An Experimental Investigation of Intrusion Errors in Memory For Script Narratives

ABSTRACT - The present study serves to extend previous research in script processing. While considerable research in psychology indicates that the reliance upon schemata can significantly influence cognitive processes, the existence and use of scripts by consumers in a marketing context has yet to be demonstrated.

Citation:

John C. Whitney and George John (1983) ,"An Experimental Investigation of Intrusion Errors in Memory For Script Narratives", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 661-666.

Advances in Consumer Research Volume 10, 1983      Pages 661-666

AN EXPERIMENTAL INVESTIGATION OF INTRUSION ERRORS IN MEMORY FOR SCRIPT NARRATIVES

John C. Whitney, University of Wisconsin-Madison

George John, University of Wisconsin-Madison

ABSTRACT -

The present study serves to extend previous research in script processing. While considerable research in psychology indicates that the reliance upon schemata can significantly influence cognitive processes, the existence and use of scripts by consumers in a marketing context has yet to be demonstrated.

As an initial effort to assess the usefulness of scripts to marketers, the present study developed scripts at the brand level and assessed the cultural uniformity of events as well as the serial ordering of events within each script. These scripts were then employed to evaluate their use by consumers. Intrusion errors in recall were observed for events central to the meaning of the script story when their location was manipulated.

INTRODUCTION

A mode of processing which is not inconsistent with the limited information processing suggested by Kassarjian (1978; also Olshavsky and Granbois 1979) is that of schematic processing. While schematic processing has exhibited considerable influence in psychology, it has drawn little attention in marketing (although see Calder and Schurr 1981; John and Whitney 1982; Rethans and Taylor 1982). This mode of processing suggests that perceptions, interpretations and inferences may be significantly affected by certain knowledge structures. The focus of this study will be upon "script processing," which is a type of schema, differentiated by the hierarchical or causal ordering of its elements.

Despite the popularity of the script concept in psychology, there has been no evidence of scripts in 8 consumer context. It is essential that the existence and homogeneity of scripts be established empirically before the marketer can utilize this body of knowledge strategically.

Bower, Black and Turner (1979) have demonstrated that it was possible to elicit scripts about a restaurant episode that exhibited high inter-subject agreement about the main events. However, they did not examine the degree of agreement regarding the sequencing of the listed events.

In the present study, we extent the Bower, Black and Turner approach to investigate the degree of inter-subject agreement relative to listed events and their hierarchical ordering for scripts at the brand level. In addition, the reliance upon scripts is demonstrated by examining the recall of non mentioned script events. The paper is organized in three sections. First, the extant literature is reviewed to indicate the possible effects of schematic (and script) processing. The data collection, analysis and results are then presented, followed by a discussion of the implications of the study.

LITERATURE REVIEW

Scripts have evolved as a variation of schema theories which have exhibited considerable influence in cognitive psychology, social psychology, cognitive anthropology and artificial intelligence (e.g., Bartlett 1932; Bransford and Franks 1971; Mandler 1979; Minsky 1975; Rumelhart and Ortony 1977; Schank and Abelson 1977). E ese theories have been proposed as a means by which individuals deal with the redundancy of information in the environment. Rather than attending to all stimuli, computing time can be minimized and information search guided by the reliance upon stored, generic concepts. events. event-sequences, etc., in a "schema."

According to Taylor and Crocker (1981) a schema "is a cognitive structure which consists in part of a representation of some defined stimulus domain. The schema contains general knowledge about the domain, including the specification of the relationships among its attributes, as-well as specific examples or instances of the stimulus domain" (p. 3). It is the basic premise of this research stream that abstract schematic concepts of how the world works often determine a "hypothesis-driven" form of information processing by individuals. Neisser (1976) has referred to this as "schematic processing" and has made an analogy between it and computer programming. Like the format statement, which allows information to enter the system, a schema enables a person to take information in. Thus, we are unable to recognize t chair unless we have a conception or schema about what a chair is.

Rumelhart and Ortony (1977) have argued that all cognitive processes perception, comprehension and interpretation involve an interaction with existing cognitive representations as opposed to operating solely on objects in the environment (also see Calder and Schurr 1981). Thus, schemata are by no means static, but are the product of a continuous cognitive process which changes with experience.

Empirical evidence has ben provided which suggests that individuals frequently draw upon different schemata for interpretative or understanding functions. These studies have shown that the recall of schema-relevant material is superior to the recall of schema-irrelevant material and that structure may be provided to ambiguous situations (Taylor, et al. 1978; Bransford and Johnson 1972; Dooling and Mullet 1973).

Once the understander has imputed meaning to a given stimulus and configuration via the interpretive or understanding function of schemata, he is facet with the problem of how to use this meaning to solve problems, set goals, or select a behavior (Taylor and Crocker 1981). Schemata can also be utilized to accomplish these latter functions. This may be accomplished by relying upon schemata to fill in missing data for a stimulus configuration (Minsky 1975) or by providing direction for further information search (Garland, Hardy and Stephenson 1975). Other studies supporting this inferential function of schemata include Cohen (1977), Bower, Black and Turner (1979), Markus (1977), Fiske and Kinder (1978), Cantor and Mischel (1977), Snyder and Uranowitz (1978), and Loftus and Palmer (1974).

Finally, it should be recognized that the reliance upon schemata may lead to a bias during the encoding of information or its retrieval from memory. These biases result primarily from the stereotypical expectations associated with schemata (Snyder and Uranowitz 1978; Markus 1977).

Scripts (Abelson 1975, 1976, 1980; Schank and Abelson 1977) are considered to be a type of schema and may serve to accomplish the same functions. They include routine, well-practiced event-sequences or stories such as departmental meetings, cocktail parties, or going to a restaurant. According to Schank and Abelson, a script is wa structure that describes an appropriate sequence of events in a particular context. A script is made up of slots and requirements about what can fill those slots. The structure is an interconnected whole, and what is in one slot affects what can be in another" (1977, p. 41). Abelson defines a script as ". . .a coherent sequence of events expected by the individual" (1966, p. 34). In effect, scripts are a stereotyped sequence of actions in which the major distinction from schemata is their causal and often temporal nature. mat is, early events in the sequence produce or at least enable the occurrence of later events (Nisbett and Ross 1980). Thus, a person may have a restaurant script which consists of 'entering,' 'being seated,' 'ordering,' 'eating,' 'paying' and 'existing' scenes. Note that certain schemata, such as that of the "waiter" may represent significant components of the script and convey a great deal of information.

It is important to note the distinction between scripts and habit or "mindless behavior" (Langer, Blank and Chanowitz 1978; Langer and Newman 1979). Unlike these latter behaviors, the act of thinking occurs in script processing. In the restaurant script of Bower, Black and Turner (1979), one must "think" when deciding what to order and where to sit. Further, a script is not the conceptual equivalent of a story (Martin 1980). Whereas the story may be embellished with unique details relating to a specific experience, the script is the essential core of the story, composed of more than just a string of unique details. The script is composed of central characters or actions (main conceptualizations), roles, setting and a sequence of events.

As knowledge structures, scripts provide a model of how people build abstract theories of how the world operates from personal or vicarious experiences by guiding lower order information processing (Abelson 1980; Martin 1980). Still, one must question whether individuals rely upon scripts as opposed to simple inferences based upon plausible explanations. For a script to warrant the status of a separate cognitive structure it must embody more than a simple inference rule. Scripts govern -a body of inferences which might serve to direct cognitive processing toward the correct inferences (Abelson 1980). Thus, in the restaurant script if one is informed that the customer waited with money in hand, the observer is able to scan the restaurant script and infer that the customer is waiting to pay his bill. This is not to imply, however, that such 'script processing' requires a search through a sequential list of script events. Clearly such a search process would not be efficient. Alternatively, individuals may search through scripts by attending to events of greater importance in the script (Nottenburg and Shoben 1980).

Script events differ in their degree of "centrality"-- some are indispensable to the script and summarize lower level actions (Abelson 1980, p. 10). Schank and Abelson (1977) referred to more central events as "main conceptualizations," around which other aspects of script processing was organized. Idiosyncratic differences would be expected for less central events arising from unique personal or vicarious experiences.

Perhaps the most widely cited empirical effort to examine the existence and use of scripts was conducted by Bower, Black and Turner (1979). In the first of a series of experiments, the authors demonstrated the existence, or "cultural uniformity," of basic actions in scripts for routine activities (e.g., eating in a restaurant, visiting a dentist). By cultural uniformity we are referring to the homogeneity across subjects in a given segment. However, while substantial agreement was found with respect to the central actions in these activities, a direct assessment of the homogeneity of the hierarchical ordering of these activities was not mate. Since this characteristic distinguishes scripts from a simple collection of event descriptions, such as assessment would seem essential.

In a subsequent experiment of that study, the recall of text narrative was investigated to determine whether subjects would falsely recall non-mentioned script events. That is, would subjects -fill in" gaps between intervening events? In addition, the authors investigated whether they -could increase the subject's belief that an unstated action had occurred in a script-based text by having him read related script stories in which the analogous or parallel actions were explicitly mentioned" (p. 189). To test this notion, subJects read a total of eighteen script stories developed by the authors with certain events deleted from the texts. Either 1, 2, or 3 versions of a script story were presented to a subject. Intrusions in recall and recognition of unstated actions were observed and were greatest when exposed to multiple versions.

Several procedures adopted by Bower et al. (1979) in their experiments examining intrusions in recall of unstated actions should be noted since they. provide the impetus for the approach taken in the current study. First, the script stories about routine activities presented to subjects and about which intrusions in recall were measured were constructed by the researchers as opposed to developing script norms from another subject pool. As such, it is difficult to conclude that intrusions are drawn from the subject's script rather than from constructive processes. In the present study, script norms are developed at the brand level from a separate subject pool. Second, Bower et al. (1979) presented subjects with 18 script stories and either 1, 2, or 3 versions. me results seem to indicate that confusion between different versions may have been responsible for some of the intrusions. In the present study, we changed the location of certain script events as contrasted to the deletion of events by Bower et al.

Rationale for Study

Since scripts describe the organization of experience in memory, it is perhaps most directly applicable to understanding how consumers' attitudes are formed and changed as a consequence of purchase experiences. As Calder and Schurr (1981) argue, attitudes are the product of a "constructive process in which incoming information is interpreted in terms of relevant stored information. (Scripts) that are triggered by incoming information determine what information (both incoming and stored) will be processed to yield attitudinal evaluation." It is apparent from this perspective that understanding script effects on consumer attitudes must involve explicating the memorial effects of triggered scripts. Only then can we understand their consequential inferential effects.

Since a script represents a stereotypic conception, it facilitates the ordering of incoming information, thereby reducing the cognitive strain resulting from encoding and retrieval processes. Thus, we would expect that information which is a good match to a script to facilitate recall. However, as noted by Taylor and Crocker (1981, p. 98), when a stimulus varies somewhat from the script, it is expected that the script information should dominate the subsequent recall. Such a result would seem most likely as short-term memory fades and greater reliance is placed upon the script.

As such, we would expect subjects to intrude more unstated script actions in recall. In the present study these memorial effects of scripts are examined for two purchase narratives.

Methodology and Results

The present study requires the identification of scripts that are relevant for the subjects in the present study. These baseline scripts were generated from a separate pool of subjects following the procedures used for script elicitation described by Bower et al. (1979). This study has been reported elsewhere in detail (John and Whitney 1982) and only the critical results are described here.

Two critical features of the elicited scripts were verified in the analysis. First, there was a substantial degree of agreement regarding the constituent actions of a McDonalds purchase episode and the purchase of a stereo from a large, local discount retailer. (Figure 1 shows these actions.) A second issue concerned the expected sequential ordering of these events, and a paired comparison analysis of the events indicated virtual unanimity in this regard.

FIGURE 1

MAIN EVENTS OF SCRIPTS

The aim of the present study is to demonstrate that subjects will intrude unstated script actions in recall. This is achieved by demonstrating certain patterns of recall across versions of script narratives developed by using the events identified in the previous study. Specifically, two types of manipulations were used to develop these narratives.

In certain of the normative versions, some main event was not explicitly mentioned in its usual place (i.e., the place where it was mentioned explicitly in the control condition or baseline script). Instead, this event was moved to another position in the sequence of events. If subjects do indeed intrude unstated script actions in their expected position, then there would be no difference between such versions and the baseline script with respect to subjects stating that such an event was mentioned in the narrative. It should be noted that this is a strong test of the effect of expectations triggered by scripts. In previous studies (e.g., Bower et al. 1979) intrusion errors were simply compared across various conditions. No equalization of memory was demonstrated as a consequence of these intrusion errors. In contrast, we expect that memory for the located event is identical regardless of whether is was mentioned or not in the narrative.

In addition, a different manipulation was used where we would expect differences between the groups. Here, unstated script intrusions were not expected. Specifically, actions were included in these versions that were not present in the baseline scripts. This produced between-group differences because no script-based expectations are triggered regarding these events.

Method

Undergraduate business students were randomly assigned to one of four conditions. Each subject read one narrative describing a purchase episode involving a stereo system. In three of the groups, they also read a narrative describing a purchase episode at McDonalds. A short distractor task followed the fast narrative in order to avoid short-term memory effects. The recall questions were administered shortly afterwards. In all, the procedure took about twenty minutes. Subjects were debriefed after the recall questionnaire was collected. me manipulations and instrument development are described below.

Independent and Dependent Variables

Four versions of the stereo purchase narrative were constructed. The baseline or control version was developed with as many high-frequency events as possible, and included some of the low-frequency items in order to make the narrative a realistic account of one individual's experience.

In the manipulated version used in the second group (ST2), the main event explicitly describing bargaining with the salesman was shifted from its position in the control script (i.e., at the stereo department while interacting with the salesperson). Instead, it was described as occurring at the cashier's counter just before the already-written order was paid for. The focal recall question (QST1) asked subjects to indicate on a 7-point scale if they were confident that bargaining had been described while discussing the terms of the sale. No differences were expected relative to the baseline because the subjects would intrude "bargaining" in its expected position by drawing on their script-based expectations.

In the version used in the third group (ST3), the main event that described the difficulty of locating a salesperson was moved from its position in the control script (i.e., after browsing around the stereo department). Instead, it was described that it was difficult to locate the salesperson once the customer was ready to pay for the item at the cashier's counter. The salesperson was needed to close the already written order. The focal recall question (QST2) asked subjects whether the action had been described as occurring in the stereo department after deciding on a system. Again, no differences were expected for the same reason as previously indicated.

In the final version (ST4), an event that was not a main event was shifted from its position in the control script. The act of driving around to locate a spot in the store's parking area was moved from the beginning of the narrative. Instead, it occurred as the subject was attempting to find a spot at the pick-up point for the purchased item. In this case, the focal recall question (QST3) asked subjects if the event had been described as occurring when the person first arrived at the store. We expected that this group would exhibit a significant difference relative to the control group since they could not be expected to intrude this event in recall if it was not explicitly described. This positive difference strengthens the integrity of the findings in two ways. First, it establishes the conditions under which script expectations do not affect memory, and further, it enables us to reduce the threat that the previous no difference findings are merely indicative of weak stimuli or inattentive subjects.

There were two filler questions included in the recall questionnaire as non-equivalent dependent measures. They asked if: (1) the subject had forgotten to lock his car (QST4), and (2) if two forms of ID were required to pay by check (QST5). Both these events had been described in all of the narratives, and no main effect of the treatments would be expected for these two items. These measures constituted another check on the validity of the expected findings. Since these events were described in all the narratives, any differences between groups would indicate artifactual group differences possibly due to attention effects or experimental demand effects.

Three versions of the Mcdonald's script were prepared. As previously, the baseline version (MC1) included as many main events as possible along with some low-frequency items. In the first manipulated version (MC2), the main event describing the waiting in line for a few minutes shifted from its position in the control script (i.e., prior to placing the order at the counter). Instead it was described as occurring when the subject returns to the counter to pick up some condiments. The focal recall question (QMC1) asked if the event was described prior to placing his order. Here, no group differences were expected to occur as subjects should intrude the unstated main event in its expected position.

In the third condition (MC3), the manipulation consisted of including two unexpected events in the script. Specifically, the script stated that the order was brought to the table by a waitress. Further, the low-frequency event describing the subject clearing away wrappers left by previous diners was deleted. Instead, it described a waitress clearing up his table after he had finished. There were two focal recall questions. (QMC2, QMC3) asking whether these events (i.e., the waitress bringing the order to the table and the table being cleaned) were described. Clearly, we can expect that group differences would exist here because the subjects in the control condition would not be expected to intrude these unexpected events.

As with the stereo narrative, two non-equivalent dependent measures were included. Subjects were asked (1) if the subject had forgotten to lock his car (QMC1) and (2) if the subject had been disturbed by a child at a nearby table (QMC5) . Neither of these events had been described in any of the three conditions, and no main effects of treatment were expected. As previously, these non-equivalent measures provide evidence of experimental artifacts. If these measures showed between-group differences, it would indicate that non-script intrusion errors were not equal across the groups. This would threaten our findings based on script-induced intrusion errors.

Results

The hypotheses that were advanced in the previous section are summarized in Figure 2. Note that these hypotheses are stated in terms of specific contrasts of the four cell 1-factor design shown in Figure 2. The results for the stereo purchase are reported first followed by the McDonald's results.

FIGURE 2

HYPOTHESIS AND RESULTS

The main effect for QST1 was nonsignificant (F=1.11; 3,86 df). Obviously, the relevant contrast (Control-ST2) is also insignificant. The results for QST2 indicate a significant main effect (F=23.9; 3,86 df). Contrary to our expectation, however, the relevant contrast (Control ST3) is significant (x1=1.08, x3=3.70, p<.05). The results for QST3 indicate a significant main effect (F=89.8; 3,86 df). More importantly, the relevant contrast (Control ST4) is significant (x1=1.32, x4=5.86, p<.05). Finally, the results of the non-equivalent dependent measures (QST4 and QST5) indicate no significant main effect (F=1.95 and F=1.27 with 3,86 df) for either variable.

The first dependent measure for the McDonalds script (QMC1) shows no significant main effect (F=3.67; 2,65 df). Obviously, the relevant contrast (Control MC2) is also nonsignificant. The QMC2 measure shows a significant main effect. The F statistic is extremely large because of virtually no within-cell variance (F=9182; 2,65 df) and the relevant contrast (Control MC3) is also significant (x1=6.9, x =1.0, p<.05). The QMC3 measure also shows a significant main effect (F=41.3, 2,65 df) and the relevant contrast (Control MC3) is significant (x1=2.25, x3=5.83, p<.05). Finally, the results of the two non-equivalent dependent measures (QMC4 and QMC5) indicate no significant effects for either variable (F=1.6 and F=0.48 with 2,65 df).

The evidence relative to the various hypotheses is summarized in Figure 2. It shows that our expectations were met in every case except one. Contrary to our expectation, we found that when the difficulty of locating a salesman was moved from its position in the control script, significant differences in recall were obtained. The implications of these results are discussed in the following section.

DISCUSSION

All of our hypothesized effects were supported except for the finding relative to the displacement of the bargaining event. A basic script-theoretic mechanism is supported by the-present data. Consumers display a tendency to intrude unstated script actions when recalling script-based narratives of purchase episodes. Events that are expected to be in a particular position as evidenced by the baseline scripts will be remembered as occurring in that position regardless of its actual position in the narrative. However, those events that are not expected in particular positions (as evidenced by their absence in the baseline scripts) are not subject to such intrusion errors.

The one unexpected effect that was found also provides further insight into the memorial effects of scripts. Intrusion errors did not equalize recall for the displaced main event that described "bargaining" in the stereo purchase episode. The focal retail outlet in the narrative is the largest local stereo store in the area. It offers no list prices and has a very heavy emphasis on deep discounting. Consequently, the bargaining aspect of the interaction is a very salient event as seen by the frequency of mention of this event in the baseline script (Figure 1). It appears, then, that very salient events are similar to the non-salient events in that they are not subject to extensive intrusion errors.

Together, these results suggest that the tendency to intrude unstated script actions in recall is bounded from below by non-main events and from above by very salient events. In other words, the absence of low-frequency events to not lead to intrusion subject errors to equalize memory; displacement of very salient events, however, also induces a similar result, and is probably due to an interruption of the instantiated script.

The intrusion errors produced in this study indicate that more attention need to be paid to developing an understanding of the intrusion errors in memory for commercials. Presently, measures of memory focus on positive responses by questions concerning the focal commercial and generally ignore false-positives. If scripts exist for advertisements in a product class, it is plausible to expect a considerable amount of intrusion errors to exist in memory for commercials of different brands in the product category. Such confusion has been documented elsewhere (McMahan 1971; Keon and Gleason 1982) although the conceptual reasons have been unclear. Script theory affords us an explanation of these intrusion errors, and thus an effective means of minimizing these errors.

Another important avenue for further research concerns the impact on consumer evaluations of script-based expectations. As indicated previously, attitudes are best viewed as the consequence of the interaction between incoming information and stored information and their evaluation. Within this framework, we have seen how scripts can affect what is remembered regarding incoming information. Thus, the same information can possibly result in different attitudinal evaluations depending on the triggered script-based expectations. In particular, we need to understand how these expectations affect the availability in memory of heavily valenced events as they are likely to have. Hopefully, these issues will be addressed in future studies.

REFERENCES

Abelson, Robert P. (1975), "Concepts for Representing Mundane Reality in Plans," in Representation and Under- standing: Studies in Cognitive Sciences, D. G. Bobrow Press.

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

Abelson, Robert P. (1980), "The Psychological Status of the Script Concept," Cognitive Science Technical Report 82, Yale University. Cognitive Science Program.

Bartlett, F. C. (1932), Remembering, Cambridge: Cambridge University Press.

Bower, Gordon H., Black, John B. and Turner, Terrence J. (1979), "Scripts in Memory for Text," Cognitive Psychology, 11, 177-220.

Bransford, J. D. and Barclay, J. D. and Franks, J. J. (1972), "Sentence Memory: A Constructive vs. Interpretive Approach," Cognitive Psychology, 3, 193-209.

Bransford, J.A., and Johnson, M. K. (1972), "Contextual Prerequisites for Understanding: Some Investigations of Comprehension and Recall," Journal of Verbal Learning and Verbal Behavior, 11, 717-726.

Calder, B. J. and Schurr, P. H. (1981), "Attitudinal Processes in Organizations," Research in Organizational Behavior, 3, 283-302.

Cantor, Nancy and Mischel, Walter (1977), "Traits as Prototypes: Effects on Recognition Memory," Journal of Personality and Social Psychology, 35, 38-48.

Cohen, C. E. (1977), "Cognitive Basis of Stereotyping," paper presented at the American Psychological Association Annual Meeting, San Francisco.

Dooling, D. J. and Mullet, R. L. (1973), "Locus of Thematic Effects in Retention of Prose," Journal of Experimental Psychology, 97, 404-406.

Fiske, S. and Kinder, D. (1978), "Schemas and Political Information Processing," paper presented at American Psychological Association, Toronto, Canada.

John, G. and Whitney, J. C. (1982), "An Empirical Investigation of the Serial Structures of Scripts," in Assessment of Marketing Thought and Practice, Walker, B. J. (ed.), Chicago, IL: American Marketing Association, 75-79.

Kassarjian, E. H. (1978), "Presidential Address, 1977: Anthropomorphism and Parsimony, in Advances in Consumer Research, Vol. 5 (ed.), H. Keith Hunt, Ann Arbor, MI: Association for Consumer Research, xii-xiv.

Keon, J. and Gleason, T. (1982), "Scaling Brands on Their Advertising Distinctiveness," in An Assessment of Marketing Thought & Practice, Walker B. J. (ed.), Chicago, IL: American Marketing Association, 443-446.

Langer, E. J., Blank, A. and Chanowitz, 8. (1978), "The Mindlessness of Ostensibly Thoughtful Action," Journal of Personality and Social Psychology, 36, 635-642.

Langer, E. J. and Newman, H. M. (1979), "The Role of Mindlessness in a Typical Social Psychology Experiment," Personality and Social Psychology Bulletin, 5, 295298.

Loftus, E. F. and Palmer, J. C. (1974), "Reconstruction of Automobile Destruction: An Example of the Interaction Between Language and Memory," Journal of Verbal Learning and Verbal Behavior, 13, 585-589.

Mandler, J. M. (1978), "A Code in the Node," Discourse Processes, 1, 14-35.

Markus, Hazel (1977), "Self-Schemata and Processing Information About the Self, Journal of Personality and Social Psychology, 35, 63-78.

Martin, J. (1980), "Stories and Scripts in Organizational Settings," Research Report No. 543, Stanford University.

McMahan, H. W. (1971), "Brand Misidentification Grows with Clutter of :30 Spots," Advertising Age, May 10, 53-54.

Minsky, M. (1975), "A Framework for Representing Knowledge," in The Psychology of Computer Vision, P. H. Winston (ed.). New York: McGraw-Hill.

Neisser, V. (1976), Cognition and Reality: Principles and Implications of Cognitive Psychology, San Francisco: Freeman.

Nisbett, Richard and Ross, Lee (1980), Human Inference: Strategies and Shortcomings of Social Judgment, Englewood Cliffs, NJ: Prentice-Hall, Inc.

Nottenburg, G. and Shoben, E. J. (1980), "Scripts as Linear Orders," Journal of Experimental Social Psychology, 16, 329-347.

Olshavsky, R. W. and Granbois, D. H. (1979), "Consumer Decision Making--Fact or Fiction?", Journal of Consumer Research, 6, 93-100.

Rethans, A. J. and Taylor, Jr., J. L. (1982), "A Script Theoretic Analysis of Consumer Decision Making," in An Assessment of Marketing Thought & Practice, Walker, B. J. (ed.), Chicago, IL: American Marketing Association, 71-74.

Rumelhart, D. E. and Ortony, A. (1976), "The Representation of Knowledge in Memory," in Schooling and the Acquisition of Knowledge, R. C. Anderson, R. J. Spiro, and W. E. Montague (eds.), Hillsdale. NJ: Erlbaum.

Schank, Roger C. and Abelson, Robert P. (1977), Scripts, Plans, Goals, and Understanding: An Inquiry into Human Knowledge Structures, Hillsdale, NJ: Lawrence Erlbaum.

Snyder, M. and Uranowitz, S. W. (1978), "Reconstructing the Past: Some Cognitive Consequences of Person Perception," Journal of Personality and Social Psychology, 36, 941-950.

Taylor, S. E. and Crocker, Jennifer (1981), "Schematic Bases of Social Information Processing," in The Ontario Symposium on Personality and Social Psychology, E. T. Higgins, P. Hermann, and M. P. Zanna, (eds.), Vol. 1, Hillsdale, NJ: Lawrence Erlbaum.

----------------------------------------