The Role of Prior Knowledge in the Acquisition, Retention, and Use of New Information

ABSTRACT - Four experiments investigating the role of prior knowledge in memory and judgment are reported. The first two studies examined the role of product familiarity on a variety of memory tasks. It was found that high familiarity subjects often recall more new information than low familiarity subjects. A variety of supplementary analyses suggest that this is due to more elaborate knowledge structures that permit the use of much more active organizational strategies. It was also found, however, that the same elaborate knowledge structures of high familiarity subjects result in more brand-attribute confusions when a large number of very similar items are presented. The final two experiments examined the role of prior familiarity in judgment processes. It was found in Experiment 3 that ratings of credibility were much less affected by repetition for high than low familiarity subjects, although it had a reliable effect in both cases. The final experiment demonstrated that although the subject's own mood state has an effect on product evaluations for low familiarity subjects, it has none whatsoever for high familiaritY subjects.

Citation:

Thomas K. Srull (1983) ,"The Role of Prior Knowledge in the Acquisition, Retention, and Use of New Information", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 572-576.

Advances in Consumer Research Volume 10, 1983      Pages 572-576

THE ROLE OF PRIOR KNOWLEDGE IN THE ACQUISITION, RETENTION, AND USE OF NEW INFORMATION

Thomas K. Srull, University of Illinois

ABSTRACT -

Four experiments investigating the role of prior knowledge in memory and judgment are reported. The first two studies examined the role of product familiarity on a variety of memory tasks. It was found that high familiarity subjects often recall more new information than low familiarity subjects. A variety of supplementary analyses suggest that this is due to more elaborate knowledge structures that permit the use of much more active organizational strategies. It was also found, however, that the same elaborate knowledge structures of high familiarity subjects result in more brand-attribute confusions when a large number of very similar items are presented. The final two experiments examined the role of prior familiarity in judgment processes. It was found in Experiment 3 that ratings of credibility were much less affected by repetition for high than low familiarity subjects, although it had a reliable effect in both cases. The final experiment demonstrated that although the subject's own mood state has an effect on product evaluations for low familiarity subjects, it has none whatsoever for high familiaritY subjects.

INTRODUCTION

The present paper is concerned with the role of prior knowledge in the acquisition, retention, and use of new information. 'While it may seem intuitively obvious that human learning, memory, and judgment in almost any domain will be heavily determined by one's previous knowledge base, the underlying mechanisms responsible for this were largely ignored until very recently. An increasing number of relevant investigations began to appear during the last two decades and now the role of prior knowledge in information processing has become a major topic of investigation in the cognitive sciences.

Several important investigations in the consumer domain .have appeared during the past several years. In one of the most impressive, Johnson and Russo (1981) outline two plausible hypotheses of how prior familiarity with a product might affect the learning of additional information about that product. The first they term the "enrichment hypothesis." Quite simply, greater prior knowledge should facilitate the subsequent learning of new information. One classic example of this is the work of Chase and Simon (1973). Expert chess masters and novice chess players served as subjects, and they were shown various board positions for brief durations. Chase and Simon found that experts remembered much more than novices when plausible board configurations were presented. Theoretically, chess masters should be able to use their prior familiarity with the same to help them identify, organize, and retrieve meaningful clusters of information. However, Chase and Simon also found that there was no difference between 'he masters and novice players when random board configurations were presented to subjects. Although this may appear inconsistent with the enrichment hypothesis, it really is not. Theoretically, prior familiarity with the game will not aid in the identification, organization, or retrieval stages when random stimuli are used. In other words, the highly organized knowledge of the masters simply is r.of relevant to the processing of completely disorganized material.

Johnson and Russo also outline an "inverted-U" hypothesis. This is based on the earlier work of Bettman and Park (1980). In brief, Johnson and Russo suggest that consumers with little prior familiarity with a product will have difficulty comprehending new information. As a consequence, memory will be relatively poor. Consumers with intermediate levels of prior knowledge should be able to comprehend new information and they will also tend to engage in an extensive memory search for many types of tasks. Thus, they will generally have very good recall of new information. In contrast, the hypothesis also suggests that highly familiar consumers will have poor memory for new information in many situations. This will often occur when they receive new information with specific processing objectives in mind. -Under such conditions, they may already "know" what is relevant and what is irrelevant information, and thus not extensively process much of it. Alternatively, in order to make certain types of judgments, they may engage in less memory search than individuals with only moderate levels of familiarity.

Using an incidental memory paradigm, Johnson and Russo found that the effect of prior familiarity on memory is highly dependent on the initial processing objectives of the subject. All subjects were given a large brand x attribute matrix concerning automobiles. Some subjects were asked to rate the favorability of each auto. Other subjects were given the same matrix but asked to choose the most preferred auto. Following these judgments, subjects were given an unexpected free recall test. Johnson and Russo found that there was a linear increase in levels of recall as a function of prior product familiarity when subjects judged each product independently. In contrast, levels of recall in the "choice" condition followed the inverted-U pattern. why would subjects with high prior familiarity recall less than those with only intermediate levels of familiarity under the choice condition? One interpretation is that the high familiarity subjects simply recognized what attributes were unimportant and/or nondiscriminating between brands, and thus allocated very little attention to them. As a consequence, overall levels of recall for the information suffered.

In general, highly familiar subjects should be able to use their prior knowledge to great advantage. The reason for this is that, similar to the chess masters studied by Chase and Simon, they will have very well articulated knowledge structures available. As a consequence, they will be able to comprehend new information easily and organize it effectively. While the inexperienced person may have to deal with isolated bits of information, highly familiar individuals will be able to deal more with "organized packets" of information. Such packets have been called schemata, scripts, frames, and memory organization packages.

Several theories describing the role of prior knowledge in information processing have received considerable support. In particular, Hayes-Roth (1977) has proposed a general model of information processing, and Burke (1982) has developed the most well articulated theory that deals specifically with consumer memory and Judgment. The studies reported below were developed within the context of these two complimentary theories. In general, the results provide additional support for each of these theoretical frameworks.

EXPERIMENT 1

The first experiment was designed to test two simple hypotheses. First, nearly all relevant theories suggest that prior knowledge in a particular domain will facilitate the learning of subsequent information, at least under conditions in which subjects are not explicitly asked to make comparisons across attributes as in the earlier work of Johnson and Russo. Second, and more important, the primary reason for this is that high familiarity subjects will use more active organizational strategies. Several previous studies have suggested that such organizational strategies are reflected in levels of category clustering (Srull in press a,. Both of these hypotheses were examined.

Subjects were undergraduate students. Following the procedure used by Johnson and Russo, subjects rated their self-knowledge of automobiles in relation to the rest of the population and a median split was used to identify low and high familiarity subjects. Each subject learned eight attributes about each of five compact automobiles. Forty descriptive statements were constructed and administered to subjects at the rate of one every six seconds. Half of the subjects received the statements in random order, and half received them "blocked" by brand. Subjects received no other instructions than to pay attention to the information. After receiving all of the information, subjects were given a one-minute interpolated anagram task, and then were asked to recall as much of the information as possible.

The mean number of items recalled by low familiarity subjects was 8.4 in the random condition and 13.6 in the blocked condition. High familiarity subjects recalled an average of 16.7 items in the random condition and 1,.2 in the blocked condition. Statistical analyses indicate that slightly more items were recalled in the blocked than random presentation format condition. As predicted, high familiarity- subjects recalled many more items than low familiarity subjects. Perhaps even more interesting is the interaction between these two variables. The recall of low familiarity subjects was much lower in the random than blocked condition (mean difference = 5.2 items), while there was virtually no difference for high familiarity subjects (mean difference = 0.5 items). The most interesting aspect of these two factors is that one is internal to the organism and one is not. That is, prior familiarity affects how the subject goes about processing the information and the blocked-random presentation format essentially provides environmental constraints on how well he/she is able to do so.

The free recall data were also analyzed using the adjusted-ratio-of-clustering (ARC) index developed by Roenker, Thompson, and Brown (1971). Consistent with previous research, high familiarity subjects manifested much greater brand clustering than low familiarity subjects. There was also much greater clustering in the blocked than random presentation format condition. Again, however, it was the interaction between these two variables that was most impressive. Specifically, while there was very little difference in the amount of clustering of high familiarity subjects in the two conditions (Ms = .68 and .74 under the random and blocked presentation formats, respectively), low familiarity subjects showed much less clustering (M = .39) in the random than in the blocked (M = .62) presentation condition. This would be consistent with the hypothesis that low familiarity subjects largely use rote rehearsal as an encoding strategy. That is, if subjects recalled the items in exactly the same order in which they were presented, they would receive a mean ARC score of .00 (chance clustering) in the random presentation format condition and 1.00 (perfect clustering) in the blocked presentation condition.

It is also interesting that this pattern of clustering scores parallels that of the free recall data. In particular, there was a positive relationship (r = .41) between the degree of (brand) category clustering and the total number of items recalled. Finally, it is worth noting that all of these results show the same pattern if one uses a tripartite rather than median split on the original familiarity scores.

EXPERIMENT 2

The first experiment demonstrated that the relatively well articulated knowledge structures of high familiarity subjects provide a distinct advantage in terms of organizing new information. There is a potential liability associated with such knowledge structures however. Because they are used so frequently to organize new information, they will often result in strong interference effects, Both the Hayes-Roth and Burke theories suggest that interference effects should be most pronounced when very- similar materials or items are used. Previous research by Burke and Srull using a standard retroactive interference paradigm has provided empirical support for these ideas. More recently Burke has demonstrated that proactive inhibition operates in much the same way.

The present experiment examined a much more subtle type of effect. Subjects were provided with 40 brand-attribute pairs, representing eight attribute values about each of five separate compact cars. They were given no instructions other than to pay attention to the information. The forty items were presented in random order at a relatively fast rate (one every five seconds). Subjects were never allowed to look back through the items. After receiving the information, they were given a five minute interpolated task in which they were asked to recall as many of the United States as possible.

Following the five minute delay, subjects were given a 120 item yes-no recognition test. The test consisted of the original 40 items and 80 distractors. Half of the distractors consisted of an old brand with an attribute value that was previously associated with one of the other brands (subsequently termed a substitute attribute). The other half of the distractors consisted of an old brand paired with an attribute value that was not presented in the previous list (subsequently termed an intrusion attribute). It should be noted that since all of the stimuli were compact cars, the attribute values were extremely similar to one another.

The mean proportion of each type of item rated as "old" by low familiarity subjects was .49, .19, and .32 for the old items, substitute attributes, and intrusion attributes, respectively. The comparable proportions for high familiarity subjects were .57, .38, and .05.

There are several things worth noting about these data. First, high familiarity subjects recognized more of the old items than aid low familiarity subjects (.57 vs. .49). High familiarity subjects were also much better at rejecting items that paired an old brand with an intrusion attribute (.05 vs. .32). Perhaps more importantly, however, high familiarity subjects also exhibited twice as many old brand-attribute confusions as did low familiarity subjects (.38 vs. .199. This suggests that high familiarity subjects exhibited more problems in discriminability. Although they were better able to reject attribute values that were not previously presented, they were no better at correctly pairing previously presented attributes with their appropriate brand. Of course, it must be remembered that all of the attributes were highly similar to one another. Nevertheless, it is noteworthy that such brand-attribute confusions are more prevalent in high familiarity than low familiarity subjects. Identifying the precise boundary conditions under which such an effect occurs should be a major topic of future research.

EXPERIMENT 3

One reasonable hypothesis that has not been previously investigated in the literature is that high familiarity subjects should be more resistant to "irrelevant" influences of presentation on judgment. That is, because greater familiarity provides a more articulated organizational structure and presumably results in a richer encoding of new stimulus information, the role of extraneous factors on judgment should be reduced or even eliminated.

One such irrelevant factor that has been shown to have an effect on judgment is repetition. In particular, it has been found across a series of studies that repeated items are judged more likely to be true than items that are not repeated. Moreover, at least within certain constraints, confidence that an item is true continues to increase as the number of repetitions increases (see Bacon 1979; Hasher, Goldstein, & Toppino 1977). It is quite possible, however, that high prior familiarity will make one resistant to such effects. This possibility was examined in the present experiment.

Subjects were undergraduate students who were classified as high or low in prior familiarity on the basis of the same procedure described above. Three separate experimental sessions were involved, each separated by an interval of one week. In each session, the subject was read a series of 100 trivia statements that were not obviously true or false (e.g., Montana has a greater range in temperature than any other of the United States). The statements were read at the rate of one every 20 seconds and, following each statement, subjects were asked to rate the validity of the item on a scale ranging from (19 certainly false to (7) certainly true. This same procedure was used in each of the three experimental sessions.

The first session included 50 statements (dispersed randomly throughout the list: pertaining to automobiles e.g., The Cadillac Seville has the best repair record of any American made automobile). The remaining 50 "filler" items pertained to geographical, historical, and scientific issues.

Subjects also received a list of 100 statements during the second experimental session, and 50 of these pertained to automobiles. Of these critical items, 20 were repetitions of items presented during the first session and 30 were "new" items that were presented for the first time. The same presentation and rating procedure was used.

Finally, a list of 100 statements was also presented during the third session. Again, 50 were filler items and 50 were automobile related. The list was arranged in such a way that 20 of the statements were now repeated twice, 15 were repeated once, and 15 were "new" or items that were presented for the first time.

In sum, low and high familiarity subjects rated the validity of statements that were seen one, two, or three times during the two week period of the experiment.

The mean validity ratings of the statements in all three conditions of the experiment are presented in Table 1. There are several things to note about these data. First, the general increase in credibility of repeated statements reported by Bacon (1979) and Hasher et al. (1977) generalizes to these stimulus items. Once repeated items are judged to be more true than non-repeated items, and twice repeated (three exposure) items are judged to be more true than once repeated (two exposure) items. Moreover, this effect occurs for both low and high familiarity subjects.

TABLE 1

MEAN RATINGS OF VALIDITY OF NEW AND REPEATED STATEMENTS

It is also important to note, however, that the increase in credibility associated with repeated items is consistently less for the high familiarity than low familiarity subjects. In the second experimental session, for example, the difference between new and once repeated items is .54 for low familiarity subjects, but only .32 for high familiarity subjects. In the third experimental session, the difference between new and once repeated items is .31 for low familiarity subjects, but only .18 for high familiarity subjects. Similarly, the difference between twice repeated and once repeated items is .42 for low familiarity subjects but only .31 for high familiarity subjects.

It is obvious that these differences in mean ratings are quite small. Nevertheless, they are surprisingly consistent across a number of separate comparisons. It should also be remembered that the items are relatively obscure statements that are repeated only one or two times during a 14 day period. Future research should be directed toward determining whether increases in credibility can be enhanced even further with more personally relevant information and/or a greater number of repetitions.

EXPERIMENT 4

The final experiment investigated another "irrelevant" influence on psychological judgment. Specifically, Srull (in press b) has reported a series of experiments that investigated the role of subjects' mood states on memory and judgment. It was found across a series of studies that affective states at the time of encoding tend to have "assimilation" effects. That is, when print ads were presented to subjects, they evaluated the products more positively when they were in a positive than neutral mood at encoding, and more negatively when they were in a negative than neutral mood at encoding. Subjective mood states at the time of judgment also played an important role but their effects were much more complex and will therefore be controlled in the present investigation (see Srull in press b for details).

Srull concluded that new information is automatically evaluated at the time of information acquisition and one's own subjective mood state influences this evaluation. Positive moods lead to more positive evaluations and negative moods lead to more negative evaluations. Moreover, since this previously evaluated information is later retrieved for use in making judgments, the effects of mood at encoding generally will persist over time.

The present study examines this process in the context of prior familiarity. It is reasonable to assume that high familiarity subjects have a much more developed and. structured basis for evaluating new information. To a-much greater extent, they will know what attributes are important and/or most discriminating and, due to their prior familiarity with the product class, they will already have evaluated many possible attribute values. For example, they may have already developed a rule such as "less than 15 mpg is inefficient and uneconomical." Thus, they may not be subject to the same types of momentary influences that have an impact on the evaluations of low familiarity subjects. This possibility was examined in the present experiment.

Subjects were undergraduate students who came into the laboratory and were put into either a positive, negative or neutral mood using the involved recollection procedure previously described by Srull (in press b). Briefly, subjects enter a quiet, dimly lighted room and are encouraged to completely relax. They are then asked to privately recall everything possible from a previous strongly affectively toned event in their personal life. _very few minutes, subjects are given a "probe" that encourages them to concentrate on every detail concerning what they were thinking and how they felt during the actual experience.

The procedure is similar to those that attempt to induce mood states with hypnotic suggestion (see e.g., Bower :?81) except that subjects are not self-selected and there is no attempt to put them into a true hypnotic trance. Nevertheless, the mood states that result can -be quite intense. In the present study, affective valence was manipulated by the type of event subjects were asked to recall.

Following the mood induction procedure, subjects were shown a single (modified) print ad. It contained ten separate attribute values for the Mazda RX7. The information was adapted from ads actually used by the company. It was changed simply by revising the sentences and putting them into paragraph form.

Subjects read the information at their own pace and were not given any instructions other than to pay attention to the information. When finished, subjects were dismissed and asked to return in 48 hours. At that time, all subjects were put into a neutral mood (to control for the effects of mood state at the time of judgment; see Srull in press b) and asked to evaluate the product without being re-exposed to the original ad. Specifically, subjects were asked, "Assuming you wanted to purchase a product similar to the Mazda RX7, how desirable no you think this particular brand would be?" Subjects made their ratings on a scale ranging from O ("very undesirable") to 20 ("very desirable").

The mean desirability ratings of the product are presented in Table 2 for each experimental condition. Subjects in a neutral mood at the time of both encoding and judgment provide an important baseline for comparison.

There are several things worth noting about these data. First, the ratings of high familiarity subjects did not differ from. their normative baseline as a function of affective state during information acquisition. Thus, whether high familiarity subjects were in a positive, negative, or neutral mood during encoding simply did not matter.

TABLE 2

MEAN DESIRABILITY RATING OF THE PRODUCT AS A FUNCTION OF PRIOR FAMILIARITY AND MOOD STATE AT TIME OF ENCODING

However, subjective mood states had a strong impact on the ratings of- low familiarity subjects. As before, these took the form of assimilation effects in both cases. Thus, subjects in a positive mood at information acquisition rated the product more positively and subjects in a negative mood at the time of information acquisition rated the product more negatively than those in the neutral baseline condition. It is also interesting to note that evaluations of the product did not differ as a function of prior familiarity when neutral moods were induced at the time of information acquisition and at the time of judgment.

In sum, the present data suggest that subjective mood states can have an important impact on product evaluations. However, they also indicate that, at least under the present set of conditions, high familiarity subjects are completely resistant to influences that have a strong effect on the evaluations of low familiarity subjects. Perhaps the most straightforward interpretation of these results is that high familiarity subjects have a strong, pre-existing context for evaluating new information. Thus, momentary influences of mood do not have any appreciable effect. In contrast, the context for evaluating new information in low familiarity subjects is more malleable and therefore more likely to be affected by irrelevant influences such as one's own subjective mood state.

DISCUSSION

It has become increasingly clear over the past several years that the role of prior knowledge is very important in a wide range of cognitive activity. The experiments reported in the present paper add to the growing data base that will ultimately have to be accounted for by any comprehensive theory.

The picture that emerges of the consumer low in product familiarity is one of a person who generally processes each new piece of information independently. The results of Experiment 1 suggest that rote processing predominates. Low familiarity subjects do not recall as much information as high familiarity subjects, and they tend to recall informational items in much the same order in which they were presented. Relative to high familiarity subjects, they are less able to reject attribute values that were not presented but they exhibit fewer old brand-attribute confusions.

It is also becoming increasingly clear that the evaluative judgments of low familiarity subjects will be influenced by a variety of factors that are not informational per se. The present studies demonstrated that more positive product evaluations occur with increased repetitions and when subjects are in a positive mood state at the time of information acquisition. There are almost certain to be other influences on the product evaluations of low familiarity subjects. and an attempt should be made to identify these in future research.

It is interesting that many of the most successful theories of human memory assume that each new piece of information is encoded into memory independently of other information known about the object (see e.g., Anderson and Bower's 1973 HAM model). Although these theories have generated a tremendous amount of empirical support they are most often tested by examining memory for random lists of words or strings of isolated sentences, The processing that is induced by such tasks, of course, will resemble that apparently used by the low familiarity subjects in the present study.

lt is important to note, however, that these models often break down when more active organizational strategies are used by subjects (see e.g., Smith, Adams, & Schorr 1978; Srull 1981; Srull & Brand in press). However, this type of processing is likely to characterize consumers high in product familiarity. As was demonstrated in the present study, the performance of such subjects often differs substantially. Under most circumstances, they will learn new information more quickly and retain it better. Their evaluative judgments of products will also be more immune to the influence of irrelevant factors.

One of the most important issues that needs to be addressed in future research concerns the relationship between memory and judgment processes. This has already proven to be an extremely complicated issue (see Anderson & Hubert 1963; Dreben, Fiske, & Hastie 1979; Riskey 1979). However, Lynch and Srull (1982) have pointed out that the execution of many decision rules requires a particular type of output from the memory system. When the "wrong" type of data are provided, many decision rules either need to be rejected or modified before a decision can be made. This, in many ways, reflects the enormous complexity of consumer information processing, and even this one issue will undoubtedly provide a major challenge to future researchers and theorists alike.

REFERENCES

Anderson, N.H., & Hubert, S. (1963), "Effects of concomitant verbal recall on order effects in personality impression formation," Journal of Verbal Learning and Verbal Behavior, 2, 379-391.

Anderson, J.R., & Bower, G.H. (1973) Human Associative Memory, Washington, D.C.: V.H, Winston.

Bacon, F.T. (1979), "Credibility of repeated statements: Memory for trivia," Journal of Experimental Psychology: Human Learning and Memory, 5, 241-252.

Bettman, J.R., & Park, C, W. (1980), "Effects of prior knowledge and experience and phase of the choice process on consumer decision processes: A protocol analysis," Journal of Consumer Research, 7, 234-248.

Bower, G.H. (1981), "Mood and memory," American Psychologist , 36, 129-148.

Bransford, J.D. (1979), Human Cognition, Belmont, Calif.: Hadsworth.

Burke, R.R. (1982), A Preliminary Model of Consumer Cognition, Unpublished manuscript, University of Florida.

Chase, W.G., & Simon, H.A. (1973), "Perception in chess," Cognitive Psychology, 4, 55-81.

Dreben, E.K., Fiske, S.T., & Hastie, R. (1979), "The independence of evaluative and item information: Impression and recall order effects in behavior-based impression formation," Journal of Personality and Social Psychology, 37, 1758-1768.

Hasher, L., Goldstein, D., & Toppino, T. (1977), "Frequency and the conference of referential validity," Journal of Verbal Learning and Verbal Behavior, 16, 107-112.

Hayes-Roth, B. (1577), "Evolution of cognitive structures and processes," Psychological Review, 84, 260-278.

Johnson, E.J., & Russo, J.E. (1981), "Product familiarity and learning new information," in Advances in Consumer Research, Vol. VIII, Kent B. Monroe, ed., Ann Arbor, MI: Association for Consumer Research, 151-155

Lynch, J.G., & Srull, T.K. (1982), "Memory and attentional factors in consumer choice: Concepts and research methods," Journal of Consumer Research, 9, 18-37.

Pryor, J.B., & Ostrom, T.M. (1981), "The cognitive organization of social information: A converging operations approach," Journal of Personality ann Social Psychology, 41, 628-641.

Riskey, D.R. (1979), "Verbal memory processes in impression formation," Journal of Experimental Psychology: Human Learning and Memory, 5, 271-281.

Roenker, D.L., Thompson, C.P., & Brown, S.C. (1971), "Comparison of measures for the estimation of clustering in free recall," Psychological Bulletin, 76, 45-48.

Smith, E.E., Adams, N., & Schorr, D. (1978), "Fact retrieval and the paradox of interference," Cognitive Psychology, 10, 438-464.

Srull, T.K. (1981), "Person memory: Some tests of associative storage and. retrieval models," Journal of Experimental Psychology: Human Learning and Memory, 7, 440-463.

Srull, T.K. (in press a), "Organizational and retrieval processes in person memory: An examination of processing objectives, presentation format, and the possible role of self-generated retrieval cues," Journal of Personality and Social Psychology.

Srull, T.K. (in press b), "Affect and memory: The impact of affective reactions in advertising on the representation of product information in memory," in Advances in Consumer Research, Vol. IA, Richard Bagozzi & Alice M. Tybout, eds., Ann Arbor, MI: Association for Consumer Research.

Srull, T.K., & Brand, J.F. (in press), "Memory for information about persons: The effect of encoding operations on subsequent retrieval," Journal of Verbal Learning and Verbal Behavior.

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