A Constraint Satisfaction Model of Cognitive Dissonance Phenomena

A constraint satisfaction network model simulated cognitive dissonance data from the insufficient justification and free choice paradigms. The networks captured the psychological regularities in both paradigms. In the case of free choice, the model fit the human data better than did cognitive dissonance theory. Cognitive Dissonance Cognitive dissonance theory (Festinger, 1957) has been a pillar of social psychology for some 30 years. The theory holds that dissonance is a psychological state of tension which people are motivated to reduce. Two cognitions are dissonant when, considered by themselves, one of them follows from the obverse of the other. The amount of dissonance is a function of the ratio of dissonant to consonant relations, with each relation weighted by its importance. Dissonance can be reduced by decreasing the number and/or the importance of the dissonant relations, or by increasing the number and/or the importance of consonant relations. How dissonance gets reduced depends on the resistance to change of the relevant cognitions, with less resistant cognitions being more likely to change. Resistance derives from the extent to which change would produce new dissonance, the degree to which the cognition is anchored in reality, and the difficulty of changing those aspects of reality. Festinger (1957) used dissonance theory to account for a number of existing psychological phenomena, including the evaluation of choices, attitude change following attitude-relevant actions, and responses to the disconfirmation of beliefs. It has since been successfully applied in a wide variety of both predictive and postdictive contexts. Consonance Model In this paper, we present a computational model of cognitive dissonance. The model is based on the idea that dissonance reduction is a constraint satisfaction problem. Such problems are solved by the simultaneous satisfaction of many soft constraints which can vary in their relative importance. In this framework, beliefs are represented as units in a network and implications among the beliefs are represented as connections among the units. The units can be variously active and the connections (weights) can vary in strength. Hopfield (1982, 1984) has worked out the mathematics for solving such constraint satisfaction problems in parallel networks. Hopfield networks are capable of simulating a variety of psychological phenomena, including belief revision, explanation, schema completion, analogical reasoning, and content-addressable memories (Holyoak & Thagard, 1989; Rumelhart, Smolensky, McClelland, & Hinton, 1986; Thagard, 1989). Unless used to model memory, these networks are generally considered ephemeral in the sense that they are created on line to deal with some particular task, although the creative process is not usually modeled. Hopfield networks function by reducing energy (equivalently, maximizing goodness) subject to the constraints supplied by the connections and any external input. Our Consonance Model for reducing cognitive dissonance is a Hopfield network lacking some of the parameters of other Hopfield networks and introducing some special parameters of its own. Maximizing the consonance (goodness) of any pair of connected units depends on the sign of the connection between them. Assume an activation range of 0 to 1. If connected by a positive weight, both units should be active in order to maximize consonance. With a negative weight, consonance is maximized when both units are not active, that is, when both are inactive or only one is active.

[1]  J. Brehm Postdecision changes in the desirability of alternatives. , 1956, Journal of abnormal psychology.

[2]  L. Festinger,et al.  A Theory of Cognitive Dissonance , 2017 .

[3]  F. Heider The psychology of interpersonal relations , 1958 .

[4]  Elliot Aronson,et al.  The effect of severity of initiation on liking for a group. , 1959 .

[5]  Elliot Aronson,et al.  Effect of the severity of threat on the devaluation of forbidden behavior. , 1963 .

[6]  J. Freedman Long-term behavioral effects of cognitive dissonance. , 1965 .

[7]  P. Tannenbaum,et al.  Theories of cognitive consistency: a sourcebook. , 1968 .

[8]  J J Hopfield,et al.  Neural networks and physical systems with emergent collective computational abilities. , 1982, Proceedings of the National Academy of Sciences of the United States of America.

[9]  J J Hopfield,et al.  Neurons with graded response have collective computational properties like those of two-state neurons. , 1984, Proceedings of the National Academy of Sciences of the United States of America.

[10]  J. Hopfield Neurons withgraded response havecollective computational properties likethoseoftwo-state neurons , 1984 .

[11]  Geoffrey E. Hinton,et al.  Schemata and Sequential Thought Processes in PDP Models , 1986 .

[12]  Paul Thagard,et al.  Analogical Mapping by Constraint Satisfaction , 1989, Cogn. Sci..

[13]  Anders Krogh,et al.  Introduction to the theory of neural computation , 1994, The advanced book program.

[14]  P. Thagard,et al.  Explanatory coherence , 1993 .