The Package Appearance in Choice


Lawrence L. Garber, Jr. (1995) ,"The Package Appearance in Choice", in NA - Advances in Consumer Research Volume 22, eds. Frank R. Kardes and Mita Sujan, Provo, UT : Association for Consumer Research, Pages: 653-660.

Advances in Consumer Research Volume 22, 1995      Page 653-660


Lawrence L. Garber, Jr., University of North Carolina at Chapel Hill

[The author gratefully acknowledges the invaluable guidance of Morgan Jones, Ray Burke, Joel Huber, Charlotte Mason, Bill Perreault and Jim Bettman through every stage of this dissertation research; the helpful comments of Rick Starr, Rick Staelin, Don Lehmann and two anonymous MSI reviewers, Darius Sabavala, Paul Bloom, Gary Armstrong, George Milne, Sharon Hodge and Tom Boyd; and the financial assistance of the Marketing Science Institute.]

Though we all know and accept the notion that the visual and graphical aspects of the product or service have an impact on consumer choice at the point of purchase, little is known or understood about how it has its effect. This research proposes a theoretical framework to account for this effect, and some empirical methods for its test. Viewing consumer decision-making as based upon a nested set of alternatives, this paper proposes a staged model of choice that explicitly considers the role of visual perception and package appearance in formation of the consideration set at the point of purchase. More specifically, the formation of a visually oriented preattention/attention set is conceived to precede and affect formation of a subsequent product-benefits-oriented consideration set.

The empirical test of this model incorporates experiments in which subjects undertake a series of six computer-simulated shopping trips across four supermarket categories: flour, raisins, spaghetti and dry cereal. The computer simulation employed is a virtual-reality supermarket representation developed and validated by Professor Raymond Burke. By this simulation, the subject is able to browse a set of supermarket facings, pull brands of interest from the shelf, turn and read the packages, and select them for purchase. Manipulations consist of the alteration of a certain few of the of the existing packages in the display. These changes are calibrated through pretesting to represent varying degrees of typicality, novelty and appropriateness for the category, while maintaining prior levels of brand identifiability. It is hypothesized that consumers will more likely attend to those new packages which are sufficiently novel or typical, and are more likely to consider those novel brands that are appropriate to the category.


George hurries into the supermarket needing a couple of items, a bottle of cola and and some dishwashing detergent. He always drinks the same cola and, knowing his mind on this product, goes to the soft drink aisle first. George quickly scans the facings, catches sight of the characteristic red color and the distinctive wavy design that identify his preferred brand, and grabs it on sight before rushing to the detergent aisle.

By contrast, George is uncertain about which dishwashing detergent to buy. Unconcerned about small differences in price and believing that any brand will "do the job", he turns into the detergent aisle intending to both see what is available and make a quick purchase decision.

George has (varying degrees of) prior knowledge about some but not all of the relatively large number of brands available in the detergent category, though no particular preferences beyond the fact that he usually buys a lemon detergent. He is already scanning the category facings as he turns the corner and heads to the yellow section of the aisle, knowing that this color typically denotes the lemon detergents.

As George searches out and attends to the brands that he remembers from past purchase occasions or from television advertising, and from which he intends to select his purchase, there are a few others that catch his eye. Some unknown brands draw his attention because they appear similar to his remembered brands; others because their appearance is so very different.

Those brands similar in appearance have bottle shapes and liquid contents virtually identical in form, consistency and yellow color. George wonders if their like appearance means that he can expect like performance.

George is also distracted by a few brands made distinctive by an appearance that departs from George's sense of what is typical for the category. There is one brand, for instance, whose bottle is opaque rather than transparent, and whose pastel yellow surface looks soft relative to the sharp, translucent yellow liquids seen through the bottles of its competitor brands. George wonders if this brand might not be a little milder and questions its cleaning ability. Another brand catches George's eye because of its distinctive barbell shape, a bulky cylinder with larger drum shapes at either end; George in this case wonders if this brand is perhaps more powerful, but thinks the bottle is ugly. A third brand stands out to George because its liquid is a contrasting yellow-green in color, and he notes that it is a lemon-lime detergent.

George is in a quandary because, with the unanticipated addition of several relatively unknown brands to his consideration set, he is no longer certain how he should decide between the brands that he is considering. Now caught up in his detergent choice but still feeling the need to hurry, George finally concludes that he is in the mood to try something different and picks the lemon-lime for its attractive and distinctive appearance, having also decided that he expects it will do a good job. George tosses the bottle into his basket and heads to the checkout.


This paper is about how consideration sets are formed, and how the appearance of brands in their packages plays a role in their formation at the point of purchase. There has been very little in the marketing literature on either topic.

Though there has been considerable evidence offered of the existence of staged models of choice (for a review, see Shocker et al. 1991), there has been less written concerning how these sets are formed at each stage. As to package appearance, it is generally accepted that the visual aspects of a product or its package have an important effect on consumer choice at the point of purchase. However, there is virtually nothing in the academic literature on the topic. I will argue in the following that appearance affects choice indirectly through its influence on which brands, and how many, enter the individual consumer's consideration set.

The purpose of this dissertation is to propose and test a model of choice that explicitly considers the role that visual perception plays in formation of the consideration set. A key feature of this model is the formation of a prior set, conceived as visual in nature, in which the consideration set is nested, and from which all the members of the consideration set are drawn. It will be called the attention set.


George's two purchases in the above scenario illustrate plausible ways in which brand or package appearance may influence choice. But in only one instance, the detergent choice, does George make a considered decision. It is the considered decision that is the subject of this paper.

In the cola example, George is predisposed to his preferred brand and grabs it on sight, without reference to other brands in the category. Package appearance in this instance serves primarily to expedite search and identification. In the contrasting detergent example, in which George scans the category facings and attends to several previously unknown or unnoticed brands that catch his eye, package appearance may also serve to alter or interrupt search (Bettman 1979), and in so doing plays a central role in (re)shaping the choice process itself. In this case, brand packages as visual stimuli play a role in differentiating between brands considered for purchase.

And yet, as important as package appearance is, there is no comprehensive theory available to account for its influence on consideration at the point of purchase. A reason for this may be the traditional tendency to treat the consumer decision process as rational, in which the decision is based upon deriving some optimal combination of performance attributes and levels. Implicit in a multiattributed approach to choice modeling, in which products are defined as some bundled combination of attributes and levels, is that consumers can only differentiate brands on the basis of these features. But what if for some reason the consumer cannot differentiate brands on the basis of performance attributes? Would all the brands then fall into the center of this consumer's perceptual map? Would he or she be unable to make a choice decision? In reality, we know that the consumer is more flexible than this. If he or she is unable to differentiate brands on the basis of one set of criteria, then the consumer can simply differentiate brands on some other basis. Our conclusion, then, is that choice decisions, given imperfect brand knowledge and brand sets which may be at parity on important dimensions, are context dependent, and many kinds of criteria may be applied, including information supplied at the point of purchase such as the text and visuals available on a brand's package, in order to differentiate brands and come to a purchase decision.

The purpose of this paper is to propose a theoretical framework that accounts for these effects, and to test three hypotheses on fundamental aspects of this model.


To encompass the several diverse roles that package appearance is hypothesized to play in consideration and choice, a theoretical framework by Roberts (1989) is adopted and extended. As shown in Figure I, Roberts casts choice as a phased process comprising three stages in time (pg. 749): "The probability of brand choice, given category purchase, can be thought to have three elements: the probability of being aware of brand j; the probability of considering brand j, given awareness of it; and the probability of choosing brand j, given awareness and consideration." This framework forms the backbone of the extended model shown in Figure II, which explicitly considers the effects of different kinds of visual stimulus on those stages of choice where they are hypothesized to play a role.

Stage O concerns the likelihood that a given consumer will attend to a given product category motivated by a pragmatic set of needs-satisfying goals, such as, "I am thirsty," or "I am out of detergent." These pragmatic goals are thought to predispose the consumer to those brands that he or she expects will provide those performance benefits. These brands are called typical for the category. Visually typical brand alternatives are more likely to be noticed and preferred by the consumer (Loken and Ward 1990), and will likely be chosen for purchase unless some distracting stimulus or event interrupts and changes the decision process. Those brands which succeed in distracting the consumer are called visually novel.

It will be argued in the following that package appearance can play important roles in defining the perceived visual typicality or novelty of brands, particularly the latter.

Stages I and II concern the formation of two brand sets that are monotonically related: first, an awareness/ attention set (hereafter called attention set), in which brands in a supermarket facing are first seen from some distance and attended to on the basis of (relative) appearance; and second, a consideration set, in which some members of the Stage I attention set are evaluated and considered for purchase. It is these two stages which are the subject of the three hypotheses.

Stage I defines a visual scanning process that begins as soon as the category facings are in sight, which is to say at some distance. This search of the facings continues as the consumer moves closer. If the number of brand alternatives is sufficiently large (Bettman 1979) C a number which may not be very large at all (Roberts and Lattin 1991) C the consumer will not likely attend to all of them, but only to those that stand out according to the sufficiency of their visually typical or novel appearance.

Where Stage I is conceived as a process of arousal and motivation, Stage II is more evaluative in nature, as the consumer studies the brands more closely, picking up and reading the packages, checking the price and so on. Consumers will tend to consider visually typical brands that pass into their consideration sets because they expect that they will perform well (Nedungadi 1990). However, consumers have no such expectations of visually novel brands. Therefore, for a novel brand to graduate from the attention set to the consideration set, it must pass an appropriateness screen. That is to say, it must be judged as appropriate for the category, or concordant with performance needs and goals.

Choice of a brand from the consideration set is accomplished by a final decision step.


Three hypotheses are developed. They concern the conjunctive screens related to the formation of the respective brand sets of Stages II and III shown in Figure II. The first two hypotheses concern the visual typicality and novelty criteria associated with Stage I attention sets. Hypothesis III concerns the appropriateness criterion associated with Stage III consideration sets.

Hypotheses 1 and 2: Formation of the Attention Set

This paper concerns itself with the relative role of package appearance. In a competitive environment such as a supermarket aisle, brands are seen in the context of their competitors in contiguous category facings such as those shown in Figure III. Therefore, the effect of brands as visual stimuli is relative to a background comprised of competitor brands, all of whom are "shouting" for the shoppers's attention.





Visual Typicality and Novelty as Visual Attributes. Acknowledging the relative nature of visual stimuli in a competitive environment, the distinctiveness of stimuli must be defined and measured according to some notion of what is visually typical, or prototypical for a category. In this paper, it is argued that there are two possible visual strategies whose objective is to increase likelihood of inclusion in the consumer's attention set. The first strategy is to appear sufficiently visually typical for the category. The second strategy is the opposite of the first, or to appear sufficiently visually atypical, or novel. These ideas are represented graphically in Figure IV, where brands in a category are represented as points in a visual attribute space.

Each brand is defined relative to a category prototype. Perceived similarity or dissimilarity in a brand's appearance, relative to the category prototype, is represented on the map by its proximity or distance from a central point representing the point of visual prototypicality. For example, if a given brand falls relatively close to the prototype on all visual dimensions, then it is deemed relatively visually typical for the category. If a given brand falls relatively far from the category prototype on any visual dimension, it is deemed visually novel for the category.

These notions are operationalized by conceiving of discrete zones fanning out concentrically from the position of the category prototype. Any brand that falls in the visually typical zone, depicted as the inner circle falling closely around the point of visual prototypicality, is declared visually typical for the category, as perceived by the consumer. Brand #1 is therefore declared typical. In contrast, any brand falling in the outermost zone is declared visually novel. Brands #3 and #4 are therefore declared novel. Any Brand falling into the doughnut-shaped zone between these two is declared neither visually typical nor novel. Brand #2 is therefore declared neither typical nor novel. In this way, all brands may be categorized according to where they fall relative to their distance from the point of visually prototypicality,



Visual typicality is defined as the look or appearance that most consumers would associate with a product category, and by which they identify brands that belong to the category. Examples would include the white bags associated with flour, as shown in Figure V, or the long, narrow boxes with a cellophane window associated with spaghetti. This definition is not to be confused with the more conventional notion of visual typicality found in the marketing literature, defined in terms of a sharing or commonality of performance attributes (Loken and Ward 1990). For our purposes, one may define visual typicality as a commonality or sharing of visual attributes. A visual strategy that employs a relatively visually typical package appearance may be thought of as a visual me-too strategy, one that emphasizes the brand's membership in the category. A visual strategy that employs a relatively novel package appearance may be thought of as a product differentiation strategy, one that emphasizes differences rather than similarities.



Attention to a visually typical brand is purposeful, the product of visual search and identification. By contrast, attention to a visually novel brand is involuntary: the eye cannot help but register the sensation of a vivid and unexpected visual element that enters its field of vision, such as a bizarre color or shape (Kahneman 1973).

The graph at the bottom of Figure IV relates perception of similarity to likelihood of attention or consideration. Put another way, a brand's position in visual attribute space, and relative to the category prototype and the zones of visual typicality and novelty that surround it, indicates the relative likelihood of attention and consideration on the part of a given consumer, according to the function that is specified. Specifically, those brands that fall within the zone of visual typicality are relatively likely to be attended to and considered. Those brands that fall within the zone of visual novelty are also relatively likely to be attended to. But those brands that fall between these zones, indicating that the brands that occupy this space are insufficiently visually typical or novel to be noticed, are therefore less likely to be attended to and considered.

Regardless of the differences by which visually typical versus novel brand stimuli have their effect, the strategic importance of the effect on likelihood of attention or consideration in both cases is as distractors to target brands at the point of purchase. Both kinds of brand stimuli may serve to distract the consumer away from some target brand, and may in turn cause him or her to add the distractor brand to their attention set. Accordingly, I propose the following hypotheses.

H1: Brand packages which are sufficiently visually typical are more likely to enter the consumer's attention set, than are brand packages which are insufficiently visually typical and not novel.

H2: Brand packages which are sufficiently novel are more likely to enter the consumer's attention set, than are brand packages which are insufficiently novel and not visually typical.

Hypothesis #3: Formation of the Consideration Set

Visual Novelty and Appropriateness. Appropriateness refers to the performance attributes or needs benefits that the consumer may infer about a product from its package. Having entered the attention set by dint of its appearance, it is argued that the brand may subsequently pass into the consideration set only if the brand is also deemed to perform sufficiently well.

All visually typical brands are by definition appropriate. Therefore, it is only with visually novel brands that appropriateness as a screening device for consideration is relevant. Having attended to a visually novel brand precisely because it is novel, because of that very novelty its efficacy as a provider of needs benefits is called into question. Therefore, visually novel brands in the attention set must pass an appropriateness screen to gain entrance to the consideration set. Accordingly, I propose the following hypothesis.




H3 Novel packages which are appropriate to the category are more likely to pass from the attention set into the consideration set, than are novel brand packages which are inappropriate to the category.

The empirical results that these hypotheses indicate are represented graphically on Figure IV. The greater likelihood of a visually typical stimulus being attended to, relative to a stimulus that is not visually typical and not novel, is indicated by the positive slope of the line labeled "H1", on the graph showing the relationship between probability of attention or consideration, and package type. The greater likelihood of a visually novel stimulus being attended to, as opposed to a stimulus that is neither visually typical nor novel, is indicated by the negative slope of the line marked "H2". The greater likelihood of an appropriate visually novel stimulus being considered, as opposed to a novel stimulus that is not appropriate, is indicated by the positive slope of the line marked "H3".


The empirical test of this model incorporates experiments in which subjects undertake a series of six computer-simulated shopping trips across four supermarket categories: flour, raisins, spaghetti and dry cereal. The parts of the theoretical framework to be tested are the elimination screens relating to Stages I and II of Figure II, the subject of the above three hypotheses.

The computer simulation to be employed has been developed and validated by Professor Raymond Burke of Harvard University (Burke, Harlam, Kahn and Lodish 1992). Figure VI shows a still photograph of the computer display. By this simulation, the subject is able to browse a set of supermarket facings, pull brands of interest from the shelf, turn and read the package, and select it for purchase. The computer accurately times and records all relevant behaviors.

To test the above hypotheses, I intend to manipulate the levels of visual typicality, novelty and appropriateness to measure their effect on attention and consideration. These manipulations will be introduced by systematically altering the packages of existing brands. These alterations are to be calibrated through pretesting to assure that they correctly represent the levels of visual typicality, novelty or appropriateness that are to be specified.

Stage O, selection of the product category, is not treated because it is assumed that the category is already selected, and the other stages are predicated on that. Stage III is also not treated, for three reasons. First, this step is seen as akin to a one-stage compensatory choice model, which has received ample treatment in the literature. Secondly, the effects of package appearance are hypothesized to effect choice only indirectly, through its effects on consideration set formation. And third, the current test is unlikely to have sufficient power to reliably detect the effects of these manipulations on choice.

The Calibration of Changed Packages by Pretesting

The stimuli in this experiment are existing brand packages that have all been changed in various ways, cleaned or altered, in order to be sure of the presence of certain visual attributes at certain levels, or to be sure of their absence. Pretesting is performed to calibrate the packages according to these qualities, and otherwise to assure that the changed packages correctly represent the qualities that are intended of them, and do not introduce extraneous factors.

The Experimental Design

I am proposing a 4X4 latin square design as shown in the following table. 160 subjects C recruited by newspaper from the town of Cambridge, Ma., and paid an incentive fee of $20 C are assigned at random to one of four groups of forty. There are three factors: a product category factor (flour, spaghetti, raisins, dry cereal), a package alterations factor (visually typical, novel-appropriate, novel-inappropriate, and neither visually typical nor novel), and a group factor.


The subject of this research is how consideration sets are formed, and how visual perception plays a role in its formation at the point of purchase. The domain of the proposed model is the competitive purchase display, where all category brands are seen together, and competitor brands thus form the visual context for each other. It is suggested that the model of consideration set formation with visual search is general in nature, given the primacy of the visual sense and the constant use of visual search at the point of purchase.

The empirical model proposes a test of the elimination criteria of Stages I and II of the model: that a sufficiency of the visual attributes typicality and novelty will indicate a greater likelihood of attention, and that the appropriateness of a visually novel brand will indicate a greater likelihood of consideration. The theoretical model is itself more broad based, and if correct suggests a research stream. Topics for follow-up research may include: the effect of individual and cultural differences; how niche brands may exploit visual opportunities; and how the pioneer package may set the visual precedent for the category.


Bettman, James R. (1979), An Information Processing Theory of Consumer Choice, Chicago: Addison-Wesley.

Nedungadi, Prakash, (1990), "Recall and Consumer Consideration Sets: Influencing Choice Without Altering Brand Evaluations," Journal of Consumer Research, 17 (December), 263-276.

Burke, Raymond R., Bari A. Harlam, Barbara E. Kahn and Leonard Lodish (1992), "Comparing Dynamic Consumer Choice in Real and Computer-Simulated Environments," Journal of Consumer Research, 19 (June), 71-82.

Kahneman, Daniel (1973), Attention and Effort, Englewood, N.J: Prentice- Hall.

Loken, Barbara and James Ward (1990), "Alternative Approaches to Understanding the Determinants of Typicality," Journal of Consumer Research, 17 (September), 111-126.

Roberts, John H. (1989), "A Grounded Model of Consideration Set Size and Composition," Advances in Consumer Research, 16, 749-57.

Roberts, John H. and James M. Lattin (1991), "Development and Testing of a Model of Consideration Set Composition," Journal of Marketing Research, XXVIII (November), 429-40.

Shocker, Allan D., Moshe Ben-Akiva, Bruno Boccara, and Prakash Nedungadi (1991), "Consideration Set Influences on Consumer Choice: Issues, Models and Suggestions," Marketing Letters 1:3, 181- 197.



Lawrence L. Garber, Jr., University of North Carolina at Chapel Hill


NA - Advances in Consumer Research Volume 22 | 1995

Share Proceeding

Featured papers

See More


Is Congruity Desirable for Brand Extensions? A Conceptual and Meta-Analytic Review

Qian (Claire) Deng, University of Prince Edward Island
Paul Richard Messinger, University of Alberta, Canada

Read More


Mere and Near Completion

Bowen Ruan, University of Wisconsin - Madison, USA
Evan Polman, University of Wisconsin - Madison, USA
Robin Tanner, University of Wisconsin - Madison, USA

Read More


Machine Talk: How Conversational Chatbots Promote Brand Intimacy and Influence Consumer Choice

Thomas Hilden, University of Geneva, Switzerland
Christian Hildebrand, University of Geneva, Switzerland
Gerald Häubl, University of Alberta, Canada

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.