# MENTALISM AND BEHAVIORISM: MERGING?

Vladimir A. Lefebvre

**MATHEMATICAL MODELING OF REFLEXIVE PROCESSES
**

MENTALISM AND BEHAVIORISM: MERGING?

© Vladimir A. Lefebvre (USA), 2003

University of California at Irvine, USA ABSTRACT The Reflexive Intentional Model of the Subject (RIMS) connects the subject’s bipolar probabilistic behavior with its mental domain. We demonstrate that the Matching Law is a formal consequence of this tie. RIMS allows us also to deduce theoretically the main patterns of animal behavior in the experiments with two alternatives where the Matching Law reveals itself. This finding inclines us to put forth a hypothesis that this law reflects the process of self programming of the subject with mental domain. As a result, the subject acquires the ability to choose alternatives with fixed probabilities. With this explanation, the relative frequencies of pressing a pedal or pecking at a key play the role of half finished products which after being downloaded into the self turn into the probabilities of choice. The Matching Law can be regarded at as an operational indication of the mental domain existence.

Introduction

Mentalism is a science about subjective matters that gives a living creature a niche for the inner world. Behaviorism is a science about behavior depriving a living crea ture of it. Both of these sciences have a common feature; in them, an organism appears as an entity. The first one focuses on a subject’s relation to the self, while the second one focuses on the relations between the subject and the environment (Tolman, 1932). For the last few decades, the border between mentalism and be haviorism has moved: a formal model of the subject has appeared which includes both its mental domain and its behavior. The model’s verification goes through its penetration into various branches of psychology, sociology, and anthropology. Be haviorism represents the most attractive field for such a penetration, because of its strict inner discipline and methodological honesty that allows us to distinguish clear ly what is understood and what is not. One of the unsolved problems in the science

REFLEXIVE PROCESSES AND CONTROL No. 2, V. 2, 2003, p. 56 76

We still do not have clear ideas about whether all living creatures are capable of probabilistic choice or only some of them. RIMS is a probabilistic model. 1959. obtaining money or food. Savage. Miller & Sulcoski. Wheeler (1987). Lefebvre (1965. For example. 1996). Various aspects of this model were considered in works by Adams Webber (1987. Mosteller & Nogee. Bradley & Terry. 1997). A great number of specialists from psychiatrists to sociologists studying criminals and ter rorists are interested in finding objective laws of moral choice. Atkinson et al.LEFEBVRE. Krylov (1994). 1960. 1967. one playing the role of the positive pole and the other that of the negative pole. 1965).b. for example. 1961. 1952.V. 1995. Luce. von Neuman & Morgenstern. Kauffman (1990). 1998). Baker (1999). 1977a. Spence. Townsend (1983. RIMS is a special mathematical representation of a subject making choice be tween two alternatives. 1987. a deal with an enemy may be more profitable than the deal with a friend. 1951. Batchelder (1987). 1957. RIMS connects the subject’s probabilistic behavior
1 RIMS is described in great detail in Lefebvre’s Algebra of Conscience (2001) and in its supplemented translation into Russian (2003). 1944. Levitin (1987). 1962. 1999a. Popper (1992). In this work we offer a solution to this problem with the help of Reflexive Intentional Model of the Subject (RIMS)1. This line of investigations changed significantly the view that behavior is a process completely determined by the environment. Audley. 1988). The utilitarian aspect relates to the behavior which is advantageous from the practical point of view. Also. 1980. 1972. Taran (1999).b). Rapoport (1990. 1960. In creating this model we tried to understand a phenomenon of “moral choice” from a purely scientific point of view.
. 1961). This model reflects two aspects of the subject’s activity: util itarian and deontological. 1927.b). McClain (1987). for example. LaBerge. Schreider (1994. Suppes & Siegel. Although effective methods have been developed to predict the results of probabilistic choice. A human mental domain must be represented in their studies as clearly and unambiguously as be havior is represented in behaviorism. choos ing between good and evil. Restle. Both these aspects are connected into a single process of behavior generation by the formal model. MENTALISM AND BEHAVIORISM: MERGING?
57
of behavior is the Matching Law (Herrnstein. Bower. It describes the ability of birds and mammals to regulate the ratio between a sequence of reinforcements and a sequence of responses. rather than from a moralistic one. 1951.A. a problem of its essence remained untouched. It may happens that the “moral” orientation of the alternative does not correspond to the utilitarian one. 1990). Zajonc (1987).. we do not know how an organism “learns” the probabilities with which it “must”make a choice in a given situation. This ability looks strange from the point of view of the util itarian common sense (see Williams. It predicts probabilities with which the subject chooses the alternatives. Kaiser & Schmidt (2003). Davidson. The deontological aspect relates to the idealistic behavior. 1959. 1992a. The idea that the subject’s choice is probabilistic appeared early in the twentieth century and was used in many theoretical models (Thurst one. Lefebvre & Adams Webber (2002).

the subject’s pencil oscillates before it makes the final mark. 1982) allow us to hypothesize that generation of a mixed state in humans is also connected with their motor activity. To choose alternatives with fixed probabilities. say of a rat. during which the subject processes information received from the environment into prob abilistic distribution. when the subject is given a task to mark the intensity of a stimulus on a scale.58
MATHEMATICAL MODELING OF REFLEXIVE PROCESSES
with its mental domain and allows us to formulate a few new hypotheses.
. As a result of such a “downloading” of the probability the subject became capable of making an instant probabilistic choice. Using a quantum mechanical metaphor we can say that immediately before the act of choice. As a result. In the framework of this model. 1999a). prior to the act of choice. This activity may also reveal itself during a process of estimation. It is worth emphasizing that the ability of the subject to make a choice between the alterna tives with fixed probabilities indicates a rather high level of the development. 1985). and even a man cannot solve this problem through its brain activity only. We presume that the “secret” of the Matching Law is that it reflects a procedure of forming a mixed state in the subject. for example. 1992a). and the act of choice is a “collapse” of the mixed state. the subject’s state is uncertain and can be characterized by the distribution of probabilities over alternative choic es. Because of that failure the entire organism becomes involved in a computational process. a pigeon. Poulton & Simmonds. We cannot exclude the possibility that human beings may download the probabilities by eye move ments. the organism must some how “download” them into the self. the subject is in a mixed state. 1979. to obtain it an organism must spend energy. The experiments with two keys in which human subjects were used (see Rud dle et al.. Sometimes it is even difficult to determine which mark is final (see. The specialists in mathematical modeling know well how difficult it is to construct a technical device which would generate a random sequence of 0’s and 1’s with a fixed probability of their appearance. Their appearance indicates the moment of an organ ism’s “liberation” from the “necessity” to respond in one only way to an external influence. It is possible that the organism’s ability to give response undetermined by a stimulus raises its chances to find food and not to become another organism’s food (Lefebvre. Wearden & Burgess. it is an external demonstration of this process. We may suppose that probabilistic behavior of organisms appears at the same time as their mental domain. We may sup pose that these oscillations are functionally analogous to rats’ running from one food hopper to another. But this ability is not “free” for the subject. whose goal is to generate frequencies which would later transform into probabilities. When an animal is running between the two feed hoppers (in the experiments in which the Matching Law is revealed). Let us note that RIMS can explain the process of categor ical estimation as well as that of matching (Lefebvre. Let us imagine that an organism. For example. the subject moves into one of the pure states.

We single out four stages in the devel opment of behaviorism and can see a move toward the fifth one. and those acquired as the life experience were called conditional. and I is the value of the variable corresponding to the model of the self. At the beginning of this stage was Thorndike (1932) who formulated The Law of Effect. correspond to the output. This transformation was called reflex. indicates the appearance of the second stage. An organism’s life actions.” etc. MENTALISM AND BEHAVIORISM: MERGING?
59
The most important difference of RIMS from the models existing previously consists in the introduction of a new special variable which corresponds to the sub ject’s model of the self (Lefebvre.). In this case.V. At the first stage. variable I can be omitted. which appeared in the Cartesian time.” which expresses a rule of behavior of an organism in concise form. Pavlov’s avoiding such expression as “a dog noticed.” For example. as a rule. neither on prohibitions on use introspective concepts (as for example.LEFEBVRE. an animal
. The fourth stage is connected with experimental methods developed by Skin ner (1938) and his followers. During the third stage. It turned out that under this condition. Unlike deter ministic models. and the environment’s demands to the input. A mechanism inside the box automatically transforms each demand into a response. In the framework of RIMS. The probabilistic models. In their experiments. The Law of Self Reflexion in a Logical Scheme of Evolution of Behaviorism
In the evolution of the science of behavior. belong to the first or second stage. they describe statistical characteristics of stimulus response rath er than functional relation between a set of stimuli and a set of responses. the organism of the subject tends to generate a line of behavior such that it reaches and holds equation B = I. The inborn reflexes were called un conditional. Pavlov’s (1927) discovery that reflexes are of two types: inborn and acquired. 1988). one may see a clear logic which does not depend much on in individual preferences of researchers. an animal response may influ ence the generation of stimuli. where B is the value of the variable which describes the subject’s behavior. Williams. today (Fig. and we obtain a behavioristic type model which can be empirically falsi fied. This principle of behavior generation we will call The Law of Self Reflexion (Lefebvre. 1).” “a dog understood. The intentional behavior is given as B = I. a cat’s organism in Thorndike puzzle box performs selections of successful manipu lations with the lock and at the end of successive trials exits cage faster than at the beginning (see also Herrnstein. it was found that the automatic response to a stimulus may change becoming more effective. We interpret the value of this variable as the subject’s intention to make a choice. which shows the ability of a living crea ture to modify its response to a stimulus depending on its “effect. called responses. 1965. an organism was rep resented as a black box with an input and output (here we use a metaphor belong ing to later times). Each stage can be described by a “law.
1. 1977b).A. 1970. 2002).

The scheme does not reflect the time order of the ideas’ appearance. 1999a. For example. Thorndike. Numerous attempt to explain this law in the frame work of behaviorism have not given us a convincing explanation.
2. the Matching Law displays bipolarity and the law of self reflexion (Lefebvre. The Matching Law
The ability of an organism to regulate relations between the sequence of responses and the sequence of reinforcements was found by Herrnstein (1961) in the experi ments with pigeons.
I
R
Law of Reflex
II
C&U
Law of conditional and unconditional reflex
III
C&U
Law of Effect
IV
C&U
Matching Law
V
C&U
Law of Self Reflexion
I
Fig.60
MATHEMATICAL MODELING OF REFLEXIVE PROCESSES
generates a line of behavior which establishes a fixed ratio between a sequence of stimuli and a sequence of responses. From the RIMS point of view. A cage had two keys. Pavlov. The broadening of behaviorism framework leads us toward the fifth stage which indi cates merging of mentalism and behaviorism (Fig. Without broadening its vo cabulary the science of behavior may not be able to explain the Matching Law. Watson and Bechter ev worked on their studies at about the same time. 2002). and by Thorndike to the third stage. Each key has an independent schedule by which rein
. by Pavlov to the second one. The concepts of the image of the self and intention lie beyond the vocabulary of behaviorism. but contributions by early Watson and Bechterev belong to the first stage. A mathematical formula for this ratio has been called the Matching Law. When a pigeon pecks a key it may result in the appearance of a grain. Logical stages in the development of behaviorism. 1. 1).

In his recent publication. 1988). MENTALISM AND BEHAVIORISM: MERGING?
61
forcement is delivered. Baum & Aparicio.2) can be reduced to
ΒP rP =c( ). the mean number of pecks.3)
where BP > BN.A. with condition rP > rN . (2.1)
Equation (2.3) are particular cases of the more general equa tion: Β2 ϕ(r2) =c( ). (2. The pair of intervals were chosen in such a way that sometimes the reinforcement appeared more often in one key. necessary to receive rein forcement.2)
where c and β are parameters which characterize a subject in a given experiment consisting of a sequence of sessions. The results of the experiments led to the formulation of the Generalized Matching Law (Baum.1) was called the Matching Law. Equations (2. and sometimes in the other. The mean interval between the appearance of grains can vary (Variable Interval Schedules. (1999) suggested that (2.4) Β1 ϕ(r1) which represents the subjects’ behavior in the experiments described above (Davi son & Jones. Besides VI other schedules were also used.LEFEBVRE.
(2.1). Baum et al. VI). There were also experiments with rats and humans. 1974):
Β2 r2 =c( )β . For example. varied. Baum (2002) substituted condition BP > BN. 1995. VR. 1999). ΒN rN
(2. Variables with subscript P relate to the more often chosen alterna tive.
3. Baum and Aparicio (1999) gave the following interpretation which expresses the dominant point of view: ”Despite
. The Attempts to explain the Matching Law within the Framework of the Science of Behavior
Why does equation (2.V. Β1 r1
(2. The experiment consisted of a series of ses sions with fixed mean intervals for each key in each session. Quite recently. and those with N relate to less often chosen one. in Vari able Ratio Schedule.2) and (2.4) hold? It is natural to assume that it is a by product of more fundamental processes (Williams. It turned out that the birds choose the line of behavior such that the ratio of the numbers of pecks to the keys (B1 and B2) is approximately equal to the ratio of the numbers of corresponding reinforcements (r1 and r2):
Β2 = Β1
r2 r1
.

but c was considered just a scale coefficient connecting utility values of re inforcements from two different sources. 2001). There are numerous general and experimental arguments pro and contra the principle of optimality as an explanation of the Matching Law (Williams. 1999). which describes the Generalized Matching Law. in which one alternative was connected with schedule VI. A usual explanation of the necessity of its introduction can be clarified with the following example. has two free parameters c and β. they chose the strategies to make equation (2. 5. 1982. all the attempts to reduce the Matching Law to opti mality look unconvincing..1) for the “sum” of utilities.1) Β1 r1 Similar argumentation was used for the experiments in which the pieces of food were equal. Baum et al. 1979. for example. 1974. and the other with schedule VR? (see.”
. RIMS
The Reflexive Intentional Model of the Subject reflects the inner domain and be havior of a subject in its interactions with two objects which we called “agencies. the researchers had to as sume that the organism of the subject was capable of finding statistical characteristics of non simultaneous factors and reflecting them onto c. Their value must be found experimentally for each subject. 75).4) hold instead. Fig. in accordance to which an animal is adapting to the environ ment in a way that it looks rational and goal oriented. all leading theories about operant choice may be seen as models of optimality” (p. The results of the experiments conducted by Mazur (1981) present a serious argument against the principle of optimality. (3.75r2 Β2 = . Sometimes for the sake of saving this argumentation. the birds did not choose the optimal strategy. Equa tion (2. we will obtain the following ratio: 0.62
MATHEMATICAL MODELING OF REFLEXIVE PROCESSES
claims to the contrary. The idea of optimality is consonant with the main thesis of behaviorism. it was said that parameter c reflects a hidden factor which changes utility measures of the same product from two different sources. 1988. instead of the number of pieces (r1 and r2). Nevertheless.75 of that from the right one. There were many discussions concerning β (Baum.2). In these cases.) The explanation of c constant value turns into an independent problem similar in its complexity with the explanation of the Matching Law.
4. Baum. How else could they explain the fixed value of c in the experiments. The conditions of the experiments permitted the researchers to easily discover the pigeons tenden cy to maximize the amount of food they received. Let us consider one more argument against the principle of optimality. Aparicio. But if we reject the interpreta tion of c as a scale coefficient. Let a piece of food from the left food hopper is 0. Wearden & Burgers. If we write (2.

x1. The subject adapted to the environment is called intentional if (4. If the value of x1 is determined by the environment. Variable x3 corresponds to the subject’s model of the self. If this equation does have a solution.V. Its value is the relative probability with which the subject intends to influence the positive agency.1) turns into the following equation in relation to X1: (4. The appearance of intention and the appearance of readiness are considered to be two independent events.A. In the opposite case. Its value is the relative probability with which the subject’s executive system is ready to influence the pos itive agency. MENTALISM AND BEHAVIORISM: MERGING?
63
One of them plays the role of the positive pole and the other that of the negative pole.1) does not hold.LEFEBVRE. Variable X1 represents the subject’s executive system. The variables not determined by the environment may take on any values which do not violate (4. x2. the intentional subject’s readiness is equal to its intention. The subject is represented with the following equation: X1 – x1 – (1– x1)(1– x2)M(x3) = 0 . Its value is interpreted in two ways. It represents the intentional domain of the subject. First. we interpret it as the influ ence received from the environment. (4.1) is supple mented with the following limitation: X1 = x3 .1) where all variables and function M(x3) take on the values from interval [0. there is a function of the type
. and x3 – intention. Variable x2 represents the subject’s experience. The absence of its solution means that the subject is not capable of intentional actions. The environment may determine all values of the variables X1. (4. x3 or only some of them. 1999b. that is. we consider the subject adapted to the environment. If the set of values determined by the environment is such that limita tion (4.2) that is.1] (Lefe bvre.2) expression (4. Function M(x3) represents the subject’s prognostic activity. it is the subject’s re quirement of the positive agency to influence him with this probability. In the opposite case. Its value is the integral estimation of the relative probability with which the positive agency influenced him in the past. The intentional subject’s choice will be called an intentional probabilistic choice. Second. Variable x1 represents the relation between the subject and the positive agency in a given moment. it is a relative proba bility of the positive agency influencing the subject.1). we will call X1 – readiness. we consider the subject disadapted. 2001). it is interpreted as the subject’s need in the environment’s influence. Under condition (4.3) X1 – x1 – (1 – x1)(1 – x2)M(X1) = 0 . Its value is the sub jective evaluation of the future positivity under condition that intention x3 turns into reality. For the sake of language simplicity.

with x1 + x2 + d > 0 X1 = x1 1 – (1 – x1)(1 – x2)(1 – d) .4) satisfying (4. While modeling some psychological processes we may introduce more func tional limitations on connections between variables X1. If (4. If X1 is not an effective function of x1 and x2. For example. that is.8). (4. This index lowers a degree of the positivity of the future. the subject is capable of making a free choice. By substituting these values into (4. when this index is maxi mal.7)
When x1 = x2 = d = 0. (4.9)
.3).3) turns into equation in relation to X1: X1 = x1 + (1 – x1)(1 – x2)(1 – d) X1.8) X1 x1 Let us connect this equation with the number of the subject’s influence on the agencies and the number of the agencies influences on the subject as follows: Ν1 n1 X1 = . all variables of which can be related to observable values. the value of X1 is not a function of x1 and x2. (4. The value of d is interpreted as an index of depression.6) holds. (4. M(x3) ≡ x3.
(4. we consider function M(x3) to be linear and look as follows: M(x3) = (1 – d)x3 .7) can be represented as ). even probability of which cannot be predicted by external observers.6a) (4. x1. the degree of positivity of the future is equal to the value of intention.5) In the framework of this work. (4. It follows from (4. RIMS then turns into a behavioristic model. x2.1]. we obtain 1 – X1 = (1 – (1 – x2)(1 – d))( 1 – x1
Ν2 Ν1
=p(
n2 n1
). hence. If x1 > 0. (4. x1 = . d = 1. in accordance with the definition. the future looks negative for the subject. so. we assume that the subject has the ability to make a choice. (4.8a) Ν1 + Ν2 n1 + n2 where N1 and N2 are the numbers of the subject’s influences on the positive and negative agencies.1) that the following inequalities hold in dependently from the type of function M(x3): x1 ≤ X1 ≤ 1 – x2 + x1x2 . (4. that is.x2) . x3 to reflect subjects’ specific peculiarities. We call this choice a free choice.64
MATHEMATICAL MODELING OF REFLEXIVE PROCESSES
X1 = f(x1.6) where d ∈ [0. we can exclude x3 from our considerations. When d = 0. and n1 and n2 are the numbers of the positive and negative agen cies’ influences on the subject. function M(x3) ≡ 0.

If a given session is first. (4. Now we write (4. which characterizes the subject’s mixed state. (4. First.11).2) for β = 1.9) corresponds to the Generalized Matching Law (2. On the other hand. Let suppose that (1) The preparation of the subject to the experiment and the conditions of the experiment determine the value of the depression index d. then the value of p corresponds to free parameter c. an organism stabilizes the relative frequencies of contacts with the agencies by holding X1 = x3. Further we demonstrate that the experiment with two keys can be modeled with the help of RIMS. being constant during the entire experiment.11) holds. Each key has its own schedule of reinforcement (type VI or VR).A. the subject’s activity aims not only at obtaining food. Modeling the Experiment with Two Keys
We assume that in the experiments with the two keys. (4.LEFEBVRE. x2=1/2. In every session.9) as follows: D1 D2 =p.9) represents the intentional subject when M(x3) = (1 – d)x3 . The subject is placed into a cage with two keys each connected with a food hopper.10) Equation (4.11)
It follows from the equivalency of (4. (4. The subject’s pecks onto keys are reinforced by pieces of food but rarely. but also at generating a mixed state (see Introduction). After stabilization. (4.V. The values n1 n2 = D1 and = D2 (4. which in a given session plays the role the positive agency.7) that the subject is intentional if and only if (4. frequency N1/(N1+N2) “turns” into the probability equal to the frequency.9). (2) The following events take place at the beginning of each session: (a) One key acquires the status of the positive agency. RIMS does not indi
. If in equation (2.2). the mean interval between rein forcements is fixed for each key. MENTALISM AND BEHAVIORISM: MERGING?
65
where p = 1 – (1 – x2)(1 – d).
5. and the other one that of the negative agency.10a) Ν1 Ν2 will be called reinforcement densities D1 and D2. The goal of this process is to form and preserve the state in which an organism has the ability to make an intentional probabilistic choice.8) and (4. B1 corresponds to the positive pole and B2 to the negative pole. The experiment consists of a series of sessions. (b) Variable x2 takes on a fixed value equal to or depending on the relative mean frequency of reinforcements in the previous sessions from the key.

The logarithmic graph C2. there is a breach. in a given session. corresponding to the reorientation of the poles. 1985). there is a breach.11) for local densities in the way similar to the one described by the model of melioration (Rachlin. the keys are polarized. The upper ray corresponds to the left key being the positive pole. Let us look at B1. Graphs A1 and A2 map the case. Let K1 and K2 be the numbers of pecks to the right and left keys. the right key is the positive pole. 2). One of them plays the role of the positive pole. At k1 < k2 the left key is the positive pole. At k1 = k2. when the right key plays the role of the posi tive pole and the left one the role of the negative pole. and the other that of the negative pole. We will call one key right and the other left. the subject behaves in such a way that this alternative is reinforced more often. (B) and (C). For those sessions in which k1 < k2. the ray lies below the diagonal in B2 and above it in C2. Because of that. (C) In each session. At k1 = k2. Each graph corresponds to a set of sessions (Fig. and N2 and n2 to the negative pole. At k1 < k2. during the entire set of ses sions Graphs B1 and B2 map the case. The left ray corresponds to the sessions with the left key being the positive pole. (B) In each session. as in B1.
6.7) and (4. Consider C1. in which the right or the left key corresponds to the positive pole only in those sessions in which it is not richer. Consider three possible relations between polarization and richness along the set of sessions. and k1 and k2 the numbers of corresponding rein forcements. In the framework of RIMS. It is possible that the subject controls equation (4. the graphs have a breach.9). as B2. When k1 > k2. the least rich alternative is the positive pole. 1973. Let us look at the differences between C2 and B2. the ray lies above the diagonal in B2 and below it in C2. The Patterns of Behavior Predicted by RIMS
We will call the alternative (key) richer if. Vaughan. Using functions (4. At k1 > k2. and the lower ray to the right one.
. consists of two rays. and the right ray to those with the right one. (A) One alternative is the positive pole in all sessions independently from be ing richer or not. At k1 > k2. Graphs C1 and C2 map the case when one of the keys (right or left) is the positive pole only in those sessions in which it is richer than the other. the right key is the positive pole.66
MATHEMATICAL MODELING OF REFLEXIVE PROCESSES
cate the strategy which the subject’s organism will choose to reach and hold equa tion (4. The logarith mic graph B2 consists of two rays going at the angle of 450 to the horizontal axis. the richer alternative is the positive pole. N1 and n1 relate to the positive pole.9) we will construct now a graph of K1/ (K1 + K2) depending on k1/(k1 + k2) and a graph of log(K2/K1) depending on log(k2/k1) for cases (A). the left key is the positive pole.

. 2. In real experi ments.LEFEBVRE. The value x2=1 means that the subject received all the preceding reinforcements from the key which is positive in the given session. (6.V. MENTALISM AND BEHAVIORISM: MERGING?
k2 k1+ k2
log
67
K2 K1
1 1
K1 K1+ K2
0 0
K2 K1+ K2
K1= N1 k1 = n1
log p {
K1= N1 k1 = n1
log
k2 k1
0 0
k1 k1+ k2
1 1
A1
k2 k1+ k2
K2= N1 k2 = n 1
A2
K log 2 K1
1 1
K1 K1+ K2
0 0
K2 K1+ K2
0 0
K1= N1 k1 = n1
K2= N1 k2 = n1
K1= N1 k1 = n1 k log 2 k1
1 1
k1 k1+ k2
B1
k2 k1+ k2
K1= N1 k1 = n1
B2
log
K2 K1
K2= N1 k2 = n1
K1 K1+ K2 K2= N1
k2 = n1
K2 K1+ K2 K1= N1 k1 = n1
log
k2 k1
k1 k1+ k2
C1
C2
Fig. The patterns of behavior predicted by RIMS
A ray shift up or down on logarithmic graphs A2. the subjects always have some experience in receiving reinforcements from the negative key. Thus.1) It is easy to see that p=1 only under condition that at least one of the values. we have to consider x2 < 1.A. x2 or d. B2 and C2 is predetermined by p = 1 – (1 – x2)(1 – d). as well. is equal to 1.

This pattern appears under condition that left and right alternatives differ essentially. A type of experimental graphs corresponding to pattern A
In the framework of RIMS we interpret this pattern as follows: the key corre sponding to B1 is the positive pole and c = p. Therefore (7. the left key is under schedule VI. Ν1 n1
(6. Williams. 3 corresponds to this case. if the subject has been led to the state with the maximal index of depression. For example. The Patterns Observed
Pattern A is well known.
(a) 1 (b)
log
B2 B1
B1 B1+ B2
log
r2 r1
0
r1 r1+ r2
1
Fig.1) r1 + cr2 B1 + B2 A type of experimental curves in Fig. In this case. (7. the index of depression is mini mal.68
MATHEMATICAL MODELING OF REFLEXIVE PROCESSES
Therefore. a set of sessions can be described by (7. 3. where c ≤ 1: B1 r1 = . and the right one under VR (see Baum. that is.1). the following equation is realized:
Ν2 n2 = x2 .1) can be written as N1 N1 + N2 = n1 n1 + pn2 .2)
may appear only at d = 1.3)
7.2). that is.2)
. 1988). the ideal correspondence
Ν2 n = 2 Ν1 n1
(6. (7. 1974. It is usually described as a case of β = 1 in (2. When d = 0.

corresponds to cases with β < 1. 5. 4. and VI one is the negative pole. We can see an example of it in the experiment by Baum and Aparicio (1999) where one alternative was worked on VR schedule with constant
. It is observed when alternatives do not differ but by the ratios of schedules. in which one key (say the left one) is controlled by schedule VI. 5). Fig. if we de scribe it with the Generalized Matching Law. 2002.LEFEBVRE. as Baum et al. (1999.
1
B1 B1+ B2
0 0
r1 r1+ r2
1
Fig.V. is a key fact for understanding the difference between utilitarian and deonto logical aspects in animal behavior. 1). This observation made Baum et al. Pattern B is also known well. MENTALISM AND BEHAVIORISM: MERGING?
69
An analysis of the experiments. 4 with RIMS we see that the positive pole corre sponds to the alternative which is less reinforced. An approximation of a broken graph (Fig.
(a) 1 (b)
log
B2 B1
B1 B1+ B2
log
r2 r1
0 0
r1 r1+ r2
1
Fig. noted. see also Baum. as we will show later. 2002). The appearance of a twist inherent to this law may be explained as an approximation of broken graph 4(a) by a continu ous power function (Fig. Pattern C can be seen rarely and only recently has been singled out as a special (Baum. This peculiar fact. A type of experimental graphs corresponding to pattern B
This pattern.A. 4a) by a continuous power function
By analyzing graphs in Fig. demonstrates that VR key is the posi tive pole. and the other (right) by VR.

70
MATHEMATICAL MODELING OF REFLEXIVE PROCESSES
mean ratio. The data of this experiments (rats 102. What are the experimental values for c in (7. 6:
(a) 1 (b)
log
B2 B1
B1 B1+ B2
log
r2 r1
0 0
r1 r1+ r2
1
Fig. 111. the appearance of a curve breach can be explained by approximation of graph 6(a) with a power function (Fig.
1 1
B1 B1+ B2
BVR BVR+ BVI
0 0
r1 r1+ r2
0 1 0
rVR rVR+ rVI
1
Fig. The experiment by Baum and Aparicio (1999) and their analysis demonstrates that pattern C can be reduced to pattern A. Approximation of broken graph Fig. Then graph 6(a) looks like Fig.6. if instead of location (right/left) alter natives are coordinated by schedules (VR/VI). We have to find an intersection point of a logarithmic graph with a
. 6(a)
Pattern C may appear in the experiments when a factor which determines a positive negative polarization of the alternatives in some sessions is connected with the left alternative and in the others with the right one. we may suppose that this pattern reveals itself in the experiments whose description with the Generalized Matching Law requires introduction of β > 1. As in pattern B.8. 120. 7). 6(a) – with a continuous power function
Fig.7.1)? It is easy to find them for patterns A. 213) can be represented in the graph in Fig. and the other one – on VI schedule and interval changing session by session. 8. A type of experimental graphs corresponding to pattern C
Following Baum (2002). Pattern (A). corresponding to pattern C in Fig.

82x. Most pattern B data were treated under assumption that the Generalized Matching Law holds.5. during the Civil War in Russia. The father finds a wealthy American. In RIMS c corresponds to p found from (4. In 1918. Then (4.
8. RIMS predicts that in this experiment. MENTALISM AND BEHAVIORISM: MERGING?
71
vertical axis in Fig. The thing was that the most important value for her that of self sacrifice contradicted to the American style of life.58.A. natural to her. c = 0.
Let us suppress our surprise (and perhaps. Therefore. (7. Mulia was confused. positive vs. we need to perform a linear approximation separately for the dots lying above the horizontal axis and those lying below it. Crane behavior was unacceptable for Mulia? This highly noble man helped fifty one families. Such a procedure was performed by Baum et al. with great difficulties. What feature in Mr.LEFEBVRE. Of course. We used their data and found the mean value of c = 0. the head of one family ends up in China. the same American helps them to move to the USA. his good deals were connected with money. while his wife is left in Russia with six children. a pigeon’s prognostic model of the future can be represented as function m(x) = 0. To find out how moral like evaluations in animals are connected with the utilitarian preferences.10) looks as follows: 0. when one key works un der VI schedule. finds children and brings all six of them to their father to China.59 = 1 – (1 – 1/2)(1 – d) . Mr. Crane “sacrificed” a lot of time and money to charity (he was helping 50 more families to stand on their feet!). 2003). be cause of this it is close to impossible to extract the value of c from them. somehow could not be applied to America. and the other under VR schedule. An idea of “living for others”. this is why his activity lacked “that special beau ty of self sacrifice. To find the value of c for pattern B using logarithmic data.V.3) from where d = 0. negative evaluations analogous to (and perhaps preceding evolution of) the human moral evaluation good vs.
Why America was not good for twenty two year old Mulia? Nostalgia? No. (1999).10). We will try to analyze this example from a nat uralist’s point of view. who with the risk to his life goes to Russia. In other words. Ten years later. bad. Here is a specific case. Animals Deontological Evaluations
We would like to put forth a hypothesis that animals have ability to make bipolar.18. But his activity did not carry that special beauty of total self sacrifice. she is executed.59. admiration of which Mulia got with her mother’s milk? (Pann. It is clear that he had to plan his activity and count money spent on each family. According to Williams (1988). 3(b). Let us find now the mean value of the depression index d for this experiment assuming x2 = 0.” It looks as if this young woman has an automatic mechanism which forbids combining utilitarian and deontological evaluations
. The oldest of the saved children feels disappointed there. indignation) of the lack of psycho logical grace in this young woman. let us consider ourselves first.

Let us emphasize that this happens if the alternatives differ only in the frequency of food delivery. Consider. reflects evaluations of the agencies related to the animal close biological needs.9) as the subject’s expenditures of appeals to the positive and negative agencies. building a cathedral) for smaller pay than they would require for similar work not connected with these values. deontological. animals’ “idealism” while being related to specific life important evaluations.
9. let us look at the patterns of behavior described in the previous sections. Sacred Shift
Altruism is understood as spending one’s own resources in order to help others. But this is not the only form of sacrificial behavior. for example. The alternative polarization opposite to their utilitarian preferences as food sources is analogous to the human act of puri fication. From this point of view. In developing this idea we can hypothesize that birds and mammals have two systems of evaluations. then the ratios
Ν1
n1
and
n2 N2
(9. but the second one is more safe (say. of separation good from practical profit. If some “non utilitarian” factor exists. In this case. Everyone may cite an example of people who agree to perform work connected with higher values (say. The first one contains more food. is connected with bipolar evalu ations positive negative on a larger time scale. hidden from view).0)
. We may hypothesize that this phenomenon reflects the work of the same mechanism which counterposes “dirty” money and “pure” intentions in humans. that is. and n1 and n2 as his revenues. whose work in humans reveals itself in a dramatic contrast between material and ideal values. If we interpret N1 and N2 in (4. is nevertheless separated from this minute preferences. which predetermines polarization of the alterna tives. We cannot exclude the possibility that this distinction is sup ported by a special mental mechanism. RIMS offers explanation for this phenomenon. Thus. while it only looks culturally conditioned. the agency richer with food gets evaluation “negative” while the poorer one receives “positive” evaluation. The first system.72
MATHEMATICAL MODELING OF REFLEXIVE PROCESSES
Let us make the next step and suppose that this mechanism has a deep biolog ical nature. then the polarization remains the same during the entire session indepen dently from reinforcement of the positive alternative. The second one. Pattern B appears when left and right food hoppers differ only in their frequency of food delivering. as we found. utilitarian. a hungry animal choosing between two feeders. In this case. This conclusion results from the interpretation of patterns A and C. Voluntary expenditure of one’s finances and energy related to creation and support of religious and moral symbols is another form of sacrificial behavior. the alternative which is less reinforced plays the role of the positive pole.

If this hypothesis proves to be correct. Aparicio.. But what about a fish or a bee. Overmatching in Rats: The Barrier Choice Paradigm. then the Match ing Law will turn into an operational criterion.
. J. Atkinson.J. 93 106. E. J. I also want to thank Victorina Lefebvre. We do not have an operational criterion which would allow us to find out whether a particular organism or a technical device has an inner world. Stefan Schmidt. Jane Ryan. as well. This belief is based only on our own subjective experience. it turns out that at average. can they suffer? In this work. and Crothers.
References
Adams Webber. to have a mental domain. are we observing an “inanimate” physical pro cess. (2001). 177 189. (1965). they can program themselves by downloading the values of probabilities with which they will make their choice.H. This conclusion is true not only for humans but for the rats and pigeons. Bower. I am grateful to my friends and colleagues.
Conclusion
We are convinced that we have a mental domain. without whose help this work would not be done. (1997).LEFEBVRE.V. MENTALISM AND BEHAVIORISM: MERGING?
73
express the mean payments the subject requires from the agencies for one appeal. We have also substantiated an assumption that the creatures with mental a domain can “self program.. It follows from (4. and Chris Weaver for valuable advice. We believe that a rat and a pigeon are able to suffer. New York: Wiley.
Acknowledgements
I am thankful to William Baum and James Mazur for important clarifications and to Robert Kessel and William Palya for critical comments. for which it holds. Journal of Social and Biological Structures.F. This problem extends beyond the framework of science and touches upon the core of our morality.C. we put forth a hypothesis that the mental domain in living creatures appears at the same time as their ability to make a probabilistic choice. the subject never requires more goods for one appeal from the positive agency than from the negative one. (1987).11) that n1 n2 ≤ . 10. PSYCOLOQUY. Furthermore. Or. G. Journal of the Experimental Analysis of Behavior. Adams Webber. we have shown that the Matching Law is an external demonstration of this self programming. An Introduction to Mathematical Learning Theory. (1995). Comment on Lefebvre’s Model from the Perspective of Personal Construct Theory. J.1) N1 N2 Therefore.” That is. 6(34).A. Adams Webber. American Journal of Psychology. on the other hand. A Pragmatic Constructivist Gambit for Cognitive Scientists. C. 75. (9. 527 541. 110. R. We will thus have grounds to con sider organisms. Self reflexion in Evaluating Others.

. V. (1995). In Wheeler. (1959). H. Lefebvre. (Ed. 92 101. (1990). (1994).. (1957). Self and Mathematics. 78. W. V. 231 242. Some Critical Issues in Lefebvre’s Framework for Ethical Cognition and Reflexion. 6 9. G. (2002). 71. (1961). In: Problemy Issledovania Sistem i Struktur. Lewiston.A.H. 1. Bower. Baum. 214 226. NM: Army Research Laboratory. 95 116. R. Stanford: Stanford University Press. New York: Peter Lang. LaBerge. Psychometrika. Choice Point Behavior. (Eds. Baum. Choice. 1 15. (1977a).A. H. Davidson. Lefebvre. (1974). (1979). & Schmidt. 64. XVII. On the Law of Effect. Journal of Social and Biological Structures. L. XXII. T. W. Journal of the Experimental Analysis of Behavior. K. Journal of Mathematical Psychology. S. Lefebvre. 1. M. Studies in Mathematical Learning Theory.: The Edwin Mellen Press. Beverly Hills: Sage. Undermatching. The Fundamental Structures of Human Reflexion..J. Lefebvre. 267 272. (1999). 13.): The Structures of Human Reflexion. Journal of the Experimental Analysis of Behavior. Krylov.M. (1977b). On Two Types of Deviation from the Matching Law: Bias and Undermatching. (1994).E.E. & Jones. 22. R. An Algebraic Model of Ethical Cognition. Journal of the Experimental Analysis of Behavior. 324 345. N.R. Decision Making. A formal Method of Investigating Reflective Processes. & Terry. 32.Yu. V. Baum.F. Suppes. S. (1992b).J. W.74
MATHEMATICAL MODELING OF REFLEXIVE PROCESSES
Audley. (1970). V. In: Bush. Los Alamos. E. (1999). 67. and Foraging Theory. Press.A. From Molecular to Molar: A Paradigm Shift In Behavior Analysis. Relative and Absolute strength of Response As a Function of Frequency of Reinforcement. Schwendiman. Applied Ergonomics.M. Terrorist Recruitment and Reflexive Control. V. On Self reflexive and Self organizing Systems. W. W. 21 24. Biometrika.A. D. Bradley. 269 281. The Good Samaritan: An Exemplary Narrative of Moral Choice. Davison. V. Journal of the Experimental Analysis of Behavior.K. (1967). 147 162. (1972). A Formal Approach to the Problem of Good and Evil. (1960).
. C. Journal of the Experimental Analysis of Behavior. & Estes. Lefebvre. (1987). and Overmatching in Studies of Choice. J. 355 373. Lefebvre. 39.Y. 36. R. 181 188. 100 128. Matching. Rank Analysis of Incomplete Block Design. Herrnstein. (2003).A. M. Baum. R. On One Model of Reflexive Behavior Distinct from Lefebvre Model.A. 243 266. Journal of the Experimental Analysis of Behavior.L. (1965). 375 396. & Aparicio. Konfliktuyushchie Struktury.D. A Recruitment Theory of Simple Behavior.A. A Psychological Theory of Bipolarity and Reflexivity. Moscow: Vysshaya Shkola. Lefebvre. General Systems. 83 120. Journal of Social and Biological Structures. 63 68. M. and Bell.. W. English translation in Applied Ergonomics. (1999). 3.H. Baker. P and Siegel. (1952). Moscow: Izdatelstvo AN USSR. 27. (1980). Kaiser. Lefebvre. A Stochastic Model for Individual Choice Behavior.M. The Method of Paired Comparisons I. R. Herrnstein. The Structure of Awareness: Toward a Symbolic Language of Human Reflexion. D.A. 4. W. V.A. 129 175. Stanford: Stanford Univ. Proceedings of the Workshop on Multi Reflexive Models of Agent Behavior. Lefebvre. 22. (1987).J. 10. General Systems.M. 75 89. Kauffman. Batchelder.A. W. Journal of the Experimental Analysis of Behavior. . 10. A Rational Equation for Attractive Proportions. Journal of the Experimental Analysis of Behavior. A Quantitative Analysis of Extreme Choice. 71. V. Psychological Review. V. Optimality and Concurrent Variable Interval and Variable Ratio Schedules.).B. (1962). Contingency Discrimination. Baum. Journal of Reflexive Processes and Control. Journal of Mathematical Psychology. 183 185. (1992a). M. B. V.

Y. Szabadi. American Statistical Association Journal. V.A. Dordrecht:Kluwer Academic Publishers.LEFEBVRE. MBS. D. Moscow: Tekst. D.A.A.W.
. Psychological Review Monographs Supplement. 420 428. (1987). Psychology of Judgment and Choice. E. (1960).2. 183 185. 55 67. Subjective Zeros. (1985). H. V. (1999a). (1990). M.V. Los Alamos. Human Reflexion and the Anthropic Principle.L. & Adams Webber. Popper.. May 24 25. V.A. NM: Army Research Laboratory. 2. 509 517. C.J. Journal of Social and Biological Structures. Poulton. E. New York: Wiley. The Law of Self Reflexion. K. 1. T. Restle. Proceedings of the Workshop on Multi Reflexive Models of Agent Behavior. Application of Generalized Reflexive Behavior: Models for Situation Specific Decisions.: The Edwin Mellen Press. M.V. Operant Matching. Thorndike. (1973). Mosteller. Behavior Theory and Learning. E. NM: Army Research Laboratory. MENTALISM AND BEHAVIORISM: MERGING?
75
Lefebvre. Appleton Century Crofts. Moscow: Cogito Centre. 371 404. Rachlin. J. (1938). Conceptual Models of Spatial and Non Spatial Selective Learning. Reflexive Processes and Control. Luce. V. Los Alamos.W. 37. 19 20. Los Alamos. Schreider. Perception & Psychophysics. (1959). Applied Ergonomics. F. 51 62.D. N. 1. (1951). Psychological Review. 95 108. McClain. 217 234.): The Structure of Human Reflexion. 1. Lewiston. N. and Moral Choice. Englewood Cliffs. (1992). A. Reflexive Model of Human Behavior: Variations on Lefebvre’s Theme. (1981). R. Lefebvre. (1951). 2. and Contraction Biases in Very First Judgments of Lightness. P. New York: Peter Lang. Pavlov. 214.P. Animal Intelligence: An Experimental Study of the Associative Processes in Animals. The Formalization of Ethical Values in Models of Reflexive Behavior. (1996). Los Alamos. E. In: Spence. Categorization. Schreider. An Experimental Measurement of Utility. 99 14. & Simmonds. Savage. PSYCOLOQUY. 91 100. J.8. V. Expanded Russian translation (2003). 29 40. Etika. 31. B. J. 69 94. Lefebvre. Oxford: Oxford University Press. and Ethics. Mazur. Spence. Conditioned Reflexes. and Bevan.S. Reflexion.A. Modeling.A.A. I. 2nd enlarged edition. Taran. In Wheeler. A Psychological Theory of Bipolarity and Reflexivity. Behaviour of Humans in Concurrent Schedules Programmed on Spatially Separated Operanda. K. Subjectively Equal Stimulus Spacing. (1961). Science. V. (Ed. 10. L. Journal of Reflexive Processes and Control. Comment on Vladimir Lefebvre’s Tonal Automata. J. Fuzzy Sets and the Structure of Human Reflexion. Optimization Theory Fails to Predict Performance of Pigeons in a Two Response Situation. No.A. & Sulcoski. Pann. Novoye Russkoye Slovo. J. (1998).” Journal of Social and Biological Structures. D. (1999b). Quarterly Journal of Experimental Psychology. (1979). F. L. 10. Miller. & Sulcoski. NM: Army Research Laboratory. Skinner.. Ruddle. Sketch of Reflexive Game Theory. Proceedings of the Workshop on Multi Reflexive Models of Agent Behavior. Rapoport. UCI. The Oldest Daughter.A. (1999a). 1. (1987). Function of Fast Reflexion in Bipolar choice. Rapoport. The Journal of Political Economy. (2002). Lefebvre. The Behavior of Organisms: An Experimental Analysis.). (1999).. (2001). E. (2003). Individual choice Behavior: A Theoretical Analysis. Miller.C. (1927). 59. Algebra of Conscience. Lefebvre. 80. 1 40. L. L. 46. Niels Bohr’s Atom and Vladimir Lefebvre’s “Inner Computer. (1911). F.A. Contrast and Matching. (1999b). (Ed. Proceedings of the Workshop on Multi Reflexive Models of Agent Behavior.: Prentice Hall. & Nogee. K.M. Proceedings of the Workshop on Multi Reflexive Models of Agent Behavior.F. Levitin. (1994). Bradshaw. Vol. 823 825. (2002). Institute for Mathematical and Behavioral Sciences. NM: Army Research Laboratory. H. A. Note in the Appendix to Lefebvre. New York: Wiley.. 204 212. 6(37). L. H. The Theory of Statistical Decision.F. P.

Zajonc.
. 161 166. R. Reinforcement. H. 43. L. Journal of Social and Biological Structures. W. Comment on Interpersonal Affiliation and the Golden Section. Lindzey. 10. von Neuman.J. and Response Strength. 461 471.. (Eds. Townsend. & Burgess. In: Atkinson. (1983). (1987).B. Purposive Behavior in Animals and Men. Journal of Mathematical Psychology. Psychology Review.A. J.) Steven’s Handbook of Experimental Psychology. 2.): The Structures of Human Reflexion. Algebra of Conscience (book review). (1987). A Constructional Biology of Hermeneutics. Vaughan. (1990). (1947). 103 123. Princeton: Princeton University Press. J.T. J. (1982). (1985). & Morgenstern.D. 27. 34. (1988).2. Wheeler. Journal of the Experimental Analysis of Behavior. Wearden. H. 28. O. 10. (1927). E. B. 339 348. Lefebvre.. In Wheeler. Tolman. Vol.L. Theory of Games and Economic Behavior. 273 286. Luce.H.T. 212 214. J. 167 244. Choice. New York: Appleton Century Crofts. R. Lefebvre’s Human Reflexion and Its Scientific Acceptance in Psychology.76
MATHEMATICAL MODELING OF REFLEXIVE PROCESSES
Thurstone. A Law of Comparative Judgment. (Ed.C. New York: Peter Lang. Vladimir A.C. Townsend. 383 405. G. Journal of Experimental Analysis of Behavior. Herrnstein. Journal of Social and Biological Structures. R. Matching Since Baum (1979).S. Choice: A Local Analysis. R. (1932).. I. Williams.