Cognitive Response, Imagery, and Scripts: What Is the Cognitive Basis of Attitude?

Bobby J. Calder, Northwestern University
ABSTRACT - It is argued that cognitive response theory must be extended to take into account memory structure and probably visual imagery coding as well. Moreover, the notion of scripts suggests a qualitatively different view of cognitive response.
[ to cite ]:
Bobby J. Calder (1978) ,"Cognitive Response, Imagery, and Scripts: What Is the Cognitive Basis of Attitude?", in NA - Advances in Consumer Research Volume 05, eds. Kent Hunt, Ann Abor, MI : Association for Consumer Research, Pages: 630-634.

Advances in Consumer Research Volume 5, 1978      Pages 630-634


Bobby J. Calder, Northwestern University


It is argued that cognitive response theory must be extended to take into account memory structure and probably visual imagery coding as well. Moreover, the notion of scripts suggests a qualitatively different view of cognitive response.


It has become increasingly clear that to understand how attitudes are related to consumer behavior we must first understand better how attitudes themselves arise. The classical definition of attitudes as relatively enduring predispositions to respond has outlived its usefulness. Attitudes are often, perhaps usually, not enduring. They derive from what the individual is thinking about at a point in time (see, for example, Salancik, 1974, 1976; Salancik and Conway, 1975). An individual may express quite different attitudes toward the same object or behavior at different times. Not only this, equally positive expressions of attitude may be differentially related to behavior depending on the nature of their underlying thoughts (Regan and Fazio, 1977).

The purpose of this paper is to lay out what I believe are the emerging issues in understanding the cognitive basis of attitude. More questions will be raised than answered. The goal is to provide a sense of direction rather than a complete theory.

Research on the Fishbein attitude-belief model has provided a strong impetus, particularly in consumer research, to investigating the cognitive basis of attitude. The contribution of this research has been to demonstrate convincingly that attitudes are not based on a simple list of beliefs, where the beliefs are "objective" in the sense that different individuals process pretty much the same beliefs (cf. Calder, 1975). The Fishbein model is an improvement over the classic assumption originating with Hovland that attitudes are based on the rote learning of completely external (message) information. But it appears necessary to go beyond the Fishbein model to allow for an even more complex and subjective basis of attitude. It should be noted that, from this perspective, attempts merely to shore up the model by adding other variables are ill-advised.

Currently the most promising direction for research is cognitive response theory. As Lutz and Swasy (1977) have shown, cognitive response theory is essentially compatible with and a step beyond the Fishbein model. A review of cognitive response studies is provided by Wright (in press). The focus here will be on the status of the theory and how it might profitably be extended. It is argued that the theory must be elaborated to take into account the structure of human information processing. A subsequent section attempts to look beyond the present concept of cognitive response. It may be necessary to allow for alternative memory codes, such as imagery, as well as structure.


The new idea of cognitive response theory is that beliefs, or the thought underlying attitudes, cannot be treated as "objective." The beliefs which are processed to yield an attitude are not just those originating in external communications; nor can they be in any sense standardized across people. Beliefs must be treated as any thought which might come to mind in a situation. The new idea of cognitive response is simply the old idea of cognitive mediation updated to a view of the person as an active information processor. Cognitive mediation is held to be better reflected in the idiosyncratic thoughts listed by a person as coming to mind after a message than by measures of message recall or standard lists of beliefs.

The cognitive response variable has stimulated research because it predicts interesting effects. For instance, Brock and his colleagues (cf. Petty, Wells, and Brock, 1977; Petty, 1977) have demonstrated that distraction can increase persuasion by inhibiting the negative thoughts which would otherwise arise with a counter-attitudinal message and that distraction can decrease persuasion by inhibiting the favorable thoughts which would otherwise accompany an attitudinally consistent message. Another illustrative effect is source credibility. In a series of studies, Sternthal (cf. Dholakia and Sternthal, in press) finds that a low credibility source produces more attitude change than a high credibility source when a person's initial attitude is already favorable. The low credibility source stimulates positive thoughts.

While such effects are intriguing, notice that there is no theoretical explanation involved beyond that of cognitive mediation. A variable's effect on persuasion is explained by whether it inhibits or stimulates cognitive response and whether positive or negative thoughts are a priori more likely. Whether positive or negative thoughts are more likely can be inferred somewhat loosely from the person's initial attitude or from the nature of the message (see Figure 1). Cognitive response is thus a mediator variable expected to proceed attitude change.

No doubt the notion of cognitive response mediation will suggest additional interesting effects. It is my opinion, however, that we must also begin to examine the nature of cognitive mediation as well as the fact of it.



My research has been concerned with how the structure of cognitive mediation affects persuasion. The most striking aspect of this structure is limitation. Evidence from cognitive psychology indicates that mediation is realized structurally by a limited capacity short-term memory store and a long-term memory store. The consensus model is that short-term memory holds information which is actively being processed. Long-term memory is a larger store of most, if not all, of the information that a person has ever processed. To be processed further, the information in long-term memory must be retrieved and transferred to short-term memory. The basis for this retrieval is the content of short-term memory of any point.

In terms of cognitive response theory, this model implies that information from a message, along with other incoming information, is initially represented in short-term memory as cognitive responses. These cognitive responses in turn trigger retrieval from long-term memory and the registration in short-term memory of further cognitive responses (see Figure 2). Note that any information in long-term memory is a potential cognitive response depending on what is triggered by short-term memory. But, since short-term memory is limited in its capacity, only so many cognitive responses can be represented.



This leads to a prediction: When a message involves unfamiliar material and there is little time for rehearsal, cognitive responses to that message will be limited by the capacity of short-term memory. If the capacity of short-term memory is exceeded, additional cognitive responses cannot be represented, and thereby cannot affect attitude.

I have tested this prediction in a series of experiments which vary the amount of information in a message. Let me describe the first of these experiments. A two-sided message was used in which the pro side consisted of either one, seven, or fourteen distinct arguments and the con side of either one or seven arguments. The arguments were testimony in a jury trial. They were equated in style and content and were counterbalanced in the experimental design. After reading the message, subjects went through a thought listing procedure and indicated their overall attitude.

Without considering cognitive structure, one would expect that the number of pro cognitive responses would increase with the pro arguments and the number of con cognitive responses would increase with the number of con arguments. Attitude would thus be a linear function of the number of arguments (see Figure 3). If the number of cognitive responses is constrained by short-term memory, however, a different pattern would be expected. Beyond some point, increasing the number of arguments on one side of the message should not affect persuasion.



In a limited capacity short-term memory, cognitive responses favorable to the pro side must in general be represented at the expense of potential cognitive responses favorable to the con side. So the number of pro cognitive responses will increase with the number of pro arguments only to the extent that the number of con cognitive responses can be reduced. At the point where the con cognitive responses cannot be further reduced, or none are left, then further pro cognitive responses cannot be represented, and more pro arguments will have no impact.

The results of this experiment indicated that persuasion did level off with increasing number of pro arguments (see Figure 3). With one con argument, increasing the pro arguments from one to seven yielded more persuasion. The thought listing data showed that the additional pro arguments were represented at the expense of con cognitive responses. Increasing the pro arguments from seven to fourteen, however, yielded no further persuasion. The number of pro and con cognitive responses remained the same as with seven pro arguments, additional pro cognitive responses were not represented.

With seven con arguments, increasing the prosecution arguments to fourteen also yielded persuasion (see Figure 3). Changes in the number of pro arguments were reflected in changes in the number of pro cognitive responses at the expense of con cognitive responses. The reverse occurred when the number of pro arguments was decreased to one. Con cognitive responses were represented at the expense of pro cognitive responses. I believe this data demonstrates rather clearly the importance of structure to cognitive response theory.

To illustrate further the importance of structure, I would like to describe one other theoretical implication. The more extensively information is rehearsed in short-term memory, the more likely it is to enter long-term memory. When a message is received, cognitive responses to the initial part should be rehearsed longer and therefore enter long-term memory. Cognitive responses to subsequent parts should be less likely to enter long-term memory. Now if attitudes are formed just after a message is received, two classes of cognitive responses are most likely to be represented in short-term memory. Cognitive responses from the end of the message are more likely because they should still be in short-term memory. And cognitive responses from the initial part of the message are more likely because they can be retrieved from long-term memory. Cognitive responses arising from the middle of the message should be relatively less available. This provides a more dynamic view of cognitive response. Not all cognitive responses are likely to mediate persuasion.

This implication was tested in a study conducted with Jerry Salancik. Subjects were asked to go through a long questionnaire checking religious behaviors which applied to them. Their attitudes toward religion were measured either before or after doing this. The idea was that responding to the questionnaire constituted a type of self-perception message. The effect of this message was indicated by a higher correlation between engaging in religious behaviors and expressing a religious attitude when the behavioral items were measured before rather than after attitudes. Reviewing one's behavior affected the attitude reported.

Several studies have shown this effect. What interested us was viewing the behavioral items as a message which would elicit cognitive responses. This message could be partitioned into parts by simply dividing the list of items into quarters. For each quarter we then computed a separate index of the number of religious behaviors checked. While we did not directly assess cognitive responses, we reasoned that they should reflect the items checked on each part of the list -- so that the four indices would reflect first quarter cognitive responses, second quarter cognitive responses, and so on.

The four indices were correlated for before and after subjects with attitudes. The results are shown in Figure 4. There were no significant differences in the correlations for subjects who gave their attitudes before their behaviors. But after subjects displayed a marked serial position effect. The indices for the first quarter and the last quarter correlated most with attitudes. We conclude that more cognitive responses stimulated by the beginning and the end of the message than by the middle must have been represented in short-term memory. Since the behavior items were randomized, we can think of no other explanation besides the memorial effect.

Important methodological issues are also raised by this structural view of cognitive response theory. One is that the thought listing procedures used in cognitive response studies themselves depend on memory. Thought listing cannot be treated as automatically dumping the short-term memory contents which yield attitude.

If other activities occur after a message and before thought listing, thought listing procedures must necessarily tap long-term memory. The original contents of short-term memory will be erased by the other activities. Moreover, cognitive responses which are represented in short-term memory after a message and which do affect attitude will nonetheless be unavailable for thought listing if they have not been rehearsed enough to be placed in long-term memory. And retrieval of the thoughts which are available will be biased by the cues present at the time of thought listing. As Nisbett and Wilson (1977) have contended for other kinds of reports on mental processes, subjects may not be able to report very well on what has affected their attitudes, and there may be good reasons why they cannot.




In a 1975 paper on the cognitive basis of Fishbein and other multi-attribute models, I pointed out that, in addition to ignoring memory structure, these models also implicitly assume that beliefs are based on linguistic verbal codes. Memory codes refer to the format in which information is stored. While it might seem that an issue as basic as memory codes should be left to the cognitive psychologist, there is a growing realization that an understanding of the cognitive basis of attitudes may require consideration of alternative memory codes as well as memory structure.

The most interesting possibility is that attitudes are based on visual imagery coding as well as verbal coding. Visual imagery is not simply pictures, or even mental pictures, though the term "quasi-pictorial" has descriptive value. It must be remembered that visual images are not objective or external, as are pictures. Imagery is a mental format for representing pictorial information. This format is thought to be the same whether it comes from visual sensation (i.e., seeing something) or from long-term memory (Hebb, 1968). Pictures are the referents of images.

If images are not mental pictures, what are they? Therein lies a controversy in cognitive psychology. Some theorists (c.g., Anderson and Bower, 1973; Clark and Chase, 1972; Polyshyn, 1973) argue against viewing visual imagery as an alternative code. The most influential argument is that of Polyshyn. He claims that all information is represented in a propositional format. Propositions are abstract logical relationships. Images and even verbal statements are surface manifestations of underlying propositions, which are themselves unavailable to consciousness. Any model of behavior must be based on propositional codes. According to this argument, visual imagery would be accorded no role in attitude theory. The experience of imagery is "epiphenominal," that is, a mental by-product. Kosslyn and Pomerantz (1977), however, have provided a convincing rebuttal to this argument. They contend that, even if propositional coding does underlie both imagery and verbal statements, the latter have emergent properties which propositions lack. Although imagery may be constructed from propositions, the properties of visual imagery are necessary in accounting for behavior. Moreover, they doubt the necessity of even postulating underlying propositions. Contrary to Polyshyn, only a set of transformational rules, not a common propositional code, is necessary. These rules would specify how imagery is mapped into verbal coding.

At this point, the construct of visual imagery does seem useful in explaining many empirical findings (cf. Kosslyn and Pomerantz, 1977). A dual code (imagery and verbal coding) appears more reasonable than a propositional one. But, as this controversy points up, though the construct of imagery coding has some integrity, it is very fuzzy. Even so, it seems to me that we should begin to amend cognitive response theory to allow for imagery as well as verbal coding.

Dual-code cognitive response theory will remain pre-theoretical, however, until we can move beyond a quasi-pictorial definition of imagery. Postulating the existence of imagery coding is in itself no help in understanding attitudes. Fortunately, a recent theoretical notion called "scripts," which is attracting wide attention in social psychology, may be helpful in further developing a dual-code cognitive response theory.


Abelson (1976, p. 41) proposes that attitude toward an object is an "ensemble of scripts concerning that object." A script is the expectation of a sequence of events learned from direct or vicarious experience. Scripts are composed of vignettes which are the individual events, or frames, of the sequence. Vignettes generally have both a visual imagery and verbal component. Abelson uses the shorthand of "picture plus caption" to describe them. Vignettes are linked together in a coherent and causal chain to yield a script. Continuing the metaphor, the script is a cartoon strip where the separate panels combine to tell a story.

Scripts may be more or less concrete. At the most concrete level are episodic scripts. They reflect single experiences. The features of multiple experiences may be synthesized into categorical scripts. Still higher levels in which scripts are reduced completely to abstract features with no episodic character are also possible.

As an illustration, consider my experience in shopping at Marshall Field on a Saturday. What I have stored is a categorical script consisting of four vignettes: I-need-something, and the subsequent it-takes-forever-to-find-a-parking-place, followed by pushing-through-wall-to-wall-people, and finally not-being-able-to-find-a-sales-clerk. Each caption being accompanied by a suitable picture. Abelson's proposal is that it is this script, as well as any others which might come to mind, which determines the attitude I express toward the Marshall Field store. Processing this script would tend to yield a negative attitude.

The script proposal is certainly in need of further theoretical specification. I believe it suggests, however, an interesting way of relating imagery to attitudes. A product might conceivably conjure up any number of images. The script notion provides a rationale for specifying which images will be important -- those that fit together coherently to represent relevant experiences.

More generally, it seems to me that it may be useful to view cognitive responses as scripts. The usual concept of cognitive response may be far too abstract. Cognitive responses are taken to be general thoughts about good or bad features where these thoughts are removed from any context. As scripts, cognitive responses would be firmly anchored in the stored experience of the individual. This is, of course, consistent with the major premise of cognitive response theory which is that persuasion in fundamentally self-persuasion and is, as contended by Tybout, Sternthal, and Calder (in press), initiated by self-perception.

To make the distinction between the usual view of cognitive response and script cognitive response clearer, consider an example. Suppose a woman sees a commercial for a new brand of shampoo. The usual view is that her attitude is based on thoughts about the features of the product plus any other general ideas that come to mind such as that most new cosmetic products are not very different. The script view is that her attitude is based on concrete event sequences. One cognitive response might be the categorical script ads-for-new-shampoo-pro-ducts-have-caught-my-eye-before, followed by I-am-always-disappointed-when-l-try-them. Another might be the episodic script maybe-this-shampoo-would-make-my-hair-shinier, followed by my-husband-admiring-my-hair. The script view is more intensely autobiographical and far more concrete. Features and ideas do not exist in isolation but in causal chains of expectations.

The concrete character of scripts cannot be overemphasized. Nisbett, Borgida, Crandall, and Reed (1976) relate scripts to people's preference for specific information. Their example is a man deciding between a Volvo or a Saab. He might encounter all kinds of information. He might read an article in Consumer Reports based on a large sample indicating that Volvo has a better record. Or he might hear about someone's brother-in-law's who has bad experience with a Volvo: "A Volvo! You've got to be kidding. My brother-in-law had a Volvo. First, that fancy fuel injection computer thing went out. 250 bucks. Next he started having trouble with the rear end. Had to replace it. Then the transmission and the clutch. Finally sold it in three years for junk." (1975, p. 129). The brother-in-law-information seems more persuasive, and there is research to support that people are biased toward such specific information. This would be expected if attitudes are based on script cognitive responses.


Two directions for extending cognitive response theory have been proposed. Cognitive response theory must take into account memory structure and probably visual imagery coding. Beyond this, the notion of scripts also calls attention to the need to view cognitive responses as ordered, contextual, and autobiographical. Scripts are a qualitatively different kind of memory. Given these complexities, it is not surprising that progress has been slow, and will continue to be slow, in understanding the cognitive basis of attitude.


Robert P. Abelson, "Script Processing in Attitude Formation and Decision Making," in J. Carroll and J. Payne (eds.), Cognition and Social Behavior (Hillsdale, NJ: Erlbaum, 1976).

John Anderson and Gordon Bower, Human Associative Memory (New York: V. H. Winston & Sons, 1973).

Bobby J. Calder, "The Cognitive Foundations of Attitudes: Some Implications for Multi-attribute Models," in M. Slinger (ed.), Advances in Consumer Research, Vol. II, 1975, 241-248.

Herbert Clark and William Chase, "On the Process of Comparing Sentences Against Pictures," Cognitive Psychology, 3 (1972), 472-517.

Ruby Dholakia and Brian Sternthal, "Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?," Journal of Consumer Research, in press.

Donald Hebb, "Concerning Imagery," Psychological Review, 75 (1968), 466-477.

Stephen Kosslyn and James Pomerantz, "Imagery, Propositions, and the Form of Internal Representations," Cognitive Psychology, 9 (1977), 52-76.

Richard Lutz and John Swasy, "Integrating Cognitive Structure and Cognitive Response Approaches to Monitoring Communication Effects," in W. Perreault (ed.), Advances in Consumer Research, Vol. IV, 1977, 363-371.

Richard Nisbett and Timothy Wilson, "Telling More than We Know: Verbal Reports on Mental Processes," Psychological Review, 84 (1977), 231-259.

Richard Nisbett, Eugene Borgida, Rick Crandall, and Harvey Reed, "Popular Induction: Information is not Necessarily Informative," in J. Carroll and J. Payne (eds.), Cognition and Social Behavior (Hillsdale, NJ: Erlbaum, 1976).

Richard E. Petty, "The Importance of Cognitive Responses in Persuasion," in W. Perreault (ed.), Advances in Consumer Research, Vol. IV, 1977, 357-362.

Richard E. Petty, Garry Wells, and Timothy C. Brock, "Distraction Can Enhance or Reduce Yielding to Propaganda: Thought Disruption versus Effort Justification," Journal of Personality and Social Psychology, (1977).

Zenon Polyshyn, "What the Mind's Eye Tells the Mind's Brain: A Critique of Mental Imagery," Psychological Bulletin, 80 (1973), 1-24.

Dennis Regan and R. Fazio, "On the Consistency between Attitudes and Behavior: Look to the Method of Attitude Formation," Journal of Experimental Social Psychology, 13 (1977), 28-45.

Gerald R. Salancik, "Inference of One's Attitude from Behavior Recalled Under Linguistically Manipulated Cognitive Sets," Journal of Experimental Social Psychology, 10 (1974), 415-427.

Gerald R. Salancik, "Extrinsic Attribution and the Use of Behavioral Information to Infer Attitudes," Journal of Personality and Social Psychology, 34 (1976), 1302-1312.

Gerald R. Salancik and Mary Conway, "Attitude Inferences from Salient and Relevant Cognitive Content about Behavior," Journal of Personality and Social Psychology, 32 (1975), 829-840.

Alice M. Tybout, Brian Sternthal, and Bobby J. Calder, "An Integrative View of Cognitive Response and Self-Perception Theory," in Advances in Consumer Research, Vol. IV, in press.

Peter Wright, "Research on Ad-stimulated Thought Processes,'' Journal of Consumer Research, in press.