Memory For Logically-Deduced Conclusions

ABSTRACT - An experiment was conducted to determine whether (1) memory for conclusions improved when logically but not illogically-related premises were also shown, and (2) false memory for conclusions occurred more often when logically than when illogically-related premises were shown. Results suggested the second but not the first effect. Subjects exposed to logically-related statements were more likely to falsely identify the conclusions of these statements as having been presented (when they were not) than were persons who were exposed to illogically-related statements. Moreover, recognition was found to be nonsignificantly related to belief strength.

Citation:

Barbara Loken (1982) ,"Memory For Logically-Deduced Conclusions", in NA - Advances in Consumer Research Volume 09, eds. Andrew Mitchell, Ann Abor, MI : Association for Consumer Research, Pages: 348-353.

Advances in Consumer Research Volume 9, 1982      Pages 348-353

MEMORY FOR LOGICALLY-DEDUCED CONCLUSIONS

Barbara Loken, University of Minnesota

ABSTRACT -

An experiment was conducted to determine whether (1) memory for conclusions improved when logically but not illogically-related premises were also shown, and (2) false memory for conclusions occurred more often when logically than when illogically-related premises were shown. Results suggested the second but not the first effect. Subjects exposed to logically-related statements were more likely to falsely identify the conclusions of these statements as having been presented (when they were not) than were persons who were exposed to illogically-related statements. Moreover, recognition was found to be nonsignificantly related to belief strength.

INTRODUCTION

Research in psychology shows that people will learn and better remember stimulus information when they are able to organize the information than when they are not. Early studies in verbal learning found that rote learning was less effective than "meaningful" learning (Katona 1940) and that organizing the items in a stimulus list improved recall of these items (Garner and Whitman 1965, Tulving and Pearlstone 1960). More recently, the effects of organization involving more complex and "realistic" situations have been examined by resurrecting the notion of schemata (Bartlett 1932). A schema is generally described as a framework for organizing information around a particular concept (Olson 1978), such as a brand name. Studies have shown that organization of information with respect to general themes (Lingle, Geva, Ostrom and Leippe 1979, Bower 1970) or rules (Bear and Hodun 1975, Potts 1972) later improved memory for this information.

In the past few years, consumer researchers have addressed similar information-processing issues as those explored by cognitive psychologists, including some speculation about the manner in which information is encoded, organized in memory, and later retrieved from memory. For example, it is generally agreed that information may be encoded, i.e., represented in memory, in several different forms and levels of abstraction. If an advertisement claims that an automobile has front-wheel drive, one consumer may encode this information isomorphically, simply as "Brand X has front-wheel drive", a second consumer may encode this as "Brand X is safer than other brands in winter", and a third consumer may encode this as "Brand X has favorable attributes". Thus, the same information may be encoded at different levels of abstraction (Russo and Johnson 1980 Olson 1978).

Furthermore, it is generally assumed that the manner in which external information is encoded and stored in memory will depend upon already-existing cognitions or "knowledge structures" (Olson 1980 Mitchell 1980). Previous knowledge and newly-acquired information are presumably integrated using one or more of several different processes or rules, e.g. perhaps an algebraic process or decision rule when an valuative decision is being made, or logical reasoning and probabilistic consistency when the validity of a proposition is being judged. The use of such rules for predicting people's decisions or judgments has been well-documented in both social psychology and consumer research.

With respect to one of these integration processes, logical reasoning, past research (McGuire 1960, Wyer and Goldberg 1960) has generally supported the notion that people's. judgments of the validity of premises are consistent with t-heir judgments of the validity of logically-deduced conclusions. Thus, logical deduction may frequently be used by people as a rule of inference. Nevertheless, logical deduction has seldom been examined as one of memory's organizing principles. Although much evidence exists for the use of thematic principles on organization, basic logical processes have largely been ignored. In fact, investigators have sometimes stressed that thematic principles of thinking are more pragmatic than they are logical (Bruner, Goodnow, and Austin 1956).

The present study investigates the use of logical reasoning as an organizing rule for memory. If, for example, a logical structure has been imposed on a set of stimulus information, propositions that are logically-related may be stored together in memory. Support for the effects of imposed structures on consumer memory has been shown in other forms of consumer research. Johnson and Russo (1978) found that memory for product information depended upon whether the purchase environment was attribute-based or brand-based. Furthermore, as Newell and Simon (1972) argue, the structure of the task that a subject performs may affect the processing strategies that will be used. In support of this agreement, Bettman and Kakkar (1977) found that consumers acquired information in the fashion that was easiest given a brand-attribute information display. Attribute processing was found when the display facilitated attribute and discouraged brand processing. Brand processing was found when the display facilitated brand processing and discouraged attribute processing.

Another research finding is relevant in discussing the use of organizing rules in memory. Memory research (cf. Wyer and Srull 1980) suggests not only that consumers may store information in memory that is quite different from the stimulus information received, but also that additional thoughts generated by the stimulus information may be stored in memory along with the stimulus information. Thus, for example, if two logically-related premises are received as information by a consumer, a conclusion deduced from these premises may be stored in memory instead of or in addition to the two premises. Further, a single premise received combined with previously-acquired knowledge may lead to a conclusion. For example, suppose a product advertisement states that a brand of mouthwash kills germs. A consumer who already believes that "killing germs prevents illness" may come to the conclusion that the mouthwash in the ad prevents illness. At a later point in time, when asked to recall the original stimulus information, a subject may be unable to distinguish information from derived conclusions (i.e. that the mouthwash prevents illness) generated at the time information was received.

The inability of a subject to distinguish presented and not presented information has been use. in cognitive psychology as a tool for understanding memory processes. Generally speaking, memory research has found that people will say a stimulus item was presented even when it was not presented if the item was consistent with a central organizational theme. For example, people often fill gaps in their visual representations. Missing information from a story, settingS or description of a person is filled in by the experimental subject when the information is consistent with or taken for granted in the overall description, and later on subjects are more likely to say that this information was presented than they are to say information inconsistent with the description was presented (Jenkins, Wald, and Pittenger 1978, Bear and Hodun 1975. Sulin and Dooling 1974, Bransford and Franks 1971). At least two possibilities exist for these results to occur, generally inseparable in memory research. One possibility is that the original information is stored along with thoughts generated by the subject at the time the information is received. A second possibility is that stimulus information is stored in memory as a simplified abstract representation, the basis for recreating and elaborating the original information at a later point in time. However, regardless of whether each of these possibilities occurs, this research suggests further possibilities for the influence of visual and conceptual organizing principles. Moreover, these findings occur not only as a result of recently learned principles (Bear and Hod 19 1975) but also as a result previous experience with similar situations (Bransford and Franks 1971).

The implications for the present research, which explores the use of logical deduction as an organizing principle, are as follows: suppose a person is exposed to two statements, of the forms "A leads to B" (or "If A then B") and "3 leads to C" (or "If B then C"), and later on is asked whether "A leads to C" (or "If A then C") was presented. If this person says (incorrectly) that s/he was exposed to "A leads to C", s/he may do so for at least two reasons. First, structural similarities between "A leads to C" on the one hand and "A leads to B" and "B leads to C" on the other hand exist in that both premises and conclusion contain the elements "A" and "C", and the elements "A", "B" and "C" may be stored separately in memory. People may confuse the context in which these elements were presented and report that "A leads to C" was originally presented. Second, false identification due to organizational factors may exist if people logically connect "A leads to B" and "B leads to C" to form the conclusion "A leads to C", store all three statements together in memory, and later on say that "A leads to C" was originally presented. It is expected that, although effects due to structural similarities may exist, organizational factors using syllogistic reasoning should occur over and above any effects due to similarity in structure. Furthermore, it is proposed that the same organizational properties should improve one's memory for the statements that were presented ("A leads to B", "B leads to C", and when it is presented, "A leads to C").

An additional issue is explored in the present investigation, concerning the relationship between a person's belief in a statement (e.g. the subjective probability that A leads to B) and her/his subsequent recognition of the statement. Studies have sometimes found that people's opinions facilitate recall of information consistent with these opinions (Levine and Murphy 1943) or that personal biases govern how one thinks (Janis and Frick 1943), although the reasons for such findings are unclear (Jones and Aneshansel 1956, Waly and Cook 1966). It may be that information is more readily organized around one's own opinions, perhaps because the opinions are highly familiar (Waly and Cook 1966). Recently, some consumer researchers have also argued that familiarity and greater prior knowledge may sometimes facilitate learning (Johnson and Russo 1978). Other studies refute the notion that information consistent with one's opinions is better recalled and claim that differences in recall or reasoning as a function of one's own position do not exist (Henle 1962, Brigham and Cook 1966). Finally, some researchers have argued that memory for propositions may be better when the information is inconsistent with one's own opinion or implausible to the subject (Wyer and Henninger 1978) since judging the validity of such statements may result in a greater amount of processing effort (Craik and Lockhart 1972). The present investigation will also explore this issue, as it relates to memory for conclusions that were precede< either by logically and illogically-related propositions.

An experiment was designed to test: (1) whether information is better remembered when it is logically related (than when it is not), (2) whether information not presented is more likely to be incorrectly or falsely identified or "remembered" when it logically follows from statements preceding it (than when it does not follow logically from statements preceding it), and (3) whether memory for presented information changes as a function of original beliefs about the information.

METHOD

Sample and Design

Subjects were 60 females, 15 in each cell of a 2 x 2 factorial design. Design factors were logical-relatedness (exposure to logically-related or illogically-related statements) and exposure to conclusions (exposure or nonexposure). Subjects participated in the experiment as partial fulfillment for a requirement in an introductory psychology course.

Stimuli and Procedure

Belief questionnaires were designed as a means for exposing subjects to information about "relevant social issues". The individual belief statements were formed such that the structural similarities between premises and conclusions were held constant for logically-related and illogically-related statements. Conclusions were of the form "A leads to C". The first premises or propositions were either of the form "A leads to B" (for logically-related statements) or of the form "A leads to D" (for illogically-related statements), i.e. both containing the element "A". The second premises were either of the form "3 leads to C" (for logically-related statements) or of the form "E leads to C" (for illogically-related statements), i.e. both containing the element "C". Thus, both logical and illogical conditions received the same number of "A" and "C" elements, although the conclusion "A leads to C" followed from the earlier statements presented in the former but not the latter conditions.

TABLE 1

EXAMPLES OF STIMULUS ITEMS, EACH CONDITION

Subjects were exposed to only one of the four combinations of belief statements (samples are shown in Table 1), but to 15 different sets of the same type, each set representing a different domain. The task required was to indicate the strength of agreement with the presented statements on a scale from O to 10 ("not at all likely" to "extremely likely"). Thus, subjects in the "logical-conclusion" condition completed 45 items, 15 of the form "A leads to B" (premise 1), 15 of the form "B leads to C" (premise 2),-and 15 of the form "A leads to C" (conclusions), in that order. Subjects in the "logical-no conclusion" condition completed 30 items, the same 15 first and second-premises. The "illogical-conclusion" condition consisted of 15 items of the form "A leads to D", 15 of the form "E leads to C", and 15 conclusions ("A leads to C"). Finally, the "illogical-no conclusion" condition included 30 items, 15 of the form "A leads to D" and 15 of the form "E leads to C". The content of these statements pertained to social and other issues with which subjects may already have been familiar. To this extent, it should be noted that results obtained using these items may not be generalizable to the use of unfamiliar or more abstract belief items. Finally, to make all belief questionnaires comparable in length, other completely unrelated belief items were added.

Following the belief questionnaire, subjects completed an unrelated intervening task to decrease rehearsal and short term memory effects of belief statements. This task lasted about 20 minutes, after which subjects completed (1) cued-recall and (2) recognition tasks. [The recognition task was always administered following the recall task. It is unknown whether the measurement of recall could have affected the subsequent recognition of items, but it seems unlikely that the experimental hypotheses would be influenced by such effects.] Subjects were not informed in advance that they would be asked to remember the content of the belief statements.

Measures of Memory

In the cued-recall task, subjects were presented with the first part of each of the belief statements (e.g. "Taking large doses of Vitamin C...") and were asked to recall the ending of the statements. Subjects were informed that for some of the statements two different endings had appeared in the earlier questionnaire, for some of the statements one ending had appeared, and for other of the statements no ending had appeared. They were instructed to write down endings to all of those statements they could remember. All conditions were presented with the same list of beginning phrases.

The recognition task was designed to test recognition of the 15 possible conclusions. Subjects were instructed to indicate whether each item was or was not ("yes" or "no") presented in the earlier belief questionnaire and a confidence rating of their judgment on a O to 10 scale ("not at all confident" to "extremely confident"). To measure recognition, the confidence rating for each conclusion was multiplied by the yes-no (+1 or -1) response. The possible range for recognition scores was therefore -10 to +10.

RESULTS

Recall and Recognition of Conclusions

Cued recall (i.e. recall of the ending of each statement) was scored dichotomously (+1 or 0) for "identification" or "no identification" of the content of each premise and conclusion. Therefore, "identification" of a statement "A leads to C" reflects correct recall in logical-conclusion and illogical-conclusion conditions but reflects incorrect recall, i.e. a false positive response (saying a statement was presented when it was not); in logical-no conclusion and illogical-no conclusion conditions. Analyses of cued recall were performed on a general measure of recall computed by summing the number of positive identifications over the 15 replications. Similarly, a general measure of recognition was computed by summing recognition scores for the 15 conclusions. Two 2 x 2 (logical - illogical x conclusion - no conclusion) analyses of variance were performed with cued recall and recognition of conclusions as dependent measures.

Results of the cued-recall analysis indicated a significant main effect for conclusion presence, F(1,56) = 145.7, p < .001, and a significant interaction effect, F(1,56) = 9.55, p < .01. Conclusions were more likely to be mentioned or identified if presented (M - 8.07) than if not presented (M - 1.1-), and, as shown in Figure 1, the number of false positive responses in "no conclusion" conditions was greater when logical (M - 2.27) than when illogical (M = .07) statements were presented. A simple effects test of this latter effect was significant, F(1,56) = 114.79, p < .01, as predicted. However, in "conclusion" conditions, correct recall of conclusions was not better when preceded by logically-related (M - 7.40) than when preceded by illogically-related (M = 8.73) statements.

FIGURE 1

CUED-RECALL OF STIMULUS INFORMATION

Results of the recognition analysis revealed significant main effects for both logical-relatedness, F(1,56) 45.56, p < .01, and conclusion presence, F(1,56) - 188.95, p < .01. The interaction was also significant, F(1,56) = 40.62, p < .01. Subjects were more likely to say conclusions were presented in logical (M - 68.30) than in illogical (M - -5.73) conditions, and trivially, in conclusion (M - 106.67) than in no conclusion (M - -44.10) conditions. Simple effects tests of the interaction term indicated that the differences between logical and illogical conditions were in the two no-conclusion conditions, F(1,56) - 86.11, p < .01, and not in the conclusion conditions, F (1,56) = .07, n.s. Results are diagrammed in Figure 2. Subjects exposed to conclusions accurately recognized the conclusions (M = 108.73 and 104.60 for "logical-conclusion" and "illogical-no conclusion" conditions, respectively) and subjects in the "illogical-no conclusion" condition accurately recognized that the conclusions were not presented (M = -116.07). In contrast, and consistent with findings from cued recall data, subjects in the "logical-no conclusion" condition made a substantial number of errors in recognition (M = 27.87). They tended to report that the conclusions had been presented when, in fact, they had not been presented.

FIGURE 2

RECOGNITION OF STIMULUS INFORMATION

Since this latter result reflects a combination of a dichotomous yes-no response and a confidence rating, it is interesting to determine whether each of these factors were affected separately. Internal analyses suggest that subjects in the logical-no conclusion condition reported a higher frequency of "yes" responses (indicating false positive identifications of conclusions) than did subjects in illogical-no conclusion conditions but a lower frequency than subjects in logical-conclusion conditions. That is, 59% of the total number of responses (15 persons x 15 y replicates - 225 total) in the logical-no conclusion condition, 1% in the illogical-no conclusion conditions and 88% V in the logical-conclusion condition were "yes" responses. (In illogical-conclusion conditions, 86: were "yes" responses.) On the other hand, subjects in the former condition reported less confidence in their recognition judgments than subjects in the latter conditions. In particular, 80% of the responses in logical-no conclusion conditions were at least moderate in confidence (responses greater than 5 on a O to 10 scale), whereas this figure was 96% in both illogical-no conclusion and logical-conclusion conditions. (This figure was 86% in illogical-conclusion conditions.) Therefore, not surprisingly, subjects in logical-no conclusion conditions misrecalled conclusions quite frequently (59% of the times), and tended to be at least moderately confident of their judgments (80% of the times), although somewhat less confident than subjects in logical-conclusion and illogical-no conclusion conditions.

Belief Strength and Memory

To determine whether memory was a function of belief strength, correlations were computed between subjects' original beliefs (scaled from O to 10) and their subsequent recognition of the items (scaled from -10 to +10), for each of the 15 individual conclusions, and separately for "logical" and "illogical" conditions. Naturally, only "conclusion" conditions were used to compute these correlations. Correlations ranged from very positive (.7:) :¦ very negative (-.68) for the individual items. However the average positive and the average negative correlations across the 15 items (converting r to z-scores, averaging the z-scores, and converting back to r) were nonsignificant (p > .05) in both logical conditions (average r 3 .108 And -.245 for positive and negative correlations, respectively) and illogical conditions (average r - .384 and -.341, respectively).

Thus, although individual differences in the items occurred, overall there was little evidence that subjects tended to better recognize items with which they agreed (cf. Levine and Murphy 1943) or items inconsistent with their opinions (cf. Craik and Lockhart 1972). Furthermore, these results do not appear to be due to general agreement or disagreement with the belief items. The mean belief strength of individual items was slightly positive (overall average was 5.60) with a range from 2.90 to 8.04.

DISCUSSION

Findings and Implications

The findings are consistent with the notion that logical reasoning is used as an organizing principle in memory. This is noted by the tendency for subjects to say that a logically-deduced conclusion was presented when in fact it was not. Both cued recall and recognition data supported this hypothesis. Furthermore, this effect could not be attributed to the structural similarities (i.e. each containing the same elements "A" and "C") between logically-related premises and conclusions. Thus, some people remember seeing conclusions they in fact have not seen, perhaps because they match the conclusions with their integrated understanding of premise statements. The present study used a belief questionnaire to expose subjects to stimuli, a technique which may have increased subjects' tendencies to process semantic aspects of the information in an integrated form.

On the other hand, exposure to logically-related premises in addition to conclusions did not improve memory for conclusions over that of exposure to illogically-related premises and conclusions. If anything, results suggest slight interference effects of premises when they are logically-related, but the nature of these effects is unclear. Furthermore, it should be noted that ceiling effects may exist with regard to the recognition data. When a conclusion was presented, most subjects were later able to accurately report this with strong confidence. Ceiling effects were less evident in the cued-recall data.

The present results may have several important implications, perhaps one of the more important being in the area of advertising research. Suppose that a product advertisement does not actually state a claim (or conclusion), but that this claim is implied by statements that are shown in the advertisement. At a later point in time, some proportion of the audience may have derived this conclusion and be unable to distinguish it from the original information presented in the ad. Furthermore, suppose a statement "Buying Brand A leads to consequence X" appears in an at, and the target population exposed to the ad already tends to believe "Consequence X leads to Y". Among some members of this audience the conclusion "Buying Brand A leads to Y" may not only be inferred, but later mistaken as having been presented. Implications for deception in advertising exist as well. The occurrence of a significantly large number of false positive responses, i.e., inaccuracies in memory such that people believe a statement was presented, raises the question of whether an advertisement is deceptive if conclusions derived logically from information contained in the ad are deceptive.

Finally, the hypothesis that people's opinions affect their memory for information was not supported. Although individual differences in stimulus items occurred, there was no general tendency for subjects to better recognize items that were consistent or items that were inconsistent with their own opinions.

Study Limitations

Two limitations of the present investigation need to be reiterated at this point. First, it is important to note that the present results do not tell us whether the conclusions of logically-related premises were derived at the time the premises were presented to the subject, or whether they were derived later on at the time of the recall/recognition task. Memory processes are relevant to both of these possibilities to the extent that the subject's error is an error in recall or recognition. In both cases, logical processes are used as organizing principles for remembering information. However, if the conclusion is inferred at the time of the memory task, it is incorrect to assume that the conclusion already exists in memory along with the logically-related premises prior to the memory task

The second limitation concerns the generalizability of results to other sets of stimuli. The present stimuli were familiar issues in the form of belief statements. If subjects had been asked, for example, simply to read the statements rather than to judge the validity of each one, perhaps they would not have thought as extensively about each of the items. Furthermore, if the statements had dealt with unfamiliar issues, with symbolic rather than descriptive statements, with nonsense statements or words rather than meaningful statements, perhaps a different set of results would have occurred.

Future Research Suggestions

Several questions are raised by this research, not the least of which is the actual stage of information processes the logical inferences are made. Research that is able to experimentally manipulate and explore the processes of memory (encoding, retrieval, etc.) is needed in all areas of memory research. Furthermore, since only a certain proportion of the subjects made false positive responses to any given item, it would be interesting to see whether the incidence of false positives is an individual difference characteristic, and if so, whether it is related to individual abilities and tendencies to reason logically.

Another research topic to explore concerns the conditions under which false positives occur. For example, the occurrence of false positives may depend upon the order in which the premises and conclusions are shown and the nature of the information shown.

Finally, the extent to which memory errors may impact on subsequent judgments, evaluations, or purchase decisions is important. For example, consumers may use derived "conclusions" or advertising claims differently depending on whether they perceive that a source gave them the information or whether they perceive that they reached a particular conclusion on their own.

REFERENCES

Bartletts F. C. (1932), Remembering: A Study in Experimental and Social Psychology, New York: MacMillan.

Bear, G. and Hodun, A. (1975), "Implicational Principles and the Cognition of Confirmatory, Contradictory, Incomplete, and Irrelevant Information," Journal of Personality and Social Psychology, 32, 594-604.

Bettman, J. R. and Kakkar, P. (1977), "Effects of Information Presentation Format on Consumer Information Acquisition Strategies," Journal of Consumer Research, 3, (March), 233-40.

Bower, G.H. (1970), "Organizational Factors in Memory," Cognitive Psychology, 1, 18-46.

Bransford, J. D. and Franks, J. J. (1971), "The Abstraction of Linguistic Ideas," Cognitive Psychology, 2, 331-50.

Brigham, J. C. and Cook, S. W. (1969), "The Influence of Attitude on the Recall of Controversial Material: A Failure to Confirm," Journal of Personality and Social Psychology, 5, 240-43.

Bruner, J. S., Goodnow, J. J. and Austin, G. A. (1956), A Study of Thinking, New York: Wiley.

Craik, F. I. M. and Lockhart, R. S. (1972), "Levels of Processing: A Framework for Memory Research," Journal of Verbal Learning and Verbal Behavior, 11, 671-84.

Garner, W. R. and Whitman, J. R. (1965), "Form and Amount of Internal Structure as Factors in Free-recall Learning of Nonsense Words," Journal of Verbal Learning and Verbal Behavior, 4, 257-66.

Henle, M. (1962), "On the Relation Between Logic and Thinking," Psychological Review, 69, 366-78.

Janis, I. L. and Frick, F. (1943), "The Relationship Between Attitudes Toward Conclusions and Errors in Judging Logical Validity of Syllogisms," Journal of Experimental Psychology, 33, 73-77.

Jenkins, J. J., Wald, J. and Pittenger, J. B. (1978), "Apprehending Pictorial Events: An Instance of Psychological Cohesion," In C. W. Sarage (Ed.), Minnesota Studies in the Philosophy of Science, Vol. 9, University of Minnesota Press.

Johnson, J. and Russo, J. (1978), "The Organization of Product Information in Memory Identified by Recall Times," in Advances in Consumer Research, Vol. 5, edited by H. Keith Hunt. Ann Arbor: Association for Consumer Research.

Jones, E. E. and Aneshansel, J. (1956), "The Learning and Utilization of Contravaluent Material," Journal of Abnormal and Social Psychology, 53, 27-33.

Katona, G. (1940), Organizing and Memorizing, New York: Columbia University Press.

Levine, J. M. and Murphy, G. (1943), "The Learning and Forgetting of Controversial Material," Journal of Abnormal and Social Psychology, 38, 507-517.

Lingle, J. H., Geva, N., Ostrom, T. M., Leippe, M. R., and Baumgardner, M. H. (1979), "Thematic Effects of Person Judgments on Impression Organization," Journal of Personality and Social Psychology, 37, 674-687.

McGuire, W. J. (1960), "A Syllogistic Analysis of Cognitive Relationships," In C. I. Hovland and M. J. Rosenberg (eds.), Attitude Organization and Change, New Haven: Yale University Press, 65-111.

Mitchell, A. A. (1980). "The Use of an Information Processing Approach to Understand Advertising Effects," ed., J. C. Olson, Advances in Consumer Research, Vol. VII, Ann Arbor: Association for Consumer Research, 171-77.

Newell, A. and Simon, A. (1972), Human Problem Solving, Englewood Cliffs, New Jersey: Prentice-Hall, Inc.

Olson, J. C. (1978), "Theories of Information Encoding and Storage: Implications for Consumer Behavior," in The Effects of Information on Consumer and Market Behavior, ed., A. A. Mitchell, Chicago: American Marketing Association, 49-60.

Olson, J. C. (1980), "Encoding Process: Levels of Processing and Existing Knowledge Structures," in Jerry C. Olson, ed., Advances in Consumer Research, Volume VII, Ann Arbor: Association for Consumer Research, 154-160.

Potts, G. R. (1972), "Information Processing Strategies, Used in the Encoding of Linear Orderings," Journal of Verbal Learning and Verbal Behavior, 11, 727-740.

Russo, J. E. and Johnson, E. J. (1980), "What Do Consumers Know About Familiar Products?", ed., J. C. Olson, Advances in Consumer Research, Vol. VII, Ann Arbor: Association for Consumer Research. 417-423.

Sulin, R. A. and Dooling, D. J. (1974), "Intrusion of a Thematic Idea in Retention of Prose," Journal of Experimental Psychology, 103, 255-262.

Tulving, E. and Pearlstone, Z. (1966), "Availability versus Accessibility of Information in Memory for Words," Journal of Verbal Learning and Verbal Behavior," 5, 381-391.

Waly, P. and Cook S. W. (1966), "Attitude as a Determinant of Learning and Memory: A Failure to Confirm," Journal of Personality and Social Psychology, 4, 280-288.

Wyer, R. S. and Goldberg, L. (1970), "A Probabilistic Analysis of Relationships Among Beliefs and Attitudes," Psychological Review, 77, 100-120.

Wyer, R. S. and Henninger, M. (1978), The Effects of Reporting Beliefs on the Recall of Belief-Related Propositions. Unpublished manuscript, University of Illinois at Urbana-Champaign.

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