You are on page 1of 28

Personality and Social Psychology Review Copyright © 2004 by

2004, Vol. 8, No. 3, 220–247 Lawrence Erlbaum Associates, Inc.

Reflective and Impulsive Determinants of Social Behavior
Fritz Strack and Roland Deutsch
Department of Psychology
University of Würzburg

This article describes a 2-systems model that explains social behavior as a joint func-
tion of reflective and impulsive processes. In particular, it is assumed that social behav-
ior is controlled by 2 interacting systems that follow different operating principles. The
reflective system generates behavioral decisions that are based on knowledge about
facts and values, whereas the impulsive system elicits behavior through associative
links and motivational orientations. The proposed model describes how the 2 systems
interact at various stages of processing, and how their outputs may determine behavior
in a synergistic or antagonistic fashion. It extends previous models by integrating moti-
vational components that allow more precise predictions of behavior. The implications
of this reflective–impulsive model are applied to various phenomena from social psy-
chology and beyond. Extending previous dual-process accounts, this model is not lim-
ited to specific domains of mental functioning and attempts to integrate cognitive, moti-
vational, and behavioral mechanisms.

In the history of attempts to discover the causes of reflect a lack of appropriate information (Friedman,
human behavior, the most widespread explanations are 1976). The second strategy postulates more than one set
based on the assumption that human beings do what of principles that may control human action. For exam-
they believe is good for them. Thus, they are construed ple, a behavior may occur mindlessly (Langer, Blank, &
as “rational animals” capable of recognizing the value Chanowitz, 1978) or automatically; that is, without di-
or utility of their actions. recting much attention to the utility of an outcome, a per-
At the same time, however, it is obvious that human son may act the way he or she has acted many times be-
beings do not always act this way; that is, under certain fore. Such habitual behaviors were the focus of many
circumstances people behave in ways that do not reflect psychological theories in which the frequency and re-
their values. To account for this phenomenon, to which cency of previous executions of a given behavior were
the Greek philosophers gave the name akrasia (e.g., seen as primary determinants (e.g., Hull, 1943). The
Mele, 1992), several strategies have been pursued. The third strategy has been to understand human behavior as
first strategy assumes ignorance or lack of knowledge a function of drives. In particular, basic needs that are bi-
on the part of the actor. Socrates, for example, claimed ologically rooted, such as hunger, thirst, or reproduc-
that if people only knew what is good for them, they tion, are seen as major forces. Their strength may over-
would act accordingly. A similar position is held by ride considerations of utility and determine behavior in
modern economists who imply that irrational decisions an immediate fashion.
Although most psychological theories have focused
This article received the 2003 Theoretical Innovation Price of the on one of these aspects, some have acknowledged that
Society for Personality and Social Psychology (SPSP). This research behaviors may be multiply determined. Most promi-
was supported by grants from the Deutsche Forschungsgemeinschaft
nently, Freud (e.g., 1933/1949) described human behav-
(DFG) to Fritz Strack (Str 264/19–1) and from the
Deutsch-Israelische Projektkooperation (DIP). The authors are grate- ior as controlled by a “psychic apparatus” that includes
ful to the members of the Würzburg Social Cognition Group who pro- several operating principles: a superego composed of a
vided valuable suggestions. Special thanks go to Dolores Albarracin, person’s values and norms, an id that operates in accor-
Axel Bühler, Bertram Gawronski, Arie Kruglanski, Ravit Levy, dance with the basic drives a person is endowed with,
Thomas Mussweiler, Eliot Smith, Bob Wyer, and five anonymous re-
viewers for their detailed comments and criticisms. Also, we thank the
and an ego that integrates and often reconciles the forces
participants of our seminar on the topic, held in summer and winter from the superego and the id. Thus, Freudian theory also
2001 at the University of Würzburg. The project has greatly profited describes how different processes may interact. In par-
from the lively discussions and the valuable input from our students. ticular, Freud deserves credit for emphasizing the im-
Requests for reprints should be sent to Fritz Strack or Roland portance of unconscious processes in the determination
Deutsch, Lehrstuhl für Psychologie II, Universität Würzburg,
Röntgenring 10, 97070 Würzburg, Germany. E-mail: strack@
of behavior. Although Freud’s thinking has greatly stim- or deutsch@ ulated psychological theorizing, evidence for the pro- posed mechanisms was largely anecdotal or based on


clinical case studies; that is, little evidence has been ated between experiential and rational processes and ap-
available to evaluate the postulated processes. plied his theory to general mechanisms of thinking.
More recently, the conviction that human behavior Finally, Sloman (1996), whose model serves as a con-
is guided by more than one underlying process has led ceptual orientation for other dual-process approaches,
to a series of research programs that have resulted in a invoked associative and rule-based processes to account
number of dual-process theories. This converging evi- for human reasoning.
dence has been extensively discussed by many authors Over the years, dual-process models have been re-
who contributed to a recent book on dual-process theo- fined and updated in various ways. For example, Smith
ries in social psychology edited by Chaiken and Trope and DeCoster (2000) have integrated new findings from
(1999; for more complex theories of mental processes, research on memory systems (J. L. McClelland,
see, e.g., Kuhl & Goschke, 1994). Of course, these the- McNaughton, & O’Reilly, 1995) to explain how asso-
ories are not entirely unique but share a considerable ciative and rule-based processing may be rooted in more
number of features. In a systematic comparison, Smith basic cognitive structures. Lieberman, Gaunt, Gilbert,
and DeCoster (2000) juxtaposed the nine most promi- and Trope (2002) described a cybernetically oriented
nent models and identified their common and distinc- two-process model while integrating findings from neu-
tive characteristics. roscience. Thus far, however, dual-process models have
Most important, all models propose two modes of directed little attention to the behavioral consequences
information processing that are distinguished, accord- of the mental mechanisms they describe, and they do not
ing to Smith and DeCoster (2000, p. 111, Table 1), by provide an alternative to rational models of human be-
the following characteristics: rule-based processing havior. In particular, they are silent on the problem of
“Draws on symbolically represented rules…That are akrasia. Although there are some accounts that focus
structured by language and logic…And can be learned primarily on behavior (Bargh, 1990), they do not pro-
in just one or a few experiences…Occurs optionally vide an integration into a comprehensive model of social
when capacity and motivation are present…And often cognition. This, precisely, is the aim of this article,
with conscious awareness of steps of processing;” in which attempts to incorporate insights from motiva-
contrast, associative processing “Draws on associa- tional science (Higgins & Kruglanski, 2000) into the
tions…That are structured by similarity and contigu- general dual-process idea.
ity…And are learned over many experiences…Occurs Although Smith and DeCoster’s (2000) survey is
automatically…And preconsciously, with awareness certainly not complete (cf. Wilson, Lindsey, & School-
of the result of processing.” er, 2000), it is obvious that the processes described in
Whereas these characteristics are shared by the ma- these models focus on judgments and information pro-
jor two-process models, Smith and DeCoster (2000) cessing, whereas behavior does not play an integral
saw the most substantive difference in whether the two role. In other words, behavior is implied in these mod-
types of processes occur simultaneously or in a mutually els primarily to the extent that it is preceded by a judg-
exclusive fashion. In addition, the existing models are ment or a decision. However, behavior is known to oc-
distinct in that they focus on different phenomena and cur without such antecedents. For example,
employ different terminology to describe various as- judgment-based behaviors may become habitualized
pects of the two types of processing. For example, through frequent execution and be carried out inde-
Chaiken, Liberman, and Eagly (1989) as well as Petty pendent of the implications of the original judgment
and Cacioppo (1986), who are among the proponents of (Aarts & Dijksterhuis, 2000; Ouelette & Wood, 1998).
dual-process models, spoke about heuristic/peripheral In addition, strong motivational forces may drive a be-
and systematic/central processing and applied their havior in a direction that is inconsistent with an actor’s
model to the domain of attitude change. In the spontane- beliefs and values (Metcalfe & Mischel, 1999; Strack
ous mode of Fazio’s (1986) model of attitude access, & Neumann, 1996). However, such direct influences
well-learned attitudes are assumed to be automatically on behavior have links to mechanisms described in the
activated by attitude objects. Only if sufficient motiva- models summarized by Smith and DeCoster (2000).
tion and opportunity are present, will deliberative pro- Therefore, it seemed necessary to integrate behavioral,
cessing determine the evaluation of an object. Devine motivational, and cognitive components into a
(1989) focused on stereotyping and assumed either au- two-system model of social behavior.
tomatic processes or more effortful stereotype suppres- Along with others (Kahneman & Frederick, 2002),
sion. Gilbert (1989) distinguished between correspon- we assume that the processes operate in parallel in-
dent inferences and attributional thinking to explain stead of being consecutively invoked, and we posit that
phenomena of person perception and biases in causal at- the two systems are concurrently active and compete
tribution. Martin, Seta, and Crelia (1990) were inter- for control of an overt response. Thus, in contrast to
ested in social judgments and juxtaposed automatic most of the dual-process models previously described,
contextual influences and more strategic and effortful the proposed model is a two-systems model (see also,
attempts at correcting them. Epstein (1991) differenti- Gilovich & Griffin, 2002) of social behavior.


To enable this model to go beyond explaining the Basic Properties and Functions
generation of social judgments, we attempt to integrate
elements from other research programs that provide For reasons of clarity, the proposed model is de-
links to social behavior. Specifically, the work by Nor- scribed in 10 theses (see Figure 1 for an overview). As
man and Shallice (1986), Cacioppo, Priester, and mentioned before, several components are shared by
Berntson, (1993), as well as Gollwitzer (1999) provided existing dual-process models. The descriptions of
important conceptual elements that added a motiva- those components are somewhat briefer than of those
tional dimension to the model. In particular, we suggest we believe to be unique to this model.
that behavior is a function of schemata that are jointly
controlled by environmental input and superordinate at- Thesis 1: Basic assumption. Social behavior is
tention (Norman & Shallice, 1986), and we propose the the effect of the operation of two distinct systems
existence of a motivational orientation that acts as a be- of information processing: a reflective system
havioral catalyst and relates valence to approach and and an impulsive system. The systems can be
avoidance (Cacioppo et al., 1993). To bridge temporal specified by different principles of representa-
gaps between a decision and its behavioral implementa- tion and information processing.
tion, we integrated a mechanism of intending
(Gollwitzer, 1999). In addition, we propose that the de- In the reflective system, behavior is elicited as a con-
privation of basic needs influences spontaneous evalua- sequence of a decision process. Specifically, knowledge
tion and preactivates behavioral schemata relevant for about the value and the probability of potential conse-
the satisfaction of the deprived needs. quences is weighed and integrated to reach a preference
Taken together, the value of this model is not that it is for one behavioral option. If a decision is made, the re-
new in each of its components. Rather, we see its merits flective system activates appropriate behavioral sche-
in its attempts to integrate elements from existing theo- mata through a self-terminating mechanism of intend-
ries and to describe how they interact at different stages ing. In contrast, the impulsive system activates
of processing. Most important, we try to tie mental pro- behavioral schemata through spreading activation,
cesses to social behavior in a nontrivial way; that is, we which may originate from perceptual input or from re-
do not assume that behavior follows inevitably from a flective processes. As described in James’ (1890)
decision and therefore does not deserve attention be- ideo-motor principle (see also Lotze, 1852), a behavior
yond its cognitive precursors. Instead, we construe so- may be elicited without the person’s intention or goal. In
cial behavior as the result of several determinants that addition, the activation of behavioral schemata may be
may operate in accord or conflict with each other. moderated by motivational orientations or deprivation.

Figure 1. Overview of the reflective–impulsive model. Note that reflective and impulsive processes are represented by solid or broken
lines, respectively.


Thesis 2: Parallel operation. Both systems oper- Representation, Storage, and
ate in parallel. However, there is an asymmetry Processing of Information
such that the impulsive system is always en-
gaged in processing (by itself or parallel to oper- Thesis 4: Relations between elements. Elements
ations of the reflective system) whereas the re- in the two systems are connected by different
flective system may be disengaged. types of relations. In the reflective system, ele-
ments are connected through semantic relations
to which a truth value is assigned. In the impul-
This model assumes that information entering the sive system, the relations are associative links
perceptual gates will always be processed in the impul- between elements and are formed according to
sive system. However, the impact of that information the principles of contiguity and similarity.
depends to a great extent on the preactivation of those
structures in the impulsive system in which the infor- The world as well as inner states of the organism are
mation is represented. Depending on its intensity and represented in different ways in the two systems. In the
the attention it receives, a stimulus may also enter the impulsive system, information is represented in pat-
reflective system. In that case, impulsive and reflective terns of activation in an associative store. In contrast,
processing occurs in parallel, and are assumed to inter- the reflective system is capable of forming proposi-
act at various stages of processing. tional representations by connecting one or more ele-
ments through the instantiation of relational schemata
Thesis 3: Capacity. The reflective system re- to which a truth value is attached. Of course, represen-
quires a high amount of cognitive capacity. tations in both systems are ultimately implemented
Therefore, distraction as well as extremely high within a neuronal network; the difference between
or low levels of arousal will interfere with its op- simple associations and propositional representations
eration. In contrast, the impulsive system re- therefore refers to the computational level of cognitive
quires little cognitive capacity and may control modeling (Marr, 1982).1
behavior under suboptimal conditions. As a con-
sequence, processes of the reflective system are Associative Links and Structures in
disturbed more easily than those of the impulsive the impulsive system
system. In this model, the impulsive system is conceived of as
a simple associative network. In accordance with most
other associative-network models (see Smith, 1998), we
One of the greatest advantages of the impulsive sys-
assume that the links between elements have different
tem is that it is fast, requires no or little cognitive effort,
strengths that are stable and change only gradually
and has a low threshold for processing incoming infor-
through learning. If an element is activated, activation
mation, whereas the opposite holds for reflective oper-
spreads to other elements in proportion to the strength of
ations. The reflective–impulsive model shares this as-
the link. The activation of elements in the network can
sumption with almost every dual-process model, and
vary rapidly. The accessibility of a content will be in-
relevant evidence is reviewed elsewhere (see, e.g.,
creased by the frequency and recency of prior activation.
Chaiken & Trope, 1999; Smith & DeCoster, 2000).
In this way, an element acquires an activation potential
However, we emphasize the role of arousal for a proper
(Higgins, 1996) that reduces the amount of additional
functioning reflective system. Resembling the
activation necessary for retrieval or further processing.
Yerkes–Dodson Law (Yerkes & Dodson, 1908), we
In general, links are created or strengthened if stim-
propose that the reflective system operates most effi-
uli are presented or activated in close temporal or spa-
ciently at intermediate levels of arousal. On one hand,
tial proximity. The resulting links reflect correlations
high levels of arousal facilitate well-practiced, domi-
between aspects of the environment and cognitive, af-
nant responses (Hull, 1943; Zajonc, 1965). Recent
fective, or motor reactions, without representing the
findings have demonstrated that arousal also affects so-
causes of such multimodal correlations. As a conse-
cial judgments and perception, in that it enhances the
quence, structures emerge in the impulsive system that
influence of stereotypic and evaluative associations
bind together frequently co-occurring features and
(e.g., Bodenhausen, 1993; Paulhus & Lim, 1994).
form associative clusters. In essence, we assume that
More important, this effect has been recently attributed
the associative store of the impulsive system works like
to a weakening influence of high arousal on controlled
a simple memory system (see Johnson & Hirst, 1991),
or reflective processes (e.g., Baron, 2000; Lambert et
which slowly forms enduring, nonpropositional repre-
al., 2003). On the other hand, very low levels of
arousal, as in the state of drowsiness, are associated
with poor reflective processing and poor self-control 1Marr (1982) distinguished between the computational,

(e.g., Baumeister & Heatherton, 1996). algorithmical, and implementational level of cognitive processes.


sentations of the typical properties of the environment Although the connections between elements in the
over many learning trials (see J. L. McClelland et al., impulsive system do not carry a truth value and do not
1995; Smith & DeCoster, 2000). The impulsive system reflect declarative knowledge about elderly people be-
has low flexibility but is fast and needs no attentional ing slow, the associative link between elderly and slow
resources. may bias perception and influence behavior if it is acti-
In addition, associative links can be formed through vated. Moreover, it is possible that motor programs that
reflective operations. This is possible because every have been executed frequently in connection with el-
propositional representation in the reflective system derly people in the past may again be activated. Re-
activates corresponding contents in the impulsive sys- search on the connection between perception and be-
tem. As a result, elements that do not co-occur in real- havior bolsters the idea that semantic concepts can be
ity but are often related to each other in the reflective directly connected to motor programs (e.g., Bargh,
system will also become associatively linked in the im- Chen, & Burrows, 1996; Dijksterhuis & Bargh, 2001).
pulsive system (cf. Smith & DeCoster, 2000). Thus, se- Recent findings on the interplay of gestures and lexical
mantic concepts will emerge in the impulsive system access indicates that this connection is bidirectional
through frequent propositional categorizations. It is (e.g., Krauss, Chen, & Gottesman, 2000). Associative
important to note that the links in the impulsive system, clusters in the impulsive system can be hierarchically
different from some other network models (e.g., Col- structured and can therefore differ in abstractness. As a
lins & Quillian, 1969), are not assumed to have any se- consequence, clusters may resemble either concrete
mantic meaning by themselves. Therefore, the only re- perceptual concepts or abstract semantic concepts or
lation between two or more elements is that of a mutual schemata.
activation. Processes in the impulsive system may be accompa-
For example, if we see an elderly person, perceptual nied by an experiential state of awareness; that is, with-
features such as hair color or body posture may activate out necessarily knowing its origin, people may experi-
specific elements in the impulsive system (see Figure ence a feeling with its distinct phenomenal quality. For
2). Because such elements have previously been paired example, a person may have a visual perception of
with other features that are correlated with advanced lightness or darkness, a pleasant or unpleasant feeling,
age, a whole cluster of elderly features will be acti- or the experience of pain or familiarity without know-
vated. As a consequence, contents of the elderly stereo- ing the concepts or categories of light, pleasantness,
type will be more readily accessible and may guide pain, or familiarity. Thus, the impulsive system can be
subsequent processing. For example, the concept of understood as a system of experiential primacy, in
slowness may become activated in the impulsive sys- which affective and nonaffective feelings are generated
tem and reflect our direct or indirect experiences with quickly and without syllogistic processes of inference
elderly people. (see also Zajonc, 1980).

Figure 2. Activation of a hypothetical elderly cluster in the impulsive system. The perception of gray hair leads to a higher accessibility
of associated contents and may facilitate associated behavior.


Figure 3. Schematic representation of a propositional categorization. The concepts elderly and person, as well as the relation is a, are re-
trieved from the impulsive system and transformed into a proposition.

The Generation of Knowledge again the perception of an elderly person. The perceiv-
Through Propositional able features have spread activation to the elderly
Categorizations and Syllogistic concept in the impulsive system. To generate a proposi-
Inferences in the Reflective System tional categorization, the relational schema of category
membership (is a) will be retrieved from the impulsive
The impulsive system can be thought of as long-term
system and combined with the label elderly and the rep-
memory, whereas the reflective system has the proper-
resentation of the perceptual input, in this case the visual
ties of a temporary storage in that the amount of infor- representation of the person. Thus, the propositional
mation that can be represented at any given time is lim- representation this is an elderly person is generated (for
ited, and the representation will fade if it is not rehearsed a detailed computational account on how relations and
(Baddeley, 1986). The reflective system generates de- their arguments are bound together to compose proposi-
clarative knowledge by assigning perceptual input to a tions, see Hummel & Holyoak, 2003; see Figure 3).
semantic category. Unlike simple associative links and The number of relations that can be applied to these
structures, knowledge in the reflective system consists contents is nearly infinite. Beside simple logical rela-
of one or more elements to which a relational schema is tions, such as is a, is not, or implies, there are also more
applied. Most important, a truth value is assigned to that complex and abstract relations, such as causality. In ad-
relation. As a consequence, this system represents states dition, there are many social relations such as friend,
of the world or the organism in a propositional format. In enemy, spouse, or partner. Of course, new relations can
this endeavor, the reflective system is driven by the prin- develop as a person’s knowledge expands.
ciple of consistency as it strives to avoid or remedy in- Once knowledge has been generated, syllogistic2
consistencies between its elements (Gawronski & rules are applied to draw inferences that go beyond the
Strack, in press). An important feature of representa- information given (Bruner, 1973). For example, a per-
tions in the reflective system is that they can be flexibly son may wonder how wise an elderly person is. Cate-
generated and changed. Thus, the reflective system can gorical knowledge about the elderly may be derived
solve a multitude of tasks, such as reasoning, planning, from the categorization of a given person as elderly.
or mental simulation. However, it is slower than the im- The quality wise may then be inferred based on this
pulsive system and requires attentional resources. categorization. Note again the fundamental difference
How are such representations generated? We assume to the mere activation of the concept, which facilitates
that the elements of the proposition—that is, one or the inference but does not generate knowledge about
more concepts and the relation that is applied to them—
are retrieved from the impulsive system. The reflective 2The term syllogistic is meant to refer to all types of rules that are
system generates semantic or episodic knowledge by as- used for the transfer of truth from the premises to the conclusion. There
signing a truth value to the concept and the relation. Take are no implications about the validity of such syllogistic inferences.


elderly people being wise. This becomes particularly within” (Bless & Forgas, 2000). In particular, feelings
apparent in attributional thinking such that the charac- of different qualities may enter into the reflective sys-
terization of a particular behavior may spontaneously tem if they are propositionally categorized and contex-
activate categories that correspond with personality tually qualified (e.g., Schwarz & Clore, 1983). A good
traits (Uleman, 1999). However, these categories may example of how feelings that emerge from impulsive
eventually not be used to categorize the person because processing can enter reflective processes and even de-
the context implies alternative causes of the same be- termine choice comes from a study by Bechara,
havior (e.g., Trope, 1986). Damasio, Tranel, and Damasio (1997). In this experi-
Operations of the reflective system are accompa- ment, participants had to choose cards from four decks,
nied by a noetic state of awareness, which consists of two of which offered high gains at the risk of an occa-
knowledge that something is or is not the case.3 A sional high loss, while the other decks offered lower
noetic state of awareness may be accompanied by an gains but with a smaller risk of a loss. Overall, the less
experiential state of awareness, which consists of a par- risky deck led to a higher payoff.
ticular feeling. For example, the process of trying to When participants were asked to choose from the
answer an almanac question may be accompanied by a four decks, they reported having a feeling which of the
feeling of knowing (Koriat, 1993). decks were good and bad long before they could name
the payoff matrix of the game. This impulsive reaction
The Role of Accessibility to the playing situation was also mirrored by changes
The assignment of a relation to elements is strongly in skin conductance before participants chose a risky
influenced by the accessibility of a category (Higgins, card, and participants frequently based their choice on
Rholes, & Jones, 1977). For example, the belief that a these feelings. Another informational basis for the gen-
particular person is slow depends on the ease with eration of knowledge about others and about the self is
which the concept slow comes to mind, which, in turn, observed behavior. As described by attribution and
depends on the associative strength between the per- self-perception theory (Bem, 1967; Kelley, 1967), peo-
ceptual input and the concept, that is, on the recency ple may infer internal states (such as attitudes) from
and the frequency of their joint activation. Of course, their own behavior or that of others, and from the con-
more elaborate inferential strategies that are the basis text in which it occurs.
of further transformational processing may also be ap- It is important to note that noetic decisions can be
plied to generate propositional knowledge (i.e., causal made not only about what is the case but also about
attribution or hypothesis testing). what is good or bad, because the syllogistic rules also
Because the reflective system uses contents from apply to evaluative judgments, such as attitudes (see
the impulsive system, reflective operations will alter Schwarz & Bohner, 2001); that is, a person may infer
the accessibility of these contents. As a consequence, that something is good or bad based on premises with
subsequent operations in the reflective system will be evaluative content. To arrive at an evaluative decision,
influenced by their predecessors. An example of this it is neither necessary nor sufficient that congruent af-
influence comes from recent work on the psychologi- fect is elicited or that the person has an affective expe-
cal mechanisms underlying the so-called anchoring rience. This also relates to the distinction between in-
bias (Tversky & Kahneman, 1974), which refers to the stant utility and the utility judgments that concern the
assimilation of an absolute judgment toward the stan- past or the future. In particular, Kahneman and collab-
dard of a preceding comparative judgment. In a series orators (for a review see Kahneman, 1999) showed that
of studies, Mussweiler and Strack (see Strack & evaluations of past situations may drastically differ
Mussweiler, 1997; for a review, see Mussweiler & from what was actually experienced. In retrospect, in-
Strack, 1999) demonstrated that this bias results from dividuals seem to rely primarily on the frequency and
the use of information that has been selectively acti- the extremity of negative or positive peaks in experi-
vated during the comparative judgment task preceding enced utility, thereby neglecting the duration of
the absolute judgment. hedonic experience (but see Schreiber & Kahneman,
2000). Judged or remembered utility refers to a noetic
Contents of Representations and judgment, whereas experienced utility accrues from an
Noetic Decisions impulsive reaction to hedonically relevant situations.
To be sure, such experiences may accompany the
The content of propositional representations can reflective processing of evaluative (and factual) infor-
take many different forms. It may be objects from the mation and facilitate or inhibit reflective processes. For
outside world but also experiences that come “from example, a recent set of experiments (Greene,
Sommerville, Nystrom, Darley, & Cohen, 2001) found
3The term noetic was introduced to experimental psychology by that incongruent affect may interfere with evaluative
Tulving (1985), who differentiated between anoetic, noetic, and judgments. Specifically, research participants had to
autonoetic kinds of consciousness. judge whether it was justifiable to sacrifice one human


life to save five. The scenarios were constructed such ing time, intention, and cognitive capacity to extract
that more or less negative affect was elicited by consid- the meaning of a negation (e.g., no money) will the re-
ering the sacrifice. This was checked by fMRI record- flective system be engaged and the task successfully
ings of the brain during decision making. It was found completed. However, if one of these conditions is not
that under the high negative affect condition, more met, the negation may be processed only in the impul-
time was required to make incompatible decisions (i.e., sive system. In this case, the negated concept (e.g.,
justifying the sacrifice) than compatible decisions (i.e., money) and the negating qualifier (e.g., no or not) will
not justifying the sacrifice). When no or less negative become activated in the impulsive system, however
affect was elicited by the decision task, the response they will not be applied to the concept. Consequently,
time did not differ as a function of the outcome of the the immediate effects of processing this information
decision. In congruence with a recent model of moral will be identical to those of processing the same infor-
judgment (Haidt, 2001), this finding suggests that peo- mation in an affirmed format or even without a quali-
ple may use their affect as a basis of such a judgment, fier attached to it.
and that the affect may be in conflict with consider- We are certainly not the first to propose that process-
ations based solely on value and expectancy. ing negations is a resource dependent process. A great
Finally, reflective operations may instigate or mod- number of studies demonstrates that negations slow
erate affect. Although in some cases merely perceiving down cognition (Wason, 1959) and are prone to error
a specific situation is sufficient to activate affect in the (Fiedler, Armbruster, Nickel, Walther, & Asbeck, 1996;
impulsive system (Le Doux, 1995; Zajonc, 1980), in Ross, Lepper, & Hubbard, 1975; Wegner, Coulton, &
other circumstances considerable reasoning is neces- Wenzlaff, 1985). Gilbert’s (1991) sequential model of
sary to infer the affective relevance of a situation (Laz- negating describes how negations may be represented in
arus, 1984). In this case, the reflective–impulsive memory. In particular, the model suggests that all infor-
model assumes that reflection activates affect-specific mation in memory is true as long as it is not “tagged” as
contents in the impulsive system that then lead to the false. This tagging process is assumed to consume cog-
experiential state. However, knowing that something is nitive resources. Empirical evidence consistent with
good (or bad) does not necessarily imply a positive (or this assumption comes from a wide array of research.
negative) feeling. In fact, the insight that a specific ac- For instance, Gilbert, Krull, and Malone (1990) asked
tion might be good can be accompanied by a bad feel- participants to learn a fictitious vocabulary by reading
ing, for example the upcoming appointment with one’s translations (e.g., “a waihas is a fish”). In addition, par-
dentist. On the other hand, positive feelings toward an ticipants were immediately told that the translation was
action may be accompanied by a negative evaluation of either true or false; in some trials, participants were dis-
what one may be about to do (i.e., temptation). tracted as this information was conveyed. In those cases,
false translations were later more often remembered as
Representing What Is Not the Case: being true than true translations were remembered as
Negations and the Future being false. In other words: If participants were dis-
tracted, negated information had similar effects as af-
As is outlined later, the assumption of different rep- firmed information. This phenomenon was replicated in
resentational principles in the reflective and impulsive other experiments using a wide range of materials (e.g.,
system has various implications for current topics in Gilbert, Tafarodi, & Malone, 1993).
social and cognitive psychology. However, it also af- Because these studies typically assessed later judg-
fects capabilities that are fundamental to our thinking ments or recollections, the results may be due to either
and acting. Specifically, the two representational prin- the actual encoding of negations or the later recall of
ciples contribute to a deeper understanding of how we the information. In a series of experiments, Deutsch,
comprehend that something is not the case (i.e., how Gawronski, and Strack (2003) used priming measures
we negate), and how we think about the future. to assess more directly the actual processing of nega-
tions. In these experiments, the effects of evaluative
Negation. Negating—that is, reversing the truth priming using affirmed versus negated words as primes
value of a proposition—is not only a widespread tool were compared to evaluative judgments of the same
of thinking (e.g., Wason & Jones, 1963) but also a con- stimuli, which were elicited under conditions that are
venient means of everyday communication (e.g., Mac- favorable for reflective processing. The results demon-
Donald & Just, 1989). Because negations are proposi- strated that under the priming conditions, effects of af-
tional statements (Horn, 1989; Mall, 1975), boundary firmed and negated stimuli were identical, whereas the
conditions for their successful computation can be de- evaluative judgments were strongly affected by the ne-
rived from the reflective–impulsive model. Most im- gations. This implies that the inefficiency of a negation
portant, Thesis 4 implies that the cognitive procedure cannot be reduced to characteristics of memory.
of negating can only be executed in the reflective sys- In addition, a study by Wegner, Ansfield, and Pilloff
tem. Consequently, only if there is sufficient process- (1998) showed that this dissociation applies not only to


cognitive tasks but also to the self-regulation of behav- generate new predictions that are open to empirical
ior. In that experiment, participants were asked to hold a testing. The main advantage of the reflective–impul-
pendulum and try to prevent it from swinging. Not sur- sive model is that these predictions follow directly
prising, they were better at holding the pendulum still if from its general principles.
they were not distracted by a secondary task. In a second
condition, however, the researchers specifically asked Representing the future. The second important
participants to prevent the pendulum from swinging consequence relates to the ability to represent the fu-
along the x axis that was marked on a piece of paper. ture. Because it is more independent from immediate
Most surprising, in the distraction condition the pendu- perceptual input, only the reflective system can explic-
lum swung more along the x axis when participants were itly generate a time perspective. Specifically, events
explicitly instructed to prevent this from happening that are expected to occur in the future may be catego-
compared to when they were given no such instructions. rized as such. This provides an important means of un-
If it is true that a depletion of cognitive resources in- derstanding courses of action and developments over
creases the likelihood that the impulsive system will time. More important perhaps, it allows for the devel-
control behavior, the instruction not to perform a certain opment of strategies to pursue goals that are remote in
behavior may have strengthened the link to a behavioral time. While the impulsive system is driven by immedi-
schema that produces the undesired behavior. ate perceptual input, the reflective system is able to ab-
Although under specific conditions, the reflec- stract from the immediate input and bridge temporal
tive–impulsive model generates similar predictions as gaps. This allows individuals to resist immediate re-
Gilbert’s (1991) sequential model of negation, the two wards and strive for more valuable future outcomes.
models are not identical. Whereas Gilbert’s model fo- Empirical evidence supporting this notion comes
cuses on the representation of negations in memory, from a recent series of experiments by Deutsch and
the reflective–impulsive model accounts for memory Strack (2002). In particular, they established learning
and retrieval, as well as for the use of retrieved nega- conditions under which people were led to emit impul-
tions in judgmental tasks. Specifically, we assume that sive reactions that were opposed to their reflective
during the encoding of a negation, a link between the knowledge. Participants where asked to open red and
negated concept and the negating qualifier will be es- blue “doors to a photo gallery” on the computer screen.
tablished in the impulsive system. Moreover, we pro- Depending on the color of the door, their actions had
pose that the establishment of such a link in memory different consequences at different points in time. In
and its use in judgments are two distinct processes. In one condition, a particular color was immediately fol-
contrast, it is implied in Gilbert’s (1991) model that lowed by a photo of an extremely negative valence
when a false tag has been added to a piece of informa- from Lang’s collection (International Affective Picture
tion, it will be always correctly recognized as being System, IAPS; Center for the Study of Emotion and
false. From the reflective–impulsive model it follows Attention, 1995), which was presented for 800 ms.
that for participants who are low on cognitive re- Four sec later, an extremely positive picture appeared
sources when they use a piece of information that had for another 4 sec. In the other condition, the valence of
been associated with a negating qualifier the effects of the two photos was reversed.
negated information will be similar to those of af- Asked which photo gallery they would prefer, people
firmed information, despite the fact that the link to the more frequently chose the contingency consisting of the
negation is available in memory. In addition, the fre- immediate but short exposure of the unpleasant and the
quent rehearsal of a negation may strongly increase the delayed but much longer exposure of the pleasant photo.
accessibility of the negated concept, as well as the ne- However, the delayed utility should not determine par-
gation itself. As a consequence, a feeling of familiarity ticipants’ reaction if a response has to be emitted under
may arise when thinking of the negated concept; at the conditions that prevent the reflective system from oper-
same time, the label false may pop into one’s mind. ating. This was the case when participants were in-
This may cause ambiguity, as the experiential cue di- structed to open the doors under the conditions of the
rectly opposes what is implied by the retrieved quali- stop paradigm (Logan, Schachar, & Tannock, 1997).
fier. Finally, the reflective–impulsive model predicts This procedure has been used as a behavioral measure of
that links between concepts and negating qualifiers impulse strength, which is inferred from the ease with
may develop either through reflective elaboration or which a response can be deliberately inhibited. As pre-
through frequent coactivation, which is independent of dicted by the reflective–impulsive model, impulse
reflection. Note again that a false tag is not sufficient strength was determined by the valence of the stimulus
for the use of a negation in the course of social judg- that was immediately associated with the color of the
ments. These predictions from the reflective–impul- door; that is, if the short but immediate valence was posi-
sive model have yet to be tested. tive, participants were less able to inhibit the elicited re-
In sum, the reflective–impulsive model can account sponse than if it was negative. In other words, the stimuli
for known phenomena in the realm of negations and acquired “hot” features (see Metcalfe & Mischel, 1999)


through learning, which influenced behavioral im- Conflicts may arise if behavioral schemata are acti-
pulses and conflicted with the “cool” knowledge about vated that are incompatible and inhibit one another. For
reward contingencies that vary over time. example, through mechanisms of the impulsive sys-
tem, the sight of food may activate an eating schema
Behavioral Control while the reflective system is executing a decision to
stop eating. The resolution of the conflict depends on
A crucial component of the reflective–impulsive the strength of the activation for each schema. If the
model is its assumption that the reflective and impulsive schemata are activated by the different systems, the
systems elicit behavior through different processes. conditions that influence their operation determine
However, the execution of impulsively or reflectively which schema will prevail. For example, the reflective
elicited behaviors is carried out by behavioral schemata system will be more likely to control the behavior if the
that are part of the impulsive system. We elaborate this necessary cognitive capacity is available. Conversely,
assumption by first discussing the development and op- the impulsive system will be more likely to have the
eration of behavioral schemata in the impulsive system. upper hand under a strong deprivation of basic needs or
Then, we describe how behavioral schemata can be acti- under a motivational orientation that facilitates the exe-
vated through impulsive and reflective processes. cution of the behavior. These possibilities will be dis-
cussed in a later section.
Final Common Pathway to Behavior

Thesis 5: Execution of behavior. There exists a Precursors of Behavior
final common pathway to overt behavior in the
impulsive system that may be activated by input Thesis 6: Precursors of behavior. The systems
from the reflective and the impulsive system. use different operations to elicit behavior. In the
This pathway consists of behavioral schemata of reflective system, behavior is the consequence of
varying abstractness. If the schema is activated a decision that is guided by the assessment of a
above a certain threshold, the behavior will be future state in terms of its value and the probabil-
executed. ity of attaining it through this behavior. In the im-
pulsive system, a behavior is elicited through the
In this model, elements in the impulsive system con- spread of activation to behavioral schemata.
sist of sensory, conceptual, affective, and motor repre-
sentations that can be interconnected (see section on
representation and processing). Typically, the following Impulsive precursors of behavior. As behav-
three elements constitute a behavioral sequence: the sit- ioral schemata are part of the impulsive system, they
uational condition, the behavior proper, and the conse- can be easily activated by impulsive processes. In par-
quences of the behavior. The model assumes that similar ticular, perceptual input may activate elements in the
learning principles hold for all types of representations. impulsive system that are associated with behavioral
Thus, associative clusters will emerge in the impulsive schemata or even a part of them. For example, seeing a
system that bind together frequently co-occurring mo- cup will activate a drinking schema. In addition, imagi-
tor representations with their conditions and their con- native input as well as reflectively activated content of
sequences. These sensory-motor clusters are called be- the impulsive system may elicit associated behavioral
havioral schemata, and similar to other contents of the schemata. Thus, thinking about a cup is assumed to ac-
impulsive system, they are subject to spreading activa- tivate a drinking schema.
tion and differ in their activation potential. In addition, if Note that the impulsive precursors of behavior do
one part of a behavioral schema is activated, the activa- not imply knowledge about valence and expectancy.
tion will spread to the remaining elements of the cluster. Although not in the form of simple reflexes, perception
Behavioral schemata and their links to other contents in is linked to behavior in a direct fashion, as described by
the impulsive system can be understood as habits (see the ideo-motor principle (James, 1890). James as-
Aarts & Dijksterhuis, 2000; Ouelette & Wood, 1998). sumed that “every representation of a movement awak-
In concordance with Norman and Shallice (1986), ens in some degree the actual movement which is its
we assume that impulsive and reflective processes can object” (p. 396). This is consistent with the assumption
lead to the activation of behavioral schemata, however that in the impulsive system, conceptual content and
the two systems differ in how they activate a behavioral behavioral schemata are bidirectionally linked. This
schema. This is described in the following paragraph. holds for either conceptual representations of anteced-
An important assumption is that more than one behav- ents or consequences of a behavior, as well as for the
ioral schemata can receive activation at a time. How- behavior itself. Research supporting the idea of a link
ever, activation must exceed a given threshold to result between conceptual and motor contents is reviewed
in overt behavior. elsewhere (Dijksterhuis & Bargh, 2001)


Reflective precursors. In the reflective system, reflective system has generated a behavioral decision to
behavior is the result of reasoning that leads to a noetic refrain from eating it. To win the contest, the reflective
decision about the feasibility and desirability of a par- system can apply knowledge about the mechanisms of
ticular action (cf. Ajzen, 1991; Bandura, 1977). If the the impulsive system. Most effectively, it may divert at-
execution of the behavior is deemed to be feasible and tention from the tempting stimulus.
its outcome positive, a behavioral decision will relate Finally, it is important to note that although both
the self to the behavioral outcome. systems may contribute to the execution of behavior,
Although this reflective procedure seems to guaran- the impulsive system can assume primary control if the
tee a reasoned action as its behavioral consequence, the operating conditions (see Thesis 3) for the reflective
previous thesis points to influences from the impulsive system are not fulfilled. As a result, behavior is less
system that are outside the deliberator’s attention. In likely to be determined by the assessed valence and the
particular, the differential accessibility of information probability of future consequences than by the imme-
about behavioral options or about relevant aspects of diate associations and the resulting hedonic quality.
the alternatives may direct reflective processing; that
is, although a procedure may be rational with respect to Intending
the normative model, it is influenced by contents
whose accessibility is determined by factors unrelated Thesis 7: Intending. In the reflective system, a be-
to rational considerations (e.g., Gregory, Cialdini, & havioral decision is linked to behavioral schemata
Carpenter, 1982). by the process of intending. Intending monitors
However, a behavioral decision may not immedi- the impulsive system for information that enables
ately elicit a goal-directed behavior. Instead, activation the behavioral implementation of the decision.
may spread to several behavioral schemata and con- The mechanism of intending is terminated if the
cepts in the impulsive system. There may be at least behavior is executed or if the goal of the preceding
two reasons why the appropriate behavior may not be behavioral decision is already fulfilled.
executed immediately. First, other behavioral sche-
mata that are incompatible with the behavioral deci- An important factor we must take into consideration
sion may be activated. This issue will be addressed in is that frequently an action cannot be executed at the
the following. Second, the behavioral decision may re- time the behavioral decision is made. In fact, the execu-
fer to a later point in time. This issue is addressed in the tion of many decisions depends on specific conditions
section on intending. that are not yet fulfilled. Thus, there may be a gap be-
tween a behavioral decision and an action. If the execu-
tion of a reflectively chosen behavior depends on the ac-
Synergistic and Antagonistic
tivation of behavioral schemata, constant activation
Interplay of Decisions and Impulses
would be necessary to bridge possible temporal gaps.
Thus far, we have described how impulsive and re- However, such a permanent rehearsal of behavioral pro-
flective processes may jointly activate behavioral grams would absorb a great amount of cognitive capac-
schema that control overt behavior. The effect on be- ity and increase the risk that the behavior is executed be-
havior, however, depends on the compatibility of the fore the appropriate conditions are fulfilled. This,
two forces. Specifically, if the reflective system and the however, does not seem to be adaptive and may even be
impulsive system contribute to an activation of the detrimental with respect to the proper pursuit of the goal.
same schema, the behavior is facilitated. Moreover, the Therefore, we suggest a process of intending as it has
cognitive capacity required to control the execution been discussed in modern theories of human motivation
will be decreased and the execution of the behavior (e.g., Gollwitzer, 1999). We assume that the temporal
may be accompanied by a feeling of fluency, which has gap between a behavioral decision and the execution of
a positive hedonic quality (Winkielman & Cacioppo, an instrumental behavior is bridged by a process that au-
2001). Together with the positive evaluation of the out- tomatically reactivates the behavioral decision and thus
come, the cooperation of the two systems has enor- activates behavioral schemata that are appropriate in the
mous motivational implications. situation. Moreover, we assume that the process of in-
However, the two systems may also compete if they tending is self-terminating; that is, when the goal of the
activate incompatible schemata or if the reflective sys- decision has been reached, the process is turned off.
tem inhibits the execution of a behavior that is impul- We predict that intending also plays a role if a goal
sively activated. Such antagonistic activation may be ac- is blocked. Because of the hierarchical organization of
companied by a feeling of conflict and temptation. For goals, people may check the instrumentality of the ob-
example, a person who is on a diet may be tempted to eat structed goal vis-à-vis a superordinate objective and
a second dessert (see Figure 4); that is, the sight of the choose a different means to the same end. Although the
dessert impulsively activates behavioral schemata that appropriate evidence is still lacking, we expect that
are directed toward consumption. At the same time, the other than through trial and error, the circumvention of


Figure 4. Impulsive and reflective activation of competing behavioral schemata. While the perception of a dessert directly activates an
approach tendency, the noetic decision to lose weight leads to the behavioral decision to go for a walk instead of eating the dessert.

obstacles requires the reflective system to establish al- avoidance. This motivational orientation may be
ternative means–ends relationships and initiate a new elicited by
operation of intending.
• the processing of positive or negative information,
Motivation • the perception of approach or avoidance,
• the experience of positive or negative affect,
So far, we have described the reflective system as a • the execution of approach or avoidance behaviors.
highly flexible system when it comes to action, whereas
the impulsive system appears to be relatively rigid. Spe- In the impulsive system, processing of information
cifically, changing evaluations in the reflective system and the execution of behavior are mediated by two moti-
may result in new decisions and their concomitant be- vational orientations (Cacioppo et al., 1993). In accor-
havioral consequences. In contrast, the impulsive sys- dance with other theorists (e.g., Gray, 1982; Lang, 1995;
tem seems to be driven by the perceptual input as it is Sutton & Davidson, 1997), we assume that these func-
connected to behavioral schemata. Changes of these tional orientations serve to prepare the organism for two
links are assumed to develop slowly following the law of fundamental types of reactions toward the environment:
effect, the law of readiness, and the law of practice approach and avoidance. Approach orientation is a pre-
(Thorndike, 1911). paredness to decrease the distance between the person
However, there are some ways in which the impul- and an aspect of the environment. This includes physical
sive system can also react more flexibly, taking exter- locomotion, instrumental action, consumption, or the
nal and internal conditions into account. For external imagination thereof. Avoidance orientation can be con-
conditions, we propose that the impulsive system can ceptualized as a preparedness to increase the distance
alternate between two distinct motivational orienta- between the person and the environment. This can be
tions that guide the processing of information and the achieved by either moving away from a target (flight) or
activation of behavior. For internal conditions, we pro- by causing the target to be removed (fight). The specific
pose a specific way by which homoeostatic type of response within both motivational orientations is
dysregulations may influence impulsive processing. determined by other influences.
Both motivational aspects are outlined in more detail in Because evidence for the existence of approach and
the following sections. avoidance systems is extensively reviewed elsewhere
(Gray, 1982; Lang, 1995), we focus here on the relation-
Motivational Orientation ship between affect, behavior, and information process-
ing within the two motivational orientations. The fol-
Thesis 8: Motivational orientation: The impul- lowing thesis is grounded in the idea that processing of
sive system can be oriented toward approach and positive information and the experience of positive af-


fect are most important for the regulation of approach serves as support for Thesis 9, which refers to compati-
behavior, whereas the processing of negative informa- bility as a basic principle of the impulsive system.
tion and the experience of negative effect are most im- Then we focus on the reversed direction of influence
portant for the regulation of avoidance behavior. and review studies that illustrate the impact of
evaluative information on behavior. The principles of
Thesis 9: Compatibility. The processing of infor- motivational orientation are represented in Figure 5.
mation, the experience of affect, and the execu-
tion of behavior are facilitated if they are compati- The Compatibility Principle I: The
ble with the prevailing motivational orientation. Direct Impact of Behavior on
Mental Processes
The following propositions can be derived directly Facial feedback. Self-perception theory (Bem,
from Thesis 9: If the impulsive system is oriented to- 1967) is among several applications that have been har-
ward approach, it facilitates the processing of positive nessed to explain a phenomenon in the domain of emo-
information, the experience of positive affect, and the tional expression. Specifically, it has long been argued
execution of approach behavior. In an avoidance mode, (e.g., Darwin, 1872/1965) that facial (and other bodily)
it facilitates the processing of negative information, the expressions serve to not only communicate feelings to
experience of negative affect, and the execution of others and thereby regulate social exchanges but also
avoidance behavior. Moreover, Thesis 9 implies the to increase or diminish the intensity of an affective ex-
principle of bidirectionality, that is, a reverse causal in- perience. Applied to the face, Darwin’s facial-feed-
fluence (cf. Neumann, Förster, & Strack, 2003). Spe- back hypothesis has been studied from a self-percep-
cifically, a motivational orientation may be elicited by tion perspective. Most prominently, Laird (e.g., 1974)
the valence of the processed information, the valence found that experimental participants who had been
of affect, or the orientation of a behavior (approach vs. asked to adopt a smiling expression gave a more posi-
avoidance). tive judgment about themselves (e.g., their own
In the reflective system, a behavior may become the well-being) and about affective stimuli (e.g., cartoons)
basis for inferences about its underlying attitude (Bem, that had been presented to them. According to self-per-
1967). This, however, requires that the behavior is ception theory, participants inferred their affective
propositionally categorized. That is to say, only if the state from their facial expression. Such an inference,
behavior is related to a category (e.g., forgetful) can it however, requires that the behavior be interpreted as
enter into syllogistic inferences. In contrast, process- the expression of a particular affective state; that is, a
ing in the impulsive system and the principle of person can only infer that she must be happy (or
bidirectionality allow a behavior to influence process- amused) if she knows that she is smiling.
ing without being propositionally categorized; that is, From the perspective of the reflective–impulsive
people are influenced by what they are doing even if model, this is not the only way in which an overt be-
the meaning of an action is not recognized. havior may influence mental processes. The inferences
In the following discussion, we first review evi- described by self-perception theory operate according
dence backing the idea that behavior may have a direct to the principles of the reflective system; the impulsive
effect on the processing of information that occurs in system, in contrast, allows for different mechanisms.
the impulsive system and is therefore not mediated by Specifically, it follows from Thesis 9 (compatibility)
syllogistic inferences. At the same time, this evidence that behavior may directly influence information pro-

Figure 5. Principles of motivational orientation.


cessing. In detail, the behavioral schemata that activate pants who were induced to believe they had scored
an emotional facial expression are assumed to be above average in a preceding task and were asked to
linked with evaluatively compatible thought contents adopt an upright body posture while they received this
and perhaps with affective experiences, and this facili- feedback reported feeling prouder than participants
tates the processing of the affective information. who received the same information while adopting a
This process may be mediated either through the slumped posture. This result corroborates and extends
hedonic quality of the emotion or through the ap- the previous facial-feedback findings (Strack et al.,
proach-versus-avoidance function of the behavior. In 1988). It shows, in a different expressive dimension,
the case of the induced smile, we assume that the im- that the impact of posture on mental processes does not
pulsive system facilitates the processing of positive in- hinge on inferences that are based on the perceived
formation. As a consequence, cartoons will be rated as meaning of the bodily action.
funnier and people will feel more amused.
However, to demonstrate that a facial expression Head movements. Although these studies clearly
may influence an emotional experience even if indi- demonstrate the phenomenon, they are not explicit about
viduals do not draw an inference from the behavior, it the exact mechanisms that afford such a direct influence
was necessary to prevent people from recognizing the of a behavior on mental processes. Subsequent studies
emotional meaning of their facial action. This was were more informative. In particular, a series of experi-
achieved by embedding the contraction of the facial ments in which we (Förster & Strack, 1996) explored the
muscle (in this case, the zygomaticus) in a task unre- effect of head movements on the recognition of words
lated to an emotional expression. Specifically, under sheds light on some underlying mechanisms. Participants
the pretext of investigating how people who must do were asked to nod or shake their head while reading posi-
so write or paint with their mouth, Strack, Martin, tive and negative words. To disguise the
and Stepper (1988) asked experimental participants to communicational meaning of the head movements as an
hold a pen either between their teeth or between their act of agreement or disagreement, we had participants
puckered lips while rating several cartoons inter- “test headphones to be used while dancing” (Wells &
spersed among other tasks. Thus, the teeth-holding Petty, 1980). Specifically, participants were required to
position activated the zygomaticus muscle, which is perform either horizontal or vertical head movements
used in smiling; the cover story prevented partici- while the words were played on a cassette recorder.
pants from interpreting their facial action as “a As expected, we found that the head movements af-
smile.” Nevertheless, people assigned to this experi- fected performance in a surprise recognition task. In
mental condition reported feeling more amused and particular, participants who had been induced to nod
rated the cartoons as funnier than people who held were better at recognizing positive words, while partic-
the pen with their puckered lips. ipants who had been induced to shake their head were
These findings (see also, Bodenhausen, Kramer, & better at recognizing negative words. Moreover, this
Suesser, 1994; Erber, 1991; Martin, Harlow, & Strack, proved not to be a response bias that affects the thresh-
1992; Zajonc, Murphy, & Inglehart, 1989) show that old for words of a particular valence. Rather, the data
the effect of facial feedback does not depend on infer- showed that when the head movement was compatible
ences from the perceived emotional meaning of a facial with the valence of the word, people were better at dis-
expression. Rather, facial feedback may also affect criminating whether the word had been presented. This
mental processes in a more direct fashion, which can suggests that the behavior influenced the processing of
be explained easily and parsimoniously by the mecha- the words at the time of the encoding.
nisms that the reflective–impulsive model ascribes to This encoding hypothesis was tested more directly
the impulsive system. in a second experiment, using a dual-task paradigm.
More specifically, while participants were explicitly
Postural feedback. Extending the same logic to instructed to learn the words while nodding or shak-
another medium of expressing emotions, Stepper and ing their head, they were also required to perform a
Strack (1993) induced experimental participants to manual dexterity task. If head movement facilitates
adopt an upright or slumped posture under the pretext (or inhibits) the encoding of information, more (or
of studying different working conditions. Based on less) cognitive capacity will be left to perform the
previous studies (Riskind, 1984), Stepper and Strack secondary task. As a consequence, we predicted a
expected that people would experience the emotion of better manual dexterity performance if the learning of
pride more intensely when adopting an upright body positive words was accompanied by nodding, and the
position. Again, those earlier studies did not rule out learning of negative words by shaking one’s head.
that people drew an inference from the quality of their Conversely, the outcome of the dexterity task should
posture about the quality of their emotional state. In the be poorer if positive words are learned while shaking
Stepper and Strack study, however, the posture was dis- one’s head, and negative words while nodding. The
guised as an ergonomic investigation. Still, partici- results supported these predictions. Moreover, the in-


clusion of neutral words showed that facilitation and whereas negative words were categorized more rapidly
inhibition occurred as a function of valence-motor when participants had to contract the extensor muscle.
compatibility or incompatibility. Finally, Seibt and Neumann (2002) applied the
These findings provide evidence for motor influ- flexor/tensor procedure to the assessment of cartoons
ences on mental processing when the behavior has no and found that participants reported feeling more
immediate evaluative implications. However, these im- amused if the flexor muscle was contracted than if the
plications are mediated by a motivational subsystem of tensor or no muscle was activated. Moreover, Seibt and
which the behavior is a part. In most cultures, nodding Neumann found that reported social anxiety (elicited
is a nonverbal signal for agreement and shaking one’s by the expectation of having to speak in front of a large
head for disagreement; these head movements are audience) was increased if participants had been in-
therefore linked to an orientation toward approach or duced to contract the tensor muscle.
avoidance that influences the way behavior is impul-
sively regulated. Tuning the impulsive system to an Direct activation of a motivational orientation.
orientation of approach facilitates the processing of In the reported studies, the link between a behavior and
positive information and inhibits the processing of the processing of evaluative information was assumed
negative information. Conversely, tuning the system to be strengthened through the prevailing motivational
toward avoidance facilitates the processing of negative orientation. To repeat, the impulsive system may be
information and inhibits the processing of positive in- oriented toward approach or avoidance (cf. also Lang,
formation. Thus, head movements and the valence of 1995), and this orientation may be triggered by the va-
words are either compatible or incompatible and facili- lence of the processed information, by the experience
tate or inhibit the processing of information. of compatible affect, and by the perception of move-
ments that imply approach or avoidance. An activated
Approach and avoidance through isometric motivational orientation results in the lowering of
muscle contractions. Similar effects were obtained thresholds for the processing of compatible informa-
for a motor action that was previously found to influ- tion (i.e., positive information for the approach and
ence positive and negative affect. In particular, negative information for the avoidance orientation),
Cacioppo and his collaborators (e.g., Cacioppo et al., and for the elicitation of compatible behaviors that
1993) discovered that pressing the palm of one’s hand stand in a functional relationship with the orientation.
from the bottom against the surface of a table, thereby One way of activating a motivational orientation is to
activating the flexor muscle of the arm, led to positive put people in a situation in which the distance between
attitudinal judgments, while pressing from the top the person and an object appears to be increasing or de-
against a table platform, thereby activating the creasing; that is, if the distance is decreasing, the impul-
extensor muscle, led to negative attitudinal judgments. sive system should be oriented toward approach; if the
Cacioppo et al. (1993) argued that through previous as- distance is increasing, the impulsive system should be
sociations, flexor and extensor actions elicit a motiva- oriented toward avoidance. As a result, the processing of
tional orientation of approach versus avoidance. compatible information should be facilitated.
The reflective–impulsive model integrates and ex- In a very direct test of this notion, Neumann and
tends Cacioppo et al.’s (1993) notion of a motivational Strack (2000) gave participants an evaluative decision
orientation that can be activated by any behavior com- task in which the valence of positive and negative
patible with approach or avoidance. It should therefore words presented on a computer screen had to be cate-
be possible to apply the described influence of head gorized. More important, these words were presented
movements to reproduction memory (Förster & Strack, in the center of circles that were either increasing or de-
1996) using the flexor-extensor paradigm. This was in creasing in size. As documented in a manipulation
fact achieved by Förster and Strack (1997) for the repro- check, the changing size of the circles created the ap-
duction of famous names; that is, contracting the flexor pearance that the words were moving either toward the
muscle facilitated the recall of positively connotated person or away from the person, although the actual
names, whereas contracting the extensor muscle im- size of the letters remained constant. As predicted by
proved the recall of negatively connotated names. the reflective–impulsive model, this apparent change
The postulated principle of compatibility and in the distance between the target and the person influ-
bidirectionality found additional support in a study by enced the evaluative decision such that the valence of
Neumann and Strack (2000), who used a different de- positive words was categorized faster if the word ap-
pendent variable. In particular, participants who were peared to be moving toward the viewer, whereas the
presented words on a computer screen had to decide if valence of negative words was categorized faster if the
these words were positive or negative in valence. As words appeared to be moving away.
predicted by the reflective–impulsive model, we found To verify that this finding does not depend on the
that positive words were categorized more rapidly evaluative nature of the task, we used the same para-
when participants had been induced to flex their arm, digm for a lexical decision task in which the stimulus


material consisted of not only positive and negative To that end, Neumann and Hess (2001) conducted a
words but also neutral words and nonwords. Partici- second experiment in which participants had to merely
pants were asked to decide whether a letter string on indicate when a stimulus appeared on a computer
the screen was a word. Again, the response latencies screen. The stimuli were positive or negative words,
were as predicted: Lexical decisions for positive words and the reactions had to be provided by either contract-
were faster if the letter strings appeared to be moving ing the corrugator or the zygomaticus. Again, the com-
toward the person, while lexical decisions for negative patibility between the valence of the stimuli and the na-
words were faster if the letter strings appeared to be ture of the motor action determined the response
moving away. This suggests that the predicted facilita- latencies. Participants indicated the appearance of pos-
tion effect is not a function of the evaluative nature of itive words faster when using the zygomaticus muscle,
the task but depends solely on the valence of the stimuli and the appearance of negative words faster when us-
(see also, Chen & Bargh, 1999). ing the corrugator muscle.
These findings provide clear evidence for the
The Compatibility Principle II: The bidirectionality postulated by the reflective–impulsive
Impact of Evaluative Information model. In particular, it was shown that facial action pro-
on the Execution of Behavior vides not only feedback (Laird, 1974) by influencing the
processing of evaluative information but also that its ex-
On several occasions, we have emphasized ecution is controlled by the valence of information.
bidirectionality as a signature characteristic of the im-
pulsive system. This implies that a behavior may acti-
vate a motivational orientation, which then lowers the Approach and avoidance through isotonic move-
threshold for the processing of compatible informa- ments. In previous studies that were based on a pro-
tion. However this also implies, conversely, that the va- cedure by Cacioppo and associates (e.g., Cacioppo et
lence of information may exert a facilitating influence al., 1993), we were able to show that the processing of
on compatible behaviors that is mediated by the moti- positive versus negative information was facilitated de-
vational orientation. Concretely, it implies that the be- pending on whether the flexor or the extensor muscle
haviors that served as independent variables in the ex- was contracted (Förster & Strack, 1997, 1998;
periments described in the preceding paragraphs Neumann & Strack, 2000). This effect was expected
should also operate as dependent variables. The avail- because the contractions are associated with move-
able evidence supports this implication. ments of approach and avoidance. The postulate of
bidirectionality predicts the opposite causal direction.
Facial action. As described in a previous sec- In particular, movements that are directed toward the
tion, unobtrusively manipulated facial expressions person should be facilitated if positive information is
were found to influence evaluative judgments (Strack processed, whereas movements away from the person
et al., 1988). Conversely, we expected that facial ac- should be easier if the information is negative.
tions would be facilitated if compatible information is First evidence for this influence came from a study
processed. This prediction was confirmed in a still un- by Solarz (1960), who presented cards with words that
published set of experiments by Neumann and Hess were either positive or negative. Depending on the
(2001), in which participants had to respond by con- words, participants had to either push these cards away
tracting their facial muscles. In particular, under the from themselves (avoidance) or pull them toward
pretext of studying whether responses to psychological themselves (approach). Solarz found that participants
tasks emitted by body parts that are closer to the brain were faster if the valence of the word and the move-
differed from responses produced by more remote ment were compatible; that is, faster reactions were ob-
body parts (such as the hand), participants had to con- served if positive words had to be pulled toward the
tract the corrugator or the zygomaticus muscle. Sur- person and negative words had to be pushed away.
face electrodes recorded the EMG activity of both This finding was replicated by Chen and Bargh
muscles while participants assessed the valence of (1999), who had participants evaluate words on a com-
words presented on a computer screen. As predicted, puter screen as good or bad by either pushing or pulling
the positivity of words was indicated faster with the a lever. In line with Solarz’ (1960) findings, lower re-
help of the zygomaticus (smiling) muscle, while the sponse latencies were obtained if the valence of posi-
negativity of words was assessed faster if participants tive words had to be indicated by pulling the lever or
were using the corrugator (frowning) muscle. As in the valence of negative words by pushing the lever. In a
previous studies, it was important to find out whether second study, Chen and Bargh (1999) showed that this
the effect depended on the evaluative nature of the task, compatibility effect did not depend on the evaluative
or whether the evaluative nature of the stimulus influ- nature of the task. Rather, it was also obtained if the
enced the processing even for tasks with no evaluative task was simply to indicate whether a word (that was
implications (see Neumann & Strack, 2000). either positive or negative) appeared on the screen.


Head movements. Vertical or horizontal move- those behavioral schemata that in the past fre-
ments of one’s head influence the encoding of negative quently led to satisfaction of those needs.
or positive information (Förster & Strack, 1996). The
reverse, however, is also the case. Wells and Petty Motivational orientations only moderate the execu-
(1980) found that people nodded their head more often tion of behavior and the processing of information based
if they agreed with the content of an attitudinal mes- on valence. Although this is true for a wide range of be-
sages they heard through headphones. Similarly, haviors, some situations may call for a more specific dis-
Förster and Strack (1996) observed that when partici- position to act. In particular, if basic needs are not satis-
pants had to nod while encoding the words, positive fied, specific behaviors are necessary to remedy this
words increased the rate of nodding. Conversely, if state of deficiency. To account for the influence of basic
people had to shake their head, negative words in- needs within the reflective–impulsive model, we invoke
creased the rate of shaking. a mechanism resembling the notions proposed in early
drive theory (Hull, 1943). In particular, we propose that
Affect as a determinant of motivational orienta- whenever a state of deprivation is successfully abol-
tion. The distinction between two motivational ori- ished, behaviors and situational circumstances that led
entations is closely related to a model proposed by to satisfying the need become strongly linked to the ex-
Lang and his associates (for a review, see Lang, perience of this deprivation (cf. Dickinson & Balleine,
Bradley, & Cuthbert, 1990), who postulated appetitive 2002). Consequently, if the same need is deprived again,
and aversive motivational systems that facilitate be- the behavioral schemata and conceptual contents that
havioral responses. Based on biological and were previously related to the satisfaction will again be
neuropsychological evidence, these researchers as- activated and produce a behavioral preparedness and a
sume that “associations, representations, and action perceptual readiness for the processing of relevant in-
programs that are linked to the engaged motivational formation in the environment (Bruner, 1957). For exam-
system have a higher probability of success” (Lang, ple, the deprivation of food will facilitate behavioral
1995, p. 377). Moreover, these motivational systems schemata of food intake and spread activation to con-
are activated by an organism’s emotional or affective ceptual representations of food. As a result, stimuli that
state. An extensive program of research with animals are related to food should be recognized more easily un-
and humans found that startle responses were in- der conditions of food deprivation.
creased when negative affect had been elicited. This This was found in an experiment conducted by Wispe
was particularly the case for the emotion of fear. At the and Drambarean (1953), who asked participants to rec-
same time, the startle response was found to be inhib- ognize words that were presented very briefly on the
ited if the organism was in the state of positive affect. computer screen. Some of their participants had not
These findings support the reflective–impulsive eaten for 10 or 24 hours, whereas others had had a snack
model in its assumption of a motivational orientation just prior to the experiment. Moreover, the words were
that links mental and behavioral processes. At the same either food related or not. As expected, hungry partici-
time, the dynamics of the impulsive system and the pants were faster at detecting food-related words than
role of the motivational orientation go beyond the neutral words. However, no such difference was ob-
mechanisms identified by Lang and his associates served in participants who were not hungry. Similar re-
(e.g., Lang, 1995). For example, the reflective–impul- sults were recently obtained by Aarts, Dijksterhuis, and
sive model treats behavior as a dependent and an inde- De Vries (2001) for thirst and drinking-related words.
pendent variable. As previously described, a particular Support for our notion that people’s attention toward
behavior may elicit a motivational orientation, which stimuli that are capable of reducing the deprivation is in-
may then facilitate the processing of compatible infor- creased by the behavioral schemata is provided by a
mation. In the reflective–impulsive model, the motiva- study by Lavy and van den Hout (1993). Based on fast-
tional orientation functions by selectively lowering the ing participants’ ratings of food valence and urge to eat,
threshold for the processing of evaluatively compatible Lavy and van den Hout concluded that attentional bias in
information and for the execution of functionally com- a Stroop task was more likely to come from an increased
patible behaviors. At the same time, the motivational urge to act than from deprived people’s emotional expe-
systems theory provides important links to underlying rience. The increased accessibility of need-relevant
biological substrates, particularly the role of the concepts may also guide the interpretation of ambigu-
amygdala. ous stimuli. For example, D. C. McClelland and
Atkinson (1948) demonstrated that hungry participants
were more likely than satiated participants to identify
Homoeostatic Dysregulation ambiguous stimuli as food-related objects.
Homoeostatic dysregulations may also modulate
Thesis 10: Homoeostatic Dysregulation. Depri- evaluative responses. As Lewin (1935) noted, the va-
vation of basic needs will lead to an activation of lence of an object “usually derives from the fact that


the object is a means to the satisfaction of a need” (p. creasing the distance (i.e., avoidance). Whether this is
78). Existing evidence shows that need-related objects accomplished by moving away from the target (i.e.,
are experienced to be more positive under deprivation flight) or by causing the target to be removed (i.e., fight)
(e.g., Drobes et al., 2001; Lavy & van den Hout, 1993). depends on determinants in the situation and in the per-
Moreover, it has been demonstrated that need-irrele- son (e.g., Keltner, Grunfeld, & Anderson, 2003), or the
vant objects are devaluated (Brendl, Markman, & level of activation. The experienced emotions of fear or
Messner, 2003). That such shifts of valence occur im- anger are the categories that are used to represent the en-
pulsively was recently demonstrated by Seibt, Häfner, tire emotional episode in the reflective system.
and Deutsch (2003) who found in two experiments that From this conceptualization it follows that affect
evaluative associations related to food as well as spon- may influence behavior in at least two ways. First, core
taneous orientations toward approach versus avoid- affect possesses a motivational orientation toward ap-
ance depended on food deprivation. proach and avoidance that facilitates concomitant be-
It is important to note that the link between depriva- haviors. Second, affective experiences may be proposi-
tion and the facilitation of behavioral schemata and the tionally categorized and become the basis of syllogistic
activation of conceptual contents is believed to develop inferences that may lead to noetic and behavioral deci-
only for needs that are immediately experienced. Other sions. Both possibilities are discussed in this article. For
dysregulations (such as that of blood pressure) are not a more detailed integration of affect and emotion into
experienced. Thus, the described behavioral and con- dual-system notions, see Smith and Neumann (in press).
ceptual mechanisms will not be elicited.
Affect and Emotion
During the past 15 years, automaticity has become
Affect and emotion are understood as products of the an increasingly important topic in social psychology
reflective and the impulsive system. Following Rus- (e.g., Bargh, 1990; Bargh & Ferguson, 2000). Even
sell’s (2003) theory of emotion, our model implies that such complex social-cognitive processes as goal acti-
the impulsive system generates a simply structured state vation (e.g., Bargh & Barndollar, 1996), trait infer-
of core affect that, by reflective processes, can be trans- ences (e.g., Gilbert, 1989), or the imitation of social be-
formed in more elaborate feelings and emotions. In par- havior (e.g., Dijksterhuis, Bargh, & Miedema, 2000)
ticular, we assume that a person’s core affect is a state have been shown to operate outside consciousness,
that can be experienced on the dimensions of hedonic without intention, to be hard to control as well as
quality and arousal. Thus, a person may feel good or feel highly efficient (cf. Bargh, 1996). It is a widely held as-
bad in a way that is accompanied by high or low activa- sumption that automaticity of cognitive procedures is
tion. Because the origin of this state is not part of the ex- achieved through frequent execution (e.g., Anderson,
perience, situationally and chronically accessible con- 1981). As Bargh (1997) put it: “Any skill, be it percep-
tents may be reflectively assigned as causes. tual, motor, or cognitive, requires less and less con-
As a consequence, a noetic awareness emerges scious attention the more frequently and consistently it
along with the experienced affect. For example, people is engaged” (p. 28).
may feel good (or bad) about themselves. At the same How can automaticity be understood from the van-
time, they may engage in behaviors that are directed by tage point of the reflective–impulsive model? Of course,
the valence of the affective quality and fueled by the it is the impulsive system to which the central criteria of
state of activation. Based on emotional schemata, the automaticity (see Bargh, 1997) apply. Particularly its in-
particular configuration of experience and knowledge dependence from intention and its high efficiency fol-
may then be categorized as a particular emotion. Then, low from its architecture as an associative network. In
a person may, for example, feel proud about a particu- contrast, the reflective system is assumed to depend
lar achievement (e.g., Neumann, Seibt, & Strack, heavily on cognitive resources and to generate inten-
2001) or disappointed at a friend’s betrayal. tions through behavioral decisions. However, according
Thus, the reflective–impulsive model does not claim to the reflective–impulsive model, the quality of the im-
that specific emotions produce specific behaviors. pulsive mechanisms differs from that of reflective pro-
Rather, it maintains that behavioral schemata are impul- cesses, which is why a direct transposition of reflective
sively influenced by habit strength, motivational orien- procedures to the impulsive system is not possible.
tation and homeostatic dysregulation. In addition, re- This raises the question of how controlled processes
flective considerations may enter into the equation. For can become more automatic? The reflective–impulsive
example, we do not assume that fighting is a necessary model implies several mechanisms that play a role (cf.
behavioral consequence of anger whereas fleeing is a Smith, 1994; Smith, Branscombe, & Borman, 1988).
necessary result of fear. Instead, and in agreement with The first and most simple mechanism is that contents
Russell’s (2003) model, we propose that the evaluative from the associative store of the impulsive system are
quality of the information orients the system toward in- used and thus activated during reflective processing. If


information is transformed through a syllogistic proce- Application to Phenomena in Social
dure (e.g., to infer the valence of a negated word), the Psychology
input to the procedure (e.g., the negated word) as well
as its output (e.g., the resultant valence) will be acti- In the remainder of this article, we discuss the im-
vated in the impulsive system. If this is repeated fre- plications of the reflective–impulsive model (see Fig-
quently, the joint activation of input and output will es- ure 6 for the full model) for various aspects of social
tablish an associative link between the two contents. psychology and illuminate its potential for explaining
Thus, the results of previous syllogistic procedures are different phenomena of social behavior. Special em-
stored in memory and eventually can be retrieved in an phasis is given to predictions where this model will go
automatic fashion (cf. Logan, 1988). beyond or differ from existing dual-process models.
This memory-based mechanism predicts only a nar-
row transfer of automatic responses to new instances or
Implicitness and Explicitness in
situations. In particular, only the reaction to those or
Attitudes, Prejudice, and Stereotyping
similar (cf. Palmeri, 1997) instances that were prac-
ticed can be elicited in an automatic fashion. However, In the past decade, research in social psychology has
the reflective–impulsive model can also account for increasingly focused on so-called implicit phenomena
more general effects of practice (see also, Anderson, (e.g., Farnham, Greenwald, & Banaji, 1999). In particu-
1993). First, the operational principles of the reflec- lar, central concepts such as stereotypes, prejudice, atti-
tive–impulsive model imply that the frequent execu- tudes, or goals were assumed to operate not only as con-
tion of reflective procedures makes the representation tents of consciousness but also outside of conscious
of these procedures more accessible for further use. awareness (e.g., Banaji, 2001; Blair, 2001). To study
Second, representations of procedures may become as- such implicit phenomena, new methods of research
sociated with situations in which they were frequently have been developed (e.g., Maass, Castelli, & Arcuri,
carried out. Finally, syllogistic procedures may be- 2000), such as the implicit association test (Greenwald,
come associated with other procedures if they are fre- McGhee, & Schwartz, 1998), evaluative priming
quently carried out in temporal proximity. Conse- (Fazio, Sanbonmatsu, Powell, & Kardes, 1986), or af-
quently, complex syllogistic operations may become fective variants of the Stroop and Simon task (e.g., De
more efficient through practice, because the more spe- Houwer, Crombez, & Baeyens, 2001). These measures
cific procedures they consist of can be quickly re- have in common that they do not rely on conscious eval-
trieved from memory. uations of the attitude object itself, thereby circumvent-
These assumptions set limits for automatic process- ing correctional processes because of social desirability
ing under specifiable circumstances. An example is the or other reasons. Instead, they tap into mechanisms that
well-practiced cognitive operation of negating. Evi- are automatically instigated by the attitude object.
dence (Deutsch et al., 2003) suggests that through the There is no doubt that the reflective–impulsive
frequent execution of this operation, semantic links model bears a family resemblance to explicit–implicit
will be created that facilitate further processing. For models in that both notions hold that different mecha-
example, the term no way has acquired its own mean- nisms may mediate between valence and behavior.
ing whereas the term no hay needs further reflective However, there are several issues on which the reflec-
processing to be understood, independent of the fre- tive–impulsive model takes a divergent position. First,
quency with which the former negation is used. It is we suggest that the two systems operate in parallel and
therefore the particular content that profits from fre- interact with one another. In contrast, most ex-
quent practice, but not the generalized application of plicit–implicit models assume a sequential mode of
the rule. In this model, the frequent execution of reflec- operation (see Smith & DeCoster, 2000). Second, the
tive mechanisms may form associations and allows the partition of the reflective–impulsive model into two
impulsive system to “take over.” However, this does systems is not based on the presence or absence of con-
not mean that the reflective procedure of negating will scious awareness (e.g., Greenwald & Banaji, 1995).
be performed automatically; rather, it will be substi- We refrain from using the phenomenal experience that
tuted by a different mechanism that fulfills the criteria may accompany mental processes as a distinctive crite-
of automaticity. rion because we do not know precisely how conscious-
Although the idea that the same psychological proce- ness arises from psychological or neural processes. In
dure can be transposed from one operational mode to an- particular, consciousness is often seen as an
other has a venerable history in psychology and goes epiphenomenon rather than a causal force or even an
back to Helmholtz’s (1867) notion of unconscious infer- integral part of cognition (Libet, Gleason, Wright, &
ences, we second Lieberman et al. (2002) in that the idea Pearl, 1983; Wegner, 2002). Moreover, categorizing a
of automatic processes as merely faster and quieter ver- process as conscious or unconscious provides little in-
sions of controlled processes “is theoretically parsimo- formation about its computational nature. Therefore,
nious, intuitively compelling, and wrong” (p. 205). the reflective–impulsive model does not invoke the ex-


Figure 6. Overview of the complete reflective–impulsive model.

istence of implicit attitudes, implicit stereotypes, im- attitude (or whatever other concept; Fazio & Olson,
plicit goals, or implicit self-esteem. 2003). From the perspective of the reflective–impul-
Some theorists in the field of implicit social cogni- sive model, explicit and implicit measures are defined
tion have begun to elaborate on more unique features by the cognitive operations that they capture. In this
of implicit phenomena. For instance, Wilson et al. sense, explicit measures tap into people’s knowledge
(2000) advanced the thesis that implicit attitudes may or beliefs, implicit measures tap into their associative
change more slowly than explicit ones and tend to in- structures.
fluence expressive and automatic behaviors. Green- Beyond redefining the implicit versus explicit di-
wald et al. (2002) specified implicit associative mecha- chotomy, the reflective–impulsive model may help to
nisms that incorporate principles of cognitive balance. understand how behaviors may be influenced by ex-
We would like to go one step further and propose to plicit and implicit mechanisms. Although it has been
ground the implicit–explicit distinction primarily on proposed that the implicit and explicit phenomena may
operational characteristics. In line with other dual-pro- take different routes to behavior (e.g., Dovidio,
cess models (e.g., Smith & DeCoster, 2000), we locate Kawakami, Johnson, Johnson, & Howard, 1997; Wil-
implicit processes in the impulsive system, whereas son et al., 2000), most models are relatively mute on the
explicit processes are thought to take place in the re- cognitive or motivational structures underlying such in-
flective system. Specifically, we prefer to use the terms fluences. The reflective–impulsive model can account
explicit versus implicit for psychological processes but for such influences through its inherent interconnec-
not for mental contents. As a consequence, an attitude tions between conceptual and behavioral representa-
is defined as a belief following from an evaluative deci- tions as well as through its motivational orientations.
sion that follows from reflection about what is good or
bad. At the same time, we talk about evaluative associ-
Automatic Attitude Activation
ations to describe links between concepts and
evaluative responses in the impulsive system. In a simi- Many studies (see Fazio, 2001) have demonstrated
lar vein, stereotypic associations are distinguished that attitudes may be activated very quickly, efficiently,
from stereotypic beliefs (Devine, 1989). unintentionally, or even unconsciously on the percep-
This conceptualization affects the interpretation of tion of the attitude object. In his influential motivation
implicit measures. Despite their popularity, the psy- and opportunity as determinants model (MODE
chological status of what they assess is still unclear (for model), Fazio (1990) specified the circumstances un-
a review, see Fazio & Olson, 2003). Some theorists ar- der which attitudes may automatically influence be-
gue that implicit measures identify implicit attitudes or havior. In particular, Fazio assumed that only strongly
implicit stereotypes (e.g., Wilson et al., 2000); others associated attitudes will exert automatic influences.
prefer to view the measure itself as implicit, but not the The main process through which attitudes may influ-


ence behavior spontaneously is by influencing the per- judgments are generated in ways that are based on vari-
ception of the situation. If motivation and cognitive re- ous simplified associative procedures that afford
sources are high, however, more deliberate decisions under suboptimal conditions.
considerations and effortful search in memory may At first glance, the reflective–impulsive model may
prevail and determine the perception of the situation seem to mirror this distinction. This, however, is not
and finally the behavior. the case, because there are no reflective versus impul-
Although the mechanisms of the reflective–impul- sive ways to social judgments that people can choose
sive model are consistent with the tenet that motivation as alternatives. Rather, we assume that judgments and
and opportunity (for thorough processing) are precon- decisions are exclusively made by the reflective sys-
ditions for reflective choice, the model proposes a tem, whereas the impulsive system operates in parallel
somewhat different mechanism for nondeliberative (see Thesis 3). How then can the reflective–impulsive
processing. Both models assume that objects can be as- model account for intuitive or heuristic judgments?
sociated with a positive or negative valence. The The answer rests on a more detailed analysis of
MODE model suggests that the activation of such posi- what we mean by judgments that are not generated in a
tive and negative associations may tune the perception systematic fashion. In general, there seems to be agree-
of the situation, which then influences behavior in a ment that they consist of shortcuts to circumvent
spontaneous fashion. Although the reflective–impul- effortful and time-consuming systematic processing.
sive model includes this mechanism as a possibility, it From the perspective of our model, there are three
also affords an alternative option that does not require types of shortcuts driven by noetic, experiential, or be-
the operation of the reflective system. Specifically, the havioral processes.
reflective–impulsive model proposes that a behavior Noetic shortcuts are conceived as simplifying infer-
may be influenced by its facilitation through a motiva- ences based on characteristics of the target. One exam-
tional orientation that is activated by the valence of the ple is the belief that a long, persuasive message is more
processed contents (see Thesis 9). Finally, the impul- valid than a short message (Eagly & Chaiken, 1993).
sive system provides for the possibility that perception Thus, the general principle length implies strength is
is directly linked to behavioral schemata, allowing for used as a premise in an inference (see Kruglanski &
an even more direct path to behavior than the MODE Thompson, 1999) that allows a person to make a deci-
model (see Thesis 5). sion without tediously scrutinizing each argument. In a
Thus, from the perspective of the reflective–impul- related fashion, the rule of similarity implies category
sive model, the behavioral component of an attitude membership may simplify propositional categoriza-
has a reflective and an impulsive meaning: reflective in tions that would otherwise require complex computa-
that it refers to a behavioral decision that is derived tions using base-rate probabilities (see Tversky &
from an evaluation, and impulsive in that it refers to ac- Kahneman, 1982).
tion tendencies that are directly associated with the At the same time, there are experiential shortcuts,
evaluative features of the attitude object. which are conceived of as simplifying inferences based
on subjective experiences. For example, the mental ef-
fort that is experienced while trying to generate a judg-
Intuitive and Heuristic Judgments
ment may become the basis for an inference about the
As mentioned earlier, theorizing in social psychol- target. The probability of an event’s occurrence
ogy has been enriched by dual-process models that dis- (Tversky & Kahneman, 1973) and the fame of a person
tinguish between rule-based and associative process- (Jacoby, Kelley, Brown, & Jasechko, 1989; see also
ing (Smith & DeCoster, 2000). At the same time, it has Strack & Neumann, 2000) seem to be determined by
been recognized that judgments may be generated in the experienced mental effort. Similarly, affective ex-
ways that differ in the ease with which they can be per- periences (such as current mood) may become the ba-
formed. In particular, it has been argued that judges sis for inferences about one’s own global well-being
may use mental shortcuts to save time and effort. Most (Schwarz & Clore, 1983) or for judgments about con-
prominent, Tversky and Kahneman (1974) identified a sumer goods (Wänke, Bohner, & Jurkowitsch, 1997).
set of judgmental heuristics that serve this purpose. Noetic and experiential components are combined in
Some dual-process theorists (e.g., Eagly & Chaiken, a third shortcut that immediately follows from a mecha-
1993) have combined the two research programs and nism described by the reflective–impulsive model: A
proposed that heuristic and nonheuristic judgments are judgment may be based on a behavioral tendency or a
generated by qualitatively distinct processes (for an al- motivational orientation that is produced by the impul-
ternative view, see Kruglanski & Thompson, 1999; sive system. Extending the logic of self-perception the-
Strack, 1999). First, there is a systematic mode in ory (Bem, 1967), the noetic and experiential representa-
which judgments are formed following rational or logi- tion of one’s behavioral tendencies may enter the
cal principles of the rule-based type of processing. Sec- reflective system before they are actually carried out.
ond, there is an intuitive or heuristic mode in which Thus, a person noting that he or she is driven to approach


a given target may translate this experience into knowl- impulsive or in the reflective system. As was outlined
edge and use it for inferences about the target. before, we believe that the impulsive system adheres to
According to the model we present, the impulsive the hedonic principle (i.e., positive stimulation facili-
system plays a passive role in heuristic and intuitive tates approach, negative stimulation facilitates avoid-
judgments. In particular, its operation may provide an ance), and that it is incapable of extracting the meaning
experiential awareness and the associative contents to of negations. Consequently, the impulsive system can-
which mental shortcuts are applied. However, the ap- not represent nongains or nonlosses in terms of propo-
plication of these simplifying rules of reasoning takes sitions, but only as associations. In particular, a person
place in the reflective system. This is indicated particu- may experience a situation in which frequent rewards
larly by the fact that judges may correct for a lack of are delivered. For instance, a very talented student may
representativeness in the noetic or experiential basis of be frequently praised by all teachers. At the same time,
the judgment. The correctional mechanism in situa- some external cues may signal the omission of reward,
tions where affect is used as information has been re- that is nongain. For instance, the student may learn that
peatedly demonstrated and discussed by Schwarz and she is frequently rewarded, except when she is in math
Clore (e.g., 1996; for a more general perspective, see class, because her math teacher does not like her and
Strack, 1992; Strack & Hannover, 1996). praises everyone but her. In this situation, the opera-
Thus, although it may seem tempting to connect tional principles of the impulsive system predict that
nonsystematic judgments to specialized intuitive or the math teacher will become associated with frustra-
heuristic processes, the explanation provided by the re- tion (cf. Higgins, Shah, & Friedman, 1997) and will
flective–impulsive model is quite different. In particu- serve as a signal of negative stimulation that is, in turn,
lar, the model suggests that although judgments and de- expected to instigate an avoidance orientation. Analo-
cisions are made in the reflective system, it is important gously, signals of nonpunishment are predicted to be-
to understand its interactions with the impulsive system. come associatively linked to positive affect and thus an
This is particularly obvious in the case of experiential approach orientation.
shortcuts where one must ask under what conditions and If information about nongains and nonlosses is con-
at what points in the judgmental sequence subjective ex- veyed verbally, however, a different pattern is pre-
perience may enter into the reflective system. dicted to occur. In this case, the valence of the event
must be inferred reflectively from negated statements,
whereas in the case of conditioning, the valence is al-
Regulatory Focus
ready stored in memory. At the same time, because
One of the most influential theories in recent moti- contents of reflective operations are assumed to be re-
vational science is regulatory focus theory (RFT; Hig- trieved from the impulsive system, the negated con-
gins, 1997), which predicts how strategies of approach cepts will receive activation, thereby instigating a con-
and avoidance result from positive and negative events gruent motivational orientation. Take the situation of a
that are experiences or expected. This topic has been student, hearing the sentence “if you do not succeed at
under research for a long time, and RFT is in line with the test you’ll not get good grades.” Within the impul-
the conventional view or hedonic principle (e.g., Gray, sive system, hearing this sentence is assumed to acti-
1982; Lang, 1995) in that considering the presence of a vate the concepts success and good, as well as not,
positive valence as an outcome (gain) elicits a type of which are of predominantly positive valence. Thus,
approach orientation that is described as promotion fo- perceiving verbal descriptions of potential nongains
cus, whereas considering the presence of a negative va- will instigate an approach orientation. Similarly, per-
lence as an outcome (loss) elicits an avoidance orienta- ceiving verbal descriptions of potential nonlosses will
tion or a prevention focus. However, RFT goes beyond instigate an avoidance orientation.
the hedonic principle when it comes to motivational ef- In sum, the reflective–impulsive model predicts that
fects of considering nongains and nonlosses as out- information about nongains and nonlosses has opposing
comes. According to RFT, deliberating about nongains effects depending on whether it is conveyed through
as possible outcomes of one’s action is connected to a discriminative stimuli that have acquired their meaning
promotion focus and should therefore elicit an ap- through associative learning or in a verbal format con-
proach-type orientation, although its overall valence is taining negations. Whereas in the former case, nongains
negative. Similarly, deliberating about nonlosses as are assumed to elicit an avoidance motivation because
possible outcomes is assumed to instigate a prevention they cue frustration and hence negative affect, in the lat-
focus and therefore avoidance-type behaviors, despite ter case an approach orientation is predicted because
its overall valence is positive. positive concepts will be activated in memory. Up to
Applying the reflective–impulsive model to this now, this prediction has not been tested in an adequate
topic leads to a more differentiated prediction. In par- experimental setting. However, some evidence, sam-
ticular, different motivational orientations are assumed pled across experiments and species, can be regarded as
to depend on whether operations are carried out in the first evidence for the viability of this prediction.


Specifically, in many of the experiments supporting is not exclusively determined by people’s attitudes.
RFT, action-outcome expectations were conveyed ver- Rather, it is important to consider impulsive influences
bally, whereas in animal research on operant and Pav- and study their interaction with the components of re-
lovian conditioning (see Rescorla & Solomon, 1967), flective determination. This framework integrates con-
cues for nonreward and nonpunishment acquired their cepts from motivational science into a dual-system the-
meaning through associative learning. It is significant ory and is able to explain a wide range of phenomena
that human and animal experimentation differs in its re- such as habitual versus intentional behavior, spontane-
sults. Studies with human participants match our predic- ous approach and avoidance behavior, perception-be-
tions for the case when information is conveyed ver- havior links, and effects of deprivation. Beyond this in-
bally. For example, some participants in one experiment tegrative attempt, the reflective–impulsive model may
by Crowe and Higgins (1997) were instructed that they contribute more than the sum of its parts in that phe-
would not have to perform a disliked task if they do not nomena, that are unexplained assumptions in other
do poorly on the exercises (prevention focus), while oth- models, follow logically from this framework and from
ers where instructed that they would have to do a disliked its operating characteristics. This includes phenomena
task if they do not do well on the exercises (promotion as diverse as the processing of negations, the limits of
focus).4 In several cognitive tasks, participants in the automatic social cognition, or the mechanisms of regu-
prevention focus were found to be more cautious and latory focus.
slow, and thus more accurate than participants in a pro- More important, the proposed perspective may help
motion focus (e.g., Förster, Higgins, & Idson, 1998; recover some phenomena that are considered to be the
Idson, Liberman, & Higgins, 2000; Shah, Higgins, & precursors of social psychology, specifically, the psy-
Friedman, 1998). In contrast, results from animal stud- chology of mass behavior (Le Bon, 1895; see also
ies match our predictions for associatively learned cues. Freud, 1921/1922; McDougall, 1920). Despite the in-
In one experiment (Ray & Stein, 1959), rats learned that sights that have been gained by replacing the mass by
a high tone was associated with the receiving an electric the group, social psychologists are ill at ease when it
shock, whereas a low tone was associated with the omis- comes to explaining social behaviors that are not
sion of such a shock. In a second part, the animals had the guided by people’s attitudes, such as panic behaviors,
opportunity to press a bar for the delivery of milk. This vandalism, riots, uprisings, and many facets of aggres-
type of approach behavior was suppressed if the signal sion and violence. Unlike Le Bon, we do not have to in-
for punishment (the high tone) was presented simulta- voke an irrational and emotional group mind that is
neously. More interesting, the signal for nonpunishment susceptible to hypnotic suggestion. Instead, our under-
(low tone) facilitated the approach behavior above base- standing of the mechanism of impulsive actions and of
line responding. Other research reviewed by Rescorla the conditions under which they occur will provide us
and Solomon (1967) as well as by Klein (1996) indicates with a more complete picture of social behavior.
that conditioned signals for nonreward facilitate avoid- Another advantage of the reflective–impulsive
ance behavior and that signals for nonpunishment in- model is that it lends itself to approaches from neuro-
hibit avoidance behavior. science (see also, Lieberman et al., 2002). Although it
In sum, the reflective–impulsive model accounts for seems difficult to link the reflective system and the im-
phenomena of regulatory focus as a joint effect of re- pulsive system as a whole to specific brain structures,
flective and impulsive mechanisms. The dissociation this may well be possible for specific psychological
of the effects of indirect verbal versus direct condi- processes described in the model. Take for instance im-
tioned activation of nongain and nonloss expectancies pulsively aggressive behavior that is due to reflective
can then be understood as the consequence of different underregulation. Neuroscientists have convincingly ar-
properties of the impulsive system and the reflective gued that such underregulation correlates with struc-
system. This interpretation, however, relies on results tural damage to the prefrontal cortex (Raine, Lencz,
that were obtained across studies and species. Future Buhrle, LaCasse, & Colletti, 2000). In addition, dam-
research should generate experimental procedures that age in the nucleus accumbens may be related to deficits
allow the orthogonal activation of direct versus indirect in the impulsive system (Cardinal, Pennicott,
of expectancies in human participants. Sugathapala, Robbins, & Everitt, 2001).
The notion that reflectively generated attitudes and
impulsive responses that express evaluations might be
due to the operation of different systems was recently
supported by a neuropsychological study conducted by
In this article, we have advanced the position that
Phelps and her colleagues (Phelps et al., 2000). Based
behavior is determined not only by assessments of
on findings from brain imaging, it would appear that
probability and value. This means that social behavior
amygdala activity was correlated with impulsive be-
havior indicating prejudice against African Americans,
4There were other conditions irrelevant to our discussion. while it was not correlated with reflective expressions


of racial attitudes. This suggests that the amygdala may that behavior is not only determined by its anticipated
be capable of detecting stimulus valence extremely fast consequences but also driven by forces outside of ra-
(Morris, Öhman, & Dolan, 1998) and before it can be tional control. However, it is not sufficient to focus on
processed by the reflective system. The distinction be- any one of these forces in isolation. To understand
tween flexible and fast acquisition of information in what people do, it seems necessary to study the dynam-
the reflective system and slowly changing representa- ics of behavior as the result of interacting influences.
tions in the impulsive system was recently linked to We have proposed that they originate from psychologi-
neocortical and hippocampal structures (J. L. cal systems that obey different operational principles.
McClelland et al., 1995). Of course, these few exam- The reflective–impulsive model of social behavior we
ples are only a sketch of possible interconnections be- have suggested is meant to provide a framework for
tween the two lines of research. A more thorough anal- this endeavor. Future research will test its merits.
ysis, however, is beyond the scope of this article.
Another area to which these principles can be ex-
tended is that of behavioral disorders. Until 1964, psy-
chological research on social and abnormal phenomena References
was published in the same scientific journals. Since
then, however, normal and abnormal behaviors have Aarts, H., & Dijksterhuis, A. (2000). Habits as knowledge structures:
been seen as distinct categories and explained by differ- Automaticity in goal-directed behavior. Journal of Personality
and Social Psychology, 78, 53–63.
ent theoretical models. From the perspective of the re-
Aarts, H., Dijksterhuis, A., & De Vries, P. (2001). On the psychology
flective–impulsive model, the difference between the of drinking: Being thirsty and perceptually ready. British Jour-
two types of behavior is only one of degree, not of kind. nal of Psychology, 92, 631–642.
What we call “abnormal” can often be described as the Ajzen, I. (1991). The theory of planned behavior. Organizational Be-
chronic determination of a behavior through only one havior and Human Decision Processes, 50, 179–211.
Anderson, J. R. (Ed.). (1981). Cognitive skills and their acquisition.
operating system. One example is addiction, a psycho-
Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
logical category that may not be confined to the abuse of Anderson, J. R. (1993). Rules of the mind. Hillsdale, NJ: Lawrence
drugs, but may also describe behaviors that are difficult Erlbaum Associates, Inc.
to control by intention, such as overeating, gambling, Baddeley, A. D. (1986). Working memory. New York: Oxford Uni-
watching TV, or surfing the Internet. As another exam- versity Press.
Banaji, M. R. (2001). Implicit attitudes can be measured. In H. L.
ple, phobic behaviors may be understood as being pre-
Roediger III, J. S. Nairne, I. Neath, & A. Surprenant (Eds.), The
dominantly controlled by the impulsive system. The au- nature of remembering: Essays in honor of Robert G. Crowder
tomatic associations of a stimulus with a fear response (pp. 117–150). Washington, DC: American Psychological As-
(cf. Bouton, Mineka, & Barlow, 2001) may then be as- sociation.
sessed with implicit measures such as the Implicit Asso- Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ:
Prentice Hall.
ciation Test (IAT; e.g., Teachman, Gregg, & Woody,
Bargh, J. A. (1990). Auto-motives: Preconscious determinants of so-
2001). Similarly, depression may be accompanied by an cial interaction. In E. T. Higgins & R. M. Sorrentino (Eds.),
automatic tendency to negatively evaluate information Handbook of motivation and cognition, Vol. 2 (pp. 93–130).
that is relevant for the self. Results from the IAT suggest New York: Guilford.
that this is the case (Gemar, Segal, Sagrati, & Kennedy, Bargh, J. A. (1996). Automaticity in social psychology. In E. T. Hig-
gins & A. W. Kruglanski (Eds.), Social psychology: Handbook
2001). Another example of application is panic reac-
of basic principles (pp. 169–183). New York: Guilford.
tions of crowds. Here, arousal is predicted to diminish Bargh, J. A. (1997). The automaticity of everyday life. In R. S. Wyer
reflective processing. At the same time, observing other & T. K. Srull (Eds.), Advances in social cognition: Vol. 10 (pp.
persons’ fearful expressions and flight behavior will 1–61). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
fuel own avoidance reactions, particularly fear and Bargh, J. A., & Barndollar, K. (1996). Automaticity in action: The
unconscious as repository of chronic goals and motives. In P.
flight and further enhance the arousal.
M. Gollwitzer & J. A. Bargh (Eds.), The psychology of action
In conclusion, this conceptualization suggests that (pp. 457–481). New York: Guilford.
social psychology should recognize the importance of Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social
an impulsive determination of social behavior and study behavior: Direct effects of trait construct and stereotype activa-
its mechanisms, particularly its interaction with the re- tion on action. Journal of Personality and Social Psychology,
71, 230–244.
flective determinants that are the primary focus of our
Bargh, J. A., & Ferguson, M. J. (2000). Beyond behaviorism: On the
theorizing. In this endeavor, it may be not enough to du- automaticity of higher mental processes. Psychological Bulle-
plicate the processes or outcomes of the reflective sys- tin, 126, 925–945.
tem and propose implicit decisions. Instead, it seems Baron, R. S. (2000). Arousal, capacity, and intense indoctrination.
necessary to study the unique mechanisms that drive the Personality and Social Psychology Review, 4, 238–254.
Baumeister, R. F., & Heatherton, T. F. (1996). Self-regulation failure:
system and understand its adaptive value in guiding be-
An overview. Psychological Inquiry, 7, 1–15.
havior under suboptimal circumstances. Bechara, A., Damasio, H., Tranel, D., & Damasio, A. R. (1997). De-
As mentioned at the outset, the notion that humans ciding advantageously before knowing the advantageous strat-
are rational beings is only part of the truth. We know egy. Science, 275, 1293–1295.


Bem, D. J. (1967). An alternative interpretation of cognitive disso- Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and
nance phenomena. Psychological Review, 73, 185–200. controlled components. Journal of Personality and Social Psy-
Blair, I. V. (2001). Implicit stereotypes and prejudice. In G. chology, 56, 5–18.
Moskowitz (Ed.), Cognitive social psychology: The Princeton Dickinson, A., & Balleine, B. W. (2002). The role of learning in the
Symposium on the Legacy and Future of Social Cognition (pp. operation of motivation systems. In H. Pashler & C. R. Gallistel
359–374). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. (Eds.), Steven’s handbook of experiemental psychology: Vol. 3.
Bless, H., & Forgas, J. P. (Eds.). (2000). The message within: The Learning, motivation, and emotion (3rd ed., pp. 497–533). New
role of subjective experience in social cognition and behavior. York: John Wiley & Sons.
Philadelphia: Psychology Press. Dijksterhuis, A., & Bargh, J. A. (2001). The perception-behavior ex-
Bodenhausen, G. V. (1993). Emotions, arousal, and stereotypic judg- pressway: Automatic effects of social perception on social be-
ments: A heuristic model of affect and stereotyping. In D. M. havior. In M. P. Zanna (Ed.), Advances in experimental social
Mackie & D. L. Hamilton (Eds.), Affect, cognition, and stereo- psychology, Vol. 33 (pp. 1–40). San Diego: Academic.
typing (pp. 13–37). New York: Academic. Dijksterhuis, A., Bargh, J. A., & Miedema, J. (2000). Of men and
Bodenhausen, G. V., Kramer, G. P., & Suesser, K. (1994). Happiness mackerels: Attention, subjective experience, and automatic so-
and stereotypic thinking in social judgment. Journal of Person- cial behavior. In H. Bless & J. P. Forgas (Eds.), The message
ality and Social Psychology, 66, 621–632. within: The role of subjective experience in social cognition and
Bouton, M. E., Mineka, S., & Barlow, D. H. (2001). A modern learn- behavior (pp. 37–51). Philadelphia: Psychology Press.
ing theory perspective on the etiology of panic disorder. Psy- Dovidio, J. F., Kawakami, K., Johnson, C., Johnson, B., & Howard,
chological Review, 108, 4–32. A. (1997). On the nature of prejudice: Automatic and controlled
Brendl, C. M., Markman, A. B., & Messner, C. (2003). The devalua- processes. Journal of Experimental Social Psychology: Special
tion effect: Activating a need devaluates unrelated objects. Issue on Unconscious Processes in Stereotyping and Prejudice,
Journal of Consumer Research, 29, 463–473. 33, 510–540.
Bruner, J. S. (1957). On perceptual readiness. Psychological Review, Drobes, D. J., Miller, E. J., Hillman, C. H., Bradley, M. M., Cuthbert,
64, 123–152. B. N., & Lang, P. J. (2001). Food deprivation and emotional re-
Bruner, J. S. (1973). Beyond the information given: Studies in the actions to food cues: Implications for eating disorders. Biologi-
psychology of knowing. New York: Norton. cal Psychology, 57, 153–177.
Cacioppo, J. T., Priester, J. R., & Berntson, G. G. (1993). Rudimen- Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. San
tary determinants of attitudes: II. Arm flexion and extension Diego: Harcourt.
have differential effects on attitudes. Journal of Personality and Epstein, S. (1991). Cognitive-experiential self-theory: An integra-
Social Psychology, 65, 5–17. tive theory of personality. In R. Curtis (Ed.), The relational self:
Cardinal, R. N., Pennicott, D. R, Sugathapala, C. L., Robbins, T. W., Theoretical conergences in psychoanalytical, social, and per-
& Everitt, B. J. (2001). Impulsive choice induced in rats by le- sonality psychology (pp. 111–137). New York: Guildford.
sions of the nucleus accumbens core. Science, 292, 2499–2501. Erber, R. (1991). Affective and semantic priming: Effects of mood
Center for the Study of Emotion and Attention. (1995). The interna- on category accessibility. Journal of Experimental Social Psy-
tional affective picture system: Digitized photographs. chology, 27, 480–498.
Gainesville: University of Florida, Center for Research in Farnham, S. D., Greenwald, A. G., & Banaji, M. R. (1999). Implicit
Psychophysiology. self-esteem. In D. Abrams & M. A. Hogg (Eds.), Social identity
Chaiken, S., Liberman, A., & Eagly, A. H. (1989). Heuristic and sys- and social cognition (pp. 230–248). Malden, MA: Blackwell.
tematic processing within and beyond the persuasion context. Fazio, R. H. (1986). How do attitudes guide behavior? In R. M.
In J. S. Uleman & J. A. Bargh (Eds.), Unintended thought (pp. Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and
212–252). New York: Guilford. cognition: Foundations of social behavior (pp. 204–243). New
Chaiken, S., & Trope, Y. (Eds.). (1999). Dual-process theories in so- York: Guilford.
cial psychology. New York: Guilford. Fazio, R. H. (1990). Multiple processes by which attitudes guide be-
Chen, M., & Bargh, J. A. (1999). Consequences of automatic evalua- havior: The MODE model as an integrative framework. In M. P.
tion: Immediate behavioral predispositions to approach or Zanna (Ed.), Advances in experimental social psychology, Vol.
avoid the stimulus. Personality and Social Psychology Bulletin, 23 (pp. 75–109). San Diego: Academic.
25, 215–224. Fazio, R. H. (2001). On the automatic activation of associated evalu-
Collins, A. M., & Quillian, M. R. (1969). Retrieval time from seman- ations: An overview. Cognition and Emotion, 15, 115–141.
tic memory. Journal of Verbal Learning and Verbal Behavior, 8, Fazio, R. H., & Olson, M. A. (2003). Implicit measures in social cog-
240–247. nition research: Their meaning and use. Annual Review of Psy-
Crowe, E., & Higgins, E. T. (1997). Regulatory focus and strategic incli- chology, 54, 297–327.
nations: Promotion and prevention in decision making. Organiza- Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R.
tional Behavior and Human Decision Processes, 69, 117–132. (1986). On the automatic activation of attitudes. Journal of Per-
Darwin, C. R. (1965). The expression of emotions in man and ani- sonality and Social Psychology, 50, 229–238.
mals. Chicago: University of Chicago Press. (Original work Fielder, K., Armbruster, T., Nickel, S. Walther, E., & Asbeck, J.
published 1872) (1996). Constructive memory & social judgement: Experi-
De Houwer, J., Crombez, G., & Baeyens, F. (2001). On the general- ments in the self-verification of question contents. Journal of
ity of the affective Simon effect. Cognition and Emotion, 15, Personality and Social Psychology, 71, 861–873.
189–206. Förster, J., Higgins, E. T., & Idson, L.C. (1998). Approach and
Deutsch, R., Gawronski, B., & Strack, F. (2003). “Not good” may avoidance strength during goal attainment: Regulatory focus
feel quite positive. Affective and behavioral consequences of and the “goal looms larger” effect. Journal of Personality and
processing negated information. Unpublished manuscript, Uni- Social Psychology, 75, 1115–1131.
versity of Würzburg. Förster, J., & Strack, F. (1996). The influence of overt head move-
Deutsch, R., & Strack, F. (2002). Evaluative learning with delayed ments on memory for valenced words: A case of concep-
gratification. Impulsive and reflective processes. Unpublished tual-motor compatibility. Journal of Personality and Social
manuscript, University of Würzburg. Psychology, 71, 421–430.


Förster, J., & Strack, F. (1997). Motor actions in retrieval of valenced Higgins, E. T., & Kruglanski, A. W. (Eds.). (2000). Motivational sci-
information: A motor congruence effect. Perceptual and Motor ence. Philadelphia: Psychology Press.
Skills, 85, 1419–1427. Higgins, E. T., Rholes, W. S., & Jones, C. R. (1977). Category acces-
Förster, J., & Strack, F. (1998). Motor actions in retrieval of valenced sibility and impression formation. Journal of Experimental So-
information: II. Boundary conditions for motor congruence ef- cial Psychology, 13, 141–154.
fects. Perceptual and Motor Skills, 86, 1423–1426. Higgins, E. T., Shah, J. Y., & Friedman, R. (1997). Emotional re-
Freud, S. (1922). Group psychology and the analysis of the ego. New sponses to goal attainment: Strength of regulatory focus as
York: Liverright. (Original work published 1921) moderator. Journal of Personality and Social Psychology, 72,
Freud, S. (1949). New introductory lectures on psychoanalysis. Lon- 515–525.
don: Hogarth. (Original work published 1933) Horn, L. R. (1989). A natural history of negation. Chicago: Univer-
Friedman, M. (1976). Price theory. Chicago: Aldine. sity of Chicago Press.
Gawronski, B., & Strack, F. (in press). On the propositional nature of Hull, C. L. (1943). Principles of behavior. New York: Appleton-Cen-
cognitive consistency: Dissonance changes explicit, but not im- tury-Crofts.
plicit attitudes. Journal of Experimental Social Psychology. Hummel, J. E., & Holyoak, K. J. (2003). A symbolic-connectionist
Gemar, M. C., Segal, Z. V., Sagrati, S., & Kennedy, S. J. (2001). theory of relational inference and generalization. Psychological
Mood-induced changes on the Implicit Association Test in re- Review, 110, 220–264.
covered depressed patients. Journal of Abnormal Psychology, Idson, L. C., Liberman, N., & Higgins, E. T. (2000). Distinguishing
110, 282–289. gains from nonlosses and losses from nongains: A regulatory
Gilbert, D. T. (1989). Thinking lightly about others: Automatic com- focus perspective on hedonic intensity. Journal of Experimental
ponents of the social inference process. In J. S. Uleman & J. A. Social Psychology, 36, 252–274.
Bargh (Eds.), Unintended thought (pp. 189–211). New York: Jacoby, L. L., Kelley, C., Brown, J., & Jasechko, J. (1989). Becoming
Guilford. famous overnight: Limits on the ability to avoid unconscious in-
Gilbert, D. T. (1991). How mental systems believe. American Psy- fluences of the past. Journal of Personality and Social Psychol-
chologist, 46, 107–119. ogy, 56, 326–338.
Gilbert, D. T., Krull, D. S., & Malone, P. S. (1990). Unbelieving the un- James, W. (1890). The principles of psychology. New York: Holt.
believable: Some problems in the rejection of false information. Johnson, M. K., & Hirst, W. (1991). MEM: Memory subsystems as
Journal of Personality and Social Psychology, 59, 601–613. processes. In A. Collins, S. Gathercole, M. Conway, & P. Mor-
Gilbert, D. T., Tafarodi, R. W., & Malone, P. S. (1993). You can’t not ris (Eds.), Theories of memory (pp. 241–286). Hillsdale, NJ:
believe everything you read. Journal of Personality and Social Lawrence Erlbaum Associates, Inc.
Psychology, 65, 221–233. Kahneman, D. (1999). Objective happiness. In D. Kahneman, E. Diener,
Gilovich, T., & Griffin, D. (2002). Introduction—Heuristics and bi- & N. Schwarz (Eds.), Well-being: The foundations of hedonic psy-
ases: Then and now. In T. Gilovich, D. Griffin, & D. Kahneman chology (pp. 3–25). New York: Russell Sage Foundation.
(Eds.), Heuristic and biases: The psychology of intuitive judg- Kahneman, D., & Frederick, S. (2002). Representativeness revisited:
ment (pp. 1–18). New York: Cambridge University Press. Attribute substitution in intuitive judgment. In T. Gilovich, D.
Gollwitzer, P. M. (1999). Implementation intentions. Strong effects Griffin, & D. Kahneman (Eds.), Heuristic and biases: The psy-
of simple plans. American Psychologist, 54, 493–503. chology of intuitive judgment (pp. 49–81). New York: Cam-
Gray, J. A. (1982). The neuropsychology of anxiety: An enquiry into bridge University Press.
the functions of the septo-hippocampal system. New York: Kelley, H. H. (1967). Attribution theory in social psychology. In D.
Clarendon Press/Oxford University Press. Levine (Ed.), Nebraska Symposium on Motivation, Vol. 15 (pp.
Greene, J. D., Sommerville, R. B., Nystrom, L. E., Darley, J. M., & 192–241). Lincoln: University of Nebraska Press.
Cohen J. D. (2001). An fMRI investigation of emotional en- Keltner, D., Gruenfeld, D. H., & Anderson, C. (2003). Power, ap-
gagement in moral judgment. Science, 293, 2105–2108. proach, and inhibition. Psychological Review, 110, 265–284.
Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Klein, S. B. (1996). Learning. Principles and applications (3rd ed.).
Attitudes, self-esteem, and stereotypes. Psychological Review, New York: McGraw-Hill.
102, 4–27. Koriat, A. (1993). How do we know what we know? The accessibil-
Greenwald, A. G., Banaji, M. R., Rudman, L. A., Farnham, S. D., ity model of the feeling of knowing. Psychological Review, 100,
Nosek, B. A., & Mellott, D. S. (2002). A unified theory of im- 609–639.
plicit attitudes, stereotypes, self-esteem, and self-concept. Psy- Krauss, R. M., Chen, Y., & Gottesman, R. F. (2000). Lexical gestures
chological Review, 109, 3–25. and lexical access: A process model. In D. McNeill (Ed.), Lan-
Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Mea- guage and gesture: Window into thought and action (pp.
suring individual differences in implicit cognition: The Implicit 261–283). New York: Cambridge University Press.
Association Test. Journal of Personality and Social Psychol- Kruglanski, A. W., & Thompson, E. P. (1999). Persuasion by a single
ogy, 74, 1464–1480. route: A view from the unimodel. Psychological Inquiry, 10,
Gregory, W. L., Cialdini, R. B., & Carpenter, K. M. (1982). Self-rele- 83–109.
vant scenarios as mediators of likelihood estimates and compli- Kuhl, J., & Goschke, T. (1994). A theory of action control: Mental
ance: Does imagining make it so? Journal of Personality and subsystems, modes of control, and volitional conflict-resolu-
Social Psychology, 43, 89–99. tion strategies. In J. Kuhl & J. Beckman (Eds.), Volition and
Haidt, J. (2001). The emotional dog and its rational tail: A social personality: Action versus state orientation (pp. 93–124). Seat-
intuitionist approach to moral judgment. Psychological Review, tle, WA: Hogrefe & Huber.
108, 814–834. Laird, J. D. (1974). Self-attribution of emotion: The effects of ex-
Helmholtz, H. (1867). Handbuch der physiologischen Optik [Hand- pressive behavior on the quality of emotional experience. Jour-
book of physiological optics]. Leipzig, Germany: Voss. nal of Personality and Social Psychology, 29, 475–486.
Higgins, E. T. (1996). Knowledge activation: Accessibility, applica- Lambert, A. J., Payne, B. K., Jacoby, L. L., Shaffer, L. M., Chasteen, A.
bility, and salience. In E. T. Higgins & A. W. Kruglanski (Eds.), L., & Khan, S. K. (2003). Stereotypes as dominant responses: On
Social psychology: Handbook of basic principles (pp. 133–168). the “social facilitation” of prejudice in anticipated public contexts.
New York: Guilford. Journal of Personality and Social Psychology, 84, 277–295.
Higgins, E. T. (1997). Beyond pleasure and pain. American Psychol- Lang, P. J. (1995). The emotion probe—Studies of motivation and at-
ogist, 52, 1280–1300. tention. American Psychologist, 50, 372–385.


Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1990). Emotion, atten- Morris, J. S., Öhman, A., & Dolan, R. J. (1998). Conscious and un-
tion, and the startle reflex. Psychological Review, 97, 377–395. conscious emotional learning in the human amygdala. Nature,
Langer, E. J., Blank, A., & Chanowitz, B. (1978). The mindlessness 393, 467–470.
of ostensibly thoughtful action: The role of “placebic” informa- Mussweiler, T., & Strack, F. (1999). Comparing is believing: A se-
tion in interpersonal interaction. Journal of Personality and So- lective accessibility model of judgmental anchoring. In W.
cial Psychology, 36, 635–642. Stroebe & M. Hewstone (Eds.), European review of social psy-
Lavy, E. H., & van den Hout, M. A. (1993). Attentional bias for chology, Vol. 10 (pp. 135–167). Chichester, England: Wiley.
appetitive cues: Effects of fasting in normal subjects. Behav- Neumann, R., Förster, J., & Strack, F. (2003). Motor compatibility:
ioural and Cognitive Psychotherapy, 21, 297–310. The bidirectional link between behavior and evaluation. In J.
Lazarus, R. S. (1984). On the primacy of cognition. American Psy- Musch & K. C. Klauer (Eds.), The psychology of evaluation. Af-
chologist, 39, 124–129. fective processes in cognition and emotion (pp. 371–391).
Le Bon, G. (1895). Psychologie des foules [Psychology of the Mahwah, NJ: Lawrence Erlbaum Associates, Inc.
crowd]. Paris: Alcan. Neumann, R., & Hess, M. (2001). The latency of facial action differ-
LeDoux, J. E. (1995). Emotion: Clues from the brain. Annual Review entiates the valence of processed information: Evidence that fa-
of Psychology, 46, 209–235. cial action is activated by evaluative processing. Unpublished
Lewin, K. (1935). A dynamic theory of personality. New York: manuscript, University of Würzburg.
McGraw-Hill. Neumann, R., Seibt, B., & Strack, F. (2001). The influence of mood
Libet, B., Gleason, C. A., Wright, E. W., & Pearl, D. K. (1983). Time on the intensity of emotional responses: Disentangling feeling
of conscious intention to act in relation to onset of cerebral ac- and knowing. Cognition and Emotion, 15, 725–747.
tivity (readiness-potential): The unconscious initiation of a Neumann, R., & Strack, F. (2000). Approach and avoidance: The in-
freely voluntary act. Brain, 106, 623–642. fluence of proprioceptive and exteroceptive cues on encoding
Lieberman, M. D., Gaunt, R., Gilbert, D. T., & Trope, Y. (2002). Re- of affective information. Journal of Personality and Social Psy-
flection and reflexion: A social cognitive neuroscience ap- chology, 79, 39–48.
proach to attributional inference. In M. P. Zanna (Ed.), Ad- Norman, D. A., & Shallice, T. (1986). Attention to action. Willed and
vances in experimental social psychology, Vol. 34 (pp. automatic control of behavior. In R. J. Davidson, G. E.
199–249). New York: Academic. Schwartz, & D. Shapiro (Eds.), Consciousness and self regula-
Logan, G. D. (1988). Toward an instance theory of automatization. tion: Advances in research (pp. 1–18). New York: Plenum.
Psychological Review, 95, 492–527. Ouellette, J. A., & Wood, W. (1998). Habit and intention in everyday
Logan, G. D., Schachar, R. J., & Tannock, R. (1997). Impulsivity and life: The multiple process by which past behavior predicts fu-
inhibitory control. Psychological Science, 8, 60–64. ture behavior. Psychological Bulletin, 124, 54–74.
Lotze, H. (1852). Medicinische Psychologie oder Physiologie der Paulhus, D. L., & Lim, D. T. K. (1994). Arousal and evaluative ex-
Seele [Medical psychology or the physiology of the soul]. Leip- tremity in social judgments: A dynamic complexity model. Eu-
zig, Germany: Weidmann’sche Buchhandlung. ropean Journal of Social Psychology, 24, 89–99.
Maass, A., Castelli, L., & Arcuri, L. (2000). Measuring prejudice: Palmeri, T. J. (1997). Exemplar similarity and the development of
Implicit versus explicit techniques. In D. Capozza & R. Brown automaticity. Journal of Experimental Psychology: Learning,
(Eds.), Social identity processes: Trends in theory and research Memory, and Cognition, 23, 324–354.
(pp. 96–116). London: Sage. Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood
MacDonald, M. C., & Just, M. A. (1989). Changes in activation lev- model of persuasion. In L. Berkowitz (Ed.), Advances in experi-
els with negation. Journal of Experimental Psychology: mental social psychology, Vol. 19 (pp. 123–205). New York: Ac-
Learning, Memory, and Cognition, 15, 633–642. ademic.
McDougall, W. (1920). The group-mind. A sketch of the principles of Phelps, E. A., O’Connor, K. J., Cunningham, W. A., Funayama, S.,
collective psychology. Oxford, UK: Putnam. Gatenby, J. C., Gore, J. C., et al. (2000). Performance on indi-
Mall, R. A. (1975). On reflection and negation. Philosophy and rect measures of race evaluation predicts amygdala activation.
Phenomenological Research, 35, 79–92. Journal of Cognitive Neuroscience, 12, 729–738.
Marr, D. (1982). Vision. A computational investigation into the hu- Raine, A., Lencz, T., Buhrle, S., LaCasse, L., & Colletti, P. (2000).
man representation and processing of visual information. San Reduced prefrontal gray matter volume and reduced autonomic
Francisco: Freeman. activity in antisocial personality disorder. Archives of General
Martin, L. L., Harlow, T. F., & Strack, F. (1992). The role of bodily Psychiatry, 57, 119–127.
sensations in the evaluation of social events. Personality and Ray, O. S., & Stein, L. (1959). Generalization of conditioned suppression.
Social Psychology Bulletin, 1, 412–419. Journal of the Experimental Analysis of Behavior, 2, 357–361.
Martin, L. L., Seta, J. J., & Crelia, R. A. (1990). Assimilation and Rescorla, R. A., & Solomon, R. L. (1967). Two-process learning the-
contrast as a function of people’s willingness and ability to ex- ory. Psychological Review, 73, 151–182.
pend effort in forming an impression. Journal of Personality Riskind, J. H. (1984). They stoop to conquer: Guiding and self-regu-
and Social Psychology, 59, 27–37. latory functions of physical posture after success and failure.
McClelland, D. C., & Atkinson, J. W. (1948). The projective ex- Journal of Personality and Social Psychology, 47, 479–493.
pression of needs: I. The effect of different intensities of the Ross, L., Lepper, M., & Hubbard, M. (1975). Perseverence in
hunger drive on perception. Journal of Psychology, 25, self-perception and social perception: Biased attributional pro-
205–222. cesses in the debriefing paradigm. Journal of Personality and
McClelland, J. L., McNaughton, B. L., & O’Reilly, R. C. (1995). Social Psychology, 32, 880–892.
Why there are complementary learning systems in the hippo- Russell, J. A. (2003). Core affect and the psychological construction
campus and neocortex: Insights from the successes and failures of emotion. Psychological Review, 110, 145–172.
of connectionist models of learning and memory. Psychological Schreiber, C. A., & Kahneman, D. (2000). Determinants of the re-
Review, 102, 419–457. membered utility of aversive sounds. Journal of Experimental
Metcalfe, J., & Mischel, W. (1999). A hot/cool-system analysis of Psychology: General, 129, 27–42.
delay of gratification: Dynamics of willpower. Psychological Schwarz, N., & Bohner, G. (2001). The construction of attitudes. In
Review, 106, 3–19. A. Tesser & N. Schwarz (Eds.), Blackwell handbook of social
Mele, A. R. (1992). Irrationality: An essay on akrasia, self-decep- psychology: Vol. 1, Intraindividual processes (pp. 436–457).
tion, and self-control. New York: Oxford University Press. Oxford, England: Blackwell.


Schwarz, N., & Clore, G. L. (1983). Mood, misattribution, and judg- Strack, F., & Neumann, R. (2000). Furrowing the brow may under-
ments of well-being: Informative and directive functions of af- mine perceived fame: The role of facial feedback in judgments
fective states. Journal of Personality and Social Psychology, 45, of celebrity. Personality and Social Psychology Bulletin, 26,
513–523. 762–768.
Schwarz, N., & Clore, G. L. (1996). Feelings and phenomenal expe- Sutton, S. K., & Davidson, R. J. (1997). Prefrontal brain asymmetry:
riences. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psy- A biological substrate of the behavioral approach and inhibition
chology: Handbook of basic principles (pp. 433–465). New systems. Psychological Science, 8, 204–210.
York: Guilford. Teachman, B. A., Gregg, A. P., & Woody, S. R. (2001). Implicit asso-
Seibt, B., Häfner, M., & Deutsch, R. (2003). Prepared to eat: How ciations for fear-relevant stimuli among individuals with snake
immediate affective and motivational responses to food cues and spider fears. Journal of Abnormal Psychology, 110,
are influenced by food deprivation. Unpublished manuscript, 226–235.
University of Würzburg. Thorndike, E. L. (1911). Animal intelligence. New York: Macmillan.
Seibt, B., & Neumann, R. (2002). The influence of unspecific mo- Trope, Y. (1986). Identification and inferential processes in
tor-programs on experiencing emotions. Unpublished manu- dispositional attribution. Psychological Review, 93, 239–257.
script, University of Würzburg. Tulving, E. (1985). Memory and consciousness. Canadian
Shah, J., Higgins, E. T., & Friedman, R.S. (1998). Performance in- Psycholgy, 26, 1–12.
centives and means: How regulatory focus influences goal at- Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for
tainment. Journal of Personality and Social Psychology, 74, judging frequency and probability. Cognitive Psychology, 42,
285–293. 207–232.
Sloman, S. A. (1996). The empirical case for two systems of reason- Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty:
ing. Psychological Bulletin, 119, 3–22. Heuristics and biases. Science, 185, 1124–1131.
Smith, E. R. (1994). Procedural knowledge and processing strategies Tversky, A., & Kahneman, D. (1982). Judgments of and by represen-
in social cognition. In R. S. Wyer & T. K. Srull (Eds.), Hand- tativeness. In D. Kahneman, P. Slovic, & A. Tversky (Eds.),
book of social cognition (pp. 99–152). Hillsdale NJ: Lawrence Judgment under uncertainty: Heuristics and biases (pp.
Erlbaum Associates, Inc. 84–98). New York: Cambridge University Press.
Smith, E. R. (1998). Mental representation and memory. In D. T. Uleman, J. S. (1999). Spontaneous versus intentional inferences in
Gilbert, S. T. Fiske, & G. Lindzey (Eds.), Handbook of social impression formation. In S. Chaiken & Y. Trope (Eds.),
psychology, Vol. 1 (4th ed., pp. 391–445). New York: Oxford Dual-process theories in social psychology (pp. 141–160). New
University Press. York: Guilford.
Smith, E. R., Branscombe, N. R., & Borman, C. (1988). Generality Wason, P. C. (1959). The processing of positive and negative infor-
of the effects of practice on social judgment tasks. Journal of mation. Quarterly Journal of Experimental Psychology, 11,
Personality and Social Psychology, 54, 385–395. 92–107
Smith, E. R., & DeCoster, J. (2000). Dual process models in social Wason, P. C., & Jones, S. (1963). Negatives: Denotation and conno-
and cognitive psychology: Conceptual integration and links to tation. British Journal of Psychology, 54, 299–307.
underlying memory systems. Personality and Social Psychol- Wänke, M., Bohner, G., & Jurkowitsch, A. (1997). There are many
ogy Review, 4, 108–131. reasons to drive a BMW: Does imagined ease of argument gen-
Smith, E. R., & Neumann, R. (in press). Emotion considered from eration influence attitudes? Journal of Consumer Research, 24,
the perspective of dual-process models. In L. Feldman Barrett, 170–177.
P. Niedenthal, & P. Winkielman (Eds.), Emotion: Conscious Wegner, D. M. (2002). The illusion of conscious will. Cambridge,
and unconscious. New York: Guilford. MA: MIT Press.
Solarz, A. K. (1960). Latency of instrumental responses as a function Wegner, D. M., Ansfield, M., & Pilloff, D. (1998). The putt and the
of compatibility with the meaning of eliciting verbal signs. pendulum: Ironic effects of the mental control of action. Psy-
Journal of Experimental Psychology, 59, 239–245. chological Science, 9, 196–199.
Stepper, S., & Strack, F. (1993). Proprioceptive determinants of Wegner, D. M., Coulton, G. F., & Wenzlaff, R. (1985). The transpar-
emotional and nonemotional feelings. Journal of Personality ency of denial: Briefing in the debriefing program. Journal of
and Social Psychology, 64, 211–220. Personality and Social Psychology, 49, 338–346.
Strack, F. (1992). The different routes to social judgments: Experien- Wells, G. L., & Petty, R. E. (1980). The effects of overt head move-
tial versus informational strategies. In L. L. Martin & A. Tesser, ments on persuasion: Compatibility and incompatibility of re-
(Eds.), The construction of social judgments (pp. 249–275). sponses. Basic and Applied Social Psychology, 1, 219–230.
Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Wilson, T. D., Lindsey, S., & Schooler, T. Y. (2000). A model of dual
Strack, F. (1999). Beyond dual-process models: Toward a flexible attitudes. Psychological Review, 107, 101–126.
regulation system. Psychological Inquiry, 10, 166–169. Winkielman, P., & Cacioppo, J. T. (2001). Mind at ease puts a smile
Strack, F., & Hannover, B. (1996). Awareness of influence as a precon- on the face: Psychophysiological evidence that processing fa-
dition for implementing correctional goals. In P. Gollwitzer & J. cilitation increases positive affect. Journal of Personality and
A. Bargh (Eds.), The psychology of action: Linking cognition and Social Psychology, 81, 989–1000.
motivation to behavior (pp. 579–595). New York: Guilford. Wispe, L. G., & Drambarean, N. C. (1953). Physiological need, word
Strack, F., Martin, L. L., & Stepper, S. (1988). Inhibiting and facili- frequency, and visual duration thresholds. Journal of Experi-
tating conditions of the human smile: A nonobtrusive test of the mental Psychology, 46, 25–31.
facial feedback hypothesis. Journal of Personality and Social Yerkes, R. M, & Dodson, J. D. (1908). The relation of strength of
Psychology, 54, 768–777. stimulus to rapidity of habit formation. Journal of Comparative
Strack, F., & Mussweiler, T. (1997). Explaining the enigmatic an- and Neurological Psychology, 18, 459–482.
choring effect: Mechanisms of selective accessibility. Journal Zajonc, R. B. (1965). Social facilitation. Science, 149, 269–274.
of Personality and Social Psychology, 73, 437–446. Zajonc, R. B. (1980). Feeling and thinking: Preferences need no in-
Strack, F., & Neumann, R. (1996). “The spirit is willing, but the flesh ferences. American Psychologist, 35, 151–175.
is weak”: Beyond mind-body interactions in human deci- Zajonc, R. B., Murphy, S. T., & Inglehart, M. (1989). Feeling and fa-
sion-making. Organizational Behavior and Human Decision cial efference: Implications of the vascular theory of emotion.
Processes, 65, 300–304. Psychological Review, 96, 395–416.