The Effects of Familiarity on Cognitive Maps

ABSTRACT - This paper investigates the impact of product familiarity on consumers' cognitive structures. Using two low involvement product categories, breakfast cereals and fast food restaurants, differences in cognitive structures derived from similarity ratings and product attribute descriptions were examined. The results indicate that, for low involvement products, product familiarity has almost no impact on perceptual structure and organization but it does have an effect on the consistency (e.g., the degree of stress) of the derived structures.


Haim Mano and Scott M. Davis (1990) ,"The Effects of Familiarity on Cognitive Maps", in NA - Advances in Consumer Research Volume 17, eds. Marvin E. Goldberg, Gerald Gorn, and Richard W. Pollay, Provo, UT : Association for Consumer Research, Pages: 275-282.

Advances in Consumer Research Volume 17, 1990      Pages 275-282


Haim Mano, Washington University in St. Louis

Scott M. Davis, Washington University in St. Louis


This paper investigates the impact of product familiarity on consumers' cognitive structures. Using two low involvement product categories, breakfast cereals and fast food restaurants, differences in cognitive structures derived from similarity ratings and product attribute descriptions were examined. The results indicate that, for low involvement products, product familiarity has almost no impact on perceptual structure and organization but it does have an effect on the consistency (e.g., the degree of stress) of the derived structures.


Consumer researchers often use perceptual mapping techniques as a basis for assessing the basic cognitive dimensions consumers use to evaluate alternatives within a product category or the relationships among the alternatives with respect to those dimensions (Carroll 1972). The use of these techniques is often based on the assumption that there is no significant variation across consumers in the manner in which alternatives are judged.

However, there are several sources of variations among consumers that may influence the judgmental process. Different consumers are often exposed to different types of information about the alternatives. Furthermore, consumers are sometimes exposed to a product directly through one or more usage occasions while in other cases information about a product is only available indirectly through word of mouth or advertising. Consumers also tend to vary in the frequency of exposure to the different information sources. The extent to which and the way information is processed are also likely to vary across consumers depending on their levels of involvement in a product category and their predispositions toward the alternatives within the category. Due to the fact that consumers normally vary in terms of their experiences with the judged alternatives or in terms of overall product knowledge, differences in product familiarity may influence the way judgments are made. In particular, knowledgeable consumers may exhibit superior capacities to process and integrate product-relevant information (Johnson and Russo 1984). Thus variations in product knowledge are likely to influence both the manner in which judgments are made and the consistency of the reported judgments.

Familiarity and expertise are major components of consumer knowledge and should have a considerable impact on consumer cognitive structures (Alba and Hutchinson 1987). Familiarity is defined as the number of accumulated product related experiences while expertise is the ability to perform product-related tasks. However, for low involvement, experience related, and nontechnical products--like those to be examined here--these two components are highly interrelated and perhaps indistinguishable.

This paper presents an attempt to examine the impact of product knowledge on cognitive representations produced-by perceptual mapping techniques. Recent research suggests that familiarity with a product category is likely to have an impact on these representations (cf. Srinivasan, Abeele and Butaye 1989). As consumers become more familiar and knowledgeable with a product category, there are at least two apparent consequences regarding the structure of cognitive representations. First, one would expect knowledgeable consumers to develop more clearly established criteria for making their judgments which in turn could lead to the development of different underlying dimensions for judging a product. Second, more knowledgeable consumers could apply these criteria more consistently; in that case, we would expect differences in familiarity to be a source of systematic variations in the perceptual maps produced by the various mapping techniques as well as in the measures of goodness of fit associated with them. Furthermore, since different mapping techniques require different types of judgments, we would expect the impact of product knowledge to depend on the mapping technique employed.


This study is part of a current research program investigating the interrelationships between consumers' evaluations and product familiarity for two low involvement products, breakfast cereals and fast food restaurants. We consider the influence of product familiarity on perceptual maps generated by two different types of data perceived similarities between product pairs and product attribute ratings.

Similarity measures are based on subjects' pairwise comparisons of stimuli (e.g., brands) and are generally analyzed by employing multidimensional scaling techniques (MDS) (e.g., INDSCAL; Carrol and Chang 1970). The results of these techniques allow one to reveal the dimensions used by consumers when evaluating the stimuli or to confirm whether a particular dimension is salient in consumers' judgments for that product category. Attribute ratings of products can also be used as a basis for deriving perceptual maps. Factor analytic techniques use correlations among attributes and products in order to determine the basic perceptual dimensions in forming product judgments.

Theoretically, if the set of attributes is relatively complete and all attribute information is adequately processed when forming similarity judgments, then maps produced by MDS and factor analytic techniques should be similar (cf. Hauser and Koppelman 1979). However, if the consistency of attribute judgments or the set of attributes in forming judgments varies systematically as a function of familiarity with product category, then we would expect the maps produced by the various techniques also to vary with product familiarity.

The analyses to be presented here will allow us to examine three issues regarding product familiarity and cognitive structure for low involvement products: (1) whether more product-familiar subjects have more robust and consistent cognitive configurations, (2) whether the relationships between brands in these configurations will differ as a function of product familiarity, and (3) whether different perceptual mapping techniques will reveal different cognitive configurations.

In this study we will consider two low involvement, experience related, and nontechnical products, breakfast cereals and fast food restaurants, which are relevant to our surveyed population: college students. A unique feature of both product categories is the large number of available brands, their relatively frequent consumption, and the high variabilities in consumption and familiarity for our surveyed population. Furthermore, employing two different product categories allows for more extensive and convergent validation of any hypothesized contentions.


In order to generate the list of products and product attributes employed in the present study, 77 undergraduate business students were asked to list the breakfast cereals they could recall while a another group of 78 students generated a similar list for fast food restaurants serving hamburgers. In addition, the students were asked to list the attributes they felt were most important in their consumption decisions of these products. From these lists, ten of the most frequently mentioned brands and product attributes were selected for use in the study.

Two weeks following the initial survey, 129 students completed questionnaires concerning perceptions and preferences with regard to alternatives and attributes of one of the two products. Based on a random assignment to one of the two product categories, 63 subjects completed questionnaires relating to breakfast cereals and 66 completed questionnaires relating to fast food restaurants. Most of these subjects were part of the initial 155 subjects used to generate the lists of products and attributes. All subjects in the study were undergraduate business students at a private midwestern university.

Questionnaires were divided into four sections which could be classified as (1) familiarity, (2) preferences, (3) perceived similarities among products, and (4) perceived attribute ratings of the products. The sequence of the sections in each questionnaire was randomized in order to minimize the impact of any learning effects that might have taken place during questionnaire completion. In addition, products and attributes were each arranged in two separate random sequences so as to minimize any effects of establishing reference points that could be attributed to the sequence of items in the questionnaire.

The familiarity section consisted of two sets of questions. The first set asked subjects to report, based on their consumption habits ova the course of the previous year, their average monthly consumption frequencies for each brand listed and for the product category as a whole. The second set of questions asked subjects to rate, on a zero to ten scale, their familiarity with each of the brands (based on personal consumption, family, friends, advertising, etc.).

The preference section consisted of five sets of questions that elicited subjects' preferences regarding products and attributes. The first two sets asked subjects for their brand preferences through ratings on zero to ten scales and rank orderings, respectively. The remaining three sets of questions asked subjects for their attribute preferences through rating the importance of the attributes on a zero to ten scale, a ranking of the ten attributes in order of their importance, and a description of their most preferred attribute levels on zero to ten scales.

The remaining two sections concerned subjects' perceptions regarding the brands tested in the study. The similarities section asked subjects to compare each product to each of the remaining nine products and rate how similar the products were on a zero to ten scale in which low ratings corresponded to an evaluation that the products were very similar while high ratings corresponded to an evaluation that the products were very different. The attribute ratings section asked the subjects to evaluate the attribute levels of each of the products on a zero to ten scale.


Overall Product Familiarity

Product knowledge can be assessed by subjective measures (i.e., perceptions), learned knowledge based on objective criteria, and usage experience (Brucks 1985). In the present study, product familiarity was assessed by subjective measures. Despite theoretical advantages of employing objective measures (Brucks 1985; Sujan 1985) and possible distortions due to memory error and mental calculations, we believe that the low involvement nature of the products used here justifies the use of subjective measures of familiarity. Furthermore, for both product categories, actual brand consumption (which reflect usage experience) and brand familiarity were highly correlated. For the cereal category, the correlation: ranged from .30 to .42 (all p's < .01) with an average of .35. For the fast food category, except for the most widely known brand (McDonald's r = .14), correlations between brand consumption and brand familiarity ranged from .37 to .71 (all p's < .01) with an average correlation of .45.

Given the convergence between brand consumption and reported brand familiarity, reported brand familiarities were used to operationally define product familiarity. Subjects were considered familiar with the product category if they rated themselves as highly familiar with the individual brands in the category. Specifically, a measure which will be referred to as Product Category Familiarity was defined as the sum of the ten questions regarding the subject's degree of subjective familiarity for each brand. Theoretically, the value of this index could range from 0 (for a subject who responded 0 for ail 10 products) to 100. Actual values ranged from 22 to 98 for cereals and from 26 to 90 for fast food restaurants. The mean familiarity rating for the breakfast cereal category was 68.96 with a standard deviation of 16.96; the mean for the restaurant category was 57.80 with a standard deviation of 13.68. Using the below the 33rd and above the 67th percentiles of the distribution on Product Category Familiarity, three subgroups of Low, Medium, and High Familiarity were formed.

Familiarity and preferences

Before investigating the interrelationships between familiarity and perceptual configurations, it is important to examine two issues: how brand familiarities and brand preferences were interrelated and whether preferences for the various brands differed for the different groups of overall product familiarity. The importance of these two questions stems from the particularly low involvement nature of the two products employed in the present study. For, if the group of higher product familiarity subjects shows a higher preference for particular brands, then any differences in perceptual configurations may be attributed to evaluative preferences and not solely to cognitive judgments.

We therefore anticipate that brand familiarities and brand preferences will be highly related for low involvement and frequently consumed products. Yet, at the same time, we expect to find the same patterns of brand preferences across the three levels of product category familiarity. In other words, because of the low involvement nature of the products, we do not expect substantial differences in the preference for, say, "Hardee's," between product familiar and nonfamiliar subjects. It should be noted, however, that this assertion is likely to hold only for low involvement products; for higher involvement products, more knowledgeable consumers are likely to express preferences for particular (e.g., higher quality) brands.

In the cereal category, the correlations between brand familiarity and brand preference were quite high; the values for the ten correlations ranged from .33 to .57 (all p's<.01) with an average of .47. Moreover, using univariate ANOVAs, no statistically significant differences were found in preferences for each of the ten cereals by the three subgroups of familiarity. The grand means for preferences arranged for the ten products were: Frosted Flakes, 7.19; Fruit Loops, 6.73; Cap'n Crunch, 6.34; Rice Krispies, 5.84; Post Raisin Bran, 5.71; Lucky Charms, 5.60; Cheerios, 5.07; Kellog's Corn Flakes, 4.76; Grape Nuts, 4.04; and Total, 3.94. As with cereals, the correlations between brand familiarity and brand preference in the restaurant category were quite high (correlations ranged from .26 to .70 with an average of .52). Except for one product (Steak y Shake, F(1,63)=4.74, p=.01), ANOVAs did not reveal significant differences in brand preferences by the three levels of familiarity. This product was given high preference ratings by those who were familiar with it, suggesting that Steak'n Shake is a desirable product that does not have a high quality image product among uninformed subjects. In general, however, these results suggest that Product Category Familiarity did not influence brand preferences even though brand familiarity was strongly related to brand preference for specific brands.

Similarity ratings and Replicated Multidimensional Scaling

MDS analyses were conducted using two different approaches. The first approach is replicated multidimensional scaling (RMDS) which allows for a joint map of objects and individuals (Young 1975). -The second approach uses the average similarities of the relevant groups to construct the multidimensional scaling maps. The difference between the approaches is that the first deals with N similar matrices (where N = number of subjects) whereas the second deals with one matrix of average similarities.

Stress coefficients and percent of explained variance associated with RMDS analysis are presented for the various familiarity groups in Table 1.

Based on the criteria of stress and interpretability, the one dimensional solution seems to be the most appropriate for the cereal category and captures the tradeoff between artificial sweetness and naturalness. The coordinates for this dimension were: Cap'n Crunch, 1.39; Fruit Loops, 1.34; Lucky Charms, 1.24; Frosted Flakes, .86; Rice Krispies, .59; Kellogg's Corn Flakes, -.64; Post Raisin Bran, -.87; Total, -.82; Cheerios,- .89; Grape Nuts, -1.01.

Simple correlations between the coordinates on the one dimensional solutions and canonical correlation coefficients between the coordinates of the two dimensional solutions, revealed that the four populations (whole, low, medium and high familiarity) resulted in extremely similar configurations (all simple and canonical correlations were above .99; p <.001).

As indicated in Table 1, in the fast food category, the one- and three-dimensional solutions for higher levels of familiarity were accompanied by lower stresses and higher levels of explained variance. However, for the two-dimensional solution, stress levels increased with familiarity; nonetheless, the percentages of explained variance remained at about the same levels. For the combined population, the two-dimensional solution seemed to be the most appropriate in terms of stress and interpretability. The first underlying dimension was interpreted as prototypicality of product offerings; the second dimension captured the perceived degree of quality of these restaurants. The whole population's two-dimensional solution is presented in Figure 1. The two-dimensional RMDS solutions for the four populations (whole, low, medium and high familiarity) were similar among themselves and with that of the whole population (all canonical R's>.80; p's<.01).





In order to further investigate the relationship between familiarity and perceptual configurations of similarity, the matrices that were based on the average similarities of the whole population and the low and high familiarity groups were also examined. For the cereal category, the one-dimensional solution seemed to be the most acceptable both in terms of stress (stress = 0.147, 94% explained variance) and interpretability. As anticipated, the stress coefficient for the low familiarity group was higher than that of the high familiarity group; (stress = 0.184, 89% explained variance versus 0.116, 96%). As in the RMDS analysis, the underlying dimension was interpreted as Sweetness-Naturalness. The coordinates in the RMDS and present analyses had a correlation of 0.97. In the fast food restaurant category, the perceptual maps based on average similarities had stress values .284, .121 and .077 for the one, two, and three dimensional solutions respectively. As indicated in Table 2, the difference in stress between the low and high familiarity groups were in the hypothesized direction with the maps of the more familiar groups exhibiting lesser stress.





Factor Analyses: Product descriptions on performance attributes

For each of the four relevant populations (whole, low, medium and high levels of familiarity), factor analyses were conducted using product descriptions on the ten attributes across subjects as observations. This analysis provides a product space whereby each brand is positioned closer to another brand if the two brands were given similar attribute descriptions.

The analyses in the cereal category for each of the four relevant populations revealed two factors with eigenvalues greater than unity. Table 3 presents the cumulative variances explained by the first two dimensions for the relevant populations. For the one factor solution, the cumulative explained variance did increase with overall product familiarity; however, this effect was not present in the two factor solution.

The whole population and the medium and high familiarity groups, had very similar one factor solutions (r's >.995; p<.0001). For the whole population, coordinates on that factor were similar with the-sweetness-nutrition dimension obtained in the RMDS one-dimensional analysis of product similarities; the correlation between the two sets of coordinates was .672 (p = .03). The low familiarity group, however, resulted in a different factor configuration; the correlation between its coordinates and those of the one factor solution of the whole population was .485 (n.s.). An examination of the two factor solutions revealed that the second factor was not easily interpretable, even though it made a considerable contribution to the explained variance. Nonetheless, the three familiarity groups had highly similar two factor configurations; the canonical correlations between the two factor solutions for the three subgroups and the whole population solution were above .97 (p's <.01). Figure 2 presents the two-dimensional solution for the whole population. This two factor solution was only moderately similar to the whole population's RMDS two-dimensional solution (canonical R = .77, p<.20).



In the fast food category, three factors with eigenvalues greater than one were revealed for all familiarity groups except for the moderate familiarity which revealed four such factors. However, the eigenvalue for the fourth factor was 1.1; therefore, due to the eigenvalue's proximity to unity and the small number of objects, this factor was excluded from further analyses. As can be seen in Table 2, the hypothesized relationship between overall familiarity and cumulative explained variance is, overall, confirmed. In particular, comparisons between the low and high familiarity groups revealed considerably higher cumulative explained variance for the high familiarity group. The two factor solution seemed the most appropriate and is presented in Figure 3. Finally, it is noteworthy that this two factor solution was only moderately similar to the whole population's RMDS two-dimensional solution (canonical R = .90, p <.02).


The results suggest that overall product familiarity may be an important determinant of consumer's cognitive structures, even for such low involvement products as the ones employed here. In particular, more product-familiar subjects had more consistent cognitive configurations than those exhibited by less familiar subjects. It appears that increased product familiarity enables the consumer to differentiate brands according to more veridical criteria (Alba and Hutchinson 1987; Johnson 1984) and apply these criteria more consistently. In terms of inter-brand relationships, however, no considerable differences were revealed among the different groups of product familiarity. Thus the major difference between the lower and higher familiarity groups' maps were not as much in terms positioning of the objects but rather on the fuzziness of the maps.

Hauser and Koppelman (1979) suggested that factor analytic techniques are superior to MDS for identifying consumer perceptions when there is variation in the way consumers perceive products in the category. In contrast, our comparisons between the two perceptual mapping techniques showed that, at the level of interpretable solutions, multidimensional and factor analytic techniques resulted in quite comparable configurations. For example, when considering the whole population for cereals, the one dimensional RMDS and the one factor solution were strongly correlated (r=.672, p = .03). The comparable analysis, however, for the fast-food restaurants revealed a correlation of .40 (n.s.). Nonetheless, when two dimensions and two factors were jointly examined for fast foods, the two solutions were very similar (canonical correlation .90, p<.02). This result suggests that the attributes contributing to the third factor were not consistently utilized by subjects when forming similarity judgments and simplifying heuristics may have been implemented when making these judgments.



Our results presented aggregate analyses of the interrelationships between perceptual configurations and familiarity. An analysis of these interrelationships on an individual level may provide useful insights into the factors influencing consumers' judgments of products. This and other related topics provide a promising field for future research.


Alba, Joseph W. and J. Wesley Hutchinson (1987), "Dimensions of Consumer Expertise," Journal of Consumer Research, 13, 411-454.

Brucks, Merrie (1985), 'The Effects of Product Class Knowledge on Information Search Behavior," Journal of Consumer Research, 12, 1-16.

Carrol, J. Douglas (1972), "Individual Differences and Multidimensional Scaling". In R. N. Shepard, A. K. Romney and S. Nerlove (eds.) Multidimensional Scaling: Theory and Applications in Behavioral Sciences, Vol. 1. Theory, New York: Seminar Press, Inc., 105-155

Carroll, J. Douglas and Jih-Jie Chang (1970), "Analysis of Individual Differences in Multidimensional Scaling via an N-way Generalization of 'Eckart-Young' Decomposition," Psychometrica, 35, 238-319.

Hauser, John R. and Frank S. Koppelman (1979), Alternative Perceptual Mapping Techniques: Relative Accuracy and Usefulness," Journal of Marketing Research, 16, 495-506.

Johnson, Eric J. and J. Edward Russo (1984), "Product Familiarity and Learning New Information," Journal of Consumer Research, 11, 542-550.

Johnson, Michael D. (1984), "Consumer Choice Strategies for Comparing Noncomparable Alternatives," Journal of Consumer Research, 11, 741 -753.

Srinivasan, V., Pierre Vanden Abeele and Ignace I. Butaye (1989), 'The factor structure of multidimensional response to marketing stimuli: A comparison of two approaches," Marketing Science, 8(1), 89-100.

Sujan, Mita (1985), "Consumer Knowledge: Effects on Evaluation Strategies Mediating Consumer Judgments," Journal of Consumer Research, 12, 31 -46.

Young, Forrest W. (1975) "An Asymmetric Euclidean model for multi-process asymmetric data." U.S.Japan Seminar on Multidimensional Scaling.



Haim Mano, Washington University in St. Louis
Scott M. Davis, Washington University in St. Louis


NA - Advances in Consumer Research Volume 17 | 1990

Share Proceeding

Featured papers

See More


Product Retention (vs. Acquisition) Choices and Preference for Person-Related Features

Liad Weiss, University of Wisconsin - Madison, USA

Read More


Unexpected-Framing Effect: Impact of Framing a Product Benefit as Unexpected on Product Desire

Monica Wadhwa, INSEAD, Singapore
Christine Kim, Hong Kong University of Science and Technology
Amitava Chattopadhyay, INSEAD, Singapore
Wenbo Wang, Hong Kong University of Science and Technology

Read More



Evan Polman, University of Wisconsin - Madison, USA
Sam J. Maglio, University of Toronto Scarborough

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.