Visual and Verbal Memory in Children's Product Information Utilization

John R. Rossiter, University of Pennsylvania
ABSTRACT - Consumer research has largely ignored product-relevant information stored in visual memory. Visually stored images may be quite sufficient to engender product choice, quite apart from attitudes, beliefs, and so on, retrieved from verbal or symbolic memory. Such may be the case when children or adults encounter familiar products on a supermarket shelf: visual recognition latency may be all that is involved in the choice process. The present experiment (1) confirmed the existence of a rich data base in children's visual memory for cereal brands, (2) demonstrated that visual information differs from verbal information and that children's choices may differ depending on which of the two types of information is situationally retrieved, and (3) pointed out measurement biases favoring verbal memory in contemporary consumer research and urged more research on visual memory, both with child as well as adult consumers.
[ to cite ]:
John R. Rossiter (1976) ,"Visual and Verbal Memory in Children's Product Information Utilization", in NA - Advances in Consumer Research Volume 03, eds. Beverlee B. Anderson, Cincinnati, OH : Association for Consumer Research, Pages: 523-527.

Advances in Consumer Research Volume 3, 1976      Pages 523-527

VISUAL AND VERBAL MEMORY IN CHILDREN'S PRODUCT INFORMATION UTILIZATION

John R. Rossiter, University of Pennsylvania

[The study was conducted at The Wharton School's Center for Research on Media and Children. The author wishes to acknowledge the support of Professor Charles S. Goodman, Chairman of Wharton's Department of Marketing, and to thank John Trent for assistance in the research administration.]

ABSTRACT -

Consumer research has largely ignored product-relevant information stored in visual memory. Visually stored images may be quite sufficient to engender product choice, quite apart from attitudes, beliefs, and so on, retrieved from verbal or symbolic memory. Such may be the case when children or adults encounter familiar products on a supermarket shelf: visual recognition latency may be all that is involved in the choice process. The present experiment (1) confirmed the existence of a rich data base in children's visual memory for cereal brands, (2) demonstrated that visual information differs from verbal information and that children's choices may differ depending on which of the two types of information is situationally retrieved, and (3) pointed out measurement biases favoring verbal memory in contemporary consumer research and urged more research on visual memory, both with child as well as adult consumers.

INTRODUCTION

Most consumer choice models based on the information processing approach presume that choices derive from a synthesis or integration of beliefs (pertaining to information about the choice object) which are largely verbalizable. That is, the models assume that beliefs are represented linguistically in memory. Thus we have our Ss exhibit their belief probabilities, belief evaluations, and so forth, via questionnaires and rating scales. These procedures rely almost exclusively on verbal task instructions.

The author and his colleagues (Calder, Robertson and Rossiter, 1975) have suggested that this reliance on verbally encoded information may overlook a great deal of nonverbal information which consumers utilize in decision making. One important type of nonverbal information is visual information. Its importance seems to be increasing in our culture with the dominance of television as an information medium, particularly as a medium for the advertising of products to children.

This study examines children's utilization of visually stored information about cereals. The general hypothesis is that children have acquired and stored in memory a rich visually encoded data base on cereal attributes. This type of data representation may differ totally from any verbal representation of cereal attributes. Indeed, it may not be amenable to verbal retrieval because a translation from one code to another would be required in the output process (see, for example, Paivio, 1971; Posner, 1973). This transformation may be beyond the capacity of young children.

The importance of studying children's visual memory has not yet become apparent in children's consumer research. Its importance lies in the fact that retrieval of visual representations of (for example) cereal attributes may, alone, be sufficient to engender product preference. The child need only retrieve a visual match for the cereal in a "familiarity" or recognition latency sense. At home or in a store, for example, greater familiarity or faster recognition of one item vs. others may be transformed into a preference indication by the child through the simple act of pointing. And even if the child can not only point to but can also name the cereal verbally, the process involved may still be one of imitative (in this case auditory or "echoic") image retrieval. This contrasts with a more adult-like symbolic information retrieval and information evaluation process.

The process distinction between child and adult modes of decision-making also has an important implication for preference or response measurement. Images are characterized as being "stimulus faithful" representations of the original stimulus or referent. Examples would be an image of a cereal carton in the visual mode or of a cereal brand name in the auditory mode. Symbolic representations, in contrast, bear only an arbitrary relationship to the original stimulus or referent. Answering a verbally stated preference question or considering an equally abstract numerical rating scale are tasks which require symbolic transformation ability. The child may only have stored the attribute data in a pre-symbolic, imagery form. Thus he or she may not be able to indicate preference in a verbal questioning procedure or on a rating scale, yet a preference may clearly exist. Alternatively, preferences may differ as a function of the particular mode of response used in measurement, especially with children. This possibility has not been explored in children's consumer decision research.

An adequate exploration of the child's visually operative data base of cereal attribute information requires an appropriate response mode and a reasonably wide range of output opportunity. This was achieved in the present study by asking children to draw cereals, freehand, from memory. They were asked to draw a generic cereal box, i.e., just "a cereal"; the back of a cereal box (a primary location for promotional material); their favorite cereal; the cereal their parents eat; and a cereal that is "healthy and good for you when you are growing up." These five tasks provided multiple opportunities to assess visually recallable information concerning brand identification, nutrition, sweetness or sugar content, premium offers, and other salient attributes of cereals.

The children were then asked to describe, verbally: their favorite cereal; parents' cereal; and a healthy cereal. For fairly obvious reasons of abstraction difficulty the children were not asked to describe a generic cereal. For the three more concrete tasks, verbal attribute emphasis could thus be compared with. visual attribute emphasis as a test of the general research hypothesis of uniquely stored visual attribute knowledge.

METHOD

Design. Because it was essential that the visual information measures not be contaminated by prior verbal recall, the drawing tasks were administered first. Children received the drawing task instructions in the same order: generic, back of box, favorite, parents' and healthy. They then received the verbal questions in the same order: favorite, parents', and healthy.

The attribute emphasis variables were scored by coding the drawings and the verbal protocols on a simple attribute present/attribute absent basis. Coding was done by one experimenter (J.T.) and checked by J.R. The dichotomous judgment procedure produced very few disagreements.

Subjects. Participants in the study were 60 boys selected from two Philadelphia area schools of similar middle class enrollment. Three grade levels were included--1st, 3rd and 5th--with 20 Ss per grade.

Apparatus and materials. The children were provided with 6 or 7 sheets of plain white 8 1/2 x 11" paper. Each child was given an identical box of 10 new felt-tipped pens for the drawings. Each box was used three or four times, but carefully checked for new pen condition before repeated use. Colors provided in the box (Bic "Bananas") were red, green, blue, yellow, light blue, orange, black, purple, brown and pink.

Procedure. The study was administered to groups of 5 or 6 children from each class, using a large room in which the children were spaced to prevent sight of one another's drawings, with their work desks facing the outer walls. Two experimenters were on hand to read the instructions, check the children's general progress (though they had unlimited time if needed) and administer the verbal questions.

Instructions for the cereal drawing task were as follows:

1. a. Draw a cereal box using these color pens.

b.(Probe:) Have you drawn in everything you can remember about the cereal box?

2. a. Would you draw me the back of the cereal box now.

b.(Probe:) Have you drawn in everything you can remember about the back of the cereal box?

3. a. Draw your favorite cereal.

b.(Probe:) Have you drawn in everything you can remember about your favorite cereal?

4. a. Draw the cereal your parents eat.

b.(Probe:) Have you drawn in everything you can remember about the cereal your parents eat?

5. a. Draw a cereal that is healthy and good for you when you are growing up.

b.(Probe:) Have you drawn in everything you can remember about a cereal that is healthy and good for you when you are growing up?

While drawing, children who asked how to spell words (e.g., the name of the cereal) were told, "Spell it as best you can." Children who asked how to draw the boxes, how big, etc., were told, "Any way you want to draw it."

Questions for the verbal attribute measures were as follows:

1. a. What is your favorite cereal?

b. What kind of cereal is it? (Probe:) What else can you tell me about your favorite cereal?

2. a. What is the cereal your parents eat?

b. What kind of cereal is it? (Probe:) What else can you tell me about the cereal your parents eat?

3 a. What is a cereal that is healthy and good for you when you are growing up?

b. What kind of cereal is it? (Probe:) What else can you tell me about a cereal that is healthy and good for you when you are growing up?

The verbal interviews were administered individually. The children were then returned to the classroom after being asked not to tell others about the questions and drawings until tomorrow (it was a one day administration per grade).

RESULTS AND DISCUSSION

The General Hypothesis

Children clearly do have strong visual imagery of cereals. The drawings themselves were quite remarkable for their overall conceptual execution (in a general "schema" sense) and for their detailed inclusion of attribute data. The imagery evident in the drawings was strong and vivid enough to indicate that children could well retrieve and utilize these iconic representations in decision making. Some most interesting evidence that visual imagery does include preference information was obtained in the fact that the generic cereals (i.e., "Draw a cereal box") were branded spontaneously by a majority of the children (Table 1). At first grade 67% of the children branded their generic concept drawings; this increased to 100% at third and fifth grades.

Dual Code Usage

Brand names are an attribute that can be represented in both the visual and verbal output modes. The extent to which children used both modes is also shown in Table 1. Brand names were "mentioned" both visually and verbally (i.e., printed or partially printed in correct visual location on the drawing [Some of the attempts by younger children to draw in the brand name were not fully decipherable; in these instances, the apparent brand identification was cross-checked with correctness of color-of-carton cues (see section on color matching).] and mentioned in response to the verbal questions) by only about 60% of first graders. However, this double code retrieval rose to 84% at third grade and 93% by fifth grade. The increase in double code retrieval held for the three tasks for which visual and verbal responses were required: favorite, parents' and healthy (the percentages just referred to are averages of these three tasks from Table 1).

Brand Matching

Of critical interest in terms of the general research hypothesis predicting unique usage of visual and verbal codes are the rows in Table 1 showing the match between visually and verbally supplied brand names.

On average, the brands nominated in the two codes (within child) matched only 52% of the time for first graders with, again, a marked increase to 59% at third grade and 77% by fifth grade.

Looked at another way, the uniqueness contributed by the visual data--brand name salience that would have been missed by the verbal procedure alone--was 16% at first grade (though remember that fewer brand names were included by the 6 to 7 year olds), rising to a very substantial 40% at third grade, and 22% at fifth grade.

TABLE 1

BRAND NAME EMPHASIS IN VISUAL AND VERBAL RECALL

The finding that uniqueness was highest at third grade, coupled with the incidence of matching at each age level, suggests a tentative set of hypotheses about children's visual and verbal code usage in their "internal search" for brand information. The 6 to 7 year-old first graders seem to be characterized by "differential focus" on the two codes, with somewhat unclear focus on either. The 8 to 9 year-old third graders seem to be characterized by a period of clear focus accompanied by "maximal discrimination and perhaps conflict" between the two codes. The 10 to 11-year-old fifth graders show a "coincidence of codes." The older children may be becoming more adult-like in their cognitions in that their visual and verbal referents are more likely to coincide.

Support for these contents comes from quite a different set of dimensional results: the colors which the children employed in their cereal box drawings (Table 2).

The top row of each sub-table provides a measure of the vividness of children's imagery as indexed by the (average) number of different colors the children used. Here it can be seen that for four of the five tasks vividness reached a peak at third grade. Taken over all tasks, first graders selected an average of 3.24 different colors from the available set of ten; third graders, 4.52; and fifth graders, 4.24. This curvilinear trend in vividness tends to support a model of maximal dual code discrimination at about age 8 or 9.

TABLE 2

VISUAL COLOR USAGE (VIVIDNESS), COLOR MATCHING, AND CORRECTED "HIT RATE"

Color Matching

Color matching gauges the extent to which the colors selected by the child matched the colors on the actual cereal box. Matching was ascertained by purchasing all of the cereals nominated by the children and comparing them with the individual drawings. A "match" was scored only if both the color and the location were correct.

Color matching provides a second type of evidence for the original research contention that visual retrieval can play a guiding role in preference since the color and its carton location would presumably have to be retrieved from visual memory. Matching data for colors are shown in the second row of the sub-tables inTable 2. Also, to adjust for any chance matches due to the sheer number of colors each child used, a corrected hit rate was computed. This is shown in the summary row of the sub-tables-

Even the youngest children were able to match correctly at a rather high rate. Again remembering that only about two-thirds of the drawings were brand-identified at this age, the first graders color matched correctly 40% of the time over the five tasks. The color match rate then rose linearly to 52% at third grade and 63% by fifth grade. It is interesting to note that color matching was highest for "favorite" cereals at the two younger grades (it may have asymptoted at fifth)-

Taken overall, the demonstration of color matching of cereal packages in the visual mode yields powerful evidence that visual imagery is alone sufficient to influence preference. All that would be required is visual recognition by color. It might be noted in this connection that if a child were asked to color match verbally, i.e., to describe from memory the colors on a cereal carton, he would almost certainly have to "read" the description from a visually stored image.

TABLE 3

ATTRIBUTE EMPHASIS

Attribute Emphasis

Three primary cereal attributes were examined in the present study: sugar (or sweetness) emphasis, nutrition emphasis, and the presence of premiums as a means of promoting the cereal.

Sugar emphasis. Sugar or sweet flavor emphasis is difficult to represent visually; and also, for two or three years now, "sweet" claims have rarely been emphasized explicitly in cereal packaging or in television (visual) advertising. It is not surprising, therefore, that apart from occasional flavor specification in the drawings, visually represented sugar attributes did not appear. Verbal attribute emphasis for sugar and sweetness, however, was most interesting (Table 3; as described earlier, verbal questions were not asked for the generic or the back of box tasks). For favorite cereals, sugar or sweetness attributes were mentioned by 42% of the first graders and 41% of the third graders, but by only 26% of the fifth graders. For parents' cereals, sugar mentions were really low: 17% or less at all grades. For healthy cereals only 5% of first graders and none of the third graders mentioned sugar or sweetness; the fact that the sugar figure then rose at fifth grade, to 20%, could indicate a very sophisticated response by these older children: Overall, the pattern is an encouraging one from an "educational" standpoint, although this was, as noted, a generally well educated research sample.

Nutrition emphasis. Nutritional attributes (mentioned as "vitamins," "natural," etc.) might be regarded as a counterpart to sweetness. These data are also shown in Table 3. Once again the results show an encouraging trend. As might be expected, the highest nutrition mentions were for the healthy cereals: 37% at first grade, 50% at third, and 60% at fifth. Also of interest was the sharp increase at fifth grade where 58% of the children also mentioned nutrition as an attribute of their favorite cereal.

Premium emphasis. Premiums are clearly a salient feature in children's visual memory of the generic concept "cereal." This emphasis increased with age. Figures for the back-of-box drawing task, where premiums are typically displayed, were 39% at first grade, 55% at third grade, and 75% at fifth. Front-of-box premium inclusion was somewhat lower, as might be expected, but still showed the increase: 0% at first grade, 15% at third grade, and then 45% at fifth grade. The latter finding is interesting because it may reflect increasing awareness and comprehension of claims about premiums (a symbolic verbal element) rather than simply iconic representation of the premium in visual memory.

The high incidence of premium inclusion in children's generic conceptions of cereals and the linear increase with age suggest that premiums do indeed become perceived as an attribute of cereals. Administration of a back-of-box drawing task for children's favorite cereals may have revealed even greater salience of premium information. This finding is of considerable interest in the light of recent controversy over whether children comprehend premiums as part of the product or as a separate choice consideration. A children's cereal without a premium may be lacking an expected attribute.

CONCLUSIONS

The intent of this study was to demonstrate that children have a rich data base of information about products (such as cereals) which is stored in visual rather than verbal memory and which can alone influence their preference process.

It was shown that retrieval of visual information can lead to quite a different indication of children's brand preferences. Visual and verbal nominations of the child's favorite cereal, for example, only corresponded in 42% of instances at age 6 to 7, 53% at ages 8 to 9, and 74% at ages 10 to 11. Visual representation therefore contains a very substantial amount of information (ranging from 8% to 44% of brand preferences) that would be missed by traditional verbal or numeric measures of children's preferences.

It was further shown that visual representations of cereals are sufficiently strong or vivid to be utilized in children's preference decisions without symbolic verbal assistance (e.g., in a supermarket situation, visual recognition then pointing). This was evidenced by the fact that even the youngest children could correctly color-match the details of-actual boxes with 40% accuracy (48% for their favorite brand), a figure which increased to 63% for the older children. Colors must almost certainly he retrieved from purely visual memory images of the cereals.

Some attributes such as sweetness or nutrition claims are, by nature, difficult to encode visually. However, it was found that another dominant attribute of cereals--promotional premiums--can be and are coded visually. When asked to draw the back of a cereal box 39% of first graders included a premium, 55% of third graders did so, and 75% of fifth graders. Unfortunately, the back-of-box task could not be included for every drawing, so the extent to which premiums are salient to the child's favorite cereal concept was not fully ascertained. Premiums did, however, receive very low verbal recall, suggesting again the uniqueness of visual retrieval.

The visual and verbal data obtained in the study also showed highly consistent trends (generally linear increases) with age. One exception was vividness of visual imagery, which appeared to peak in curvilinear fashion at third grade, an age at which the discrepancy between visually and verbally produced preferences also happened to be at a maximum. This Suggests at least one innovative direction for further research. The challenge from a research design standpoint and the fascinating output of visual memory measurement should be enough to stimulate many others.

To reiterate the issue with which this research on visual memory began: consumer preferences may be based on information other than that represented in the form of verbal (or verbalizable) beliefs. Consumers may utilize information stored uniquely in visual memory. This data base may not be amenable to retrieval via the typical verbal measurement procedures such as those employed, for example, in conjunction with "Fishbein-type models."

The present study focused on visual and verbal memory in children's consumer behavior. Visual memory may be more important than is commonly realized in adult decision making. For example, all of us from time to time consult our "mental maps" of shopping environments and store layouts to facilitate our consumer decision processes. Once we are in the store, in a supermarket for instance, we may then choose to "breeze through," making many of our brand selections in a rather uninvolved, rather automatic way. In doing so we may be relying on momentary evocation of attitudes or preference indications from a data base located primarily in visual memory.

APPENDIX

Tabular Base Levels

Tables are based on N = 20 per grade level with the following exceptions: 2 first grade children could not conceptualize a generic cereal even after two or three instruction repetitions; 1 first grader, 3 third graders and 1 fifth grader did not have a particular favorite cereal; the parents of 8 first graders, 8 third graders and 3 fifth graders did not eat cereal, thus making the parents' cereal task irrelevant for them; and 1 first grader could not conceptualize a healthy cereal. Base levels for proportion data in the tables were adjusted accordingly.

REFERENCES

Bobby J. Calder, Thomas S. Robertson and John R. Rossiter, "Children's Consumer Information Processing," Communication Research, 2 (July, 1975), 305-316.

Allan Paivio, Imagery and Verbal Processes (New York: Holt, Rinehart & Winston, 1971).

Michael I. Posner, Cognition: An Introduction (Glenview, Illinois: Scott, Foresman, 1973).

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