The Role of Previous Knowledge in the Recognition of Product Information

ABSTRACT - Although dramatic differences have been reported for recall performance between high and low knowledge consumers, the differences are attenuated, if not non-existent in recognition performance. These results question the role of retrieval in recognition. In this paper theories of recognition which are independent or dependent on retrieval are compared. A study is described in which the ability and opportunity to retrieve information varies. The results of the study offer some initial evidence that the level of previous knowledge moderates the relationship between retrieval and recognition.



Citation:

Elizabeth Cowley (1996) ,"The Role of Previous Knowledge in the Recognition of Product Information", in AP - Asia Pacific Advances in Consumer Research Volume 2, eds. Russel Belk and Ronald Groves, Provo, UT : Association for Consumer Research, Pages: 68-72.

Asia Pacific Advances in Consumer Research Volume 2, 1996      Pages 68-72

THE ROLE OF PREVIOUS KNOWLEDGE IN THE RECOGNITION OF PRODUCT INFORMATION

Elizabeth Cowley, University of Western Sydney, Nepean

ABSTRACT -

Although dramatic differences have been reported for recall performance between high and low knowledge consumers, the differences are attenuated, if not non-existent in recognition performance. These results question the role of retrieval in recognition. In this paper theories of recognition which are independent or dependent on retrieval are compared. A study is described in which the ability and opportunity to retrieve information varies. The results of the study offer some initial evidence that the level of previous knowledge moderates the relationship between retrieval and recognition.

INTRODUCTION

It has often been reported that the advantage of the high knowledge individual on memory tests is attenuated when the task is to recognise instead of to recall previously seen information (Alba 1983; Alba, Alexander, Hasher and Canaglia 1981; Chiesi, Spilich and Voss 1979). The explanation provided for this finding is that recall is heavily reliant on the organisation of information in memory. Well organised information results in the provision of more effectve retrieval cues. High knowledge individuals are better at organising information and, therefore, should be better at recalling information. The explanation for the equivalence in recognition performance is that the process of recognising is not dependent on the presence of retrieval cues, as the test itself provides sufficient information for assessment of previous exposure. Therefore, recognition does not improve as the information in memory becomes more organised. In this case, the advantage of the high knowledge individual disappears (Lynch and Srull 1991).

The independence explanation is tenuous as recognition is used extensively elsewhere in the investigation of the organisation of knowledge (Anderson 1974; 1976; 1983; Reder and Anderson 1980; Reder and Ross 1983). In these studies recognition performance varies significantly when the organisation of information changes. The objective of this paper is to look at how the ability to organise information (which varies with previous knowledge) influences recognition judgments, and whether the opportunity to use the organisation of presented information (which varies with the relatedness of the information) results in systematic differences in recognition performance. The results will be discussed in the context of the role of retrieval in the process of recognition and the moderation of the level of previous knowledge.

The following discussion includes consideration of: the role of retrieval in recognition, the impact of the level of previous knowledge on the process of recognition, and an empirical study designed to test for the effects of varying the opportunity, and ability to use information in recognition judgments. The results are reported and their implications for the process of recognition and the moderation by the level of previous knowledge are discussed.

RECOGNITION

If organisation does not matter. Recognition has been modelled as a process involving a single operation of discrimination, or a familiarity check. The familiarity dictated by the presentation of an item is compared to some criterion level (Lynch and Srull 1991). Recognition is not dependent on the retrieval of any of the original information from memory.

Support for the independence model has been claimed by manipulating the intentionality of learning. Intentional learning results in more organisation of information during encoding. When intentional learning was encouraged, there was no affect on recognition performance (Watkins and Gardiner 1979).

If this model is representative of the process of recognition, then the ability to organise information at encoding is not relevant for recognition performance. Further, any differences in recognition performance between individuals of various knowledge levels should not be systematic, but random.

If organisation does matter. Recognition has also been modelled as either a single process of retrieving information and comparing it to the presented information (Anderson 1974) or as a dual process including both a familiarity check and the retrieval of the context at study. When recognition is a single process, retrieval is guided by the context supplied by the test, and the result is compared to the to-be-recognised information. When recognition is a dual process, the degree of familiarity is assessed first, if the feeling of familiarity is strong or weak enough, it is used as an indicator of previous exposure. However, if the degree of familiarity is judged to be a undiagnostic or an insufficient indicator, retrieval occurs (Mandler 1980).

Support for the dual process model has been found by manipulating the number of categories and the number of items in a category at encoding. The significant correlation between the number of categories and recognition performance is claimed as a demonstration of the importance of the organisation of the information in memory in recognition jdgments (Mandler, Pearlstone and Koopmans 1969).

If recognition does include retrieval at some point in the process, then the ability to organise information at encoding is important for recognition judgments. There should be systematic differences in the recognition performance as the ability to organise information varies. The ability to organise information should facilitate retrieval of information given a learning or study context.

THE ORGANISATION OF INFORMATION AND RECOGNITION

The Fan Effect

Research investigating associative network models of memory use recognition to test the influence of various configurations of information on the probability of retrieval. A robust finding in these studies is that as more facts are learned about a concept, the retrieval of any specific fact becomes more difficult, as reflected in recognition latencies and errors (Anderson 1974; 1976; Anderson and Bower 1973; Hayes-Roth 1977; Lewis and Anderson 1976; Reder and Anderson 1987; Reder and Ross 1983; Whitlow, Smith and Medin 1984). This effect has been called the fan effect and is explained in Anderson’s spreading activation model (1983) as due to sharing of the activation spreading down the links between the concepts (Reder and Ross 1983). The spread of activation is a limited capacity operation, therefore an increase in the number of associations for a particular concept reduces the activation level for each of the associations. The result of a reduction in the level of activation is a decrease in the likelihood of correctly recognising any of the specific facts. An increase in the number of #misses’ (old statements judged as new) in recognition judgments is the consequence (Anderson 1974; Reder and Anderson 1980; Reder and Ross 1983).

The Related Fan Effect

The explanation of the fan effect was criticised because it did not predict an advantage for a high knowledge person when retrieving information about a concept for which considerable information is held (Smith, Adams and Schorr 1978). Reder and Anderson (1980) responded to these criticisms by further specifying the fan explanation. They established that an increase in unrelated facts caused a fan effect, while an increase in related facts tended to attenuate the fan effect, if not reverse it (facilitate retrieval).

One explanation for the attenuation effect is that as individuals accumulate facts which are consistent with a theme they will use sensibility judgments ("all of the attributes for this brand were related to the ease of use benefit, this one is not, I must not have seen it before" or "all of the attributes for this brand were related to the ease of use benefit, this one is also related to that usage situation, I must have seen it before") instead of retrieving the information from memory. If this explanation is correct, as the subject becomes more reliant on the sensibility judgment, there will be a marked increase in the incidence of false recognition (#false alarms’) when the foils (or #new’ items) are related to the theme, and therefore judged as #old’ because they are sensible (Anderson and Reder 1987).

Another explanation for the attenuation of the fan with related facts is that each new fact requires the activation of the other facts during integration. The increase in activation during this process should strengthen the links from brand to attribute to usage situation, therefore allowing for more accurate recognition judgments. In other words, as each new related fact is learned, all of the related facts will be reactivated. An increase in the number of related facts should improve the probability of the retrieval of any of the specific facts due to the frequency and recency of ativation. Target statements then, will be #missed’ less often and foils of any variety should be less often falsely recognised, compared to the unrelated fan condition. In both cases, the information must be organised and integrated to see the relatedness of the fact to a concept.

THE LEVEL OF KNOWLEDGE AND RECOGNITION

If the information is integrated at encoding in the manner expected in the both Smith’s and Anderson’s studies then the organisation of the information at encoding certainly appears to have an affect on the ability to make recognition judgments. In this case, high knowledge individuals, who are better able to organise information, should be influenced by the relatedness of the facts to the concept. Low knowledge individuals, on the other hand, may not be affected by the organisation of the information. Notice that the influence is not always positive. It is possible then that the overall performance on a recognition test could be equivalent although the high knowledge group is affected by the organisation of the information at study while the low knowledge group is not.

Hypotheses:

Overall recognition performance may be equivalent for high and low knowledge individuals, but that does not necessarily mean that the organisation of the information is not an important factor in the process of recognition. High knowledge individuals will be influenced by the organisation of the information at encoding. Specifically, the high knowledge individual:

1] will be less accurate in the recognition of statements when three brands are associated to the usage situation (the fan effect),

2] will be more accurate in the recognition of statements when the same brand is presented with three attributes associated to the same usage situation (related fan),

3] will make more errors in recognition when the foil for a brand is related to the same usage situation presented at study (make sensibility judgments).

The recognition accuracy of the low knowledge individual will not change as the fan configuration varies. The ability to recognise statements will be constant across configurations. Low knowledge individuals will be more likely to miss the target statements (say #no’ to a previously seen item), and less likely to commit false alarms (falsely recognise foils). False alarms will occur when both the brand and attribute were present at study, but not paired together. The relatedness of the foil to a usage situation will not affect the low knowledge individuals evaluation of previous exposure.

EMPIRICAL WORK

Design

To test for the influence of the organisation of information on recognition, subjects of different levels of product knowledge made recognition judgments on information organised into an unrelated fan, a related fan or no fan configuration. Product information in the format of brand-attribute associations were presented to the subjects. In each statement the relatedness of the attributes to a usage situation or the number of brands associated to a usage situation were manipulated.

The design is a 2 x 3 with one measured between subject factor, thelevel of knowledge (high, low) and one manipulated within subject factor, the organisation of the information (fan, no fan, related fan). See Figure 1 for further explanation of the design.

Stimuli

A list of twenty four brand-attribute statements about hypothetical camera brands were presented to each subject. The list was composed of eighteen target statements. Three statements were included in each of the fan configurations (3 statements x 3 configurations), and two sets of each of the three fan configurations were included in the list (2 x 3 x 3). Six of the statements were buffer statements placed at the beginning and the end of the list. No two statements concerning the same brand or the same usage situation were seen consecutively at study or at test.

The order of the statements was counterbalanced on two separate versions of the list.

Brand names. The brand names were randomly assigned to the attributes. The brand names were five letter nonsense words. The words were tested for similarity to other existing brand names and for memorability.

Pretest One-Similarity to Existing Brand Names Pretest

Forty undergraduate students were presented with twenty brand names and asked to:

i] list any characteristics associated with cameras that came to mind when they read,

ii] rate their familiarity with each of the hypothetical brand names,

iii] state whether the brand name reminded them of other existing brand names, and

iv] identify other similar and related brand names.

Two of the original set of brands reminded more than one subject of an existing camera brand, three of the brands (the previous two included) reminded more than one of the subjects of the same #other brand name’. Many of the comments were country of manufacture speculations. Although there was consensus in terms of the country itself, the corresponding associations to cameras manufactured in that country were not systematic.

FIGURE 1

Pretest Two-Memorability Pretest

Another group of twenty students were asked to rate sixteen of the brand names in terms of the #ease of use’ suggested by the brand name. The task was employed to ensure that the subjects processed the brand names. Ten minutes later they were asked to recall as many of the brands as possible. The presentation order was counterbalanced between four lists. When the presentation order was removed as a factor there were no significant differences on the probability of recall between the brands.

The Usage situations. The two usage situations for the related fan are:

1] a camera for use by a professional photographer and

2] a camera that is suitable for a beginner.

The two usage situations for the usage situation fan are:

1] a camera that is appropriate for travelling and

2] an underwater camera for scuba divers and snorkelers.

The attributes were pretested to ensure that subjects could identify the relatedness to the usage situation in the related conditionand the lack of relation in the unrelated conditions (no fan, unrelated fan).

Pretest Three-Relatedness of the Attributes

Attributes were designated as related or unrelated to a usage situation by twenty undergraduate students. This information was used in the preparation of related or unrelated brand-attributes statements. Another group of twenty undergraduate students read the associations and then were asked to rate the brands as to their appropriateness to a usage situation on a ten point scale of ranging from #inappropriate’ to #appropriate’, the middle of the scale was anchored with "don’t know". Subjects also indicated how much they knew about cameras.

Appropriateness ratings for both high and low knowledge students did not differ significantly from #don’t know’ for brands associated to unrelated attributes (no fan and unrelated fan). This implies that the unrelated attributes do not suggest that the brands are appropriate to the usage situation. Brands associated to the related attributes were rated as appropriate to the usage situation (related fan). There were no differences in ratings between knowledge levels.

The Recognition Stimuli. The list used for recognition judgments included thirty six statements. There were eighteen target statements and 18 foils. The foils are either related attribute foils or misassociated attribute foils. Related attribute foils included an attribute which was not seen at study, but was related to the usage situation associated to the brand. Misassociated attribute foils included a brand and an attribute which were seen at study, but not together.

Dependent Measures. The accuracy of recognition judgments is used as the dependent measure.

Independent Measures. The knowledge measure is composed of both subjective and objective knowledge measures. Subjective measures are self-rated items: 1] knowledge, 2] familiarity, and 3] usage. Objective measures are comprised of a score on: 1] terminology questions, 2] attribute definition questions, and 3] brand name listing (actual brands).

Procedure

Thirty six undergraduate university students were awarded course credit in their third year marketing course for their participation in this study. Subjects were told that the purpose of the study was to test brand names for a new brand of camera. Subjects were informed that would see a number of claims about cameras. They were warned not to panic when they saw the claims and did not recognise the brand names. The subjects were instructed to read the claim and rate it as to the appropriateness of the brand name considering the attribute featured in the claim, on a scale ranging from #not at all appropriate’ to #very appropriate’. They were encouraged to take their time and really try to think about the claim. Subjects then saw two sets of each of the three fan conditions in twenty-four brand attribute statements on the screen.

Following the rating session, subjects spent ten minutes on an unrelated task. Subjects were then given instructions for the recognition task and asked to indicate whether each of the statements on the screen had been included on the list seen earlier.

TABLE 1

RECOGNITION PERFORMANCE - PROBABILITY OF CORRECT JUDGEMENTS

RESULTS

As expected, the overall recognition performance for high and low knowledge individuals was not significantly different. Neither an ANOVA of correct recognition (hits), nor an ANOVA of correct rejections resulted in a significant knowledge level factor. A closer look at the results however, uncovers systematic differences in the ability to recognise previously seen product information between subjects of different knowledge levels (see Table 1).

Targets

If the organisation of information matters, there should be differences in the ability to recognise information when organised into a fan or related fan when compared to the no fan configuration. The ANOVA of correct recognitions (hits) mentioned above reveals that although there is an insignificant effect for the level of knowledge factor and the encoding condition, there is a significant interaction of the two effects (F (2, 107) = 4.37, p > .015). A closer look at hit rates reveals that high and low knowledge individuals were as likely to recognise the items in the no fan condition, note that this is the base case in which only one statement was presented per brand and one brand was presented for any particular usage situation. High knowledge individuals however, varied in their ability to recognise the target statements as the relatedness of the information varied from the base case. In the unrelated fan condition, the probability of recognising an item was reduced when compared to the no fan condition (unrelated fan = .63, no fan = .83, p > .007). The fan effect was attenuated when the facts were related (related fan = .83) for high knowledge individuals.

Low knowledge individuals did not vary significantly in their ability to recognise items when the relatedness of the information in each of the configurations varied at encoding. Although the performance of the low knowledge individual remained consistent, the high knowledge individual’s performance varied, resulting in certain situations where the recognition of the low knowledge subject was actually better than the high knowledge individual. Low knowledge individuals were better able to recognise targets in the unrelated fan condition (high knowledge = .63, low knowledge = .79, p < .05), and less able to recognise the targets in the related fan condition (high knowledge = .83, low knowledge .68, p < .04) than were high knowledge individuals.

Are high knowledge individuals more accurate in the related fan condition or are they making sensibility judgments in this condition because all three of the statements made about a brand in this condition are related to the same usage situation? In both cases recognition of the target statements would be more accurate. A closer look at the recognition performance for foils is required to answer the question.

Foils

If organisation matters then errors should be systematically made and meaningful in relation to the organisation of the information. High knowledge individuals made significantly more errors when identifying foils in the related fan condition than in the no fan or the unrelated fan conditions (related fan = .50, no fan = .22, p < .0005, and related fan = .50, unrelated fan = .23 , p < .001). This indicates that as the high knowledge individual learns information about a particular usage situation and the appropriateness of one brand to that situation (related fan), they are likely to begin changing the strategy used in their recognition judgments. Without actually retrieving information from memory, they will assess the likelihood of previous exposure the item’s relatedness to the usage situation (sensibility judgment). This strategy is not possible (with any sort of accuracy) when the accumulated information about a usage situation relates to a number of brands (unrelated fan).

What kind of false recognitions were made?

High knowledge individuals were much more likely than low knowledge individuals to falsely recognise related foils in the related fan condition (high knowledge = .42, low knowledge = .19, p < .02).

Low knowledge consumers were much more likely to falsely recognise misattributed foils (the proportion of related to misattributed errors = .26/.74). This was particularly apparent in the related and unrelated fan conditions compared to the no fan condition. High knowledge individuals falsely recognised related foils as often as they falsely recognised misattributed foils (the proportion of related to misattributed errors = .50/.50).

DISCUSSION

What does all this mean in the context of the role of organisation in recognition judgments?

If recognition was simply a process of assessing the familiarity of a particular item, then whether the brand was the only brand associated to a particular usage situation (no fan), one of three brands associated to a particular usage situation (unrelated fan), or the only brand associated to a usage situation but with three separate attributes (related fan) should have no bearing on the probability of recognition. If however, the process of recognition does involve some retrieval of the information studied earlier or the context within which the information was learned, then the organisation of the information should be influential in recognition performance.

The low knowledge individual, who is thought to be less able to organise information at encoding, is not affected by the number of associations to a usage situation or a brand. The high knowledge individual, thought to be better able to organise information at encoding, is affected by the number of associations to a usage situation or a brand during study. In summary, the organisation of information does have a role to play in recognition judgments as it affects retrieval and the choice of strategy used in making the judgment.

What does all this mean for the role of previous knowledge in recognition judgments?

Low knowledge individuals found it more difficult to reject foils when they were constructed of studied materials, they were not as confused when faced with foils which were related thematically, but not previously seen. It is not clear whether the low knowledge individual retrieves information or assesses previous exposure by the degree of familiarity at test.

High knowledge individuals appear to be retrieving some information during recognition. Further, it seems that the high knowledge individual will shift their recognition strategy from a retrieval based approach to a sensibility strategy, if the accumulated knowledge about a brand allows the use of a different, more efficient approach.

There is plenty of room for further research to understand how and why the recognition performance between consumers with more or less knowledge varies. This study does provide some initial evidence indicating that the organisation of information in memory does have a role in the process of recognition and that the relationship is moderated by the level of knowledge of the individual processing the information.

Future Research

The results discussed in this paper beg some interesting questions for future research. Perhaps most interesting are the errors made in the recognition of foils. First, it appears that it in some cases HKCs do not actually retrieve information when making a recognition judgment, instead they rely on the sensibility of a particular brand having a certain attribute. This strategy is efficient and probably accurate in the main, particularly considering the technical knowledge of the product held by the HKC. However, the results offer preliminary evidence that the HKC draws inferences based on earlier information without acknowledging that the inference is not a fact retrieved from memory. Further research is required to identify the when the HKC realises that the inference is not a piece of information retrieved from memory.

Second, it is interesting that the LKC appears to recognise the brands and the attributes from the study session, but has a lot of difficulty remembering what brand was associated with what attribute. It is likely that the LKC would be quite susceptible to brand confusions or mix-ups of brand and attribute, particularly if xposed to a great deal of information at one time (on a trip to a camera store or while looking through a camera magazine for instance). What might be useful for marketers is to understand when the errors are made by LKCs and how confident the LKC is in these incorrect judgments.

REFERENCES

Alba, Joseph W. (1983), "The Effects of Product Knowledge n the Comprehension. Retention, and Evaluation of Product Information," in Advances in Consumer Research, Vol. 10, Richard P. Bagozzi and Alice M. Tybout (eds). Ann Arbour, MI: Association for Consumer Research: 577-580.

Alba, Joseph W., Susan G. Alexander, Lynn Hasher and Karen Canaglia (1981), "The Role of Context in the Encoding of Information," Journal of Experimental Psychology: Human Learning and Memory; 7, 283-292.

Anderson, John R. (1974), "Retrieval of Prepositional Information from long-term memory," Cognitive Psychology, 5: 451-474.

Anderson, John R. (1976), Language, Memory and Thought. Hillsdale, NJ: Erlbaum.

Anderson, John R. (1983), "A Spreading Activation Theory of Memory," Journal of Verbal Learning and Verbal Behavior, 22: 261-295.

Anderson, John R. and George H. Bower (1973), Human Associative Memory, Washington, D.C.: V.H. Winston.

Anderson, John R. and Lynne M. Reder(1987), "Effects of Number of Facts Studied on Recognition Versus Sensibility Judgments," Journal of Experimental Psychology: Learning, Memory and Cognition, 13(3): 355-367.

Chiesi, Harry L., Spilich, George J. and Voss, James F. (1979), "Acquisition of Domain-Related Information in Relation to High and Low Domain Knowledge," Journal of Verbal Learning and Verbal Behavior, 18, 257-273.

Hayes-Roth, Barbara (1977), "Evolution of Cognitive Structure and Process," Psychological Review, 84: 260-278.

Lewis, Clayton H. and John R. Anderson (1976), "Interference with Real World Knowledge," Cognitive Psychology, 8:311-335.

Lynch Jr., John G. and Thomas K. Srull (1991), "Memory and Attentional Factors in Consumer Choice: Concepts and Research Methodologies," in Perspectives in Consumer Behavior, Fourth Edition, Harold H. Kassarjian and Thomas S. Robertson (eds). Englewood Cliffs, NJ: Prentice-Hall, Inc.: 101-129.

Mandler, George, Zena Pearlstone and H.J. Koopmans (1969), "Effects of Organisation and Schematic Similarity on Recall and Recognition," Journal of Verbal Learning and Verbal Behavior, 8:410-423.

Mandler, George (1980), "Recognising: The Judgment of Previous Occurrence," Psychological Review, 87: 252-271.

Reder, Lynne M. and John R. Anderson (1980), "A Partial Resolution of the Paradox of Interference: The Role of Integrating Knowledge," Cognitive Psychology, 12: 447-472.

Reder, Lynn M. and Anderson, John R. (1980), "A Partial Resolution of the Paradox of Interference: The Role of Integrating Knowledge," Cognitive Psychology, 12: 447-472.

Reder, Lynne M and Brian H. Ross (1983), "Integrated Knowledge in Different Tasks: The Role of Retrieval Strategy on Fan Effects," Journal of Experimental Psychology: Learning, Memory and Cognition, 9(1): 55-72.

Smith, Edward E., N. Adams and D. Schorr (1978), "Fact Retrieval and the Paradox of Interference," Cognitive Psychology, 10: 438-464.

Watkins, Michael J. and John M. Gardiner (1979), "An Appication of Generate-Recognise Theory of Recall," Journal of Verbal Learning and Verbal Behavior, 18:687-704.

Whitlow Jr., J.W., Edward E. Smith and Douglas L. Medin (1984), "Retrieval of Correlated Predicates," Journal of Verbal Learning and Verbal Behavior, 21:383-402.

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

Authors

Elizabeth Cowley, University of Western Sydney, Nepean



Volume

AP - Asia Pacific Advances in Consumer Research Volume 2 | 1996



Share Proceeding

Featured papers

See More

Featured

Approach and Loss Aversion: Consumer Responses to Approaching and Receding Stimuli in Advertising

Lana Mulier, Ghent University, Belgium
Iris Vermeir, Ghent University, Belgium
Hendrik Slabbinck, Ghent University, Belgium

Read More

Featured

The Re-Mediation of Consumer/Brand Relationships Through Voice Shopping: The Case of Amazon Echo

Johanna Franziska Gollnhofer, University of Southern Denmark, Denmark

Read More

Featured

The Influence of Goal Specificity on Planning Effectiveness: A Meta-Analysis

Ekaterina Napolova, University of Technology Sydney
Francois Carrillat, University of Technology Sydney
Leona Tam, University of Technology Sydney

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.