Affect and Memory: the Impact of Affective Reactions in Advertising on the Representation of Product Information in Memory

Thomas K. Srull, University of Illinois
ABSTRACT - The present paper serves as a progress report of a current program of research investigating the role of subjective mood states on the processing of information presented in advertisements. Across a series of studies assimilation effects in product evaluations were found for mood states at the time of information acquisition. Independent of these effects, however, there were also contrast effects on judgment when the mood state at the time of retrieval was inconsistent with the major evaluative implications of the stimulus information. A preliminary model is outlined that suggests that while the former effects are due to an alteration of the encoding process, the latter effects are due to a change in retrieval strategies.
[ to cite ]:
Thomas K. Srull (1983) ,"Affect and Memory: the Impact of Affective Reactions in Advertising on the Representation of Product Information in Memory", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 520-525.

Advances in Consumer Research Volume 10, 1983      Pages 520-525

AFFECT AND MEMORY: THE IMPACT OF AFFECTIVE REACTIONS IN ADVERTISING ON THE REPRESENTATION OF PRODUCT INFORMATION IN MEMORY

Thomas K. Srull, University of Illinois

ABSTRACT -

The present paper serves as a progress report of a current program of research investigating the role of subjective mood states on the processing of information presented in advertisements. Across a series of studies assimilation effects in product evaluations were found for mood states at the time of information acquisition. Independent of these effects, however, there were also contrast effects on judgment when the mood state at the time of retrieval was inconsistent with the major evaluative implications of the stimulus information. A preliminary model is outlined that suggests that while the former effects are due to an alteration of the encoding process, the latter effects are due to a change in retrieval strategies.

INTRODUCTION

The present paper will address three extremely general, but highly interrelated, issues in the study of advertising and its effect on consumer memory and judgment. The first is related to what has become one of the most vexing problems in advertising and consumer behavior. This concerns the way in which properties of stimulus information (such as that presented in an advertisement) interact with characteristics of the consumer that are independent of the stimulus per se during information processing activities.

The second issue has to do with the relationship between the cognitive and affective systems. The suggestion that cognition and affect represent two major classes of determinants of behavior has been made by philosophers for centuries, and the same general assumption has become part of contemporary psychology and consumer behavior as well. It is clear, however, that these two systems interact with one another almost continually and they are probably best conceived as two distinct, but highly interdependent, systems.

The final issue addressed is closely related. It concerns the relationship between memory and (evaluative) judgment. One might intuitively assume that the judgments that are made at any given time are a direct function of the implications of the information that is recalled at that time. In fact, however, the few studies that have explicitly examined the two have found only a very weak relationship (see e.g., Anderson & Farkas 1973; Anderson & Hubert 1963, Dreben, Fiske, & Hastie 1979; Riskey 1979). The experiments reported in the present paper will add to this body of literature. However, they will also provide evidence that memory and Judgment are often affected by separate variables, or by the same variable in different says. Thus, the lack of any systematic relationship between the two becomes somewhat less of a mystery.

The experiments reported are specifically concerned with the effects of subjective mood during information acquisition, retrieval, and Judgment. One's own affective state or "mood" can vary along a number of dimensions. Two of the most prominent of these are valence (positive vs. negative' and intensity (weak vs. strong). Although both of these can be represented along an underlying continuum, it is convenient, at least in the initial stages of investigation, to talk about the two extremes.

The experiments reported in the present paper examine the effects of valence and intensity on both memory and judgment processes. Moreover, since it became clear very early in the investigation that the effects of both valence and intensity are partly a function of properties of the stimulus, characteristics of the stimulus information were varied as well.

MOOD INDUCTION PROCEDURES

Since the approach used in the present investigation is somewhat novel, and the methodology remains fairly constant across studies, it is useful to begin with some background. Three separate procedures for manipulating both the valence and intensity of the subjects' mood states have been used. At one extreme is the procedure introduced by Velten (19689. He has developed a series of self-referent statements that are designed to put subjects in either a positive or negative mood by having them read each one aloud. Valence is manipulated by the set of statements given to a subject to read and intensity is varied by the ratio of "critical" statements to i "filler" or neutral items provided. Although this is a fairly simple procedure, researchers in social (Snyder & White 1982), clinical (Natale 1977), and cognitive (Leight & Ellis 1981) psychology have had a surprising amount of success with it.

The major advantage of the Velten procedure is that it t is easily administered and it can be used with virtually all subject populations of interest. However, there are at least three disadvantages associated with such an approach. The first is that mood state is completely con, founded with the semantic content of the statements. The extent to which this is problematic in any given research domain is an empirical question, but the potential interpretive difficulty that results from such a procedure must be acknowledged. The second disadvantage of the procedure is that it seems high susceptible to experimenter demand effects. While some (e.g, Polivy & Doyle 1980) have argued that the potential for demand is severe, others (e.g., Snyder & White 1982) have recorded effects that do not appear to be consistent with a demand interpretation. The final disadvantage, at least in our experience, is that the empirical effects obtained with-such a procedure are quite small. Thus, it does not appear to be an ideal procedure for investigating small, but theoretically important, differences in mood states.

At the other extreme of experimental procedures that might be used to manipulate mood states is hypnotism. With this procedure, subjects are hypnotized and instructed to remember in great detail and "live through" ar. affectively toned event from their past life. Valence is manipulated simply by the nature of the event recalled. Intensity is also easily manipulated through instructions that determine the length of time and amount of detail associated with the recalled episode. The major advantage of hypnotism is that the mood states elicited can be extremely intense. Moreover, they have the appearance (both to the subject and the experimenter) of being completely real and, at the empirical level, they tend to produce quite large effects. Although the semantic content associated with the various recollections is uncontrolled, it is at least different for each subject. Thus, semantic content is not systematically confounded with mood state, as it is with the Velten procedure.

The major disadvantage of the procedure is that not all subjects are able to enter a deep hypnotic trance . In fact, only about 20% of the population is highly hypnotically susceptible with little or no training. Thus, when using hypnotism, there is always a subject selection factor operating that must be considered.

There is another mood induction technique that can be used. This procedure essentially uses the same method as that described above but subjects are not self-selected and there is no attempt to put them into a hypnotic trance. Subjects simply come into a quiet, dimly lighted room and privately recall everything possible from a past affectively toned event in their personal life. he have found it helpful to provide subjects with a "probe" every few minutes that encourages them to concentrate on every detail possible concerning what they were thinking and how they felt. Once again, valence is manipulated by.the event chosen and intensity is varied by the length of time and amount of detail involved in the recollection. In many ways, this approach combines the advantages from both procedures noted above. On the one hand it is applicable to all subjects and relatively easy to administer. On the other hand, although not quite comparable to the hypnotism procedure, the mood states are often quite intense and one typically observes fairly large effects on memory and judgment. Thus, this has become our preferred experimental procedure for manipulating subjects' mood states. For the sake of brevity and consistency, only studies using this procedure will be reported below

TYPES OF STIMULI INVESTIGATED

There were also three separate types of stimulus materials used in the present investigation. Initially, American and European magazines were searched for print ads that: (a) were as complex as possible in the sense that they contained a large amount of attribute information, and (b) pertained to products with which subjects were likely to have little prior familiarity. For example, one ad was five paragraphs long and described a computer-based instrument that could translate English prose into five separate languages with correct syntactic structure. Not surprisingly, these ads generally portray the product in a very favorable light and they will be referred to as positive Ads in the present investigation.

The actual print ads were then modified in two ways. First, the actual attribute information was kept constant but all other "extraneous" comments were eliminated or modified to create a more neutral context. These are subsequently referred to as prescriptive Ads. Second, negativeLy evaluated attributes were embedded in the co-y of the original ad to create a more negatively evaluated product. These are subsequently referred to as Negative Ads.

EXPERIMENT 1: THE EFFECTS OF VALENCE AND INTENSITY ON MEMORY FOR SINGLE ADS

The first study was simply designed to examine the effects of mood valence and intensity, if any, on memory for single advertisements. In particular, the study was designed to determine whether two effects from the verbal learning literature generalize to these types of materials. there is a substantial amount of evidence to suggest that intensity regardless of valence results in -tetter memory (see e.g., Dutta & Kanungo 1975). This possibility was examined. The possible interaction between mood valence and type of ad was also of interest.

There is some evidence to suggest that internal affective states can serve as retrieval cues (see e.g., Bower 1981). If these findings generalize, the Positive Ads should be better recalled by subjects in a positive mood while the Negative Ads should be better recalled by those in a negative mood.

Subjects were undergraduate psychology students who were run in individual sessions. Positive or negative moods were induced using the recollection procedure described earlier, and intensity was also varied. After being placed into a particular mood , subjects read either two Descriptive, two Positive, or two Negative Ads with no instructions other than to pay attention to the information. Following this, they were given ten five-letter anagrams and told to solve as many as possible within one minute. At that point, they were asked to recall as much of the at information in as much detail as possible.

Since the specific content of the three ads varied slightly, and the-major comparisons of interest were ma-e within the various ad conditions, the study was considered for purposes of analysis as three separate 2 (positive vs. negative mood) X 2 (moderate vs. intense mood state, factorial designs.

The dependent variable was the total number of distinct "idea units" (or propositional statements, recalled from the two ads. The mean number of idea units recalled in each condition are displayed in Table 1. There are several things to note about these data. First, there is reasonably strong support for the "intensity hypothesis" described by Dutta and Kanungo (1975). That is, more intense moods, regardless of valence, produced better overall levels of recall for each of the ads. Second, valence of mood and type of ad interacted in a guide unexpected way. Specifically, the Positive Ads were better recalled by subjects in a negative mood, while the Negative Ads were better recalled by subjects in a positive mood. These differences were equivalent across both levels of intensity.

TABLE 1

MEAN NUMBER OE IDEA UNITS RECALLED AS A FUNCTION OF TYPE OF AD AND VALENCE AND INTENSITY OF MOOD STATE

One interpretation of these data is that more intense mood states are more arousing, leading to better overall levels of recall (even in the Descriptive Aa condition. In addition, however, the arousal may also lead to a focusing of attention, particularly on mood-incongruent information (cf. Easterbrook 1959). These differences in recall are important and will be returned to shortly, as they are particularly relevant to the judgment data presented later in the paper.

EXPERIMENT 2: EFFECTS OF VALENCE AND INTENSITY ON EVALUATIONS BASED ON SINGLE ADS

The second experiment used a similar procedure with several minor variations. First, a "neutral mood" condition was included to provide a baseline for evaluating the judgments. Second, after reading through the two ads and completing the anagram task, subjects were asked 'Assuming you wanted to purchase a product similar to the , how desirable do you think this particular brand would be?" Subjects made one rating for each of the two products on a scale ranging from O ("very undesirable"> to 20 ("very desirable").

Again, since content varied slightly and the major comparisons of interest were within the various ad conditions, the study was analyzed by three separate 3 (mood valence) X 2 (mood intensity) analyses of variance.

The mean ratings of each product in each condition are presented in Table 2. Several things are worth noting about these data. First, when the Descriptive Ads are used as stimuli, there do not seem to be any consistent mood effects. However, there are reliable effects in response to the Positive and Negative Ads. As one can see, the Positive Ads are evaluated more favorably by those in a positive than in a neutral or negative mood, and this is true in both the moderate and high intensity conditions. Tn contrast, the Negative Ads are evaluated less favorably by those in a negative than in a neutral or positive mood, and this is also true in both the moderate and high intensity conditions.

TABLE 2

MEAN DESIRABILITY RATINGS OF EACH PRODUCT AS A FUNCTION OF VALENCE AND INTENSITY OF MOOD STATE

In sum, judgments are shifted from their normative baseline only when the subject's mood state is consistent with the major evaluative implications of the ad. Thus, there is no difference between subjects in a negative and neutral mood when a positive Ad is presented. Similarly, there is no difference between subjects in a neutral and positive mood when a Negative Ad is presented. Intuitively, one would expect the mood states to have parallel effects on recall and judgments. However, this clearly was not the case across the first two experiments. In particular, the "mood congruence" effect in the present study contrasts sharply with the "mood incongruence" effect in the first experiment. Possible reasons for this discrepancy will be addressed shortly.

EXPERIMENT 3: RELATIONSHIP BETWEEN MOOD VALENCE AT TIME OF ENCODING AND TIME OF JUDGMENT

The results of the second experiment are interesting but the effects of mood valence at the time of encoding can not be separated from the effects of mood valance at the time of judgment. The third study allows these effects to be empirically isolated. Subjects came into the laboratory and were put into either a positive, neutral, or negative mood. They then read either a positive Ad or a Negative Ad at their own pace, with no instructions other than to comprehend the information. All subjects were then dismissed and asked to return in 24 hours. At that time, they were again put into either a positive, neutral, or negative mood and given the judgment task without being re-exposed to the original ac.

The mean ratings of the Positive Ad are presented in Table 3. There appear to be two independent types of effects. First, mood valence at the time of encoding appears to have "assimilation" type effects. Thus, subjects in a positive mood at the time of encoding rate the product more favorably than those in the neutral condition, and those in a negative mood at the time of encoding rate the product less favorably than those in the neutral condition. Independent of this, however, there also appears to be "contrast" effects for those subjects in a negative mood at the time of judgment. Considering only those subjects in a positive moo. at the time of encoding, for example, those subjects in a negative mood at the time of judgment give more favorable ratings to the product (M = 14.48) than those in a positive (M = 14.27) or neutral (M = 13.98) mood.

TABLE 3

MEAN DESIRABILITY RATINGS OF EACH PRODUCT AS A FUNCTION OF MOOD AT THE TIME OF ENCODING AND MOOD AT THE TIME OF JUDGMENT

This "weighting scheme" for interpreting the judgments predicts the rank order of the means observed perfectly.

The mean ratings for the Negative Ad are also presented in Table 3. Again, there appear to be assimilation effects at the time of encoding. In this case, however, there appear to ne contrast effects for subJects in a positive mood at the time of judgment. The appropriate weighting scheme predicts the rank ordering of the means with only one exception. That is, according to this scheme, subjects in a negative mood at encoding and a positive mood at judgment (M = 7.23) should give less favorable ratings than subjects in a negative mood at encoding and a neutral mood at judgment (M = 7.13). With this one exception, however, the order of the means is predicted perfectly.

In sum, there appear to be assimilation effects at encoding and contrast effects only for mood incongruent events at the time of judgment. Since this scheme was not predicted by any a priori theoretical framework, however, two important questions are raised. The first concerns generalizability. It is noteworthy in this respect that, despite the large number of groups involved, the present scheme accounts for every single difference in ratings across the two conditions except one. Perhaps more impressive, the present scheme can also be used to reinterpret the results of Experiment 2. Since the time interval between encoding and judgment was only one minute, it is reasonable to assume that subjects in the second experiment were in the same mood at both encoding and judgment. Table 2 indicates that, when a Positive Ad was rated, subjects in a positive mood rated the product more favorably than those in a neutral mood. Theoretically, this is due to the assimilation effect at the time of encoding. On the other hand, there was no difference between those in a neutral and Negative mood. However, there should be both an assimilation and contrast effect for subjects in a negative mood. These effects should cancel one another out, and nc difference between the negative mood and neutral mood groups would be predicted. It is also worth noting that these differences generalize across both the moderate and high intensity conditions.

When the Negative Ad is used, there should be assimilation and contrast effects for subjects in the positive mood condition. Thus, there should be no difference between these subjects and those in the neutral mood. However, since there should only be assimilation effects for those in the negative mood, the ratings of these subjects should be depressed. This was, in fact, the case and these differences also generalize across both the moderate and high intensity conditions. Interestingly, this scheme is also consistent with the pattern of means in the Descriptive Ad condition, although the differences did not reach statistical significance ( see Table 2).

Perhaps the more important issue that is raised concerns the determinants of such contrast effects. That is, what type of mechanism is responsible for such mood-incongruent contrast effects at the time of judgment? One clue is perhaps provided by the results of Experiment 1. The data presented in Table I clearly indicate that mood-incongruent ads are much better recalled than mood-congruent ads. Again, however, the question arises as to whether mood at the time of encoding or mood at the time of judgment is the critical determinant of these effects. This is the focus of the final experiment

EXPERIMENT 4: EFFECTS OF MOOD VALENCE AT THE TIME OF ENCODING AND RETRIEVAL ON MEMORY

The final experiment was designed to empirically isolate the effects of mood at encoding from the effects of mood at time of test on retrieval processes. A large number of attribute values for new automobiles and stereo systems were scaled by subjects from the same population, and a long Consumer Reports type of passage was constructed for each. Each passage contained twenty separate attributes that ranged from extremely negative to extremely positive.

A separate group of subjects were brought into the laboratory and put into either a positive, negative, or neutral mood. They then read each passage (counter-balanced for order across subjects) at their own pace with no instructions other than to comprehend the information. At that time they were dismissed and asked to return 24 hours later. Upon returning, they were put into either a positive, neutral, or negative mood. They were then asked to recall as much of the information from the two passages in as much detail as possible.

Following the recall test subjects were given five minutes to list as many of the United States as possible. They were then given a recognition confidence rating task with the original 40 items and Lo lures. Each item was rated along the following scale: (1> positive the item was not presented, (2) fairly certain the item was not presented, (3) undecided, but think the item was not presented, (4) undecided, but think the item was presented, (5) fairly certain the item was presented, and (6) positive the item was presented.

The mean number of distinct "idea units" recalled are presented in Table 4. One can see that subjects in a neutral mood at the time of retrieval recalled fewer items than those in either a positive or negative mood at retrieval. However, there was no difference between these latter two groups . Similarly, there do not seem to be any systematic effects due to mood valence at the time of encoding.

TABLE 4

MEAN NUMBER OF ITEMS RECALLED, MEAN RECOGNITION CONFIDENCE RATINGS, AND CORRELATIONS BETWEEN RETRIEVAL RATIOS AND FAVORABILITY RATINGS OF INDIVIDUAL ITEMS

The mean recognition confidence ratings are also presented in Table 4. The rating were uniformly quite high and there were no reliable differences due to mood at the time of encoding, mood at the time of retrieval, or their interaction.

The most important results are related to a comparison of the recall and recognition responses. In virtually every instance in which a subject recalled an item, he/ she gave it a rating of 6 on the recognition task. However, subjects also gave recognition confidence ratings of 6 to many items they did not recall. Following the conceptual work of Tulving and Pearlstone (1966) on the distinction between "availability" and "accessibility" and the empirical investigation of it by Brewer (Brewer & Dupree in press; Brewer & Treyens 1981), the accessibility of information in long-term memory was examined by computing a "retrieval ratio" for each item. These retrieval ratios were determined by computing the number of times an item was recalled divided by the number of times it received a recognition confidence rating of c. A high retrieval ratio (approximating 1.0) reflects the fact that most subjects who gave a rating of 6 to an item also recalled it (high accessibility as well as high availability in the terms of Tulving and Pearlstone). A low retrieval ratio (approximating 0.0) reflects the fact that most subjects who gave a rating of 6 to an item were unable to recall it (high availability but low accessibility in the terms of Tulving and Pearlstone). A correlation was then computed between these retrieval ratios and the previously scaled favorability ratings of the items in each condition of the experiment. These correlations are presented in the right-hand column of Table La.

The pattern of correlations displayed in Table 4 indicates that the ability to retrieve attribute information is strongly affected by the subject's mood state at the time of retrieval. Consistent with the results of Experiment 1, subjects in a positive mood at the time of retrieval are much more able to retrieve negative than positive attributes. In contrast, positive attribute valueS are more likely to be retrieved by subjects in a negative than positive mood at the time of retrieval. However, no comparable differences were observed for mood state at the time of encoding. Finally, it is worth noting that all of these differences occur despite the fact that subjects have equally good recognition memory for positive and negative attributes in all conditions.

DISCUSSION

One general model of consumer memory and judgment suggests that when subjects acquire new attribute information, they simply encode it into memory. At some later time, when a specific Judgment is required, that information (or some subset of it) is retrieved and evaluated at that time. The data reported in the present paper would appear to be inconsistent with such a model. In particular, the strong and consistent assimilation effects of mood at the time of encoding are problematic for this type of theoretical framework.

An alternative approach assumes that evaluation is a natural component of the comprehension process. Thus, even in the absence of explicit instructions to do so, subjects will naturally evaluate new information at the time of input. When a specific judgment is required at some later time, according to this model, subjects will retrieve the previously evaluated information to use as a basis for making their judgments. The present data would appear to be more consistent with this type of theoretical model.

The results reported suggest that initial evaluations are strongly influenced by subjective mood states at the time of information acquisition. In every single case, assimilation effects were observed. Thus, positive mood states led to more favorable evaluations and negative mood states led to less favorable evaluations than those associated with a neutral control group. In addition, however, there are noteworthy interactions between mood state at the time of retrieval and type of information provided. These consistently took the form of "contrast" effects when the mood state at the time of retrieval was inconsistent with the major evaluative implications of the information presented. It appears that these effects are due to a tendency to recall more items inconsistent than consistent with the mood at the time of retrieval.

Perhaps the best explanation for why these latter effects may occur can be made in terms of cue-overload theory (Watkins 1979; Watkins & Watkins 19763. Consider the case in which subjects are in a positive mood at the time of judgment for example. Although the positive mood may serve as a cue for the positive attributes, it will also be associated with nearly ail other positive events in the person's life (cf. Bower 1981). It is well known that the more things that are associated with a single cue, the less effective that cue becomes (see e.g., Mueller & Watkins 1977; Watkins 1975; Watkins 1979; Watkins & Watkins 19759. In contrast to the positive attributes, however, the negative attributes will remain highly distinctive. Thus, all else being equal, cue overload theory would predict that a greater proportion of negative than positive items should be recalled under these conditions. This is exactly what occurred, and the corresponding "contrast" effects on judgments were also observed.

There are two situations in which such effects might be eliminated or even reversed, and they are worthy of close attention in future research. First, a highly vivid and distinctive context in which the ads are originally presented may eliminate the tendency of subjects to rely on their internal affective states at the time of judgment as retrieval cues. It is important, in this regard, to recall that the "ads" in the present studies consisted simply of lines of printed text. Moreover, the mood induction procedure itself was certainly the most distinctive aspect of the experimental setting.

There is also some reason to believe that the effects observed may be eliminated or even reversed under conditions of long delay. After a long period of delay, all of the attribute information will be very difficult to retrieve. Under these conditions, overall levels of recall are likely to be quite low and the advantage of one's own affective state in reducing the search set may offset the fact that mood-consistent items are non-distinctive. Evidence of such a process in a slightly different domain has recently been reported by Graesser, Woll, Kowalski, and Smith (19809. Both of these possibilities should be examined in future research. More gene-ally, it is important for researchers to begin to establish a rich descriptive data base concerning the ways in which one's own affective state influences various information processing activities. Until that time, our understanding of the psychological processes involved in consumer memory and judgement will remain quite limited.

REFERENCES

Anderson, N.H. , & Farkas, A.J. (1973), "New light on order effects in attitude change," Journal of Personality and Social Psychology, 28, 88-93.

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.

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

Brewer, W.F., & Dupree, D.A. (in press,, "Use of plan schemata in the recall and recognition of goal-directed actions," Journal of Experimental Psychology: Learning, Memory, and Cognition.

Brewer, W.F., & Treyens, J.C. (1981), "Role of schemata in memory for places," Cognitive Psychology, 13, 27-230.

Dreben, E.e.., 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.

Dutta, S., & Kanungo, R.N. (1975), Affect and Memory, Oxford: Pergamon Press.

Easterbrook, J.A. (1959), "The effect of emotion on cue utilization and the organization of behavior," Psychological Review, 66, 183-201.

Graesser, A.C., Woll, S.B., Kowalski, D.J., & Smith, D.A. (1980), "Memory for typical and atypical actions in scripted activities," Journal of Experimental Psychology Human Learning and Memory, 6, 503-515.

Leight, K.A., & Ellis, H.C. (1981), "Emotional mood states, strategies, and state-dependency in memory," Journal of Verbal Learning and Verbal Behavior, 20, 251-266.

Mueller, C W., & Watkins, M.J. (1977), "Inhibition from part-set cuing A cue overload interpretation," Journal of Verbal Learning and Verbal Behavior, 16, 699-709.

Natale, M. (1977), "Effects of induced elation-depression on speech in the initial interview," Journal of Consulting and Clinical Psychology, 45, 45-52.

Polivy, J., & Doyle, C. (1980), "Laboratory induction of mood states through the reading of self-referent mood statements: Affective changes or demand characteristics?" Journal of Abnormal Psychology, 89, 286-290.

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

Snyder, M., & White, P. (1982), "Moods and memories: Elation, depression, and the remembering of the events of one's life," Journal of Personality, 50, 149-167.

Tulving, E., & Pearlstone, Z. (1966), "Availability versus accessibility of information in memory for words," Journal of Verbal Learning and Verbal Behavior, 5, 381-391.

Velten, E.C. (1968), "A laboratory task for induction of mood states," Behavior Research and Therapy, 6, 473-482.

Watkins, M.J. (1975), "Inhibition in recall with extra-list 'cues'," Journal of Verbal Learning and Verbal Behavior, 14, 294-303.

Watkins, M.J. (1979), "Engrams as cuegrams and forgetting as cue overload: A cueing approach to the structure of memory," in Memory Organization and Structure, C. Richard Puff, ed., New York: Academic Press.

Watkins, M.J, & Watkins, O.C. (1976), "Cue-overload theory and the method of interpolated attributes," Bulletin of the Psychonomic Society, 7, 289-291.

Watkins, O.C., & Watkins, M. J. (1975), "Buildup of proactive inhibition as a cue-overload effect," Journal of Experimental Psychology: Human Learning and Memory, 1, 442-452.

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