The Application of Prototypes and Catgorization Theory in Marketing: Some Problems and Alternative Perspectives

S. Ratneshwar, University of Florida
Allan D. Shocker, University of Washington
[ to cite ]:
S. Ratneshwar and Allan D. Shocker (1988) ,"The Application of Prototypes and Catgorization Theory in Marketing: Some Problems and Alternative Perspectives", in NA - Advances in Consumer Research Volume 15, eds. Micheal J. Houston, Provo, UT : Association for Consumer Research, Pages: 280-285.

Advances in Consumer Research Volume 15, 1988      Pages 280-285

THE APPLICATION OF PROTOTYPES AND CATGORIZATION THEORY IN MARKETING: SOME PROBLEMS AND ALTERNATIVE PERSPECTIVES

S. Ratneshwar, University of Florida

Allan D. Shocker, University of Washington

Concepts such as "prototypes" and "typicality", adopted from the psychological literature on categorization, have been applied in a number of different areas in marketing in recent years. The investigations that have used these concepts range from studies in consumer behavior (e.g., knowledge representation, choice set formation, information processing in salesperson encounters) to those concerned with more strategic issues (e.g., relationship between hierarchical market structures and categorical structures, order-of-entry effects in new product introduction). Of present interest is the fact that such a categorization perspective has proven to be useful in providing a psychological explanation for empirically observed "order of entry" effects on sales of competing brands in product-markets (Carpenter and Nakamoto 1987; Marlino and Sujan 1987). Here the argument is made that the first entrant in a category "establishes the category" and as a consequence is most likely to become the prototypical brand or exemplar for that category. The first brand to enter a market and establish a category might enjoy higher acceptance (sales) than others (Robinson and Fornell 1985; Urban, et al. 1986) and thus gain for itself a "pioneering advantage". The recent experience of Coca Cola in rescinding its decision to drop original Coke in the face of substantial customer outcry can, for example, be "explained" in these terms. As the first carbonated soft drink, it could be said to have established the category and become the prototype for a significant number of customers. The concept of an improved now product could not be accepted by those customers, since Coke was the standard. (Since different customers enter the product-market at different times, they may form similar categories with different brands, c.g., Pepsi, as prototypes. Such a process would still permit the first to enter to possess the largest market share, yet not necessarily force such a result.)

Marlino and Sujan (1987) have elaborated on the above perspective by suggesting that individuals form expectations of the product category as a whole based on the pioneering exemplar and reference such expectations when making evaluations of subsequent product category entrants. This is consistent with the psychological research tradition dating back to Bartlett (1932) which holds that the role of past experience is to provide abstract cognitive tools (e.g., schemata) that help an individual to cope with new experiences. But in addition to its explanatory contribution to a psychological theory of consumer behavior, research that is capable of revealing such categorizational (and expectational) structure may have substantive normative value if it can lead to coherent product-market definitions.

The purpose of this paper is to note certain important problems and possible caveats which need to be better understood, and addressed through future research, if the above contribution to managerially relevant phenomena such as market structure, order of entry effects, etc. is to be realized from paradigms based on categorization theory. While the resolution of some of these problems may lie in terms of defining boundary line conditions within which generalizations can be made from particular investigations, this may be an opportune time to selectively review the problems, particularly those which are of broader relevance.

Category Formation and Identification: An important issue that needs to be addressed is the process by which pre-existing cognitive categorical structures are modified to accommodate new categories. From a research standpoint, this issue is germane to an identification problem: How does one draw the boundaries around a new category, and can one offer a defensible definition for distinguishing this category from other collections or groupings of objects? New category formation does not appear to be an automatic process in response to an encounter with something new, or even a process that is easy for the marketer to initiate. The marketing of a technologically "new to the world" product may serve to stimulate consumer interest or awareness, but this need not imply a new category will necessarily form around it. Consumers may at first try to fit the new object into their existing categorical knowledge structures, and to ascribe meaning to the new thing in terms of what they already know. Quite independent of technological or producer-side evaluations of the "degree of newness", the new product may be assimilated to cognitive structures with minimal effects on the latter, or it may be simply ignored. The following example from Rogers and Shoemaker (1971) serves to illustrate this point.

A change agent sent by the public health service department attempted, over a two-year period, to persuade the families in a peasant village in Peru to adopt an innovation. The innovation involved boiling water on a regular basis before consumption, to minimize the risk of water-borne disease from water sources that were known to be contaminated. Ultimately the persuasion campaign failed with barely 5% of the families adopting the innovation. Rogers and Shoemaker (1971, p. 5) attribute this failure chiefly to the manner in which the traditional (cultural) beliefs of the local population related to the innovation. Boiling water was categorized into a more general category of "making cold foods hot"; the peasants were highly familiar with the latter activity, but - unfortunately for the change agent- only in the context of something a person did if he was sick or chronically invalid. The novelty of the innovation was lost in the process of assimilation to existing belief or knowledge structures.

If one accepts the argument that what really matters is the meaning conferred upon the new product by potential adopters (Bartlett's, 1932, "effort after meaning"; Piaget's "assimilation'), rather than producer-side perceptions of newness, then changes in preexisting categorical structures have to be viewed as a gradual, adaptive process. From a research standpoint, this clearly complicates the problem of identification. It can be argued that nothing is "really" now. Coca Cola may have been first perceived by consumers merely as carbonated water with a different taste; the automobile as a horse-less carriage; television as a movie within the home or as radio with pictures. Sociologists talk of continuous and discontinuous innovations is an attempt to acknowledge this fact. But even from a producer standpoint, it may be difficult to identify the discontinuities. Crawford (1983) has discussed this issue in the context of the Jewkes, Sawers and Stillerman (1959) study, The Sources of Invention. Crawford points out that the lack of sharp breaks or discontinuities led the above authors to be highly situational in their identification of inventions and inventors.

The above identification problem is compounded if one makes a distinction between technological invention and successful innovation (Crawford, 1983). The latter involves the additional effects of new product management. The ability of the firm to promote and market its innovation will also affect the likelihood that the new product comes to serve as exemplar of some new category. A product called Lestoil was the first liquid cleanser to be marketed, but it is no longer available. Mr. Clean, Handy Andy, and Ajax are later brands that have survived in this category. Presumably, one important reason for this lies in the inferior marketing capability or inadequacy of the resources devoted to marketing the first entrant. Primacy can only be important if the first to enter effectively markets its advantage and does whatever things lead to new category formation. The rate at which eventual customers encounter the products in a category should possibly also have an effect on which becomes the more accepted exemplar. If competition enters a market vert quickly or enters and promotes its products at a time when rapid growth is capable of occurring, then a later entrant could well become the exemplar to the largest number of customers. The presence of these confounds suggests that historical data on new product successes need to be interpreted with caution when such success is interpreted in terms of first entrants becoming prototypes of new categories [Caution in such interpretation is also indicated from a theoretical standpoint, given that classic studies on visual patterns such as Posner and Keele (1968) and Franks and Bransford (1971) did not find serial position effects in prototype abstraction.].

Importantly, the entry of directly competing, similar objects (e.g., additional brands) may accentuate differences perceived between new and existing products and imply that this collection of "new objects" requires distinct categorization. If imitative competition should fail to develop, perhaps the existing categorization structure would be more likely to remain unchanged [This provides an illustration of a case where a firm may welcome direct competition because its presence makes it easier to distinguish the entire product class and by so doing achieve differential growth.]. Such cues may serve to enrich categorical structure as individuals learn to discriminate and, through that process, pay attention to different properties or features of the new objects. Seven-up may have helped distinguish itself (and lemon-lime beverages generally) from other carbonated beverages by promoting itself as the "uncola" and later emphasizing that it never had caffeine. The appearance of numerous products based upon digital technology and the existence of high resolution computer monitors likely help in establishing digital televisions as different from conventional receivers and thus may help to justify their higher prices for some customers. While empirical research in a consumer setting is needed to establish the validity of the above speculations, a theoretical foundation for such research is available in the psychological literature through the work of Gibson (1969) on perceptual learning, and those of Garner (19 14) and Tversky (1977) on the "diagnosticity" provided by the "context of alternatives". Objectivist vs. Relational Approaches to Categorical Structure: Two closely related constructs, similarity and typicality, underlie most contemporary psychological accounts of categorical structure (Smith and Medin, 1981). These constructs are "explained" in various structural models by decomposition of objects into elements ("primitives") referred to as features, attributes, properties, etc., with such terms used almost interchangeably. For example, Tversky's (1977) well-known set-theoretic treatment of similarity represents this construct as a linear combination, or contrast, of measures of the common and distinctive features of a pair of objects. Rosch and Mervis (1975) relate typicality to "family resemblance", which is a measure of the extent to which an object shares attributes with all other category members.

But as Rosch and Mervis (1981) acknowledge, there has been much controversy over the level of abstraction at which a particular kind of attribute decomposition can or cannot be said to occur. Attributes that are considered as "primitives" are usually believed so through default rather than by any explicit logic. In most empirical investigations subjects have routinely been given attribute listing tasks (e.g., Rosch and Mervis, 1975) and it has been assumed that subjects somehow "automatically" find the "right" level of analysis. A major criticism of "objectivist" decompositional approaches (e.g., see Murphy and Medin, 1985), it was originally referred to as the problem of presupposition of attributes by Cassirer (1953). Consider for instance the several different levels of aggregation at which one might choose to define "feature" in an automobile exhaust system. The level of analysis is likely to depend on situational goals (e.g., a car emission inspection is imminent), and also on one s expertise in perceiving how various components are meaningfully related into organized "wholes" (e.g., the differences in perspective between the car owner and the skilled repair mechanic).

A closely related problem to the above is that the standard (objectivist) account of categories looks upon attributes as properties of objects in themselves, rather than as properties that get defined in a relational sense out of the purposeful interactions that people have with their world (For elaboration of this criticism see Nelson, 1974; Bransford and McCarrell, 1974; Miller and Johnson-Laird, 1976; Lakoff and Johnson, 1980). Interactional properties, such as the "graspability" of a tennis racket, cannot be defined independent of characteristics (e.g., physiological) of people. Attempts at accommodating such "functional features" into the objectivist view (e.g., Smith and Medin, 1981, p. 26-27) ignore, at the very least, the problem created by variability in the relevant subject level characteristics.

The problem of using objects in isolation as the unit of analysis is acknowledged by Rosch (1978). Rosch points out that in order to find a unit of analysis relevant to the real world, attention has to be turned to the contexts in which objects occur: 'To the culturally defined events in which objects serve as props," (1978, p. 42). In the same vein, Murphy and Medin (1985) have pointed out the inadequacies of using similarity as the basic construct for categorical structure, and state that "Human interests, needs, goals, and theories are ignored" (p. 295) in current theories.

In contrast to the objectivist accounts, Nelson (1974, 1985) has drawn on the work of Piaget and Cassirer to propose a relational theory wherein the formation or acquisition of the category concept involves attempts by the individual to comprehend instances of the category within an overriding function or relational rule rather than through the specification of attributes. Nelson (1974) suggests an all-important distinction between what the category concept means, and how instances of the category can be identified. Nelson points out that the psychological literature has not generally distinguished between the two, with the result that identifying attributes have often been taken to be the basis for categorical concepts (for example, in decompositional structural models). But from a process perspective, analysis of the whole (object) into its parts (attributes) may be unnecessary to a meaningful category concept. On the contrary, the object may first be comprehended in terms of important functional relationships: What Cassirer (1953) terms "logical acts". Perceptual analysis which leads to attribute decomposition may be a secondary process, and not a priori essential to the existence of a meaningful category concept.

The above distinction between what a category concept means, and how instances of the category can be identified, is likely to be less important in the case of simple perceptual stimuli such as geometric figures, colors, etc. But the distinction becomes highly important when the phenomena of interest are objects of pragmatic use (or as William James would put it, "sensible objects"). For example, inter-object similarity may be based on many different dimensions (Nelson, 1914), of which the static, perceptual dimension of shape or form, which is the dominant dimension in identification, is only one. Others include function, action, usage and affect. It should be noted that these latter dimensions, unlike the first one, are usually relational: They involve relations between the object and people, events and other objects. Nelson's (1974) position is that a person's concept of an object category may originally develop out of a synthesis of the various relations and acts into which a new object enters. Feature representation, necessary for identifying other instances of the category, may be a later development.

There are several important research implications to be derived from the above discussion. First, it poses problems for use of the standard, objectivist categorization literature as a theoretical base for the study of consumer behavior. Thought needs to be given to whether a literature that is, self--admittedly, so critical of its own remoteness from human needs, goals, and events can really be related to the purposeful behavior of consumers. The recent work of cognitive psychologists such as Barsalou (1983, 1985) on goal-related categories indicates that it may be possible to bridge the gap, but only with considerable revision in the nature of the structural constructs. Second, it suggests the value of a relational approach to the study of useful constructs such as similarity and typicality. The attribute listing tasks commonly employed in many studies may be primarily accessing the identification or cue validity component of categorical representation in memory. But as mentioned earlier, this component may be only a small part of the total category concept. For example, Ratneshwar (1987) investigated similarity and typicality judgments in the snack food category on the basis of structural models where the predictor variables involved judgments of product usage appropriateness across situations. It was verified that a linear composite of the number of common and distinctive usages of pairs of products explained a substantial portion of variance in similarity. It was also verified that prototypicality was highly correlated with the extent to which a product shared usages in common with all other category members. An incidental finding was that typicality was equally well explained by an index of product versatility, namely its overall appropriateness across a variety of usages. Attribute listing data were also obtained as part of this study, and some interesting differences were found in the cases where the usage-based model predicted similarity better than the attribute-based model, and vice versa. The conclusion was that the two methods addressed partly overlapping dimensions of similarity. Third, it suggests that the effect of the pioneering exemplar, that "sets up' the category, on consumer evaluations of subsequent category entrants may have more to do with the manner in which it affects identifying attributes rather than the manner in which it "molds" preference. Consumers may infer such identifying attributes from their experience with the pioneering brand, and use them as a heuristic to reduce risk when faced with uncertainty in answering questions such as "Is x (brand) a y (product), and thus likely to possess functional attributes f1, f2, f3?" Such an explanation would leave consumer sovereignty intact, and would not imply that the pioneering product derives its advantage out of its ability to shape the consumers utility structure.

The Role of Usage Context: Consider two sample usage contexts in which beverages might be consumed:

(1) "A beverage to drink after returning home from a workout on a hot summer day."

(2) "A beverage to drink while stopping for breakfast at a roadside diner while on a long car trip."

Contrary to one's intuitions, the standard, deterministic theories of category structure do not predict differences in the instances of the beverage category brought to mind in the different contexts. This is because typicality of individual beverages is supposed to be independent of context, and since the recall of instances ("exemplar production") is expected to be a function of typicality, it is also not expected to be altered by context. This issue clearly is of considerable significance in view of its implications for choice set formation. Roth and Shoben (1983) discuss various ways in which context may alter the degree of representativeness of particular objects. Context may lead to highly specific expectations for the referent (e.g., only one product or brand could prove useable), or it may serve to shift the focus to specific brands (e.g., Colgate may be thought to be the more prototypical toothpaste when no context is specified, but Crest may be perceived more prototypical in situations where cavity protection is a major buying motive), or it may reorganize the entire typicality structure based upon constraints imposed by the context (e.g., a topping for ice cream could include not only products specifically designated as toppings, but jams and preserves, certain liqueurs and cordials, pancake syrups, crushed fruit, nuts, and melted candies, etc.). An important implication of such ad hoc categorization is that the appropriate product alternatives can go beyond the members of a nominal product category and even include novel or customer-manufactured alternatives. Indeed, in the preceding example only subsets of other product categories could be appropriate to the new use, e.g., all liqueurs would not be considered as ice cream toppings (e.g., brandy). Both Roth and Shoben (1983) and Smith and Medin (1981) have pointed out that several modifications need to be introduced in the standard categorization theories to accommodate context effects, and that there is a substantial element of ad hocism in these modifications.

From a "contextualist" perspective, the static or deterministic conceptualization of category structure may be flawed because (a) it does not recognize that an object's meaning may be acquired by an individual through experience with its role in events (Nelson, 1974; Bransford and McCarrell, 1974), and (b) it does not suggest the inherent flexibility in the "remembering" of objects as a function of the context or setting (Bransford et al., 1977). The implications of the former were outlined in an earlier section of this paper. As for the latter, Ratneshwar (1987) has emphasized the role of usage situation as a goal context that provides a frame of reference to the consumer when faced with a choice problem. Such a goal context may play a functional role in focusing ("attuning") the consumer's thought processes to the product alternatives appropriate to the particular usage situation, and on the product attributes relevant to situational goals. This in turn suggests that typicality at the "basic" or "common" category level may not be a very good predictor of products or brands brought to mind in specific usage situations. This hypothesis was verified by Ratneshwar (Study II, 1987); it was found that typicality measured at the usage situation- based, ad hoc category level, when compared to typicality measured at the "common" category level, proved to be a much better predictor of the frequency with which products were recalled in particular usage situations (R2 = 0.80 vs. 0.25).

The above discussion has emphasized the role of usage context in terms of its effects on category structure and choice set formation. A similar emphasis on usage context is to be found in consumer-oriented approaches to determining market definition and structure (Day, Shocker and Srivastava, 1979; Srivastava, Alpert and Shocker, 1984; Stefflre, 1971). One other normative implication may be briefly mentioned in this regard. To the extent consumer category structures are dominated by the first, presumably prototypical, market entrant, later entrants may find it more worthwhile to position their products by end-use(s) rather than on the basis of category membership. But the former strategy often implies more limited market potential compared to the latter, and the trade-offs involved in these alternative strategies are certainly worthy of further research.

SUMMARY AND CONCLUSIONS

Rogers and Shoemaker (1971) draw on the anthropological tradition to propose that there are three aspects of an innovation which need to be recognized by change agents:

1. Form, which deals with the directly observable physical appearance and substance of an innovation.

2. Function, which is the potential utility or contribution of the innovation to the adopter.

3. Meaning, which is the inherently contextual significance of the innovation to the potential adopter.

A considerable part of this paper has been devoted to issues which in one way or the other involved contrast of "form" with "meaning". It was pointed out that the standard categorization literature with its Aristotelian leaning (attributes of objects "in themselves") tends to ignore the relational nature of object concepts. There is little recognition of the fact that significances are the results of interactions of people (having particular characteristics) with objects having particular form and function, in particular usage, event and cultural contexts. It appears that this literature has not,really been able to free itself of its links with the tradition that the philosopher Pepper (1942) termed as "formism", even though concepts such as prototypicality have replaced the "classical" view that categories could be defined in terms of "necessary and sufficient" attributes. Pepper (1942) specifically refers to the tradition known as "transcendent formism' as one which stresses similarity of objects (belonging to the same category) to a plan or a norm; Pepper gives examples such as a shoemaker making shoes to the same design/plan, or different oak trees that approximate the same norm or prototype. (Plato stressed the artisan, and Aristotle the natural categories.) Pepper's definition of a norm or prototype as a "center of a rather vague extensity, claiming as exemplifications objects which closely approximate it and making lesser and lesser claims toward the periphery" is retained in the above literature. He points out that formism is identified with the root metaphor of similarity; that it stresses hierarchical categorical structure; and that its favored method of scientific investigation is through descriptive structural representation. All these attributes are once again to be found in the "standard" categorization literature that has emerged since the early 1970's.

The linkages between "form" and "function" are clearly of crucial importance to marketers. Psychologists who work within the ecological tradition emphasize that it must be the case that there is a non-arbitrary relationship between the two (Gibson, 1979). But the distinction made earlier between identifying attributes (which likely emphasize "form") and core meaning (which emphasizes "function") indicates that the linkages may be complex, and possibly also phenomenon--specific. Empirical evidence indicates that there is substantial correlation between form-oriented variables such as typicality, and function-oriented variables such as preference (Barsalou, 1983, 1985; Nedungadi and Hutchinson, 1985; Ratneshwar, 1987), but causal directions remain unclear. Clearly, further research is needed to clarify the relationship between the formist categorization literature, with its emphasis on structural description, and mainstream marketing literature that deals with the prediction of preference-based choice behavior (e.g., through fixed or random utility models). An emphasis on "meaning" is a continuation of the tradition that Pepper (1942) refers to as "contextualism", which is associated with people such as William lames and John Dewey. Pepper points out that the root metaphor of contextualism is the "dramatic event". By this he means that the contextualist tends to be preoccupied with phenomena such as "quality" or 'meaning", which involve the understanding of things in the context of interconnected events or activities with continuously changing patterns. Unlike formist theories, contextualist theories tend to be dynamic and relational. They tend to emphasize pragmatism in human behavior, and response that is "situationally appropriate". (See Gibson, 1979; Shaw and Bransford, 1977; Bransford, et al., 1977 and Nelson, 1974 for contextualist perspectives on perception and cognition.) Contextualism implies overlapping categorical structures, and emphasizes evolutionary change without sharp discontinuities. In the marketing domain, such a perspective implies that what a product means or signifies to a consumer is not a question that can be answered independent of situational context. We have has argued in favor of the value to be obtained from this perspective. This is not to say that there are no insights to be obtained from the use of paradigms based on the standard categorization literature; rather, it is to suggest that such insights need to be balanced against, and related to, alternative perspectives.

Finally, we would note that the inability to control (i.e., consumer sovereignty) suggests that such managerial findings as "first mover advantage" may not always be actionable. Discovery that there is an advantage to being first does not tell one how to become first. The research which documents a first mover advantage is, after all, retrospective; it looks at categories which have already been formed. Advice that suggests it is desirable to become first is prospective; and its actionability depends upon the manufacturer finding ways to suggest new categories and establish exemplar status for its innovations. Because the firm can suggest but cannot compel, its ability to realize these objectives may be poor (as the firm in Los Angeles discovered when it set out to build a better ice cream by increasing its butterfat content only to learn that consumers refused to perceive it as ice cream and saw it more as sweet, flavored butter, or the beer company in Pittsburgh found out when it attempted to add fruit flavoring in an attempt to produce a better tasting beer and learned it had invented a beverage that was no longer perceived to be beer). We speculated that the marketing efforts of the firm might enable it to anticipate and affect such results and that research might usefully be devoted to learning how. If the product offering is too similar to what already exists, it is likely to be assimilated by many to an existing category, if not ignored entirely. But if it is very new and different, the likelihood of a new category resulting may also be low. No matter how novel something is, people will attempt to understand it in terms of what they already know. Inability to comprehend for many could result in no categorization and work against the new product's success. Again research aimed at examining boundary conditions (what is too little and too much "newness") should hold pragmatic value for many marketers. The major advantage of a categorization perspective on marketing and strategic issues is that it provides a new way of looking at some old issues. What is a coherent criterion for defining a product-market? What is "newness" in a product and in a category? What is really "new" about a given new product? When do attempts to improve quality lead to new products rather than higher quality old ones? These are just a few of the issues we have touched upon. Relating them to categorization reveals their connectedness. Our hope is that there will soon be better answers as a consequence of the different perspectives that research into cognitive processing and structures can afford.

REFERENCES

Bartlett, F. C. 1932. Remembering: A Study in Experimental and Social Psychology, Cambridge, England: Cambridge University Press. Barsalou, L. W. 1982. Context-independent and context-dependent information in concepts. Memory & Cognition, 11 (3), 211-227.

Bartlett, F. C. 1985. Ideals, central tendency, and frequency of instantiation as determinants of graded structure. JEP - Learning, Memory and Cognition, 11 (October), 629-654.

Bransford, J. D. & N. S. McCarrell. 1974. A sketch of a cognitive approach to comprehension. In Cognition and the Symbolic Processes, W. Weimer & L. D. Palermo (eds.), Hillsdale, NJ: Erlbaum.

Bransford, J.D., J. J. Franks, N. S. McCarrell, dc K. E. Nitsch 1977. Toward unexplaining memory. In Perceiving, Acting and Knowing, R. Shaw & J. D. Bransford (eds.), Hillsdale, NJ: Erlbaum.

Carpenter, G. & K. Nakamoto 1987. Market pioneering, learning and preference. Paper presented at the 1987 Association for Consumer Research Conference.

Cassirer, E. 1953. Substance and Function and Einstein's Theory of Relativity, New York: Dover Publications.

Crawford, C. M. 1983. New Products Management. Homewood, IL: Irwin. Day, G. S., A. D. Shocker & R. K. Srivastava 1979. Customer-oriented approaches to identifying product-markets. Journal of Marketing, 43 (Fall), 8-19.

Franks, J. J. & J. D. Bransford 1971. Abstraction of visual patterns. Journal of Experimental Psychology, 90, 65-74.

Garner, W. R. 1974. The Processing of Information and Structure, Potomac, MD: Erlbaum.

Gibson, E. J. 1969. Principles of Perceptual Learning and Development. New York: Appleton.

Gibson, J. J. 1979. The Ecological Approach to Visual Perception, New York: Houghton-Miffin.

Lakoff, G. & M. Johnson. 1980. Metaphors We Live By, Chicago: University of Chicago Press.

Marlino, D. L. & M. Sujan 1987. Category expectations and market pioneering. Paper presented at the 1987 Association for Consumer Research Conference.

Miller, G. A. & P. N. Johnson-Laird 1976. Language and Perception, Cambridge, MA: Harvard University Press.

Murphy, G. L. & D. L. Medin 1985. The role of theories in conceptual coherence. Psychological Review, 92, 289-316.

Nedungadi, P. & J. W. Hutchinson 1985. The prototypicality of brands: relationships with brand awareness, preference and usage. Advances in Consumer Research, 12, 498-503.

Nelson, K. 1974. Concept, word, and sentence: Interrelationships in acquisition and development. Psychological Review, 8, 267-285.

Nelson, K. 1985. Making Sense: The Acquisition of Shared Meaning, Orlando: Academic Press.

Pepper, S. 1942. World Hypotheses: A Study in Evidence, Berkeley: UC Press.

Posner, M. I. & S. W. Keele 1968. On the genesis of abstract ideas. Journal of Experimental Psychology, 77, 353-363.

Ratneshwar, S. 1987. Investigations of the Role of Usage Context in Consumer Behavior. Unpublished Ph.D. dissertation.

Robison, W. T. & C. Fornell 1985. Sources of market pioneer advantages in consumer goods industries. Journal of Marketing Research, 22, 305-317.

Rogers, E. M. & F. F. Shoemaker 1971. Communication of Innovations: A Cross-Cultural Approach. New York: Free Press.

Rosch, E. 1978. Principles of categorization. In Cognition and Categorization, E. Rosch &c B. B. Lloyd (eds.), Hillsdale, NJ: Erlbaum.

Rosch, E. & C. B. Mervis 1975. Family resemblances: Studies in the internal structure of categories. Cognitive Psychology, 8, 382-439.

Roth, E. M. & E. J. Shoben 1983. The effect of context on the structure of categories. Cognitive Psychology, 15, 346-378.

Shaw, R. & J. D. Bransford 1977. Introduction: Psychological approaches to the problem of knowledge. In Perceiving, Acting and Knowing, R. Shaw & J. D. Bransford (eds.), Hillsdale, NJ: Erlbaum.

Smith, E. E. & D. L. Medin 1981. Categories and Concepts, Cambridge, Mass: Harvard University Press.

Srivastava, R. K., M. I. Alpert & A. D. Shocker 1984. A customer-oriented approach for determining market structures. Journal of Marketing, 48 (Spring), 32-45.

Stefflre, V. J. 1971. New Products and New Enterprises: A Report of an Experiment in Applied Social Science, Irvine, CA:- University of California.

Tversky, A. 1977. Features of similarity. Psychological Review, 84, 327-352.

Urban, G. L., T. Carter, S. Gaskin, & Z. Mucha 1986. Market share rewards to pioneering brands: An empirical analysis and strategic implications. Management Science, 32, 645-659.

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