The Role of Usage Context in Consumer Choice: a Problem Solving Perspective

Luk Warlop, University of Florida
S. Ratneshwar, University of Florida
[ to cite ]:
Luk Warlop and S. Ratneshwar (1993) ,"The Role of Usage Context in Consumer Choice: a Problem Solving Perspective", in NA - Advances in Consumer Research Volume 20, eds. Leigh McAlister and Michael L. Rothschild, Provo, UT : Association for Consumer Research, Pages: 377-382.

Advances in Consumer Research Volume 20, 1993      Pages 377-382

THE ROLE OF USAGE CONTEXT IN CONSUMER CHOICE: A PROBLEM SOLVING PERSPECTIVE

Luk Warlop, University of Florida

S. Ratneshwar, University of Florida

[The authors thank Cynthia Huffman for helpful comments on an earlier draft of this paper.]

A considerable amount of prior research has documented that usage situations or contexts play a major explanatory role in consumer preference/choice (eg., Stefflre 1971; Belk 1975; Dickson 1982; Miller & Ginter 1979; Srivastava, Alpert, & Shocker 1984). Much of this literature has tended to take a "black box" approach with respect to situational effects. There is little understanding at this point of the role of usage context in the cognitive processes that are involved in consumer decisions (Ratneshwar & Shocker, 1991). The present research seeks to contribute to this area by examining usage context in terms of its role in consumer problem solving.

We start from the premise that the usage context of a product is an ecological factor that can aid consumer decision-making by imposing, in cognitive terms, constraints on the problem-solving process and therefore also on the possible solutions (Bransford & McCarrell, 1974; Ratneshwar & Shocker 1991). Based on this premise, we suggest that usage context can have two kinds of effects on the choice process. First, it may directly facilitate remembering of "known solutions" to the choice problem by directing retrieval processes toward context-appropriate product alternatives. We shall later offer a more fine-grained description of this process. Second, situational problem definition might frame the construction of a problem representation in terms of the product benefits that should be sought in the usage context and the constraints the situation imposes on a solution set.

Both of these processes may well go on in parallel in many situations. But it is probably reasonable to assume that the constructive process becomes prominent only when on account of lack of familiarity with the situation there are no "known solutions" available in memory. This process may also be invoked when none of the physically available alternatives match the retrieved set. The present paper focuses on the cognitive processes that are involved in both of these routes to finding solutions to contextually-situated choice problems, and how these processes might be moderated by situational familiarity. We report two exploratory studies. The first one utilizes verbal protocols to investigate the influence of usage context on memory-based processes in the formation of consideration sets. The second one examines how usage context might affect solution recognition in stimulus-based choice.

STUDY 1

Our efforts in study 1 were mainly directed at arriving at a better understanding of the role of usage context in the problem-solving process with regard to the generation of potential solutions from memory. Our approach to this was a discovery-oriented, verbal protocol analysis study. 13 undergraduate student subjects were each presented with different product usage situations printed on cards and asked to come up with acceptable alternatives in those situations, while verbalizing their thought process as it led to such a set of alternatives. Situations were manipulated to be either relatively familiar to the subjects (eg., "A beverage that you might drink while relaxing at a swimming pool with friends during summer vacation") or relatively unfamiliar (eg., "A beverage that you might carry in your backpack and drink while hiking all day on a steep trail in the Grand Canyon"). The target product categories were beverages and snacks; they were selected on the basis of prior research that has demonstrated usage situation influences in these categories (see Belk 1975). There were four possible situations for each category, two familiar and two unfamiliar. The subjects saw one situation in each product category, and saw either all familiar or all unfamiliar situations, to reduce the likelihood of demand awareness. The target situations were separated by filler situations, that were not included in the analysis.

The use of protocol analysis to study decision processes is based on the assumption that during decision making the information available for use, which is the output of lower order retrieval and recognition processes, is present in working memory. We conducted the study in accordance with Ericsson & Simon's (1984) guidelines to maximize the reliability of the protocols. The study was introduced as one in which we were interested in "how people like yourself relate to different consumption situations." Each individual's session commenced with a training/familiarization phase with three trial situations during which the experimenter first emphasized the need to verbalize everything that "passes through your head while you think about this situation" and then provided corrective feedback. Sessions were audio taped and later transcribed for analysis.

Considering the limited number of subjects involved in the study, we regard this as an exploratory investigation. We used the protocol analysis in an inductive manner; the examples given should be regarded as mere illustrations, and the ideas we develop remain to be tested in a more systematic study. Our discussion of study 1 is organized so as to highlight our distinction between the direct retrieval of known solutions (when the usage context is familiar) and the more constructive processes used to generate potential solutions (when the usage context is unfamiliar).

Familiar Situations and the Direct Retrieval of Known Solutions

In accordance with our initial expectations, the protocols obtained from subjects presented with familiar situations were characterized by little or no evidence of constructive thought processes: virtually no mention of product attributes or benefits, and hardly any mention of the goals and constraints implied by the situation. Instead, these protocols suggested three qualitatively different processes by which consumers might directly retrieve known solutions when the usage context is familiar: (1) Retrieval based on contextually-organized product category structures, (2) Retrieval based on event scripts cued by the situation, and (3) Retrieval based on episodic memory of similar situations.

(1) Although subjects were encouraged throughout the training period to verbalize everything that came to mind, in many cases the resulting protocol was hardly more than a list of context-appropriate alternatives. Consider, for example, this subject's response when given the situation "A snack that you might eat when you are at a movie theater with a friend":

(M. M.): "Popcorn, M & M's, raisin, raisin nuts, raisin nut things, a chocolate candy bar, on of them sour tart candy type deals, I want to pick up a soda pop or a drink, something like that, licorice, that's about it."

The same subject produced the following when presented with the "swimming pool" situation described earlier:

(M.M.): "Beer..some kind of beer, light beer, soda or ice water it doesn't matter and that's about it, that is all I would drink, ice water, soda or beer, light beer."

These protocols suggest that the subject is directly accessing instances of categories organized around familiar usage contexts. Research on the graded structure of usage context-defined categories (Ratneshwar & Shocker, 1991) has shown that differential accessibility of brands and products in a given usage situation, as measured by the order of recall, is positively (and strongly) associated with typicality judgements of the same products in that situation. These results suggest that for familiar (i.e., well-established) usage contexts, the structure and organization of the stored knowledge base of potential solutions may come to resemble the graded structure of a taxonomic category (see also Barsalou 1991). The mental representation of the encountered usage situation could then serve as a retrieval cue that selectively accesses this contextually-defined, categorical knowledge base. Consequently, it cues retrieval of appropriate solutions in the order of their accessibility in the relevant category.

(2) In some other cases, the retrieval processes resembled the mental running-off of an event script. Such script recalls are characterized by (a) frequent use of verbs (e.g. "go", "get", "do") denoting activities that are typical of the event script, and (b) mentioning of "irrelevant" objects that are thematically related to the script. For example, the movie-going protocol discussed earlier mentioned "pick up a soda pop" with the list of snacks. But in the next example the evocation of an event script comes through much more clearly. The situation was a "A snack that you might eat between classes on a busy weekday when you don't have time for a regular lunch."

(H. U.): "... I just like to go to the vending machine and I get a peanut butter cracker and a cup of coffee, that is what I do... When I am in this area, like the business buildings here, I may just go across the street and go to Burger King and just get a cheese burger or whatever..."

This subject appeared to rely on an abstract or "scripted" memory of a familiar life event rather than a specific past occurrence of a similar event. Research by Nelson and her colleagues has documented the importance of scripts in the development of a conceptual knowledge base (see, e.g., Nelson 1988). Nelson suggested a close relationship between a child's comprehension of the functional role of objects in different life events and the individual's formation of category concepts. A child's repeated experiences with similar events (eg., different occasions of "breakfast") lead to a generalized event representation or script in which different objects become slot fillers. For example, both "pancakes" and "cereal" become slot fillers for the slot "what I eat" in the breakfast script. Other developmental psychologists also suggest that situational or "thematic" categorization forms the basic organizational principle of young children's knowledge (Markman & Callanan, 1984).

(3) Both of the processes described earlier involved the use of abstract knowledge representations based on past experience to access known solutions. However, our protocols revealed that directCthat is, not mediated by controlled search activityCretrieval of alternatives could also be based on specific episodes of similar situations in an individual's past. For example, given the situation "A snack that you might eat between classes on a busy weekday when you don't have time for a regular lunch":

(D. A.): "I am starting to envision the little stands that go through campus, the guy that parks over here in the business quad. Running over to the library to grab a banana from him, especially when I have a hard time staying awake...I am grabbing a banana from him and a diet Mountain Dew from the soda machines right next door and then run back to class..."

Unfamiliar Situations and the Construction of Potential Solutions

In the previous section we discussed different cognitive processes by which consumers, when given a familiar situation, might directly retrieve solutions appropriate to the problem defined by the usage context. When the situation is unfamiliar, however, the decision-maker may have to rely on more strategic processes to construct possible solutions to the choice problem. Barsalou (1991) proposes that the construction of ad hoc and goal-derived categories proceeds through an active, top-down, and relatively effortful process in which existing knowledge about object attributes is manipulated in working memoryCa process he refers to as conceptual combination. Many choices among products in relatively unfamiliar usage situations can be construed as mediated by the formation of an ad hoc category. When faced with a situation such as "a beverage to carry on a hiking trip on a steep trail in the Grand Canyon during summer", people are very unlikely to have stored in memory a well-established category with potential solutions that corresponds to the situation. Neither are they likely to have individualized (i.e., episodic) or scripted event memories for that particular event. On the other hand, none of the elements of the situation nor any of the beverages that might be potential candidates are as such unfamiliar. Most people should be able, based on their general "world knowledge", to derive the benefits a beverage in those circumstances should provide (eg., fluid replenishment and energy) and the constraints that are imposed on the solution by the usage context (eg., shouldn't spoil easily in the heat, a light-weight and sturdy container.) Based on the benefits and constraints that are relevant to the situation, the product attributes and attribute values that are most desirable in the particular situation (cf. Barsalou's 1991, "ideals"; also see Bransford & McCarrell 1974) can be derived, and the resulting ideal representation compared to the available alternatives.

We tentatively propose that the retrieval of alternatives in unfamiliar usage situations occurs in three stages. First, the consumer forms a reduced cognitive representation of the "total" consumption situation. This is followed by a recognition of benefits and constraints involved in the elements of that representation. Then, by a process of association, attributes and alternatives are retrieved that fit these benefits and constraints. The more of the diagnostic situational information that is implicated in the generation process, the less likely it is that non-appropriate alternatives are retrieved.

We coded the idea units in the protocols as either referring to goals (constraints or benefits), lower-level attributes, alternatives, or exclusions (an alternative or class explicitly excluded). Compared to the familiar situations, where hardly any goals were mentioned, they were abundant for the unfamiliar situations, and mostly they preceded the mentioning of alternatives. Also, often subjects point out which class of alternatives they would definitely not consider, before mentioning the alternatives they find appropriate. The following protocol was typical:

Situation: "A snack that you might eat at home shortly before going out on a Saturday evening date."

(M. S.): "I want something that is real quick, and that won't give you bad breadth or anything. You don't want to eat a sandwich or something to that effect. A piece of fruit or something, something to tide me over till dinner. It will not bother your stomach or anything like that. And nothing that is messy, in case you are all dressed up. You wouldn't want to spill on your clothes ... I'd like an apple or something, or some grapes, or some crackers, and you wash it down with something like water. That's about it."

This subject clearly constructed a solution to the problem, based on the constraints and goals that are relevant to the situation. The appropriate alternatives were found through a process of conceptual combination and were clearly not accessible immediately. In this example the retrieval of situation-relevant information and the generation of alternatives were more or less separated, and problem-solving proceeds in a straightforward, linear fashion. In other cases, subjects went back and forth between goals and related alternatives, ultimately to come back to a few alternatives that seem to satisfy all criteria. In other words, subjects seemed to gradually "zoom in" on a solution. For example:

Situation: "A snack that you might eat in the evening shortly after a workout or an aerobics class when you know you will be meeting some friends for dinner in a couple of hours"

(A. F.): "Something like TCBY yoghurt, because it is not that heavy. Fruit, ... it is like an appetizer ... like a watermelon, an orange or something like that, and I would stay away from meats, because they seem to stay on you, but fruit definitely, or some yoghurt. Nothing fattening, like ice cream, or anything like that, because it tends to stay with you and spoil your appetite. Something that can carry me over, even a Snickers bar, something that will hold me over, you know, Snickers satisfies the hunger in you..., something with less preparation, that I can go buy quick. No burritos or sandwiches, maybe some crackers and cheese, something not as solid, not a solid substance, that is easily digestible and that is quick. Maybe fruit, because that has that quick energy and it gets you OK. Something with a lot of potassium or carotene, even a few slices of bread or something.

An interesting observation was that some of the problem solving behavior was reminiscent of the reliance on surface characteristics by novices in technological domains. For example, when reacting to the situation "A snack that you might eat shortly after a workout or an aerobics class when you know you will be meeting some friends for dinner in a couple of hours", quite a number of subjects emphasized that the snack should be "healthy", "no fastfood" or provide "quick energy". They seemed to rely more on the "workout" element and its health-related associations than on the (normatively) more diagnostic cue that they would have to eat dinner in a couple of hours. Admittedly, the examples we have discussed are the clearest and the most demonstrative in the whole set. Many protocols were less clear cut. For example, for some undoubtedly unfamiliar situations some subjects came up with alternatives immediately. Speculatively, what seems to have happened there is that subjects disregarded much of the situational information and concentrated on a few elements that were most salient, actively reducing the situational representation to something that is familiar.

In summary, one basic pattern clearly emerged from the study 1 data. In familiar usage contexts, subjects in almost all cases evoked a series of (situationally appropriate) alternatives without, or before, mentioning any attribute or goal-related information. In unfamiliar situations the subjects' behavior was more variable. Subjects occasionally seemed to be able to come up with alternatives directly when they were able to reduce the complexity of the situation by ignoring detail; otherwise, they relied on constructive problem solving.

STUDY 2

Our discussion so far has focused on the cognitive processes by which usage context impacts on the alternatives evoked from memory. This issue is of considerable significance since it affects the manner in which consideration sets are formed by consumers. But equally important to consumer problem-solving are the processes by which usage context affects solution recognition: the ability to "pick out" (discriminate) among currently available choice alternatives.

Consistent with our previous suggestions, situational familiarity is likely to play an important role in governing the processes by which the consumer discriminates among available alternatives. Consider, first, the case where the situation is highly familiar, as indexed by the number of previous encounters with "similar" situations. As we discussed, "known solutions" to the choice problem (in the form of context-appropriate product alternatives) should be retrieved from long-term memory in a more or less automatic fashion. The consumer is then likely to compare this set of alternatives retrieved into working memory with the alternatives currently available in order to detect "acceptable" solutions.

In contrast, when the features of the usage context are relatively unfamiliar, the consumer might perceive this unfamiliarity and not bother to search long-term memory for an appropriate analogue (category, script, or episode) to the current situation. Alternatively, the search might take place but fail to reveal "known solutions." In such novel situations, the goals and constraints imposed by context might still facilitate discrimination, but by a qualitatively different process. These problem-defining goals and constraints might have an orienting effect (William James's "point of view") wherein the consumer's attention is attuned to a small number of contextually-relevant product attributes on the basis of which the available choice alternatives might be evaluated. Thus, context might shape the attentional gate which regulates the information that is processed with regard to the alternatives in the environment. But to reiterate the point made earlier, this process may come to the fore only when the novelty of the situation prevents the spontaneous retrieval of known solutions, or when none of the retrieved solutions are available to the consumer.

In the preceding paragraphs we have suggested that usage context plays a key role in consumer problem-solving by impacting the discriminability of choice alternatives and that the implicated processes might vary with situational familiarity. Since higher levels of discriminability between alternatives generally should result in faster decisions, an important consequence for consumers might be in terms of decision-making time. Consumers make numerous "low involvement" decisions during the course of a day, and often these decisions are made from potentially very large numbers of options open to them. To the extent that many of these decisions take place in well-defined situations or contexts, contextual goals and constraints might help the consumer to discriminate acceptable alternatives from a much larger available set. Consequently, consumers might be able to able to make these constrained decisions quickly and with relatively low cognitive effort. For example, consider a usage context wherein a consumer has to make a choice among beverages upon returning home from a work-out on a hot summer day. The context might facilitate swift solution-recognition in terms of suitable alternatives such as Gatorade and also facilitate "disconfirmation" of contextually inappropriate options such as coffee. Even when the situation is relatively less familiar or "scripted" and the decision-maker has to take a more constructive approach to evaluating the alternatives, our earlier discussion suggests that situational constraints might still facilitate discrimination (and quick decisions) by focusing the consumer's attention on context-relevant product features.

We investigated some of the aforementioned speculations in an exploratory study in the category snack foods for which normative data on product preferences in various usage contexts were available through prior research (see Ratneshwar & Shocker 1991). The experimental task required our undergraduate subjects (n = 50) to make choices from pairs of product alternatives whose names were displayed on a PC screen while the computer recorded decision-making time (DMT). All subjects made choices from 12 different choice pairs on 12 different "trials." But on each trial the display of the stimulus choice pair on the screen was preceded by a description of the choice condition for that trial: either a specific usage context or merely the category name (the latter was the control condition). Note that the time taken to read this description was not included in DMT. Six usage contexts that varied in familiarity were selected and the design was balanced so that the twelve choices made by a subject were equally divided between category-level and context-level choices. Further, the stimuli were designed so that the usage context either enhanced discriminability (ED) for the particular choice pair or it did not (ND). For the ED choice pairs, the prior data on preferences revealed that one of the two products was appropriate for the particular usage context (mean = 84%), while the other was inappropriate (mean appropriateness = 21% of subjects). For the ND choice pairs, both products had been judged appropriate (mean = 73%), with little difference in contextual appropriateness between the products in a pair (mean difference = 5.5%). In order to familiarize them with the task, subjects were first guided through six practice trials where choices were made in categories unrelated to snack foods. Subjects then made choices on 36 trials out of which every third trial related to snack foods, while the remaining filler trials dealt with unrelated categories and situations. After completing the choice task on the computer, subjects filled out questionnaires with measures designed to check the experimental manipulations and also to assess the familiarity of the usage contexts.

We report here some of the key results. The grand mean DMT was 2.53 seconds, and it ranged from 0.75 to 9.29 seconds. Across the 50 x 12 = 600 observations, as expected, a significant negative correlation (r=- .21, p .01) was found between DMT and stimulus discriminability as indexed by the absolute value of the difference in preference scores for the two alternatives. An ANOVA was conducted on DMT after carrying out a logarithmic transformation in order to stabilize variance. The predicted interaction between choice pair type (ED vs. ND) and choice condition (usage context vs. category) was significant, F (1,576)=10.03, p<0.01. However, this was qualified by a significant three-way interaction of usage context block x choice pair type x choice condition (p<0.01) suggesting that the hypothesized interaction was not obtained uniformly across all usage contexts.

As anticipated, subjects made significantly faster choices at the usage context level (vs. category level) for ED choice pairs (2.37 vs. 2.77 seconds, p<0.05). The opposite was true of ND pairs: Subjects were significantly slower in their choices when provided with the usage context than at the category level (2.64 vs. 2.32 seconds, p<0.05). Thus, for example, when provided with a snack usage context that depicted "a Friday evening party while drinking beer or other beverages," subjects made faster choices (2.58 vs. 3.55 seconds at the category level) between the ED pair "mixed nuts" and "oatmeal soft cookie"; not surprisingly, they also displayed more consistency in their choice, with 92% picking the former alternative at the context level vs. 44% at the category level. In contrast, with the same usage context, subjects made slower choices between the ND pair "crackers" and "hot dog," presumably because the context did not enhance their discriminability (2.91 vs. 2.21 seconds at the category level).

Interestingly, and consistent with the earlier theorizing, the study revealed evidence of faster decisions even in unfamiliar usage contexts. For example, DMT was lower for the ED pair "grapes" and "cereal" when subjects were provided with the usage context of a snack "you might eat while waiting for your Saturday evening date to show up." This context had the lowest familiarity rating of the six that we used, yet subjects were significantly quicker at picking the preferred alternative (grapes) given the contextual description (2.49 vs. 3.65 seconds at the category level). However, when provided the same usage context with the ND pair "raisins" and "potato chips," subjects were significantly slower in their decisions (3.29 vs. 1.77 seconds at the category level).

While this study was only exploratory and more research is needed before any definitive conclusions can be reached, the results were encouraging in terms of the present perspective. The cognitive constraints resulting from usage situation or context can play a problem-definition role such that consumers discriminate effectively and efficiently among available alternatives. Further, discrimination may be facilitated even in relatively unfamiliar situations, which we attribute to the attentional constraints imposed by usage context.

DISCUSSION

Problem-solving research in consumer behavior has been closely modeled after issues in the general problem-solving literature where processing differences due to differential domain expertise have often been the central issue (e.g., Sujan, 1985). We deal with a conceptually different issue: one wherein consumers are likely to possess the "technical" knowledge of a product category (e.g., beverages or snack foods) necessary to solve the problem (the choice of a product), but where the "problem" essentially resides in the use of that knowledge base to arrive at appropriate solutions in the context of familiar and less familiar usage situations.

At its current stage our research has necessarilly been exploratory in nature; but the findings of the two studies, one dealing with the retrieval of alternatives from memory and the other with stimulus-based choice, suggest some tentative conclusions. First, we found support for the idea that the role of usage context in consumer choice is one of constraining the problem-solving process, and, as a consequence, one of guiding the search for, and evaluation of, potential solutions. Second, we found some tentative support for the hypothesis that the actual processes by which solutions are found are qualitatively different in familiar and nonfamiliar situations.

In familiar situations, the cognitive representation of the situation seems to cue a set of potential solutions directly, that is, not mediated by a controlled search of memory. The protocol data from study 1 suggest that such direct retrieval may be based on category structures and/or event scripts that are well-established in memory as well as specific past episodes that are analagous to the current situation. The retrieved set is then likely to be compared to the the alternatives available in the environment to detect "acceptable" or satisfactory solutions.

When the situation is relatively unfamiliar, alternatives (potential solutions) are retrieved from memory through more constructive processes that involve cognitively mapping the situation on to contextually-relevant goals and constraints in terms of the benefits desired of the product, and then retrieving products that are associated with those benefits. Further, such contextual attunement likely guides the decision-maker's attention to relevant product attributes in order to discriminate between the available alternatives. In general, the particular features that get the decision-maker's attention are likely to be those that have relevance for the goal context of the ongoing situation and/or those that have high diagnostic value for the local context of alternatives (see Garner 1974; Tversky 1977). For example, in an unfamiliar situation such as "A snack that you might eat while waiting for your Saturday evening date to show up", discrimination between the two alternatives, potato chips and "Doritos" tortilla chips (a garlic-flavored product), might be enhanced by focusing on whether a product affords "good breath". Suppose the choice set had also included a third alternative, grapes. It is possible that doing so might enhace the contrast value of the garlic-flavor feature and thus further expedite the rejection of Doritos as an acceptable option.

The problem-solving perspective we employ is related to the means-end chain tradition (see, e.g., Guttman 1982; Olson 1988), but in line with the more traditional expectancy-value approaches to consumer motivation (see Cohen & Warlop 1992), we limit the scope of the means-end chain to goals and benefits consumers take into account when thinking about products, rather than to the terminal "values" from which they (in part) may have arisen. Aditionally, our emphasis on the role of usage context in solution recognition contrasts with the context-free approach that seems to be taken by most means-end applications (but see Olson 1988).

Our exploratory studies served to discover some interesting questions and speculations rather than decisive answers. For example, recent research by Hutchinson, Mantrala, and Raman (1992) suggested that consumers are able to screen out exemplars of nontarget taxonomic categories during retrieval from memory, but that typical category brands not in the situation-specific subcategory could not be screened out. Our results on the other hand indicate that situation-based selectivity (e.g., on the basis of event scripts) during retrieval is possible. When, and how, consumers are able to selectively access situationally appropriate alternatives are important issues, because to the extent that they fail to do so, it may be possible for marketers to influence consideration sets by creating top-of-mind awareness through advertising (see also Ratneshwar and Shocker 1991).

A second interesting research question is to identify the conditions under which the proposed attentional gating mechanism fails. Our framework assumes that consumers faced with a product choice "in-context" form a mental representation of the situation. In forming such a problem representation of a less familiar usage situation, however, consumers may be subject to "functional fixedness" (Duncker 1945), ignoring diagnostic situational detail, and relying on a superficially similar, but familiar analog. In that case they might bypass the solution construction process, and retrieve potentially nonoptimal "known solutions".

In conclusion, it might be remarked that the present contribution is part of an emerging stream of research that seeks to emphasize the role of context and goals in consumers' learning, their representation and organization of knowledge, and their use of that knowledge in decision-making (see, e.g., Huffman & Houston 1993; Park & Smith 1989; Ratneshwar & Shocker 1991). Traditional information-processing research in consumer choice behavior has typically contented itself with stimulus and subject task manipulations (see, e.g., Bettman, Payne, & Johnson 1991). The "new look" (cf. Bruner 1957), however, is firmly based on the notion that all consumer behavior ultimately is purposive, and to understand in cognitive terms such behavior, it is important to understand the ongoing context in which it is grounded.

REFERENCES

Barsalou, L. W. (1991), "Deriving Categories to Achieve Goals," in The Psychology of Learning and Motivation, Vol. 27, ed. G. H. Bower, pp 1 - 64.

Belk, R. W. (1975), "Situational Variables and Consumer Behavior," Journal of Consumer Research, 2 (December), 157-164.

Bettman, J. R., J. W. Payne, and E. J. Johnson (1991), "Consumer Decision Making," in Handbook of Consumer Behavior, ed. T. S. Robertson and H. H. Kassarjian, Englewood Cliffs, NJ: Prentice Hall, 50 - 84.

Bransford, J. D. and N. S. McCarrell (1974), "A Sketch of a Cognitive Approach to Comprehension: Some Thoughts about Understanding what it Means to Comprehend," in Cognition and the Symbolic Processes, eds. W. Weimer and D. S. Palermo, Hillsdale, NJ.: Erlbaum, 189 - 229.

Bruner, J. S. (1957, "On Perceptual Readiness," Psychological Review, 64, 123-157.

Cohen, J. B. and L. Warlop (1992), "Motivational and Cognitive Perspectives on Means-End Chains," Working Paper, University of Florida.

Dickson, P. R. (1982), "Person-Situation: Segmentation's Missing Link", Journal of Marketing, 46 (Fall), 56-64.

Duncker, K. (1945),"On Problem Solving," Psychological Monographs, 58 (No. 270).

Ericsson, K. A. and H. A. Simon (1984), Protocol Analysis: Verbal Reports as Data, Cambridge MA, MIT Press.

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

Gutman, J. (1982), "A Means-End Chain Model Based on Consumer Categorization Processes," Journal of Marketing, 46 (Spring), 60-72.

Huffman, C. and M. J. Houston (1993), "Goal-Oriented Experiences and the Development of Knowledge," Journal of Consumer Research, forthcoming.

Hutchinson, J. W., M. K. Mantrala, and K. Raman (1992), "Finding Choice Alternatives in Memory: A Markov Model of Brand Name Recall," Working Paper, University of Florida.

Markman, E. M. and M. A. Callanan (1984), "An Analysis of Hierarchical Classification," in Advances in the Psychology of Human Intelligence, ed. R. J. Sternberg, 325 - 366.

Miller, K. E. and J. L. Ginter (1979), "An Investigation of Situational Variation on Brand Choice Behavior and Attitude," Journal of Marketing Research, 16 (February), 111-123.

Nelson, K. (1988), "Where Do Taxonomic Categories Come From ?," Human Development, 31, 3-10.

Olson, J. C. (1988), "Theoretical Foundations of Means-end Chains," Working Paper, Penn State University.

Park, C. W., and D. C. Smith (1989), "Product-level Choice: a Bottom-up or a Top-down Process?," Journal of Consumer Research, 16 (December), 289 - 299.

Ratneshwar, S. and A. D. Shocker (1991), "Substitution in Use and the Role of Usage Context in Product Category Structures," Journal of Marketing Research, 28 (August), 281 - 295.

Srivastava, R. K., M. I. Alpert, and A. D. Shocker (1984), "A Consumer-Oriented Approach for Determining Market Structures," Journal of Marketing, 48 (Spring), 32-45.

Stefflre, V. (1971), New Products and Enterprises: A Report of an Experiment in Applied Social Science, Irvine, CA: Unversity of California, Irvine

Sujan, M. (1985), "Consumer Knowledge: Effects of Evaluation Strategies Mediating Consumer Judgements," Journal of Consumer Research, 12 (June), 31 - 46.

Tversky, A. (1977), "Features of Similarity," Psychological Review, 84, 327 - 352.

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