory and Cognition. 117–130. (1986). Neuropsychological correlates of positive symptoms: Delusions and hallucinations... A dictionary and bibliography of discrete distributions. M. & Neufeld. D. 102. (2003). 95. I. J. (1983). Perceptual comparisons through the mind’s eye. J. Mental representations: A dual coding approach. (2004). M. & Bandettini. Batchelder. Paivio. Pantelis. & Batchelder. J. D. E. A. C. Santor. New York: Cambridge University Press. New York: Hafner. Nelson. Stochastic processes. Similarity scaling and cognitive process models. A conceptual and methodological checklist for conducting a taxometric investigation. Nicholson. Ross. 85–118). S. Smith. W. S. McClelland.. R. W. M.. & Manifold. 3. A dynamic vulnerability perspective on stress and schizophrenia. J. Behavior Therapy. (2001).). San Francisco: Holden-Day. 403–447. Positron emission tomography and functional magnetic resonance imaging. J. Schizophrenia: A neuropsychological perspective (pp. M. 2). L. & Ramsay. Psychological Assessment. American Journal of Orthopsychiatry. In J. A. 3–13.. 205–235). (2002). Stochastic processes (2nd ed. P. S. M. D. Journal of Abnormal Psychology. 184–201.. E.).). London: Wiley. Jorenby. & Ruscio. S. E. & G. & Neufeld. A. G. & T. (1987)... (1998). (1988). R. D. D. Progress in the technology of measure- . New York: Oxford University Press.. 14. R. Journal of Abnormal Psychology. D. Psychological Review. (2000). Reiman. A. G. G. G. (1996). W. Smoking withdrawal dynamics: 1. Barnes (Eds. Bernston (Eds. H. R. Abstinence distress in lapsers and abstainers. Mathematical models in the social and behavioral sciences. (1996). & Townsend. In C. I. & pdp Research Group. (1971). Cacioppo. New York: Holt. J.. T. Annual Review of Psychology. The friendship of science and religion (Plenary Lecture: Veritas Forum). Nosofsky. T.. 259–270. J. H. 635–647. B. (1975). New York: Wiley. W. 25–33. (2003). Imagery and verbal processes. Parzen. London: University of Western Ontario. Psychological Review. 35. V. R.. pp. E. H. 62. 370–392. A. D. Classification of the schizophrenias according to symptomatology: A two-factor model. (1992). M.81 Formal Clinical Cognitive Science Neufeld. 318–339.. R. Roe. Cambridge ma: mit Press. 108. J. M. R. W.. Rumelhart. R. Tassinary.. Lane.. 43. 112. Rappaport. O. Van Petten. T.. T. & Williamson. (1993). R.. M.. Knapp.. Multialternative decision field theory: A dynamic connectionist model of decision making. Riefer. Parallel distributed processing: Explorations in the microstructure of cognition (Vol. & Joshi. A. Piasecki.. Bamber. C. & Baker. (1962).. Patil. B. Nicholson. M. Ruscio. Multinomial modeling and the measurement of cognitive processes. Riefer. Rinehart & Winston. (1992).. R. J. Fiore. Polkinghorne. Paivio. Paivio. E. (1968). Handbook of psychophysiology (2nd ed. P.

H. “The distemper of learning . Townsend. 117–123.. 102. L. Y. (1978). Type I errors and power of the parametric bootstrap goodness-of-fit test: Full and limited information. Townsend. parallel processing: Sometimes they look like Tweedledum and Tweedledee but they can (and should) be distinguished. E. The trouble with physics: The rise of string theory. J. S. Mowrer (Eds. 46–54. Mulaik. Quarterly Journal of Experimental Psychology. (1983). 110. Introduction to multidimensional scaling: Theory. (1997).. (2006). Castellan & F. (1990). B. In J. Journal of Mathematical Psychology.”: A review of S. J. Journal of Abnormal Psychology. (2004). (2003). F. 502–507. & Baker. 48. L. Introduction to the special section on structural equation modeling. M. 506–507. (1975). 393–423. Reynolds. Klein and R. . (1983). G. T. T. (1984). . R. Townsend. (2001). Psychological Review. Smith. T. Tomarken. Staddon. & Mooijaart.) Contemporary learning theories: Instrumental conditioning theory and the impact of biological constraints on learning... 4–5. Townsend. methods and applications. J.). Staddon. A.. H. A. New York: Academic. by Frederick David Abraham]. J. Townsend. (1994). S. Tollenaar. Psychological Assessment. 1. Noncentrality interval estimation and the evaluation of statistical models. W. Boston: Houghton Mifflin. Mahwah nj: Erlbaum. R. J. J. (1995). & Ashby. Parallel versus serial processing and indi- . 91. (1984).. & Fouladi. Smolin. S. Psychophysically principled models of visual simple reaction time. American Journal of Psychology.. 271–288. British Journal of Mathematical and Statistical Psychology. the fall of a science. Hillsdale nj: Erlbaum. Takane. Cognitive theory (Vol. 363–400. Strauss. and what comes next. Psychometrika. 221–257). 3. J. Memory scanning: New findings and current controversies. 112. Serial vs. Steiger (Eds. L. J. Social learning theory and the dynamics of interaction. 28.82 modeling complex systems ment: Applications of item response models. Sternberg. & Young. Uncovering mental processes with factorial experiments. J. (1991). Psychological Review. & Fific. J. In L. J. Restle (Eds. 27. E. T.. S. & Ashby. N. 345–359. A visual approach to nonlinear dynamics: Just cartoons or a serious pedagogical device? [Review of A visual introduction to dynamical systems theory for psychology. T. 56.). Journal of Abnormal Psychology. Methods of modeling capacity in simple processing systems. Multidimensional scaling models for reaction times and same-different judgments. Steiger. P. F. & Sergent. 36. 567–593. Harlow. E. M. 200–239). F. 107. Cambridge: Cambridge University Press. Contemporary Psychology. & J. R. J. Townsend. (2003). T. (1981). M. Methodology for identifying specific psychological deficits: Introduction to the special section. 523– 525. What if there were no significance tests? (pp.. Schiffman. G. Stochastic modelling of elementary psychological processes. A. Psychological Science. 1–32. 10. T. R. L. A.

Baker (Eds. J. (2006). & Waldorp. serial. Special issue on model selection: Theoretical developments and applications. (1977). Weinstein. (1994). J.. T. Newbury Park ca: Sage. Basic response time tools for studying general processing capacity in attention. 341–367. E. Psychological Review.. & Wenger. Psychonomic Bulletin and Review. . Townsend. Assessment of mental architecture in clinical/cognitive research. (Eds. M. J. Spatio-temporal properties of elementary perception: An investigation of parallel. N. Wiggins. G. Tversky. N. A. The serial-parallel dilemma: A case study in a linkage of theory and method. 97–116. (2004b). D. N. J. J.. N. (1995). How to fit a response time distribution. Tversky. 111.. (2004a). A. J. P. Wright. (1974).83 Formal Clinical Cognitive Science vidual differences in high-speed search in human memory. Kraut. 67–99. University of Western Ontario... 953–962. B. Choice by elimination. D. (1972b). Psychonomic Bulletin and Review. Elimination by aspects: A theory of choice. Vollick. Reading ma: Addison-Wesley. Psychological Clinical Science: Recent Advances in Theory and Practice. Manuscript in preparation. Effect of noise on intellectual performance. & Neufeld. Journal of Mathematical Psychology. L. M. (2000).. A theory of interactive parallel processing: New capacity measures and predictions for a response time inequality series. A. Townsend. 7. 289–299. Treat. & Neufeld. T. Solving measurement problems with the Rasch model. & Townsend. T. (1998). Hillsdale nj: Erlbaum. & T. Townsend. T. and cognition. Waller. 1003–1035. Journal of Mathematical Psychology. Townsend. R. 79. Stochastic modelling of encodinglatency means and variances in paranoid schizophrenia. S. J. A. Wagenmakers. W. 548–554. (1972a). Journal of Educational Measurement. Fific. & Meehl. 391–418.). 14. J. Multivariate taxometric procedures: Distinguishing types from continua. J. 11(3). Perception and Psychophysics. (2004). Van Zandt. D. Unpublished doctoral dissertation. Psychological Review. Journal of Applied Psychology.. T. M. (2000). Vollick. J. Journal of General Psychology. (1973). (in press). E. J. 9. M. & Wenger. Faculty of Graduate Studies. Stochastic models of encoding-latency means and variances in paranoid schizophrenia. D. and coactive theories. G. W. 66(6). Journal of Mathematical Psychology. R. 321–359.). 424–465. 50(2). perception. 59. & Nozawa.. In T. 127. 39. Wenger. Personality and prediction: Principles of personality assessment. T.

.

and action. psychopathology is a genuinely interdisciplinary field. • Biology focuses on genetics and on neuronal processes associated with psychological dysfunction. and practitioners in this field (traditionally. Researchers in this field explore maladaptive cognition. thereby starting from a number of different perspectives. psychiatrists) commonly apply a range of quite different instruments and interventions when they. such perspectives are usually given by the scientific disciplines from which such exploration was initiated: • Psychology focuses on cognitive and behavioral models of psychopathology. for instance. emotion. combine psychotherapy with neuroleptic medication in a milieu-oriented ward of a psychiatric hospital. . • Sociology identifies social and societal factors that are associated with the disorders in individuals. • Pharmacology studies chemical compounds that act on the neuronal substrate. Thus.A Dynamics-Oriented Approach to Psychopathology Wolfgang Tschacher and Zeno Kupper University of Bern Theoretical Framework: Optimality in Self-Organizing Systems Psychopathology addresses disorders of the psyche.

workers in artificial intelligence (ai). and many others contribute. the distinction between the computationalist perspective and the dynamic approach has been debated intensely. Consequently. in principle. Motivational processes are. Bridging assumptions of dynamic systems approaches to cog- . a natural foundation from which psychopathological states of cognitive functioning may be regarded. be simulated using digital computers.” which is based on the physical-symbol-systems premise (Newell & Simon. is derived from computationalist theory. the mind is regarded as a physically implemented symbol system. 2002). Cognitive science is. 1997. as well as the cognitive-behavioral therapies currently in use in psychiatry and clinical psychology. 2003) has challenged this idea of cognition as computation. We will argue that cognitive acts can be viewed as coherent patterns emerging spontaneously in a complex cognitive system.” Cognitive science is a field to which neuroscientists. The major part of today’s cognitive psychology. According to this premise. philosophers of mind. This field has gained momentum in the last decades especially because of advances in brain research (e. we will outline our theoretical guidelines and suggest how the two main ingredients of our approach—cognitive science and dynamic systems theory—may be combined.86 modeling complex systems The perspective for psychopathology that we wish to put forward in this chapter is in itself interdisciplinary—of its two ingredients. consequently. 1972).g. Pattern formation via self-organization is the cornerstone of this approach to psychopathology. Pfeifer & Scheier. 1999. the other cognitive science. It is with respect to such motivational gradients that cognition is more or less “optimal. Clark.g. one is dynamic systems theory. The dynamic approach to cognitive science (e. In this introductory section. defined as those forces or gradients that “drive” the complex cognitive system. the use of brain-imaging techniques and the profound implications of those techniques for psychiatry) and because of the perennial questions of consciousness that have been receiving increased attention recently (Carter. thus. 1999. psychologists and psychiatrists.. The computationalist perspective in cognitive science elaborates on the theory of information processing and continues the tradition of “classic ai. Among the current approaches in cognitive science. which can. Tschacher & Dauwalder.. mental processes are identical to symbol processing.

task affordances in psychological systems). 1983. 1995. Knoblich. those constants of a physical environment that confine behavioral systems) and further environmental conditions that “energize” or “drive” the systems (e. Synergetics starts from the observation that the behavior of many systems is strongly determined by environmental conditions. These conditions may be divided into constant conditions or constraints (e.g. the emergence of new qualities is studied.g.87 Approach to Psychopathology nition have been identified: “offline” reasoning is continuous with “online” motor-control strategies in the sense that abstract cognition may be decoupled from behavior in an actual environment but still work with the same principles (Mechsner. In many cases control parameters have the form of externally applied gradients. been shown that a large number of systems are accessible to unifying mathematical and conceptual approaches. Thus. The main question is whether there are general principles that govern the behavior of complex systems when qualitative changes occur. . Kelso. By means of their interaction. Then it may tend to leave its state and develop a new structure or behavior. such features are often directly observable as patterns. the colloquial distinction between software and hardware makes sense only with respect to computers. free energy in physical systems. the system may become unstable.. as in the computationalist perspective. 2001). but are emergent properties of the complex brain/mind system. & Prinz. There is agreement in the dynamics community that cognitive patterns are not provided by programs. When one or several control parameters are changed. not with respect to naturally occurring cognition.. the core part of the theory of complex systems. In the mathematical approach these latter environmental conditions are taken care of by control parameters. Kerzel. 1996. 1977). Self-organization theory (synergetics) deals with systems composed of several or many components. By these assumptions cognitive science and dynamic systems theory have joined forces to formulate a non-Cartesian theory of cognition. Emergent properties are modeled in the mathematical framework of self-organization theory (Haken. Nicolis & Prigogine. in fact. In other words. These situations are probably of particular interest. And it has. The general strategy of synergetics is as follows: It sets out from a state of a system that is already known under a certain control-parameter value. these components can produce new qualitative features on macroscopic scales.

the individual components react on the order parameters and. An example of visual perception is shown in Figure 1. This implies an enormous information compression because it suffices to describe the order parameters instead of all the components. Perception of these patterns is mutually exclusive—it is impossible to perceive both at the same time. control parameters such as task affordances can . the relation between order parameters and components is based on circular causality. Moreover. Bistable Gestalt array (Rubin. Synergetics shows that the behavior of the system close to instability points is described and determined by order parameters. Either two (black) faces or a (white) vase is recognized (for details. What are viewed as noses in the domain of one pattern become merely background as soon as a vase is perceived. the—in general few—order parameters determine the behavior of the many individual components. According to synergetics. a bistable Gestalt array. 1921). Quite often order parameters show very simple behavior. Rubin’s (1921) well-known vase-face Gestalt. bi or multistability. even generate the order parameters. that is. such Gestalt flips are analogous to phase transitions in dynamic systems theory. for instance. see the text). transitions between the patterns occur frequently. In this figure two order parameters compete with each other: perception of the pattern face(s) or perception of the pattern vase. On the other hand. Under neutral circumstances. a system can acquire different states under the same external conditions.88 modeling complex systems Figure 1. Both order parameters organize the components (the black-and-white details of the display) in a specific fashion. in this way. Thus.

Words of a somewhat ambiguous quality are presented (see the left column of Figure 2). cells 1–4) that give rise to the perception of either chaos or order. depiction of respective potential landscape (for details. The stability of the percept chaos resulting from this stimulus is represented by the potential sink depicted in Figure 2 (see the right column next to cell 1). The state of the system.89 Approach to Psychopathology 1 "chaos" 2 "order" "chaos" 3 "chaos" "order" 4 "order" Figure 2. Here again. for example. the percept chaos emerges as the order parameter. Figure 1 is presented in a series of similar displays together with the instruction to determine the gender of faces in the pictures. then. Right column. however. the vase pattern is hardly ever perceived. altered gradually in the sequence of cells 2. If. The stimulus is. see the text). still read chaos in cell 2 and cell 3 (the system remains in the respective attractor of the potential land- . and 4. symbolized by the black circle. in cell 1 the word chaos is quite reliably identified. the qualitative pattern recognized instantaneously in the stimulus of cell 1. 3. Most viewers. After Tschacher (1997). Sequence of Gestalt-like written words (left column. A further essential property of pattern formation via self-organization can be illustrated with the help of Figure 2. modulate the emergence of perceived patterns. rests in the minimum of a potential landscape (the “attractor” of the system).

hysteresis is an additional fingerprint of attractors governing the perceptual system. and often does. Interestingly. the attracting capacity of the Gestalt chaos has given way to the Gestalt order. As we cannot go into more detail here. In the movement sciences. . Singer & Gray. Mahoney. the phase transition occurs in cell 1 instead of cell 4. for instance. 1991) and in theoretical psychiatry (Ciompi. Varela. we have shown that Gestalt perception may be analogous to pattern-formation processes described in the framework of dynamic systems theory. in psychotherapy research (Grawe. In neuroscience.90 modeling complex systems scape). the circle dwells in the potential minimum on the right-hand side) and passes through cells 3 and 2. 2003). cognitive coordination and movement coordination are modeled along such lines (Kelso. it is found that the timing of the phase transition may depend on sequence. 1995. 1999. when the stimulus sequence is reversed in experiments with several subjects. We will briefly explore both topics in the rest of this section. In this way the stability of emergent patterns can be explored: a Gestalt may. Arnold. it must suffice to point to some active fields of research where pattern formation is at the center of investigations. a phase transition occurring only in cell 4. This phenomenon of delayed phase transitions depending on past sequences is called hysteresis. therapeutic strategies have been developed on the background of dynamic systems concepts. Self-organized patterns have attractor-like dynamic properties. 1997). remain unchanged in spite of alterations of the actual stimulus. In the present context. In clinical psychology and psychiatry. Here. a phase transition has taken place. We claim that these are pivotal processes found throughout cognition and action. 1995). Hysteresis is usually viewed as a hallmark of nonlinearity in system dynamics. In summary. for example. Patterns of the mind emerge from the complex mind/brain system via self-organization processes—this general premise puts great emphasis on the connectivity among the components as well as on the driving of systems by environmental-control parameters. & Taub. Witte. When the state of the system is started from order (see cell 4. Braun. 2004. synchrony in transient aggregates of neurons in the cortex is seen as the hallmark of the self-organization of cell assemblies (Miltner. and the state of the system has been qualitatively changed.

” The research program of connectionism is. such as the perception of Gestalts and the fingerprints of phase transitions introduced above.91 Approach to Psychopathology connectivity There are two general strategies employed to investigate neurocognitive systems with respect to their interaction and self-organization. the historical roots of contemporary emergentism and of complexity theory lie in Gestalt theory. are the components of (neuro)cognition? Defining an elementary cognitive act is notoriously a nontrivial task in phenomenological psychology. top-down and bottom-up. at least in principle.g. for example. In focusing on the connectivity of systems components.e. have remained rather hypothetical. Attempts to define the components of behavior in a cognitivebehavioral fashion. Gestalt psychology insisted on aggregates possessing qualities that are absent at the level of components. however. What.. psychology departed from associationist psychology—whereas associationism stated that properties of aggregates can be derived from the properties of their components. We have argued that the synergy (i.. In a biopsychological framework. Köhler. however. observable. Connectionist models are good examples of emergent complex systems (e. The top-down strategy begins with observable (“macroscopic”) phenomena. . the task appears much easier—the activity of a single neuronal cell in the brain is a good candidate for a component because it is well defined and. The crucial point about the bottom-up approach is that we must clarify the nature of the components. Ensembles of neurons are formed in concordance with Hebb’s law: “Cells that fire together wire together. thus. by which one can easily be lead into the shallow waters of introspectionism. The bottom-up approach first identifies components of a system and then bases the pattern emerging from the system on a description of the components.. the working together) of many components leads to the emergence of macroscopic patterns. The comparable term coined by the Gestalt-psychological tradition is nonsummativity. based on the connectivity of single cells. Therefore.g. Haken. The Gestaltist holistic position claimed that aggregates are not identical with the sum of their components (e. as conditioned stimulus-response associations. With this core tenet of Gestalt theory. 1920). 1987). we would adopt a bottom-up approach.

are associated with specific motivational and affective deficits or excesses. often have fluctuating levels of negative symptoms. Gonzalez. in addition to the obvious affective disorders. can be seen as a misconnection disorder characterized by altered interactions among brain regions. A related approach views the ion channels of glutamatergic synapses as responsible for the “loosening of associations” that Bleuler (1911) identified as the core symptom of spectrum disorders. in experiments on inattentional blindness (Simons & Chabris. 1998) predicts that the circuitry between areas of the (especially frontal) cortex. 2003) and decreased “binding” capabilities (Garcia-Toro. such . some order parameters may never be generated. task affordances and attention/vigilance determined what was being perceived. Such energetic driving of cognitive systems is traditionally termed motivation. or in situations with competitive motivational parameters. Schizophrenia. This has been demonstrated. The cognitive-dysmetria hypothesis (Andreasen. the thalamus. for example. In the previous two perception examples (Figures 1 and 2 above). for instance. for example. The prototypical courses of schizophrenia. Blanco. we pointed out that the driving of the system by control parameters constitutes a necessary condition for self-organization to occur. even if they appear obvious to unbiased observers. Gestalt perception changes (Phillips & Silverstein. which may result in a host of cognitive impairments. 2001). control parameters In this section. Most psychopathological states. the emphasis is again on changed neuronal connectivity. we introduced the synergetic theory of pattern formation. 1999) where observers were found to be functionally and consciously blind to certain items in a visual scene when their attention was focused on an alternative task. Paradiso. In addition to the connectivity of many components in a complex system. and the cerebellum is unbalanced in schizophrenia spectrum disorder. & Salva. even if this task was performed using the same visual stimulus material. for example. Here. & O’Leary. In the absence of motivation.92 modeling complex systems Several issues in psychopathology have recently been approached with a focus on connectivity.

93 Approach to Psychopathology as flat affect and emotional withdrawal (see the next section). We think that this may turn out to be an encompassing “big picture. Thus. that is. We argued that. we will examine the evidence that has been accumulated and put forward ideas about how one might proceed in this line of research on psychopathology. when applied to cognitive systems. explorative studies of the dynamics of psychopathology are presented. 1994). in turn. it is to be expected that the motivational driving of the cognitive system in such patients may undergo marked changes during the course of the illness. So far we have concentrated on the “control side” of the system-environment coupling. exert an influence back on the control parameters. this principle of efficiency or “optimality” in reducing environmental gradients was a key to understanding intentionality in cognition.” from a natural science perspective. These studies used various approaches to the dynamics of psychopathology. In the following. Admittedly. motivation interacts with cognition in a manner analogous to the driving of complex systems by gradients of free energy (also called exergy in ecology. this big picture must be filled out with empirical findings and scrutinized for its predictive value. We conclude the theoretical introduction to this chapter by mentioning the interaction of (cognitive) pattern formation and (motivational) control parameters. In the remainder of this chapter. Schneider & Kay. however. how control parameters affect the generation of patterns. Empirical Research in Psychopathology: From Description to Testing Dynamic Hypotheses Understanding phenomena by observing change is a common and often successful path to scientific insight. As we have claimed above and discussed elsewhere (Tschacher & Haken. The study questions and corresponding methodological . In this theoretical section we have sketched a formal systems model for psychopathology. in press). Yet this reflects only one side of the coin of interaction: it is self-evident that the patterns. emergent patterns are consistently found to present the most efficient ways to reduce the gradients that caused their emergence in the first place. of how cognition works. In such systems.

” In other words. 1999). Constant changes in symptoms have. 1994. trajectories can be traced back to underlying attractors that describe the influence of forces that act on the evolution of the process variables (Figure 3). A second level of analysis investigates “how things interact in time. Systems theory provides mathematical models for the evolution of mental disorders (Globus & Arpaia. It is important to note that these different levels of analysis are complementary rather than mutually exclusive. From this perspective. Finally. a third level of analysis is related to the emergence of patterns—nonlinearity. Both descriptive and more refined analytic approaches are necessary to illuminate self-organizing systems. the resulting longitudinal patterns have for a long time been important categories. 1998). attractors. Systematic interrelations result in stable patterns of a system. 1999. temporal evolutions are termed trajectories. Tschacher. the interrelation of the systems’ variables is analyzed. On a first level of analysis. 2000). As an example.94 modeling complex systems approaches can be summarized under three levels of analysis. & Giel.” The results on this level are descriptions of longitudinal developments. On a descriptive level. This type of analysis is mainly descriptive. instances more than a trait of a person. several typical temporal patterns of psychopathology can be distinguished (Kupper. level of analysis 1 : trajectories and attractors What is commonly termed psychopathological impairment is in most. “chronic” forms of mental disorders show considerable temporal stability of psychopathological impairments. the study questions relate to “how things change in time. Except for the case of purely stochastic dynamics. that is. called trajectories in the language of dynamic systems theory. been described as a hallmark of mental disorders (Wittchen. instead. The corresponding attractors are . Examples of prototypical dynamics are trajectories that tend to a stable constant state. degrees of order. Nienhuis. and pattern formation. & Schuster. even in chronic mental disorders. Lieb. Kupper. if not all. Pfister. Nevertheless. By definition. 1997). Slooff. “insidious” versus “acute” onset of disorders has been related to different prognoses of schizophrenic disorders (Wiersma.

Prototypical dynamics and attractors. Figure 4 shows the symptom processes of four exemplary pa- . called points attractors. To date. however. In a study of symptom trajectories. whereas the cross-sectional structure of schizophrenic symptoms has been studied extensively. Finally. A novel time-series approach was used to identify initial phases of response and other descriptive features of the trajectories. little is known about the development of symptoms during acute episodes. Another prototypical dynamics is characterized by oscillations: limit cycle attractors. 2002). 2001).95 Approach to Psychopathology Prototypical Dynamics and Dynamics Convergent Cyclic Chaotic Attractors Time Time Time Attractors Point Limit Cycle Chaotic 0-dimensional 1-dimensional >2-dimensional Figure 3. 46 schizophrenia spectrum patients (18 females. Psychopathology in acute schizophrenia. is often characterized by dramatic and repeated changes in symptoms. unpredictable yet nonrandom fluctuations are produced by certain nonlinear dynamics that have been discovered only recently (Rössler. In such a way psychopathological systems may be classified as exemplars from a “zoo” of prototypical attractors arising from various linear and nonlinear dynamics. mean age 24. research on the dimensions of psychopathology in schizophrenia has been mainly cross-sectional in nature (Peralta & Cuesta. for example.7 years) were examined daily during an average treatment period of 104 days (Kupper & Tschacher. 1976). Correspondingly.

These factors summarized the temporal structure of symptom evolution throughout the psychotic episodes. Symptom trajectories of psychotic episodes. termed psychoticity. excitement.96 modeling complex systems Figure 4. The examination of a subsample of 19 patients with relapse indicated a . tients.” consisting of trends. depicting the positive symptoms. and withdrawal. and durations of these trajectories that were assessed in two periods of treatment. anxiety and tension. They tended toward a lower level of positive symptoms and showed a less prominent response to treatment. and negative symptoms together with depression observed in patients. (4) enduring negative symptoms. and (5) duration of response regarding psychoticity. Each process consisted of three trajectories. the initial improvement phase and the ensuing stabilization phase. Compared to patients with an acute schizophrenia-like psychotic disorder. The time-series method used in this study allowed the identification of “dynamic factors. levels. schizophrenia and schizoaffective disorder patients ranked higher on factor 4 (enduring negative symptoms). (2) duration of nonspecific response. Five such dynamic factors were found: (1) overall level of positive symptoms. (3) slope of response in all symptom domains.

and (5) unstable at a low level of functioning.97 Approach to Psychopathology prolonged duration of initial treatment response regarding psychoticity. and variability of each trajectory were calculated. common process. Examples from groups 1 and 4 are shown in Figure 5. and measures of self-concept. whereas levels of symptoms are. few studies have examined individual courses by means of repeated and frequent observations. 2000). tending toward a slight descent. (4) steep descent of functioning. in fact. these results suggested that the different aspects of psychopathology might be related to some single. trend. often independent cross-sectionally. At program termination pronounced differences were found among the groups . Time-series regression was applied on weekly behavioral ratings of psychosocial functioning. locus of control. their change within a given patient was often strongly interrelated. Therefore. The five groups of patients with different trajectories varied at intake with respect to psychopathology. Compared to a merely cross-sectional view. Although considerable research has been undertaken on psychosocial treatment and rehabilitation of patients with chronic schizophrenia. The mean. (3) middlelevel functioning. (2) fluctuating at a middle level. These results supported the validity of this approach for the description of symptom trajectories. this approach provided a dynamic analysis of the dimensions of schizophrenic psychopathology. cognitive dysfunction. Moreover. provided by the nurses’ observation scale (nosie). The trajectories of psychopathology found suggested that. A more dynamic view of rehabilitation was thought to disclose patterns of response useful for both understanding and treating symptoms and disabilities associated with chronic schizophrenia. Cluster analysis revealed five groups of trajectories: (1) stable at a high level. In this study the symptom change throughout psychotic episodes showed distinct temporal features. it was possible to separate the stable and the variable components of psychopathology in the patients. In schizophrenia meaningful temporal patterns in psychopathology are not restricted to acute phases of the disorder. In a study of rehabilitation (Kupper & Hoffmann. courses of psychosocial functioning during intensive vocational rehabilitation of schizophrenia patients were examined. and coping. In this exploratory study time series of 35 schizophrenia outpatients participating in a vocational rehabilitation program were examined by a novel quantitative approach by which dynamic patterns were identified.

e. it is observed that psychotherapy can initially amplify the differences among clients. The time-series methods used in the studies discussed above allowed the description and clustering of such evolutions but gave no indications as to the attractors and patterns underlying them. these trajectories arose from patients’ characteristics interacting with environmental constraints that were induced by the introduction of intensive rehabilitation with increased levels of stress. thus. It should be noted that. these few examples showed that the descriptive level of “how things change” in time provides information that is complementary to traditional research approaches.. For this type of exploration additional methods must be applied. leading to a potential destabilization of the system. From a systems-theory point of view. Cross-sectional analysis cannot explain this rich source of variation. the driving forces of longitudinal evolutions of psychopathology. in vocational reintegration. environmental-control parameters). On the grounds of dynamic systems theory. This study exemplified that a broader use of dynamic designs can substantially clarify the variety of reactions of patients to psychosocial interventions.98 modeling complex systems Figure 5. Psychosocial functioning in two vocational rehabilitation courses of schizophrenia patients. This resulted in a transfer either to a higher level of functioning or to a deterioration in functioning. Similar patterns are known from other forms of psychological intervention. we may suppose that the diverse courses described here have resulted from attractors. generally. The different trajectories can. The task affordances related to the rehabilitation environment acted as gradients (i. As an example. be understood as typical pathways linking patient characteristics to rehabilitation outcome. methods that .

4 years. In Figure 6 the arrow starting from withdrawal at day t 1.65 to . excitement. The interrelations among different components are expected to reveal information about underlying mechanisms that themselves are not directly observable. in this patient high withdrawal one day entailed increased excitement on the following day.44 (the T-value of 2. var models the day-to-day interrelations between symptom factors on the basis of the data of a single course (for an example. 2002). In a study of psychotic episodes (Tschacher & Kupper. level of analysis 2 : the interrelation of system variables On this level the temporal interrelations of components of a system are analyzed. see Figure 6). the goal of these analyses was to model mechanisms that are supposed to fuel the courses described in the previous section. The symptoms were monitored by regular daily staff ratings using a scale composed of three factors: psychoticity. 64% males) of a community-based acute ward were examined to identify dynamic patterns of symptoms and to investigate the relations between these patterns and treatment outcome.3. neither in cross section nor in longitudinal monitoring. Each of the 84 symptom trajectories was analyzed by time-series methods using the vector autoregression (var) approach (Lütkepohl. pointing toward excitement at day t. 1993). mean previous admissions: 1. . that is. Patients showed moderate to high symptomatic improvement documented by effect-size measures ranging from . This notion can be tested empirically on symptom courses by utilizing methods of dynamic systems research. The horizontal arrow in Figure 6 (withdrawal → withdrawal) indicates positive autocorrelation of withdrawal.94 is significant at the 1% level). Examples of such data are shown in Figure 4 above. This var parameter may be labeled withdrawal → excitement. expresses a positive association between these two variables with a weight of . and withdrawal. increased withdrawal on any one day predicts increased withdrawal the following day.99 Approach to Psychopathology provide a closer look at the mechanisms of change leading to observable dynamics. In other words.92. the symptom courses of 84 schizophrenia spectrum patients (mean age: 24. Thus. such as time-series analysis.

Two var parameters were found to be associated significantly with favorable outcome in this sample: withdrawal → reduction of psychoticity as well as excitement → increase of withdrawal. It may be assumed that this illuminated mechanisms that drove the symptom evolutions in this sample. Individual time-series model of a patient (see the text). we combined this analysis of schizophrenia courses with a comprehensive assessment of neuropsychological functioning and psychopathology (Spaulding et al. and anxiety-depression.100 modeling complex systems Figure 6. negative symptoms. * p < . As a further step in a consecutive unpublished study. as hypotheses about the underlying attractors of the symptom dynamics..05. be regarded as preliminary assumptions. The rating scale was composed of three factors: positive symptoms. These findings were interpreted as indicating how patients cope with psychotic episodes. In the Schizophrenia Process Study. or tep). The courses were analyzed by time-series methods using var to model the day-to-day interrelations between symptom factors. A 10-item scale for daily symptom assessment was applied (Today’s Evaluation of Psychopathology.01. however. Comprehensive neuropsychologi- . Multiple and stepwise regression analyses were then performed on the basis of the parameters of var models of all 84 patients. These findings must. ** p < . 1999). symptom courses of 114 schizophrenia spectrum patients were observed using daily ratings of psychopathology (mean duration of observation: 88 days).

and an increase of outward cognitive orientation. On a first level of analysis. reduce the levels . neurocognition. and cognitive orientation. in fact. The results of this study. “inward” centered). on average. notably the outcome domain reduction of social anxiety. that is. Again. The interrelations between positive and negative symptoms were associated with outcome.6 days). The time-series models predicted the treatment effect.101 Approach to Psychopathology cal assessments were performed at the beginning and at the end of the observed courses. and outcome were calculated. Time-series modeling (level of analysis 2) showed that. From the results it was concluded that crisis intervention should focus on having patients increasingly engage in outward cognitive orientation. the expected linear trends were found pointing to an improvement of mood. 2002). If positive symptoms entailed negative symptoms. the remediation processes after psychosocial crises were modeled. A sample of 38 inpatients who were assigned to treatment in a crisis-intervention unit was monitored in order to study the interrelations of process variables. a reduction of tension. tension. var models of the process data were computed to describe the prototypical dynamic patterns of the sample and explore process-outcome associations. which were assessed three times a day throughout hospitalization (on average. Preliminary results (Kupper. again. & Hoffmann. outward cognitive orientation entailed improved mood (see Figure 7). The focus of this study was on “cognitive orientation” (“outward” events centered vs. outcome tended to be unfavorable. The process data consisted of patients’ self-ratings of the variables mood. Associations between the day-to-day patterns of symptoms. yet were unrelated to the domain of symptom reduction. over 22. We included patients with diagnoses such as adjustment disorder and an affective disorder who were often hospitalized with a status of suicidal behavior or ideation. In a study of crisis intervention (Tschacher & Jacobshagen. Outcome of crisis-intervention treatment was evaluated by pre/postquestionnaires. In cross-sectional studies different symptom domains as well as symptoms and neurocognition are often found to be independent. suggested that the various symptom domains as well as symptoms and neurocognition are. Tschacher. 2004) revealed specific relations between longitudinal symptom patterns and both outcome and neurocognition. associated if longitudinal patterns of symptoms in individual patients are examined.

34* TENSION .01.97** COGNITIVE ORIENTATION . and relationship problems. Therapy outcome was evaluated by pre/ postquestionnaires and direct measures of change. whereas the therapeutic-bond factors did have less of an effect on the dynamics (Figure 8). The process data consisted of therapists’ and patients’ session reports. eating disorders. The dynamics-outcome findings showed that direct measures of change were associated with a spe- . *** p < . ** p < . The dynamic patterns of client-centered therapies differed from other modalities. and activating the resources of patients. anxiety and phobic disorders.102 modeling complex systems Whole Sample (N=38) t-1 MOOD .001. client-centered.74** t . Similarly. A sample of 91 courses of dyadic psychotherapy using different treatment modalities was analyzed in order to study session-by-session dynamics. This therapeutic approach should support stabilizing mood. * p < . the largest diagnostic groups being adjustment disorders. 2000).05. Baur. of self-focused attention. schema-theoretical cognitive psychotherapy). After data reduction by principal-component analysis. It was found that the factor patient’s sense of selfefficacy/morale governed the observed dynamics of the sample. “driving forces” of change were extracted from psychotherapy courses (Tschacher. reducing anxiety. linear time-series models of the resulting factors were computed to describe the prototypical dynamic patterns of the sample and of the modality subsamples (cognitive-behavioral. & Grawe.90*** Figure 7. The presented problems generally belonged to the spectrum of neurotic disorders. Interrelations among system variables in crisis intervention.

The observation period lasted 3 weeks.103 Approach to Psychopathology Figure 8. 2003). Clinical observations and recent findings suggested different acceptance of morphine and heroin by intravenous drug users in opiate-maintenance programs. an average of 70 injec- . we assumed different mechanisms of pharmacological action underlying the manifest effects. Interrelations among process factors in psychotherapy. supported by other process factors.1 years) were randomly allocated double-blind to the substance groups. The average daily dose per participant in the heroin condition (n = 17) was 491 mg. In a psychopharmacological study (Tschacher. Thus. Haemmig. The interrelations between both types of effects were determined to test the hypothesis of a differential mechanism of action.7 years.01. Such programs are conducted nationwide in Switzerland. ** p < .05. Thirty-three patients (5 females. This time-series approach was also utilized to uncover psychopathological changes and side effects related to different substances employed in psychopharmacology. mean age 30. * p < . 28 males. cific process pattern in which the patient’s sense of self-efficacy was. We measured the desired and adverse effects of high doses of injected morphine and heroin in patients included in the program. & Jacobshagen. especially how desired and adverse effects of both drugs interacted. mean duration of previous street heroin use 10. we postulated that this different acceptance may be caused by differences in the perceived effects of these drugs. in the morphine condition (n = 16) 597 mg. in turn.

This may explain the better acceptance of heroin in opiate-assisted treatment of intravenous drug patients. this second level of analysis allowed for the identification of interrelations among process variables of many different origins and in various time frames. Ratings were summarized into the factors euphoria and adverse effects. These results pointed to different mechanisms of action of the two opioids when the perceived drug effects are evaluated in a field setting. such transformations of systems leading to the formation of new patterns are a striking feature of self-organizing systems. We view such interrelations as core attributes of dynamic systems that give rise to the patterns and attractors described above.104 modeling complex systems tions was received.g. As described in the introductory section of this chapter. Some premises for the application of level of analysis 2 methods must be regarded.. The next section therefore presents studies of nonlinearity and pattern formation in psychopathology. In addition to the results concerning the studied sample. The interrelations of systems variables can be studied under naturalistic conditions. The monitoring units may be fixed (e. that the approach assumes stationarity of the processes. at each injection of a substance). using the var format described above. daily ratings).. In general. Time-series models were computed for each participant on the basis of the factor scores.g. the finding of a generally higher level of adverse effects in morphine was replicated. various aspects of drug effects were recorded systematically. For this reason we used several techniques in the aforementioned empirical studies to ensure stationarity. transitions in order and pattern formation were analyzed. After each injection of either substance. Additionally. whereas adverse effects of morphine preceded subsequent lower euphoria. we proposed the method as a promising pharmacological tool for the comparison of substance groups other than opioids. at each psychotherapy session). or a combination of both (e.. notably. A highly significant difference between the substances was found in the interrelation between euphoria and adverse effects. . giving this approach high heuristic value and ecological validity.g. On a next level of analysis. detrending of courses prior to modeling. Adverse effects of heroin preceded higher euphoria. for example. natural (e.

Order and Disorder in Complex Systems We studied the hypothesis of pattern formation in a variety of data sets in clinical psychology as well as in acute and rehabilitative psychiatry. Landsberg order was used as described in Banerjee. and Maze (1990. both computed on the basis of the variance-covariance matrix of the p t window (p = . and if phase transitions between stable patterns occur. an increase of order (to be defined below) should be observable. Using the method of therapy-session reports. 1999). see Shiner. data on 33 variables per weekly session were collected. Sibbald.105 Approach to Psychopathology level of analysis 3 : nonlinearity and pattern formation In this section we will address some of the more specific predictions of the dynamic systems perspective proposed in the first section. 3 client centered. & Landsberg. Second. The hypothesis of order increase by pattern formation rested on our conceptualization of the therapeutic alliance as a complex psychological system that is embedded in an environment of various driving parameters. 40–90 weekly sessions) were investigated. In this study 28 courses of dyadic psychotherapy (10 behavioral. nonlinearity. as soon as self-organization structures a complex system. one may determine whether systems evolve in a nonlinear manner by detecting signs of chaotic dynamics and. one may explore the stability attributes of self-organized patterns and how they are linked with psychopathological symptoms. Scheier. all individual sessions of each therapy were assessed by the therapist (a 14-item questionnaire with items such as: “I have the impression that the client came up with what really bothers him/her”) and likewise by the client (a 19-item questionnaire with items such as: “Today I felt comfortable with the therapist”). Davison. Third. If the notion of pattern formation by self-organizing dynamics is true. normalized by the potential entropy. For each therapy course. the following fingerprint phenomena should be expected. A systematic study was performed in the field of psychotherapy (Tschacher. & Grawe. more generally. This measure estimates the degree of order in a multivariate data set by calculating the ratio of the actual entropy. The multiple time series generated by this monitoring procedure served as the foundation of order analysis. 1998). 9 heuristic. 6 schema oriented. First.

These associations pointed to order and order formation as being functional for good treatment results. One minus this ratio then provides the measure of order. The same methodological approach to order formation as in the previous study was . In these data the assessment of order was based on daily symptom ratings using the instruments introduced above and on weekly evaluations of progress in a psychiatric vocational rehabilitation program (Kupper. In a second step of this study we addressed a question that follows naturally from this finding: Is there a connection between order and therapy outcome? The general impression achieved by the outcome study was unequivocal: Order of the therapy system as well as increase of order were related to positive outcome of therapy. In the study of acute treatment we analyzed psychopathology courses of 21 patients (mean age 24 years.. 1999). order and order formation were calculated using the measure Landsberg order. t = session number). Results clearly supported the hypothesis of a significant increase of order during the course of psychotherapy in a comparison of first versus last sessions of each therapy course. order and order formation in courses of 30 participants (mean age 27 years. We computed this measure for the 20 first and the 20 last consecutive sessions using the 33 items recorded in therapy-session reports. Order generally increased throughout the treatment of psychotic episodes. Daily ratings of psychopathology were performed. 1997) and to study rehabilitation processes in psychiatry. 10 of 21 female) during a psychotic episode. 2000). As in the study reported in Tschacher et al. 20 of which were correlated with Landsberg order in a statistically significant way. In the vocational rehabilitation data (Kupper. The results of this study showed that order and order formation correlated negatively with the overall level of psychopathology and with the final level of psychopathology at discharge of patients. Psychopathology of the psychotherapy patients was found to be connected to increase of order. (1998). 10 of 30 female) with schizophrenic disorders were explored. 1999). This finding was corroborated using an extended data set of a later study (Tschacher et al. Order was also computed to investigate the course of schizophrenia in acute treatment (Kupper. Tschacher. We monitored 41 outcome measures. & Hoffmann.106 modeling complex systems items of session reports. Improvement of the self-image of patients and a decrease in social anxiety were linked with both order and increase of order.

the majority of results in these data were consistent with what we labeled order effect (level of Landsberg order correlating to favorable outcome) and ordering effect (increase of Landsberg order correlating to favorable outcome. Overall. job placement. meaningful differences between studies are evident from Table 1. . family interventions. however. Generally.4–0. This program for vocational rehabilitation included vocational training. here regarding vocational reintegration. from both participants’ and supervisors’ perspectives. appear to start from a stable entry point. and support by job coaches. The primary outcome was level of vocational reintegration after the conclusion of the program. The results showed that order and order formation were again related to a positive outcome.7) Middle phase: positive outcome (level of integration) Acute Treatment Initial phase: more negative symptoms and depression (outcome) Middle phase: less confusion and tension (outcome) Final phase: no correlation Yes Initial-middle: less overall tension. Summary of Order and Order-Formation Results from Studies of the Treatment and Rehabilitation of Schizophrenia Patients Vocational Rehabilitation Initial phase: no correlation Correlates of order (r approximately 0. see Table 1). The increased stress induced by these demands does not allow a general ordering effect but leads instead to largely varying developments among patients.107 Approach to Psychopathology Table 1. The assessments were weekly ratings of behavioral and emotional indicators of functioning. social skills training. Moreover. Rehabilitation programs with their higher demands. fear. Acute treatment seems to consist in a transition from a disordered acute state to a more stable state toward the end of treatment. and ambivalence Initial-final: less negative symptoms and depression Ordering effect present Correlates of ordering No Positive outcome (level of integration) used. No general increase of order was found.

These investigations were compatible with the hypothesis that schizophrenia may be characterized by chaotic evolutions. thereby pointing to nonlinearly generated dynamics of the observed data. therefore. that the decay of predictability typical of deterministic chaos was present in the larger . 1997).” “linearly autocorrelated stochastic process. The review of Paulus and Braff (2003) lists the recent dynamic approaches used to investigate the temporal architecture and the disease process underlying schizophrenia. In a study of daily ratings of psychopathology in 14 schizophrenia spectrum patients where observations were possible for extended periods of time (between 200 and 760 consecutive days). It was. Then we applied a series of bootstrap tests that implemented a hierarchy of potential process attributes. Visual inspection of forecasting accuracies showed.” and “linear stochastic process with identical power spectrum. The analysis suggested that 8 of 14 schizophrenia patients presented nonlinear dynamics. & Hashimoto. however. Fingerprints of Nonlinearity and Chaos One of the hypotheses that have been derived from dynamic systems theory is that seemingly random temporal behavior may be generated by relatively simple deterministic nonlinear systems. 1997). Whenever significant differences occur. Therefore. Scheier. To do this. we may conclude that the dynamics of the observed time series was not sufficiently explained by the process attributes. we first estimated the forecastability of the time series of each individual patient. Chaos theory. one cannot assess the chaoticity of the schizophrenia courses we studied. in these studies order reflected the stability of the mental system of the patients.” Finally. especially schizophrenia (Ciompi. the actual forecastability of the observations was compared to the forecastabilities of the bootstrap time series.108 modeling complex systems it can be hypothesized that. such as “random. Yet direct proof of deterministic chaos in relatively short empirical time series is unattainable. we modeled the attributes of the disease process (Tschacher. suggested that nonlinear systems might provide explanations for the often unpredictable trajectories of psychopathological disorders. therefore. the tacit undertone of the word disorder (meaning mental illness) seems justified. has pointed out how such systems may provide both stability (attractor-driven dynamics) and innovation (sensitivity to environmental changes). even when we define order in the strict mathematical sense. a branch of systems theory.

1952) views the evolution of schizophrenia by a succession of stages: everyday cognitive patterns may be impaired prior to the outbreak of psychosis (derealization as Gestalt loss) and may then be resurrected in a bizarre way (delusions and hallucinations as malfunctional Gestalt re-formation). depression. A number of empirical studies have.g. in these early stages stimulus features are grouped and “bound together. 1958. Matussek.” The large majority of these studies suggested that perceptual organization is dysfunctional in schizophrenia. The tradition of Gestalt psychiatry (Conrad. both behavioral and neurophysiological. 2005). Stability of Cognitive Patterns in Schizophrenia Perceptual and cognitive patterns are characteristically changed during psychotic experience. the prerequisites for the generally very sophisticated dynamic methods are seldom realized in psychopathology data. In addition to chaos representing a quite ephemeral quality of (some) nonlinear systems. We think that the chaos hypothesis of schizophrenia and other diseases (e. focused on the changes in (predominantly perceptual) organization in schizophrenia spectrum disorder (Uhlhaas & Silverstein. 1996) must be treated with considerable caution. ubiquitous phenomenon and will concentrate on fingerprints of cognitive patterns and their stability in the following. These changes in perceptual organization may reflect a more encompassing impairment in the coordination and binding of spatial and temporal information and of information originating from different sensory modalities (Phil- . Pezard et al. Although there is some evidence from a variety of data sources. this allows that chaotic processes may underlie psychopathology in schizophrenia. it is probably not their most important property. We already introduced pattern formation as a functionally valuable and. Existence of chaos may well remain undetected because of the constant fluctuations found in any environment outside a physicist’s laboratory. consequently. These phenomenological studies thus pointed to a variety of alterations in Gestalt perception and cognition indicating that both disorganization and delusional hyperorganization should be found. Cognitive psychology views Gestalt-like processes as essential during preattentional stages of parallel stimulus processing... Together with the rejection of alternative hypotheses (especially linearly correlated stochasticity). at the same time.109 Approach to Psychopathology part of the sample.

Gestalt dysfunction does not necessarily go hand in hand with patients showing decreased task performance. cognitive (disorganization). the stimulus presentation is inherently multistable (Kruse. correlate with symptomatology. Although well in need of treatment. sex. A number of clinical symptoms were linked with Gestalt perception. especially. Thus.110 modeling complex systems lips & Silverstein. in press). negative. we investigated several apparent motion paradigms. 1992) because the same display can give rise to different illusions in the same person under . or that of a rapid oscillation of the two apparent motions. 34% inpatients. Pavlekovic. for example. Clinical symptomatology of the patients was measured by the Positive and Negative Symptom Scale (panss). In a recent study of Gestalt perception (Tschacher. The mean reaction time was increased by 1. most patients were not acutely symptomatic. 81% males. more accurate than controls in counting tasks where grouping of stimuli did not support performance. The impression is always either that of a wandering of the dots in a clockwise or counterclockwise circular direction. Meier. Stadler. In their review Uhlhaas and Silverstein (2005) cited evidence for a correlation of Gestalt dysfunctions with cognitive disorganization as well as with both positive and negative symptoms.7 standard deviations in patients. Circular Apparent Motion (cam) The frames displayed in Figure 9 were presented alternately for 500 ms each.3 years) and 32 control subjects matched for age. and Hyman (1995). and education. Grochowski. excitement. 2003). mean age 27. pointing to their generalized impairment of cognitive functioning. receiving average panss scores below 3 (“low symptoms”). Interestingly. & Junghan. Gestalts in the shape of movement illusions result. Research has shown that alterations of perceptual organization differ between subgroups of patients and. & Gheorghiu. Nevertheless. which consists of the five factors positive. We included 32 schizophrenia patients (66% dayclinic patients. in fact they were. Dubouloz. Place and Gilmore (1980) and others found that schizophrenia patients did not present a generalized deficit when compared to controls. Generally. this schizophrenia group had reaction times markedly decreased compared to controls. For all computations we used the factorization of the panss according to Lindenmayer. and depression.

Stroboscopic Apparent Motion (sam) Corresponding to the previous task. flip events are enforced (initially.e.. respectively. 1995). In the present sample. We monitored the timing of these flip events by having subjects press a button when a flip occurred. . identical circumstances. The perceived durations of respective Gestalts were suggested as a measure of cognitive stability (Kruse & Stadler. the more likely hsam is seen. The Gestalt impression depends on the horizontal distance d. vsam transiting to hsam. the frames displayed in Figure 10 were presented alternately for 500 ms each. then the reverse.111 Approach to Psychopathology Figure 9. as was seen above. Two qualitatively different Gestalt illusions are generated. With d gradually reduced from a maximum value to 0 and then increased again. the perception is of A and B moving to C and D. producing the motion perception. multiple regression showed that durations of Gestalt perceptions were significantly linked to the panss factors. Especially the negative symptom factor was associated with prolonged durations.and the right-hand panels are shown alternately for 500 ms each. The left. Display shown to test circular apparent motion (cam). In the case of vsam. dot A is perceived as moving down to dot D and B as moving up to C. Haken (1996) discussed the size of the hysteresis effect as a prominent measure of Gestalt stability. the hysteresis effect) is a hallmark of phase transitions between stable attractors. The smaller is d. vertical apparent motion (vsam) or horizontal apparent motion (hsam) of the dots. In the case of hsam. Yet there was no difference between the patient and control groups. The asymmetry of the timing of the initial and the reversed flip events (i. hsam giving way to vsam).

a Gestalt phenomenon in the sense of figureground rivalry. The display contains three stationary dots colored bright yellow and a grid of dimly illuminated blue crosses. The subjects are asked to press a key whenever. apparent blindness occurs. The five panss factors explained 48% of the variance in this variable in a whole model test. whereas positive symptoms accelerated the occurrence of flip events.112 modeling complex systems Figure 10. no group effect was observed in our data. It was our premise that this task is.and the right-hand panels are presented alternately for 500 ms each. Gestalt stability measured by the hysteresis effect did not generally deviate in patients. and Sagi (2001): Subjects are instructed to observe the middle of a display such as that depicted in Figure 11. Cooperman. The left. and for the duration that. The phenomenon that most observers perceive is “apparent blindness”—one or all of the yellow dots subjectively disappear from the visual field for a period of time. In other words. in fact. The grid is rotated slowly. The timing of the initial Gestalt flip (vsam to hsam) was. This exchange of figure and ground may entail the subjective perception of subjects becoming blind to the yellow dots. strongly linked with symptomatology. The mib phenomenon is similar to inattentional blindness as investigated by Simons and Chabris (1999). Display shown to test stroboscopic apparent motion (sam). with the result that the yellows dots are put in the background. Again. The blue grid may acquire the function of figure after some time has elapsed. however. The variable of interest in this paradigm was the number of mib . sometimes for several seconds. The negative and the cognitive (disorganization) factors both delayed the initial Gestalt flips. Motion-Induced Blindness (mib) The following phenomenon was described recently by Bonneh.

Lewkowicz.3. Two disks are presented moving from either sides of the screen toward the opposing sides. The five panss factors explained 49% of the variance in this variable. auditory and visual) into a coherent perceived pattern.1 in the control group). vs. 2006). Schuler. There was a tendency in the patients’ group toward fewer mib counts (29. The positive and the excitement factors were both related to increased numbers of mib phenomena (Tschacher. In multiple regression analysis of the patients group. following the same trajectory (Figure 12). Similar .113 Approach to Psychopathology Figure 11. & Junghan. yet this difference did not reach the significance level. they are perceived either as “bouncing” off each other or as “streaming” through one another at the moment they overlap visually. Intersensory Binding The final task of the Gestalt test battery addressed temporal binding of different sensory modalities (here. When the disks approach one another in the middle of the screen. 42. and Shimojo (2003). We used a program developed by Scheier. we found a strong association of number of mib phenomena with symptomatology. phenomena perceived by each participant during a 3-min period. however. Screen display (schematic) for motion-induced blindness (mib).

In the patient group. apparent causality depended on symptoms. a strong relation was found between symptoms and preferred causal attributions. at the time of their “collision” and visual overlap. the Gestalt test battery showed few general group differences between patients and matched controls. There was a tendency toward patients having fewer bouncing perceptions. we found that positive symptoms were linked with Gestalt instability and faster production of Gestalt perceptions. Screen display (schematic) for the intersensory binding task. however.114 modeling complex systems Figure 12. Two disks wander from the sides of the screen toward one another. As a whole. In addition to the optical stimuli. Additionally. The observer is instructed to indicate which of the two possible events (bouncing or streaming) was perceived. More specifically. however. Within the patient group. We report here the findings regarding the probability of bouncing versus streaming perceptions (Tschacher & Kupper. an acoustic-click stimulus is presented with varying temporal latency relative to the time of disk overlap. effects that can explain the weak significance of group comparisons. Which causal relation is attributed to this scenario. negative symptoms and the cognitive disorganization factor with persistence of Gestalts. Attri- . 2006). The positive factor and the cognitive (disorganization) factor of the panss were strongly linked with increased probability of bouncing perceptions. the panss ratings suggested that specific associations exist between psychopathology and Gestalt perception. was unaffected by psychopathology. Psychopathology has opposing effects on perception. Hysteresis. a click sound is given. bouncing or streaming? displays have been investigated in studies of apparent causality by Michotte (1954). This may be attributed in part to the low symptom burden in the patient group.

especially in complex open systems. In this. and the stability of Gestalt-like attractors. We have quantified and described the trajectories and processes ubiquitously found in various fields of practice. consequently. some of the specific hypotheses of the dynamic approach have been tested in the data sets. The main results were these: Studies show a variety of dynamic patterns in courses of psychopathology concerning psychiatric rehabilitation and acute treatment of psychoses. Structure and stability are. especially var models. We concluded that these models were psychologically meaningful because the attractors were related to outcome and/or patient characteristics. The final part of the second section addressed some of the core assumptions of dynamic systems theory applied to fields of psychology. On the basis of the empirical evidence we may claim that the search for fingerprints of nonlinear dynamics has been successful. even more so as some psychopathological phenomena are inherently unstable or show unpredictable and erratic dynamics. . nonlinearity. Such fluctuations have frequently been perceived as a problem for the validity and reliability of quantitative research. We proposed that these models reflect the attractors underlying the observed dynamics. we have demonstrated the validity and feasibility of this approach. seen as the secondary results of dynamic processes that lead to the emergence of attractors. The dynamics-oriented approach views change as primary. such dynamics poses not so much a scientific problem as an opportunity for a novel approach to modeling. a long-standing call for longitudinal studies is met. that are capable of capturing the essence of the change processes. particularly order formation. the resulting attractors and patterns have been detailed. In the second section of this chapter.115 Approach to Psychopathology bution of causality (the “bounce” perception) was a marker of positive and disorganization symptoms. Discussion A common finding in research on mental disorders is that the level of psychopathology usually varies in each individual case in the course of time. For a wide range of settings and problems we applied methods by which the trajectories could be translated into condensed models. With the theoretical and empirical approach presented here.

From these applications of the dynamic approach can be drawn a series of implications for psychiatry and psychology. limitations that are brought about by the demands of time-series methodology. this has significant clinical implications. Therefore. the relation could be affected by unknown third variables that have not been monitored. come within reach as soon as favorable patterns are reliably identified. such as order and stability of patterns. allows the distinguishing of favorable from unfavorable patterns. The aggregation of many single systems.116 modeling complex systems The data again suggested that the phenomena of nonlinearity. The kind of longitudinal assessment that we proposed stands in contrast to the established cross-sectional approach. It is. As time-series methods enable the modeling of single systems. necessary to be aware of the limitations of dynamic methods. such as rehabilitation and psychotherapy. thus. nevertheless. and the theory of cognitive science. during the necessary observation period the underlying causal mechanisms must remain unaltered (the premise of stationarity). as demonstrated repeatedly in the previous section. New methods of online control and the adaptation of interventions in the context of quality control may. especially as regards the fields of therapy. When these limitations . Time-series models are so-called Granger-causal models—as in all correlational models. The most important benefit lies with the possible identification of change mechanisms in therapeutic settings. These demands are the large number of sequential measurements necessary for modeling. reflect causal mechanisms. diagnosis. the effective mechanisms can be assessed directly in the respective fields of practice and in natural clinical situations with uncompromised ecological validity. but need not necessarily. therapy We expect that time-series methods can elucidate mechanisms of action in practice fields of psychiatry and psychology. dynamic approaches should go hand in hand with classic experimental therapy research wherever possible. Another caveat must also be kept in mind: var models give an account of sequential interrelations among the process variables that may likely. are associated with markers of psychopathology and clinical outcome.

this theory is . go beyond psychopathology: How can we address cognition in general on the basis of the dynamic view? Arguments for a novel view of cognition. has largely abandoned. Current dsm-style diagnostics is based on set theory. however. this understanding of cognition deviates from the computationalist approach of cognitive psychology. Bélair. however. is a computationalist notion that cognitive science. Categorization along the lines of prototype theory (Rosch & Lloyd. Glass. First. & Milton. five of nine criteria must be fulfilled. 2003). Dynamics-oriented theory is more encompassing in several respects. especially dynamic cognitive science. Yet the dynamics-oriented approach holds further implications for psychiatric diagnoses. We propose that psychiatric diagnosis must be reformed at its very basis by considering what is known about psychological categorization.117 Approach to Psychopathology are taken into account. implications that have as yet received little attention. one that treats cognition as an emergent property of the mind. 1978) has clear advantages and. the concept of dynamic diseases (cf. 1995. theory Our final remarks. This is in line with what was found above. the dynamic approach has proved fruitful in process-outcome research. for example. much as a computer program would run on silicon chips. Emrich & Hohenschutz. where cognition is an abstract semantic structure running on neuronal hardware. Tschacher & Dauwalder. 1992) deviates from the common structural concept in that (some) diseases are viewed as emergent states and attractors. with a vast potential for further studies of therapeutic applications. an der Heiden. to diagnose borderline personality disorder. As we said in the introductory section. The definition of concepts in terms of a list of criteria..g. have accumulated in the recent past (e. in addition. would be more compatible with the approach favored in this chapter. For example. diagnosis Categorical diagnosis and the dynamic view of psychopathology follow conflicting concepts of disorder.

E. therefore. Being there: Putting brain. (1990). Bonneh. The main demand on cognition is just this—simplifying the world by the extraction of significant patterns on the basis of motivation.. Journal of Theoretical Biology. Cambridge ma: mit Press. Bleuler. dynamic systems theory deals with complexity.. J. Die beginnende Schizophrenie. (1997). Quantifying the dynamics of order and organization in biological systems. The exploration of cognition by abstracting from its very driving forces is increasingly viewed as unfeasible in psychology and neuroscience. Stuttgart: Thieme. Göttingen: Vandenhoeck & Ruprecht.. London: Weidenfeld & Nicolson. an der Heiden. 91–111. N. Paradiso. D. Second. References Andreasen. L.. Die emotionalen Grundlagen des Denkens: Entwurf einer fraktalen Affektlogik.). & Maze. of the utmost survival value. R. D. Bélair. Ciompi. & Milton. 143. two separate sources of complexity must be addressed. and world together again. (1997). Conrad.. 798–801. (1995). & Sagi. Dementia Praecox oder die Gruppe der Schizophrenien. Motion-induced blindness in normal observers. Nature. in cognitive science. Third. “Cognitive dysmetria” as an integrative theory of schizophrenia: A dysfunction in corticalsubcortical-cerebellar circuitry? Schizophrenia Bulletin. (2001). Dynamical disease. S. This integrated view of cognition in context (embodied cognition) is a principal attribute of our model. U. Consciousness. P. (1998).. Glass. Continuous simplification is. Carter. in the dynamic framework. 411. Sibbald. (Eds.118 modeling complex systems in accordance with transdisciplinary systems theories in the natural sciences that study and predict self-organization phenomena. This unitary approach has advantages with respect to the dissemination of both methods and phenomenological findings across disciplines. (1958).. cognitive systems always deal with a complex world. On the other hand. 203–218. A. (1911). via a self-organizational process. 24. Cooperman. K. A. Y.. a world that enforces and affords complexity reduction. body. cognitive systems themselves are always complex.. information processing and cognition is never separable from emotion and motivation. S. Woodbury ny: American Institute of Physics Press. J. L. On the one hand. We have demonstrated that. as only complex systems can. (2002). Banerjee. Leipzig: Deuticke. C. S. S. J. Clark. . R. & O’Leary. produce adaptive pattern formation.

35. B. Grawe. 204–212). Z.. Synergetics: An introduction. J. A. In W.. & E. (Eds.. Tschacher. Haken. Comprehensive Psychiatry. 272. H. (1983). (1994). K. (1992). Haken. Kupper. J. Psychiatry and the new dynamics. In H. 14. Psychological therapy. C. S.). Lindenmayer. Schizophrenia Research. & Hoffmann. H. Kupper. Kupper. Tschacher. M. & Hoffmann... Psychopathology and the binding problem. G. A. Blanco. Berlin: Springer. (1995). Cambridge ma: mit Press. Medical Hypotheses. Self-organization and clinical psychology (pp. Symptom trajectories in psychotic episodes. Berlin: Springer. Pavlekovic. S. Singapore: World Scientific. J. Schiepek.. In W. Die physischen Gestalten in Ruhe und in stationärem Zustand. (2003). Dauwalder (Eds. (2002). Bavaria.).... Brunner (Eds. & Gheorghiu. Poster presented at the Joint Conference on Complex Systems in Psychology/6. H.. and psychotherapeutic interventions. 2–23).). Schizophrenia Research. Z. psychological experiments. Kelso. M. Brunner (Eds. G. W. Kupper. (1996). Garcia-Toro. S. The dynamical systems approach to cognition (pp. Berlin: Springer. 26(3). 352–364. Psychiatric disorders: Are they “dynamical diseases”? In W. (1987). & Salva.). Berlin: Springer. 681–698.119 Approach to Psychopathology Emrich. W.. & Hohenschutz. Computational systems—natural and artificial: Proceedings of the International Symposium on Synergetics at Schloss Elmau. Grochowski. Herbstakademie. Kelso. & Tschacher. Five factor model of schizophrenia: Replication across samples. Globus. Dynamische Modelle für chronische psychische Störungen.. Haken (Ed. B. (1995). H. J.).. Cognitive coordination dynamics. G. Biological Psychiatry. & Stadler. 1987 (pp. Haken. Lengerich: Pabst Science. J. Braunschweig: Vieweg. March). Seattle: Hogrefe & Huber. behavior. 43(4). Stadler. Gstaad. M. (2004). & Hyman. Tschacher. C. Kruse. Gonzalez. W. Correlates of order and ordering in the time course of schizophrenia. P. (1995). W. P. Dynamic patterns: The self-organization of brain and behavior. Z. Berlin: Springer. 718–723. Instability and cognitive order formation: Self-organization principles. May 4–9. Longitudinal symptom patterns in psychotic episodes are related to neurocognition and outcome. 311–318. 102–117). R. Ambiguity in mind and nature. P. G. V. H. H. Principles of brain functioning: A synergetic approach to brain activity. Schizophrenia Bulletin. and cognition. & E. Z. Z. 45–70). & Arpaia. Suppl. H. Berlin: Springer.. Tschacher & J. J. (2001). Self-organization and clinical psychology (pp. Schiepek. Tschacher. Köhler.. (2000).. Synergetic computers for pattern recognition and associative memory. .. Course patterns of psychosocial functioning in schizophrenia patients attending a vocational rehabilitation program.-P.-P. Kupper. & Hoffmann. (1999). (1920).). 229–234. 67(1. 57(6). (1992). Kruse. J. (2004). A. (1997.

S. Nandrino.. 991–999. (1996). How many and which are the psychopathological dimensions in schizophrenia? Issues influencing their ascertainment. A. & Prigogine.. & Scheier. Englewood Cliffs nj: Prentice-Hall. (1954). D. J. T. (2003). Singapore: World Scientific. Davison.. (1991). Scheier. Mechsner... C. & Martinerie.. Biological Psychiatry. M. 397–398.). F.. D. P. Pezard. 397. & Landsberg. (1976).. D. Copenhagen: Gyldendalske. & Silverstein. Newell. J. synergetics. (2003). W. L. M.. Peralta. P. Massioui. 65–138. B.. 39.120 modeling complex systems Lütkepohl. Lewkowicz... . 3–11. (1994). J. Rosch.. G. O. C. Understanding intelligence. (1980).. 69–73. Perceptual organization in schizophrenia. Rössler. Shiner. autonomous agents: Nonlinear systems approaches to cognitive psychology and cognitive science (pp. An equation for continuous chaos. Cognition and categorization.. I. Schizophrenia Research. (2001). (2001). Depression as a dynamical disease. Life as a manifestation of the second law of thermodynamics. Dynamics. J. E. Knoblich. (2003). 57a. J. A. & Lloyd. Archive für Psychiatrie und Zeitschrift für Neurologie. Paulus. J. G. Phillips. (1999).-P. A. Gesammelte Werke. Berlin: Springer. M.. Witte. Braun. Visuell wahrgenommene Figuren. (1977). Hillsdale nj: Erlbaum.. E. 279–319. & Simon. F. Rubin. (1999). J. (1993).. Miltner. C. H.. M. 434–436. G. Human problem solving. Place. Self-organization in non-equilibrium systems. Mitteilung. 409–418. E. E. Varela. Nicolis. Untersuchungen über die Wahrnehmung: 1. L. Michotte. Chicago: Basic. J. Kerzel. 269–285. 414. A. Arnold. Mathematical and Computer Modelling. Mahoney. & Braff. 25–48. E. 53. H. B. Perceptual basis of bimanual coordination. & Gilmore. (1978). (1999). 49–63). & Shimojo. D. Matussek. J. Bern: Huber. Renault. Chaos and schizophrenia: Does the method fit the madness? Biological Psychiatry. V.-F. 19. 49(3). Dauwalder (Eds. E.. Tschacher & J. 6..). 233–244. (Eds. Schneider. (1972). S. Allilaire. Coherence of gamma-band eeg activity as a basis for associative learning. & Cuesta. In W. & Kay. M. C. Convergence of biological and psychological perspectives on cognitive coordination in schizophrenia.-L.. W. 180.. 26. Nature. New York: Wiley. H. S. & Taub. J. F. Physics Letters. (1952). (1921). J. & Prinz. Developmental Science. W. On measures for order and its relation to complexity. Sound induces perceptual reorganization of an ambiguous motion display in human infants. R. M. Pfeifer. P. E.. Nature. Introduction to multiple time series analyses. Journal of Abnormal Psychology. 89.. Human change processes: The scientific foundations of psychotherapy. Behavioral and Brain Sciences. B. Cambridge ma: mit Press..

Reed. The dynamical systems approach to cognition. Schizophrenia Research.. Dynamics. 41.. 28. (1999). M. Psychiatry Research. C. W. & Jacobshagen. Varela. 28. & Hashimoto. Schuler. Singapore: World Scientific. C. W. Biological Psychiatry. 618–632. N. J. Biological Research..). 296–309. & Grawe. Wiersma.-P. Meier. Storzbach.. (2006). R. F. W. W. Gorillas in our midst: Sustained inattentional blindness for dynamic events. Baur.-P. P. J. Singapore: World Scientific. W. & Jacobshagen. N..... Psychology and Life Sciences.. & Lam. 195–215. (in press). Tschacher. Psychiatry Research. D. J. synergetics. S. Tschacher.. Tschacher. Prozessgestalten—die Anwendung der Selbstorganisationstheorie und der Theorie dynamischer Systeme auf Probleme der Psychologie. & Kupper. W. & Grawe.. Göttingen: Hogrefe. S106–S112. Z. W. D. 131.. U. 81. J. Crisis. Tschacher. (2000). Schizophrenia Bulletin. 275–289. W. 555–586. & Junghan. Tschacher. 2. H. J.. (1999). A.. (1999). W. Slooff. 32. & Dauwalder. W. Tschacher.. K.. Nienhuis.. W. M. 127–137. S. N. Temporal interaction of process variables in psychotherapy. (2005). Time series modeling of heroin and morphine drug action. Order and pattern formation in psychotherapy. & Kupper. Analysis of crisis intervention processes. M. (2003). D. Tschacher. 25. F. Nonlinear Dynamics. K. & Giel. R. Perception. (1995). U. D. (2003). W. 81–95. F.. D. W. 23. & Junghan. Psychopharmacology. Dubouloz. C... Time series models of symptoms in schizophrenia. 113. Tschacher. Resonant cell assemblies: A new approach to cognitive function and neuronal synchrony. Scheier. Dynamical analysis of schizophrenia courses. (1997). Scheier. Natural course . 18. Sullivan... J.).. J. Spaulding.121 Approach to Psychopathology Simons. Psychological Bulletin. Tschacher. (1998). autonomous agents: Nonlinear systems approaches to cognitive psychology and cognitive science. Tschacher. Uhlhaas... R.. & Haken. Schizophrenia Bulletin. Tschacher. C. Visual feature integration and the temporal correlation hypothesis. Y. 261–267. C. Cognitive functioning in schizophrenia: Implications for psychiatric rehabilitation. (1998). & Gray. New Ideas in Psychology. (2006). (1995). (2002). Altered perception of apparent motion in schizophrenia spectrum disorder. Tschacher. W. Fleming. Tschacher. (Eds.. Psychotherapy Research. 10. (1997). Perception of causality in schizophrenia spectrum disorder. Reduced perception of the motion-induced blindness illusion in schizophrenia. Haemmig. Singer. 59–67. M. P.. F. (2002). K. Annual Review of Neuroscience.. D. & Dauwalder. 188–193. Intentionality in non-equilibrium systems? The functional aspects of self-organized pattern formation. 428–437. Z. (Eds. W. 165. 1059–1074. (in press). & Chabris. & Silverstein.. Perceptual organization in schizophrenia spectrum disorders: Empirical research and theoretical implications.

H. 122–132. The waxing and waning of mental disorders: Evaluating the stability of syndromes of mental disorders in the population. . Pfister. Wittchen. & Schuster.... 24. Comprehensive Psychiatry. P. R. Lieb. U.122 modeling complex systems of schizophrenic disorders: A 15-year follow-up of a Dutch incidence cohort. (2000). H. 75–85. 41. Schizophrenia Bulletin.

the second who or what does the modeling. is the importance of modeling performance. There are still three key questions regarding modeling. My use of the term is close to that of the online Merriam-Webster dictionary. Even though there is some truth to the notion that practice makes perfect. from a psychological perspective. regardless of the population or the subject material studied. Bandura (1977) defined it in terms of how learning can be facilitated by observing others perform the task in question.Developing ComputerBased Learning Environments Based on Complex Performance Models Susanne P. . Lajoie McGill University A universal theme in my research. The relevance to learning of such modeling is that learners practice deliberately rather than indiscriminately (Ericsson. modeling has been defined in a similar manner. Expertise can be developed when practice is sufficiently deliberate with feedback regarding performance. 2006). which refers to a model as being something set or held before one for guidance or imitation. we also know that practicing the piano for 10 years does not make one a concert pianist. Interestingly enough. The word model can be defined in many ways. The first one is what to model. The benefit of modeling would be that learners do not always have to learn from their own mistakes but can sometimes learn by observing others do a task efficiently and then incorporating the ideas thereby obtained.

body. 2003). 1987. and the strategies that lead to effective problem solutions can all be modeled for the learner. Chi. What to Model In learning and performance contexts it is often the cognitive processes involved in problem solving that need to be modeled. 2003).124 modeling complex systems and the third how to model effectively. who wanted a brain. It seems that the cognitive scientists are often considered “heartless” in the sense that they focus on the mind and how people . spirit connection—where affect and emotions are considered to go hand in hand with the mind (James. For instance. Some answers to these questions are presented. The domain knowledge. Glaser. 1890)—has long been recognized. 2003. The groundwork for examining expert/ novice differences was summarized in Chi et al. self-monitoring and metacognition are key elements of expertise. 2002. who wanted a heart. & Lajoie. are used today. Glaser. and the methodologies for studying expert/novice differences. result in helping learners become more competent. and motivation theory to the Scarecrow. rather. Ericsson. the structure of the knowledge. 1988. followed by examples from my own research that will help contextualize the arguments made in this chapter. Instructional improvements are based on modeling these dimensions of expertise. however. we need to identify what experts monitor in a specific context before we can model these processes for novices. The assumption is that identifying dimensions of expertise can lead to improvements in instruction that will. comparing cognitive psychology to the Tin Man in The Wizard of Oz (Baum. Studies of expertise have been criticized for being too coldly cognitive (Alexander. Many of the principles of expertise discovered then. (1988). we are still struggling as a community to see how these elements interact. using such knowledge and interacting with it with specific feedback. Although the mind. Lajoie. but it is not simply observing things that leads to learning but. ultimately. Lesgold. Lepper (1988) reintroduced the connection when he called for the intersection of cognition and motivation theories. & Farr. 1969). Studies of expertise and proficiency have helped in the identification of such processes in specific contexts and have also led to the discovery of commonalities in such components across expertise theories (Alexander.

& Schauble. and motivational theories in interesting combinations that lead us to a better understanding of human performance (see esp. and social situations are becoming elements of research design (Cobb. where higher learning and cognition are linked to emotion. Pellegrino et al. 2003). learning theories are increasingly more inclusive in that cognition. and team members are part of the situation. diSessa. Burke. 2000. motivation. 1997). Pellegrino et al. Confrey. report on a large body of scientific knowledge about the processes of thinking and learning and the development of competence in specific curriculum areas.125 Developing Learning Environments learn. support a curriculum-instruction assessment triad based on . New models must consider the interleaving of such principles in specific contexts. and Goodwin (in this volume) speak to this issue with reference to developing expert teams in natural settings. and the social context in which learning takes place are considered as interconnected (Anderson. and Glaser (2001) provides insight with regard to what to model across the K–12 curriculum. Mayer. motivation. However. Their chapter addresses the notion of distributed expertise. The intersections of learning. Salas. Stagl. Cordova & Lepper. Chudowsky. Mahoney (in this volume) describes lifelong learning from the perspective of embodied cognition. 32. Alexander has edited a special issue on the topic of expertise in the journal Educational Researcher (Vol. Learning situations in the real world are often social in nature. 1996. modeling competence across the curriculum The National Academy of Science report by Pellegrino. pointing toward a shift to a more inclusive theoretical model that parallels the discussion in the editors’ introduction to this volume of the confluence of social. Examining competency or proficiency within a specific context is the first step in elaborating a model of thinking that can help the less competent become more proficient in a specific domain. & Simon. 8) that points to how researchers study expertise today. which is important in real-world settings. Greeno. No. Lehrer. 2003). Examining learning processes as they occur within “situations” or meaningful contexts goes hand in hand with personal interests when considering what is meaningful. cognitive. Alexander. Reder.

The implications for teaching and the design of instruction and assessment (Pellegrino. 2002) are clearly based on identifying what students know and what they need to know and having them reflect on the processes. Analyses of avionics troubleshooters revealed no ideal solution paths and demonstrated that experts solve similar problems differently (Lesgold et al. a cta of the clinical decision making of surgical nurses revealed that expert nurses reached similar decisions. The same principles would apply in complex learning situations in the real world (see Lajoie. we cannot scaffold learning unless we understand the learning process. actions performed. only then can we model these processes for novice performers. Again. Studies of expertise have provided the foundation for developing models of what students need to know with respect to complex performance across domains. Such studies supported the design of methodologies used today to support our understanding of verbal data (see Chi. Azevedo. Differences were observed in hypothesis generation. 1998). and overall solution paths. hence. actions. In fact. They report that effective learning environments are learner centered and. Ericsson & Simon. Lajoie. and mental models (Lesgold. interpretation of the results. demonstrating that there are many paths to solving problems (Lajoie. 2003). . results of evidence gathering. albeit by different routes. Similarly. Logan. 1993). & Fleiszer. & Eggan. that instruction must provide scaffolding for solving meaningful problems and supporting learning and understanding. planning of medical interventions. heuristics. 1990). 1990). which are central to documenting problem-solving activity. avionics troubleshooting and clinical diagnostic reasoning. Cognitive task analysis (cta) is still an appropriate means of discovering what experts know with regard to their problem-solving plans. Two examples from complex decision-making activities in the real world. reveal that experts can differ in their solution strategies and steps but be guided by common underlying mental models to the correct solution. However.126 modeling complex systems principles about learning and knowing that can assist learners along a learning trajectory within a specific field. 1997. similarities were found in the underlying mental models that guided their problem solving.. cta is an excellent mechanism for documenting an effective problem space that reflects a more inclusive analysis of how individuals who differ in their performance skills go about solving problems.

was designed on the basis of the cta described by Lajoie et al. 1992). Pre and posttutor assessment was done using verbal troubleshooting techniques as well as a paper-andpencil test. However. the average gain score for the group using Sherlock was equivalent to almost 4 years of experience.9 and 53. Similarly. “Sherlock. & Gluck. Statistical analyses indicated that there were no differences between the treatment and the control groups on the pretest (means = 56. Bunzo. a computer tutor for nurses working in a surgical intensive-care unit (sicu).0) performed significantly better than the control group (mean = 58. Lajoie. The tutor teaches troubleshooting procedures for dealing with problems associated with an f-15 manual avionics test station. Such an approach was employed in the avionics troubleshooting example mentioned above. Jones. Two groups of subjects per air force base were tested: (1) subjects receiving 20 hours of instruction on Sherlock and (2) a control group receiving on-the-job training over the same period of time. (1998). & Alley. knowing what to model can lead to better performance. respectively).9) and equivalently to experienced technicians having several years of on-the-job experience (mean = 82. 1992). Pokorny. That is. on the verbal posttest as well as on the paper-and-pencil test. The cta led to the development of a cble. Hence. 2003).127 Developing Learning Environments Documenting such problem spaces gives clear guidance for how we might tutor individuals in their problem-solving decisionmaking strategies. The analysis led to the design of . A study was conducted evaluating Sherlock’s effectiveness using 32 trainees from two separate air force bases (Nichols. Gott. Assessment must consider learning at a higher level of abstraction and a wider effective problem space in which assessment of learners takes place (Lajoie.” that provided a practice environment allowing first-term airmen to become proficient troubleshooters (Lesgold.4. Developing computer-based learning environments (cbles) and intelligent tutoring systems on the basis of cta can validate performance models by demonstrating that learning does occur in the predicted manner and that providing feedback based on such performance models can lead to greater proficiency in trainees. Lajoie. sicun. 2000). the treatment group (mean = 79. & Eggan.2). All this is to say that identifying robust performance models in complex domains can lead to effective instruction for those less proficient in problem-solving performance (Shute.

and who and what should do the modeling abound (Derry & Lajoie. The discussion presented above has described how decisions can be made about what to model to learners. Lajoie. but the subjects tested became more planful in their patient assessments. For example. 1993. A full-scale evaluation of this tutor was not conducted. The increasingly integrated models of learning. the tutor would prompt them about the results of their assessment. with decision trees designed to encourage self-monitoring and comparison to expert problem solvers at various phases of problem solving. Who or What Does the Modeling? Discussions of cognitive performance models. What and whom you choose to model complicate these questions. However. This struggle seems to parallel changing conceptions of learning theories as well as the difficulties associated with methodologies used to instantiate such theories. and motivation will further challenge us in answering these questions. goals. The issue of deciding who or what does the modeling is another matter entirely. how complete they are. whereas in the past they were described . 2000). the issue of computers as modelers is more contentious than that of human tutors. capillary refill. actions. edema. Learning theories today are described in the context of knowledge construction through realistic tasks and situations. plans. they might check pulse rate as well as skin (for swelling. Nurses who worked with sicun were required to post their goals before conducting patient assessments and then specify outcomes of their assessments.128 modeling complex systems a system that encourages self-monitoring of decision-making processes. tutor. Prior to their moving on to a new goal or body system. and temperature). if their goal was to check the patient’s circulatory system for adequate blood supply to the heart. A simple answer to the “who or what models?” question is that both human beings and computers can model. or coach individuals and groups of learners. coloration. one that can lead to choices about how to successfully model. Hence. social processes. Members of the artificial intelligence (ai) and education community continue to struggle with the question of performance models. and outcomes were all built into the system.

thus ensuring opportunities for multiple representations of knowledge (Kozma. Pea. 1996. Jones. Salomon. but. we are not content without robust testable models. computers can still be used as cognitive tools and used in conjunction with human coaches and peers to scaffold the process (Palincsar & Herrenkohl. suggesting that extracting expert models in more ill-defined problems takes too much time and involves less certainty that all the possible solutions patterns can be captured with computer models. Computers can serve as cognitive tools that can amplify. 1985. teachers. & Wiseman. Instead of living with incomplete computer models. Jonassen & Reeves. 2002). and enhance learning (Jonassen. Rather than debating whether human or computer coaches are better at modeling. Almond. & Johnson. & Globerson. They can do so by providing a mechanism for learners to generate and test hypotheses in complex problem-solving situations (de Jong & van Joolingen. and tutors have all the possible models in their heads. Perkins. Mayer & Moreno. when we build computer tutors. Elliott Soloway (1990). Marx. well-defined laboratory tasks that had solutions. cbles can be designed to provide scaffolding that is in the form of diagnostic feedback embedded in the learning situation to be offered when assistance is needed (Lajoie. for example. . 2001. If such sophisticated models are not available. Breyer. given the identification and use of complex student models. extend. 1991). 1998).129 Developing Learning Environments in the context of contained. Lajoie & Lesgold. Faremo. it might be better to consider the two in tandem than to replace one with the other. even the most seasoned ai experts have moved away from the modeling camp. It is somewhat paradoxical that we do not question whether human coaches. Russell. In addition. 1996. 1992. & Davis. Steinberg. such tools can be designed to provide information through multiple modalities. 1996. Musen (in this volume) describes how intelligent databases and expert systems can serve as vehicles for expertise but that domain ontologies that reflect problem solving in a domain are becoming increasingly more important in assisting learners. 2002). Well-defined tasks are easier to model than some of the ill-defined real-world tasks being modeled today. Mislevy. Soloway moved away from computer models to human models involving teachers and peers serving as guides on the side. 2002). Furthermore. In moving toward more complex problem-solving tasks.

I have described methods for documenting the content knowledge earlier in this chapter. methods. is supported by the expert chef. The pedagogical methods employed by this framework involve modeling expertise and coaching learners with feedback in a manner that supports differences in knowledge acquisition. and sociology. Williams. Brown. an apprentice chef observes and learns from the head chef and usually starts with easier tasks. in this example. & Newman. who provides coaching in. For example. 1989. Gott. Coaches scaffold knowledge by providing the right amount of support to learners while they are performing a task. 1991. Palinscar & Herrenkohl. Lave & Wenger. Ellery. Brown. 1994. & Newman. The cognitive-apprenticeship (Collins. A cognitive apprenticeship is more difficult to construct since cognitive rather than physical processes are being modeled for the learner. such as apprenticing to become a chef. cleaning and preparing vegetables. methods for externalizing these decision-making strategies must be used in order to build a model of complex performance prior to modeling for novice learners. 1998. sequence. the correct cutting procedures. and the situation in which learning occurs. When considering a traditional apprenticeship setting. & Duguid. 1992) provides a template for connecting abstract and real-world knowledge by creating forms of pedagogy that are based on a model of learners. The novice. Hence. 1989. 1989) and communities-of-learning (Brown. we think of novices learning from a head chef by observing and modeling physical skills. The cognitive-apprenticeship model (Brown. There are four aspects of the pedagogical model. and whose active intervention gradually fades into the background when the novice demonstrates that he or she can perform the task independently. rather than planning menus and meals for heads of state. other frameworks have led me in deciding modeling performance itself. Collins. Not everyone needs the . for example. The literature on expertise reveals that experts are not always able to articulate their problem-solving strategies or their thinking processes since those things are highly routinized or automated (Anderson.130 modeling complex systems How to Model? Just as theories of expertise have guided my decisions about what to model. as described above. namely. 1989. the task. Collins. 2002) frameworks lead the way in terms of linking cognitive science and pedagogy. Brown. content. 1983). & Campione.

2001). and exploration.131 Developing Learning Environments same amount of help or scaffolding. the scaffolding is gradually removed since the goal is for students to become independent learners and problem solvers. 1999). The sociology of such real-world tasks usually occurs in small groups or teams of people working together. created. with tasks requiring cognitive skills. Brown et al. and empirically studied. are being described. The social sharing of information in this approach is essential to the final outcome of the task. Reflection can consist of self-assessments. Once students demonstrate that they can do something on their own.and out-of-school learning situations.. Similarly. Hence. Wenger. communities of learning (col. Similarities exist between these models and the cognitive-ap- . the work can be distributed among a team of individuals working toward a common goal. allowing students of such professions as medicine and law to work together to solve cases. and it can include comparisons of student processes with more proficient models of performance. The sequence of instruction takes the complexity of the content material into account. each student in the community playing a role as the community progresses toward the goal of learning. In this way they are still participating in producing a final product and developing a conceptual model of how the overall task is performed rather than working in isolation. The methods include opportunities for articulation. Once again. Brown. We see similar considerations with training situations in the real world (Cannon-Bowers & Salas. students participate in the global or overall task by performing smaller components of the overall task prior to engaging in the specific local skills from the start. ensuring that students are challenged with meaningful tasks but not overwhelmed by information that they are not ready for. Barab & Duffy. reflection. Learning in this way must take the sociology of the group into consideration. Exploration and active problem solving in meaningful contexts are key features of the model. 2000. head chefs have a team of sous chefs who help in the overall preparation of a banquet. Learning theories and pedagogical models alike are considering issues of shared cognition more carefully. In both in. The col model engages students in meaningful research activities. 1994. Learners articulate what they know either through performances or through communication of some sort. 1998) and communities of practice (cop. respectively. The cop model is employed in real-world situations.

Both approaches include in their descriptions reference to the sociology of the environment or the notion of community. model proficiency. Lavigne.132 modeling complex systems prenticeship framework discussed above. Lajoie. 1997). Both suggest ways of making schooled learning more meaningful as a way of easing the transition between school and practice. Both approaches use the models of expertise within their design of effective instruction. Patient cases were developed with the help of experts. BioWorld serves to instruct. hence. The remainder of this chapter will contextualize these ideas with examples from my own research. In the area of high school science there has been a call for instruction that provides more meaningful. and a cta established robust models of such ill-structured problems. experiences that foster the development of reasoning abilities that can apply to real-world practice (Johnson & Lawson. Metz. 2001). In this regard I developed “BioWorld. . Guerrera. Munsie. Greer. 1995. and richer learning experiences. & Munsie. and students work collaboratively to collect evidence to confirm or refute hypotheses. 1998. Students learn about diseases associated with the body systems they are studying. and role-playing to serve as new ways of sharing expertise and constructing learning within a community. & Aleong. 1995. and assess knowledge. who or what can do the modeling. Both use models of scaffolding that are based on the assumption that multiple zones of proximal development (Vygotsky. What to Model: Differences in Diagnostic Reasoning My research-and-development efforts aim at integrating instruction and assessment in contexts that reflect real-world situations. turn taking. and how to model cognitive processes. that feedback must adapt to those developmental zones. So far I have described ways of discovering what to model. The col approach is unique in the types of participant structures that it provides as examples of modeling. Guerrera. Patient cases are presented in the context of hospital simulations. Wilkie.” a cble that serves as a platform for evoking evidence of knowledge and capturing complex performance as high school students develop scientific-reasoning skills in the context of problem solving about patient cases (Lajoie. more challenging. 1978) exist within the group and.

These data determined the unique profiles of experts in terms of plans.. in an attempt to get a robust expert model of problem solving for each case. By identifying what experts know.133 Developing Learning Environments Figure 1: The current BioWorld interface: case history screen. In determining what to model. La- . strategies. Figure 1 presents an annotated screen shot of the BioWorld interface. Expert problem-solving traces were monitored and collected dynamically. you might select cirrhosis as a hypothesis. we asked medical experts to solve the patient cases designed for BioWorld. as were verbal protocols. as can one’s beliefs about said hypotheses. it is possible to develop assessment genres for examining transitions in learning in novice performers. The patient case is presented and evidence from this case can be collected and stored in the evidence table. Hypotheses can be selected and changed throughout the problem-solving activity. and actions within BioWorld demonstrating different clusters of competencies that served to establish benchmarks of performance (Lajoie et al. 1995). but your confidence that it is a correct hypothesis might be weak. Results have shown that learners’ confidence levels increase as they acquire more knowledge (Lajoie. For instance.

similar findings are found in clinical psychology. in this volume). use systematic plans and actions throughout problem solving. perform diagnostic tests pertinent to the hypothesis (see Figure 2).. Consultation is also provided online.134 modeling complex systems Figure 2. Ordering diagnostic tests. they select relevant evidence from the patients’ medical histories. As experts solve patients’ cases. BioWorld also presents an online library where factual information is provided. Confidence and knowledge building are intertwined. and make final arguments based on evidence (see Figure 3). students became more systematic in their . where treatment outcomes are related to increases in clients’ self-efficacy (Tschacher & Kupper. et al. With practice. Interestingly enough. vigne. as is a patient chart. allowing learners to practice ordering diagnostic tests. collect pertinent information in an online medical library. Records of expert competency clusters are established and used to monitor novice students electronically in terms of the proportion of overlap of their actions with each predetermined cluster. 2001). Students also learn to self-assess by comparing their reasoning with that of an expert (as shown in Figure 3). Instructional feedback was based on the dynamic assessment of actions throughout problem solving.

Decisions regarding what to model led to design decisions regarding the types of cognitive tools that BioWorld provides to learners. Faremo (2004) examines how medical students and residents solve medical cases. BioWorld highlights relevant information in the decision-making process. et al. Rather than modeling a sequence of problem-solving processes. reasoning and more strategic about evidence collection and hypothesis testing. They can reflect on their own thinking and also compare their solutions with an expert’s path at the completion of the problem. In order to better understand developing expertise. indicating that instruction based on dynamic assessment can lead to transitions in learning (for evidence of these claims. The success of BioWorld with high school students has led us to consider ways to reinvent this cble for medical students. Lavigne. In essence the evidence palette provides a mechanism for learners to monitor what they thought was important to the solution process.135 Developing Learning Environments Figure 3. 2001). see Lajoie. The consult button was designed to model expertise in the form of feedback. Expert evidence as a reflection device for self-assessing final argument.. compared .

This type of in-depth analysis of diagnostic reasoning across cohorts can eventually lead to robust models of problem solving that can extend the use of BioWorld to higher-education classrooms and complete the picture of emerging expertise in this domain. Figure 4 presents a representation of how a medical student’s problem-solving trace compared to that of a resident and expert in diagnosing a patient with celiacs. Figure 4 demonstrates the relation between the plans or goals experts had. lab studies. Celiacs is a chronic intestinal malabsorption disorder caused by intolerance to gluten. and sodium and (2) elevated alkaline phosphotase and prothrombin time. as experts. It also shows that the students identified some of the same information as relevant. It presents in a variety of ways. Her analysis of the cognitive processes involved in diagnostic reasoning represents the knowledge levels and qualitative differences in argumentation and reasoning patterns in this medical cohort.136 modeling complex systems to experts. 2000). evidence identification. weight loss. The patient case took the form of a narrative that presents a 27-year-old male with a 2-year history of intermittent diarrhea. and X-rays (antigliadin antibodies are the gold standard test for detection). and factual and procedural knowledge specific to a task. and the hypotheses they generated in the context of reasoning about this case. Who or What to Model: Human and Computer Coaches in Medical School Medical tutorials present a naturalistic mechanism for studying apprenticeship. potassium. Iron deficiency is found in affected adults. The disorder is diagnosed by means of symptoms. abdominal discomfort. As an example. Through verbal-protocol analyses she was able to capture a picture of emerging expertise as individuals reason through cases. & Faremo. Her coding schemes are based on the plans. As with any apprenticeship there are strengths and weaknesses. The strength of . and distention) may be present. calcium. the types of data they collected and reviewed. and there can be (1) low albumin. Wiseman. We studied human tutors in medicine to develop a model of the medical tutorial situation for studying clinical problem-solving tasks (Lajoie. but did not engage in the same systematic planning and evidence evaluation. and anxiety. and some direct clues (diarrhea.

Figure 4. Expert model with student overlay for the case of a celiac. .

2001) for instructors’ use in building their curriculum of cases in areas that they want their students to have more knowledge about. we have designed the “CaseBuilder” (see Lajoie. Faremo. articulation and reflection. in fact. Two lessons were learned from this setting. there is no standardization of curriculum or guarantee that all students will see the types of patient cases that will build on their existing knowledge. but they can also add feedback at key locations where they anticipate instruc- . & Wiseman. all part of the tutorial process. not only do the instructors build the case. The first lesson was that contextualized and diagnostic feedback could be studied in these settings and that elements of scaffolding and gradual removal. In other words. and reflection. whereby small groups of individuals shared the problem and distributed their learning through shared representations. Hence. articulation. How to Model: The Social Context We are in the process of using the above findings in the construction of better tools for medical tutorials.138 modeling complex systems the setting is that realistic patient cases are seen and handled. The elements of the CaseBuilder will provide the instructor with tools to build in links to library and test information. 2003). with a human tutor scaffolding as needed. As we move the BioWorld cble into a medical setting. The second lesson was that this environment had to consider the sociology of the group and the roles of each member if it was to be an authentic tutorial. and more complex studies of the link between knowledge representation and decision making in the context of information-access systems are needed (Nakamura & Lajoie. What we discovered was that a col approach was common in these tutoring sessions. The library becomes more complex with medical populations. We were interested in documenting the types of diagnostic feedback that human tutors gave to students as well as describing the learning context. were. In doing so our attempt was to develop a cognitive-apprenticeship model for diagnostic feedback that might be built into a cble. The weakness of the natural apprenticeship is that what you see is what you get. an argument could be made that the pedagogical methods of the cognitive-apprenticeship model were used and could be documented. More important.

. selecting what to model must be specific to the domain and problem tasks in question. Medical students learn from experienced practitioners and teachers (experts) and other students and residents (intermediates) during their training. I have attempted to address the intersection between performance modeling and the design of cbles. to make knowledge representations more visible to the teams involved. to facilitate the structuring and tracking of information between small groups. and residents). On a clinical teaching ward. We are currently looking at ways to build a more distributed model of learning for this population in the context of problem-based learning situations. Lu and Lajoie (2003) describe three types of cognitive tools that will be considered in this situation: visualization tools.139 Developing Learning Environments tional moments for their students. They can also run the case with students to see whether they need to revise it for better instruction and assessment purposes. These questions are: What do we chose to model? Who or what should we model? How do we model complex decision making? Theory-based arguments were provided in response to each question. participate in problem-based groups. to help people add on to existing knowledge in specific ways. These experts also continuously interact within a hierarchy of novice to intermediate learners (medical students. interns. and participate in discussions about hospital patients. Students attend lectures. We are still researching ways of designing tools that will facilitate these exchanges in a somewhat controlled setting. The literature on expertise and competency was discussed in the context of the empirical findings from this literature as well as the methodological tools that were offered from that literature. and management tools. Conclusion By asking and answering a set of questions. observe others. Better-informed models of competency can provide transparent trajectories for learning in such domains. In essence. knowledge-building tools. there are experts in many different content areas collaborating hierarchically to work on a given case. allowing individuals to share knowledge collaboratively. Incorporating the social context into performance modeling is more difficult.

Note Research reported in this chapter was made possible through funding provided by the following granting agencies: the Canadian Social Sciences and Humanities Research Council. Thomas Patrick. Janet Blatter. the complexity of assessment is increased. approaches that support specific pedagogical methods in order to promote learning through effective models. Not all modeling is done through explicit coaching. . Office of Educational Research and Improvement. The design of cbles with cognitive tools can support learning by constraining problem solving through support of the overall learning process. Special thanks to Arshad Ahmad. whether the tutor be human or computer. and cop approaches were described. As modeling moves beyond the individual learner and includes a small group of learners. Valorisation Recherche Quebec. Sonia Faremo. Science and Technology. Subsequent to these discussions.140 modeling complex systems Decisions regarding who or what should do the modeling are based on guiding learning paradigms. Carlos Nakamura. Lucy Cumyn. the Quebec Ministry of Industry. For instance. Claudia Guerrera. Although it is widely recognized that real-world learning often takes place in small groups. Commerce. Gloria Berdugo. Many graduate students (former and current) have contributed to the work that is reported here. social constructivists may prefer that modeling occur in groups. Andrew Chiarella. and Jeffrey Wiseman. col. There are still unanswered questions. Susan Lu. Genevieve Gauthier. Cognitive constructivists may agree that models provided in a cble can extend learning effectively when individuals interactively engage in the problem-solving activity. and new ways of sharing and documenting the group process will need to parallel authentic practice. Roger Azevedo. In answering the how-to-model question. Nancy Lavigne. the cognitive-apprenticeship. examples from science and medicine were presented to demonstrate how such approaches guide the design of instruction and assessment in complex learning situations. New ways of supporting differences will need to be developed. it is difficult to develop both individual and group models of performance when deciding appropriate levels of feedback to move the problem-solving process forward. with more expert peers assisting the learning process.

Goldman (Eds. Hillsdale nj: Erlbaum. P. Mahwah nj: Erlbaum. Chi. & Lajoie. J. L. L.. M. (1983). A.). I. & Farr. G. J. G. (1993). Land (Eds. & Salas. 22(2). (2001).. Cannon-Bowers. K. R. J. The wizard of Oz. S. Review of Educational Research.. J.. Resnick (Ed. J. Collins. Situated cognition and the culture of learning. thinking and activity.. The architecture of cognition. & Duguid.. Introduction to M. J.. Jonassen & S. A middle camp for (un)intelligent computing. From practice fields to communities of practice. Perspectives on learning. Englewood Cliffs nj: PrenticeHall. Reder. (1996).. Derry. Glaser. H. S. 32(1). C. (1989). & Campione. Quantifying qualitative analyses of verbal data: A practical guide. 29(4). The development of expertise: The journey from acclimation to proficiency.. Baum. Theoretical foundations of learning environments (pp. A. Attaining excellence through deliberate practice: In- . A. Educational Researcher. 179–201. J. A. L. Journal of Organizational Behavior. Brown. (1994). 11–13.. P. S. Ericsson. and mathematics. Educational Researcher. Reflections on shared cognition. A. H. Cambridge ma: Harvard University Press.. Chi. 2(8). Journal of Educational Psychology. In S. Lajoie & S. A. Anderson.. P.). Cordova. (2003). 715–730. Derry (Eds. J. Journal for the Learning Sciences. 10–14. L. Hillsdale nj: Erlbaum. T. S. Hillsdale nj: Erlbaum. 453–494). Educational Researcher. & Newman. (1998). R. (2002). and choice. Collins. Greeno & S. Chi. writing. & Schauble. 271–316.. B. Chicago: Children’s Press. 6(3). T. M. R. Social learning theory. Brown. A. Educational Researcher. Cognitive apprenticeship: Teaching the craft of reading.). Intrinsic motivation and the process of learning: Beneficial effects of contextualization. T.. In J. Computers as cognitive tools (pp.. (1969).. Lehrer. F. 195–202..1– 11). Design experiments in educational research. H. H. Farr (Eds. M. and instruction: Essays in honor of Robert Glaser (pp. J. 18(1). 68(2). (2003). T. & van Joolingen... diSessa. P. S. S. J. In L. Cobb. P. S. learning. 32–42. Confrey. J. & Duffy. (2000). (1997). 88(4). W. M. Brown. H. 341–367). The nature of expertise (pp. Glaser. Knowing. L. A. (2000). M. Barab. The advancement of learning. M. A. R. Bandura. Scientific discovery learning with computer simulations of conceptual domains. J. E. 32(8).141 Developing Learning Environments References Alexander. R. Ellery. (1988). Greeno. In D. E. 25–55). Creating zones of proximal development electronically. xv–xxvi). 4–12. Thinking practices: A symposium in mathematics and science education (pp. R. & Lepper. (1977). (1998).). Hillsdale nj: Erlbaum. T. & M. personalization.. Educational Researcher. (Original work published 1900) Brown. de Jong.). A. (1989). D. & Simon. A. R. Anderson. 9–13.

Faremo.. Jonassen (Ed. H. Vosniadou. New York: Macmillan. A. S. 2. 693–719). P. No more walls (pp. Columbus oh: Prentice-Hall. Johnson. & Lawson. Ericsson. (1996). D.. Lajoie (Ed. P.. Mahwah nj: Erlbaum. New York: Henry Holt. C. & Simon. Cambridge ma: mit Press. 3. (1995). R.142 modeling complex systems sights from the study of expert performance. Establishing an argumentation environment to foster scientific . Journal of Educational Computing Research. J. In S.. N.. The principles of psychology..). Ronning. Lajoie. Gott. A. S.. In M. Glaser. (2001). Ferrari (Ed. The influence of experience and deliberate practice on the development of superior expert performance. Computers in the classroom: Mindtools for critical thinking.). The pursuit of excellence in education (pp. J.. 41–60). 293–309. E. Educational Researcher. Review of Research in Education. S. C. J. S. Learning with technology: Using computers as cognitive tools. linked representations to facilitate science understanding. S. & Fleiszer. Charness. P. C. 21–25. (1996). The Cambridge handbook of expertise and expert performance (pp. 205–235. pp.). The use of multiple.. Breaking camp to find new summits.. 18(3). Lajoie. Hillsdale nj: Erlbaum. & Robert R. Hillsdale nj: Erlbaum. Wilkie. P. 21–55). 35(1). & J.. In R. (1996). Witt (Eds.). Conoley.. In D. Glover. R.). Lesgold. & Aleong. H. H. (1890). R.). E. Ericsson. S. 97–169. Glaser.. (2004). Lajoie. Ericsson. Russell. (1998). International Journal of Artificial Intelligence and Education. Apprenticeship instruction for real world cognitive tasks. K. Faremo. D. S.. & H. Marx.. M. A. Montreal. What are the relative effects of reasoning ability and prior knowledge on biology achievement in expository and inquiry classes? Journal of Research in Science Teaching. 32(8). B. Hillsdale nj: Erlbaum. J. S. & Reeves. S. & Lajoie. & Davis. Kozma. Protocol analysis: Verbal reports as data (Rev. Guerrera. Azevedo. N. 12(3). Tutoring strategies for effective instruction in internal medicine. Lajoie. Cambridge: Cambridge University Press. A. R. A. Transitions and trajectories for studies of expertise. Lajoie. (2003). Examining medical problem solving in a computer-based learning environment. P. 223–241). Munsie. International perspectives on the design of technology-supported learning environments (pp. D. Computers as cognitive tools: Vol.). Feltovich. (1998). P. W. R. In K. 89–103. Unpublished doctoral dissertation. P. Hoffman (Eds. V. Jonassen. D. Handbook of research for educational communications and technology (pp. P. The influence of cognitive psychology on testing (Vol. Mandl (Eds. 41–85). J. (1989). S. xv–xxxii). (1987). C. T. (1993). Jones. 15.. A. (2000). Cognitive tools for assessment and learning in a high information flow environment. P. Jonassen. T. In S. J. De Corte. T. Toward a cognitive theory for the measurement of achievement. & Wiseman. A.. ed. James. J. H. E. McGill University. Greer. (2006).. K.

& Lajoie. 67(1). Lavigne. Chee.. C. 12(1). N. E. 93–127. On the complex relation between cognitive developmental research and children’s science curricula.. R. P. S.. 31(3/4). (2003). N. R. Motivational considerations in the study of instruction.). New York: Cambridge University Press. (1990).). 201–238). Mayer. & Eggan. Paper presented at the Educational Media Conference. S. Lu.. 5(4). G. Lepper.143 Developing Learning Environments reasoning with Bio-World.. Review of Educational Research. G. . Nakamura. 29(2). 65(2).. L. Lajoie. J. 155–186. G. J. & D. R.. C. (2002). Lee. Breyer. D. 337–340). Making sense of data from complex assessment. Lajoie. R. Farr & J. Constructing knowledge in the context of BioWorld. Guerrera. In M. S.. 363–389. Lajoie.. Lave. Computer assisted instruction and intelligent tutoring systems: Shared goals and complementary approaches (pp. Tutoring strategies for effective instruction in internal medicine. Learners as information processors: Legacies and limitations of educational psychology’s second metaphor. S. New York: Taylor & Francis.. E. A. P.. Shafto (Eds. Lesgold.. Educational Psychologist. R. (1992). Metz. P. Mislevy. (1995). H.)... J. (2003. S.. 15(4). S. & Johnson. Lajoie. Honolulu hi. & Munsie. F. K. Montreal. & Faremo. Chabay (Eds. Charlottesville va: aace. Larkin & R. In J. 107–119. Cognitive task analysis approaches to testing.. Glaser. R. Diagnostic monitoring of skill and knowledge acquisition (pp. K.). C. Steinberg. L. 289–309. (1991). P. S. (1992). In N. & Eggan. S. Learning and Instruction. E. 325–350). S. 151–163. R. & Lesgold. Intelligent instruction by computer: Theory and practice (pp. J. Hillsdale nj: Erlbaum. Lesgold. D. Review of Educational Research. Apprenticeship training in the workplace: A computer-coached practice environment as a new form of apprenticeship. Psotka (Eds. M. Logan. & Wenger. (2002). In D. 89–96).. Mayer. & Moreno. Applied Measurement in Education.. A. S. P. In Y. Suthers (Eds. Lajoie. Proceedings of the International Conference on Computers in Education (pp. (2001). Reassessment of developmental constraints on children’s science instruction. Frederiksen. Lesgold. June). Aids to computer-based multimedia learning. A. (1997). Informational and instructional content in digital learning environments. (1997). (2000. Fostering medical problem solving in a collaborative Web-based learning environment. S. M. Jonassen & G. Wiseman. Instructional Science. & M. E. (1988). Cognition and Instruction. Metz.). A. W. K. & Lajoie. Situated learning: Legitimate peripheral participation.. Hillsdale nj: Erlbaum. Bunzo. E. Charlottesville va: aace. June). M.. Paper presented at the International Conference on Intelligent Tutoring Systems. P. 151–161. sherlock: A coached practice environment for an electronics troubleshooting job. P. Law. 15–36). Proceedings of the International Conference on Computers in Education (pp. J. Almond. McCalla (Eds.

(1992). S. 26–32. G. (1999). 2–9. (Eds. Brooks City tx: Brooks Air Force Base. S. T. (2001). L. & Herrenkohl. July). (2002). Knowing what students know: The Science and Design of Educational Assessment. Pokorny... Evaluation of an avionics troubleshooting tutoring system (Tech. S. Beyond amplification: Using the computer to reorganize mental functioning. (1992). Vygotsky. Science.). Manuscript in preparation. Designing collaborative learning contexts. and identity.. G. Salomon. Soloway. Perkins.. Paper presented at the annual National Science Foundation K– 12 Math. V. Rep. Individual and group approaches to training.. Shute. Human Resources Directorate. Washington dc: American Psychological Association. 171–207). (1985). & Gluck. New York: Cambridge University Press. N. Pea. Armstrong Laboratory. instruction and assessment. Washington dc: National Academy Press. E. Reston va.. D. R. Pellegrino.. & Globerson.144 modeling complex systems Nichols. A. Partners in cognition: Extending human intelligence with intelligent technologies. Wenger. Tobias & H. L. E. Gott. E. Mind in society: The development of higher psychological processes. Calgary. meaning. J. Pellegrino. R. P. Jones. & Alley. Journal for the Learning Sciences. P. K. (2002. 2(4).. Educational Psychologist. N. M.. S. P. 41(1). Cambridge ma: Harvard University Press. (2000). Communities of practice: Learning. Paper presented at the nato Advanced Studies Institute. Palincsar. S. Understanding how students learn and inferring what they know: Implications for the design of curriculum. J. 20(4). Chudowsky.. Lajoie. Theory into Practice. . D. February). Putting case-based instruction into context: Examples from legal and medical education. R. J. & Glaser. In S..). and Curriculum Implementation Projects Conference. 367–427. 167–182.). Williams. (1991). Educational Researcher. O’Neill (Eds. (1978). (1990. 20(3). W. Handbook on training (pp. Interactive learning environments.

1999). 1986). From ancient peoples who first tracked the stars in the night sky to contemporary industrial workers who wish to clarify best practices for their organizations. They evaluate our applications for credit. Musen Stanford University The Vision of Artificial Intelligence People in industrialized societies are surrounded by computer systems that demonstrate intelligent behaviors (Stefik. 1995. and application of knowledge are fundamental aspects of being human. Winston & Prendergast. All people share a basic motivation to share knowledge within their social groups and to codify knowledge so that others can take advantage of it. They help us navigate our cars. we live in “the age of the smart machine” (Zuboff. communication. It is no wonder that computers and their associated technologies—the most powerful communication vehicles that have yet to be invented—provide the infrastructure for much of our attempt to manage knowledge in modern society (Liebowitz. there is an undeniable drive to elucidate and to share information. They help us prepare our tax returns. 1988). . We take it for granted that human knowledge can somehow be embedded in computer systems and that computer systems can help people solve knowledge-intensive tasks—or can solve such tasks completely autonomously. For better or for worse.Technology for Building Intelligent Systems: From Psychology to Engineering Mark A. The formulation.

and suggesting methods by which such systems can be built and maintained. form abstractions and concepts. only in science fiction. investigators began to study how computers might be programmed to do intelligent things. and economics at Dartmouth College to study a new field that McCarthy dubbed artificial intelligence. nearly as soon as computers became available for general research. The emphasis is on the representation of human knowledge in computers and on the primitive building blocks that make such representations possible. What is surprising is that.146 modeling complex systems Because computers have been mass-produced for only a few decades. outlining the obstacles that developers have overcome. and for processing natural language. no one has created ai in the composite sense. Some programs can analyze or generate natural language. The 1950s saw the birth of quite sophisticated programs for playing checkers. For a host of reasons. and represent a class of computer programs that have been enormously successful in our society. & Shannon (1955) declared: “The study is to proceed on the basis of the conjecture that every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it. philosophy. solve kinds of problems now reserved for humans. Minsky. Enormous progress has been made on components of this montage of capabilities. for proving mathematical theorems. McCarthy. The story begins with a close alignment between developers of intelligent systems and scholars of cognitive psychology. The computer programs that help us prepare our taxes and that offer driving directions are extremely useful. however. it is not surprising that the field of artificial intelligence (ai) is relatively young. Rochester. with a perspective that construes the devel- . It ends. This chapter traces some of the history of intelligent systems.” This grand vision for ai has been achieved. An attempt will be made to find how to make machines use language. Some can solve problems typically reserved for humans. nevertheless. psychology. In the original proposal for the Dartmouth conference. We refer to such programs as intelligent systems. but no computer system can be viewed as generally artificially intelligent. Others can form abstractions from primitive data. and improve themselves. to date. By 1956 there was sufficient interest in the challenge of getting computers to demonstrate intelligent behavior that John McCarthy and several colleagues convened a two-month meeting of workers in computer science.

This is a book chapter. Ted Shortliffe and Bruce Buchanan had built mycin. maintainable. they usually think of computer programs that reason using large sets of rules. We must recognize that the computer systems that will be most useful to society are not necessarily those that might replicate human cognition but rather those that allow us to store. The elucidation of the methods that are needed to enable system builders to create such software artifacts can be just as exciting as the elucidation of human cognition itself. however. Carl Djerassi. & Lindsay. Buchanan. however.147 Building Intelligent Systems opment of intelligent systems more as software engineering than as applied psychology. 1984). and. The broad picture of how developers of intelligent computer programs have viewed the role of psychology in their craft over the years tells a very interesting story. not a book. By not attempting to replicate human cognition but instead attempting to take advantage of what is known about the construction of good software systems. a program that could interpret mass-spectroscopy data from the chemical laboratory (Lederberg. a program that could diagnose causes of infectious diseases and recommend treatment as well as physicians could (Buchanan & Shortliffe. By the 1970s. I have to be selective in what I say. and apply human knowledge in specific contexts when it is needed. developers now can build computer programs that are extremely useful. communicate. when Ed Feigenbaum. I find it particularly helpful to look at the evolution of thought regarding knowledge-based systems in a historical context. must be more focused than those of the participants in the 1956 Dartmouth conference. By the end of the 1970s. reliable. The Rise of Rule-Based Systems When most computer scientists think of intelligent computer programs. and—many would say—intelligent. and Joshua Lederberg teamed up at Stanford University to create dendral. 1980). there were scores of wellknown expert systems that used rule-based programming and the explicit representation of domain-specific knowledge to solve a variety . therefore. Enthusiasm for rule-based systems started in the 1960s. The reader should be aware that the need to streamline the presentation requires that I simplify (and sometimes oversimplify) parts of the story. Our goals. Feigenbaum. Sutherland.

(In this chapter. A user of mycin presumed that the patient to be described had either bacteremia (bacteria in the blood) or meningitis. which is now becoming dated. there were hundreds of rules such as the following: Rule 124 if: 1) The site of the culture is throat. I avoid the term expert system. asking questions of the user to gather all the relevant data. The mycin system used the rules to reason about descriptions of patients entered into the program (Figure 1). the mycin inference engine would examine the first clause of the rule and ask: “Is the site of the culture the throat?” .148 modeling complex systems of difficult tasks in application areas that ranged from medicine to geology to computer-hardware configuration. for example. when evaluating Rule 124 (given above). For example. regardless of whether it uses rules. the computer could translate them into understandable English. At the time of the user’s interaction with mycin. I use the terms intelligent system and knowledge-based system as synonyms that refer to any computer system that encodes domain-specific knowledge for the purpose of driving problem solving.) Rule-based systems encode domain knowledge in an intuitive if-then formalism. and 2) The identity of the organism is Streptococcus then: There is strongly suggestive evidence that the subtype of the organism is not Group D Although the rules were actually written in a dialect of the lisp programming language. 1984). if a computer behaved intelligently. the patient ostensibly was ill. A program known as an inference engine interpreted the rules in mycin and allowed one rule to invoke other rules in order to reach the program’s final conclusions. but it was not clear what the source of the infection was. there was an assumption that the program was using rules. The notion of rulebased programming became so popular that the terms rule-based system and expert system began to be used nearly interchangeably. The program would analyze the case description. In the mycin system. and then determine (1) the organism or organisms most likely causing the patient’s infection and (2) the antibiotics that should be administered for optimal treatment (Buchanan & Shortliffe.

149 Building Intelligent Systems Figure 1. then the rule would fail. If. however. If the organism was. indeed. then the rule would fail. A trace of a user’s dialog with the mycin system. then the rule would succeed. mycin would suspend its evaluation of Rule 124 and go on to evaluate those other rules. What if the identity of the organism under consideration were not known one way or the other? The mycin inference engine would examine the rule base to see whether there were any rules that had not yet fired that would allow the system to conclude whether the identity of the organism was streptococcus. The program posed questions to the user. once it was known whether the identify of the organism was streptococcus. Each question seeks information necessary to determine the truth value of one of the program’s many production rules. If there were such rules. then the mycin inference engine would consider the second clause of the rule and ask: “Is the identity of the organism under consideration Streptococcus?” If the organism was not streptococcus. streptococcus. indeed. and further evaluation of Rule 124 would not take place. If the site from which the culture under consideration was taken was not the throat. the throat. however. the mycin inference engine would return to Rule 124 to determine whether the second clause was true . allowing the entry of information regarding a patient to be treated for presumptive bacteremia or meningitis. and mycin would conclude that the subtype of the streptococcus is not Group D. the site of the culture was.

. then the system would conclude that the subtype is not Group D. The inference engine contained no knowledge of infectious diseases or their treatment. If so. Examination of these rules may cause yet other rules to be invoked. if not. It was designed only to invoke rules to follow chains of inference. the inference engine would invoke rule after rule recursively to conclude everything it could about the case at hand so that the program could then determine the appropriate course of action. It was extremely appealing to view knowledge bases as collections of rules. Rules were viewed as independent of one another and not subject to interaction effects as a knowledge base grew. The wonderful thing about the mycin inference engine was that it was a simple procedure. Thus. Starting with an overarching goal rule (basically: “if you know what are the likely organisms causing the infection then recommend treatment”). This style of computer programming is known as a production system. The procedure stops when the truth value of the goal rule becomes known. The rules were viewed as modular “chunks” of knowledge that developers could drop into an expanding knowledge base as new capabilities were defined. Each production rule in the system has a left-hand side (a condition) and a right-hand side (a conclusion). workers in ai at the time believed that mycin-style rules had verisimilitude with components of actual human problem solving. then the inference engine considers the truth value of other rules that can “produce” as their conclusion the condition of the rule currently under consideration. If the second clause evaluated true. mycin’s inference engine thus used the hundreds of rules in the knowledge base to evaluate the case at hand and to generate recommendations for the user. The idea of rule-based systems caught on like wildfire. Most important. all the domain-specific knowledge was encoded strictly in the rules.150 modeling complex systems or false. then the rule “produces” its conclusion. it was a relatively straightforward matter to apply mycin-style inference engines to an entirely different rule base—allowing the system to evaluate cases and make recommendations in new application areas far afield from medicine. An inference engine such as the one in mycin starts with the goal rule and determines whether the condition predicating the rule is true.

and select moves in chess. Perhaps most influential to workers on ai was the research that Newell and Simon (1972) had done to study human problem solving. viewing human problem solving as a process in symbolic reasoning. Although not all ai researchers believed that the goal of automating intelligence would. (Developers of intelligent systems have borrowed heavily from this same methodology when attempting to elicit the knowledge relevant for building electronic knowledge bases. A cryptarithmetic problem. many computer scientists were greatly inspired by investigation to developer comprehensive models of human information processing. Rule-Based Systems as Cognitive Models The surge of interest in rule-based systems in the 1970s followed the explosion of interest in cognitive psychology that took root in the 1960s. prove theorems in logic. Newell and Simon studied how people solve cryptarithmetic puzzles (Figure 2). They described problem solving as the serial processing of symbols transferred from long-term memory to short-term memory in response to external . and the actions that the subjects considered when confronted with a task to address (Ericsson & Simon. the goals to which the subjects seemed to be responding. 1993). Newell and Simon (1972) asked subjects to replace each letter consistently with a digit in order to create a mathematically correct formula. They used the term protocol analysis to define the manner in which they carefully recorded the stimuli to which their subjects seemed to attend. They presented subjects with well-defined problems to solve and carefully recorded and analyzed statements made by their subjects as they “thought out loud” about the problems that they were given. require a deeper understanding of psychology. ultimately.151 Building Intelligent Systems Figure 2.) Newell and Simon’s (1972) empirical analysis of protocols from scores of laboratory experiments encouraged them to adopt an information-processing perspective on cognition.

it was commonplace for developers of intelligent systems to view the problem of building electronic knowledge bases as one of eliciting rules from experts. such as Feigenbaum (1984). Pioneers in the field. using mechanisms that had been well studied by computer scientists and philosophers for decades. By the end of the 1970s. Granted. The ai community paid some attention to these concerns. Dreyfus (1981). 2001)—as the simple application of rule-based reasoning.” where the goal was to get the rules out of the expert’s brain and into the computer. and the builders of intelligent systems can construct their electronic knowledge bases by finding out what those rules are. this work had a rather unguarded implication: human knowledge must surely be encoded as production rules.152 modeling complex systems and internal stimuli. For the ai community. The prevailing view was that the rules in the computer system were precisely the same rules that the experts used to solve real-world problems. but there was an obvious attractiveness to this conjecture. Building an intelligent system was seen as a problem in the “transfer of expertise. 803). pointed to the qualitative transformation of problem-solving abilities that seems to take place as novices become experts and argued that the tacit nature of much professional knowledge makes it impossible for computers ever to reason like human experts. More important. They intentionally were guarded when they claimed to “confess to a strong premonition that the actual organization of human programs closely resembles the production system organization” (p. for example. but the apparent success of mycin and a host of other rule-based systems seemed too compelling to ignore. capturing “golden nuggets” of expertise that could be added to a growing knowledge base. Human problem solving suddenly could be understood in terms of production systems (Young. there was widespread belief that the rules in the knowledge base had correlates in representations in the head of the expert who served as the informant. frequently used a metaphor in which system builders were said to “mine” the knowledge of human experts. By definition. There was face validity to the claim that human expertise could be packaged within a com- . every knowledge-based system became a cognitive model. They saw in their data a common problem-solving procedure that looked as if subjects were performing rule-based reasoning. one rule at a time. there were some vocal naysayers at the time.

1986). military. sprung up. as did new companies that built special-purpose computers finely tuned to run special-purpose programming languages such as lisp. The U. These companies offered inference engines that grew in their capabilities (and in their complexity) to perform increasingly sophisticated reasoning. Companies such as Digital Equipment Corporation with great fanfare installed rule-based systems to help configure the complex backplanes of their minicomputers (McDermott. Indeed. But the boom was beginning by the early 1980s.S. Schlumberger touted the benefits of employing rule-based systems to assist in drilling for oil. When the International Joint Conference on Artificial Intelligence convened in Los Angeles in 1985. had visions of autonomous robots fighting battles in which there were no human casualties.153 Building Intelligent Systems puter. American industry became obsessed with the idea of mining human expertise and putting it in a box. There were new companies selling programs to help developers edit and debug rules. building machines to perform the complex cognitive problem solving previously believed to be the realm of only highly trained people (Winston & Prendergast. Rule-based systems for playing chess routinely—and quickly—lost to human grand masters. the inference engines often became so complicated that yet other companies were founded to advise customers what new hardware and software to buy and how to use their purchases most effectively. 1982). and the excitement was like nothing that computer scientists has seen previously. A new “knowledge industry” had been born. Companies with names . A system called “Prospector” was credited with the discovery of a new deposit of molybdenum ore (Duda & Shortliffe. 1983). the city had not seen such crowds since the 1984 summer Olympic games. New professional organizations. such as the American Association for Artificial Intelligence and the International Association of Knowledge Engineers. The ai Boom and the ai Winter mycin never was asked to diagnose a real patient. which had funded much of the work in ai. and Newell and Simon’s suggestion that simple production systems could form the basis of human cognition seemed to be playing out—at least in silico.

even to control the environment of the buildings in which we work. and Symbolics were busily busing hundreds of conventioneers—including bright-eyed graduate students who had no money at all to spend—to exclusive locations in Beverly Hills. It proved much more difficult to encode and to maintain electronic knowledge bases than most ai practitioners had ever imagined. to audit our income tax returns. however. nevertheless. they saw the construction of intelligent systems as a problem in relocating knowledge. As we shall see in the next section. We have learned. Within a few years. but also by the unprecedented per capita honoraria that they received. offering abundant free food and a chance to see the companies’ products in action. still remain essential commodities and are woven into the fabric of modern society. to determine the appropriateness of requested medical procedures. Such systems continue to review our applications for commercial credit. Scruffy-looking researchers who had agreed to give tutorials on basic aspects of rule-based systems were stunned. The Breakdown of the “Transfer” Metaphor When the early developers of knowledge-based systems spoke of the transfer of expertise from humans to machines. allowing at least one to brag about buying a new sports car as a result! The word on the street was that human intelligence could be mined and put into machines. however. Rule-based systems. By 1988. They spoke of the parallels between transferring knowledge— .154 modeling complex systems such as Intellicorp. ai had become a $1 billion industry in the United States. to assist in industrial manufacturing. not only by the numbers of attendees cramming into their lecture halls. building intelligent systems is an enterprise that is extraordinarily more complicated than simply mining nuggets of knowledge. The expectation that knowledge-engineering groups would be springing up in every cranny of the corporate world was wildly optimistic. to interpret our electrocardiograms and other medical tests. Teknowledge. the expert-systems industry would be largely gone. that the knowledge needed to drive such systems does not exist as sets of production rules waiting to be plucked from the heads of human experts and dumped into computer programs.

and the goal was to get it from one place to another. Johnson. there is concurrence regarding the qualitative changes that occur in the way in which people seem to retrieve information during problem solving. Knowledge was a commodity. much of the expertise often had to be invented in the first place. there is the cognitive stage. Human cognitive skills appear to be acquired in at least three generally distinct stages of learning (Fitts. the models eventually converge. but suddenly the warnings were beginning to be take root. both the developers and the experts continually revise their respective mental models. At the core of the matter is the automatization of professional behavior. When developers interact with application specialists to build knowledgebased systems. during . Although different authors have used different terms to describe the three phases. 1987). The problem of building knowledge-based systems was no longer seen as one of transferring expertise. Dreyfus (1981) and others had been warning about these difficulties all along. Builders of intelligent systems began to accept that the application experts. 1983). In the course of building the system. of course. and revise their respective models (Regoczei & Plantinga. often have no preexisting mental model of how they do their work. whose professional acumen was to be encoded as a knowledge base. Although the developers and the experts may have very different mental models at the outset of their collaboration. The view of knowledge-based development as the transfer of expertise hindered the now-obvious recognition that building electronic knowledge bases is a creative and inventive activity. Initially. By the 1990s developers of knowledge-based systems began to realize that the nuggets of knowledge were not always there for the mining. the experts cannot begin to verbalize how they actually go about solving problems. the developers form mental models of how the experts solve problems. publicly examinable form—typically the emerging knowledge base—and (2) because the frequent consideration of examples and test cases forces the system builders to assess. a substance. the experts. have mental models of their own that attempt to capture their professional problem-solving behavior. Often. This convergence is possible (1) because the development process forces all parties to commit their mental models to a fixed. corroborate. 1964.155 Building Intelligent Systems generally in the form of production rules—from human experts to computer programs and transferring knowledge from computer programs to naive users.

(2) unaware that a response has been affected by a stimulus. and effortlessly—without thinking. yet the skills that these professionals actually practice are procedural in nature (Anderson. when professionals are asked to report on their compiled expertise. In experimental situations. Instead. the learner often verbally rehearses information needed for execution of the skill. With repetition and feedback. The consequence is that the special knowledge that we would most like to incorporate into our intelligent systems is often that knowledge about which experts are least able to talk. the person performs the actions appropriately. inaccessible to their consciousness. In this stage. subjects give verbal reports of their cognition based on prior causal theories from their nontacit memory (Nisbett & Wilson. Then. and (3) unaware that a cognitive response has even occurred. Furthermore. The knowledge that experts acquired as novices may be retrievable in a declarative form. The knowledge has become tacit (Fodor. as humans become experienced in an application area and repeatedly apply their know-how to specific tasks.156 modeling complex systems which an individual identifies the actions that are appropriate in particular circumstances. There is substantial evidence that. 1977). Suddenly. . The inability of experts to verbalize these compiled associations is well accepted (Lyons. the learner compiles the relations from repeated practice to the point where he or she can perform them without conscious awareness. 1986. because Western culture mistakenly teaches us that accurate introspection somehow should be possible (Lyons. they often volunteer plausible answers that may well be incorrect. 1968). Nisbett & Wilson. their knowledge becomes compiled and. in which the relations noted during the cognitive stage are practiced and verbal mediation begins to disappear. 1977). The problem for builders of intelligent systems is that. subjects have been shown to be frequently (1) unaware of the existence of a stimulus or cue influencing a response. either as a result of direct instruction or from observation of other people. people freely explain and rationalize their compiled behaviors without recognizing that these explanations are frequently incorrect. Next comes the associative phase of learning. thus. the person begins to apply the actions accurately in a fluent and efficient manner. 1986). in the final autonomous stage. 1987). Johnson (1983) has identified this phenomenon as the paradox of expertise. Experts lose awareness of what they know. proficiently.

there had been a conviction that the secret to creating ai was rooted in a deeper understanding of human psychology. By the time the ai community had declared that “winter” had settled in at the end of the 1980s. be observed. They are creating de novo theories of professional problem-solving behavior and representing those theories in terms of electronic knowledge bases. and easily disseminated to other people in need of advice. The intelligent systems that result from this work may not achieve the same level of nuanced performance associated with the procedures actually used by domain experts. Psychologists who were studying human problem solving not only reaffirmed that much professional behavior involves knowledge that is completely tacit but also noted that the stimuli that drive problem solving in the first place are often environmental and difficult to isolate. The notion of situated action is more a perspective on behavior than it is . extended. Knowledge could never be “extracted” and formalized since it could never be decoupled from the situations in which it was applied. but that do not reproduce. As such. Clancey. thus. nevertheless. knowledge bases instead represent only models of surface-level behaviors—models that attempt to approximate. the relation between psychology and knowledge-based systems seemed increasingly disconnected. however. It is simply incorrect to view an electronic knowledge base as an embodiment of actual professional knowledge. Intelligence increasingly was seen as a holistic interaction involving the problem solver with the environment. 1987. Further Assaults on Knowledge Bases as Cognitive Models At the time of the Dartmouth conference in 1956. A major salvo came from the situated-action community (Suchman. The developers serve the important function of detecting gaps in the articulated knowledge of their informants and helping them fill in those gaps by defining plausible sequences of actions that can achieve the necessary goals. The underlying knowledge bases can. 1989). intelligence could never be decontextualized. 1997). the actual problem-solving steps used by humans (Clancey.157 Building Intelligent Systems Developers of knowledge-based systems are. not mining preexisting nuggets of knowledge.

158 modeling complex systems a formal psychological theory (Kushmerick. This move toward a software-engineering perspective was reinforced by other events in the ai community. ultimately. There were no preexisting abstract representations in the brain to define general plans of behavior. The situated-action perspective cast aside Newell and Simon’s hypothesis that human problem solving is fundamentally grounded in the processing of symbols (Newell. Such responses might require the dynamic creation of a mental representation to guide the agent’s behavior. no production rules) to drive cognition. Deep Blue demonstrated convincingly that knowledge-based systems would be successful when they could . but not because the program was cleverer than Kasparov or because it modeled the thought processes of a better chess player. Workers at CarnegieMellon University built Deep Thought (reborn at ibm as Deep Blue) and achieved outstanding game-playing performance.g. and the ability to use brute force to look ahead many moves to see the consequences of potential actions (Newborn. They did so. Intelligent behavior was seen instead as the consequence of myriad specific. but there were no prefabricated representations for the abstract problem space (e. Perhaps none was more important than the appearance of computer programs that could. The emphasis shifted almost completely away from psychology toward the construction of useful pieces of software.. 1998). The primary consequence of the “situated” perspective was that virtually no developers of knowledge-based systems claimed that they were creating cognitive models anymore. not by building a computer system that mimicked the cognitive behaviors of human experts. beat grand masters in games of chess. Deep Blue ultimately beat Gary Kasparov at his own game. Although the emergence of the situated-action perspective caused a minor uproar in the cognitive-science community. 2002). 1996). there were only particular behavioral responses to particular situations. the execution of multiple computations in parallel on distributed processors. but by capitalizing on things that only computers can do: blazingly fast computation. finally. its effects on the developers of knowledge-based systems were. but it had a jarring effect on the way people viewed the role of representations in cognitive science. often-unknowable stimuli in the environment leading to myriad specific actions. 1980). more muted (Menzies & Clancey.

Some investigators. The claim was that knowledge engineers needed a skill set that was different from the one required by computer programmers. The kads project (Schreiber et al. The emphasis was placed squarely on better techniques for systems engineering. 2000).159 Building Intelligent Systems take advantage of the unique abilities of the hardware and software of which they were made. They maintained vehemently that the building of intelligent computer systems was different from the development of conventional computer programs. particularly in Europe. The kads consortium argued that the construction of knowledge bases and the construction of conventional software share an important element: the need to manage complexity. rather than on better cognitive modeling. kads emphasized the role of predefined. Creating intelligent systems involved the representation of large amounts of content knowledge and almost no programming of traditional computer code. stereotypic patterns of problem solving that could help developers elicit knowledge. 1987). Knowledge-Based Systems Are Software During the mycin era. Just as workers in traditional software engineering were grappling with methods to help computer programmers manage complexity and generate solutions creatively (Brooks. Psychology was important in the development of intelligent systems—not because of the need to model the cognitive processes of human informants. but because of the need for the system builders themselves to be able to keep track of a huge network of knowledge representations and the relations among them. the kads project strove to create methods allowing knowledge engineers to keep track of elicited knowledge and to identify gaps in the emerging problem-solving solution. nothing to compile.. The approach also stressed ways of categorizing knowledge into different “layers” of specificity to help structure the emerging knowledge-based . thought otherwise. began in the early 1980s to provide a comprehensive methodology for building knowledge-based systems. that knowledge engineers were fundamentally different from software engineers. for example. There was nothing to flowchart. most developers of intelligent systems argued strongly that knowledge engineering was not the same thing as software engineering.

An essential contribution of the kads project was the view that. like conventional software. First developers should build a conceptual model that defines what task the system is to perform in a functional manner: What are the inputs? What are the outputs? How does the nature of the inputs constrain the nature of the outputs? Can the task be decomposed into subtasks. From a conceptual model of system performance. the conceptual model would identify patient data as the input and a set of antibiotics to prescribe as the output. developers create a design model that makes a commitment to an abstract pattern for how the system might be created using software. methodical practice for building complex software systems that manifested intelligent behavior. Developers should never simply sit down and try to build a large. Stages in the development of intelligent systems. commonkads—transformed the process of knowledge engineering from an informal enterprise of mining nuggets to a stepwise. instead. The kads project viewed system building as the construction of a set of models and argued that those models should be made explicit and examinable. The design model then drives the implementation. they need to approach the design process in a principled manner. kads was a software-engineering methodology for building intelligent systems. complex system all at once. both to simplify the computation and to make the system more cognitively tractable? In the case of mycin. The overall task of determining what presumptive therapy to administer might be decomposed . For many development groups. system. At its core. intelligent systems need to be developed in stages (Figure 3).160 modeling complex systems Figure 3. kads—and its subsequent refinement. for example. which results in a physical software system that users can run.

2000). A design model for mycin as it was originally implemented might describe how production rules could be used to program the organism-identification task. 2004). the authors of kads and commonkads insisted that builders of intelligent systems needed to articulate a coherent design before they actually began to encode the production rules and write the program components that would form the basis of the executable system (Schreiber et al. When creating conventional software. rather. From the design model would come an implementation. The design model is just that. The kads project initially argued for a “waterfall” model of software development (Sommerville. Just as developers of methodologies for software engineering were arguing that design must precede implementation.. Because any creative task is simplified when it can be decomposed into more tractable elements. together. There is no discussion of how the computer might actually perform the tasks required by the finished system. it is conceptual. how a constraint-satisfaction program could be used to determine the minimal set of antibiotics to prescribe.161 Building Intelligent Systems into two subtasks: (1) identify what classes of organisms most likely were causing the infection and (2) construct a small set of antibiotics that. object-oriented programming often plays an important role in helping developers decompose the implementation task into the construction of cognitively tractable software modules. ideally there are software components that can serve as building blocks for developing the implementation. I will have much more to say about possible building blocks for the design model in the next section. Next comes the creation of a design model. It is not a working piece of computer code but. The implementation would be the software artifact with which end users would interact and that actually would provide them with advice. and so on. it is a design. Those primitives could be the components of expert-system-building shells or abstract algorithms for performing tasks such as constraint satisfaction. were likely to be effective against the most likely organisms. an outline of how software components might be brought together to implement a working system. Construction of a design model is facilitated if developers have access to a set of primitives that can serve as building blocks for constructing a design. The conceptual model is just that. with information flowing .

Thus. Unfortunately. they must revisit their conceptual models and design models as they gain experience with a particular implementation. for implementing intelligent systems? How can a set of building blocks help developers construct their models in a way that ensures comprehensive and useful designs? How can a set of building blocks ensure that the software that implements an intelligent system is maximally reliable. the focus for research shifted radically to elucidating components both for design models and for implemented systems. but suddenly the community was much more concerned about software components and elements of design than it was about nuggets of expert knowledge. System maintainers can guard against such drifts when they use building blocks that have well-understood meanings and behaviors to create their design models and implement their systems. and traceable in its behavior? As the knowledge-engineering community evolved in the 1990s. Similarly. the design model may become increasingly irrelevant for explaining the programming choices made in the implementation. then developers can have confidence in the consequences of replacing one building block with another. If each of the building blocks is fixed and invariant in its semantics. A primary problem with the approach is that developers rarely get things right the first time. but their attention was turning to the cognitive processes of the system developers—with the goal of creating modeling and implementation primitives that would help software engineers manage . well-meaning developers can choose to tinker with the implementation and ignore the design model. the behavior of the knowledgebased system can come to represent something different from the specification documented in the original conceptual model. so to speak.162 modeling complex systems from the conceptual model to the design model to the implementation. Psychology still was important to the builders of knowledge-based systems. 1992). maintainable. In time. the conceptual model can rapidly become out-of-date unless system maintainers are particularly conscientious about documenting all their changes. As system builders have thought about the development process shown in Figure 3 above. The emphasis was still on nuggets. Any drift in behavior can be cleanly ascribed to a change in the components and can be documented fairly easily (Krueger. ultimately. a number of questions surfaced: What are the optimal building blocks for creating design models and.

In the knowledge-based-systems community.g. In any modeling activity. Partly influenced by the kads project. And an abstraction—the word’s most literal meaning tracing back to its Latin roots—takes away from some reality. These classes of knowledge describe (1) propositions that we know about the world—concepts and the relationships among concepts—and (2) the procedures by which we can use those propositions to solve problems. what the kinds of essential elements are). ontology was an abstract science. developers consider two broad classes of elements when modeling knowledge. thus. Let us consider these two modeling elements in turn. Components for Modeling Knowledge Models are abstractions.. the branch of metaphysics that concerned the study of existence. Aristotle worked to categorize objects and to define inherent properties of those objects that could differenti- . leaving just the essential elements. but there is also surprising consensus. ontologies: defining what exists For hundreds of years. Philosophers since the time of Aristotle have tried to define and categorize what exists in the world. the notions of deciding what are the right building blocks for creating models and of deciding what is the right language (or languages) for writing down descriptions of those building blocks. carving up his world into those objects that were immaterial (spirits) and those that were material. it becomes convenient to think about knowledge in terms of ontologies and problem-solving methods. there are two key questions: (1) What are the kinds of essential elements to abstract? (2) How should modelers talk about those elements? There are.163 Building Intelligent Systems the tremendous complexity inherent in the construction of intelligent systems. those that were inanimate and those that were animate. To use the current buzzwords. partly influenced by observations from workers such as Chandrasekaran (1986) and Clancey (1985). there is ongoing discussion regarding the details of how knowledge is best modeled (e. those that were humans and those that were beasts.

The rediscovery of the notion of ontology in the late 20th century. 1999). The very idea stemmed from centuries of work in metaphysics. 1998). The study of ontology remained a relatively obscure branch of metaphysics until the last decade of the 20th century. much of human cognition centers on the creation and application of categorizations and taxonomies (Bowker & Star. however. to Roget and the thesaurus of the English language. provided a principled way for computer scientists to begin to think about the world that their programs were attempting to model and allowed them to put their models into a well-established context (Guarino. hardly something that the Internet community could claim as its own. other commercial Web sites such as Amazon. 1998). but also that of software engineers trying to model the concepts and relationships needed to define some reality for the purposes of building a system (Guarino. put an article in front of it. of course. not only the work of philosophers. 2004). This notion of ontology was. Suddenly. Indeed. to e-commerce transactions (Staab & Studer. Yahoo! initially had enormous success because it provided easy entrée to the Internet by allowing users to browse a detailed ontology of the kinds of Web pages that people had created. . Ontology became a big business as the Internet revolution required computer programs to be able to represent detailed descriptions of hierarchically organized information—from Web sites. In the modern era the formal classification of entities—from Linnaeus and the speciation of living organisms. to commodities. There are now conferences and workshops devoted exclusively to the subject of building ontologies. and began to speak of an ontology as a description of the concepts and relationships among concepts that existed in some world being modeled. to Dewey and the classification of literature—had had profound effects on human thought. when computer scientists co-opted the word.164 modeling complex systems ate each class into subcategories. There are specialty journals and an explosion of books on the subject. the elucidation of ontologies was.com offered a detailed ontology of books and music that allowed Web surfers to examine an enormous inventory of online products. The study of ontology has swelled because formal data structures that can be used to represent some aspect of the world and to provide an enumeration of essential categories have become more important commercially than any philosopher could ever have imagined. Soon.

including the authors who developed the protocol. The ontology models the entities and relations needed to specify how medical care should be given over time to patients who have cancer. eligibility criteria that determine whether a clinical trial actually applies to a given patient.. the eligibility criteria that determine the set of patients to which the protocol applies. For example. Clinical trials have a number of attributes. In the ontology.. The ontology describes what may exist—what are the possible components of a protocol (e. Figure 4 shows an ontology of a medical application area—standard protocols for treating patients who have cancer that specify how chemotherapy and other treatments should be administered to patients. tests to perform that may predicate decisions in the clinical algorithm) and what are the relationships among those components (e. and the end points that determine when treatment should terminate.g.165 Building Intelligent Systems Figure 4. When building knowledge-based systems. clinical-trial protocols are a kind of clinical guideline. algorithms that specify the sequence of treatments to offer. either as part of a clinical trial or in accordance with a clinical management guideline that attempts to define a set of best practices. An ontology of oncology protocols. that .g. developers now concentrate on fashioning a formal model of the entities in the domain and the relationships among those entities.

Protégé provides an interface with which users can enter instances of that entity (e.. The form on the right-hand side of the screen shows the attributes of whatever class is highlighted on the left.g. such as guideline. trial protocol). The interface also allows developers to specify relationships among the instances (e. At the top of the hierarchy may be an object with a very general description (e. for example.g. In the figure. Ontologies are typically represented within computer systems as hierarchies of objects. Each object in the hierarchy has attributes—some of which are particular to that object. Gennari et al.g. the Protégé system can generate a user interface with which developers can enter the content knowledge structured by that ontology (Figure 5). Fergerson. Crubézy.. the concept guideline has three distinct subclasses: management guideline. instead laying out the classes of entities that exist in protocols in general. Some of the attributes of trial protocol are inherited from more general concepts. The tree browser on the left-hand side of the screen shows the entities in the ontology. thing). a particular clinical trial for treating breast cancer). In Figure 4. The subclasses in turn will have subclasses. 2003). others of which may be inherited from objects higher up in the hierarchy. There will be subclasses of this object that represent more specialized concepts. In our laboratory.. Noy. This workbench is known as “Protégé” (Musen. 2000.g.. consultation guideline. & Gennari. and trial protocol.166 modeling complex systems steps in the clinical algorithm are related temporally). The protocol ontology can then form the basis for acquiring the content knowledge for specific protocols—providing a structure that allows developers to build multiple knowledge bases such that each knowledge base encodes the description of a particular type of guideline or clinical trial (Musen.e. For each generic entity in the ontology (e. we have developed a computer-based workbench that helps system builders use ontologies to create detailed knowledge bases. attributes of trial protocol include items such as authors and eligibility criteria.. The ontology does not describe any particular protocol (i. a specific plan of care for a specific medical problem). For any ontology. Grosso. the sequence of steps that determine how chemotherapy and other inter- .. 1998). The screen shot in Figure 4 shows how an ontology appears within the Protégé tool.

Tools such as Protégé divide the process of creating an electronic knowledge base into two phases: (1) description of a general ontology of the application area and (2) instantiation of that ontology to define particular knowledge bases. the particular blanks to be filled in and the choices in the palette available for drawing diagrams). users fill in the blanks and draw diagrams to specify the content knowledge that constitutes particular knowledge bases. The particular protocol compares the effects of conventional chemotherapy and radiotherapy with those of high-dose chemotherapy followed by bonemarrow transplantation.167 Building Intelligent Systems Figure 5. The ontology defines the classes of entries that a user can make (in this case. We and many other groups routinely use Protégé (and tools like Protégé) in this two-step process of knowledge-base definition. A knowledge-entry tool for the specification of clinical trials. Here. the user is entering knowledge about a clinical trial for treating patients who have breast cancer. ventions should be administered over time). the Protégé-generated tool allows developers to create knowledge bases that define clinical trials. In this example. Protégé generated the user interface for this tool directly from the ontology shown in Figure 4 above. .

that were very general purpose in nature and could. thus. Chandrasekaran (1986) and his colleagues at Ohio State University were identifying broad patterns in the problem-solving behavior of knowledge-based systems that seemed reusable from one system to another. The second major class of modeling components for knowledge-based systems—problem-solving methods—specifies what to do but typically takes no position on what exists (Eriksson.g.. 2004). Newell and Simon (1963) identified a number of “weak” problem-solving methods. therefore. a description of a clinical trial). the methods required only a single task-specific evaluation function that could inform the problem solver whether a selected means would lead to some desired ends. 1995. such as means-ends analysis and generate and test. In the early 1980s. when the hype regarding knowledge-based systems was reaching its peak. A problem-solving method defines what a system should do with the propositional knowledge in an ontology (Crubézy & Musen.168 modeling complex systems problem-solving methods: defining what to do Ontologies specify what exists in the world (e. but they typically take no position on what should be done with that information (e. & Musen. be applied to a wide range of problems. planning. The notion of problem-solving methods has been around since the early days of ai.. The Ohio . The distinction between problem-solving methods and ontologies mirrors the distinction in psychology between knowing how and knowing that. whether some generated solution would test satisfactorily. All knowledge about the desired outcome—and. Puerta.. and so on. they must model both the ontology of the application area and the problem-solving method with which the system will address the task at hand.g.g. A problem-solving method is a general procedure for solving some well-defined task (e. using a clinical trial description to recommend treatment for a particular patient). Shahar. some workers in ai were taking a fresh look at the idea of a problem-solving method. 1988). Any time developers choose to build an intelligent system. McDermott. classification. about whether the problem solver might have achieved its goal—was built into the evaluation function. The methods were considered weak because they could not take advantage of explicit domain-specific knowledge to drive all their inferences. constraint satisfaction). Tu.

For example. the use of data about the case at hand to infer generalizations about the patient to be treated involved a collection of inferences known as feature abstraction.169 Building Intelligent Systems Figure 6. State group described recurrent problem-solving approaches that. which performed diagnosis by examining a hierarchy of possible diagnoses (the most general diagnoses were at the top of the hierarchy) and performed a top-down analysis of the hierarchy to determine the most specific diagnosis that applied to the situation at hand. Clancey maintained. Heuristic classification in mycin. For example. Rather. and (3) solution refinement. Clancey (1985) showed that the behavior of mycin was not the emergent result of hundreds of modular rules interacting seemingly at random. unlike weak methods. could take advantage of domain knowledge to address tasks in more sophisticated ways than are possible with a single evaluation function. additional domain knowledge was considered by the problem solver. for example. Chandrasekaran described a method that he called establish-refine. In Figure 6. Problem solving in the mycin system can be construed in terms of three major inference patterns: (1) feature abstraction. Soon thereafter. mycin is shown to use its domain knowledge to reach the conclusion that the patient under consideration may be a “compromised host” either because the . (2) heuristic match. Clancey had fully dissected the problem-solving activities of the mycin system. the developers of mycin had deliberately—although perhaps unintentionally—built into mycin a coherent set of inference patterns that collectively he called heuristic classification (Figure 6). After Clancey (1985). Each time a diagnosis was established and then refined into a more specific diagnosis.

that the organism causing the infection may be of the class Gram-negative rod). those rules merged into a singular. biography). Other inferences allow mycin to perform solution refinement. For example.. and match those generalizations heuristically to classes of minerals that might be present (Duda . Grundy would interview the borrower. More important. and (3) solution refinement. Clancey identified dozens of other contemporaneous knowledge-based systems that also seemed to be using heuristic classification as their principal problem-solving method. thus. then use heuristic match to identify classes of books in which the borrower might be interested (e. heuristics links the abstractions related to the condition of the patient to abstractions related to the possible causes of the infection. In this case.170 modeling complex systems white-blood-cell count is low or because the patient is an alcoholic.g.. adopting an approach much like establish-refine.g. In the case of mycin. making recommendations to potential borrowers of books to read. that the patient is a compromised host) with an element of another abstraction hierarchy (e. and then use additional knowledge to be able to recommend specific books. Even if the developers of mycin had talked about their work as the elicitation and engineering of hundreds of production rules that related directly to the rules expressed by infectious-disease experts.. Workers at sri International had built a system called “Prospector” that would identify geographic areas in which to drill for minerals. Elaine Rich had built a system called “Grundy” that served as a kind of automated librarian. mystery novels. use feature abstraction to determine appropriate generalizations.g. Special inferences. allow mycin to associate an element of one abstraction hierarchy (e. perform feature abstraction to conclude some generalizations about the client. Prospector would elicit features about a given geographic area. using domain knowledge to distinguish among a group of organisms that may be causing the infection. (2) heuristic match. The program was not just triggering production rules. The system is. mycin uses knowledge of microbiology to identify the most specific organism that may be causing the infection. known as heuristics. Clancey (1985) demonstrated a new coherency in the way in which mycin solved problems. using input data about the case to determine resulting abstractions about the case. science fiction. it was following a well-defined pattern of inference that involved (1) feature abstraction. consistent problem-solving method when an observer could obtain sufficient distance.

the input data must be mapped to some classification). What is essential is that the method must be programmed in a manner that allows the data on which the methods operate to be provided in terms of an appropriate domain ontology. on the other hand. and solution refinement. Putting the Pieces Together Because each problem-solving method has a method ontology that defines the problem solver’s inputs and outputs. one based on feature abstraction.g. and (3) software that actually implements the method as a computer program that can execute on some machine. Tu. 1988). developers have identified and written computer programs to implement several reusable problem-solving methods that can be applied to a variety of tasks (McDermott.. Rothenfluh. These constructive methods encode algorithms for planning. Problem-solving methods themselves can be viewed as having three components: (1) a definition of the algorithm that includes the problem-solving steps that the method carries out (often in some formal language) and the relationships that must hold between the method’s input data and the method’s output data once the algorithm has finished running (e.g. laboratory-test results) to the cor- . and so on—problems where it is impossible (or impractical) to preenumerate the solution set in advance. simulation. These contributions include methods such as heuristic classification and establish-refine. in that the solution set is preenumerated and the problem solver selects one or more solutions from that set. where the problem is one of classification. (2) a description of the format of the method’s input data and output data (often referred to as a method ontology. in the case of heuristic classification. Gennari. constraint satisfaction. In subsequent years. 1983). but all of them used a shared problem-solving paradigm. Other problem-solving methods. The systems that Clancey (1985) studied were designed for different application domains. 1994). Problem-solving methods can be implemented in virtually any computer language. heuristic match. require the problem solver to construct a solution on the fly..171 Building Intelligent Systems & Shortliffe. & Musen. a patient with an infectious disease. developers can relate application-specific concepts defined by a domain ontology (e.

the mycin antibiotic-selection task might be best solved using a constraint-satisfaction problem-solving method. 2004). the domain ontology defines the propositional knowledge of the application area. Thus. the problemsolving method defines a generic strategy for obtaining a solution for a certain class of problems.172 modeling complex systems responding application-neutral requirements defined for the method (e. A separate method would be needed to compute the most parsimonious set of antibiotics that can provide protection against all the potential microorganisms named by the organism-identification task. possible classes of infecting organisms. 1994). the relationships between signs. (2) that the patient receives the fewest number of antibiotics possible. symptoms. 2004. Thus.. the laboratory tests that physicians might request. and antibiotic. the manner in which the developer relates elements of the domain ontology to the data requirements of the problem-solving method determines how the method actually will be used to solve particular cases in the application domain (Crubézy & Musen. Imagine that a developer were to re-create the mycin system using these types of building blocks (Figure 7). Developers can relate the entries of an ontology of microorganisms to the abstract solution set evaluated by the heuristic-classification method. The system builder would need to relate the data requirements of both the heuristic-classification method and the constraint-satisfaction method to the classes of data described in the domain ontology. Gennari et al. a case to be classified. This latter method needs to ensure (1) that the physician will administer at least one antibiotic that will treat each of the possible pathogens. and test results.. they can relate definitions (such as that of compromised host) to the inferential knowledge required for feature abstraction. The reengineered mycin system would use the heuristic-classification problem-solving method to address the task of identifying the most likely infecting organisms (as in Figure 6 above). organism. Attributes of these concepts would specify the signs and symptoms that patients might demonstrate. In practice developers specify this wiring of domain ontologies to problem-solving methods by declaring explicit mappings between these two kinds of building blocks (Crubézy & Musen.g. The domain ontology would include concepts such as patient. and (3) that the patient does not receive any antibiotic to which he or she is allergic or that may interact with any other drug that he or she is taking. input data). and so on. .

as there was when rule-based systems debuted in the early 1980s. Ontologies and problemsolving methods exist at two different levels—both as conceptual entities (abstract models of concepts and of generic problem-solving approaches) and as software artifacts (implemented object systems . most important. The evolution in thought has taken place rather quietly. Propositional knowledge would be represented as a domain ontology. These classes of building blocks have enabled a wide range of developers to create extremely complex systems and. to maintain those systems over time as needs evolve and situations change.173 Building Intelligent Systems Figure 7. and reusable problem-solving methods (Musen. The knowledge-based-systems community has gained considerable experience over the past decade building intelligent systems using explicit domain ontologies. The problem-solving knowledge required for the organism-identification task would be encoded using heuristic classification. 2004). Although the vast majority of small-scale knowledge-based systems continue to be built using the rule-based approach. knowledge bases that add detailed content specifications to those ontologies. The system would require an additional constraint-satisfaction problem-solving method to address the task of constructing a minimal set of antibiotics that can cover for all likely pathogens. There has been no overbearing excitement. Reengineering the mycin system with modern components. knowledge-engineering groups are increasingly turning to the use of ontologies and problem-solving methods to construct systems in a way that makes it clear (1) what knowledge about the world actually has been encoded in the systems (what is the ontology) and (2) how the systems go about solving the tasks with which they are confronted (what is the problem-solving method). There has been no great hype about any of this.

the replication of human cognition in silico was viewed as essential for the creation of systems that displayed ai. 1983). however. 1984). In the past. When thinking about the design stages depicted in Figure 3 above. however. it is clear that—abstractly—developers can use ontologies and problem-solving methods as building blocks for purposes of conceptual modeling and—concretely—developers can use these same building blocks to implement actual computer programs. Building Intelligent Systems Fifty years after the Dartmouth conference. that same central nervous system. We have a long way to go .174 modeling complex systems that store concept descriptions and implemented problem solvers). Once. workers sought to build cognitive models—sometimes replete with obviously erroneous human behaviors such as the Tversky and Kahneman (1974) biases—in an effort to duplicate the perceived thought processes of real-world experts. by mimicking the perceived functioning of the most advanced problem-solving system that we know of (the human brain). fails miserably whenever it tries to program a rule base of more than a few hundred entries or to keep track of how all the rules might interact. Deeper understanding of the psychology of intelligence and of the underlying neuroscience unfortunately did not lead developers to create better computational artifacts. we could create ai of the highest order. It once seemed natural to believe that. Builders of early expert systems. few psychologists come to scientific meetings that concern the development of knowledgebased systems. The human genome is able to guide the creation of a central nervous system that contains more than 100 billion neurons in an utterly graceful and remarkably reproducible manner. did not anticipate that the human brain that they sought to imitate was so poor at articulating its own reasoning (Johnson. They chose approaches that were consonant with the way in which humans seemed comfortable addressing problems and even avoided algorithms that might be more normative or more computationally efficient. They did not anticipate either the ultimate complexity of the rule-based systems that they would create or the enormous difficulty of maintaining those rule bases over time (Bachant & McDermott.

with propagation of signals upward through the network. it is the ability to define a formal engineering discipline for the cre- . by building a large number of systems—many of them successful. thus. cognitive models do not lead to practical computational artifacts that are either highly scalable or easily maintainable. not through psychology or through neuroscience. These “connectionist” architectures consist of layers of nodes (sometimes called neurons) linked together in a manner that is inspired by neural circuitry in the cerebral cortex but that is not representative of any specific neural pathway. Our collective experience has resulted in our ability both to engineer intelligent systems from reusable software components and to maintain and enhance those systems in a maximally efficient manner. The major lesson of the past five decades is that the most pragmatic route to the creation of intelligent systems is. This activity ultimately causes the nodes in the topmost layer to signal an appropriate output—a combination of nodes that can be inferred to represent the classification of the input signal into some category. Despite our best intentions. Although artificial neural networks continue to be used to address multivariate classification problems. Although there were no traditional engineers invited to the Dartmouth conference in 1956. they do not solve the general problem of engineering computer systems that demonstrate intelligent behavior. During the 1980s there was a surge of excitement that diverted the attention of some workers in ai away from modeling psychological phenomena to the enticing prospect of modeling basic neurological structures. They are incapable of performing constructive problem solving. but rather through engineering. be preenumerated.175 Building Intelligent Systems before the human brain will be able to understand its own cognition or to manage in an unaided fashion the complexity of computer systems that attempt to simulate human reasoning. This lesson has been learned the hard way. The networks can only transform input signals at the bottom layer to output signals at the top layer. 1991). Artificial neural networks came on the scene with tremendous fanfare and the promise that they would overcome the limitations of many of the cognitive models of intelligence (Hinton. by definition. however. many of them not. Artificial neural networks have proved to be extremely successful in a large number of classification tasks. Particular values of input data trigger specific nodes in the bottommost layer. and. the solution space (represented by the top layer of nodes) must.

Much of our work is manifest in the suite of tools and the supporting development methodology that we call “Protégé. have had the greatest effect on the field. • Use of the domain ontology to generate a knowledge-acquisition system with which users can enter the detailed content knowledge that defines a knowledge base for the particular application area (see Figure 5 above).” Protégé provides a workbench that assists in the construction of ontologies (see Figure 4 above). • Creation of mappings between the domain ontology and the method ontology that define how the abstract data and knowledge requirements of the selected problem-solving method are to be met using the particular concepts in the domain ontology. The importance of the engineer—who cares about the design. defining the major classes and relations that need to be modeled (see Figure 4 above).176 modeling complex systems ation of intelligent systems that may. The development of intelligent systems using Protégé proceeds in accordance with a sequence of well-defined steps: • Creation of a domain ontology that captures the structure of the application area. ultimately. however. and testing of novel manufactured entities—has eclipsed that of the psychologist. and in a variety of other tasks related to the construction of knowledge-based systems. 2004). • Selection of a problem-solving method that can automate the task at hand. The construction of intelligent systems is a problem in constructing artifacts. The goal of Protégé. emphasizing the development of tools and methods to support the engineering of these systems from domain ontologies and abstract problem-solving methods. Protégé is in many ways like the computer-assisted software-engineering (case) tools that developers use routinely in the construction of conventional computer programs (Sommerville. is to assist in the engineering of intelligent systems. in the use of ontologies to guide the entry of content knowledge (see Figure 5 above). The Protégé System My own laboratory has adopted an engineering approach to the construction of intelligent systems. implementation. .

There is no need for either the users or the participants to have innate beliefs about the particular ontology in advance—the ontology is merely a social convention that specifies the concepts in the application domain and the attributes of those concepts. For readers of this volume. Protégé has provided a framework for building knowledge-based systems where much of the ownership belongs to the community of developers at large. 2000). and use of domain ontologies and problem-solving methods and aids developers in linking the two types of building blocks together. There is no need for the problem-solving method in any way to approximate the approach that a person might take in solving the given task. where software engineers have augmented the original Web-browser metaphor with countless extensions that go well beyond the original vision for the World Wide Web. Similarly.. central role when developers use Protégé to build intelligent systems. indeed. is in studying the way in which developers might interact with systems such as Protégé and in . and the ways in which users choose to support their creativity by augmenting the basic Protégé system seem to be almost limitless (Musen et al. the choice of a particular problem-solving method to solve a given task is often pragmatic. It is built in a modular fashion that facilitates the development of “plug-ins” that can extend the functionality of the basic system. come as a disappointment that psychology does not play an obvious. The building blocks that Protégé users manipulate to create intelligent systems—like the intelligent systems themselves—are simply artifacts. selection. Like the Internet itself. adaptation. a domain ontology must define a structure for an application area in a manner that system builders and system users can agree on. Protégé is used regularly by thousands of developers worldwide. The growing community of Protégé users (more than 10. Building intelligent systems is inherently a creative process. it may.000 developers at the time of this writing) have implemented a wide range of plug ins—many of which address functions that our original development group could never have anticipated.177 Building Intelligent Systems Protégé assists in the creation. Protégé enables developers to work with domain ontologies and problem-solving methods as reusable building blocks for system development without the need to buy into any kind of psychological theory. rather than to the original programming team. Where psychology does take center stage. however. From the perspective of Protégé.

Now we live in a world were computers are commodities. who may need to examine and assemble these intricate building blocks. We are beginning to see the emergence of new methods to help in the visualization of complex knowledge bases and ontologies (Storey. no World Wide Web. the advent of the Internet and the Web have changed qualitatively the nature of information access in our society. visualize intricate relationships. there is an essential role that psychologists can play in analyzing the human problem solving required to build and debug intelligent systems from reusable building blocks. Noy. 1985) has its origins in cognitive psychology. Fergerson. We are already at the stage where system builders may incorporate into their work ontologies that contain hundreds of thousands of concepts and problemsolving methods of enormous computational complexity. Most interaction with computers over the next two decades would take place via punched cards. and assess system performance in ways that lend themselves to formal psychological study. there is an urgent need to provide more guidance to system builders. Developers of intelligent systems must manage complexity. Moving into the Next Generation of Development Tools At the time of the Dartmouth Conference in the summer of 1956. 1986) and computer-programming behaviors (Weinberg. Best. Although desktop publishing and even hypertext existed long before the Web. Musen. For our development methods to continue to scale up. The modern study of human-computer interaction (Norman & Draper. We take it for granted that we can launch a World Wide Web browser and read information stored on computers almost anywhere in the world. 2002).178 modeling complex systems evaluating how alternative design choices in the workbench might affect the ways in which system builders design ontologies and knowledge bases and map their ontologies to selected problem solvers. no computing grid. there was no Internet. in government laboratories. They are ubiquitous and interact with one another seemingly instantaneously over wide networks. but there is clearly much more work to be done. & Ernst. . and in large corporations. The first commercially successful personal computers were still 25 years away. Computers were housed in specially built rooms at academic institutions.

not by humans reading html pages. There is no model for human cognition that matches the behavior of an intelligent agent that determines dynamically how to optimize the processing of huge amounts of data in parallel and that takes advantage of arbitrary memory stores and processing units that may happen to be available anywhere in the world. the scientific community has become captivated by the possibility of extending the scope of intelligent systems throughout cyberspace and by the potential of intelligent systems that can draw on vast. There are no psychological correlates for intelligent problem solvers that are based on building blocks distributed across the Internet. knowledge bases. . 2004). and. Although. intelligent systems that will use the Semantic Web and the computing grid will be successful primarily because of their ability to exploit particular computational technology. The vision is that intelligent agents will be able to locate on the Internet the right building blocks to solve realworld tasks. intelligent systems can be fashioned out of components distributed throughout cyberspace (Benjamins. but instead by computers that will surf the Web in search of appropriate ontologies. creating the Semantic Web remains a research goal. and to use them in their problem solving. at present. Like Deep Blue. Just as users can translate print documents into hypertext markup language (html) and post them as Web pages anywhere on the Internet. 2003). and problem-solving methods (Berners-Lee. Problem-solving methods such as heuristic classification can be made available as Web services. a World Wide Web designed for access. The result will be what investigators have come to call the Semantic Web. 2001). to access those building blocks. this new generation of intelligent systems will still use the same kinds of ontologies. developers now can “publish” ontologies and knowledge bases on the Web and can make them available to any computer anywhere in the world. suddenly. Nevertheless. but now those building blocks will be situated throughout the Internet. & Lassila. The next generation of intelligent systems will comprise the same kinds of components that have become popular as building blocks for creating stand-alone systems. Hendler. not because they will model known cognitive processes.179 Building Intelligent Systems The Internet and the Web are about to change qualitatively the nature of intelligent systems. virtual libraries of resources that have no physical limitations and to which a global community of developers can contribute (De Roure & Hendler.

and of new methods that can make the use of those tools cognitively tractable. developers will face new challenges in conceptualizing ontologies and navigating their mappings to the various Web services that will provide problemsolving support. Conclusion Early developers of knowledge-based systems viewed their work as a problem in applied psychology. Their belief was that.180 modeling complex systems knowledge bases. of new tools that support those principles. it was difficult for rule-based systems to scale up to the requirements of many real-world tasks. and problem-solving methods that have become important in the engineering of current intelligent systems. As rule-based systems were reaching the peak of their popularity. The availability of the entire Internet as a frontier for the creation of new kinds of intelligent applications is extraordinarily inspiring. Expanding the scope of automated intelligence from programs that run on stand-alone computers to applications that will draw on the resources of the Semantic Web will require a new class of development tools and engineering methods. As those models individually increase in scope and collectively become distributed in cyberspace. such as those now developed using Protégé. however. the better one’s model of an actual cognitive process in humans. In practice. The Semantic Web will become a reality only with the advent of appropriate engineering principles. The Internet alone is not enough to guarantee our success. The principal issue in building intelligent systems is one of managing complex models and their interrelations. It is most helpful to view the construction of intelligent systems . however. The need to visualize the dynamic flow of information across networks and to debug systems when their components are widely distributed will stimulate the creation of a new generation of development tools and techniques. the better would be the computational artifact that would attempt to solve the same task in the world. Rule-based systems offered appealing computational architectures because much of human problem solving could be modeled in terms of production systems. psychologists were strongly questioning the claim that such systems could even be viewed as appropriate cognitive models in the first place.

& McDermott.181 Building Intelligent Systems as a problem in engineering. The advent of the Semantic Web will entail the creation of new kinds of intelligent systems from the same kinds of components that we use now—but the components will be distributed in cyberspace. (1984). r1 revisited: Four years in the trenches. (2003).Stanford. References Anderson. rather than in psychology. Intelligent systems can be created from reusable building blocks that include (1) domain ontologies. Note The Protégé system is supported as a national biotechnology resource by the National Library of Medicine under grant lm007885. Development of Protégé has been supported by contracts from the National Cancer Institute and the Defense Advanced Research Projects Agency and by grants from the National Library of Medicine and the National Science Foundation. and problem-solving methods provide services. Web services solve problems. Benjamins. 192–210.edu). V. A. J. Bachant. J. 5(3). Psychologists will have an invaluable role to play in helping design and evaluate new methods and tools for piecing together the widely distributed building blocks. ieee Intelligent Systems. 76-77. Psychological Review. Protégé is the most widely used workbench for building intelligent systems from reusable components and demonstrates the utility of this approach. (2) knowledge bases that provide content knowledge structured by those ontologies. . 94. 21–32. and (4) mappings that relate the concepts in domain ontologies to the knowledge and data requirements of problem-solving methods. (3) problem-solving methods that automate stereotypical tasks such as heuristic classification and constraint satisfaction. Skill acquisition: Compilation of weak-method problem solutions. (1987). There is an exciting challenge to build a new generation of usable development tools that will address the enormous complexity that will result when systems can be assembled dynamically from computing resources that are scattered across the Internet.. R. 18(1). Currently. J. The system may be freely downloaded from our Web site under an open-source license (see http://protege. ai Magazine.

Computer. 161–204). N. E. 79. Rothenfluh. Scientific American. Shahar. Expert systems research. Eriksson. 10–19. W. J. A. L. 246. Heuristic classification.. B. (1994). Dreyfus. Bowker. J. A. Artificial Intelligence. W. W. (1999). Rule-based expert systems: The mycin experiments of the Stanford Heuristic Programming Project. P. (1993). H. Generic tasks in knowledge-based reasoning: High-level building blocks for expert system design.. A. Gennari. & Tu. M.. W. A. & Musen.). R.. (1981). Protocol analysis: Verbal reports as data (Rev...). M. J. No silver bullet: Essence and accidents of software engineering. (1989). Annals of the New York Academy of Sciences. (1997). 91–107. Handbook on ontologies (pp. Viewing knowledge bases as qualitative models. D. De Roure.. Categories of human learning. In A. O. R. International Journal of HumanComputer Studies. Clancey. Puerta. G. 321–341). O. W. (1964). New York: Academic. A. The evolution of Protégé: An environment for knowledge-based systems development. ieee Intelligent Systems. 23–30. P.). & Star. Cambridge ma: mit Press. H. Reading ma: Addison-Wesley.. (1983).. Melton (Ed. A. Ericsson. & Shortliffe.. Buchanan. Hendler. Musen. S. Studer (Eds. Ontologies in support of problem solving. 261–268. H. 65–71. Situated cognition: On human knowledge and computer representations. K. New York: Cambridge University Press. A. (1995). 41. (2001). Clancey. 18–23. The appeal of tacit knowledge in psychological explanation. 293–326. . Noy.. Crubézy. J. J. A. 89–123. 27.182 modeling complex systems Berners-Lee. 627–640. Clancey. Task modeling with reusable problem-solving methods. J. (1986). Duda. Tu. H. International Journal of Human-Computer Studies. Crubézy. & Simon. E. Fodor. W. 19(1). S. Cambridge ma: mit Press. A. 20. S. (1968). The Semantic Web. T. & Lassila. From micro-worlds to knowledge representation: ai at an impasse. F. E. 289– 350. Journal of Philosophy. R. F. (2003). Y. (2004). 65. ieee Expert. ieee Expert. A. Science. Haugland (Ed.). Mapping domains to methods in support of reuse. Fergerson. M. Cambridge ma: Bradford. Fitts. Perceptual-motor skill learning. 58(1). In S. M. H.. & Hendler. H. Brooks. T. B. Berlin: Springer.. & Shortliffe. Chandrasekaran. W. Staab & R. In J. E. Gennari.. Tu. 279(5). W. E-science: The grid and the Semantic Web.. C. H. (1984). Mind design (pp. J. M. ed. A.. S. Knowledge engineering: The applied side of artificial intelligence. Feigenbaum. L. 220(4594). H. 34–43. (1985). E.. Sorting things out: Classification and its consequences. 1(1).. (1987). Grosso. (2004). M. 399–424.. & Musen.. Eriksson. (1984). J. 4(2).. G. M.. Artificial Intelligence. & Musen.

Boston: Kluwer Academic. China. M. W. A.. (1982). W. Fergerson. 24(2). (2004).. 3–15). Cognitivism and situated action: Two views on intelligent agency. (1998). J. Formal ontology in information systems: Proceedings of fois ’98 (pp. Lyons. Amsterdam: ios. Boca Raton fl: crc... Lederberg.html. L. R. 77–97. A. (1980).. & Lindsay.). The disappearance of introspection. 4(2). (1998). 135–183. M. Amsterdam: ios. August. 46. Deep blue: An artificial intelligence milestone.. 37. 18–22). McDermott. J. (1963). New York: McGraw-Hill. & Shannon. 15(5). W. New York: Springer. In J. (1996). In S. L. T. E. A. Musen. & Simon. Connectionist symbol processing [special issue]..). E. Applications of artificial intelligence for organic chemistry: The dendral project. Knowledge management handbook.. In Proceedings of the Conference on Intelligent Information processing (iip 2000) of the International Federation for Information Processing Sixteenth World Computer Congress (wcc 2000). J. Grosso. gps: A program that simulates human . (1980). Ontology-oriented design and programming. A.). Guarino (Ed. A.stanford. Krueger.. Formal ontology and information systems.. Domain ontologies in software engineering: Use of Protégé with the eon architecture. H. Computers and Artificial Intelligence.. N. Knowledge engineering and agent technology (pp.. Menzies. What kind of system should an expert be? Journal of Medicine and Philosophy. McDermott. G. (1955). & Clancey. M. McCarthy. 19. K. C. Feigenbaum. J. & Gennari. Liebowitz. acm Computing Surveys. R. r1: A rule-based configurer of computer systems. Preliminary steps toward a taxonomy of problemsolving methods. (1998). Cognitive Science. M. Treur (Eds. Newborn. 3–16). 2000 (pp. A. (2000). N. 49(6).edu/jmc/history/ dartmouth/dartmouth. 39–88.. (1991). Cambridge ma: mit Press. Laxenburg: International Federation for Information Processing. E. J. P.. Buchanan. Automating knowledge acquisition for expert systems (pp. Crubézy. Methods of Information in Medicine. (1999). J. (1992). Physical symbol systems. Marcus (Ed. 767–769. N. Artificial Intelligence. A. M. Cuena. Retrieved from http://www-formal. Newell. Noy. Musen. Minsky. Kushmerick. W. Garcia. N. (1983). Johnson. F. 8. & J. A. In N. Beijing. B. G. International Journal of HumanComputer Studies. E. Rochester.183 Building Intelligent Systems Guarino. 1–258. Demazeau. The challenge of situated cognition for symbolic knowledge-based systems. M. Software reuse. Component-based support for building knowledge-acquisition systems. 225–256). Artificial Intelligence. Newell. (1986). A proposal for the Dartmouth Summer Research Project on artificial intelligence. (1988). (2002). Sutherland. C. Hinton. 393–417. J. J. Y. W. Musen. 540–550. 131– 183. H.

International encyclopedia of the social and behavioral sciences. A. C. 231–259. W.). Creating the domain of discourse: Ontology and inventory. (2004). Tversky.. K. Psychological Review. Newell. (Eds. Production systems in cognitive psychology. In the age of the smart machine: The future of work and power... D. & Plantinga. M.. S. The psychology of computer programming. & Ernst. (1972). T. B. M. S. Anjewierden. Storey.. Norman.. Boston: Addison Wesley. & Draper. & Kahneman. L. Science. M. Stefik. Fergerson. (2002).. (2000). 235–250. Handbook on ontologies. Software engineering (7th ed. Weinberg. (2004). Musen. O. W. D. A. Noy. Winston. (1985). Cambridge ma: mit Press. Nisbett. & Wilson. Suchman. N. A. Schreiber.. R. E. Best. I. A. . W. M. (1987). (Eds.. Cambridge ma: mit Press. Zuboff. & Studer. R. H.184 modeling complex systems thought. (1987). Introduction to knowledge systems. International Journal of Man-Machine Studies.. Englewood Cliffs nj: Prentice-Hall. Computers and thought (pp. & Prendergast. (Eds. G. D. H. van de Velde. (1986). G.. Plans and situated actions: The problem of human-machine communication. 239).. E. New York: McGraw-Hill. Judgment under uncertainty: Heuristics and biases. A.). de Hoog. (1988). Baltes (Eds. New York: Elsevier. 1124–1131. Human problem solving. N. A. (1995). Jambalaya: An interactive environment for exploring ontologies. New York: Van Nostrand Reinhold. Knowledge engineering and management: The Common kads methodology. H. Shadbolt. B. 84. (1986). S. & Wielinga. Englewood Cliffs nj: Erlbaum. (1977). J. Smelser & P. S. Feigenbaum & J. R. Sommerville. In Proceedings of the Seventh International Conference on Intelligent User Interfaces (p. San Francisco: MorganKaufmann. Telling more than we can know: Verbal reports on mental processes. & Simon. A.. R. Young. M. A. P... N. User-centered system design. The ai business: Commercial uses of artificial intelligence.). In E. New York: acm.)..). P. (2001). In N.-A. 27. A. Akkermans. Staab. Feldman (Eds.. 185. R. Cambridge: Cambridge University Press.). 279–293). Berlin: Springer. (1974). F. New York: Basic. Regoczei..

Army Research Institute Organizations seeking to foster teamwork in fluid environments are increasingly turning to teams as a preferred performance arrangement.Fostering Team Effectiveness in Organizations: Toward an Integrative Theoretical Framework Eduardo Salas. At one time in history the sun never set on the British Empire. As organizations become more comfortable with. C. today.S. Multinational corporations facing unprecedented domestic and foreign competition are utilizing teams as a means to organize production and service around a 24-houra-day process. . Kevin C. Goodwin University of Central Florida. University of Central Florida. U. and dependent on. University of Central Florida. technology. and Gerald F. however. The ongoing proliferation of teams is due in part to the rise of global market opportunities. Once restricted to collocation. The spread of teams is also due in part to the ongoing technological revolution. technology now makes it possible and profitable for team members to be distributed in space-time. Shawn Burke. we believe that opportunities for team-based collaboration will abound. the sun never sets on Intel as software engineers collaborate around the globe in order to meet organizational financial forecasts. Stagl.

there is a concomitant increase in team-effectiveness theory building. Hollenbeck.186 modeling complex systems Unfortunately. Later in this chapter we extract a representative sample and expound on how each of these initiatives has taken a different path to framing team effectiveness. Those concerned with developing a deeper understanding of teamwork in order to promote team synchronicity have proposed a wide variety of theoretically grounded models and frameworks that address both team development and team effectiveness. 2003. Stagl. The proliferation of teams and the concomitant focus on orchestrating team effectiveness so that team members. and team maturation (Morgan. many recent tragic incidents (e. the space shuttle Columbia) have been attributed to breakdowns in teamwork. history has repeatedly shown that team performance is an elusive.. developing a thorough understanding of how teams interact in a synchronized fashion to achieve shared goals is critical to meeting short-term tactical objectives and to cultivating long-term strategic success. including midpoint effects (Gersick. & Zaccaro. Ilgen. in a recent literature review. These failures are due in part to the often fluid and sometimes chaotic environment in which teams operate. and complex phenomenon (Salas. Mathieu. Hackman. Therefore. 2003. Edmondson. during the past century a plethora of research initiatives have been undertaken with the goal of more fully illuminating the complexities of teamwork and. temporal entrainment (Ancona & Chong. 2004). Bell & Kozlowski. Occidental’s Piper Alpha platform. 2000. 2004. episodic team processes (Marks. Although dozens of intriguing conceptualizations of teamwork have surfaced in the past 25 years. 1999). and organizations can reach their full potential is supported by a growing body of theoretical and empirical research (see Belbin. We suspect that this body of empirical evidence will only continue to expand exponentially in step with the ongoing proliferation of teams. Beyerlein. Thus. the uss Stark. Indeed. advancing the science of teams. Salas. 1994).g. the uss Vincennes. we identified more than 800 articles and chapters that present empirical evidence addressing some aspect of team effectiveness. 2001). however. teams. there has not been a compre- . In fact. Cooper & Robertson. multiteam systems. Johnson. 1990. dynamic. & Beyerlein. 1988). thereby. 1993). In addition to the growing number of empirical investigations noted above. & Sego. & Glickman. 2002. it is no surprise that. Major. & Burke.

models. 2001. & Smith. 1999. 1992).187 Team Effectiveness in Organizations hensive review of team-effectiveness models and frameworks conducted since the early 1990s (see Campion. in this chapter we present a multilevel. Specifically. 1999. Likewise. we believe that undertaking a comprehensive review and integration of team-effectiveness models and frameworks will be beneficial to team researchers by revealing common ground on the substantive team constructs of interest. 1993. If there are. there are several problems associated with having such wide diversity in the definition and measurement of team constructs. Kozlowski & Bell. West. teamwork. team performance. and team effectiveness. and much has been learned about teams (see Campbell & Kuncel. 2003). Once these preliminary issues are addressed. and frameworks in the current body of relevant literature. organizations have changed in many ways. Edmondson. Also. & Salas. & Higgs. indeed. then it is likely that they include constructs. Hackman. Our review of the team literature resulted in the identification and subsequent integration of 138 initiatives that modeled or framed aspects of team performance or effectiveness. Since that time. there are clear positives associated with maintaining a common language when engaging in team research (Salas & Cannon-Bowers. Tjosvold. and proposed construct relations that incrementally add to our understanding of the complex picture of team effectiveness. multidisciplinary synthesis of team-effectiveness models and frameworks advanced during the last 25 years. Given the above arguments. Beard. 1990b). accumulating and integrating what is known to date are challenges confronting both team researchers and industrial organizational psychology (Campbell. construct operationalizations. Ilgen. This state of affairs suggests that there are a growing number of unincorporated team-effectiveness research findings. Tannenbaum. Medsker. Therefore. we describe a set of criteria utilized to guide our literature review. a number of recent research initiatives undertaken to frame teamwork that were not already addressed by prior comprehensive reviews. 2000). in this chapter we begin by defining the terms team. This endeavor was undertaken to collect and integrate current comprehensive models and frameworks of team effectiveness in order to present a more coherent picture of the factors constituting and impinging on teamwork. 2003. Furthermore. A representative sample of 11 models and frameworks that . 2002.

For example. we discuss a few practical applications (i. ultimately. and long-term goals that can serve to guide future team research. Clarification of Key Constructs The current diffusion of existing conceptualizations of team constructs mandates that. and settings that our team-effectiveness framework generalizes to and across (Cook & Campbell. coordinated effort among all team researchers is needed to accomplish it. multilevel framework of team effectiveness. The descriptions that follow also offer insight into the persons. training.e. we expect that our framework will serve as a departure point for team researchers concerned with addressing the salient features and conditions that constitute and affect teamwork in specific types of teams. teamwork. and team effectiveness. we take a moment to clarify and elaborate on some of the qualities that characterize teams. team performance. This chapter concludes with a call for the generation and. widespread adoption of a set of short-. we advance an integrative. at a broad level our proposed framework is tailored to teams characterized as having members who hold shared goals and act interdependently to achieve them. albeit preliminary. mid-. 1979). Much work remains to be done. Furthermore. We believe that team research has reached its teenage years and that tackling the substantive issues that remain to be addressed by team researchers will help curtail the haphazardness that has characterized the selection of constructs investigated up until this point in time.188 modeling complex systems typify cutting-edge team advancements are reviewed in detail. Thus. measurement. in this section we attempt to foster a shared understanding between the authors and the reader of the conceptualizations of these related but unique phenomena that are applied throughout this chapter. . After describing the novelties of our approach to framing team effectiveness. before advancing further. staffing) and how they can be more fully appreciated when viewed through the lens afforded by a theoretically grounded framework.. The advanced framework was designed as a simple yet rich heuristic that reflects the current state of the art within the team domain. On the basis of this research initiative. times. and a collective.

Dickinson. a plethora of other perspectives have been advanced during the past 25 years to define team (Guzzo & Dickson. marketing teams. (4) have shared or common goals. forensic-science teams. construction teams. and (5) hold meaningful task interdependencies. oil-rig crews. 1996). (4) intense time pressure. submarine teams. teamwork Teamwork is defined in the online Merriam-Webster dictionary as work done by several associates with each doing a part but all subordinating personal prominence to the efficiency of the whole. and (5) severe consequences in cases of error (Orasanu & Salas. (2) no “optimal” answers. cockpit crews. 1992). including emergencymanagement teams. modern operational environments are characterized by a historically unparalleled accelerating rate of change that mandates team flexibility. and it is (9) embedded within an organizational/environmental context that influences and is influenced by ongoing processes and performance outcomes (Salas. such as financial-analyst teams. & Tannenbaum. In fact. (3) information overload. archaeological teams. All these team exemplars often perform in fluid environments typified by (1) rapidly evolving and ambiguous situations. This definition exemplifies action teams that perform some of the most critical work in the global economy. and fast food teams. We suspect that operational environments will increase in fluidity in step with emerging technologies. adaptability. It also typifies many types of teams appearing in more traditional business settings. sports teams. and resilience. it (6) is hierarchically structured and (7) has a limited life span. in it (8) expertise and roles are distributed. 1993). For our purposes we define team as follows: it is a complex entity consisting of (1) two or more individuals (2) who interact socially and (3) adaptively. Converse. control-tower teams. In addition to this conceptualization. command and control military teams. . and space-exploration teams. research-and-development teams.189 Team Effectiveness in Organizations teams The term team is defined in the online Merriam-Webster dictionary as a number of persons associated together in work or activity. gene-sequencing teams. medical teams.

1994). skills. In other words. not surprisingly. Acton. Teamwork consists of the knowledge. and mission. and other characteristics (ksaos) underpins and enables the enactment of the steps constituting a given process. adaptive performance and the completion of taskwork objectives” (pp. Therefore. For example. However.g. 2005. Stevens & Campion. and Klein (2004): “Teamwork is a set of flexible behaviors. actions. Furthermore. In the previous paragraph. Tannenbaum. Fleishman & Zaccaro. we noted that a set of knowledge. as it would be difficult to imagine any interdependent collective that did not on occasion require communication or coordination. Salas. 2001. teamwork is a set of interrelated thoughts. Marks et al. Dickinson & McIntyre.. Sims. the ksas presented in the remainder of this section are most appropriate for teams with characteristics similar to those mentioned above. what constitutes teamwork? The answer to this question. guided-missile teams). 1995. McIntyre & Salas. skills.. Sims. we defined teamwork as adaptively enacted processes undertaken to achieve shared goals. objectives. Essentially. As noted in the previous section. 497–498).. the question remains as to which ksas are central to teamwork. and feelings that combine to facilitate coordinated. The four teamwork skills advanced by McIntyre and Salas are (1) mutual performance moni- . 1998. cognitions. This definition illustrates the core of what we mean when we refer to teamwork throughout the remainder of this chapter. and attitudes (ksas) that are displayed in support of one’s teammates. 1997.190 modeling complex systems Although this definition is elegant. attitudes. & Mathieu. is highly contingent on the type of team under consideration. Marks. & Burke. 1992. Zacarro. & Volpe. 1995. and attitudes that interact to achieve desired mutual goals and adaptation to the changing internal and external environments. Salas. this chapter emphasizes a particular type of team (e. 2000. Smith-Jentsch.g. Zeisig. There have been many systematic initiatives undertaken to illuminate the ksas that constitute teamwork (see Cannon-Bowers. & McPherson. with tight interdependencies). However. this caveat is not meant to imply that some of the ksas advanced below do not generalize to all types of teams. McIntyre and Salas (1995) delineated teamwork skills and performance norms from their programmatic work with navy tactical teams (e. a more detailed one has been advanced in Salas. one that is hierarchically structured.

and (13) teamwork skills. (5) team mission. a thorough description is provided in a later section reviewing extant theoretical models and frameworks of team effectiveness. (4) leadership/team management. suggest that teams require several knowledge competencies. This effort resulted in the advancement of the ksa competencies proposed to underlie teamwork in organizations. thus. (4) team cohesion. (6) coordination. but dynamically. (2) shared situational awareness. (1995) made great strides in advancing our collective understanding of the core content constituting teamwork. (3) shared vision. (3) performance monitoring and feedback. (2005) has made a significant contribution to our collective understanding of the dynamic interdependencies existing among the components of teamwork. and resources. (6) relation to the larger organization. (2) collective efficacy. (12) boundary-spanning role. including (1) team orientation. not in isolation. (3) role responsibilities. (11) accurate problem models. (7) task sequencing. Another approach to understanding teamwork was taken in Cannon-Bowers et al. and Salas et al. (4) shared task models. (3) closed-loop communication. they propose that teams require a variety of attitudinal competencies. including (1) adaptability. and (4) backup behavior. Cannon-Bowers et al. It will satisfy for now to state that teamwork ksas operate. (5) interpersonal relations. (1995). . (9) procedures for task accomplishment. Marks et al. (6) mutual trust. (7) communication. (8) team role-interaction patterns. (2001). including (1) cue-strategy associations. and recursively as they unfold over time to emerge as team performance. (7) collective orientation. (2) teammate characteristics. While McIntyre and Salas (1995) and Cannon-Bowers et al. These three particular initiatives are critical enough to warrant closer inspection. the more recent research of Dickinson and McIntyre (1997). (10) accurate task models. (2) feedback. Furthermore. and (8) decision making.191 Team Effectiveness in Organizations toring. simultaneously. and (8) a belief in the importance of teamwork. Specifically. (5) interpersonal relations. norms. These researchers collected and synthesized prior research addressing teamwork. skills that can be clustered into eight higher-order dimensions. Also identified were two performance norms: (1) a team’s self-awareness and (2) the fostering of within-team interdependence. and. Finally. objectives. they suggest that teams require dozens of skills.

This is because operationalizations of performance are contingent on the timing of measurement.e. and their colleagues’ assertions provide a basis for understanding human performance at a conceptual level.e. in practice the timing of performance measurement often leads to confusion. However. cross-sectional research is typified by the measurement of performance outcomes or the result(s) of expended energy (e. performance has been defined within the domain of industrial organizational psychology as goal-directed behavioral or cognitive action (Campbell. p. We assert that team performance consists of the display by one or more team members of taskwork competencies (e. Dunnette. Campbell. action) and effect (i. This speaks to the issue of what constitutes team performance. backup behavior).. Dunnette. & Weick. Specifically.e. team performance remains either a behavioral or a cognitive action and outputs of these actions are best conceptualized as performance outcomes. 1970. 1990a. . 1963).. adaptation).g.g. 2003). a point echoed elsewhere in recent scientific literature (Kozlowski & Klein. accomplishment). a record can be compiled of actions and outcomes at multiple levels in the conceptual space. Lawler. What is noteworthy in this definition is the inclusion of both cause (i. While Campbell. In fact. regardless of whether longitudinal or cross-sectional research is conducted. Campbell (1990a. 2000). and integrated team-level action (i. Our point is that. In both longitudinal and cross-sectional team research.g. number of products produced). it is often conceptualized as a series of actions. when performance is examined longitudinally. 704) states: “Performance is not the consequence(s) or result(s) of action. This line of thinking leads to the conclusion that team performance is a multilevel phenomenon.. coordination.. running a computer program). it is the action itself.. This mingling of antecedent and consequence is rejected by those researchers concerned with modeling human performance.” Following this line of thinking. teamwork competencies (e.. prior research has often indiscriminately conceptualized performance as either a series of behavioral or cognitive actions that unfold over time or the outcomes resulting from those actions (Kozlowski & Bell. Thus. Dunnette.192 modeling complex systems team performance Performance is defined in the online Merriam-Webster dictionary as either (a) the execution of an action or (b) something accomplished.

193 Team Effectiveness in Organizations Essentially, team performance can be conceptualized as a bottom-up emergent process unfolding from individuals to teams. In other words, team performance begins with team members and emerges upward to teams. However, team performance is also simultaneously a top-down process. This line of reasoning suggests that higher-level context has a direct or moderating effect on team performance. However, this does not preclude the foundation of team effectiveness residing in the cognition, behavior, and affect of individual team members. In fact, team members’ thoughts, actions, and feelings via social interaction, dyadic role exchanges, and amplification have emergent properties that manifest at higher levels (i.e., bottom-up emergence). The assertion that team performance is both a top-down and a bottom-up multilevel phenomenon that emerges over time is consistent with the guiding tenets of general systems theory (von Bertalanffy, 1956) and its variants, which collectively acknowledge that teams are hierarchically nested, interdependent, open systems (Kozlowski & Klein, 2000).

team effectiveness
Effectiveness is defined in the online Merriam-Webster dictionary as producing a decided, decisive, or desired effect. Thus, effectiveness is, not the outcome produced from team performance, but rather the result of a judgmental process whereby an output is compared to a subjective or an objective standard. Similarly, effectiveness has been defined within industrial organizational psychology as the evaluation of the results of performance (Campbell, 1990a). While this conceptualization of team effectiveness seems straightforward, Cohen (1994) disagrees: “Most group effectiveness researchers think about effectiveness as a multidimensional construct, but do not agree as to the criteria of work group effectiveness” (p. 71). For example, team effectiveness has been operationalized as (1) archival records of productivity, (2) self-report measures of employee satisfaction, and (3) manager judgments of effectiveness (Campion et al., 1993). Subsequent research reported in Campion, Papper, and Medsker (1996) operationalized team effectiveness as (1) immediate managers’ judgments at two points in time, (2) senior managers’ judgments, (3) peer managers’ judgments, (4) subordi-

194 modeling complex systems nates’ judgments, (5) archival records of satisfaction, and (6) archival records of performance appraisals. A wide variety of other operationalizations of team effectiveness have also been advanced. For example, Cohen (1994) adopted a tripatriate perspective asserting that the variables contributing to team effectiveness can be clustered into three broad categories, including (1) team performance, (2) team members’ attitudes about quality of work life, and (3) withdrawal behaviors. According to Cohen (1994), each of these three global factors encompasses a number of effectiveness-related variables. Specifically, the performance factor includes (1) controlling costs, (2) increasing productivity, and (3) increasing quality. The team members’ attitudes factor incorporates (1) job satisfaction, (2) team satisfaction, (3) satisfaction with social relationships, (4) satisfaction with growth opportunities, and (5) organizational commitment. The withdrawal behaviors include both (1) absenteeism and (2) turnover. Hackman (1987) also asserts that group effectiveness can be conceptualized in terms of three components. The first component consists of a judgment by those stakeholders who review the work of teams in terms of whether it meets their standards for quality and quantity. The second is whether the needs of group members are satisfied by their team participation. The third is whether group interaction has served to maintain or strengthen the group’s ability to work together at some future date. Finally, Sundstrom, DeMeuse, and Futrell (1990) suggest that team effectiveness is composed of (1) managers’ and customers’ judgments about the acceptability of performance and (2) team viability, where team viability is defined as commitment on the part of team members to continuing to work together. We suspect that some of the ambiguity underlying the nature of team effectiveness lies in an incomplete specification of effectiveness across levels and time. To date, most team research has operationalized and measured effectiveness at both the individual (i.e., team members’ self-reports of satisfaction/commitment) and the team (i.e., ratings of team coordination) levels. In other words: “Team effectiveness is different from individual effectiveness” (Yetton & Bottger, 1982, cited in Tannenbaum et al., 1992, p. 117). Furthermore, individual effectiveness and team effectiveness are distinct from, but interdependent with, multiteam-system and organizational effectiveness. Certainly, a comprehensive discussion of team effective-

195 Team Effectiveness in Organizations ness must extend beyond the individual and team levels of analysis typically addressed in research in order to incorporate effectiveness factors at the multiteam system and organizational levels. This is because the criteria relied on by stakeholders to form effectiveness judgments are contingent on whether the system under consideration is the team member, the team, the multiteam system, or the organization. In addition, team effectiveness judgments are intertwined with temporal factors such as team maturation. As a team and its members mature, a given criterion, once heavily relied on by senior stakeholders in forming their judgments about the effectiveness of a particular aspect of the team, may no longer be salient. Moreover, the type of criteria adopted may change as teams pass through the early stages of formative development en route to achieving process gains. For example, during the norming stages of team development, judgments of effectiveness may be directed at the accuracy of team-member mental models. However, as the team matures, the emphasis may shift from individual mental models to shared and compatible mental models that serve as a cognitive reservoir for fueling coordinated adaptation to novel challenges. As noted above, the criteria adopted by stakeholders to form their effectiveness judgments change over time. Specifically, we noted that temporal dynamics such as team maturation influence what criteria are relevant and how those criteria are weighted. The criteria utilized by stakeholders to form their effectiveness judgments may also vary over time as a function of the rater’s level in the organization. In this example, the stakeholder’s organizational level is meant to convey expertise or systems thinking (see Senge, 1990). Certainly, a strong argument can be made that, as incumbents navigate a given organizational hierarchy and develop a history with the target unit under consideration, there is a concomitant evolution in their perspective on what and how much of a given set of qualities constitute effectiveness. The above-noted line of reasoning is supported by empiricalresearch evidence collected from the naturalistic decision-making paradigm suggesting that expert decision makers (e.g., firefighters, chess masters, police officers) emphasize different criteria than novices do when forming judgments about stimuli (see Klein, 1993). Thus, it may be plausible to conceive of an interaction between the

196 modeling complex systems target unit being rated and the level of the stakeholder (i.e., rater) when predicting both (1) the types of effectiveness criteria attended to and (2) how chosen criteria are weighted to form global effectiveness judgments. If these arguments are valid, they have implications for the design of performance-appraisal systems and for rater training. Thus, these assertions warrant closer attention in future theory building, research, and practice. Essentially, effectiveness is a value judgment that is influenced by a number of factors. In this subsection, we noted how a particular criterion or set of criteria can change as a function of (1) the target unit’s level in the conceptual space, (2) the target unit’s level of maturation, and (3) the idiosyncratic characteristics of raters such as systems thinking. Although these three factors are undoubtedly important in understanding stakeholder judgments of team effectiveness, a number of other pertinent issues also exist. For example, a host of individual (e.g., latent cognitive resources, available cognitive resources, biases), team (e.g., performance phase), situational (e.g., time pressure, novelty), and organizational (e.g., culture, reward systems) factors also influence judgments of effectiveness. In sum, the substantive similarities and differences of the content constituting and the process underlying effectiveness judgments across system levels and time remain to be addressed in future research endeavors.

Extant Team-Effectiveness Models and Frameworks
In the preceding section we broadly defined some of the more salient characteristics of teams, teamwork, team performance, and team effectiveness for the purpose of framing our assumptions regarding the nature of these interrelated phenomena. Furthermore, we contextualized our thinking about these phenomena by describing the complex and fluid nature within which teamwork often occurs. Thus, the foregoing section served to set the stage for our thinking about teams as presented throughout the remainder of this chapter. In this section we describe the results that were generated from a review of the team literature. The literature review, guided by our above-noted assumptions, resulted in the identification of 138 teamperformance and -effectiveness models and frameworks that met

197 Team Effectiveness in Organizations Table 1. Team Models and Frameworks
Aldag & Fuller (1993) Alper, Tjosvold, & Law (2000) Ancona & Caldwell (1992) Annett & Cunningham (2000) Argote & McGrath (1993) Arrow, McGrath, & Berdahl (2000) Avolio, Kahai, Dumdum, & Sivasubramaniam (2001) Baldwin & Bedell (1997) Balkundi & Harrison (2004) Barrick, Stewart, Neubert, & Mount (1998) Barry & Stewart (1997) Cooper, Shiflett, Korotkin, & Fleishman (1984) Coovert & Dorsey (2000) Cuevas, Fiore, Salas, & Bowers (2004) Cummings (1978) Deeter-Schmelz, Kennedy, & Ramsey (2002) de Jong, Bouhuys, & Barnhoorn (1999) Denison, Hart, & Kahn (1996) DeSanctis & Poole (1994) Devine & Clayton (1999) Dickinson & McIntyre (1997) Gibson, Zellmer-Bruhn, & Schwab (2003) Gist, Locke, & Taylor (1987) Gittell (2000) Gladstein (1984) Goodman, Ravlin, & Schminke (1987) Gupta, Umanath, & Dirsmith (1999) Guzzo & Shea (1992) Hackman (1983) Hackman (1987) Hackman (1990)

Doolen, Hacker, & Hackman & Oldham (1980) Van Aken (2003) Beck (2002) Drexler, Sibbet, & Forrester (1988) Hall (2001) Bolman & Deal (1992) Driskell, Radtke, & Salas (2003) Haward et al. (2003) Brandes & Weise (1999) Druskat & Kayes (1999) Heinemann & Zeiss (2002) Brodbeck & Greitemeyer (2000) Durham & Knight (1997) Helmreich & Foushee (1993) Bunderson & Sutcliffe (2002) Edmondson, Roberto, & Hertel, Konradt, & Watkins (2003) Orlikowski (2004) Burke, Stagl, Salas, Pierce, & Entin & Serfaty (1999) Higgins & Routhieaux (1999) Kendall (2006) Campion, Medsker, & Erez, LePine, & Elms (2002) Higgs & Rowland (1992) Higgs (1993) Campion, Papper, & Fleishman & Zaccaro (1992) Hinsz, Tindale, & Vollrath Medsker (1996) (1997) Choi (2002) Flemming & Monda-Amaya (2001) Hoegl, Praveen, & Gemuenden (2003) Cohen (1994) Flood, Hannan, Smith, Turner, West, Hogan, Raza, & & Dawson (2000) Driskell (1988) Cohen & Bailey (1997) Furst, Reeves, Rosen, & Hollenbeck et al. (1995) Blackburn (2004) Cohen, Ledford, & Gersick (1988) Hollingshead & McGrath (1995) Spreitzer (1996) Janz, Colquitt, & Noe (1997) Millitello, Kyne, Klein, Shea & Guzzo (1987) Getchell, & Thordsen (1999) Jung, Sosik, & Baik (2002) Mischel & Northcraft (1997) Sheard & Kakabadse (2001) Kahai, Sosik, Avolio (1997) Morgan, Dailey, & Kulisch Sheehan & Martin (2003) (1976) Karau & Kelly (1992) Morgan, Salas, & Smith-Jentsch, Zeisig, Acton, Glickman (1994) & McPherson (1998) Kelley (2001) Navarro (1994) Sonnentag, Frese, Brodbeck, & Heinbokel (1997)

198 modeling complex systems Table 1. (cont.)
Kirkman & Shapiro (1997) Kirkman, Tesluk, & Rosen (2004) Kline (2001) Kline & MacLeod (1996) Kolodny & Kiggundu (1980) Kozlowski, Gully, Nason, & Smith (1999) Kristof, Brown, Sims, & Smith (1995) Kuo (2004) Leon, List, & Magor (2004) Losada (1999) MacMillan, Entin, Entin, & Serfaty (1994) Marks, Mathieu, & Zaccaro (2001) Mathieu, Goodwin, Heffner, Salas, & Cannon-Bowers (2000) McGrath (1984) McGrath (1991) McGrath (1997) McGrath, Arrow, & Berdal (2000b) McGrew, Bilotta, & Deeney (1999) Spreitzer, Cohen, & Ledford (1999) Neck, Connerley, & Manz (1997) Spreitzer, Noble, Mishra, & Cooke (1999) Neuman & Wright (1999) Stewart & Barrick (2000) Nieva, Fleishman, & Stoker & Remdisch (1997) Reick (1978) Oetzel (1999) Strasser & Falconer (1997) Pagell & LePine (2002) Sundstrom & Altman (1989) Pearce & Ravlin (1987) Poulton & West (1993) Poulton & West (1999) Priest, Stagl, Klein, & Salas (2006) Rangarajan, Chonko, Jones, & Roberts (2004) Rentsch & Hall (1994) Robertson, Maynard, Huang, & McDevitt (2002) Rouse & Rouse (2004) Ruel (2000) Salas, Dickinson, Converse, & Tannenbaum (1992) Salas, Sims, & Burke (2005) Sundstrom, DeMeuse, & Futrell (1990) Susman & Ray (1999) Tambe (1996) Tannenbaum, Beard, & Salas (1992) Tata & Prasad (2004) Tjosvold, Wong, Nibler, & Pounder (2002) Tompkins (1997) Neal & Hesketh (2002)

Tubbs (1994) Walker & Vines (2000) Werner & Lester (2001)

West, Borrill, & Unsworth (1998) Schippers, Den Hartog, Koopman, Yeatts & Hyten (1998) & Wienk (2003)

our criteria for inclusion (see Table 1). Specifically, inclusion was dictated by whether a particular initiative addressed team performance or effectiveness and described the relations between three or more constructs or construct categories. The latter of these two criteria was designed to screen out the thousands of research initiatives that addressed only a few of the myriad constructs currently believed to constitute the complex nomological network of teamwork. Although our literature review, which included searches of both electronic databases and our own extensive collection of team literature, resulted in an achieved population of 138 models and frameworks, only 11 of the identified theories are reviewed here. A strong case can, undoubtedly, be made that the remaining 127 sources are

199 Team Effectiveness in Organizations significant in their own right. Unfortunately, space constraints preclude a thorough examination of all identified models and frameworks. Space limitations also restrict the detail with which any given theory can be addressed. Thus, the inclusion of the 11 theories expounded on in this section is not intended as a covert endorsement of a particular line of research. However, with this caveat in mind it should also be noted that we believe that the sample of models and frameworks selected for review in this chapter is quite representative of the larger population of theoretical research prevalent within the team domain. In fact, the 11 models and frameworks reviewed below were deliberately sampled because they illustrate some of the key advancements contributing to our collective understanding of teams over the past 25 years. The particular initiatives selected for review in this section were vital in shaping our thinking about the key constructs constituting and contributing to team effectiveness. Thus, this section provides critical background information for the reader interested in gaining a deeper appreciation of why a given model or framework was strongly influential in the development of the conceptually integrative team-effectiveness framework presented later in this chapter. Prior to launching into a full discussion of the salient features of the 11 models and frameworks, we begin by describing the criteria utilized to select these particular research efforts from the larger population of initiatives identified via our literature review.

criteria for model/framework selection
Prior to discussing the particularities of the 11 models and frameworks, we will briefly describe the criteria that guided the process whereby a given initiative was selected for review. Three criteria guided our decision to expand on a given model or framework: (1) model form, (2) model dynamism, and (3) model focus. Specifically, we feel that the 11 models and frameworks reviewed in this section are representative of cutting-edge advancements in the teams domain, advancements that we capture in our three criteria. After briefly discussing these three criteria, we chronologically review each of the 11 models and frameworks.

The epistemological foundation for ipo models originates within general systems theory and its many derivatives (see von Bertalanffy... Thus. and these innovations will be reviewed in the following subsections describing model dynamism and model focus. 1960) and open systems theory (Katz & Kahn. it should be noted that there is ongoing debate about the adequacy of this approach (Hackman. 1978).g.. the use of ipo models to frame team effectiveness is consistent with derivatives of general systems theory. 1990). Ilgen. including both sociotechnical systems theory (Emery & Trist. However. performance). Model Dynamism A second critical difference between the models and frameworks of team effectiveness identified in our literature review is that they tend to diverge on the degree to which they explicitly consider how team processes may differ across the dynamic task cycle of teams. processes. While initial attempts by researchers to describe . Generally speaking. Thus.200 modeling complex systems Model Form Perhaps the most prevalent theme within the team-effectiveness literature reviewed was the widespread use of ipo (inputprocess-output) models as a guiding framework (McGrath. Sundstrom et al. Although ipo models are predominant in the teams domain. in press. and resulting outputs but also recognizes that team outputs can be fed back into input variables and that teamwork does not occur in a vacuum (Tannenbaum et al.. ipo models highlight the importance of throughputs as mediators or moderators of the relations between input factors (e. as witnessed by the fact that the preponderance of models and frameworks identified by our literature search adopted this format. there have been other advances with regard to team effectiveness that are even more intriguing. 2002. & Jundt. the use of an ipo model to frame team effectiveness is particularly important for capturing the dynamic interactions and emergent states that constitute teamwork and illuminating the nomological network of these processes. The use of ipo models reflects the current state of the art within the teams domain. team and individual characteristics) and outputs (e. describing teamwork through an ipo lens has advanced our collective understanding of the factors that constitute and promote team effectiveness. Hollenbeck. Johnson. Thus. 1956). 1992). Describing effectiveness through an ipo lens not only emphasizes the importance of the interactions between inputs.g. 1964). team satisfaction.

The stream of thought concerning entrainment is grounded in the premise that team members’ behavior becomes entrained to . A similar approach was utilized by Ancona and Chong (1999) to map existing group theory. -feedback. then performancemeasurement. & Schemmer. (2001) has implications for human-capital-management systems. more recent models and frameworks of team effectiveness have served to illuminate the “black box” of team process. research such as that presented by Dickinson and McIntyre (1997) and Marks et al.201 Team Effectiveness in Organizations teamwork focused primarily on team inputs as a mechanism for manipulating team outcomes (Goldstein. These assertions are theoretically supported by a branch of mathematics known as coupled oscillations (i. 1996). For example. 1985). Price. 1993. 1992. Fleishman & Zaccaro. 1996). therefore. Modeling the dynamic nature of interdependent action blurs the distinction between traditionally held notions of predictors and criteria and. mandates the building of increasingly sophisticated theories in order to capture the inherent complexities. such as midpoints and stages of team development. is relatively static in nature. 1997.. Marks et al. The movement to model fluidity reflects a growing recognition within the teams community that collective task performance requires adaptive moment-to-moment interteam and intrateam interaction (Dickinson & McIntyre.e. Eisner.. to the temporal role of team context. Model Focus A third primary difference between the research efforts identified in our literature review concerns the overall focus of the work. the research advanced by Dickinson and McIntyre (1997) adopts a more dynamic position. Guzzo & Dickson. 1993. In addition to raising a number of intriguing directions for future endeavors. although inclusive of many of the key constructs believed to relate to team performance. 2001. Campion et al. if particular processes take precedence over other processes at specific phases of teamwork. The focus factor pertains to whether a particular model or framework defines performance as being driven primarily by internal team factors or by external context. For example. Conversely.. and -training systems should be appropriately tailored to reflect these contingencies. These authors assert that the key to understanding team behavior lies in a shift in focus from internal factors. Shiflett. entrainment). Campion and his colleagues’ research (Campion et al..

many of the models and frameworks delineated in the material covered in our literature search conceptualize performance as a phenomenon that is actively driven by members and events within the team.g. However. For example. . entrainment is witnessed in the recurring flurry of goal-directed activity that takes place within a top management team as the fiscal quarter comes to a close. Efforts to understand teamwork via entrainment suggest that performance is largely determined by contextual factors in the team’s external environment. McGrath. and context for the meaning of time. Kozlowski. 1999. A simultaneous internal and external focus could illuminate insights into critical processes such as the timely delivery of scarce resources. Given the compelling argument supporting entrainment. Gully. For example. interventions meant to enhance teamwork have traditionally targeted variables inside the team for change. 1991).202 modeling complex systems Figure 1. the environment serves as a pacer. Fleishman. & Reick. creator of opportunity. Conceptual model of team performance (Nieva. In fact. naturally occurring contextual cycles. rhythm setter. it seems that traditional approaches emphasizing internal factors should be augmented by an understanding of key external factors. Nason. some initiatives have already been undertaken to examine the joint influence of both internal and external forces (e. 1978).. As the management team’s cycles become entrained to the quarterly business cycle. & Smith. source of interrupts. The differences between these perspectives have important implications for framing team effectiveness.

g. (3) team characteristics (e.. abilities. (2) member resources (e.g. and (4) task characteristics (e... This model incorporates individual-level input factors. communication. propose that team performance is composed of both individual task performance and team-level performance functions. 1986). social context. complexity).. Fleishman. and Reick (1978) (see Figure 1). standard operating procedures).203 Team Effectiveness in Organizations Figure 2. Model of group effectiveness (Gladstein. Nieva et al. and outputs (see Figure 2). Gladstein (1984) presents another relevant model that depicts the relations between group inputs. structure.g. models and frameworks of team effectiveness One of the earliest conceptualizations of teamwork was advanced by Nieva. personality characteristics). including group compo- . processes.g. This model also specifies four categories of team-performance antecedent variables: (1) team environment (e. training). individual skills. 1984). Particularly noteworthy is the fact that Gladstein’s model has empirical support from a large sample (Goodman.

g. heterogeneity) and group structure (e.. organizational context.. sition variables (e. and interdependence moderate the relations between group processes and outcomes such as satisfaction.g. There are a number of propositions that flow from this model. which. Normative model of team effectiveness (Hackman.g.. are related to team effectiveness (e.g. supervisory control). group design) are related to team processes. education.204 modeling complex systems Figure 3.. the model highlights the importance of fostering an organizational context that supports and reinforces teamwork via rewards. it includes organizational-level input factors such as resources available (e. rewards. customer satisfaction with product.. group . including that input factors (e. capability of team members to work together over time).. More specifically. work norms). The relations between individual. In addition. According to Hackman.g. This model illustrates how a broad range of variables can influence teamwork. uncertainty. in turn. member satisfaction.and organizational-level input factors and team effectiveness are proposed to be mediated by group processes. consulting) and organizational-structure variables (e. training. and availability of information. Group design is a second input factor illustrated that is proposed to relate to team processes.g. Hackman (1987) also advanced a model of team effectiveness (see Figure 3). 1987). The model also indicates that group task complexity. formal leadership. skills.

(b) group composition. Thus. the relations between team inputs and team processes are moderated by the ability of the group to minimize process losses (i. (1992). at the midpoint. A fourth model of team effectiveness reviewed was advanced by Gersick (1988) and is known as the punctuated-equilibrium model (pem) (see Figure 4). First. if there are not enough material resources. and (c) appropriateness of task-performance strategies. the relations between team processes and team effectiveness are moderated by the material resources available to the team. and (c) appropriate group norms regarding teamwork. skill. the researchers reviewed many of the then-current ipo models of team effectiveness and developed an integrative framework that built on prior initiatives. which reported one of the first integrative efforts to frame team effectiveness (see Figure 5).205 Team Effectiveness in Organizations Figure 4. the task may not be completed. 1988). design consists of such things as (a) task structure. The pem is empirically grounded in the findings of research conducted with a sample of eight diverse teams. Second. the model proposes two variables that may moderate the depicted relations. Punctuated-equilibrium model of team performance (Gersick.. In Tannenbaum et al. The evidence accumulated from Gersick’s research suggests that teams determine an initial method of performance during their first meeting and adhere to this method until the midpoint of the target objective is reached. team members become aware of the time left to completion and switch their strategy accordingly. Gersick argues that. Hackman’s (1987) model also specifies process criteria of effectiveness that can serve to guide the diagnosis of team weaknesses. this research makes an important contribution by showing that teamwork is dynamic and that teams adapt their performance strategies in accordance with temporal contingencies. gain group synergy). no matter how well team members interact with one another in terms of effort. Although fairly simple in its depiction. and performance strategies. This framework is more complex than the pre- .e. (b) amount of knowledge and skill. Finally. These process criteria include (a) level of effort.

viously described Gladstein (1984) model but includes some of the same variables. Specifically. quality. quantity. resulting from team performance and performance outcomes. A final addition of this framework over previous models is the recognition that organizational and situational characteristics affect team effectiveness. (2) work characteristics. Furthermore. adaptability) that occur over time. both the individual team member and team processes are proposed to affect team-performance outcomes (e. coordination. and (4) team characteristics. Not surprisingly. it recognizes that training or teambuilding interventions may moderate the relations between inputs and processes as well as those between processes and performance outcomes.. (1993) (see Figure 6). not just at the input stage. Synthesizing all five team-effectiveness models already illustrated. cycling back as subsequent system input. backup behavior.g. & Salas. devel- . Campion et al. including (1) task characteristics.206 modeling complex systems Figure 5. A sixth model of team effectiveness reviewed herein was advanced in Campion et al. The framework also depicts system feedback. time. 1992). it illustrates four distinct types of input variables. but throughout the entire ipo process. It suggests that these input factors affect each other and also serve to affect both team members and team processes (e. Beard. errors).g.. (3) individual characteristics. Team-effectiveness framework (Tannenbaum.

and process. job . Synthesized model of group effectiveness (Campion. composition.. Medsker. 1993. Although it may oversimplify the dynamic. oped a metamodel of team effectiveness. This hybrid model incorporates only those constructs proposed to directly affect team effectiveness. 1996). Specifically.. this programmatic theoretical formulation has received empirical support from two diverse samples of teams (Campion et al. Campion et al.207 Team Effectiveness in Organizations Figure 6.’s (1993) metamodel describes five categories of variables that are proposed to affect team effectiveness: job design. 1993). Campion et al. thus ignoring key mediators and moderators of the relations between team inputs and outputs.’s model parsimoniously frames the complexities of team effectiveness. Campion et al. interdependence. context. & Higgs. recursive nature of team performance.

Figure 7. * Gersick (1988). 1994). Salas. ** Tuckman (1965). . Team-evolution and -maturation model (Morgan. & Glickman.

Taskwork represents the “task-orientated skills that the members must understand and acquire for task performance” (Salas et al. and (d) environmental context. by illustrating behaviors that occur within each developmental stage. The advantages of the Morgan et al. workload sharing. First. Context is proposed to cover training. the model serves to frame how teamwork develops during task performance and training. Dickinson and McIntyre (1997) proposed a model that describes the interrelations between essential teamwork processes such as communication. Interdependence is proposed to encompass task interdependence. relative size.209 Team Effectiveness in Organizations design subsumes self-management. both task and teamwork. and after task performance (see Figure 7). In this model communication acts as the glue linking together all .. (b) individual expertise. feedback. managerial support. task significance. team leadership. backup behavior. team orientation. Morgan et al. Subsequent research has provided empirical support for this assertion (Glickman et al. and communication/cooperation between groups. it illustrates that there are two sets of skills that must be mastered in order for a team to be effective. monitoring. Composition is proposed to subsume heterogeneity. there are two tracks of skills that must be mastered before it can perform effectively: taskwork and teamwork. 1965). task variety. More recently. social support. 1987). Second. Morgan et al. The process category includes potency. 1992. and attitudinal responses that must be mastered before a team can work together effectively. Conversely. and interdependent feedback/rewards. flexibility. model over previous theoretically grounded models are twofold. The specific stage at which a given team begins and how quickly the team progresses through the proposed stages depend on such characteristics as (a) members’ experience as a team. The model proposes that task-orientated teams progress through a series of developmental stages at varying rates (see Tuckman. and preference for group work. and coordination (see Figure 8). 10). Building on the work of Gersick (1988) and Tuckman (1964). during. and communication/cooperation within groups. participation. (1994) developed a model that illustrates the stages that teams progress through before. and task identity. goal interdependence. as a team progresses through these stages.’s model also proposes that. p. (c) task characteristics.. cognitions. teamwork skills reflect the behavioral interactions.

. team leadership) and team attitude competencies (e. team orientation). this episodic model of team process consists of a series of recursive ipo loops proposed to occur sequentially and simultaneously during both a transition stage and an action stage of performance. undertaken to illuminate the complexities of team effectiveness. 1997). advance a temporally based framework of team effectiveness that extends recent notions of team process by categorizing throughputs into recurring phases. they synergistically serve as a platform for team coordination.. This systematic research has markedly advanced our understanding of the content constituting team process and the fluid nature of teamwork as processes unfold over time. is the research of Marks et al. goal specification) .g. when all the aforementioned teamwork competencies are occurring in unison. In turn. Distinct competencies characterize the action (e.g. the feedback resulting from team coordination serves as input back into team processes. mission analysis. Marks et al. (2001) (see Figure 9).210 modeling complex systems Figure 8. Model of teamwork (Dickinson & McIntyre. team leadership and team orientation are proposed to facilitate a team member’s capability to monitor his or her teammates’ performance. the model proposes that performance monitoring drives both the content of feedback and timely backup behaviors. It also suggests that. Specifically. it fails to model many of the critical antecedents and outcomes of team process.. The ninth initiative reviewed in this section. Another aspect of the model is the integration of both team skill competencies (e. Although Dickinson and McIntyre’s (1997) model incorporates many of the skill competencies underlying teamwork. Together. Furthermore. other teamwork processes.g.

(2005). proposed by Salas et al. The rhythm of team task accomplishment (Marks. There is “nothing as practical as a good theory” (Lewin. coordination) stages. 2001). assessing.’s (2001) effort to model teamwork is theoretically as well as practically intriguing. Interpersonal processes are proposed to occur during both stages. and the transition (e. & Zaccaro. suggesting that certain ksas take precedence depending on the timing of performance. 1951. Marks et al. Mathieu. this attempt to model teamwork highlights the centrality of five core teamwork processes. systems monitoring. The implications for those charged with observing. including (1) team leader- . and rewarding team performance include the need to develop a new appreciation of the timing of their measurements as well as new human-capitalmanagement systems that account for which processes are predominate at a given time.. As previously noted. and this maxim is nowhere more applicable than when utilized to describe meaningful research that offers a more robust specification of the dynamic nature of teamwork.g. This model was developed in an effort to highlight the “essence of teamwork” by illustrating the relations between the processes that they argue constitute the core of interdependent interaction (see Figure 10). 169).211 Team Effectiveness in Organizations Figure 9. Specifically. Another recently advanced theoretical initiative reviewed was the “Big Five” model. p.

and (3) mutual trust. the Big Five model also illustrates the importance of three ancillary team products and processes. Furthermore. The Big Five teamwork model (Salas. including (1) shared mental models. & Burke. Taken together. and (5) adaptability. Sims. However. (3) mutual performance monitoring. At a broad level the Big Five model of teamwork proposed in Salas et al. For example. (2005) adopts a similar approach to that of the above-described theory advanced by Dickinson and McIntyre (1997).212 modeling complex systems Figure 10. these eight constructs are dynamically related to one another and collectively form teamwork. ship. the Big Five model proposes a set of constructs and construct interrelations that differ somewhat from Dickinson and McIntyre’s attempt to describe the dynamic nature of teamwork. 2005). (2) closedloop communication. (4) backup behavior. . (2) team orientation.

Figure 11. Salas. & Kendall. Pierce. . under review). Stagl. ito model of team adaptation (Burke.

cognitions (i.. Thus.e. moves beyond modeling teamwork behavior to incorporate affective competencies (i. this applied research initiative defines team adaptation as an emergent phenomenon that coalesces over time from the unfolding of an adaptive process whereby one or more team members utilize their resources to functionally change current behaviors. therefore. research could be carried out to develop a deeper understanding of the similarities and differences between team and contextual performance. Certainly. Stagl. The 11th and final programmatic effort undertaken to investigate team effectiveness reviewed here was advanced by Burke. We believe that the similarities and differences between contextual performance (see Borman & Motowidlo. that proposed by Salas et al.. feedback is generated that subsequently serves to revise shared cognition and adaptive input factors.. Specifically.214 modeling complex systems while Dickinson and McIntyre’s theory primarily highlights the centrality of skill competencies. As this adaptive process is carried out. it is our opinion that additional theory building should be undertaken to develop meaningful models of (1) task. shared mental models). cognitions. The team-adaptation model illustrated in Figure 11 is the only initiative reviewed in this chapter that has radically departed from the traditional predominant focus on either task or team performance. 1993) and team performance warrant closer attention in future research endeavors. For example. (2) team. team members draw from their individual and shared resources to detect. frame. and Kendall (2006). Pierce.e. or attitudes to meet expected or unexpected demands. and emergent team phenomena (i. Essentially. and. These researchers proposed a model of team adaptation within an ipo framework (see Figure 11). and act on a set of cues that signal the need for functional change. Perhaps one fruitful avenue for future exploration involves framing these unique but interrelated aspects of performance in a nomological network of lawful relations. Salas. adaptability). mutual trust). and (4) adaptive performance. this state of affairs is indicative of an area ripe for exploration. emphasizes the centrality of an adaptive process that unfolds over time to emerge as team adaptation. It may be particularly interesting to examine whether indexes of contextual performance . The model advanced by Burke et al.e. the adaptive process is recursive by nature. (3) contextual. which seem to share somewhat similar content and emphasis.

The prominent differences between the ipo models reviewed lie (a) in the specific primary variables or constructs highlighted and (b) in the moderators and mediators included (e. Thus. For example. (1994). of the efforts reviewed above. Toward an Integrative Multilevel Theoretical Framework The 11 models and frameworks discussed in the previous section were utilized to guide an inquiry into the properties that seem to be essential to any truly integrative theory of team effectiveness. or Marks et al. Tannenbaum et al. These points of similarity and distinction provided us with clues as to potential content to include in our integrative framework of team effectiveness.215 Team Effectiveness in Organizations capture a practically meaningful amount of unique variance in substantive criteria beyond that already explained by indexes of teammember performance.. each of these initiatives acknowledges that inputs. with the notable exception of Dickinson and McIntyre (1997) and Salas et al. Fur- . whose models focus strictly on team processes. (2001). models and frameworks summary The 11 team-effectiveness models and frameworks described above have commonalties as well as differences. Campion et al. Salas et al. (2006). followed by Nieva et al. (1992). Another difference lies in the degree to which the models explicitly consider how the relations proposed differ across the dynamic task cycle and/or life span of teams. Burke et al. and Morgan et al.’s (1993) metamodel. situational/organizational characteristics). For example. and outputs need to be examined in order to gain a holistic understanding of team effectiveness. (1978). (2005). they do not present as detailed a picture of the fluidity of teamwork as that modeled by Dickinson and McIntyre (1997). Hackman’s (1987) model. 1964). and Gladstein (1984) are the least dynamic. processes (throughputs). one commonality shared by all the above-noted efforts to understand team effectiveness is that they adopt an ipo model (see McGrath. (2005). Gersick (1988). Although the last four adopt a less static lens than the first three.g.

Integrative framework of team effectiveness. .Figure 12.

via this ongoing process. On a broad level. team inputs are actively interpreted by team members. the accuracy of individual and shared cognition. shared mental models. The organizational environment is characterized by a cue stream that is also interpreted by team members.. form stable yet malleable expectations regarding the nature of their obligations.g. In the remainder of this section we review some of the critical components of the framework while emphasizing how it expands prior programmatic efforts to understand team effectiveness. over time. and episodically enacted over time.. Shared cognition (e. can serve to change both the organizational inputs and the teams inputs available to subsequent performance episodes. psychological safety) accrues as teamwork occurs and. . team-situation awareness. The integrative framework of team effectiveness illustrated in Figure 12 incorporates four categories of input factors.and team-performance outcomes. team leadership influences. Performance outcomes produce system feedback that. team characteristics. Furthermore. the framework illustrates the moderating role of individual-level cognition (i. In the remainder of this section. influences subsequent teamwork activities. As the processes constituting teamwork are dynamically.217 Team Effectiveness in Organizations thermore. they lead to shared cognition. the various aspects of the framework are detailed in greater specificity. simultaneously. team characteristics. task characteristics. who.e.. task characteristics. The multilevel integrative framework depicted in Figure 12 offers researchers and practitioners a simple but meaningful heuristic that illustrates some of the most important aspects of team performance. including individual characteristics. we drew heavily on what we considered intriguing constructs and ideas from the broader domain of team literature. Essentially. work structure) in promoting teamwork and team performance. The framework also suggests that team performance results in individual. Team members with accurate expectations are more likely to know which team processes to engage in and when particular activities should occur.e. expectations about roles and requirements) on the relations between team inputs and throughputs. individual characteristics. In addition. in a recursive fashion. The outcome of this inquiry is a preliminary integrative framework of team effectiveness (see Figure 12). thereby contributing to both individual and shared cognition. and is influenced by. the framework highlights the role of team inputs (i.

& Ardison. 1997. O’Shea. team-level openness to experience. The category of team characteristics also includes myriad constructs that may either have been formally established (e. Steiner. Both theoretical research and empirical research suggest that task characteristics are. 1994). Goodwin.218 modeling complex systems and work structure. LePine. encom- . The task-characteristics category includes task organization. 1978. power structure. 1972). work structure. additional research should be undertaken to investigate other critical factors such as team climate and culture. openness to experience. such as task ksas. 2004). These differences will be important to future research initiatives undertaken to develop team type–specific models and frameworks of team effectiveness. 1984. task type. the task-characteristics category reminds us that not all teams are alike and that task variations leading to different levels of interdependence can have a profound effect on how teams interact.g. critical to fostering team effectiveness (Herold. 1980. mental models. research results suggest that personality variables such as extraversion. Given the paucity of attention afforded team characteristics to date. Mullen & Cooper.. and personality. empirical work has supported the relation between team cohesion and team performance (Dailey. 1994) can be applied to analyze communication patterns and that the findings of this process can be utilized to model.. 1992. indeed. 1987. and task complexity. motivation. One implication of our emphasis on work structure is that social-network analysis (see Krackhardt & Brass. predict. the results of empirical research have repeatedly supported the importance of a team orientation (e. Interestingly. For example. teamlevel team orientation).g.. and adjustment are essential for coordinated performance (Driskell. Each of these four broad categories is an umbrella for multiple specific constructs. Eby & Dobbins. In turn. the category of individual characteristics covers a wide range of phenomena.g. Driskell. The fourth category of input variables. Driskell & Salas. & Salas. For example. each of these constructs is related to team processes and team-performance outcomes to varying degrees. Hogan. Salas. 2003). and manage team-member and team performance. team orientation. Likewise. For example. McGrath. performance arrangements) or have emerged upward from the unfolding of coordinated dyadic role exchanges of team members over time (e.

1990). For an equally impressive yet somewhat different set of team-input variables. Shared cognition such as shared and compatible mental mod- . For instance. team norms. This is important because. Essentially. team-member expectations. while we frame many of the same inputs (i. moderate the relations between inputs and processes. 1996) metamodel of team effectiveness as reproduced in Figure 6 above. task characteristics. flexible expectations about their roles and requirements will be better positioned to engage in appropriate team-member and team processes at the optimal point in time. Furthermore. which is essential for negotiating what one wants to accomplish. Also. our framework suggests a novel purpose for these components. (1992). Together.. work structure) that are traditionally advanced within the team-effectiveness domain. 1993. ongoing teamwork. while there exists a general consensus about the nature of the broad categories of input variables. Shared cognition also plays a prominent role in the hybrid framework. Our framework draws heavily on many of the recent advancements appearing in the body of team literature and perhaps most of all from cutting-edge research in cognitive psychology. and the organizational context. and communication structure. however. Thus. Campion et al. team characteristics.e. teams consist of both a formal work structure and a unique but interdependent social structure. Work structure is critical to team performance because. work characteristics such as communication structure dictate who has access to what information and when. team norms have a considerable influence on what behaviors are deemed appropriate (Hackman. individual characteristics. team leadership.219 Team Effectiveness in Organizations passes work assignment. the socio and technical systems serve to shape both what cues team members attend to and how they react to those cues. It should be noted. as open systems. team members with accurate. see Campion and colleagues’ (Campion et al. The hybrid framework depicted in Figure 12 reflects our belief that the myriad relations between input variables and teamwork processes are moderated by team-member cognition. established and changed by team inputs. that the particular constructs exemplified in the above descriptions were largely adopted from the programmatic theoretical framework advanced by Tannenbaum et al. the specific constructs proposed to be encapsulated within these categories vary from research program to research program...

220 modeling complex systems els and shared situation awareness serves to form templates that are drawn on by team members during teamwork. Shared templates or shared frames of reference imply processing objectives that constitute the social reality that team members share (Hinsz.. thinking. the teamwork processes displayed in a single revolution may not be homogenously directed at the accomplishment of a single goal. Shiflett et al. & Vollrath. implicit perceptions. Also. roles.. negotiated dyadic role exchanges) most would agree that team leaders and the leadership processes that they enact are essential to promoting team performance. 2000.. however. 1997. and feeling.. as teamwork occurs. and/or team leaders intervene. and episodic fashion in order to foster individual and team effectiveness. teamwork processes will still ensue. competencies. simultaneous. 2005). in press). episodic nature of activity in high-performance teams by illustrating a sample of core teamwork processes in a cylinder. 1997). process losses will likely abound. our hybrid framework reflects the cyclic. In the absence of this cognitive reservoir. functions. Tindale. our framework also incorporates advancements made in modeling the dynamic nature of teamwork (Dickinson & McIntyre. and (3) when to enact ksaos in support of processes in order to meet tactical and strategic objectives. Specifically. Furthermore. Salas et al. Salas.. . The point we are trying to communicate by framing teamwork processes with a cylinder is simple: that both team members’ and team-level competencies must be displayed in a dynamic. In fact. Marks et al. this set of instantiated cognitive structures enhances teamwork processes because it provides team members with the insight required to understand (1) what they should and should not be doing. & Burke. 1985. traits. several revolutions may be needed to meet a single objective. unfortunately. In addition to having a distinctive cognitive flavor. Specifically. Fleishman & Zaccaro. Our integrative framework of team effectiveness also acknowledges the centrality of team leadership throughout the life span of the team (see Marks et al. (2) how to go about accomplishing stated objectives. Stagl. adaptation. as multiple objectives are interlaced or additional interrelated but unique projects undertaken. the organizational environment changes. While the construct of team leadership is a complex phenomenon with a variety of conceptualizations (e. 2001. 1992. shared mental models and other forms of shared cognition are revised.g.

. performance. Consequently.. who provide a more complete discussion of this topic. the inquisitive reader is directed to Campbell and Kuncel (2001). and (3) team ksas. albeit at a broad level. in press). teamwork/team performance is composed of the dynamic display of (1) team-member taskwork ksas. recent meta-analytic evidence highlights the importance of team-leadership behaviors in achieving team outcomes (Burke et al. therefore. and after task episodes (Hackman. has seldom been explicitly considered. While prior programmatic efforts to frame teamwork have included both individual-level team members’ competencies and team-level competencies. Essentially. as well as their symbiotic interdependencies. Moreover.and team-level processes.. the processes displayed in our framework are meant as a small sample of the number of competencies that must be fluidly interlaced as team performance unfolds over time. collective efficacy) are simultaneously proximal products of interdependent interaction and serve as subsequent inputs to new performance episodes. In our framework. Incorporating multiple performance outcomes mandates the need for multiple ef- . 1997) that result from a team’s experiences while navigating its operational challenges. team leaders act as major drivers and maintainers of team development. coordination) processes are highlighted. they serve to yield team performance. Furthermore. The individual. the distinction between these ksa sets.g.g. These emergent states (e. and effectiveness over time. In fact. during.and team-level performance outcomes. adaptive.. it is intuitive that it should also include both individual. As noted above. cohesion.and team-level performance outcomes depicted in the model are meant to include both traditional output indexes such as the number of goods produced as well as emergent states (Marks et al. (2) team-member teamwork ksas.221 Team Effectiveness in Organizations and effectiveness. 2001) or psychosocial traits (Cohen & Bailey.g. our framework illustrates both individual. both individual-level (e. 2002). As the competencies constituting these three sets unfold over time. It is through the leadership process that team leaders act to synchronize task and developmental cycles. the interdependencies between team members’ competencies and team competencies. Another unique aspect of our integrative hybrid framework is the fact that it illustrates. they institute the conditions that teams and their members draw on before.. For further clarification of this issue. interpersonal) and team-level (e.

Specifically. our framework acknowledges that the particular level of effectiveness achieved by a given team member or team will vary depending on the dimension(s) under consideration. hundreds of research studies have been guided by the team-effectiveness theories presented in this chapter. and (3) describing key antecedents to team performance and effectiveness. (2) illustrating some the interdependencies between team members’ competencies and team-level competencies. heuristic that can be called on to quickly understand the most important aspects of interdependent performance. and staffing teams.222 modeling complex systems fectiveness evaluations. With regard to this latter point. Practical Implications During the last 25 years. (2) formulating empirically testable propositions. in the next section we review the practical implications of our hybrid framework for performance measurement. summary The multilevel integrative framework described in this section offers researchers and practitioners a useful. our hybrid framework advances the science of teamwork by (1) adopting a cognitive approach to framing the dynamic and multilevel nature of team effectiveness from an ipo perspective. On further conceptual refinement. yet simple. In essence this suggests that teams and their members can be described in terms of effectiveness patterns. The integrative framework illustrated in Figure 12 above is intended to highlight often-ignored issues in team-effectiveness research. and (3) designing interventions to facilitate human-capital management. the performance-outcome dimensions that are most relevant to the evaluation process will change as a function of time. training. . Furthermore. Thus. The degree to which variables constituting these patterns can compensate for one another when contributing to a global index of effectiveness that can subsequently be utilized by organizations for practical decision-making purposes remains an area ripe for exploration. we expect that our hybrid integrative framework will be useful in (1) defining a nomological net of teamwork.

and task-based simulations. depending on whether the team-effectiveness model adopted incorporates the notion of time. Furthermore. the accumulated findings from this research have been shaped into principles and guidelines (e. Salas. developing.. The importance of team-effectiveness models and frameworks to this process cannot be understated. it seems that. teameffectiveness models and frameworks illuminate what factors are potentially important in a given situation. The delineated principles and guidelines offer essential insight to those concerned with designing. because most initiatives to describe team effectiveness have adopted an ipo lens. & CannonBowers. and staffing teams will be addressed in this section. conducting developmental interventions. 2004). instructional strategies. As discussed above. for those individuals charged with the application of measurement tools. Thus. Salas. the resulting models and frameworks specify how and why a factor is important. there is “nothing as practical as a good theory.” Keeping with this line of thinking. Burke. it can also provide guidance as to when particular factors are important. The answers to the above questions provide the informed practitioner with clues on what factors to measure. Finally.223 Team Effectiveness in Organizations In some cases. serve to increase the utility of a given intervention. & Stagl. This purchase of information is better informed when guided by a theoretical rationale such as those provided by the team-effectiveness models and frameworks described in this chapter. measuring team performance Team-effectiveness models and frameworks are critical to building appropriate team-measurement systems because they serve to illustrate both the complexities of engaging in teamwork and the many factors surrounding team process. as is reflected in the statement that performance measurement can be viewed as a purchase of information that a researcher makes in order to help guide his or her decision making (Brannick & Prince. Burke.g. developing measures of teamwork must be guided by a number of . and when to measure them and. 2000. the use of theoretically grounded team-effectiveness models for measuring performance. as we have seen. Furthermore. 1997). how to measure them. and delivering human-capital-management interventions. thus.

In turn.224 modeling complex systems factors. researchers have also offered general guidance that is essential for developing instructional strategies.. the nature of incorporated stimuli. Marks et al. 116). multilevel systems. in order for team members to develop the adaptive expertise required for effectiveness in a chaotic context. most of these differences can be reconciled in a set of core universal team processes. In addition to the insight provided by models and frameworks of team effectiveness. This is because team-effectiveness models and frameworks illustrate critical ksa competencies or individual and team-level processes and their nomological net. including the purpose of measurement. developing. These processes are illustrated in most exhaustive attempts to frame teamwork (see Cannon-Bowers et al. these ksa competencies serve as the core content when designing. focused on the development of adaptive individual and team skills” (Kozlowski. and anticipated costs (McIntyre & Salas. 2001) and were described earlier in this chapter. While measures can be designed to assess any part of the ipo system. training team members In addition to providing a platform for performance measurement. team-effectiveness models and frameworks also offer insight into the types of training that can be utilized to develop high-performance teams and team members. Training must be shifted to the work environment. For example. Care . “training systems must shift in orientation from off-site. p. By illuminating the specific competencies underlying teamwork. Essentially.. measurement timing. single episode. 1998. and delivering instructional strategies to train teams and team members. 1995. While there are slight variations in the labels and definitions produced by those initiatives undertaken to examine teamwork. the comprehensive research initiatives listed above provide guidance on the development of instructional strategies. capturing diagnostic information mandates measuring processes as raw materials are transformed into finished products and services as well as team-performance outcomes. target ksa competencies. models and frameworks of team effectiveness help practitioners answer the question of how to turn a team of experts into an expert team. on line. individual-level skills delivery to multi-episode. 1995).

1995. 450) holds true at all levels in the conceptual space. Cannon-Bowers. Huselid. & Mount. a few endeavors have already been undertaken to delineate guidelines for staffing teams (see Driskell et al. Following this line of thinking. Thus. Neuman & Wright. Particularly noteworthy are those investigations that offer compelling theoretical rationales for how individual-level characteristics synergize and emerge as team-level phenomena (see Barrick.225 Team Effectiveness in Organizations must also be taken when designing interventions to help ensure that both horizontal transfer and vertical transfer ensue (Kozlowski. all teams must import energy in the form of human capital in order to avoid entropy and enjoy the prosperity that interdependent interaction can produce (see Katz & Kahn. staffing teams There is a growing awareness of the importance of managing human capital at multiple levels (see Campbell & Kuncel. Schneider. LePine. 2003. Stewart.. Although. 1987. Ployhart & Schneider. Brown. 1998. 2001. 2005. Together. researchers have begun to explore the importance of individual differences for both team members and team performance. Once dominated by a largely individualistic emphasis. additional research is required in order to more fully refine current processes as well as to develop new staffing strategies that simultaneously enhance both individual and team effectiveness.. Specifying the mechanisms whereby individual-level characteristics emerge upward to collective phenomena within various team types characterized by varying levels of interdependencies is an issue warranting closer attention. p. Weissbein. Smith. 1978). Jackson & Ruderman. the traditional concerns of person-job fit and person-environment fit must be balanced with equal attention directed at person-team fit (Hollenbeck et al. theories of team effectiveness and principles delineated from their application can be a powerful source of information for fostering team performance and effectiveness. & Sipe. 1995). 1999). 1987. 1996.and public-sector organizations now recognize that the maxim “the people make the place” (Schneider. 2000). private. Stevens & Campion. . 2000. 1999). 2002). Neubert. & Salas. Essentially. Klimoski & Jones.

226 modeling complex systems

Toward an Integrated Science of Teams
In this chapter we synthesized the current state of the literature with regard to the ongoing proliferation of team-effectiveness models and frameworks. This endeavor resulted in the identification of 138 team-effectiveness models and frameworks that have been proposed since the early 1980s. Specifically, our review identified 15 initiatives published in the 1980s, 68 models and frameworks published in the 1990s, and 51 efforts that had already been advanced by early 2004. As anticipated, the results of our literature review suggest that there was a 257% increase in the number of theoretically grounded models and frameworks proposed in the 1990s as compared to the prior decade. Furthermore, team researchers are on pace to eclipse all previous historical periods combined in just the first decade of the 21st century. The above-noted findings are indicative of the proverbial double-edged sword. If you typically view the glass as half full, you can find solace in knowing that the quest to understand the complex, dynamic, and elusive nature of teamwork continues and is, indeed, quickening. Those viewing the glass as half empty, however, have undoubtedly by now surmised that our review also suggests that, while there are some commonalties among the models (Ruel, 2000), there remains a significant amount of intereffort variation. The accelerating rate of production, coupled with the concomitant disparities among the initiatives advanced during the past 25 years, foreshadows a looming crisis in the teams domain, an impending conundrum paralleling the general lack of theoretical integration plaguing the behavioral sciences today (see Campbell, 1990b). Why are there such a wide range of perspectives on supposedly the same underlying team phenomena and the corresponding intereffort variation, as noted above? We believe the answer is simple: team researchers and, in fact, whole disciplines fail to communicate. Why is this so? The answer is certainly complex. However, aside from environmental and interpersonal issues, the answer is relatively straightforward: team researchers are speaking different languages, different in terms of underlying paradigms, assumptions, values, goals, and missions. Similar problems often undermine mergers, acquisitions, and multinational teamwork. So what can be done when social psychologists are speaking

227 Team Effectiveness in Organizations Latin while cognitive psychologists are versed in the queen’s English? We believe the answer is twofold. Foremost, the current state of affairs characterizing team research finds the area largely devoid of collectively set, overarching short-, mid-, and long-term goals. Agreed-on tactical and strategic objectives must be articulated if the domain of team research intends to continue maturing in an integrative, productive fashion. After all, differentiation ultimately results in chaos without integration. A second reason for the rampant disjointedness is the lack of an integrative structure that, if present, could serve to house what is currently known, guide future endeavors, and be seamlessly updated by distributed stakeholders to reflect accumulating findings. For example, O*Net has radically reshaped the outmoded Dictionary of Occupational Titles. Perhaps it is time for T*Net to emerge. A failure to systematically encode and integrate is far from the only problem contributing to the frailty of research investigating teams. Unfortunately, team research is also, on occasion, methodologically weak. For example, far too often, in their haste to advance the field, team researchers settle for single-source expediency at the expense of scientific rigor. Specifically, increased use of triangulation methodology through the use of multiple measures is needed. Transparent Likert scales, subject to a host of contrived response protocols, continue to dominate the landscape of team members’ attitude assessment when seemingly superior measurement techniques such as Thurstone scaling and the conditional-reasoning approach are readily available. Is the extra effort and cost of developing a conditional-reasoning instrument not offset by the more accurate picture of a respondent’s standing on an underlying attribute that would be provided? Keeping with the above line of thinking, a similar issue warranting considerable attention concerns the measurement of team cognition (Salas & Fiore, 2004). Perhaps there is no single greater concern, as it is widely accepted that shared cognition is the foundation of the coordinated dyadic role exchanges that underlie much of team performance. However, for all the fanfare, there is little known about measuring cognition in teams (Cooke, Salas, Cannon-Bowers, & Stout, 2000; Cooke, Salas, Kiekel, & Bell, 2004). What is known suggests that team research would substantially benefit from the refinement of proved techniques for measuring shared cognition, such as

228 modeling complex systems card sorting, concept mapping, textual analysis, relatedness ratings, laddering interviews, and multidimensional scaling. Perhaps the situation is not so bleak or the outcome of theoretical chaos as of yet predetermined. However, the increasing diffusion without the proper structure in place to define, guide, and integrate forthcoming advancements will, in our opinion, only contribute to the fragility that currently characterizes the teams domain. In order to curb the ongoing diffusion and other substantive problems noted above, we assert that, where appropriate, new initiatives should build on what is enduring from the past. Future endeavors should be informed by the insights afforded by the models outlined in this chapter. Furthermore, we argue for a return to basics because it is only by beginning anew that we can hope to set the substantive goals and structure that will thrust us forward in an organized, meaningful manner.

Concluding Comments
The complexity of contemporary work environments compels a recognition that it is no longer economically viable to navigate current challenges via an exclusive reliance on individual workers (West, Borrill, & Unsworth, 1998). In fact, an accelerating rate of change, driven by the ongoing technological revolution, is sweeping away the last remnants of a business landscape once dominated by an emphasis on colocated individuals (Priest, Stagl, Klein, & Salas, 2006). The lesson learned is clear: organizations as open systems are always growing in differentiation and integration. Thus, organizations must adapt their vision, structure, and human-capital practices in order to avoid entropy and enjoy prosperity in the 21st century. It is our hope that the research effort described in this chapter serves to further the understanding of teams, teamwork, team performance, and team effectiveness by providing common ground or a shared mental model of the science of teams. Furthermore, we eagerly anticipate the day when team researchers everywhere step up to the challenges laid at their feet herein and overcome their differences to collectively define the future of team research.

229 Team Effectiveness in Organizations

Note
The views expressed in this work are those of the authors and do not necessarily reflect official army policy. This work was supported by funding from the Army Research Institute under sbir topic #osd02-cr01 and under baa #dasw01-04-r-0001. We would also like to acknowledge Dana Kendall for conducting a thorough literature review.

References
Aldag, R. J., & Fuller, S. R. (1993). Beyond fiasco: A reappraisal of the groupthink phenomenon and a new model of group decision processes. Psychological Bulletin, 113, 533–552. Alper, S., Tjosvold, D., & Law, K. S. (2000). Conflict management, efficacy, and performance in organizational teams. Personnel Psychology, 53, 3, 625–642. Ancona, D. G., & Caldwell, D. F. (1992). Bridging the boundary: External activity and performance in organizational teams. Administrative Science Quarterly, 37, 634–665. Ancona, D., & Chong, C.-L. (1999). Cycles and synchrony: The temporal role of context in team behavior. In R. Wageman (Ed.), Groups in Context, 2 (Research on Managing Groups and Teams, Vol. 2, pp. 33–48). Amsterdam: Elsevier. Annett, J., & Cunningham, D. (2000). Analyzing command team skills. In J. M. Schraagen, S. F. Chipman, & V. L. Shalin (Eds.), Cognitive task analysis (pp. 401-415). Mahwah nj: Erlbaum. Argote, L., & McGrath, J. E. (1993). Group processes in organizations: Continuity and change. In C. L. Cooper & I. T. Robertson (Eds.), International review of industrial and organizational psychology (Vol. 8, pp. 333–389). New York: Wiley. Arrow, H., McGrath, J. E., & Berdahl, J. L. (2000). Small groups as complex systems: Formation, coordination, development, and adaptation. Newbury Park ca: Sage. Avolio, B. J., Kahai, S., Dumdum, R., & Sivasubramaniam, N. (2001). Virtual teams: Implications for e-leadership and team development. In M. London (Ed.), How to evaluate others in organizations (337–358). Mahwah nj: Erlbaum. Baldwin, T. T., & Bedell, M. D. (1997). The social fabric of a team-based M.B.A. program: Network effects on student satisfaction and performance. Academy of Management Journal, 6, 1369–1397. Balkundi, P., & Harrison, D. (2004, August). Networks, leaders, teams and time: Connections to viability and performance. Paper presented at the meeting of the Academy of Management, New Orleans. Barrick, M. R., Stewart, G. L., Neubert, M. J., & Mount, M. K. (1998). Relat-

230 modeling complex systems
ing member ability and personality to work-team processes and team effectiveness. Journal of Applied Psychology, 83, 377–391. Barry, B., & Stewart, G. L. (1997). Composition, process, and performance in self-managed groups: The role of personality. Journal of Applied Psychology, 1, 62–78. Beck, A. P. (2002). Work team effectiveness, group efficacy, and the role of informal leaders. International Journal of Group Psychotherapy, 2, 311–315. Belbin, R. M. (2000). Beyond the team. Oxford: Butterworth-Heinemann. Bell, B. F., & Kozlowski, S. W. J. (2002). A typology of virtual teams: Implications for effective leadership. Group and Organization Management, 27, 14–49. Beyerlein, M., Johnson, D., & Beyerlein, S. (2003). Team based organizing: Advances in interdisciplinary studies of work teams (Vol. 9). Oxford: Elsevier Science. Bolman, L. G., & Deal, T. E. (1992). What makes a team work? Organizational Dynamics, 1, 34–44. Borman, W. C., & Motowidlo, S. J. (1993). Expanding the criterion domain to include elements of contextual performance. In N. Schmitt & W. C. Borman (Eds.), Personnel selection in organizations (pp. 71–98). San Francisco: Jossey-Bass. Brandes, W., & Weise, P. (1999). Team performance as a constellation of forces: A general model. Kyklos, 52, 573–590. Brannick, M. T., & Prince, C. (1997). An overview of team performance measurement. In M. T. Brannick, E. Salas, & C. Prince (Eds.), Team performance assessment and measurement: Theory, methods, and applications (pp. 331–356). Mahwah nj: Erlbaum. Brodbeck, F., & Greitemeyer, T. (2000). A dynamic model of group performance: Considering the group members’ capacity to learn. Group Processes and Intergroup Relations, 3, 159–182. Bunderson, J. S., & Sutcliffe, K. M. (2002). Comparing alternative conceptualizations of functional diversity in management teams: Process and performance effects. Academy of Management Journal, 45(5), 875–893. Burke, C. S., Stagl, K. C., Klein, C., Goodwin, G. F., Salas, E., & Halpin, S. M. (in press). What type of leadership behaviors are functional in teams? A meta-analysis. Leadership Quarterly. Burke, C. S., Stagl, K. C., Salas, E., Pierce, L., & Kendall, D. (2006). Understanding team adaptation: A conceptual analysis and model. Journal of Applied Psychology, 91(6), 1189–1207. Campbell, J. P. (1990a). Modeling the performance prediction problem in industrial and organizational psychology. In M. D. Dunnette & L. M. Hough (Eds.), Handbook of industrial and organizational psychology (pp. 687–732). Palo Alto ca: Consulting Psychologists Press. Campbell, J. P. (1990b). The role of theory in industrial and organizational psychology. In M. D. Dunnette & L. M. Hough (Eds.), Handbook of industrial and organizational psychology (pp. 39–74). Palo Alto ca: Consulting Psychologists Press.

231 Team Effectiveness in Organizations
Campbell, J. P., Dunnette, M. D., Lawler, E., & Weick, K. (1970). Managerial behavior, performance, and effectiveness. New York: McGraw-Hill. Campbell, J. P., & Kuncel, N. R. (2001). Individual and team training. In N. Anderson, D. S. Ones, H. K. Sinangil, & C. Viswesvaran (Eds.), Handbook of industrial, work and organizational psychology: Vol. 1. Personnel psychology (2nd ed., pp. 272–312). London: Sage. Campion, M. A., Medsker, G. J., & Higgs, A. C. (1993). Relations between work group characteristics and effectiveness: Implications for designing effective work groups. Personnel Psychology, 46, 823–850. Campion, M. A., Papper, E. M., & Medsker, G. J. (1996). Relations between work team characteristics and effectiveness: A replication and extension. Personnel Psychology, 49, 429–452. Cannon-Bowers, J. A., Tannenbaum, S. I., Salas, E., & Volpe, C. E. (1995). Defining competencies and establishing team training requirements. In R. A. Guzzo, E. Salas, et al. (Eds.), Team effectiveness and decision making in organizations (pp. 333–380). San Francisco ca: Jossey-Bass. Choi, J. N. (2002). External activities and team effectiveness: Review and theoretical development. Small Group Research, 2, 181–209. Cohen, S. G. (1994). Designing effective self-managing work teams. In M. M. Beyerlein & D. A. Johnson (Eds.), Advances in interdisciplinary studies of work teams: Theories of self-managing work teams (pp. 67–102). Greenwich ct: jai. Cohen, S. G., & Bailey, D. E. (1997). What makes teams work: Group effectiveness research from the shop floor to the executive suite. Journal of Management, 23, 239–290. Cohen, S. G., Ledford, G. E., Jr., & Spreitzer, G. M. (1996). A predictive model of self-managing work team effectiveness. Human Relations, 49(5), 643– 676. Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: Design & analysis for field settings. Chicago: Rand McNally. Cooke, N. J., Salas, E., Cannon-Bowers, J. A., & Stout, R. (2000). Measuring team knowledge. Human Factors, 42, 151–173. Cooke, N. J., Salas, E., Kiekel, P. A., & Bell, B. (2004). Advances in measuring team cognition. In E. Salas & S. M. Fiore (Eds.), Team cognition: Understanding the factors that drive process and performance (pp. 83–106). Washington dc: American Psychological Association. Cooper, C. L., & Robertson, I. T. (Eds.). (2004). International Review of Industrial and Organizational Psychology (Vol. 19). New York: Wiley. Cooper, M. A., Shiflett, S., Korotkin, A. L., & Fleishman, E. A. (1984). Command and control teams: Techniques for assessing team performance. Bethesda md: arro. Coovert, M. D., & Dorsey, D. W. (2000). Computational modeling with Petri nets: Solutions for individual and team systems. In D. R. Ilgen & C. L. Hulin (Eds.), Computational modeling of behavior in organizations: The third scientific discipline (pp. 163–181). Washington dc: American Psychological Association.

232 modeling complex systems
Cuevas, H. M., Fiore, S. M., Salas, E., & Bowers, C. A. (2004). Virtual teams as sociotechnical systems. In S. H. Godar & S. P. Ferris (Eds.), Virtual and collaborative teams: Process, technologies and practice (pp. 1–19). London: Idea Group. Cummings, T. G. (1978). Self-regulating work groups: A socio-technical synthesis. Academy of Management Review, 3, 625–634. Dailey, R. C. (1980). A path-analysis of R&D team coordination and performance. Decision Sciences, 11, 356–369. Deeter-Schmelz, D. R., Kennedy, K. N., & Ramsey, R. P. (2002). Enriching our understanding of student team effectiveness. Journal of Marketing Education, 24(2), 114–124. de Jong, R. D., Bouhuys, S. A., & Barnhoorn, J. C. (1999). Personality, self-efficacy and functioning in management teams: A contribution to validation. International Journal of Selection and Assessment, 7(1), 46–49. Denison, D. R., Hart, S. L., & Kahn, J. A. (1996). From chimneys to crossfunctional teams: Developing and validating a diagnostic model. Academy of Management Journal, 39(4), 1005–1023. DeSanctis, G., & Poole, M. S. (1994). Capturing the complexity in advanced technology use: Adaptive structuration theory. Organization Science, 5(2), 121–147. Devine, D. J., & Clayton, L. D. (1999). Teams in organizations. Small Group Research, 30, 678–711. Dickinson, T. L., & McIntyre, R. M. (1997). A conceptual framework for teamwork measurement. In M. T. Brannick & E. Salas (Eds.), Team performance assessment and measurement: Theory, methods, and applications (pp. 19–43). Mahwah nj: nea. Doolen, T. L., Hacker, M. E., & Van Aken, E. M. (2003). The impact of organizational context on team effectiveness. ieee Transactions on Engineering Management, 50(3), 285–296. Drexler, A. B., Sibbet, D., & Forrester, R. H. (1988). The team performance model. In W. B. Reddy & K. Jamison (Eds.), Team building: Blueprints for productivity and satisfaction (pp. 45–61). Alexandria va: National Institute for Applied Behavioral Science; San Diego: Pfeiffer. Driskell, J. E., Hogan, R., & Salas, E. (1987). Personality and group performance. Group Processes and Intergroup Relations: Review of Personality and Social Psychology, 9, 91–112. Driskell, J. E., Radtke, P. H., & Salas, E. (2003). Virtual teams: Effects of technological mediation on team processes. Group Dynamics: Theory, Research, and Practice, 7, 297–323. Driskell, J. E., & Salas, E. (1992). Collective behavior and team performance. Human Factors, 34, 277–288. Druskat, V. U., & Kayes, D. C. (1999). The antecedents of team competence: Toward a fine-grained model of self-managing team effectiveness. Research on Managing Groups and Teams, 2, 201–231. Dunnette, M. D. (1963). A note on criterion. Journal of Applied Psychology, 47, 251–254.

Norwood nj: Ablex. Time and transition in work teams: Toward a new model of group development. Process variables critical for team effectiveness: A Delphi study of wraparound team members. and effectiveness. G. J.. 18. B. L. (2004). & K. Fleishman. & Elms.. R. Journal of Management.. Gist. E. 9–41. Edmondson. Personnel Psychology. G. L. (2002). 14(3). S. L. Team effectiveness in multinational organizations: Evaluation across contexts. C. Journal of Organizational Behavior. Collectivistic orientation in teams: An individual and group-level analysis. Adaptive team coordination. . 203–231. Locke. & Monda-Amaya. 275–295. New York: Pergamon. Managing the life cycle of virtual teams. A. E. J... J. 6– 20. Gibson. A. 237–257. 44. In M. 158–171.). Academy of Management Journal. W. consensus decision making. (1997). D. Edmondson. (2000). M. West. (1992). In R. Flemming.. Organizing work to support relational co-ordination.. J. D. Tjosvold. International Journal of Human Resource Management. E. Models and Techniques (Vol. Furst. Gersick. W... team-set goal difficulty. & Knight. 312–325. Psychological safety and learning behavior in work teams. A. S. (1987). T. (2003). 83–97).. and top management effectiveness. E. G. 235–256). 11(3). Effects of rotated leadership and peer evaluation on the functioning and effectiveness of self-managed teams: A quasi-experiment. Hannan. 72(2)... & Schwab. F. Managing the risk of learning: Psychological safety in work teams. Chief executive leadership style. 28(4). 13(2). pp. & Trist. 297–325. M. A.. Reeves. Leadership Quarterly. Sociotechnical systems. Teams: Their training and performance (pp. & Taylor.. E. Administrative Science Quarterly. Toward a taxonomy of team performance functions. 517–539. Human Factors. M. J. C. 18(2).. E.233 Team Effectiveness in Organizations Durham. Emery. C. Zellmer-Bruhn. E. (1999). A. (1999). E. 55(4). & Dawson. 401–420. & Zaccaro. T. 41. Chichester: Wiley. (1988). (1997). 929–948. Flood. Smith (Eds. B. A. L. D. 31. Erez. H.. Edmondson.. (1960). Rosen. C. Entin. H. Churchman & M. 31–56).). D. & Serfaty.. H. A. M. Organizational behavior: Group structure. M. (2003).). and tactics on team effectiveness. A. 350–383. A dynamic model of top management team effectiveness: Managing unstructured task streams. Academy of Management Executive. Roberto. LePine. Group and Organizational Management. International handbook of organizational teamwork and cooperative working (pp. 2. A.. & Dobbins. Organizational Behavior and Human Decision Processes. 22(3). Turner. Swezey & E. S. C. M. process. (2001). (2000).. A. efficacy. Salas (Eds. Eby.. 444–474. P. D. West. P. Verhulst (Eds. & Watkins. E. 9(3). A. Effects of leader role. M. & Blackburn. Gittell. Remedial and Special Education. E. S. Smith. K. G.. Management Sciences. In C. (2003). European Journal of Work and Organizational Psychology.

21(3). (1987).. R. & Sainsbury. L. Goldstein. E. P.. Hackman. Group performance and intergroup relations in organizations. 269–313). 9.). 327–341. R. Morgan.. (1987). 2). R. 11. S. R. M. School Leadership and Management. 47. V. R. G. Driskell. Cummings & B. Gustafson (Chair). S. Z. Amir. The design of work teams. Training in organizations (3rd ed.. Dunnette & L. J. G.. & Shea. P. J. (1980). Hackman. G.. Hackman. S.. 121–173). Research in Organizational Behavior (Vol. In J.. P.. . G. 29. J. J. Handbook of organizational behavior (pp. (1987). C. E. Goodwin. Zimmer.. Groups in context: A model of task group effectiveness.). (1993). The evolution of teamwork skills: An empirical assessment with implications for training (Tech. pp. In L. Salas. R. 89(1). Understanding groups in organizations. A. B. W.. M. Breast cancer teams: The impact of constitution. (1986).. (1990). W. Staw (Eds. M. D. E.). P.. 499–517. and methods of operation on their effectiveness. San Francisco: JosseyBass. & Dickson. Leading teams: Setting the stage for great performances. New Haven ct: Yale School of Organization and Management. 87-016). Handbook of industrial and organizational psychology (2nd ed. Gupta. J.). A. Guerette. (1996). D. In S.234 modeling complex systems Gladstein. Teams in organizations: Recent research on performance and effectiveness. Making conditional reasoning tests work: Reports from the frontier. Supervision practices and audit effectiveness: An empirical analysis of gao audits. O’Shea. Borrill. P. Ravlin. Annual Review Psychology. Chicago. B.. In M. & Oldham. P. West.. Groups that work (and those that don’t): Creating conditions for effective teamwork. (1992). Belmont ca: Wadsworth. M. Haward. British Journal of Cancer. J. Rep. 27–49.. Hackman. M.. Umanath. Goodman. Goodman. (2001). Reading ma: Addison-Wesley. R. Dawson. A. Glickman. & Salas. (1999). Montero. W. Behavioral Research in Accounting. Guzzo. Englewood Cliffs nj: Prentice-Hall. Hackman. J. E.. (2003). F. (1983). Guzzo. 15–22. Symposium conducted at the 19th Annual Conference of the Society for Industrial and Organizational Psychology. (2002). J... (1984). M. Scully. J. & Schminke. R. What makes a good team player? Development of a conditional reasoning test of team orientation. Campbell. Palo Alto ca: Consulting Psychologists Press. No. A normative model of work team effectiveness (Technical Report No. & Ardison. J. S.. new cancer workload. pp. Lorsch (Ed. R. San Francisco: Jossey-Bass. R. 315–342). Orlando fl: Naval Training Systems Center.. N. Boston: hbs. P. Hough (Eds. Management teams in education: An unequal music. L. C. Designing effective work groups. Hall. S. (2004).. & Dirsmith. Administrative Science Quarterly. R. 307–338. Greenwich ct: JAI. Work redesign. I. L..

. Structural contingency theory and individual differences: Examination of external and internal person-team fit. et al.. J. (1997).. Ellis. European Journal of Work and Organizational Psychology. Hertel. 87. Raza.. D.. . and corporate financial performance. G. A. Whitney & R.. D. R.235 Team Effectiveness in Organizations Heinemann. New York: Plenum. In E... 349–362. Hoegl. Konradt. J. R. Guzzo. B. Ilgen. R. 599–606. Tindale. Herold. & Driskell. M. Ilgen. In G. 29–42).). 23(4). In P. C. & Foushee. (1988). S. D. J. E. R. Journal of Applied Psychology. A. (1999). Why crew resource management? Empirical and theoretical bases of human factors training in aviation. Hogan.. The emerging conceptualization of groups as information processors. & R. & Sheppard. Computer-assisted groups: A critical review of the empirical research. New York: Academic. Managing distance by interdependence: Goal setting. D. West. & Rowland. A. Sego. Helmreich (Eds. B. D. Ochsman (Eds. team performance. D. H. Hinsz. Salas. (1995). J. 1–28. Helmreich. A multiple-level analysis of hospital team effectiveness. A.. and team-based rewards in virtual teams. Kanki. J. 17. & Phillips. H.). 635–672. & Routhieaux.). task interdependence. M.. B. productivity. J. (1995). J. Hollenbeck. H. J.). Journal of Engineering and Technology Management. (1999). G. 43–64. R.. Cockpit resource management (pp. 3. Improving the performance effectiveness of groups through a task contingent selection of intervention strategies. B. (1992). G. & McGrath. Wiener. (2002). U. Moon. Higgs. A model of team performance. 1–13. Hollingshead.. San Francisco: Jossey-Bass... (Eds. E. L. Health Care Supervisor.. When teamwork really matters: Task innovativeness as a moderator of the teamwork-performance relationship in software development projects. 129–139. 54. (1978). B. Team effectiveness and decision making in organizations (pp. Hollenbeck. and organizational context. M. L. R. A. E. A. S. Psychological Bulletin. P.. 13. All pigs are equal? Management Education and Development. B. R. 2. Major. M. M. (2004). V. 93–103). & Vollrath. & Orlikowski. (2002). Psychology and productivity (pp. Team performance in health care: Assessment and development (pp. 292–316. 83(3). & Zeiss. 20. 315–325. Heinemann & A.. Multilevel theory of team decision making: Decision performance in teams incorporating distributed expertise. S. (1995). L. Huselid. American Psychologist. R.. 281–302.. D. Praveen. G. (1993). & Gemuenden. L. 3–45). Personality. Ilgen. In R. 121. L. 46–78). D. A. Teams embedded in organizations... M. Zeiss (Eds. The impact of human resource management practices on turnover. Academy of Management Review.. Academy of Management Journal. (2003). P. Hedlund. D. Journal of Applied Psychology. Higgins. D. New York: Kluwer Academic/ Plenum.

B.. (1996). J. Washington dc: American Psychological Association. J. & Rosen. 6.. (1997). Norwood nj: Ablex. Personnel Psychology. San Diego: Academic. 26(2). (1978). E. L. Team effectiveness and decision making in organizations (pp. A. Personnel Psychology. Johnson. 245–270).. N. 29(3). H. Kirkman. R. Journal of Specialists in Group Work. Team effectiveness: Contributors and hindrances.. B. Human Systems Management. & Kahn. (1992). Hollenbeck. J. & Avolio. Klein. A recognition primed decision (RPD) model of rapid decision making. K.). L. (1993). Tesluk. M. J. Academy of Management Executive. 185–197. B.). D.. J. Kolodny. 50. 121–125.. R. I. Karau. R.). 15(2). The effects of time scarcity and time abundance on group performance quality and interaction process. Staffing for effective group decision making: Key issues in matching people and teams. & Shapiro. Guzzo & E. Kelley. Klimoski. Salas (Eds. Colquitt. J. A. Keys to effective virtual global teams. A. J. J. Jung. B. Katz. 183–186. 730–757. and contextual support variables. & Noe. E. Chemers & R. L.. Towards the development of a . & Kiggundu.. Group and Organizational Management. E.. & Jones. & Baik. T. (2004). Kahai. & Jundt. 6. J. R. R. E. 153–171. Investigating work group characteristics and performance over time: A replication and cross-cultural extension.. G. & Ruderman.. L. In M. Teams in organizations: From i-p-o models to imoi models. Janz. Ayman (Eds. & Sego. M.. Orasanu. J.. T. J. R. P. B. Academy of Management Review. R. 138–147). D. R. D. G. Sosik. Hollenbeck. The impact of cultural values on employee resistance to teams: Toward a model of globalized self-managing work team effectiveness. S. (1995). 22(3). D... 132–133. Kline. B. Leadership theory and research: Perspectives and directions (pp... & C. M. 50. A. Calderwood. 15(3). (2001). Effects of leadership style and problem structures on work group process and outcomes in an electronic meeting system environment. Team research in the 1990s. Sosik. (1997). The impact of demographic heterogeneity and team leader–team member demographic fit on team empowerment and effectiveness. Ilgen. Journal of Experimental Social Psychology. R. F. San Francisco: Jossey-Bass. Predicting team performance: Testing a model in a field setting.. B. S. In R. Group Dynamics.. In G.. Knowledge worker team effectiveness: The role of autonomy. (2001). B. Jackson. interdependence. 542–571. (2002). M.236 modeling complex systems Ilgen. Zsambok (Eds. J. D. D. J. 291–332).. The social psychology of organizations (2nd ed. Klein. (1996). 877–904. Kline. Kirkman. D. (1980). (1993). S. D. team development. (in press). New York: Wiley. & Kelly. Annual Review of Psychology. S. Diversity in work teams: Research paradigms for a changing workplace. (1997). R. Decision making in action (pp. D. N.). & MacLeod. Major. M. 334–368. J.

). & Salas. T. San Francisco: Jossey-Bass. research and methods in organization (pp. 2. Mathematics and Computational Modeling.. Greenwich ct: jai. pp. Cambridge. W. E. 179–192. Sims. E. 12. Kozlowski. Losada. Wasserman & J.. & Magor. 27–39. 623–645. pp. Ilgen.). Work groups and teams in organizations. P. 88. temporal. (2003). A. Kozlowski. Journal of Applied Psychology. W. C.. R. and emergent processes. A. & S. Advances in interdisciplinary studies of work teams (Vol. Research on impacts of team leadership on team effectiveness. S. K. Advances in social network analysis: Research in the social and behavioral sciences (pp.. & Brass. 266–277.. (1951). In W. Intraorganizational networks: The micro side. London: Wiley. Kozlowski. D. G. G. (1999).. 3–90). J. A. Ilgen & E. and new directions (pp. Weissbein. J. D. J. San Francisco: Jossey-Bass. Cannon-Bowers. The complex dynamics of high performing teams. Multilevel theory. A. Personal experiences and team effectiveness during a commemorative trek in the high arctic. K. Environment and Behavior. H. S. research. and methods in organizations: Foundations. personnel actions. E. J.. E. principles. 157–210). 30. 333–375). Galaskiewicz (Eds. In K. J. & Smith. Klein & S. K. Klirnoski (Eds. W. (2004). In K. W.. Entin. & Bell. A.. S. Nason. D. List. B. LePine. 33. Training and developing adaptive teams: Theory. M. Kuo. Entin. D.. W. Handbook of psychology: Industrial and organizational psychology (Vol. Kozlowski (Eds. Journal of American Academy of Business. J.). R. Klein (Ed. C. Structuring and . Thousand Oaks ca: Sage. (2000).). D.237 Team Effectiveness in Organizations systems model in woodlands mechanical harvesting. 386–401. Team adaptation and postchange performance: Effects of team composition in terms of members’ cognitive ability and personality. J. & Serfaty. S.). Multilevel theory. Beyerlein. J. Kozlowski. (1994). K. M. B. San Francisco: Jossey-Bass. E. S.. New York: Harper & Row. J. & R. Leon. (2000). J. In M. Making decisions under stress: Implications for individual and team training (pp. Developing adaptive teams: A theory of compilation and performance across levels and time. R. In J.).). Cannon-Bowers & E. MacMillan.. M. In D. (1998). J. 5. (1994). D. Brown. 15–153). Kristof. extensions. 36. Johnson. N. J.. Beyerlein (Eds. S. 229–253).. D. M. Lewin. W. S. Kozlowski. and research. A multilevel approach to theory and research in organizations: Contextual. G. Salas (Eds. 207–229). 240–292). (2003). & Klein. Brown. K. Borman. Pulakos (Eds. (1995)... L. Krackhardt. (2004). Field theory in social science.. E. and development (pp.. Human Relations. A multilevel approach to training effectiveness: Enhancing horizontal and vertical transfer. & Smith. R. (1999). The changing nature of work and performance: Implications for staffing. Gully. Washington dc: American Psychological Association. The virtual team: A case study and inductive model. In S.

. New York: Holt. Academy of Management Proceedings.. R. “I think we can. M. The study of groups: Past. (1994). Englewood Cliffs nj: Prentice-Hall.. B. Newbury Park ca: jai. and performance (TIP). 131–158. Rinehart & Winston. McGrath.. 6. K.. Small group research. E. pp. Jr. 1. Markovsky. J. M. J. Journal of Applied Psychology. present and future.238 modeling complex systems training high-reliability teams (Technical Report No. L. McGrath. (2000). (1964). 3(2). In B. The composition of effectiveness and ineffectiveness in R&D project teams. E. L. & Zaccaro. B. 19. E. McGrew. J. (2000).. G. Mathieu. J. (1994). L. A theory of groups as complex. International Journal of Cognitive Ergonomics. Salas. 22(2). G.). E. Advances in group processes (Vol.. & Kulisch. . Marks. Mischel. & Northcraft.. 120(3). & Mathieu. Group Dynamics: Theory. S. Journal of Applied Psychology. J.. J. (Eds. (2001). J. J. B. Groups: Interaction and performance.”: The role of efficacy beliefs in group and team effectiveness. 356–376.. J.. L. An analysis of team evolution and maturation. E. Time. (1984).). J. R. Performance implications of leader briefings and team-interaction training for team adaptation to novel environments. 30. & Cooper. J. Getchell. E. McGrath. H. (1999). Research. A synthesized model of team performance.. E. & Deeney. A. E. B. M. C. . Arrow. Burlington ma: Alphatech. & Cannon-Bowers. J. 85. J. M. adaptive. Measuring and managing for team performance: Emerging principles from complex environments. Academy of Management Review. Morgan. G. Bilotta... (2000a). Social psychology: A brief introduction. G.. Team effectiveness and decision making in organizations (pp. 149–203). dynamic systems. The relation between group cohesion and performance: An integration. M. 971–986. J.. McIntyre. C. & Glickman. & Berdahl. and Practice. 500–504. F. Salas. Journal of General Psychology. 209–234. S. (1997). M. (2000b). J.. that once and future field: An interpretation of the past with an eye to the future. Lawler (Eds. (1991). P. T. (1999).. 1. S. & Thordsen. 7–27. L. J. Zacarro. Personality and Social Psychology Review. 26(3). E. (1997). A.. E. 177–197). McGrath. Lovaglia. McGrath. Millitello. Salas. A. et al. Psychological Bulletin. F. 147–174. Klein. C... Goodwin. I think we can . Mathieu.. Kyne. M. H. 95–106. 210–227. A temporally based framework and taxonomy of team processes. Mullen. (1995). McGrath.. Greenwich ct: jai. & Salas. In R. 14. 277–291. San Francisco: Jossey-Bass. The influence of shared mental models on team process and performance. & Berdahl. Small Group Research. Heffner. interaction. Software team formation and decay: Extending the standard model for small groups. J. 115. Marks. Small Group Research. J. Arrow. A. J. J. E. S. A. E. & E. Dailey. W. M. E. (1976). 273–283. Guzzo.. 632). Morgan. G. ...

& Schneider. In N. & Ravlin. A. C. M. and their relationships (Contract No. London: Sage. Norwood nj: Ablex. (1999). R. E. Journal of Operations Management. Anderson (Eds. & Wright.. and performance among boundary-spanning teams: A conceptual framework and propositions for research.. L. J. In G. C. In C. T. Orasanu. their measurement. J. Chonko. R. In A. Making effective work teams with people. B. dahc 19-78c-0001). Oxford: Blackwell. Neck. & Manz. & S. pp. M. (1997).. The design and activation of self-regulating work groups. Productivity in organizations.). 40. Beyerlein (Eds. 185–212). Viswesvaran (Eds. R. J. Poulton. Handbook of industrial. Nieva. and models. Neuman.. Stagl. R. Salas. & West. Neal. P. Pearce. learning. M. Multilevel selection and prediction: Theories. 376–389. S. Anderson. L. M. O. Jones. The determinants of effectiveness in primary health care teams. Advances in interdisciplinary studies of work teams (pp. (2004). Ployhart. 619–639.. Johnson. (1993). K. & Salas. Evers. (1994). (2006). & C. A. Virtual teams: Creating context for distributed teamwork. G. C.. work and organizational psychology: Vol. K. Effective multidisciplinary teamwork in primary health care. E. J. Members of great teams think alike: A . (2002). 193–216). (1994)... In M. A. & N. 317–342. C. J. E. A. J. Washington dc: Response Analysis Corp. & West.. & Hall. C. (1999). Ones. 41. 7–24). 495–516). methods. C. & Roberts. Poulton. & F. 84.. 7–18. Sinangil. E. A. & C. B. (1993). J. E. Smit-Voskuyl. and networks (pp. Journal of Organizational Computing and Electronic Commerce. Journal of Advanced Nursing. W. 751–782. Organizational psychology (2nd ed. Handbook of personnel selection (pp. A. & LePine. Organizational variables. H. E. A. Decision making in action: Models and methods (pp. Journal of Applied Psychology. Klein. (2002)... sales force perceptions of readiness for change. Rangarajan. D. machines. Human Relations. M. B. Team effectiveness: Beyond skills and cognitive ability. G.).). A. E. Toward a continuum of self-managing team development. & Hesketh. A. Industrial Marketing Management. Klein. (1999). A. C. Team decision making in complex environments. J. V. Zsambok (Eds. E. Rentsch. J. (1978). & Reick. 2.. Orasanu. J.. Jentsch (Eds. Beyerlein. J. & Salas. Bowers. Calderwood. C. 252– 271.).. 13. Greenwich ct: jai. H. 18. Team dimensions: Their identity.239 Team Effectiveness in Organizations Navarro... (2005). Multiple case studies of team effectiveness in manufacturing organizations. Psychological Bulletin. 289–305. 918–925. Computer supported self-managing teams.).. Priest.. D. Pagell. 327–345). Washington dc: American Psychological Association. A. A. Developing a theoretical model of intercultural small groups: Understanding the effects of culture and cultural diversity on work group processes and outcomes. 20(5).. Journal of Interprofessional Care. Connerley. (1987). 4. D. B. Oetzel. Fleishman. 33.

In M. & Sipe. 325–355).. A. International Review of Industrial and Organizational Psychology (Vol. 24. Norwood nj: Ablex. Advances in interdisciplinary studies of work teams: Team development (pp. D. E. S. & S. J. S. E. T. I. Mahwah nj: Erlbaum. 779–802. Maynard. Proceedings of the ieee. 25 years of team effectiveness in organizations: Research themes and emerging needs. T.. S.. Greenwich ct: jai. Schneider. Den Hartog. & Wienk. M. Salas. Leader development for transforming organizations (pp. Personnel Psychology.). 555–599. C. Salas. & Burke. & Klein. (2004).). Halpin (Eds. M. Teamwork in the performing arts. 92. S. In R.. D. M. New York: Wiley. Beyerlein & D. & Burke. S. J. (2000). Paper presented at the 46th annual meeting of the Human Factors and Ergonomics Society. & Rouse. Schneider. Team cognition: Understanding the factors that drive process and performance. In D. & Tannenbaum. (2004). J.. Amsterdam: jai. Cooperation and teamwork at work. Encyclopedia of applied psychology (Vol. pp. Zaccaro. A.. Huang. K. Johnson (Eds. (2000).. A. D... (2000). Ruel.). H. J. (2003). L. Developing teams and team leaders: Strategies and principles. E.). Cooper & I. E.. 36(5). Toward an understanding of team performance and training. K. In M. Reflexivity and diversity in teams: The moderating effects of outcome interdependence and group longevity. H. 4.). Johnson. & S.. W. pp. & Fiore. Koopman. 339–356. Journal of Organizational Behavior. International Journal of Management Reviews. & Stagl. D. & Cannon-Bowers.. In C. 323–331). (2004). A. In C. A. (2004). Rouse. A. Beyerlein (Eds. W. M. In M.. Salas.. Is there a “Big Five” in teamwork? Small Group Research. 1. K.. Salas. Day.. Speilberger (Ed. Schippers..). 173–186). C. Norwall ma: Kluwer Academic. E.... D. W. Robertson (Eds. M. (1987). present and future (pp. Smith. L. 40. M. Stagl. B. 437–453. & McDevitt. Y. 19. D. P. E. & Cannon-Bowers. R. (2000). E. Robertson. (2002). L. E. Sims. (2005). 3– 29). W. (1992). Personnel selection psychol- . 1. C. E. Baltimore. Beyerlein. S. S. N. (2004). P. C. Burke. Teamwork: Emerging principles. C. Salas. C. S. Burke. Work teams: Past. M.240 modeling complex systems model of team effectiveness and schema similarity among team members. E. The people make the place. B. C. 47–91). D. pp. 497–505). San Diego: Academic.. Converse. Reconsidering our team effectiveness models: A call for an integrative paradigm.. Telecommuting: An overview of emerging macroergonomic issues. J. Beyerlein (Ed. Sims. J. B. Washington dc: American Psychological Association. Dickinson.).. Teams in organizations: Lessons from history. Salas. M. S.. Salas. Advances in interdisciplinary studies of work teams: Theories of self-managing work teams (Vol. 606–615. Salas (Eds. Salas. M. T. Teams: Their training and performance (pp. Swezey & E. 223–261).

Conger & R. Zeisig. J. (1985). Team dimensional training: A strategy for guided self-correction. and methods in organizations: Foundations. Brodbeck. (1998).). J. 340–366. M. (1999). S. A. Sheard. S.. & McPherson. The practice of leadership (pp. M. J.. Fort Benning ga Fort Benning Field Unit. The fifth discipline: The art and practice of the learning organization. K. pp. Spreitzer. M. M. 443–454. San Francisco: Jossey-Bass. S. E. 71–92). A. pp. Developing effective self-managing work teams in service organizations..241 Team Effectiveness in Organizations ogy: Multilevel considerations. A. Steiner. 25. Best practices in team leadership: What team leaders do to facilitate team effectiveness. research. Sonnentag. & Guzzo. N. Washington dc: American Psychological Association. W. & Heinbokel. & Schemmer. Research in personnel and human resource management (Vol. 503–530. 5. Noble. Journal of Management. Group processes and productivity. 133–151. A.. Rowland & G. New York: Doubleday. A. & Campion. K. From loose groups to effective teams: The nine key factors of the team landscape.. In K. (1972). J. A. L. 24. New York: Wiley. Research on managing groups and teams: Groups in context (Vol. 9. (1994). C. Greenwich ct: jai. A.). P. Neale. Staffing work teams: Development and validation of a selection test for teamwork settings. Salas (Eds. .. R. 55. G.. Army Research Institute for the Behavioral and Social Sciences Report). Making decisions under stress: Implications for individual and team training (pp. & Ledford. and new directions (pp. and ability requirements for teamwork: Implications for human resource management. Kozlowski (Eds. (1999). Frese.. (1999). 172–198). Senge.. (1997). 21(2). A. Orlando fl: Academic. 271–297). In K. A. International Journal of Human-Computer Interaction. Sheehan. Salas.. 323–356). skill. Stevens. S. Mishra. (1990). E. Acton. P. K. Journal of Management. & Martin. R. & Cooke. (2003). A. S. D. R. Wageman (Eds. B. Journal of Management Development. J.. Spreitzer. Use of design methods. 3–90). P. & Kakabadse. J. E. G.. A. Ferris (Eds. M.). and team effectiveness: A follow-up study in software development projects. (1987). Stevens... M. (2001). E. 207–228.. G. M. W. C. Cohen. Multilevel theory. Australian Journal of Psychology. A. Group and Organizational Management. G. In J. Stagl. 2... Eisner.. Greenwich ct: jai. extensions. L. D. Groups as human resources. team leaders’ goal orientation. Price. Mannix. Klein & S. G. In M. In J. G. Shea. & Campion. 20. S.. M. Predicting process improvement team performance in an automotive firm: Explicating the roles of trust and empowerment. M. (in press). Riggio (Eds. T.S. F. The knowledge. R. Smith-Jentsch. & R. & Burke. F. The definition and measurement of small military unit team functions (U. M.. E. Cannon-Bowers & E.. A. Understanding diversity to maximize work team effectiveness: Field studies designed to unravel the complex relationship between diversity and team effectiveness.). Shiflett. J. C. 144.). I.

& Altman. M. Tubbs. July). A. Personality. Tompkins. D. 61. (2000). (1999). (1997). Walker. D. (2000).). (2004). In M. 135–148. M. J. A. 120–133...). T. Team self-management. Leading work teams: Directions for team effectiveness. pp. E. S. 1. W. Tjosvold. structure. & Barrick. Journal of Engineering and Technology Management. von Bertalanffy. Teamwork in real-world dynamic environments. Johnson (Eds. 39–66. 131–138. B. I. M. Advances in interdisciplinary studies of work teams (Vol. D. K. 175–209. DeMeuse. Academy of Management Journal. 16. Tata.. Wong. Developmental sequence in small groups. 248–265. M. 1–10. . 45. J. R. Tambe.. Greenwich ct: jai. G. & Falconer. 27. W. A developmental approach to organizational learning teams: A model and illustrative research. 15–27. Beyerlein & D. (1956). General systems.. Physical environments and work group effectiveness. Tuckman. 16. & Salas. Test of a model of organizational contributors to product development team effectiveness. 278–284. B. Australian multi-unit residential project construction time performance factors. Linking treatment to outcomes through teams: Building a conceptual model of rehabilitation effectiveness.. Paper presented at aaai Fall Symposium on Plan Execution. & Futrell. (1996. & Prasad.. L. 4(18). Susman. Research in Organizational Behavior. (1965).242 modeling complex systems Stewart. L. M. & S. D. & Pounder. & Ray. Sundstrom.. D. E. jai. J. 79–96). Beard.. Kelley (Ed. 43. Advances in interdisciplinary studies of work teams (pp. Topics in Stroke Rehabilitation. E. Sundstrom. (1964). American Psychologist. (1990). J. Work team structure and performance: Assessing the mediating role of intrateam process and the moderating role of task type. and judgments of team effectiveness.. (2002). 117–153). J. C. (1989). (1994). P. Tannenbaum. & Remdisch. Amsterdam: Elsevier. 223–245. R.. Advances in Interdisciplinary Studies of Work Teams. 281–302). L. Nibler.. A. S. S. Journal of Managerial Issues. I.. Issues. Work teams: Applications and effectiveness. Sociometry. 11. I. (1992). and research in industrial/organizational psychology (pp. Yearbook of the Society for Advancement of General Systems Theory. (1997). Engineering Construction and Architectural Management. 63. A. 469–487. Johnson. In K. G. organizational structure. Tuckman. & Vines. L. In M. The historical roots of self-managing work teams in the twentieth century: An annotated bibliography. Stoker. Teamwork and controversy in undergraduate management courses in Hong Kong: Can the method reinforce the message? Swiss Journal of Psychology. R. T. theory. S. Beyerlein. S. W. C. Beyerlein (Eds. (1997). H. Strasser. Psychological Bulletin. 384–399.. Team building and its influence on team effectiveness: An examination of conceptual and empirical developments. T. group composition and group functioning. 4.). I. 7(3). 1.

). Robertson (Eds. M. (1982). & Unsworth. 385–402. & Lester. T. Yeatts. S. International Review of Industrial and Organizational Psychology (Vol. G.. Borrill. K. E. 12(4). & Hyten. D. P. (1998). S. In C. J. International handbook of organizational teamwork and cooperative working. . C. P. (2003). Organizational Behavior and Human Performance. Human Resource Development Quarterly.. (2001). W. pp. Chichester: Wiley.243 Team Effectiveness in Organizations Werner. A. M. 1–48)..). Individual versus group problem solving: An empirical test of a best-member strategy.. Chichester: Wiley. C. Cooper & I. (Eds. Yetton. K. D. 13. A.. L. M. & Smith. (1998).. Thousand Oaks ca: Sage. 307–321. High-performing self-managed work teams: A comparison of theory to practice. Tjosvold. Applying a team effectiveness framework to the performance of student case teams. 29. West. Team effectiveness in organizations. West. L.. C. & Bottger. W.

.

a later expression of his early inclinations toward integrative and holistic thinking. Mahoney University of North Texas In the early 1970s. Mahoney showed a homemade film to illustrate his position. The first scene opened on a bleary-eyed graduate student. Personally.” For the next 30 years. then a graduate student. which he opened to lap up one of several puppy treats. He then returned to his task at the typewriter. One of this volume’s editors. and the camera followed him to the refrigerator. As the scene faded. working at a typewriter. Change and stability are the twin engines of human consciousness. too. attended a debate between Mahoney and some of the old-guard figures of behaviorism. He became a leading figure in the constructivist movement. but so. when clinical psychology was undergoing its cognitive revolution. therefore I think I am. The story gets complex. Suddenly the dog stopped. Mahoney delighted the psychology community with such demonstrations and parables. the camera panned to the typewriter to reveal what the dog had been writing: “I type. 2006. His work on chaos theory and related ideas in development and holistic psychotherapy led to his participation in this volume of the Nebraska Symposium on Motivation. A broad scholarly community was deeply saddened when he passed away on May 31. This was repeated until all the treats had been consumed. This volume of the Symposium is dedicated to his memory. a young psychologist named Michael Mahoney became a voice for paradigmatic change. who paused to stagger to the refrigerator and quaff a soda—uniquely human self-reinforcement? The scene faded to a dog pawing a typewriter. That story has fascinated many over the course of human history and the short span of our individual lives. do humans. ultimately. How do people change? What is it that changes when a person changes? Why do some of us seem to change more easily than others? Why is change so difficult for so many? And. With a style and wit that became legendary. what should we do? . human change processes have long been at the heart of my passions as a scientist.Constructive Complexity and Human Change Processes Michael J.

This means. To better serve those whose primary interests may be more practical than theoretical or evidential. a list that has been considerably winnowed. however. A common comment made by colleagues. I will conclude with a brief discussion of how appreciations of dynamic complexity might inform professional life counseling. It is the ratio. Before offering those gestures. and practitioners. Technicalities of theory and research evidence can be found among the resources cited in the reference list. The growing interest in complexity is reassuring to those of us who have been following its scientific scent for some time. Why that sense of awe transformed itself into formal philosophical reflection remains unknown. Being Boundaries and Knowing: A Brief History It is hardly coincidental that ancient Greek philosophy emerged from a sense of awe regarding the phenomenon of change. Pythagoras had anticipated this idea and had suggested that the ultimate language of such a realm would be mathematical. that serves as . and changeless forms in a world that lay beyond our senses. however. The ultimate truth about reality would be expressed in numbers and their ratios. and—as Alfred North Whitehead put it—all simplification is oversimplification. at the beginning. But Aristotle asserts that the original motivation for philosophy was awe. but some of it is also due to the use of technical terms that are unfamiliar. students. is that the literature on complexity is too complex. They find it difficult to understand. Thales and his followers wondered about a central paradox: How is it that everything changes and yet there is permanence? What is the relation between change and stability? When water changes its form—from ice to rain to steam—what is it within or behind these changing forms that remains the same? An influential answer was later offered by a young man nicknamed “Broad Shoulders. Let me begin. incidentally. of course. My goal is to use language and concepts that are familiar and accessible. permanent. Plato suggested that there was a realm of perfect. however.” We call him Plato. that I will be simplifying. let me explain the nontechnical tone of what follows. Some of the difficulty may have to do with the subject matter.246 modeling complex systems I shall not presume to offer more than reverent gestures in the direction of responses to such questions.

Some cell parts became specialized by serving various life-support functions. Sexual reproduction involves the exchange of nuclear material. increasing connection can itself promote an expression of inner order. For a very long time. that ideas. . We like to call these central ordering processes nuclear. Boundaries separate. Western civilization has been steeped in three assumptions: that stability (order) is more permanent. Other parts of the inside begin to serve the coordination of the other parts. there came the first centers. It had some disadvantages. nucleated cells (eukaryotes) were able to adapt themselves to more planetary environments than their less-centered peers (prokaryotes). It was literally the archetypal contrast. and the body. and Cycles of Exchange Suppose one were asked to list the four greatest leaps in the history of life on our planet. They also do not diversify. something we now call a cell membrane began to form. The insides of a cell become more connected.247 Complexity and Human Change Processes the base concept of Western rationality. but they also serve to connect. Out of the chaos of primordial soup. and the intellect are better sources of knowing than experiences. This continued as the only form of reproduction until there was sex. Over time. Life-forms that reproduce asexually do not die. and that the mind and the body are separate. numbers. real. It introduced novel combinations of the centers of living creatures. First came boundaries. Coordination and complexity became both more common and more demanding. growth was achieved by means of two kinds: enlargement (hypertrophy) and division (hyperplasticity). sensations. The Evolution of Planetary Life: Edges. Thanks to Pythagoras and Plato. Nucleation was the second major leap in life on earth. but. What do boundaries do? They separate. Cells enlarged to the limits of their membranes and then divided into new cells. Chemical concentrations become different on one side of a membrane than on the other. Sex was the third major leap of life on the planet. by and large. After the first edges (boundaries). and beautiful than change (disorder). Sexual reproduction is credited with two contributions to life: diversification and death. Centers. These three assumptions are being challenged by complexity studies. They simply replicate themselves through endless divisions.

We become complex. Change and Complexity Those who seek the origins of complexity studies often find themselves chasing rabbits that appear and disappear across eras. and cultures.g. all of whom are social. Symbol use occurs only among a few life-forms. arts.. and it is through our diversity and connectedness that we continue to develop. Science progresses by means of well-connected networks of clever and careful seekers. Darwin’s circle of acquaintances encouraged him toward conjectures about change. So do centers. To summarize this brief history of biological evolution. and we change at accelerated rates. if the earth itself showed evidence of change. But. it is apparent that rates of change in everyday worlds increased sharply during and after the European Renaissance. if life-forms had changed. might not there have been changes in the consciousness that had emerged at the primate end of the spectrum? Darwin’s “dangerous idea” has influenced much of what has followed. Instead of exchanging only their nuclear material. Sociality and symbolic processes seem to be intimately connected. The essence of science is connected knowing—a term recently popularized by feminist scholars. and we continue to seek and create order in our lives.248 modeling complex systems The fourth and most recent leap made by life on earth has been the creation and exchange of symbols. We are life-forms with a legacy of coordinated centering. Archaeology and geology were still infant disciplines. symbol-using creatures exchange information and expressions of their own experiencing. The discoveries of Copernicus and Galileo emerged out of communities of inquiry that were becoming increasingly connected. civilized community. An important part of what emerged out of that era was the scientific revolution. In Western civilization. But the real godfather of change was Charles Darwin. continents. Francis Bacon’s valuable contribution was to organize the support and encouragement of such communities. was it not possible that life itself had changed? And. . science. life as we know it originated in contrasts. We diversify and exchange. technology). Boundaries continue to serve critically important roles in our lives. All that we think of as modern is an expression of our accelerating exchange of symbols (e. Again reflecting the importance of connections.

etc. It is in many ways a new expression of ideas that have been cycling through multiple cultures for . We tend to classify. Aristotle’s earliest work was on categories. An appreciation of complexity and the dynamics of developing systems is now sweeping through many of the sciences and large sectors of the humanities. Contrasts and the tensions they reflect or create are also at the heart of our planetary life. He formalized logic. areas of study. beings.1 This shift is made easier by a conceptual revolution in which we are now living. psychology. Why do we classify? Because it is a way of creating order out of chaos. But it is not simply a new kid on the block. among other things. not separate species. Reason and experience need not be cast only in terms of conflict. and neuroscience. And order and chaos are generative tensions.249 Complexity and Human Change Processes Darwin’s two greatest contributions were his emphasis on mutability (change) and emotionality. Our emotions reflect our embodiment. We are moved to change—put into literal motion—by a complex interaction of processes. It certainly does embrace and elevate the role of disorder in development. The “new look” in the cognitive sciences and psychotherapy is one that vigorously embraces embodiment and emotionality. Just look at psychology. The words themselves reflect this legacy. He suggested that there are two kinds of people in the world: those who believe that there are two kinds of people in the world and those who do not. Motivation and emotion are based in movement. Before elaborating much more of this story. In mammals—and particularly in social primates like us—these dimensions are related. it might be wise to recall an aphorism attributed to Arthur Schopenhauer. How do we classify? By creating contrasts. He spent much of his life classifying (things. and this realization has resulted in the erosion of mind-body dualism. that we are categorical beings. and we use his logic to classify and organize our lives. Body and mind can no longer be meaningfully separated. Constructivism: A New Expression and Ancient Wisdom Let us now move closer to the issue of how we are structured as human beings and how we change. This means.). It is those as well. Others call it complexity studies or dynamic systems theories. That appreciation is drawing particular attention in biology. Some call it chaos.

With such a breadth of themes. the following individuals have played roles in the history of constructivism: Lao-Tzu. (4) the relational matrix of experiencing. In the interest of focus. Throughout a fascinating diversity of expression. Johann Herbart. Arthur Schopenhauer. What is constructivism? Succinctly. existential-humanistic. The emphasis is on reciprocity and the efficacy of an agent who is participating in life by influencing the world by which it is being influenced. and connected lifelong learners. however. most apparent in our immersion in social and symbolic networks. William James. Given this remarkable conceptual concordance. Heraclitus. This consideration of the person as an active agent is an important contrast to former portrayals that viewed humans as passive pawns in a mechanical universe. Humans are not simply reactive to the forces of their environments. Giambattista Vico. Thus. and (5) lifelong developmental unfolding. complex. (3) the centrality of an embodied identity—often called a self—in the organization of experiencing. constructivism is a term now commonly applied to a family of theories—a metatheory—that share a view of human beings as active. Immanuel Kant.250 modeling complex systems millennia. In terms of name croppings. behavioral. . constructive theorists share five basic themes: (1) the active agency of the human organism. Viktor Frankl. Friedrich Hayek. We are proactive as well. Alfred Adler. We choose. Hans Vahinger. Lev Vygotsky. active agency The active agency of the person is a key feature of constructivism. and Albert Bandura. Bandura has termed it reciprocal determinism. Jean Piaget. not surprising to find that former theoretical factions have encountered one another on common conceptual ground. (2) the importance of order or meaning in human experience. I shall forgo a digression into its rich legacy in Eastern and Western philosophy and science. it is. A popular term for connecting these perspectives is constructivism. and psychodynamic perspectives have all used the term constructivist to describe some of their most recent developments. perhaps. a closer examination of the five themes is warranted. Jerome Bruner. We change our environments in ways that change us. representatives of cognitive. Buddha. George Kelly.

. Note the practical and ethical implications for respecting individual phenomenology. Biological self-organization is a demanding task of infancy. human infants organize their cycles of sleeping and being awake in congruence with cycles of light and dark (day and night). the real self. establish. identity: the mysteries of a changing stable self The third theme of constructivism is connected to individuality in these expressions. Parental care of offspring is a common example. Most relevant here is the role of emotions in organizing our styles of being. We seek to find. It is not coincidental that one of the pioneering classics in the women’s rights movement was titled Our Bodies. emotions have contributed to our flexibility in adapting to new challenges and in deepening our capacities to relate to one another. Personal identity is embodied and emotional. Among other things. One of the most powerful expressions of this organization is apparent in early emotional development.251 Complexity and Human Change Processes ordering processes The second theme of constructivism is that much of human activity is devoted to organizing experience. Recurring patterns in nature have helped. We are all order freaks. or a true self. The paradox of the changing yet stabilizing self is the focus of considerable study. some more so than others. Ourselves. We experience from a reference point—an embodied perspective or positionality that is paradoxically stable and yet ever changing throughout life. What is clear—and a point that is emphasized in constructivist writings—is that each person experiences his or her world in a unique manner. for example. what we call personality and personality traits are fundamentally emotional patterns—stylized expressions of our literally feeling our way through life. there is little debate about the body being the base from which a person experiences. Although there are debates about whether there is a stable psychological entity that could be called the self. Within their first year of life. Emotions are expressions of life organization that made their first appearance with the emergence of mammals about 165 million years ago. As extensive evidence now reflects. and maintain a stable order in and from which to live our lives.

we often change most in the midst of desperate attempts to avoid changing. This is the essence of the fourth theme of constructivism. All of the so-called higher cognitive processes—including the symbolic—are founded in our emotional.252 modeling complex systems relatedness: social symbolic processes But individual selves cannot be separated from their support systems. Again and again we make new gestures toward regaining our equilibrium as a dynamic living system. In humans. Our warrant for such awe is further amplified by the fifth and final theme of constructivism. Order begins at home. we are psychologically relational by our very nature. in a base camp of biological embodiment. We live in and from relationships. We live in multidimensional webs of relationships that are woven in large part by our symbolic processes. Relationships change. This is a lifelong fact. And this is where the five themes come full circle. relational embodiment. Indeed. it refers to the fact that we continue to unfold over the course of our lives. a sense of self develops out of formative emotional relationships and expressive symbolic skills. Activity takes the form of order seeking. We are often thrown off balance. We continue to change. Again and again—for a lifetime—we seek a stability that we never quite achieve. Our worlds change. whether we want to or not. Skills in the use of symbols have helped us learn from and teach toward generations we can barely imagine. In more accessible terms. Our thinking is not just “in our heads. lifelong development: idiodynamics The complexity of being human is itself awesome. Not only are we biologically dependent on our early caregivers. Our bodies change. . Our inner life—or at least the small part of which we become conscious—is saturated with symbols. Situations change. Instead of thinking of them as a linear sequence of stages.” however. we would be living in and from symbolic relationships. Even if we were to live in the isolation of a cave on a remote mountain. imagine them as points on the face of a clock. Our life order is challenged. In technical terms it is dynamic dialectical development. Symbols share features that include the re-presentation of that which is currently absent. But life keeps coming at us.

Some of these change-requiring circumstances may happen suddenly—for example. There are small.” a “tipping point. yet we live—literally and figuratively—in the essential tensions that mark the edges of chaos. clothes. Complex-systems scientists tell us that this is a necessary fact of our existence. with the loss of a loved one. an accident or injury. hair. .253 Complexity and Human Change Processes There is an ironic paradox in human complexity. This is what Gregory Bateson liked to call first-order change. Sometimes an equally demanding need for change emerges from within. Contrast this with second order-change.g.” or an intense internal “itching. two kinds of change There are two kinds of change in a person. We tend to change as little as possible to get by. Permit me to translate my limited understanding of these matters in the direction of some practical suggestions for how we counsel professionally. But we are sometimes forced toward larger and more fundamental changes. or a change in where we work or live. we prefer to change smaller and more superficial dimensions (e. off-balance lives? I do not believe this to be the case. It often involves changes in appearances and refinements in serving a present life organization. We long for order. This existential fact needs to be addressed with practical compassion. in which there are basic changes to the foundations and operational organization of a life. Biologically and psychologically. But not all catalysts for change arrive suddenly or from the outside. What does all this chaos and complexity mean for us in our everyday lives? Are we doomed to live hypertense. It may take the form of a “crystallization of discontent.. vehicles). gradual adjustments that amount to a kind of remodeling. In these instances. Human Change Processes How we change is intimately related to how we stay the same (Appendix A). When given a choice. we are relatively conservative creatures.” The point is that there come times when a personal transformation is required. there is an identifiable event that has challenged us to change.

Much of contemporary psychotherapy—particularly time-limited and manual-driven psychotherapy—amounts to problem solving. Beyond solving their problem(s). But rare is the client without multiple recurring problems. This is often what clients want. There is. their occupation. and psychotherapy that begins with a focus on a single problem may not remain so focused. These are often emotionally charged preoccupations. and the distinction between kinds of change is more complex than I have here presented it. But not all problems are simple or soluble. or their life partner and community. perhaps. More often than not. Let us call them problems. For some clients at some points in their lives.254 modeling complex systems A second-order change is often called a transformation or a personal revolution. and processes Consider the prototypical circumstances surrounding clients’ presentation for psychological services. In some instances people may actually change their name. In my own work as a therapist I find it useful to imagine three interacting levels of focus. They are typically struggling and. it may be called a quantum shift. Perhaps a translation into clinical practice will help clarify that complexity. Recurring problems may reflect patterns. many clients want to understand the patterns they are experiencing. or both) and that something needs to be done. or change the way you want them to be. of course. When a second-order change happens suddenly. and they seek professional counsel on what is happening in their life and what they can or should do. then there are two primary ways of solving problems: change the way things are. patterns. Broadly defined. The felt sense of my clients may be that there is something wrong (with them.” If this definition works. They are often puzzled. a problem is a painfully felt discrepancy between the way things are and the way they “should be. But suddenness is not common. “Why do I repeatedly put myself in abusive relationships? Is there something wrong with me? Does this mean I am crazy? Am I [fill in a dsm classification]? Am I this way because [fill in a developmental event or context]?” Their . the level of problems. their world. problems. their self-presentation will focus on one or more current life concerns. I serve best by offering suggestions as to what might be done. suffering.

embodied. Such work appeals to a wisdom that cannot be localized in the left hemisphere of the neocortex. and healthy expression of our essence as order-seeking creatures. It was not formally recognized or respected until the beginnings of dynamic systems writings among the Scottish moral (social) philosophers of the 18th century. of course. Teleological and Teleonomic Order There are also two kinds of order evident in most human activity and development. and experimental.3 Although formal exercises (in session and as homework) can add structure to these ventures.2 To be optimally helpful in developmental processes. exploratory. I call it the process level. This deterministic directionality is so familiar to us that it is difficult to imagine any alternative. Process-level work can. If we are at point A and we desire to be at point B. But there is also a second kind of orderliness in development. We are. All rational planning and concrete problem solving reflect teleological directionality. This level of inquiry might be called a pattern level. Process-level work in psychotherapy tends to be more spontaneous. Activity is at the heart of the third level of focus in psychotherapy. our direction of movement is determined by point B. Teleological order is direction imposed by a specific destination. and emotional than the solution-focused planned interventions of problem solving. Work at the level of process tends to be experiential. The quest for meaning is a natural. three levels of focus. Let me introduce another contrast that may further serve an understanding of how we construct orderings to cope with complexity. two kinds of change. their stochastic movements cannot be completely anticipated or predicted. be among the most challenging and rewarding (for both client and therapist). The term teleonomy refers to a spontaneous . common.255 Complexity and Human Change Processes requests for explanations are requests for order. But this is where—I reluctantly confess—much of psychology’s theorizing has failed to serve constructively. Note that I am using contrasts to structure (organize) my remarks: two kinds of people. always “in process.” even when we do not acknowledge or label that fact. order seeking must reflect and serve activity. therefore. Clients are searching for meaning.

We have inherited the burdens and blessings of the Greeks’ early love of rationality and its applications to the design of our social systems. Their patterned orders reflect the operation of principled processes. It is the result of human action.256 modeling complex systems order that emerges in complex open systems. But there are undeniable principles that constrain and organize what . But the orders reflected in biological evolution and a person’s life-span psychological development are different from planned. The two most common examples of teleonomy are biological evolution and human personality development. Adam Ferguson. I here call it flow to emphasize the contrast with mechanistic force. Hayek called this the primacy of the abstract (abstract implies “tacit”). but one can never hope to perfectly predict the particulars. He distinguished between the level of principles and the level of particulars. and David Hume. An appreciation of that difference is critical to our being able to provide responsible professional counsel on human lives and their conduct (Mahoney. a significant limit to what can be rationally designed and institutionalized without causing harm. Viewed backward— historically—both reflect a remarkable orderliness. Teleology is direction (orderly movement) defined by a specific destination. in press). predicted. however. They follow rules of order that can never be completely specified. One can approximate a description of the principles by which complex systems operate. Complex interactive systems exhibit complex expressions of chaos and both kinds of order (teleological and teleonomic). They recognized that there is a second kind of order that appears in human conduct. teleonomy is direction (orderly movement) without a single or specific destination. An example in psychotherapy is that of language production. Flow reflects the order that spontaneously emerges from within the ongoing activity of its constituent agents. Open. Those words are the particulars. but not of rational design. This was one of the central insights of Bernard Mandeville. teleological orders. complex systems express endless. rational. There is. Neither we nor our clients can accurately predict the exact words that may emerge from either of us in the coming moments. or controlled. dynamic exchanges. Early Greek political philosophers liked to distinguish between the laws of nature and laws designed by humans to govern their own conduct. Concrete particulars are the consequences of multiple interacting forces in a widely distributed and ever-changing network. Adam Smith.

Our insides are teeming with activity. A few. We are teeming with excitatory impulses. social custom. or ideas). will offer novel solutions to adaptive challenges. Every person is a live-in developmental laboratory. as David Hume noted. Retention processes strive to conserve life order—the very order that is being challenged. epistemic authorization. selection. our current concerns).257 Complexity and Human Change Processes does emerge (e. the languages we speak. that we have apparently internalized—literally. It respects a kind of wisdom in the distributed yet connected systems that are the developing person. Yes. Such work reflects an appreciation for evolutionary processes within the individual. voluntary and otherwise. Variation (diversity) is key. therefore. Their control. What does all this mean.). behaviors. One family of processes is bent on liberal protection of degrees of freedom in the expression of new forms. is exercised pri- . among other things. Process-level work respects the operation of three essential processes in developmental change: variation. etc. all development—requires variation or novelty. The variance generators live in essential tension with the order protectors. With a little luck these variants will be retained (through genetic coding. Process-level work in psychotherapy honors this second.4 All learning—and. New forms are a necessary part of adaptive change (whether expressed as organisms. our emotions are often far more powerful than our intellect would like to admit. incorporated (brought into our bodies)—the very processes from which we (and other earthly life-forms) have emerged.g. more conservative family. but. habitual use. This family of liberation is at odds with another. however. They will not be selected (naturally or otherwise) as “keepers” by the environments in which they are expressed. practically speaking? It means. rules of grammar. and we are quite protective of our boundaries. more complex and teleonomic kind of order. Most new forms will not be viable. and retention. Much of what is inside has become organized into systems and centralized structures. we are bounded beings.. It is worth remembering Sechenov’s discovery—published in Darwin’s time—that the primary activity of the nervous system is inhibitory. We like to believe that we are being rational in our selection processes. neural structuring. Karl Popper and Don Campbell—both pioneers in evolutionary epistemology—noted that the first and third processes in complex developing systems are in dialectical contrast.

Moment by moment (within sessions). of course. in the technical language of dynamic systems theory. At the other end of this hypothetical-kinds spectrum are clients who are stuck. There is too much stability in their experiencing. but they offer a starting point—a bifurcation. I may serve them best by first bearing compassionate witness to their difficulties and then helping them develop or regain a sense of center. Clients who are off balance often present with self-descriptions that emphasize crisis. anger. Many clients present a mixture of these extremes. Such clients may be best served by interactions and exercises that challenge their overly regular stability. I believe that our gestures can be refined. These skills become a secure base to which the client regularly returns as our work together moves toward later emphases on new ways of experiencing. Anxiety. and depression are common. Level on level. it is the equivalent of psychological constipation: a painful absence of movement. incoherence (confusion. 2001). we can see development as a dialectical (contrast-generated) development of lifelong gestures of balance. I often begin therapeutic work by assessing and encouraging refinement of centering skills. Let me therefore turn to some practical strategies that can be useful in psychotherapy (Appendix B). and a sense of having lost a sense of their center or life order. oversimplifying. distraction. As they move through their cycles of expansion and contraction. They follow the same ruts or routines of activity even though these routines are dysfunctional or unsatisfying. As one client put it.258 modeling complex systems marily through inhibitory processes. We live as and within essential tensions (Mahoney & Mahoney. my professional tasks are to comfort and challenge as my clients idiodynamically . All “two kinds” contrasts are. and most exhibit oscillations and dynamic cycles of looseness and tightness in their experiencing. Practical Implications: Two Kinds of Clients There are two kinds of clients in the world: those who are off balance and those who are stuck. falling apart). They may be experiencing a wide spectrum of emotions. and across sessions. my professional responsibility is to sensitively attune my counsel toward appropriate mixtures of comforting and challenging them.

it is a recurrent skill to be revisited and refined throughout the course of our work together. can encourage a sense of stability and order. their breathing). People are instructed to focus their gaze on a point or to pay attention to a repetitive sound or sensation (e. progressive muscle relaxation. are less likely to serve their clients as well as therapists who practice a relaxation process in the presence of their clients. which—after initial awkwardness—may invite further explorations of both literal and figurative . In addition to centering meditation exercises. and breathing exercises. Mantras and repetitive rituals can also induce feelings of order in the midst of chaos. Standing up into stability may also have metaphoric meanings for clients and therapists accustomed to traditionally seated “talking heads” psychotherapy.. for example. involve attention and balance. and there are important individual differences in what works best. Centering is not just a preliminary emphasis. It is important to note. however. guided imagery. It invites the risk of a novel experience. a “standing-center” exercise asks an individual to stand and focus attention on the sensations in the soles of the feet. Centering: Dynamic Balance Skills There are many different ways to teach and practice centering skills. Therapists who simply recite scripted instructions for relaxation. There are many different paths to a sense of center. for example.259 Complexity and Human Change Processes contract and expand. I like to teach centering skills through postural stability. I defer discussion of biomechanical aspects and laboratory evidence. It is a literal “grounding” experience that many people find inherently stabilizing. Consciously focusing on a fixed object or a rhythmic pattern. that one of the most important aspects of the teaching of centering skills in live iterations with clients is the continued learning and practicing of those skills by the teacher. I often use these in combination with some basic meditation exercises. Practically. Two of the most common skills in centering. For the sake of brevity. This is an approach that emerged in my early work with Olympic athletes and was later incorporated into embodiment exercises with psychotherapy clients. however. Examples include highly structured behavioral rituals.g. This is a common component of diverse forms of meditation.

even when we momentarily can’t find it. I may ask clients to close their eyes while standing in order to focus their attention inward and on the sensations in their feet.260 modeling complex systems movement (e. What are often termed disorders in psychology are. Note again that the healthy functioning of the whole person requires a dynamic tension between stabilizing and destabilizing processes. Other clients may best be served by an approach to psychotherapy that incorporates an ongoing practice of centering skills with exploratory exercises that encourage an appropriately paced experimentation with novelty. in fact. and home (or center) is always there. our body-brain processes intuitively know well how to find their way home. walking—if only in circles—can positively affect the course of therapeutic process). Becoming aware of it can serve to illustrate more general lessons about balance in life: e. which is essentially a disorder of overordering. these may be invaluable. We are usually unaware of this movement because it is so basic and habitual. Many of the other formally diagnosed disorders also reflect the hyperstabilization of routine patterns of thought. This is most apparent. perhaps. we are always moving (even when we are unaware of that fact). in obsessive-compulsive disorder. change often depends on exploratory activity and self-experimentation. For clients whose current needs are primarily for structure and . which may last for only a few minutes.. centering skills and a comforting presence may be the most important things I can offer in psychotherapy.g. Edging: Explorations and Experiments in Experiencing For some clients. and behavior. we are most aware of our center (or balance) when we lose it. This is particularly the case for clients who present as being more toward the “stuck” (overstabilized) than the unstable end of an order-disorder continuum.. Eye closure will also increase postural sway—the slight rocking motion by which the body continually finds and adjusts its balance. In time-limited therapy with individuals who are struggling to regain a sense of stability and structure in life. feeling. For persons suffering from such patterns. overstabilized routines of being in the world.g. can also illustrate some lessons that are both literal (embodied/concrete) and figurative (metaphoric/abstract). The standing-center exercise.

The fundamental strategy is to encourage selfcompassion. From this dynamic systems perspective. and enjoyments). movement meditation (which literally encourages clients to move toward new strengths. but laboratory. psychotherapy is a special form of human relationship in which and from which clients can find compassionate safety and—each when they are ready—begin to explore alternate ways of experiencing themselves. development is often expressed in cycles and spirals. Cycles and Spirals of Expansion and Contraction In complex open systems like ourselves. The idea is to introduce more variance into their life order. “Do something different” is a frequent homework assignment heard by my clients. their worlds. and feelings). and mirror time. our complexly whole beings move back and . clinical.261 Complexity and Human Change Processes stability. With such individuals a constructively challenging interpersonal style can be balanced with the more comforting style required when they feel destabilized. For clients whose current needs are primarily to break out of a rutted order. skills. I emphasize the development and practice of orderly routines. I emphasize exploration with new possibilities. Oscillations are always in process. thoughts. rearrange your closet. 2003a). Take a different route home. Just as our bodies move back and forth through a gravitational center when we are standing. Mirror time is a self-relational exercise that asks the individual to spend brief regular periods of time in front of a mirror. and field research have suggested that self-relational skills are important components in life quality and well-being. The only constraints are that these ventures into novelty be self-caring and socially responsible. Exercises for exploring and experimenting with new skills and different possibilities for experiencing cover the entire spectrum of techniques employed in psychotherapy and other forms of developmental life counseling (Mahoney. Not all clients will take kindly to the exercise or themselves. and their possible development. A common format is to suggest alternate minutes with eyes open (looking in the mirror) and eyes closed (focusing internally on bodily sensations. tune in to different channels on your radio or television. but the process must be paced and structured according to individual starting points and learning styles. Among my favorites are personal journaling.

262 modeling complex systems forth through an ever-changing base of viability. The same is true of our interactions in relationships. Episodes of openness and connection may alternate with episodes of closure and distance. Part of the unique challenge of being a mental health professional is the demand to develop an operating center that is large and flexible enough to safely hold all the destabilized centers being served. Psychotherapy can be a destabilizing experience for therapists, and this should not surprise us. We therapists are changed by our work. The life-span psychological development of psychotherapists is often accentuated and accelerated. Therapists should, therefore, be encouraged to prioritize self-care and to explore contexts and experiences that serve their own unique developmental path and pacing.

Concluding Remarks
Appropriately, I hope, I shall close with an opening. I began with a reference to my passions as a scientist fascinated with human change processes. Science and passion have been in a strained relationship for too long. The sources of the strain are less important here than are the means by which we might bring them back into creative balance. As a gesture in that direction, I shall conclude by taking a risk. One finds very little creative writing in the literature of scientific psychology, yet it is often through creative expressions—gestures toward novel experience—that clients and therapists best learn their dance. I encourage my clients to write, draw, sing, and sculpt their hearts out. Occasionally, they share some of their private creations with me. One client wrote a poem expressing her feelings as we moved toward the conclusion of our work together. She asked if I would write something she could take home with her. The poem that emerged for her is what I now leave with you: It’s a season of transition and you’re on the move again On a path toward something you cannot disown; Searching for your being in the labyrinths of heart And sensing all the while you’re not alone. Yes, you seem to keep on changing for the better and the worse

263 Complexity and Human Change Processes And you dream about the shrines you’ve yet to find; And you recognize your longing as a blessing and a curse While you puzzle at the prisons of your mind. For as much as you seek freedom from your agonies and fears And as often as you’ve tried to see the light, There is still a trembling terror that your liberation nears As you struggle with the edges of your night. For your Reason is a skeptic and rejects what it desires, Playing hard to get with miracles and signs; Till a Witness gains momentum and emerges from within To disclose the patterns well above the lines. Then a window has been opened and you’ve let yourself observe How the fabric of your Being lies in wait; And you want to scream in anger and you want to cry for joy And you worry that it still may be too late. For the roller coaster plummets with a force that drives you sane As you tightly grasp for truths that will abide; Never fully understanding that your need to feel secure Is the very thing that keeps you on the ride. You survive the oscillations and begin to sense their role In a process whose direction is more clear And you marvel as your balance point becomes a frequent home, And your lifelong destination feels like “here.” So with gentleness and wonder, with questions and with quests You continue on the path that is your way; Knowing now that you have touched upon the shores of inner life, And excursions deeper can’t be far away.

264 modeling complex systems There will be so many moments when an old view seems so strong And you question whether you can really change; And yet, from deep within you, there’s a sense of more to come And your old view is the one that now seems strange. Take good care, my friend, and listen to the whispers of your heart As it beats its precious rhythm through your days; My warm thoughts and hopes are with you on your journeys through it all . . . And the paths of life in process find their ways. Do be gentle, Process Pilgrim; learn to trust that trust is dear, And the same is true of laughter and of rest; Please remember that the living is a loving in itself, And the secret is to ever be in quest . . .

Appendix A: Human Change Processes, a Synopsis
1. Humans are active participants in organizing their experiences of themselves and their worlds. Dynamic and continuous ordering processes construct, maintain, and revise activity patterns. 2. Active ordering processes are primarily tacit and unique to each individual. 3. Those self-organizing processes most vital to life support and individual functioning—what might be called core ordering processes (cops)—are given special protection against changing. 4. cops organize experiences and activities along interdependent dimensions that include: a) emotionality and valence, b) reality status, c) personal identity, and d) power (control/efficacy/agency). 5. Resistance to change—even desired change—is common, especially when the change is experienced as “too much” or “too

265 Complexity and Human Change Processes quickly.” Such resistance reflects basic self-protective processes that serve to maintain the coherence of the living system. 6. Ordering processes always operate in relation to their own contrasts, which are disordering processes. Order and disorder are facets of the same dynamic diamond. 7. The ongoing interaction of ordering and disordering processes creates a simultaneous interplay of both familiar and novel experiences. 8. Novelty is necessary for learning and development. Novelty involves contrast. 9. Familiarity and consistency are essential to life support, systemic coherence, and human well-being. 10. The dynamics between familiarity and novelty lie at the heart of human change processes. 11. Order is necessary for development. 12. All learning and development involve transitional disruptions or perturbations in systemic functioning. At all levels, development feeds on disruptions and then digests them into familiarity. 13. It is in the context of disorder that a living system exhibits both its greatest rigidity and its greatest variability. When rigidity reigns, stereotyped behavior or frozen passivity is common. Variability may emerge, particularly in a safe relationship that encourages ventures into flexibility. From this flexibility in being and these expressions of variability, more fulfilling and functional activities may emerge and vie for potential selection in the expression of that person’s life. 14. When novel experiences—contrasts—are deficient relative to an individual’s capacities and developmental needs, stagnation and “hardening of the categories” are likely. 15. When novel experiences greatly exceed the individual’s capacities to balance, feelings of being overwhelmed are common. Episodic or chronic disorder and “breakdown” may result. 16. When new experiences are well timed and suited to the individual’s current developmental capacities and edges, developmental “breakthroughs” may emerge and effect whole-system transformations in experiencing. 17. Although disorder may be experienced and expressed in highly patterned processes of human activity, it is diverse, individually unique, and systemic; we shall advance in our attempts at con-

266 modeling complex systems ceptualization and classification only as we are willing to embrace the limits of symbol systems to capture human uniqueness and the ultimate ineffability of complex system dynamics. 18. Persons can become “trapped” in disorder. Many psychiatric disorders are, in fact, rigidly ordered patterns. 19. Change is a nonlinear process. It is neither continuous nor cumulative. Rather, change processes reflect many small shifts punctuated by occasional sudden leaps and frequent returns to earlier patterns of activity. 20. Change is often experienced in waves or multirhythmic oscillations. Anchors for dimensions of oscillation are often abstract conceptual polarities, such as life/death, right/wrong, good/bad, real/ fake, and sacred/profane. Descriptions of experiences of oscillation often include references to opening and closing, expansion and contraction, loosening and tightening, or approach and withdrawal. 21. Change emerges from a shifting matrix of competing possibilities. Old, tenured patterns of activity compete with new and experimental possibilities. Like all other forms of evolution and revolution, this competition is never finished. New patterns, when they gain dominance in the competition, themselves become the “old” and familiar order in contrast with which newer patterns emerge and compete. 22. Ongoing competitions in development are neither “won” nor “lost” in reference to allegedly absolute criteria. Some competitors (i.e., impulses of activity) are selected to assume temporary positions in the “driver’s seat” of the body. The old patterns remain as contenders, and they may “win” occasional episodes of ascendancy in future situations. Old habits are not eliminated completely, but they can be effectively displaced by new ones. 23. Selection processes always include human agency. What “decides” the ongoing competition among activity patterns is a complex dynamic system that emerges out of and expresses a human will. 24. Selection processes must mature into retention processes if a chosen activity is to become an influential pattern in ongoing self-organization. A change—to become a change that makes an enduring and, therefore, significant difference—must be actively practiced. 25. Successful (adaptive) change is facilitated by a rhythmic orchestration of exploratory, selective, and perpetuating activities (i.e., experiments in living, evaluative choices regarding which experi-

267 Complexity and Human Change Processes ments are “working,” and action patterns that serve to perpetuate and elaborate valued experiences). 26. Self-relationships, which emerge in social and symbolic relationships, powerfully influence life quality and resilience under stress. Awareness, acceptance, and celebration are common quests in self-relating. 27. Interpersonal relationships involving strong emotions are powerful contexts for psychological development. Safe and loving intimacy is an expression of trust, which lies at the core of optimal contexts for development. 28. Symbol systems including language and the arts may offer valuable structure and welcome stimulation in personal development. 29. Conscious practices, both spiritually and secularly pursued, are central to qualities and directions of life experiencing. Intention and action are key (even when the goal is stillness). 30. Psychotherapy should reflect an appreciation of the history and motivational power of personal realities, the role of interpersonal, symbolic, and self-relational processes in the maintenance and change of personal realities, and the complex existential agency and responsibility of the socially embedded individual. 31. Love is basic to life. What is basic to life is basic to psychotherapy. Spiritual and wisdom traditions that embrace these insights may be precious sources of comfort, companionship, and direction in the complex processes of life-span personal development.

Appendix B: Recommendations for Constructive Practice
1. Prepare for each session in private reflection. Even if only for a few seconds, take the time to nurture a sense of your own center and your intention to serve another developing being. 2. Honor the complexity and uniqueness of each client. Do not presume to know your clients. Let them be themselves, and invite them to share as much of their uniqueness as they are ready. 3. Give yourself and your client permission not to know and not to fully understand. Life is much more than figuring things out; effective life counseling does not require complete understanding or definitive explanations.

268 modeling complex systems 4. Let the clients set the pace, and honor their process. Some clients will want to move very quickly; some will not want to move at all. Pushing often results in pushing back or digging in. Expansions and contractions tend to alternate. Recognize and respect the importance of timing. 5. Encourage (but do not force) emotional expression. Allow your clients to feel, and to feel freely, but do not convey the message that they must be emotionally expressive if they are to make progress or win your respect or caring. When feelings emerge, invite elaborations and explorations. Seek to understand beyond familiar words and summary labels. Locate felt emotions in bodily sensations. 6. Allow and invite yourself to feel emotional in the process of counseling. Let your heart lead your helping. Open yourself to feeling. Learn your patterns around pain. Cultivate compassion, loving kindness, balance, and trust. Recognize that a primary responsibility of your role as a professional helper is to maintain a spirit of centeredness (balance) large enough to accommodate the combined energies of your client and your self. You will be challenged in your balancing abilities, particularly if you allow yourself to be emotionally alive to your interactions with clients. When you are overwhelmed (e.g., by intense emotions, whether your clients’ or your own), take a moment to breathe deeply and to refocus on your intentions to help the person you are serving. 7. Trust that your clients can endure their pain and be strengthened by the process. You cannot take their pain away, although you might often wish that you could. But you can hold a steady course of confidence in their capacities. 8. Emphasize safety, and offer as much structure as your client needs. Use routines, rhythms, and rituals to create a sense of order and familiarity. Give your clients freedom to create their own path, yet be ready with suggestions and illustrations when they request direction or structure. 9. Affirm and encourage experimentation and exploration. Invite clients to experiment with safe and socially responsible excursions into new patterns of experiencing—especially new patterns of action, interpretation, explanation, or meaningmaking. Affirm the processes and feelings involved; express genuine respect for the challenges of changing, and offer generous encouragement of your clients’ active engagement with those challenges.

and their brains for the pain they are experiencing. and pain ranks high as a motivator of human inquiry. Guidano. For their contributions to the development of the ideas here presented. please recall. the author is grateful to Albert Bandura. Some clients are well served by having a name for their problem and a sense of community in their struggles and suffering. Psychology is full of labels and theories that encourage naming and blaming. Problems tend to be associated with pain. When a diagnostic label results in a client’s identifying with the problem. but also one’s core sense of identity. and Walter B.269 Complexity and Human Change Processes 10. culturally transmitted inclinations to “name and blame. Likewise. or the like. 2004).com. and self-care. developmental change. which they viewed as less scientific and experimental. For further information. skills. and culpabilities rather than resources. and appreciated capacities.d. Thomas S. Campbell. 2003b. additional resources are included in the reference list without being specifically cited in the text. Donald T. Some clients are temporarily relieved to blame their parents. Naming people associated with the creation of pain is often involved in a blaming process. Naming things sometimes helps reduce their frightening or puzzling aspects. Friedrich A. Identification with one’s problem(s) results in an added challenge—to change. Teach self-care: set a good example. both camps were and are “radically empirical” in William James’s sense (Mahoney. Historically. Weimer. change may become more (rather than less) difficult. Kuhn. Notes These remarks are elaborated in Mahoney (1991. Activity. Encourage forgiveness. 1. when an explanation suggests that people are a certain problematic way because of their brain chemistry. the Web site of the Society for Constructivism in the Human Sciences. not only the problem(s). deficiencies. n. Help your clients develop compassion for themselves and others. their partners. clients may resign themselves to insight without action. mid-20th-century behavior therapists liked to distinguish themselves from the experiential humanistic-existential tradition. Vittorio F. It is also at the heart of our potential capacities for constructive. forgiveness.) . Teach compassion. However. We do it every day on both individual and collective levels. psychology and psychiatry have aided and abetted strong. Until recently. personality. or the journal Constructivism in the Human Sciences. 2. Most relevant here is the practical consequence that categorical names and no exit/no action explanations often disserve a client. one must change. 3. I have elsewhere elaborated my concerns about the categorical preoccupations of our profession. developmental history.” Emphases have been placed on pathologies. see www. To preserve textual flow. 2003a.constructivism123. is at the heart of our being alive. Hayek.

and the quest for coherence. and evolutionary epistemology (including scientific progress). (1997). The critical approach to science and philosophy: Essays in honor of K. Cambridge ma: Harvard University Press.). F. T. Hillsdale nj: Erlbaum. M. Ford. F. J.). Hayek. Chicago: University of Chicago Press. B. difference. 332–349). Goldberger. These are the three essential processes in biological evolution. (1976). Arciero. Guidano. literature. Constructions of disorder: Meaning-making frameworks for psychotherapy (pp. (1990). N. New York: Basic. The self in process. Societies of brains. G. Clinchy. Acts of meaning. Making stories: Law. References Anderson. Tarule..). A. & Guidano. T. W. Neimeyer & J. Law. Guidano. The truth about the truth: De-confusing and reconstructing the postmodern world. V. An invitation to social construction. Anderson. Chicago: University of Chicago Press. (1987). 4. Gergen. Hayek. (1964). (1996). Raskin (Eds. F. The theory of complex phenomena. explanation. W. behavioral development. J. (Ed. New York: Basic. & Belenky. Self-efficacy: The exercise of control. New York: Free Press. New York: Tarcher/Putnam. New York: Farrar Straus Giroux. (1998). Collins. Damasio. Gergen. The sociology of philosophies: A global theory of intellectual change. The saturated self. (1999). (2002). (1991). A. New York: Freeman. Bunge (Ed. Realities and relationships: Soundings in social construction. and liberty: Vol. London: Sage. T. In R.). San Francisco: Harper & Row. and power: Essays inspired by women’s ways of knowing. (1997). Knowledge. Cambridge ma: Harvard University Press. Popper (pp. Bruner. R. A. (1999). Anderson. New York: Guilford. (1994). Complexity of the self: A developmental approach to psychopathology and therapy. Freeman. 2. K. Reality isn’t what it used to be. The future of the self: Inventing the postmodern person. K. New York: Putnam. (1990). Washington dc: American Psychological Association. legislation. . Hayek. Cambridge ma: Harvard University Press. Humans as self-constructing living systems: A developmental perspective on behavior and personality. life.270 modeling complex systems because of their emphasis on praxis (action) that takes the learner to the embodied and emotional edges of his or her live experiencing. The feeling of what happens: Body and emotion in the making of consciousness. Bruner. D. W. The mirage of social justice. H.. V. In M. A. W. (1995). New York: Harcourt Brace. 91–118). Bandura. J. F. F.. Hillsdale nj: Erlbaum. F. The sensory order. Experience. J. (1995). (Eds. J. (1952).. D. (2000). New York: Guilford. K. A. A. J. (1991). R. Gergen. (1987). V.

Mahoney. imagination. J. Mahoney. (n. (Original work published 1976) Mahoney. Journal of Psychotherapy Integration. A. J. Mahoney. 659–665). In R. & Johnson. Constructive suggestions for the practical education of professional life counselors. 43–62). Behaviorism. In over our heads: The mental demands of modern life. In M. S. (Eds. Chicago: University of Chicago Press. (2000b). J. (1995). Kegan. G. A. (2003b). 337–352. Mahoney.). New York: Norton. Mahoney. New York: Basic. The handbook of humanistic psychology (pp. Human change processes. Power. Journal of Psychotherapy Integration. Mahoney. A. Constructions of disorder: Meaning-making frameworks for psychotherapy (pp. Mahoney. J. M. S. Lakoff. A. S. J. J. (2004). (1999). philosophy. and psychotherapy: A constructive caution on unification. T. Plainfield il: Kinder Path. J. . (2003a). economics. 15(3).271 Complexity and Human Change Processes Hayek. 1179–1184. Washington dc: American Psychological Association. & J. M. (2000a).). Clinton Corners ny: Eliot Werner. M. (2005a). (1987). Mahoney. The heart minding science: A personal history of psychology. Chicago: University of Chicago Press. A.. Dynamic patterns: The self-organization of brain and behavior. F. and constructivism: Reflections on persons and patterns in my intellectual development. R. Kelso. J. F.d. J. M. Constructive psychotherapy: A practical guide. G. 183–200). At home in the universe. Cambridge ma: Harvard University Press. M. Mahoney. Washington dc: American Psychological Association. and reason. (1978). 61(9). Scientist as subject: The psychological imperative. J. (in press). (2000). Kegan. & Mahoney. T. Kuhn. M. M.). M. How therapists change (pp. Schneider. Goldfried (Ed. Philosophy in the flesh: The embodied mind and its challenge to Western thought. Kauffman. M. Pierson. New studies in philosophy. Chicago: Chicago University Press. Mahoney. J. Thousand Oaks ca: Sage. Unpublished manuscript. Kauffman. M. Bugental. S. Suffering. and psychotherapy. Journal of Clinical Psychology. R. The psychology of personal constructs. The evolving self. (2001). J. politics.). Johnson. A. M. (1982). politics. In K.. D. Pilgrim in process: Collected poems. J. M. and the history of ideas. (2005b). Neimeyer & J. (1994). New York: Guilford. New York: Basic. S. Raskin (Eds. The body in the mind: The bodily basis of meaning. The road since structure. Living within essential tensions: Dialectics and future development. M. Kelly. Cambridge ma: mit Press. Cambridge ma: Harvard University Press. cognitivism. A constructive view of disorder and development. The origins of order: Self-organization and selection in evolution. Oxford: Oxford University Press. Oxford: Oxford University Press. M. J. (1993). F. R. (1995). (1955). (1991).

(2002). J. D. (2002). Segal. Segal. 105. A.. F. Petitot. Cambridge ma: mit Press.. J. D. Stolorow. and the sociology of knowledge.). (Eds. R. Stolorow. Hillsdale nj: Analytic. J. & Roy. N. (1994). G.. V. & Freeman. New York: Guilford. & Rosch. & Raskin. & Atwood. Z.. J. & Varela. Evolutionary epistemology. L. R. Thompson. Varela.272 modeling complex systems Maturana. B. A.. Evil: An alternative history of philosophy.. J.. Self and Identity. Bowling Green oh: Imprint Academic. S. Varela. M. Neff. Contexts of being: The intersubjective foundations of psychological life. B.. H.-M. LaSalle il: Open Court. (Eds. W. Stanford ca: Stanford University Press.). Varela. (1987). Thelen. E. & Mahoney. P. (Eds.. (1999). F.. & Orange. Affect regulation and the origin of the self. F. W.. (1995). & Shear. K. Between ourselves: Second-person issues in the study of consciousness. (1987).). Neimeyer. Van Geert. A dynamic systems approach to the development of cognition and action. Atwood.. & Bartley. J. W. Núñez. theory of rationality. & Teasdale. Self-compassion: An alternative conceptualization of a healthy attitude toward oneself. (2001). (2002). J. Mindfulness-based cognitive therapy for depression. M. Hillsdale nj: Erlbaum.). Washington dc: American Psychological Association. (1999). Schore. Radnitzky. intention and emotion. (2003). J. (1979). E. Naturalizing phenomenology: Issues in contemporary phenomenology and cognitive science. Constructions of disorder: Meaningmaking frameworks for psychotherapy. New York: Elsevier North Holland. New York: Basic. L. G. The tree of knowledge: The biological roots of human understanding. (1991). 2. Varela. Reclaiming cognition: The primacy of action. Constructivism in psychotherapy. Princeton nj: Princeton University Press. J. (1994). (1986). D. D. A dynamic systems model of basic developmental mechanisms: Piaget. G. P. Psychological Review. (2000). (1999). Current Directions in Psychological Science. E. . Boston: Shambhala. J. (Ed. Williams. Pachoud. G. Thompson. (Eds. R. & Smith. Tuscon az: Imprint Academic. The dynamics of general developmental mechanisms: From Piaget to Vygotsky to dynamic systems models. Principles of biological autonomy. The embodied mind: Cognitive science and human experience. D. (Eds. Charlottesville va: Imprint Academic. Neimeyer. Worlds of experience: Interweaving philosophical and clinical dimensions in psychoanalysis. E. M. J. 634–677. E. A. R. R. New York: Norton. E. F. R. The view from within: First-person approaches to the study of consciousness. (1992). F...). The dream of reality: Heinz von Foerster’s constructivism. Van Geert.. Neiman. (2000). Cambridge ma: mit Press.. (1998). 9. 64–68.. 85–102. J. Washington dc: American Psychological Association. J.). Vygotsky and beyond.

257–296). Weimer. 267–311). Hillsdale nj: Erlbaum. Weimer. In R. Notes on the methodology of scientific research. and knowing (pp. . (Ed. P.). Watzlawick. Hillsdale nj: Erlbaum. W. (Ed. Amsterdam: Benjamins. New York: Norton.). W. Shaw & J. M. (1984).). acting.). (1987). Investigating phenomenal consciousness: New methodologies and maps.273 Complexity and Human Change Processes Velmans. B. Perceiving. Centripetal forces in the universe (pp. Spontaneously ordered complex phenomena and the unity of the moral sciences. Weimer. Radnitzky (Ed. (1977). In G. A conceptual framework for cognitive psychology: Motor theories of mind. Bransford (Eds. (2000). The invented reality: Contributions to constructivism. New York: Paragon. (1979). W. B. B.

.

the Madonna model—meant to model the quite molar “complex process” of clinical rehabilitation—is presented. While the specific model presented here is not derived from other sources. a specific model developed in Nebraska. conduct. To illustrate the point. Hence. coordinated rehabilitation programs. are addressed in contributions in this volume. clinical rehabilitation research with an explicitly practical focus. it may also be seen as representing complementary areas that together have a great deal to offer with respect to the design and delivery of rehabilitation and to rehabilitation-related research. molar to molecular. Several levels of analysis. and analysis of contextually based. However. this. the final chapter of this volume is a brief overview of one application of modeling complex processes associated with the practice of rehabilitation. research. the collection may in some senses be seen as quite eclectic. from the perspective of a multifaceted system like that seen in the world of practical clinical rehabilitation for injured and/or ill individuals. it is to be acknowledged immediately that the individually designated aspects of the model together constitute a set of .Editors’ Postscript: Modeling Complex Processes in a Rehabilitation Application In keeping with the theme of balancing theory. and modeling approaches with potential practical applications. it is also a model to support design. Developed as a framework to systematically design and deliver comprehensive.

intended to emphasize the interdependence of the various facets in the design and delivery of individualized treatment programs for individuals requiring comprehensive rehabilitation. A rehabilitation research agenda can be expressed as a study of the quantitative and qualitative relations between elements of the model and treatment outcomes. Moreover. differs among diagnostic categories and. Most.276 modeling complex systems foci that would probably be endorsed as relevant to the design and operation of comprehensive rehabilitation programs by most experienced rehabilitation providers and recipients. The model (see Figure 1) is represented in a circumplex form. When these facets are addressed in isolation. facets of the rehabilitation model presented here are addressed in a typical rehabilitation treatment plan or program. Although using the model to guide treatment planning naturally introduces some degree of complexity—relative to a simple “critical path. and maximize efficiency. outcomes are much more likely to be suboptimal. The distinctive focus of such research is the . interdependent nature of these foci. interdisciplinary rehabilitation teams are most likely to work successfully with the range of stakeholders to create a plan that will optimize functional outcomes. The model can also serve as a coherent guide for research. Invoking and using the model at the outset of treatment is also a useful vehicle for beginning the process of educating patient and family about the nature of their circumstances and some dimensions of their prospects in the coming months or years. owing to differences in patient characteristics and life circumstances. the pattern of relations between the aspects of the model. indeed. along with brief excerpts drawn from the present collection illustrating the relevance of these areas of research and thought to research and practice in clinical rehabilitation. by carefully evaluating the relations that exist—or are necessary to establish—among these aspects in each unique clinical situation. maximally successful program. minimize unanticipated complications. and the relative importance of addressing each aspect. if not all.” for example—it is crucial to recognize and respond accordingly to the linked. In practice.1 The individual aspects of the model are discussed only in broad outline below. Examination of these relations and incorporation of them in the initial planning process is an important ingredient in designing and delivering a coherent. across patients within the same diagnostic group.

they are treated by multiple clinicians and receive many different modalities. many studies have been conducted examining the utility of a range of physical-therapy evaluation and treatment techniques for addressing patients’ mobility outcomes. minimize complications. As we learn more. but interacting. we will be able to furnish clinical treatment teams with the . It is the interaction of all these interventions. joint influence of these many influences. examination of the relative contributions on important functional outcomes of the many separately delivered. one needs to look at the combined. However.277 Editors’ Postscript Figure 1. However. and increase efficiency. that produces the final result. For example. to maximize functional outcomes. many patients receive comprehensive programs. comparatively little is known in specifics about how these “independent variables” interact to produce the outcomes of interest. Therefore. The Madonna model: an integrative focus guiding treatment and research. along with their own organismic vitality. interventions that a rehabilitation patient receives.

The chapters by Lajoie.278 modeling complex systems information they need to ensure inclusion of crucial ingredients in the treatment plan and omission of non-value-added components. and practical data representation to aid in decision making. treatment and research endeavors are reciprocal elements of one integrated process. ultimately. reflecting the mechanisms through which elements of the later components are applied in the context of an individual treated by the team. service delivery. Although empirical research and. and Goodwin can all be read profitably as suggesting models of treatment-competency development. Neufeld. developing high-performing teams. is the challenge currently facing rehabilitation workers. communication. and Salas. and team decision-making processes. in real time. Ideally. Musen. clinician accountability. thus. explicit guidelines or recommendations illuminating clinical-treatment-team structures and decision-making processes are underrepresented in the rehabilitation literature. Research in this domain focuses on issues like finding ways to optimize collaborative functioning. Finding ways to meaningfully include these stakeholders. members of the community are all stakeholders in the outcomes sought. Model Elements team characteristics The initial component of the model reflects the functional primacy of clinical-treatment-team structures and decision-making processes governing the organization and applications of the model components that follow sequentially. it is important to conceptualize the “team” in an inclusive way: in addition to patients/consumers and families. managing emotional issues at the team level effectively. payers and. all of which will be increasingly challenging—but crucially important—as rehabilitation moves from confined treatment environments into (complex) community venues. and developing and maintaining a team social climate that enhances individual team members’ psychological growth and levels of satisfaction and engagement. Stagl. . Burke. Central aspects of this initial component include attention to issues such as interdisciplinarytreatment-team skills and the team culture influencing issues such as planning.

socially-embedded. and social-network engagement. Identifying efficient and reliable ways to discover and tap these strengths should be a major research focus.279 Editors’ Postscript patient/family engagement This aspect of the model refers to application of processes and procedures to optimize patient and family involvement. without which the effects of the implementation of subsequent components will be suboptimal. motivation. From the psychological perspective. These ideas are explored more fully elsewhere (Mahoney. Fredrickson. Lopez & Snyder. 2003). Michael Mahoney’s chapter in this volume. with its emphasis on “constructivism”—explicitly recognizing “human beings as actively complex. 2002) are promising approaches to systematic and proactive development and application of strategies to optimize this aspect of the rehabilitation model. Instru- . this segment of the model includes assessments and treatments designed to minimize trauma associated with injury or illness and to promote positive adaptations to associated challenges. 1992. Identification of patient and family strengths and resources early in the rehabilitation process is crucial to a successful outcome. inviting maximum independence for patients and families. Seligman. and developmentally dynamic self-organizing systems” (as the tagline on the cover of Constructivism in the Human Sciences describes his journal’s mission)—offers many avenues to explore in the process of promoting and maintaining patient. Attention to this dimension includes management of a treatment setting’s psychosocial milieu to promote an environment communicating optimism and encouragement. Clifton & Nelson. Lajoie’s insights and findings as presented in her chapter in this volume would seem to be as applicable to “educating” patients and families as to training professional personnel. Themes in psychology currently associated with variants of “positive psychology” and “strengths-based” psychology (Buckingham & Clifton. 2001. therapeutics This element of the model refers to the various evaluation and treatment techniques available for use in rehabilitation programs. and engagement in the rehabilitation process. 2001. 2003. family.

all fall within the scope of this facet of the model. Clinicians and clinical teams identify promising new approaches or identify particular challenges. including determination of the relative contributions of the constituent interventions. and other domains has tremendous promise with respect to helping individuals with disabilities successfully meet the challenges confronting them in their homes and communities. The rapidly developing area of biomedical engineering reflects the increasing importance of technology in rehabilitation as well as in medicine generally. From a research perspective it represents a focus on improvement and application of specific forms of patient evaluation and treatment interventions. range and delivery of therapies. and/or cognitive impairments participate more effectively in the activities of daily life and life roles. research is designed and trials implemented to refine and validate evolving resources and procedures. Each paper suggests innovative approaches that can serve to enrich clinicians’ repertoire of tools to enhance rehabilitation practice. computer sciences.or the physical-medicine realm. medications. civil engineering. For example. In addition to individual therapeutic modalities. other research foci should include determination of: • • • • Optimal timing of treatments following an accident or illness Optimal intensity of treatments Optimal duration of treatments Optimal mix of the interventions provided to a patient. Progress in the fields of electrical engineering. computerized augmentative and alternative communication devices are available now to help individuals with speech. whether in the psychiatric. All the Symposium speakers provided valuable insights in this area of focus.280 modeling complex systems mentation and equipment. Laser technology is being employed to help individuals with significant motor impairments control their environments and . language. technology The technology component of the model places an emphasis on development and application of the rapid advances in engineering sciences and derivative technology. and so on. mechanical engineering. industrial engineering.

. home. and so on. enhancing productivity. Implementation of emerging technologies is an important emphasis in the work reported in the chapters by Salas et al. workplace). Increased research in this area will facilitate development of rehabilitation and engineering practices that are effective and cost-efficient and that ensure that patient reintegration into his or her living setting is optimized. “automatically” altering the environment in ways that best conform to an individual’s unique strengths and needs. thereby increasing prospects for optimal participation in important life activities. wireless communication technology. Lajoie. participation This element of the model emphasizes the signal importance of maintaining a rehabilitation focus on identification of effective and efficient ways of linking individuals and families to resources in . living setting The living-setting facet of the model reflects the importance placed on individualized. and maximizing positive quality of life. This would seem to be a very useful area to explore in a future Symposium.g.. laser-sensitive surfaces.281 Editors’ Postscript participate more fully in the important domains of their lives. To foster independence and quality of life for many individuals following disabling illnesses or injuries. via the application of electrical sensors. living settings need to be redesigned. school. These environments are designed to be responsive to differences in individuals’ capacities (e. discharge-destination-specific rehabilitation-program design.g. This aspect of the model is not represented with marked emphasis in this year’s Symposium volume but should be addressed in any comprehensive research or practice efforts. global positioning technology. computers.. physical. and Musen in particular. incorporating structural modifications and/or installation of assistive and alternative technologies. Socalled smart environments that recognize specific individuals are now being engineered and refined. reflecting the salience of contextual factors in human experience and functioning. and communicative) in a given environment (e. cognitive.

vocational. Program improvement. Perspectives and traditions resonant with this component of the model include ecological psychology (e. life roles. Program integrity (qa). again with the overarching aim of promoting a successful resumption of significant and productive life roles. Quality of collaborations. the work of Roger Barker and his colleagues on the concept of behavior settings and the implications of how such settings are “manned”) and community psychology (e.282 modeling complex systems their community. providing timely informational feedback to team members and administrators concerning the outcome of the activities of the program/system.. Research in this area. This dimension of the rehabilitation model is understood to include the individual and his or her family. Heller & Monahan. 1988). Information regarding improvements in patient functioning. also includes development.g. Sarason. and implementation of effective strategies to ensure that community resources are fully accessible as well as efforts targeting timely and adequate education of community members concerning strengths and needs of their disabled members. Cost efficiency. Rappaport. This aspect reflects the strong emphasis on “activities and participation” articulated in World Health Organization (2001). number and pattern of short. examination. Typical measures include • • • • • Patient outcomes. school.. though often appropriately focused on individual change and empowerment. 1977. . and relative efficiency of service delivery is developed and made available on a regular basis.and long-term complications associated with recovery processes. taken at the broadest of levels.g. program evaluation The program evaluation component of a rehabilitation model is crucial to maintenance of an evidence-based culture. social network. and community as well as the environment as a system. Operationalization of this facet of the model underscores the importance of rehabilitation-oriented input into community-planning activities. and recreational settings: that is. 1997. in the effort to optimize reintegration into home.

. such as computer-based decision trees.283 Editors’ Postscript Evolving team/program activities include the integration of new protocols as well as the successful implementation of relevant research findings. more complex procedures are rapidly being developed and incorporated. clinical decision support Clinical decision support in a rehabilitation setting is a practical exercise in modeling complex systems. the content of which is derived from program evaluation. and as time and resources available for rehabilitation steadily decrease.g. Simple examples of decisionsupport aids include basic tools such as diagrams and clinical paths. calculated conditional probabilities. This highly significant component of the model represents the mechanism for translating information derived from program evaluation into practical understanding and effective responding to the relations that exist between all the other elements of the model as they apply to a given individual. However. community context) in order to develop truly useful decision-support . and clinician decision making during rehabilitation. family. clinical-decision-support systems—linked to a guiding model of rehabilitation aspirations and activities—are increasingly important. should operate to enhance the efficiency and effectiveness of patient. Graphic representations of patient performance over time and statistical tools for prediction of patient outcome based on prognostic data are beginning to appear. It is also a mechanism for gaining an understanding of the interplay of factors influencing a rehabilitation setting at the level of a treatment program. and sensitivity analyses to determine the relative impact of changes and interactions between specific variables on the final outcome(s) produced by a combination of several components. The development and application of computer-based clinical-decision-support systems. temporal context. It is important that the design of such decision-support tools be informed by awareness of the importance of naturalistic contexts (e. hospital context. Regular dissemination of this information with other organizational teams as well as with colleagues at other rehabilitation settings is an indispensable mechanism for improving rehabilitation practices. As the amount of information and complexity of treatment technologies increases.

.Figure 2. The context of rehabilitation: illustrative aspects.

a bruised shoulder sustained in a softball game might require a few brief sessions of ultrasound therapy. often involving multiple organismic systems. At the other end of the spectrum. one can work to build and elaborate a successful research program aligned with the essential elements of an effective rehabilitation program. However. . Using this model as a guide. computer-based clinical decision support. one can see many areas of application for the thoughts presented in the chapters in this volume that can be brought to bear on various aspects of the rehabilitation process. but would not require the use of assistive technology. modification of the home environment. the flow chart presented here is an obvious oversimplification.285 Editors’ Postscript systems. but readers are urged to reflect on the potential application of these ideas to rehabilitation and psychological practice in general as they read and reflect on the wealth of stimulating ideas offered in this volume. Space does not allow description of these applications. model summary The model described here represents an approach to focused rehabilitation treatment and research that is probably compatible with a general rehabilitation philosophy that has been elaborated over the years in a number of settings and that is comprehensive and true to the multifaceted nature of the rehabilitation process. and Musen are all quite relevant to consult when contemplating this aspect of the decision-making process. For example. Salas et al. along with a home exercise program. Many of these dimensions are well addressed in the chapters in this volume. Indeed. or a community-reintegration program. there are a number of elements involved in the rehabilitation process. The notion of comprehensive rehabilitation is meant to apply to clinical situations in which afflictions and impairments are substantial.. rehabilitation following stroke frequently requires a comprehensive approach because multiple capacities are affected. As indicated in the schematic of Figure 2. The chapters by Neufeld. Note 1.

286 modeling complex systems References Buckingham. Constructive psychotherapy: A practical guide. 218–226. Community psychology: Values. New York: Dell. O. Washington dc: American Psychological Association. L. (Original work published 1972) Seligman. (1988). & Clifton. Geneva. B.. E. D. American Psychologist. K. Rappaport. Soar with your strengths. J. (2003). P. J. & Nelson. and action. Clifton. . New York: Free Press. Positive psychological assessment: A handbook of models and measures. (1992). 56. B. Mahoney. research.).. Sarason. discover your strengths. (2002). New York: Free Press. Psychology and community change.. (1977). New York: Guilford. Fredrickson. P. D. Brookline ma: Brookline. The creation of settings and the future societies. International classification of functioning. The role of positive emotions in positive psychology: The broaden-and-build theory of positive emotions. O. J. & Snyder. M. Rinehart & Winston. Homewood il: Dorsey. New York: Holt.. S. (2001). S. disability and health. Heller. & Monahan. (2001). (Eds. J. M. Lopez. (1997). R. Now. World Health Organization. Authentic happiness. (2003). C. (2001). M.

leadership. Burke earned her doctorate in industrial/organizational psychology from George Mason University and is a member of the American Psychological Association. team training. Dr.Contributors C. Gerald F. She is currently investigating team adaptability and its corresponding measurement. and the impact of stress on team process and performance. assigned to the Leader Development Research Unit (ldru). Shawn Burke is a research scientist at the University of Central Florida. Burke has presented at 66 peer-reviewed conferences. the Society for Industrial and Organizational Psychologists. and assisted organizations in evaluating aviation-related team-training programs and reducing medical errors. and the Academy of Management. and the training of such teams. Institute for Simulation and Training. Army Research Institute for Behavioral and Social Sciences.S. and team effectiveness. measurement. published 33 articles in scientific journals and books related to the above topics. Her expertise includes teams and their leadership. team adaptability. evaluation. Goodwin is a research psychologist at the U. She serves as an ad hoc reviewer for Human Factors and Quality and Safety in Healthcare. He received his ms and PhD in industrial/organizational psychology from Pennsylvania State Uni- . issues related to multicultural team performance. Dr.

using variants of time-series analysis and cluster analysis to identify characteristic patterns of schizophrenia patients’ response to treatment and rehabilitation and modeling stable and unstable states in the course of psychotherapy. He was previously employed at the American Institutes for Research. He is a research and clinical psychologist at the University Hospital of Psychiatry. Dr. training evaluation. in 1998. Goodwin’s current research focus is on leader and team effectiveness issues. Lajoie received her doctorate from Stanford University in 1986. She has numerous publications. In addition to being the James McGill Professor she is chair of the Department of Educational and Counselling Psychology at McGill University. He currently serves as an ad hoc reviewer for the Journal of Applied Psychology and Human Performance and has recently joined the editorial board of Human Factors. the American Psychological Association (apa). He is a member of the Society for Industrial and Organizational Psychology. Zeno Kupper received his PhD in clinical psychology from the University of Fribourg. where his project work included test development. including two volumes on computers as cognitive tools published by Erlbaum. Dr. for example. Switzerland. statistics. particularly with regard to the joint. and performance modeling. Lajoie has engaged in a wide array of innovative research and scholarly activities where she applies cognitive theories in the design of computer-based learning environments for classroom and real-world applications. Susanne P. and medicine. Switzerland. He also provides statistical and methodological support within ldru. interagency. He was trained in clientcentered and cognitive-behavioral psychotherapy and has extensive clinical experience in the rehabilitation of patients with severe mental illness. and multinational context. In his research he has been engaged in developing analytic techniques to identify and measure complex patterns in the course of mental disorders. This kind of analytic technology aims to translate the insights of dynamic systems theory into clinical research.288 modeling complex systems versity. employment-litigation support with regard to statistical analysis. She uses a cognitive approach to skill identification and applies her research to the design of computer-coached practice environments in the areas of science. and apa Division 19 (Military Psychology). . University of Bern.

where he is head of the Stanford Center for Biomedical Informatics Research. reusable ontologies and knowledge representations. As cognitive constructs became acceptable and then mainstream in behavior therapy. and biomedical decision support. the Semantic Web. At the time of his Nebraska Symposium presentation. Mahoney was a fellow of the American Association for the Advancement of Science. He received a . Professor Mahoney was a leading thinker in conceptualizing behavior change from a systemic perspective. Pilgrim in Process. Dr. Musen conducts research related to intelligent systems. Most recently. was published in 2003 by Kinder Path. The “motivation to change” plays a key role in these conceptualizations. but with the systemic view that motivation is a complex interactive process that includes both organismic and environmental components. the American Psychological Association. A collection of his recent poems. and the World Academy of Art and Science. Mark Musen is professor of medicine (biomedical informatics) and computer science (by courtesy) at Stanford University. he received the Young Investigator Award for Research in Medical Knowledge Systems from the American Association of Medical Systems and Informatics. Mahoney was a professor at the University of North Texas. A recipient of many honors. derived from chaos and self-organizing systems theories. Mahoney moved on to even more complex models of the psychotherapy enterprise. an endeavor to systematize one of the most complex and difficult-to-quantify decision processes in the human repertoire. Dr. His seminal 1974 Self-Control: Power to the Person (with Carl E. In 1989. Thoresen) heralded the contemporary age of cognitively oriented. social-learning-theory-based therapies. He was also distinguished consulting faculty at Saybrook Graduate School and Research Center. San Francisco. His analyses of the role of verbal representations in behavior acquisition and change were compelling demonstrations of the limitations of “cognition-free” behaviorism. Dr. Mahoney’s application of systemic models to psychotherapy was at the frontier of this area. Mahoney was a pioneer in the cognitive-behavior-therapy movement in the 1970s. He was also the executive director of the journal Constructivism in the Human Sciences. holding public debates with Joseph Wolpe and other classical behaviorists in the newly formed Association for the Advancement of Behavior Therapy.289 Contributors Michael J.

J. 1997) and the coeditor-in-chief of the journal Applied Ontology. where he also is a core faculty member of the Program in Neuroscience. and Psychological Assessment. Philosophy. he received the Donald A.290 modeling complex systems Young Investigator Award from the National Science Foundation in 1992. Neufeld is a professor in the Departments of Psychology and Psychiatry. Professor Neufeld has authored or edited 7 books and journal special sections. His 155 publications and 21 technical reports have appeared in journals ranging from the Journal of Mathematical Psychology. Richard W. being the first psychologist recipient. Musen sits on the editorial boards of several journals related to biomedical informatics and computer science. and Social Science at the Rhode Island School of Design and in the Department of Psychology and the Science and Society Program at Brown University. the Ontario Mental Health Foundation Senior Research Fellowship. 1994) and a coauthor (with William Spaulding and Mary Sullivan) of Treatment and Rehabilitation of Severe Mental Illness (Guilford. Jeffrey Poland received an ma in clinical psychology from the Southern Connecticut State University in 1982 and a PhD in the philosophy of science from the Massachusetts Institute of Technology in 1983. He has held academic positions at Colgate University and the University of Nebraska–Lincoln. Professor Neufeld is a past associate editor of the Canadian Journal of Behavioral Science and of Psychological Assessment. He is the coeditor of the Handbook of Medical Informatics (Springer. the British Journal of Mathematical and Statistical Psychology. Dr. and he currently teaches in the Department of History. and the University of Western Ontario Faculty of Social Science Research Professorship. He is the author of Physicalism: The Philosophical Foundations (Oxford University Press. In 2006. the Journal of Consulting and Clinical Psychology. 2003). He has received the Joey and Toby Tannenbaum Schizophrenia-Research Distinguished Scientist Award. B. He also holds an appointment as program . Lindberg Award for Innovation in Informatics from the American Medical Informatics Association. Eduardo Salas is Trustee Chair and professor of psychology at the University of Central Florida. and the Psychological Review to the Journal of Abnormal Psychology. University of Western Ontario.

with an emphasis on identifying effective applications of information technology to improve medical rehabilitation evaluation and treatment services for children and adults with disabilities. Salas has coauthored over 300 journal articles and book chapters and has co edited 15 books. Dr. His research interests include modeling team decision making in rehabilitation. He was a postdoctoral fellow in mental health research and teaching.291 Contributors director for Human Systems Integration Research Department at the Institute for Simulation and Training. Dr. . learning methodologies. with Mary Sullivan and Jeffrey Poland. Cromwell. clinical decision making. he conducts research on the nature of mental illness. law enforcement. of Treatment and Rehabilitation of Severe Mental Illness (Guilford. and performance assessment. decision-making under stress. He is currently designing tools and techniques to minimize human error in aviation. During this period. where he is currently a professor. team training. He has coedited two previous volumes of the Nebraska Symposium on Motivation and is the author. under the mentorship of Rue L.and recovery-oriented services for people with severe and disabling mental illness. Nebraska. In addition to training clinical psychologists. in the Department of Psychiatry of the University of Rochester. He received his PhD in clinical psychology from the University of Nebraska in 1985. Bill Shuart is the director of the Institute for Rehabilitation Science and Engineering at Madonna Rehabilitation Hospital in Lincoln. Will Spaulding received his PhD in clinical psychology from the University of Arizona in 1976. 2003). and the outcome of treatment and rehabilitation. After his postdoctoral work. He also practices as a clinical psychologist in rehabilitation. and medical environments. Salas served as a principal investigator for numerous researchand-development programs focusing on teamwork. where he retains a clinical affiliation with the Department of Psychology. advanced training technology. he was a senior research psychologist and head of the Training Technology Development Branch of navair-Orlando for 15 years. He currently edits the annual series Advances in Human Performance and Cognitive Engineering Research for Elsevier. Previously. he joined the faculty of the University of Nebraska–Lincoln’s Clinical Psychology Training Program.

Organizational Frontiers Series. Dr. He also received psychotherapy training in systemic therapy at the Institute of Family Therapy. where he was awarded a professorship in 2002. Switzerland. where he received his PhD in 1990. and International Review of Industrial and Organizational Psychology. he received the Venia legendi in 1996 at the University of Bern. Munich. After his habilitation in psychology. leadership. The lessons learned and practices distilled from this effort have appeared in scholarly outlets such as the Journal of Applied Psychology. Wolfgang Tschacher studied psychology at the University of Tübingen. . development. He works at the University Hospital of Psychiatry and is currently head of the Department of Psychotherapy. His main interests are in empirical psychotherapy research and experimental psychopathology. Leadership Quarterly. and adaptation. Stagl is an organizational consultant with Assessment Technologies Group and formerly served as a research scientist and research assistant at the University of Central Florida’s Institute for Simulation and Training. Stagl’s research investigates team performance. with an emphasis on dynamic systems approaches and phenomena of cognitive self-organization.292 modeling complex systems Kevin C.

259–260 attitudinal competencies. problem-solving. 255. 246. 257–258. 86.: The Wizard of Oz. 69–72 base rates. 153–154. technology artificial neural networks. Albert. 213 Adler. 248 balance skills. numerical. 3 analytic derivations. 250 algorithms. 212–214. 156 attention skills. Alfred. See also computer-based learning environments (cbles). 259–260 balance theory. Musen’s study of. Mark A. 128–129. L. 12 apprenticeship. 171 Amazon. 145–147. and performance models. 126. 49–53. 130. 68 Bateson. Francis. 124–125 Bayesian theory: on bold conjecture and falsifiability. 175 associative stage of learning. overview of. 58. 250 base distributions. Simon’s research on. xii– xiii. 191 attractors. 165–166 algorithms. 8 anova. 2. 269n2 adaptation.com. active agency. current . 250 activity. clinical. team. xx–xxii Bandura. 127 Bacon. 73n2. 147. 249 artificial intelligence (ai): A.Subject Index Page numbers in italics refer to illustrations. 10. 164 American Association for Artificial Intelligence. A. 95. 61. 151–153. F. in field of cognitive science. 253 Baum. 8. 136–138 Aristotle. 176 algorithms. 39–42. 94–99. 163. 8. Newell and H. boom and winter of. 156 avionics trouble-shooting. computer software. 153 American Psychological Association. 115 autonomous stage of learning. 50–51. history of. Gregory.

in diagnostic reasoning. dynamic systems theory. Roger Barker on. 8 bootstrap tests. posterior distributions. formal theory application of. and estimations of base rates. 32–38. 88 clinical assessment. and team performance. fingerprints of. 10–11. 157–160. Bruce. systems theories chess. 247. and parameter distribution. frameworks of. 250 Buchanan. 248. and order. 110–111. 108–109. 140. 137 blaming. theorem. 49. and motivation theories. 158 Case Builder (cble). 251. 17–21. 19–22 behavior: and intelligent system construction. 124–125. 85. 27. 227–228 cognitive-apprenticeship model. and mathematical theory. 247 centers. 140 cognition. xi. 54. 253. 217. 140. 19–22.) developments of. 55. 10–11 circular apparent motion (cam). 55–56. and neurological mechanisms. 257 capacity utilization. 86 Bruner. 270n4 behavioral field. and process latency models. 256–257. and rule-based systems. human. 158 choice. 117– 118. 132–136. 136. and psychopathology. and statistical property issue. 219–220. 174–176. xxxviii–xlii. and parameter estimation. 258–260 change processes. 137 cell membrane. ix biology. xii. 89–90. monitoring changes in. 25. 248. and stress susceptibility. See also apprenticeship cognitive-dysmetria hypothesis. 43 bodies. 283–285 clinical paths. 105. 14–17. and human lifespan. as learning tools. model frameworks of. xxxi–xxxiv. 245–249. 61. xxii–xxiii. 209. 138 categorization. 283 closure. 130–131. 109– 115. 43. 247. Don. 101– 102. 17–21. 114–115 celiacs. Jerome. and performance models. See also constructivism chaos. and orientations. 92 cognitive orientation study. 270n4 BioWorld (cble). xviii coaching. 53–55. 139. 130–132. 249. xix–xx behavior settings. 192. 133. 249. 257–260 bold conjecture. schema of. 31–39. shared. 264–267. modeling techniques. 9. 134. 282 behavior theory. 61. 102 cognitive processes: and clinical assessment. 212 biological utility. xxv–xxvi Big Five model. 89 chaos theory: in constructivism. 64–66. 108 boundaries. . 257–258 brain research. 249. 18 clinical decision support. 67–68. xxii. 151–153. 269n2 blood-oxygen-level-dependent (bold) response. and nonlinearity. item. 40. order. 136. 17. 18. 135. 130. 212 behavioral development. 111 circular causality. xiii. xxxv–xxxvi. 249. 253–255.294 modeling complex systems Bayesian theory (cont. 69–72 Carnegie Mellon University. Gestalt perception of. in schizophrenia. 24 causality. 147 Buddha. 252. 250 Campbell. 211–214. See also complexity studies.

248 connectivity. 53–55. See also BioWorld (cble) computers. psychological. 57–59. 247. xiv–xlii. 55–56. 279. frameworks of. xiii. and treatment-program efficacy. xxxi–xxxiv communities-of-learning (col) model. 191. 209–210. 173. 214. 171 constructivism.295 Subject Index 45–46. 11–12 cognitive science: description of. themes of. engineers and. Bayesian extension of. 248–249 decisional control. stochastic modeling in. for problem solving. 210. technology computer software: as engineering tools. 27–30. 2–3. for mixing-distribution parameters. 176 computer-based learning environments (cbles). 140. xxii– xxviii complexity studies: approaches to. kads project. 47 construct validity: assessment of. 87–89. and capacity utilization. 131–132. 125–128 Dartmouth College conference. 224 complexity. 61–66. 138. team. 66–67 cognitive stage of learning. 155–156 cognitive task analysis (cta). 248 Copernicus. 248 coping. Charles. computer-based learning environments (cbles). 136–140. 218 communicative acts. See artificial intelligence (ai). and change. and intelligent system construction. 245. 227 conditioned responses. 86. 46–49. 252–253. 212. 90–92 constraint satisfaction. 59–60 control parameters. Protégé (computer system) conceptual models. 17. 57–59 coordination. 92–93. 102 cryptarithmetic problems. and model parameters. 161 constructive methods. xxvii–xxviii congruence. formal theory of. xxxiii connected knowing. 247 complexity. 262 crisis intervention study. 67. description of. 45–46 decision-and-choice models. paradox in human. 127–129. 283. 140 community. 98 convergence. 176. 86 computer-assisted software-engineering (case) tools. 209–211. 130–132. xii. 158–163. 264 coupled oscillations. 248–249. 19. 45–46 core ordering processes (cops). and rehabilitation decision making. xiii–xiv. 267–269. 151. and time variance. history of. in statistical summaries. 250. 192. mycin (computer program). 140 communities of practice (cop). 101–102. 60–61. 126–128 Columbia (space shuttle). 249–253 construct representation. 171. See social situations competencies. ix–x. xii. 201–202 creativity. See also BioWorld (cble). 146 Darwin. 185. cellular. See also chaos theory computationalist theory. 151 curriculum modeling. 186 communication. 10–11 decision making: through cognitive . and team performance. Michael Mahoney and. 221. 249. 160–162 conditional probabilities. definition of. description of. 283 conditional-reasoning measurement technique. computer software. recommendations for practice of.

and definition of clients. in psychopathology. 117. 250. 25. 258. 280–281 entrainment. 44. xxxvi emotions: Charles Darwin on. See also order distribution moments. 60. 86–87. . 117 dual-coding theory. 227 Digital Equipment Corporation. 257–258 Dictionary of Occupational Titles. 36. lifelong development of. 257. 283 deductive systems. and diagnosis. 69–73 Djerassi. and models in clinical cognitive science. 46. xx. David Krech on. xii. 7. 19. 103. in rehabilitation. 30–31. 64–66. 10 encoding process: and cognitive debilities. 165. 9. stochastic models of.) task analysis. 147–150. 161–162 diagnosis: computer-based learning tools for. 181. 270n3 Educational Researcher. 117. and individualized parameter estimation. 108. xxix. 158–159 dendral (computer program). 175–178. 10–12. 90. and estimations of base rates. 52. psychological. 180. nonlinear. rules of. 147 domain knowledge: in intelligent system construction. and self-organization. Kurt Lewin’s model of. 195. 96–99. 29. and order seeking. xii–xiii. and parametervalue mixing distributions. 218 empirical equivalence. 49. 153 disorder. decision making in. 138. 169. and convergence. in model frameworks. 26. limitations of. in Madonna model. 32. 108. and statistical property issue. 105. 168–174. 57. and expertise modeling. 249. 283 dialectical contrast. 7 elimination-by-aspects model. frameworks of. on team-effectiveness models and frameworks. in psychopathology. and modeling effects of stress. as motivational factors. in mycin. 150. 28. and problem solving. 132–136. 116–117. See lifelong development dynamic systems theory: and chaos theory. patterning in. 126. 135. 223 decision trees. 58. and establish-refine method. 260–261. 137. and stochastic model distributions. 148 dsm-style diagnostics. See also chaos theory ecology. 102–104. xxxv–xlii Einstein’s limiting constant. 11 emergent social aspects. 9. 283–285. 40–42. xxviii– xxx. medicine diagrams. and psychopathology. 33. and performance-latency model. 278. 56. 147 design models. 130. 3–7. See also medical students. 23 dynamic dialectical development. 181. 125 E-E (environment-environment) unit. xiii. 115–118. 48–49. of infectious diseases. 282 edging. Carl. 251 empirical data: and construct validity. and constructivism. 201–202 environment: in apprenticeship models.296 modeling complex systems decision making (cont. 10 Deep Blue. xxxiv–xlii. 93–115. 176–178. and team effectiveness. 48. 249. 43. 10. 252. 34–39 engineering. and posterior problem of group membership. 126–128. classification of models of. 158–159 Deep Thought.

250 Galileo. 270n4 excitement. 26. 253 flow. 8 feature abstraction. Viktor. 283 groups: in apprenticeship models. demarcations of. biological. 13–14 equilibrium. 270n4 evolutionary epistemology. 260–261 falsifiability. 256 fmri. 214 Feigenbaum. 91. 169–170. 150 good figure concept. 195 expert systems. 16–17. 93 exogenous variance. 3–6. 256–257. 58. xviii–xxi goodness-of-fit tests. 19–22. 116. base rates and distributions. 96. 10–13 experiencing. 52–53. and model frameworks. 17. 70–72. xx. 139–140. 34–39. 30. 64–66 Gram-negative rod. 88–89. 8–10. parameters of. 111. 140. 203. 89 global market. 256 field (concept). 219–220 equations. 27. 53. 39–42. 13–14. likelihood of schizophrenic symptoms in. xviii–xxii. xli–xlii. 157–158. and strength of predictions. 185. xxxv. 96. xiii. 138. 92. 57–60. 99–104 exergy. paradox of. and capacity utilization. and cognitive processes. on evolution of schizophrenia. 24–27. 147 Ferguson. 189 goal rule. and psy- . 280–281. models of. 169–171 events. 98. 124.297 Subject Index 131–132. Adam. 52. See intelligent systems explorations. historic role of. and stress susceptibility. 210. 248 gamma distributions. and problem solving. 217. 93. 130–131. in statistical summaries of performance. 206. 202. assets of. written word sequence in. 50–51. and team-effectiveness models and frameworks. Ed. 126. 109–115. xv. and self-organization. 138–140. 45– 46. Fritz Heider on. 28. 55–56. 170 Granger-causal models. xxv evolution. 133–136. See also time-series method graphic representations. diagnostic. 209. and hysteresis effect. 169 feedback: in behavioral field. 131–132. Roger Barker on. 59–60 erp. cognitive functioning in schizophrenic. relevant dimensions of. xxxiv first-order change. xvii good form. vase-face in. 6–8. xxxiv–xlii. and team effectiveness. xiv Gestalt psychology: and connectivity. xix–xxi. in Bayesian extension. 137. 12. 39–42 formal theories: aesthetic appeal of. Gladstein model of effectiveness of. description of. 25–27. in clinical cognitive science. 156. 269n3 expertise: in intelligent system construction. and process-latency model. and rehabilitation. 14–17. 260–261. xl–xli Erlang distribution: aesthetic appeal of. 152–157. 88. barriers to. in learning process. 13–14. 20 establish-refine method. 58. 189. xxii–xxviii. 2–3. 73n3 Frankl. 203. 90. 20. 30. xvii–xviii. 87. xii. 89–90. xi. psychological. team. 69–73 “general systems” framework. 58.

teamwork The Growth of the Mind (Koffka). 62–65. Event Structure. 162. 179–180. Johann. 177–180 intersensory binding. 146–147. 168–171. 19–22. and mycin system. and problem solving methods. 129. 256. 24 item-response theory (irt). David. 30–34. 33. inference patterns of. 152–157. 166. 158–163. and team performance measurement. Protégé (computer system) interaction. 103–104 heuristic classification. William. 92 individuals: in behavior settings. and the Psychological Environment. 269n3 kads project. 113–115. 170. 206. xiii Intel. engineering of. 8. 169–172. cognitive processes of developers of. 39–42. 136–138 Hume. 111–112 human models. 173. team effectiveness. 185 Intellicorp. 215–222. 169. 64. 159–162 Kant. 179. 111–112. stages of development of. and team effectiveness. 217–218. 153 International Joint Conference on Artificial Intelligence (1985). 32. 160. teams. 160–162. 256 Hebb’s law. 158 . xxxi International Association of Knowledge Engineers. 250 heroin. Fritz: On Perception. psychological applications of. 169. 91 Heider. 188. 174–176. Friedrich. team performance. cognitive functioning in schizophrenic. perception and behavior of. 214–215. xviii Grundy (computer system). See also communities-of-learning (col) model. 169–172. 155–159. 213. 133–136 hysteresis. 164. 59–60. xiii–xlii inattentional blindness experiments. 154 intelligent systems: advances of using Internet. uses of. statistical summaries of performance of. 193. 174–178. 158 ideas. 162–163. 205.) chometric measures. 147–150. and team processes framework. and team effectiveness. 148–150. 203–204. 195 horizontal apparent motion (hsam). 257 hypotheses test. 74n8 James. 165–167. 217 hierarchies. 250 Kasparov. 250 Herbart. and software programs. xxxv–xxxvi. 200. xxxi– xxxiv. Gary. Immanuel. 173 inference engines. See also artificial intelligence (ai). 90. 210–211. 181. history of construction of. 175–178. 153 informal theories. and ontologies. xix. 153–154 Internet. 148. 221 infection: diagnosis of. 165.298 modeling complex systems groups (cont. 250. communicative behavior between. 180–181. 206. 224. 114 investigator role. 16–17 isomorphism. and expertise transfer. 172–174. rule-based. 114 ibm. 25. xl–xli. 17. 2–3. 146–147. 73n4 input-process-output (ipo) models: and team adaptation framework. 250. definition of. xxiii–xxv item encoding. 170 Hayek. xv Heraclitus. history of.

xi. and attitudes (ksas). xxvii–xxviii learning theories. 46. Kurt: The Growth of the Mind. 16. 135–139. 219–221 learning: model construction for. 139 knowledge industry. 176 knowledge-based systems. 31. 137 medicine. 153 living settings. 46. tacit. skills. 155–156. 172. 27–30. and intelligent system development. Wolfgang: Physical Gestalten. 275–285. skills. 167. xx. 105–107 language production. 224 knowledge. 165–167. 157. 281 L-mode. and stress. 181 mathematica (computer-algebra program). 152–153. 277. as human aspect. 32–34. 148. attitudes. 5 local independence. xvii Landsberg order. 227 limit cycle attractors. interdependence of facets of. 124 method ontology. 131–132 Lederberg. 16 mathematical theory. 147–150. 153–154 Madonna model. xii–xiii. parameters of. Michael. 5 McCarthy. 125–129. 19. 250. 154–159. in production systems. xviii Kohler. See also domain knowledge knowledge. design of. 139 Mandeville. 137. 30. 176–177. 73n5. 10 Los Angeles ca. 256–257 Lao-Tzu. 7–8 medical students. 5 management tools. 69–73. See also processlatency model latency variance. 22–24 memory-trace theory. and performance measurement. Joshua. 103–105. 155–156 learning curves. 250 latency distributions: and Bayes’s theorem. 49. 5 Maxslope. 24 metacognition. 252–253 life space. 132–140. in individualized parameter estimation. xxxvii–xxxviii. Roger Barker on psychological principles of. 153–154 Koffka. John. 261 mixture models: base distributions in.299 Subject Index Kelly. 276 Mahoney. 180. 140. 146–147. 165 memory search paradigms. 180. in cognitive process measurement models. 221. and other skills (ksaos). 50–51. Bernard. 211. 162– 171. 249. . stages of. 145. 156. elements of. 87. 24. 171–172 mind-body dualism. 16 mappings. 41 Maxcov. 52. and real-world situations. George. 34. 21–22. 217. 209–212. 191. 190–191 knowledge-acquisition system. xxxiv Likert scales. 146 mean latency models. 95 lisp programming language. 37. 19–21. 20. 250 knowledge: competencies. 11–12 leadership. 256 maple (computer-algebra program). 190–191. 250 mirror time. 49. 42. latency distributions in. 147 lifelong development. 31–34. 178. 94 matlab Optimization Toolbox. xix. 278–285. See intelligent systems knowledge-building tools. 245 Mambac. 39–42. in parametervalue mixing distributions. 5 Maxeig. 57–59.

fingerprints of. in self-organization. emotional and perceptual factors of. 169–170. influence of. 153. xxiv–xxviii order: and centering skills. 158 Newton’s universal law of gravity. human.. 249. 165. 108–109. 17. and user interfaces. 195. 92–93 motivational processes. and the Psychological Environment (Heider). 16 nonlinearity. 86 mri. 269n2 naturalistic settings. as rule-based system. 90–92. Edward L. 160–161. 163–168. and cognitive processes. 186 off-balance clients. history of. 105–108. xxxviii optimal complexity. xxxix. xxiv Neural Process Complexity. 43–44 multidimensional scaling. 260–261 Occidental’s Piper Alpha platform. 264–267. Event Structure. 160–162. for team effectiveness. characterization of. and rulebased systems. 123–124. xii. 94. 168–169 O*Net. design model of. 251. and Internet. 172–174. simple and complex. excessive. subjects for. social context of. 43. 151–153. 125–128. 163–167. 246–248. 73n5 neuropsychological symptom patterns. 161–162. in . 169. 171–174. definition of. in lifelong development. xxviii– xxx. and chaos theory. 113 motivation: and behavior settings. xiv–xv molecular learning theory. 103–104 opportunities. 173. 227 On Perception. in intelligent system construction. 128–132. domain knowledge in. and E-E units. 126–128 nurses’ observation scale (noise). xiv–xlii neural nets. and ontologies. 97 obsessive-compulsive disorder. 155–157. xxiii– xxv neurological mechanisms. 176. diagnostic reasoning in. xii. 200 opioids. 152. 179–180. 283–285 Nebraska Symposium. 147–150. 92. 132–136. 127. 259–260. 165–166. 252. 24 mycin (computer program): conceptual model of. statistical property issue in. and problem solving. protocols of. xv ontologies. and performance. 149 naming. 258. 128–129. 250. 247 nurses. 105–116. 6–7. 166–167. 100–101 neuroscience. A. 138–140. problem solving in. 199–202 molar learning theory. 186–188. 258 Ohio State University. 36–38. in nonlinear systems. 124–125. xiii. 124– 125. in human change processes. xi.300 modeling complex systems mixture models (cont. 139–140. 167 open systems theory. 260–261. 174– 175. xxxvi. xxii. 115–116 nonsummativity. 151–153. 91 nucleation. 12. Walker’s theory of.) and posterior problem of group membership. 54–55 modeling: computer v. 20–21. curriculum of. 112–113. xiv–xv morphine. 16. 35–39. 117–118. 249 Newell. 97. 180. 64–66 mle(v). 103–104 motion-induced blindness (mib). 136–138.

mixing-distribution. A. 10–11 problems. 86. 73n1. xxxi– xxxiv pharmacology. 61–66. 46–61. 55–56. in dynamic systems theory. 19. See also self personal revolution. models of cognitive. 43–44. 219 parameters: construct validity of distribution. 255–258. 246–247 plug-ins. 44. 177 points attractors. 43–45. Gestalt perception of. 17. 25. definition of. 94–99. 106–108. 17–21. 254 problem solving: A. 57–59. model selection and function. 31. 110–114 posterior latency distributions. trajectories and attractors in. 281–282 Pascal distribution. 70. xxxi–xxxiv. of schizophrenics. systematic interrelations in. 123–124. 61. 279. 85 phase transitions. xvii–xviii. 107 Physical Gestalten (Kohler). xxxv–xxxvi. 28–29. 66 posterior probability. in mixing distributions. 86 Piaget. 52. 34–36. 30–31. 25. 57–58. 59–60. 27– 30. 21–22. 250 Plato. estimation of in schizophrenia. 126.301 Subject Index schizophrenia treatment and rehabilitation. 107. 39–42. 251. See also team performance performance-latency distributions. statistical summaries of prototypical. 110. 30. 95 Poisson distributions. 47–49. and selforganization. 27. xxii orientations. 70–73 Popper. 90. estimation of in diagnostic-group level analysis. estimation of in individuals. 212. 89–90. 47–49. 104 pattern levels. Jean. 88. Newell and H. sample of in Bayes’s theorem. xvi–xviii. 42. 255. See also chaos theory order. Simon’s research on. 105. 50. xvii physical-symbol-systems premise. xxviii–xxx. for selected base distributions. in stochastic models. 107 order parameters. and fmri technology. 257 Positive and Negative Symptom Scale (panss). 253–254 person model. of process-latency models. and nonlinearity. 91. and artificial neural . 218. 58–59 patterning. 109–115 performance: construct validity paradigms affecting. 128–132. 90. parameters of in process-latency models. 255 perception. 256–257. xx person-object relationships. xxxv. 92 organization of behavior. 88–90. and expertise studies. 89 order effect. teleological and teleonomic. 3–9 primacy of the abstract. 107. xxi predictions. Karl. 19 performance-model parameters. 104. See also control parameters participation. transitions in. 105–115. 74n7. 57. 99–104. seeking of in problem solving. xxxii p-o-x system. 256 probabilities: cumulative and multiple distributions of. 151– 153. 17 personality. 34–42 Postman-Bruner hypotheses. ix. xv–xvi. 209–210. 139–140. and parametervalue mixing distributions.

scope of field of. 92. 136– 139. 52 Pythagoras. 282. 116–117. 94–94. 96. and team performance. 114. 95 psychiatry. 255. medical tutorials of. and ontologies. 58. 62–65. and domain ontologies. and motivation. 61. 170 Protégé (computer system). Paul Meehl on. 163. and posterior problem of group membership. and model selection. 35–39. 94–99 psychopharmacology. xii. 85. ecological and community. 103–104 psychosocial crises. 257 production rules. 165. 165. 94–99 psychopathology: connectivity in. and cognitive task analysis. 162. 99–104. 90. and kads project. 50–51. 155 production systems. and problem solving. 25–27. 2. 166–167. and dynamic systems theory. 105–108. description of field of. 173. 249. in computerbased learning environments. 150. diagnosis and treatment efficacy in. 100–101. 205 punishments. 53–54. 134. 192. and statistical summaries of performance. methods of in infectious organism-identification. inconsistencies in research of. 167. 105–115. nonlinear patterns in. 99–104 psychoticity. 146–147. 159. xxxiv–xlii. 155–159. 117. 85–86. 181 protocols: analysis of in rule-based systems. 168– 174. language production in. 180–181. 249. trajectories and attractors. 180. xv psychology: classification in. 246–247 . 269n2 Psychological Issues. base distributions in. 1. 69–73. and environment. therapy. xxxii pure death system with a linear death rate. and dynamic systems theory. 53–55. 93–115. empirical research in. 46. 256–257. 258–262. 254–255 process-completion potential. and order in pattern formation. 175. 92–93. and Gestalt psychology. 254–255. diagnosis of.) networks. 74n8. and molar and molecular learning theories. 101 psychotherapy: and applications of dialectical development strategies. parameters of. and intelligent system development. 140. See also construct validity psychopathological impairment. “positive. 155–157. 56. 165–166. 157–158. Bayesian extension of. 58–59. and psychotherapy. “name and blame. and intelligent system construction.” 269n2. and BioWorld. 176–178. 174–178. 133–136.” 279. components of. 126–127. 47–49. symptom patterns of. 49–53. systematic interrelations in. 176–178. 171.302 modeling complex systems problem solving (cont. 115–116. 124 Prospector (computer system). 151. 116–117 psychotic episodes. 90. 205. 176 prototypical dynamics. 57 process-latency model. in construct validity. 180–181 proficiency models. xiv–xv. 96. 99–104 punctuated-equilibrium model (pem). 152–153. 129. 219–220 psychometric validity. and time-series methods. 137. 36–38. 90. 153. 51. xi. 59–60 process level. 179–181. and complexity studies.

134–135 self-organization: biological. and individualized parameter estimation. 98. cognitive patterns in. symptom patterns of. 92. 250 rehabilitation: and complex-process modeling. 94–115. and living settings. 109–115. 261. 279. xv–xvi. pattern formation by. task affordances in. 22–24. 106–108. 106–107. 278. 98. 170 rule-based systems: as cognitive models. motivation in. 260n4 rewards. 256 . and posterior problem of group membership. rise of. 92–93. and time-series methods. comprehensive. 86. 157–158 Smith. xxv response parameters. 173. 151–153. likelihood estimation of symptomgroup base rates of. difficulties associated with. and environment. 66–67. connectivity in. efficacy assessment of. 251 self-monitoring. description of. 253–254 selection. 147–150. 127–128 Simon. trajectories and attractors in. 158 simplicity. and stochastic modeling. 94–99 Schizophrenia Process Study. phase transitions in. 282–283. 116–117 repeated activation. 257–258 second-order change. 34–39. stimulus-encoding dysfunction in. and Madonna model team characteristics. and therapeutics. 275–285. 3. 88–90.. relationship to. 153 Schopenhauer. 110–115. 131. 21. 87–89. 284. 18. 128. 107. 283–285. 262 science instruction. 39–42. 285n1. illustrative aspects of. xxii. 98. and parameter-value mixing distributions. 254 reciprocal determinism. in dynamic systems theory. Elaine. measurements of progress in. xii. 181 sensitivity analyses. 251. 32. 279. and order. 17–46. 281–282. xli–xlii. successful applications of. 99–104. 92–93. 153–154. and memory-search paradigms. 74n6 responsibilities. 174. Ted. connectivity in. 247 Sherlock (cble). 99–104. 107. 127 Shortliffe. 249. 9 quantum shift. Ivan. 100–101 Schlumberger.303 Subject Index qualitative research methodology. order in treatment of. 61–62. 6. practical model for. 283 sexual reproduction. 281. patient and family engagement in. Adam. xvii–xviii. nonlinear. A. parameters of. 280–281. 108–109. 250 science. 180. control parameters of. 104 Semantic Web. 2–3. decision making in. 260n4 self: behavior regulation of. 87. 279–280. 257. 97–98. 147 sicun (computer tutor). 87. 281–282. and technology. of schizophrenics. xxxviii retention. 180 schizophrenia: and Bayes’s theorem. 33. 132–136 Sechenov. xxxv. 151–153. xxxii Rich. 280–281. 31. and motivation. participation in. program evaluation of. 257. Arthur. 90–92. 27–29. 179. systematic interrelations in. stability of. H. 124. xxxv. 250. nonlinearity of. 105–115. xi. definition of. 92–93. 283–285. xxi situated action.

252 social situations. and rhythm of task accomplishment model. and evolution of mental diseases. 74n6 task affordances. mental models of. xx sri International. 19. 204. 7. 215. xiv. 129. 60. 211. 248. 2–3. 4. 110. 49. 45–46 stroboscopic apparent motion (sam). 111–112. 252 synergetics. 195. and problem solving methods. See encoding process stochastic models: characteristics of. measuring of. 87. and ipo models. 98 task performance: in kads project. 170 spatialized psychology. 209. 212. and stochastic modeling. predictive models of. 248. integrative framework of. 7. Roger Barker on. 154 symbols. 48. 283 statistics: in formal theory. 19. psychological: and empirical performance configurations. modeling of. and model focus. normative model of. 44–45. of performance samples. 91. 160–161. 216. 215. 17. 215–222. 258. xxvi stimulus-encoding process. xxxv–xxxvi. symbols in. 43. 220–221. dynamics of. 282. xli–xlii. multiple processes in. 204. 202. and neurological measurement. xxx–xxxiv. integrative research of. overview of models and frameworks of. 27. xviii. 2–3 stress. 59–60. 221 team effectiveness: conceptualizations of models and frameworks of. concept of. 279. 195. xx. 186–188. 7. and formal theory. 203. 98. distributions of. 4. of cognitive performance. 215. xxiii. 219–220. 168. See also selforganization systematic interrelations. 44. in schizophrenia. 217. definition of. 214. 61. and performance. 10–12. 112 structural-equation causal modeling (sem). 259–260 Stanford University. selection of models and . xxi standing-center exercises. 147 Stark. and effects of stress. 196–199. multisectored. 203–215. 4–5 structure: of cognitive process models. susceptibility to. uss. 61–66. and statistical property issue. and Kurt Lewin’s models. 7. xxxvi. 99–104 systems theories: characteristics of. 10. 138–140 sociology. 250. xvi. 260–261 superadditivity. 9 Switzerland. in clinical cognitive science. 18. 36. and ipo models. 197–198. 205. 193. 64–66 stimulus complexity. 20. 226–228. and stochastic modeling. 17. 250. 206–211. 103 Symbolics. in study of schizophrenia and psychological stress. 94. 223–224. See also cognitive task analysis (cta) tasks. 131–132. 202. 200. See order standard state. xxi stuck clients. 205–206. 88–89. See also chaos theory target levels of inference. 217– 218 taskwork. 193–196. 170 stability. 64–66. 94. 210–211. 218–220 social processes: in rehabilitation. 200.304 modeling complex systems social network analysis. 200 solution refinement. 124– 125. 186 statistical tools for prediction. 206. 138 sociotechnical systems theory. 85.

definition of. 11–12. 267–269. Lev. 185–186. Hans. See also team performance team performance: appraisal systems of. skills of. 223–224. 269n3 therapy. 209. 222. integrative framework of. See also latency distributions time-series method: attractors and trajectories in. 208. 115. 202. Giambattista. 153 Vahinger. 58. measuring of. 210. xii. computer-based learning environments (cbles). 250 Vincennes. 191. computer software Teknowledge. viability of. and skill competencies. 186. conceptual models of. 203–215. 215–222. 283–285. 209. in Big Five model. 255–258 tension. and process attributes of schizophrenia. 139 visual processing. dynamic interdependence in. 224–225 trajectories. 260n4 vector autoregression (var) approach. 99–101. 219. xiii. definition of. 23 time: and probability distributions. 212. 205. staffing of. See also team effectiveness team processes. 206–209. overview of models of. in integrative team-effectiveness framework. 217. 206. 262. 161–162 Weibull distribution. 191. 255–258 teleonomic order. 225. See psychotherapy. 51. science of. 58. 223. 250 waterfall models. 211. See also artificial intelligence (ai).305 Subject Index frameworks for study of. 207. and staffing. 186 visualization tools. 116–117 Today’s Evaluation of Psychopathology (tep). 55. 192–193. 278. 189–191. 224. 6 vertical apparent motion (vsam). 154 teleological order. and team effectiveness. 253–254 translational research. 203. 214–215. 190–191 technology. 257. 100 training. xxii–xxviii. 100. and therapy. and systematic interrelations. norms of. in team-adaptation model. 95–98. 99–104. 210–212. synthesized model of. 246 therapists. 224 teams: characteristics of. 224. 106–107 Vygotsky. 206.S. 15. 196. and training team members. 52. 227. 227 Thurstonian discriminal-difference size dispersions. 199– 200. model of. military. 111–112 Vico. 221. x–xi U. 94–99 transformation. 217–218. rehabilitation therapy-session reports. 104. 189. punctuated-equilibrium model of. team-evolution andmaturation model of. 209–210. and team performance measurement. 194 teamwork: and competencies. 225. 205. 70–72 . 3–4. 224–225. definition of. research on. 105–106 Thurstone scaling. 250 variation. xiii. uss. 116 verbal reasoning. 205. 280–281. history and nature of. xx Thales. xvii–xix vocational rehabilitation. 194. and psychological events. 209. 108.

96. 99–104 The Wizard of Oz (Baum).306 modeling complex systems Whitehead. 124–125 work structure. See Internet Yahoo!. Alfred North. 216. 96. 246 withdrawal. 218. 164 . xxxv World Wide Web. 219 World Health Organization.

F. S. 117 Anderson. B.. 123. 178 . xv... B. D. W. D. J... 201 an der Heiden. P. M. 9. J. 11. J. 19.... R.. Acton.. A. N. 250 Banerjee. G. 179 Best. 59 Berger. B. V. A. 27. 227 Bell. 174 Bacon. G. R. 43 Bandura. 206 Bélair.. 19 Berners-Lee.Author Index Page numbers in italics refer to illustrations.. B. R. 40.. 60 Bandettini. R. 18.. 127 Almond. 129 Ancona. P. R. W. 186. 5 Baker. 130 Andreasen. 5. S. A. R. E.. 1 Bailey.. 192 Benjamins. C.. R. 190 Aguirre. J. J. R. 124 Baur. 282 Barrick. R. 186 Bell. S. C. J. 11 Bamber. Roger. 132 Alexander. O.. 43 Aleong.. 221 Baker... 156 Anderson.. 56.. M. B... M. J. P.. 10 Baum. 102 Beard. xxxiv–xlii. J. 69 Averill. E. 117 Belbin. xiv. 186 Bell. H.. N. 218 Arnold. F. 90 Arpaia. K. D. A. 94 Ashby.. 225 Batchelder.. D. T. 131 Barker.. N. U..... S. 73n4. 105 Barab. T. 60. 126 Bachant. L. 125. Francis.. F. 187. 179 Benn. L... 125 Alley. 92 Ardison. 124.. 68 Batsell. G.. P. 19.. A. A... 187.. 45 Azevedo. K. R. 5 Balakrishnan. 52.

K. 105 de Jong. 125 Cramer. 193. A.-L.-P. H. 92 Boksman.. P.. 172 Cuesta. A. T. 147. 225 Carbotte. xi. 92 Bleuler... 214 Borowski. L. 226 Campion.. 225 Campione. 186. 129 Delucchi. M. T. M.. A.. 215. 117 Davis. L. 212. J. 129 Davison. 164 Boynton. 92. xvi. xxxv. S. J. 130.. S. J. F.. 220. M. R. S. 10 Bonneh.. 221 Collins. 62 Carter. R. C. T. D. 47 Crubézy. 228 Borwein. 108 Clancey. J. 46. B. J. 201 Chudowsky. xxxviii. F. L. 60. 170 Clark. K. A.. J. C.. 187.. 112 Borman. R. 64 Brooks. 188 Cooke. 9 Checile. xlii Brown. W.. 62. 158. Lee J.. D. S. M.. C. xl. 12 Brenner. I. 225 Browne. 64. J. J.. G. 131 Brown. 227 Cooper. S. 168 Chapman. 74n6 Campbell. 218 Dauwalder. L. 126 Chong... 8. Joe. R. 129 Broga. 192. 191.. D... 130 Colonius. 224. 225... B.. P. L. 60 Chi. 223 Brass. J.... 187.. A. C. 148 Buckingham. 190. D.. 163... W.. Egon. J. 186. M. 223. K... 189 Cook. M. 6 Brannick. 215. C. 279 Burke. C. 190. M. 186 Beyerlein... E... 168. I. A.. 124. M. 43 Braff. 194 Cohen.. 62. 90. 11 Confrey. R. 20.. A. 7. 279 Cobb. 74n5 Carter. B. 5. 206– 209.308 modeling complex systems Beyerlein. M. D. 193. H. 166. K. 62 Breyer. 188 Campbell. A. E. P. P. 86.. L.. L.. 221. 3.. J. R. 10. 86 Casti. xix. J. 186. A.. N.. 219. 218 Cooperman. 131. J. 112 Cordova. xli.. M. 125 Ciompi. C.. G.. 62 Chapman. R. C.. Y. 86 Clark. 193.... J. 130 Cannon-Bowers. 73n1 Borrill. 64 Brunswik. 95 Dailey.. 5 Buchanan... 125 Cohen. J.. 62 Chechile. 125.. 59. S. 223. 187. M. 46. 157. M. 194 Bowker. 43 Bollen. T. C. A.. 186 Blanco. D. 10 Cronbach.. J. J. J. Shawn. 62. 46. 68 .. 213. C. 8. J. M...... xiii. N. 221.. 15 Chabris. 169. T.. A. A. O. 109 Converse. x Burke. 19. H. E. 12 Brown. 278 Busemeyer.. 24. M. C. 112 Chandrasekaran. 108 Braithwaite. 125 Conrad.. R.. C... J. xlii Bryk. 68. 218 Braun.. 14 Clifton.. 130 Brown. K... C. P. D. J. A. 90 Braun. W. 159 Brown. 214. G. 10. 11. F. C. G. R. 190. J. 12. 73n1 Bottger. 163..

D. 186.. A. M. H. K.. 155 Fleishman. M. 203. J. S. 220 Dickson. P.. 117 Engel. T. 208. J.. 128 D’Esposito. D.. 125 Glass. Gerald F. G... J. D. M.. K. L. xvi. D. 13 Friston. M. 209. R. 28. 212–214. L. 172 George. P. D.. S. G. 70 Gennari... 125 Greer.. 218 Doob.. xlii Giel. 191.. 200 Emrich. 210.. A. 203 Goodwin. 5 Fiore. J. 4 Draper. N. J. A. 186. 131 Dunnette. 155 Driskell. 125.. C. 205. H. D.. 19. O. 124. 123... A.. 220 Fleiszer. R. E... 117 Gleick. M. 202. L.. 135–136. 105 Gray.. 43 Futrell. 70 Fergerson. G. S... O. C. 186. R. 46 Fitts. 201. 126. M.. M. 13 Gilmore. M. J. 23 Gersick. A.. M. 187 Eggan. A. D. 130 Embretson.. E. R. S. 94 Glover. 13 diSessa. 68 Garcia-Toro. H.. 126.309 Author Index DeMeuse... S. 215 Gheorghiu. A.. xiii. 46. M. 227 Fisher. 208. 9. James. 166. B... W. D.. 194 De Roure. 13 Glickman. 218. I. 225 Dubouloz. 28. M. 201 Gonzalez. 129 Golden. P. C. J.. 192 Eby. 138 Farr.. 179 Derry. C. J. 166. 189. W. T.. R... R.. 63 Goldstein. L.. A. A. 215 Glaser. 8. L. L. 201 Ellery. xi Fiske. 190. J.. 43 Gluck. Ronald... W. 7 Glass. L. K. R. 92 Gardner.. 189. 46.. 102. 43 Ericsson.. 127 Eisner. 151 Eriksson. 153. S.. S. 129. W.. 125 Dobbins. T. 43 Dickinson. K. R.. K. 168 Ernst. 90. S. 124 Feigenbaum. K. P. E... J. 127. 201. V. 2 Frackowiak.. W. 215. 170 Duffy. 178 Dreyfus. 49 Faremo.. M... G. G. F. 147. 201 Dirac.. L. 47 Emery. T. 92 Goodman. L. E. H. 5. S. 110 Gladstein. 171. M. E. L. M. 126 Fodor. P. H. G.. 94 Gilden. 52 Gibson. J. A. C. 52. C. S. M.. 152 Feller. 49 Fiore.. 156 Fokas... 64 Geddes. S.. H. J. 279 Freedman.. W. 218.. Z. A. 206. G. 110 Duda. A.. E. 178 Fific.. 178 Evans. S. 209 Globus. 64 Fouladi. R. R. 124.. 43 Fredrickson. 132 ... 194 García Pérez. 209.. 218 Edmondson.. P.. 203. 110 Gibbon. 90 Greeno.. T.. 278 Gott. 66. M... G. 127 Goberson. 190.... 130 Grawe. 43 Forster. 12. D. 152..

103 Haken. 64 Knoblich. 129 Johnson... B. 40 Huber. L.. S. P. 187.. 16 Higgs.. E. G. L. 174 Jonassen. 6 Hockley.. 87.. V. Kurt. 10 Grochowski.. 10. 73n2 Koch. R.. 225 Jacobshagen. xlii Heller.. M. 228 Klein.. J. 46... 12 Homan. xv–xxii. Fritz. G. 56 Johnson.. 117 Hollenbeck. V. xvii. 87 Koçak. 186. 6 Huettel. 87. 110 Ilgen. R.... 200 Johnson.. N.. 62. 52 Harrison. R. D. 5 Joshi.. 156.. D. D. A. 113 Kahn. N. S. 200 Junghan.. 221 Haemmig. W. 202 Guzzo. 215. A. R.. 43 Huselid. 60 Heeger. 213... D. 57. J.. 200. 194. E... D. R. xvi Klein. 206–209 Highgate-Maynard. J. M. N.. E. B... 101. ix Koffka. U. J. A. 193 Klimoski. 91 Kotz. 43 Heider. H. 200. 132 Gully. C. xv. L. J. 24 Hinsz. 282 Hendler.. R. S. R. xxxviii.. J. 8. B. 186.. 218 Herrenkohl. 204. S. J. W.. 97.. 166 Guarino. W. S. H. E. 71 Kelso. 43 Hashimoto. 189. S. S. G. F.. G. 218 Hohenschutz. 71 Kerzel.... 110.. 129 Jorenby. 106 Hogan... G. D. R. T. M. M. P. 60 Harris.. R. A. 201 Hackman. 101. 129 Jones. 11 . 225 Hyman.... 53. 225 Kline. 87 Kiekel. 49 Johnson. xix. E. H. S. D. W. L. George S. 5 Knapp. R.. R. 192.. Y. 219. 36 Katz. M. 27. K. L.. 103 James.. S. E.. 60 Knight.. 129 Herrnstein. xvii. P.. W. D.. H. E. C. M.. 225 Holmes. M. 225 Kay. 225 Kahneman. D. G.. 8 Haynes.... A.. K. 225 Jones.. T. 187.. 124 Jetté. J. 59 Jundt.. 23 Hoffmann. R. N. S. D... M. A. 190. 132 Johnson. R. 12 Huberman.310 modeling complex systems Gregson.. C. N. Immanuel. 90 Kendall.. 108 Hastings. 11. D. 10 Klein. 200... 186 Johnson. 200. 46.. 164 Guerrera. A. xvi. D. C. 227 Kirk.. 22.. M. xx Köhler. 93. G. S. 111 Hamilton.. A... L. 127 Jones. 1 Karabatsos. H. 195 Klein. 175 Hintzman. J. R.. A. R. 174 Kant. 6 Kline. A.. 110 Grosso. 220 Hinton.. 91.. xxxiv. 200 Jackson. xiv. 179 Herold.. 93 Keeping. 186.. 155... 214 Kenny. B. S. B. 187.

. M. 218 McIntyre. 97.. M. 62 Kozlowski. 278.. 147 Link. C. xvi. P. 211 Matussek.. A. xx. L. 10 Lehrer. A. 225 Kozma. J.. 60 Marinakis. J. 132 Lederberg. R. W.. E. R. 179 Lave. R. S. 126. T. M. 40 Lütkepohl. 190. 186 Mandelbrot. 132. 129 Mathieu. 13 Manifold. 2.. 187. 191. V. R. 12 Krueger.. 11 Luke. 127 Logan.311 Author Index Koyfman.. P.. Michael J. R. S. 210. 193. 94 Liebowitz. 190.. H. 193. 186. xix.. 73n5 McDermott. 187.. D. 147 Leeper. T. 192 Lawson. 211. 111 Kuhn. S. 126. 43 Marks. R. 211 Lewkowicz. A. xxxiv. 225 Lepper. 110. 114 . J.. A... N. C. J. 201.. 127. D. J.. 192. 6 Lloyd. R. 52 McGrath. 117 Logan. 129 Lewin. 135.. V. A. 224 McPherson. A. J. T. 220. 202. xii. 190. O. 136. 206–209 Meehl. B... 46.. 5 Mach.. 59 McClelland. 133–134. 146 McCarty. N. 124.. E.. 4. 126. J.. 74n6 Marx. 186. 174 McFall. xxix. 218 Krech. W.. 212–214. 186. D. x. 101... J. C. 43 Lassila. 109 Mayer.. 132 Michotte. xi. P. David. D. J. 94. xi.. 220. M. E. 221. 269n3. 156 Lysaker. xxi. Frank. G. 261. 224.. xvii Mahoney. E.. 87 Medsker. 63 Meier. 43 Lanius. 215. 99 Lyons. N. K. 47. 62 MacCallum.. 125. H. D.. Benoit B. W. 209. 221.. 225 Kupper. 279 Lu.. 105 McCarthy. 134 Kushmerick. 99. Paul E. 258 Major. 43 McCarthy.. D. 113 Lieb.. 279. 128. A. W. Robert Ward. E. M. 9. 11 Marr. R. 8. 4. 258. D... P. 129 Krackhardt. 168. 130 Lavigne. 190 Mechsner. 1. Suzanne P. 124. J.. 1 Kukde... 218. S. 125 Lesgold. R.. xiii. J.. J. A. 281 Landsberg.. 158 Metz. 138.. 106. P. R..-P. B.. 202. E.. 95. 9 McGill. 46. Zeno. 139. 90. C. A.. A. 215.. 153. J. 125 LePine.. S.. 129 Maze.. G. 110 Menzies. 145 Lindenmayer. R. B. 256.. J. Kurt.. 45 Lefebvre. 279 Mahoney. 5. N.. 7. 201. J.. 132. A.. F.. M. 224 Marley. M.. xxviii–xxx Lees.. K. xxviii Lopez.... 129.. 127. R. 215. J.. 133. A. 187. 110 Lindsay. J. D. 200. 114. 171. 210. xiv. xv. J. R. T. xxix Krieg.. J. 105 Lane.. 10 Kuncel. R. 135 Lawler. Ernst. J... W. A. 124. 191. 162 Kruse. 158 Lajoie. 139 Luce.

156 Nissen. 166. 60. 62. A.. 138 Nason.. 10. J. J. 225 Neufeld.. 158.. 155 Ployhart.. 8. V.. O.. 19. 68. 8 Pearson. xv. 94 Phillips. G. G. R. 186.. 64 Nakamura. R. D. 14. 25. Jr. W. L. 218 Munsie. 125 Nelson.312 modeling complex systems Miles. F. M.. E. B. R. 129 Pezard... 146 Mislevy. 46. J. 110 Plantinga. I.. 28. 129 Papper. J.. 223 .. S. 21. D. R. 130 Newton. I. 145. B. S. B. 218 Paivio. A. J. M. 215 Morrison.. 168. E. M. C.. 12 Noy.... 73n5 Pea.. R. F. J. J. 87 Nienhuis. 214 Mount. B.. W. xi Pellegrino... B. 225 Mullen. S. 13 Prendergast. 62 O’Leary. 60 O’Donnell. 45 O’Brien.. 5 Pierce. 94 Nieva.. E. 10 Price. A.. R. J. 86 Pfister.. 178 Nozawa. 11.. 214 Place. 16 Nichols. xx. 5. 31. 23. B. xii. 66.. 172. 24. 108 Pavlekovic.. 23 Musen. E. G. 171. M.. A. J. xiv. 52.. K. S.... P. 158 Newcomb. P. 125 Penny. P. 278. 30.. 86. 7. 173. A. W. 151. V... 129 Monahan. M. 68 Moreno. 110 Piasecki. L. 213. N. 74n5.. 38. S. C.. Karl. A. G. 177.. W. S. J.. 95 Perkins.... R. Richard W. 12. 193 Paradiso. 209. D. 129 Morgan. 201 Priest.. 92 Orasanu. K. 9. S.. K. 55. 278. 10 Polkinghorne. 10 Miltner... 6 Miller. 24. 202 National Academy of Science. S. 168 Newman. A. 225 Newborn. G. Jr. 53. 208. N. 178. 228 Prigogine. 56. 20.. M. A. H. J. 282 Mooijaart. E. B. D. Isaac. M. W. J. 46 Nicolis. 59 Paulus. E. M. 25. 12 Patil.. 129 Peacock. ix Norman. P... D. 279 Neubert. T. J. 46 Mothersill. 45. G. I. R. 285 Myung.. 203. R. M... 27. 38. C. 215 Nisbett. O.. H. J. 132 Murdock. A.. L.. 92 Parzen. 92. 46. M. Mark A.. B. 26.. 166. M. G.. Theodore. 153 Pressing.. 10. B.. 59. 29.. 43 Peralta. D. 87 Prince. 23 Palincsar.. 178 Nosofsky.. 281... C... 189 O’Shea.. 202. 109 Pfeifer. 64. xxx–xxxv Newell. E. 127 Polking. F.. xi. 22. H. E. 129. 285 Neuman. 117 Minsky. 90 Milton... 127 Nicholson. 20 Motowidlo.. P.. R.. 110 pdp Research Group. 72 Morrison.. J. 225 Pokorny.

Kevin C.. 211.. Eduardo. 105 Shortliffe. J. E. 2 Steinberg. 161 Schuler. A. H... 164 Stefik... V... J. 108.. 9. R.. xiii. 171 Rubin. 109. 228. R. A.. 117 Ross. 73n5 Ruscio. 201 Schiffman. S... E. 68 Robertson. S. 12 Shahar... T. K. 159. 187. 52. B. 213. 113 Schuster. L. S. 43 Reynolds. P.. 94 Smith. J.. 282 Raudenbush. 212 Singer.. L. I. 125 Scheier. 87 Puerta. M. O. 146 Roe. S. M... R. M. 20 Rosch. W. C. 208. 64. xiii.. P. 220.. J. 111 Stagl... 100 Staab. 44 Smith.. G. 224.. 6. M. S. 105 Silverstein. D. 92 Santor. A. D. S. S.. D. 113 Shiner. 161. 155 Reick. 5 Russell. 282 Schauble. E. 125... 129 Regoczei. E. 10 Rappaport. E. E. S. 226 Rumelhart. 90 Sipe... 220 Shimojo.. 278. D. A. J. 190.... D. 129 Steiner. 164 Staddon. R. P. S.. 86. 129 Sommerville..... 281.. 105. 223. 195 Sergent. J.. 129 Salva. D. J. H. H.. 223. 73n4 Stadler. E.313 Author Index Prinz. J. 279 Soloway. W. 92. D. M. J. M. 151.... 8. 191. J.. L. 70 Sego... xiii. 190. M. C. E. 225. 187 Smith. 145 Steiger. 5 Reder. Y. 60. 153. 43 Spaulding. 93 Schreiber. 7 Singer. D. D. E. 189.. 12 Schneider.. 74n8 Sarason... 147. 218 .. W.. K. E. 225 Schneider. A. C. 49. S. 285 Salomon. W. 92. 13 Snyder. 218. 202.. H. 278 Star. 279 Senge. 206. 126. 110 Simon.. 203 Reiman. 146 Shiflett.. J. A. M. 168 Shannon.. 228. 12 Riefer. F... T. 168 Simons. 127 Sibbald. 201. 110. 125. E. J. 176 Song. D. P. 212. D. 225 Smith. C. 11. R. L. 186. 227. 148. 220. 225 Slooff. 86.. I. 186 Seligman. 186 Rochester. M. P. 88 Ruderman. 129 Sagi. E.. I. 171 Shute. M. W. 215.. 113 Schemmer.. 70 Rössler. 62.. 225 Ruel. 125.. T. E. J.. S. M. 214... C. 94 Scott... Elliott. 168 Ramsay.. N. M.. G. B. J. O. 214. 5 Ruscio.... G. D.. T. P.. M. A. 2. L. 74n8 Rappaport. 209. L.. S. 5 Smith-Jentsch. 131. M. 112 Salas. 213. S.. M. 125 Reeves. S. N. W. 186. 202 Smith. L. A. I. J. 190 Smolin.. S. 112 Sims. 95 Rothenfluh.

69 Wertheimer. R. A. G. 174 Uhlhaas. 6. 94.314 modeling complex systems Sternberg. S. 187. B. 208.. D. H. E. 90 Viken.. S. G. 200 Sutherland. R.-A. 12. L. 211. H. C. W.. 94 Wiggins.. xxii– xxviii Waller. D. W. 132 Williams. 90 Tindale. R... 10. 9. 190 Zuboff. J. 45. 19. 99. xv. 219 Taub. xv. 106. 45.... D. B. 105. J. 136. xiv. E. Max. 113. L. M.. 94 Witte. A. E.. 5 Townsend. A. 47 Wilkie. 4 Vinacke. 9. 43 Van Zandt. 90 World Health Organization. 205. 12 Tannenbaum. 64 Walker. 187. 164 Suchman.. C. T. 44. 228 Wicker.. E. S.. 19. F. 49. E. 18. 23 Wright. I. 8. J. 190. S. 60 Weissbein. 56. D. 220 Tjosvold. P. 20 Weick. 200. M. 18. 4.. 66 Vollrath. 186. 22 Stevens. 93.. 86. xvii West. 74n6 Trist. 11..... K. 43 Zeisig.. 9. 132 Wagenmakers. 201. 209 Tversky. R. 12 Young.. 145 . xi... P.. 95. J. 10. 52. E. 200 Vygotsky. 103. 225 Storey. W. D. M. L. 24. 200 Tschacher. M. 178 Weinstein. Wolfgang. 130. J. 68 Tolman. P. L. G. E. 108... 129 Van Petten. 62 Winston... 5 Wang.. 110.. 194. 194. W. 110 Unsworth.. W. 220 Zarahn. 228 van Joolingen. 153 Wiseman. 66 Varela. 220 Volpe. P... A.. E.. 52. J.. D. C. A. 19 Studer.. 206. V. D.. L. 225 Stewart.. J. 13. 56. 156 Wilt. W. 147 Takane. U. 157 Sundstrom. J. 138 Wittchen. 178 Stout. Edward L.. T.. 24.. 215.. T. Y. N. L. L. A. 101.... T. S. C. 131 Wenger. 194 Young.. 64 Waldorp. J.. 102. J. 27.. 134 Tu... 187 Tollenaar.. R. 225 Wenger.. 129. 27. 114. xlii Wiersma. L. 6. 69. R. M. S. 190. 145. S. F.. K. L. D... M. 25. J. 28. 190..... xxix Tomarken. 171 Tuckman. M. 109. 152 Zaccaro. xxxv. S. 66 Wilson. 190 von Bertalanffy. H.. N. x Vollick.. S. 27. E. 192 Weinberg. J. 130 Williamson. 8. 27... 168. 7. 193. 225 Yetton. J. 117. Y. A. N.. 282 Wright.. 189. W..