Journal of Consumer Research

Editors

Darren Dahl (Editor in Chief)Eileen FischerGita JoharVicki Morwitz

3.125
5.003
16 out of 115

When Do Incidental Mood Effects Last? Lay Beliefs versus Actual Effects

Anastasiya Pocheptsova , Nathan Novemsky
992-1001 First published online: 1 April 2010

Abstract

In a series of studies we examine how incidental mood present at the time of an experience affects judgments made long after the mood has dissipated and compare this to lay beliefs about how mood affects memory-based judgments. We find that memory-based judgments are affected by incidental mood only when there is an external prompt to evaluate the stimulus in real time. This is contrasted with lay beliefs about the effects of mood, which are not sensitive to delay or to the presence of real-time evaluations. The mismatch between lay beliefs and actual effects leads consumers to distort previously unbiased memory-based judgments when they are reminded of the source of the incidental mood.

• Affect/Emotions/Mood
• Memory
• Beliefs
• Judgment and Decision Making

Many consumer decisions involve memory-based judgments, where the decision is based on the evaluation of past experiences. These could be consumers’ own past experiences or others’ experiences communicated by word of mouth. For example, when consumers decide whether to come back to the restaurant they visited last week, buy a DVD that their friends saw in a movie theater a couple of months ago, or return to the same vacation destination 1 year later, they have to rely on memories of past experiences.

Those past experiences are frequently affected by temporary contextual factors such as incidental mood (Schwarz and Clore 2007). While many studies have demonstrated the impact of incidental mood on judgments made at the time of experience (real-time judgments), in this research we focus on the effect of incidental mood on memory-based judgments: judgments made after a time delay when the context of the experience is no longer present. For example, suppose that last week you went to a restaurant and consumed a well-prepared meal. Further imagine that you went into the restaurant in either a good or a bad mood perhaps because it was a sunny or rainy day. A week later, would you be more likely to praise the restaurant or return to it yourself if your earlier experience happened on a sunny day? What would happen if you remembered that the earlier visit was on a sunny day?

In the present research, we examine how incidental mood influences judgments made after a delay. We compare actual effects of incidental mood to individuals’ lay theories about these effects. We find that while real-time evaluations and delay between the mood and the judgment influence memory-based judgments, individuals do not intuit these effects. As a result, individuals are likely to adjust their unbiased judgments on the basis of their incorrect beliefs about long-term mood effects.

The Role of Lay Beliefs in Long-Term Effects of Incidental Mood

Many studies have documented the effects of incidental mood on evaluation (e.g., Schwarz and Clore 1983, 2007). According to “affect-as-information” theory, when evaluating an object, individuals ask themselves, “How do I feel about it?” Since this question probes current feelings, any influence on momentary affect can influence evaluations, resulting in judgments that assimilate toward current mood. For example, Schwarz and Clore (1983) found that on a sunny day (in a positive mood) people give higher ratings of life satisfaction than on a rainy day (in a negative mood). Thus mood produced by the weather was incorporated into judgments of overall well-being. Similarly, Isen et al. (1978) found that people who were put in a positive mood by a free gift gave higher evaluations of products that were entirely unrelated to the gift.

In this article we strive to expand our understanding of the effect of incidental mood on judgments by examining the effect of incidental mood on memory-based evaluations. We examine delayed judgments regarding experiences that included incidental mood, where the mood is not present at the time of the judgment. We believe that whether or not the context is recalled at the time of judgment plays a critical role in mood effects in memory-based judgments, so we outline our predictions as a function of whether or not mood is remembered at the time of judgment.

First, consider memory-based judgments in which the context of the earlier experience (e.g., that a past visit to a restaurant occurred on a rainy day) is not recalled at the time of judgment. Building on the affect-as-information framework (Schwarz and Clore 1983, 2007), we expect that if real-time judgments of an experience are formed at a time when incidental mood is present, such evaluations are likely to be influenced by incidental mood. Once a real-time evaluation is made, this evaluation is likely to strongly influence later evaluations of the experience (Novemsky and Ratner 2003; Srull 1983, 1987), and as a result, memory-based judgments are likely to incorporate incidental affect that was present at the time of the experience. However, when making a memory-based judgment of an experience that was not evaluated in real time, one has to rely on episodic memory of the experience. Emotions are not generally stored as a part of episodic memory (Robinson and Clore 2002), so people are unlikely to recall a mood state that is unrelated to the content of the experience. Hence, this incidental mood would not be incorporated into a memory-based judgment. In sum, when context is not recalled during the formation of memory-based judgments, the presence of real-time evaluations will determine whether incidental mood affects memory-based judgments.

The above analysis supposes that the earlier context is not specifically remembered when making a judgment after a delay, but on many occasions the previous context will be salient when making a memory-based judgment. For example, people might remember the rainy weather that canceled their barbecue and caused them to eat at a restaurant. Additionally, as a part of recounting the experience to others, context can be a natural part of the conversation. For example, when telling a friend about a recent restaurant experience, one might be asked about dinner companions and recall that the meal was consumed with good friends. The context itself might also be quite memorable because it was very extreme or unusual.

When the source of an incidental mood is recalled at the time of judgment, people may use their beliefs about the effect of incidental mood when forming a memory-based judgment (Wegener and Petty 1995; Wilson and Brekke 1994). If consumers believe that incidental mood influenced their judgments, they will adjust their evaluations to counteract this bias.

Lay beliefs about past experiences often diverge systematically from the experiences (e.g., Novemsky and Ratner 2003; Ross 1989), and when they do, any adjustment based on these theories may introduce rather than attenuate bias in memory-based judgments. To our knowledge, lay beliefs about the effect of delay on incidental mood effects have not been studied. However, past research does suggest that lay beliefs about long-term mood effects may differ from actual effects. A number of studies suggest that time passage is generally not a salient factor when people are making judgments. For example, Kahneman and colleagues have shown that people are insensitive to the duration of experiences when making summary judgments (Kahneman et al. 1993; Redelmeier and Kahneman 1996; Varey and Kahneman 1992). Other research suggests that people tend to compress time intervals between consumption occasions (Read and Loewenstein 1995). On the basis of this literature, we expect that beliefs about the effect of incidental mood on judgment will not be sensitive to time delay, even when the actual effects of incidental mood do decay over time as memories of past feelings fade. Furthermore, since people may not recognize that memory for their feelings fades, they are unlikely to realize that real-time evaluations are critical for incidental mood to have long-term effects. Hence, people are likely to believe that incidental affect will have a prolonged effect on judgments whether or not there was an evaluation of the experience in real time.

Given this potential discrepancy between lay beliefs and actual long-term mood effects, we expect a reminder about a past mood state to cause a bias in memory-based judgments rather than attenuate it when memory-based judgments are made following an experience that did not include a real-time evaluation. Because memory-based judgments would not generally be affected by incidental mood, any reminder about this mood will cue an erroneous lay belief that the judgment is biased and induce an unnecessary adjustment.

In sum, we predict that long-term effects of incidental mood will depend on whether the context of the experience is recalled at the time of memory-based judgment. We propose that when one does not recall a prior incidental mood, that incidental mood will have a fleeting effect on judgments in the absence of real-time evaluations. However, when a real-time evaluation takes place, the effect of incidental mood becomes a part of the long-term memory of the experience and is present in later memory-based judgments. When the prior mood is salient at the time of judgment, the judgment will be adjusted in accordance with lay beliefs about how mood influenced those judgments. We propose that lay beliefs about incidental mood are not sensitive to delay or to the presence of real-time evaluations, both of which substantially influence actual incidental mood effects. This mismatch between lay beliefs and actual effects may lead to distortions of previously unbiased judgments when one is reminded about incidental mood at the time of experience.

In the next section, we present experimental evidence to support our propositions. In studies 1a and 1b, we examine the effect of incidental positive and negative mood on real-time and memory-based judgments in the presence and absence of external prompts to evaluate an experience in real time. In studies 2a and 2b, we measure people’s beliefs about the effect of incidental mood on real-time and memory-based judgments. Finally, in studies 3a and 3b, we manipulate whether the context of the experience is salient at the time of judgment and examine how memory-based judgments are affected by lay beliefs about incidental mood. We conclude with a discussion of our findings and directions for future research.

Study 1a: Effect of Incidental Negative Mood on Memory-Based Judgments

Study 1a examined the effect of an incidental negative mood at the time of viewing a painting on evaluations of that painting made several days later. We predicted that judgments of participants who were prompted to make real-time evaluations would be affected by incidental mood both at the time of viewing the painting and after a delay. This would happen because when making a memory-based judgment, people will recall their real-time evaluations that were biased by incidental affect. However, in the absence of real-time evaluation, participants will rely on their memory of the features of the painting rather than their past affective state and will be more likely to have context-free memory-based judgments. Therefore, we expect a negative mood to lower memory-based evaluations only when consumers make real-time evaluations. Further, we expect that memory-based judgments of neutral mood participants will not vary as a function of real-time evaluations, since relying on either the features of the painting or overall evaluation should lead to similar judgments.

We choose paintings as our target stimuli because previous research has shown that ambiguous experiences are more likely to be affected by incidental context (Ha and Hoch 1989; Hoch and Ha 1986; Prescott and Young 2002). Art enjoyment is subjective and open to multiple interpretations, which makes it a rather ambiguous experience. Therefore, using paintings as the stimuli in our studies raised the likelihood that we would observe effects of incidental mood at the time of the experience.

Method

One hundred forty-one undergraduate students participated in this study as a part of a longer survey session. Participants were assigned to one of four conditions based on a 2 (negative mood, neutral mood) × 2 (real-time evaluation, no real-time evaluation) design. During the survey session (time 1), participants in the negative mood conditions completed two seemingly unrelated studies: a mood manipulation and an art study. Participants in the neutral mood conditions completed only the art study. The mood manipulation involved reading and answering questions about a story about the inhumane treatment of pregnant horses to produce drugs for humans.

The story was pretested on a separate group of participants ($$mathtex$$ $N=31$ $$mathtex$$ ). In the pretest, half of the participants read the story and then reported their mood using the Positive and Negative Affectivity Schedule (PANAS; Watson, Clark, and Tellegen 1988), and the other half completed only the PANAS. After reading the story, participants reported significantly lower positive affect than the control group (19.69 vs. 26.50; t(29) = 2.71, p < .05) and significantly higher negative affect (17.85 vs. 13.72; t(29) = 2.19, p < .05).

After the mood manipulation, participants saw a reprint of a painting by Cézanne, which depicted a family in a house, labeled Family. Participants in the real-time evaluation conditions were asked how much they would enjoy having the poster of this painting hanging on their wall at home on a 7-point scale from “not at all” to “very much.” Participants in the no real-time evaluation conditions were asked to rate how much they enjoy art (mood did not affect this rating in any of our studies and will not be discussed in further analyses). Five days later (time 2), all participants were contacted via e-mail and asked to rate how much they would enjoy having a poster of the painting Family hanging on their wall at home (without being re-exposed to the painting).

Results and Discussion

In a replication of prior research on immediate mood effects, participants in a negative mood rated the painting lower in real time (M = 1.73) than participants in a neutral mood (M = 2.43; t(70) = 2.34, p < .05). A 2×2 ANOVA on ratings of the painting at time 2 revealed a significant interaction of mood and real-time evaluation (F(1, 137) = 5.21, p < .05; see fig. 1). Simple effects analyses confirmed that the effect of mood persisted at time 2 for those participants who gave real-time ratings (Mnegmood = 1.84, Mneutrmood = 2.7; F(1, 137) = 7.05, p < .01). However, participants who did not make a real-time evaluation showed no effect of mood at time 2 (Mnegmood = 2.63 vs. Mneutrmood = 2.44; F < 1).

Figure 1

Study 1a: the effect of negative mood on memory-based judgments

As expected, we found that the evaluations of the painting assimilated to ambient mood at the time of viewing. A few days later, only those participants who made real-time evaluations had context-dependent judgments. Interestingly, incidental mood had no effect on the judgments of participants who did not make real-time evaluations, suggesting that they retrieved a context-free memory of the painting.

In a separate study we tested the external validity of the results of study 1a by examining whether incidental mood at the time of the experience would affect real consumer choices made after a delay. This study used two critical cells from study 1a: negative mood/real-time evaluation and negative mood/no real-time evaluation. The only difference in procedure from study 1a was that at time 2 participants were asked to choose among various products including a poster of the painting they saw at time 1. Participants were told that we would randomly select one out of every 10 participants to give a product based on their stated choices. As expected, we found that participants chose the poster over other products more often in the no real-time evaluation condition (72%) than in the real-time evaluation condition (55%; t(75) = 2.43, p < .05). These results are consistent with the results of study 1a showing that when participants have an experience in a negative mood, the incidental affect had a prolonged effect on memory-based choices only in the presence of real-time evaluations.

Study 1b: Effect of Incidental Positive Mood on Memory-Based Judgments

Previous research has found significant differences in the way people process information while in a positive versus negative mood. For example, compared to a negative mood, a positive mood results in more global rather than local processing (Gasper and Clore 2002) and higher reliance on scripts (Bless et al. 1996). Therefore, we wanted to extend our findings to the effect of positive mood on memory-based judgments. Study 1b is similar to study 1a except that the mood manipulation used an uplifting story rather than a negative one.

Method

Sixty-seven undergraduate students participated in this study as part of a longer session. Since in study 1a we found that the presence or absence of real-time evaluations did not affect memory-based judgments of participants in the neutral mood condition, in study 1b we ran a three-cell design: neutral mood/real-time evaluation, positive mood/real-time evaluation, and positive mood/no real-time evaluation. The procedure was very similar to study 1a except that study 1b used a positive mood-inducing story and a painting by Monet labeled Haystacks. A pretest confirmed that the story increased positive mood as measured by the PANAS (28.78 vs. 22.56; t(28) = 2.08, p < .05).

Results and Discussion

As expected, participants in a positive mood rated the painting more positively at time 1 (M = 5.69) than participants in a neutral mood (M = 4.19; t(54) = 2.54, p < .05). More important, a one-way ANOVA on time 2 ratings was significant (F(2, 57) = 4.34, p < .05). Planned contrasts show that for participants who made real-time ratings, time 2 judgments also revealed mood effects (Mneutrmood,r-t = 4.25, Mposmood,r-t = 6.0; F(1, 57) = 8.12, p < .01; see fig. 2). Consistent with our predictions, the effect of incidental mood was not present in the memory-based judgments of participants in the positive mood, no real-time evaluation condition (M = 4.58). Their memory-based evaluation was significantly lower than that of those who made an online rating under the influence of mood (M = 6.0; F(1, 57) = 4.45, p < .05) but not significantly different from the neutral mood condition (M = 4.25; F < 1).

Figure 2

Study 1b: the effect of positive mood on memory-based judgments

One possible alternative explanation for the results of studies 1a and 1b is that in the absence of real-time ratings, individuals’ memory for the painting is weak, and hence their ratings tend to gravitate toward some general evaluation of paintings. To rule out this explanation, we ran a separate study in which we compared memory-based ratings of two different paintings while manipulating whether participants made real-time evaluations. The purpose of this study was to show that memory-based evaluations will reflect the differences between the paintings whether or not there are real-time evaluations, indicating that participants hold equally accurate memories of the quality of these paintings regardless of the real-time evaluations.

One hundred and eight undergraduate students were randomly assigned to one of four conditions based on the following design: 2 (real-time rating, no real-time rating) × 2 (painting 1, painting 2). In this study we used two paintings: Family (the same one that we used in study 1a) and a painting called Fields with a country theme similar to that of Haystacks. We substituted Fields for Haystacks because the latter is a fairly famous painting, and we wanted a conservative test for memory of the paintings.

The procedure was similar to that used in the neutral mood conditions of study 1a. Examining real-time ratings, we found that painting 1 was evaluated more positively (M = 5.22) than painting 2 (M = 3.69, diff = 1.53; t(52) = 2.25, p < .05). Consistent with our predictions, an ANOVA of memory-based ratings revealed only a significant main effect of the painting (F(1, 104) = 11.8, p < .01). Simple effects analyses showed that the difference between the paintings remained significant and was nearly identical in the real-time (Mpainting1 = 5.53 vs. Mpainting2 = 4.02, diff = 1.51; F(1, 104) = 5.54, p < .05) and no real-time evaluation conditions (Mpainting1 = 5.01 vs. Mpainting2 = 3.45, diff = 1.56; F(1, 104) = 6.49, p < .05). Therefore, observed effects in studies 1a and 1b are unlikely to be accounted for by weaker memories of the paintings for participants who did not make real-time judgments.

Thus far we have shown that incidental mood at the time of an experience does not generally affect memory-based judgments. We suggested that this happens because the source of the incidental affect is not retrieved spontaneously at the time of judgment. However, we propose that if the source of incidental mood is made salient when making a memory-based judgment, people are likely to adjust their memory-based judgments on the basis of the lay theories they hold about the effect of incidental mood. We believe that these theories are not sensitive to the delay between experienced incidental mood and the memory-based judgment, nor to the presence or absence of real-time evaluations. In studies 2a and 2b we measure lay beliefs directly to test these propositions. First, study 2a was designed to measure participants’ predictions about the effect of incidental mood in studies 1a and 1b. Next, study 2b extends our understanding of lay theories of the effect of incidental mood on memory-based judgment by looking at the basis for the predictions observed in study 2a.

Study 2a: Lay Theories about Long-Term Effects of Incidental Mood

Method

Two hundred fourteen participants from an online subject pool participated in this study. Participants were assigned to one of six conditions based on a 2 (negative mood, positive mood) × 3 (real-time judgment, memory-based judgment preceded by real-time evaluation, memory-based judgment not preceded by real-time evaluation) design. Participants were asked to imagine themselves in the position of participants in study 1a or study 1b. Next, one-third of the participants were asked to imagine making a real-time evaluation of the painting. The rest were asked to imagine making a judgment after a 5-day delay. Half of these participants were told that they had made an evaluation at the time of the experience whereas the other half were told that they had not made a real-time evaluation.

Next, participants were shown the evaluation scale used in studies 1a and 1b. Further, participants were told that in a neutral mood they would have rated the painting as 2 on the 7-point scale, corresponding to the actual mean observed in study 1a. All participants were then asked for the rating they would have given to the same painting provided that at the time of the exposure they read either a positive or a negative mood-inducing story (manipulated between subjects).

Results and Discussion

There are three possible ways in which incidental mood present at the time of an experience can affect memory-based judgment: via assimilation (rating increased [decreased] under the influence of a positive [negative] mood), via contrast (rating decreased [increased] under the influence of a positive [negative] mood), or no effect on judgments. To see which of these three types of beliefs are held by the participants, we compared the proportion of these three beliefs across our six conditions. The majority of participants believed that incidental mood would have an effect on their judgments, and this belief did not differ by the timing of the judgment (real-time = 64.9% vs. memory-based = 63%; χ2(1) < 1). Further, participants did not intuit the role of real-time evaluations in extending the effect of incidental mood and believed that mood would affect memory-based judgments regardless of real-time evaluations (memory-based with no real-time evaluations = 60.4%, memory-based with real-time evaluations = 65.6%; χ2(1) < 1).

Since we did not observe any significant differences in participants’ beliefs between the two memory-based evaluation conditions (with and without real-time evaluations), we collapsed those two cells in the following analysis. We did not find reliable differences between the two mood conditions in the belief that mood would have an effect on either immediate (positive mood real-time judgment = 68.4%, negative mood real-time judgment = 61.1%; χ2(1) < 1) or memory-based judgments (positive mood memory-based judgment = 67.6%, negative mood memory-based judgment = 59.4%; χ2(1) = 1.08, NS). Interestingly, though the majority of participants in the positive mood conditions believed that mood would have an assimilating effect on judgments (65.6%), only one-third of participants in the negative mood conditions had a similar belief (25.8%; χ2(1) = 43.7, p < .001). Significantly more participants believed that a negative mood would have a contrasting effect on judgment (34.1%) than a positive mood (0.8%; χ2(1) = 50.6, p < .001). This difference in beliefs occurred despite the fact that neither a positive nor a negative mood produced any actual contrast effects.

An analysis treating the difference between the rating stated by the participants and what they were given as the neutral mood rating as a continuous measure generated similar results. An ANOVA revealed only a main effect of mood (F(1, 208) = 24.1, p < .001). Participants believed that both positive mood and negative mood would bias judgments upward (Mposmood = 1.55 vs. Mnegmood = .58). Predictions about real-time judgments did not differ from predictions about memory-based judgments for either positive (Mposmood,r-t = 1.61 vs. Mposmood,mem = 1.5; F < 1) or negative mood (Mnegmood,r-t = .68 vs. Mnegmood,mem = .54; F < 1). Finally, varying whether a real-time evaluation was made prior to memory-based judgments did not reliably affect the predicted judgments in the positive mood condition (Mposmood,r-t = 1.64 vs. Mposmood,nor-t = 1.39; F < 1) or the negative mood condition (Mnegmood,r-t = .28 vs. Mnegmood,nor-t = .79; F(1, 208) = 2.34, p > .1).

The results of study 2a show that naïve theories of the effect of incidental mood are not sensitive to a delay between the experienced incidental affect and the judgment. Further, participants did not believe that the presence of real-time evaluations influences long-term mood effects. Interestingly, the present study reveals a belief among many participants that negative mood will increase evaluations, even though we observed assimilation in real-time judgments for both positive and negative mood. We return to this finding later.

Study 2b: Understanding Lay Theories of Long-Term Effects of Incidental Mood

Study 2a provided us with initial evidence that people believe their memory-based judgments are biased by incidental mood in contrast to results of studies 1a and 1b. To gain further insight into participants’ beliefs about long-term mood effects, study 2b uses a design similar to study 2a with the addition of questions to gauge beliefs about assimilation and contrast.

Method

We conducted a pretest asking participants to respond to the same questions as in study 2a and then asked them an open-ended question about why they indicated a belief in contrast, assimilation, or no long-term mood effects. Upon examination of those responses, three themes consistently emerged, and we used those themes to generate three statements for use in study 2b that are described below.

Two hundred and twenty six participants took part in study 2b as part of a course requirement. First, participants were asked to indicate how their rating of a painting would change if at the time of exposure they read either a positive or a negative mood-inducing story as in study 2a. In contrast to study 2a, all participants were told that they were not asked to make immediate judgments of the painting at the time of viewing. Further, in contrast to study 2a, the stated rating of the painting for a neutral mood was 3 on a 1–7 scale. Next, participants were asked to indicate their agreement with three statements that expressed different ways that mood might affect judgments (on a 7-point scale from “strongly disagree” to “strongly agree”). The statements measured agreement with the following: (1) feelings at the time of viewing would be associated with the painting affecting judgments 5 days later, (2) feelings at the time of viewing would spread to the reaction to the poster affecting judgments 5 days later, and (3) feelings arising from the poster would be compared to one’s current mood influencing judgments 5 days later.

Results and Discussion

Consistent with the results of study 2a, we found that the majority of participants (66.8%) think that a positive mood would have a prolonged effect on judgment. Similarly, the majority of participants (58.6%) believe that a negative mood would have a prolonged effect on judgment. As in study 2a, more participants believed that there would be assimilation for a positive mood (66.1%) than for a negative mood (31.5%; χ2(1) = 26.98, p < .001). Again, more participants, in contrast, believed in the negative mood condition (27%) than in the positive mood condition (7%; χ2(1) = 16.27, p < .001). Further, an analysis of the mean difference between the rating stated by the participants and what they were given as the neutral mood rating revealed a similar pattern of results (Mposmood = .89, Mnegmood = .04; t(221) = 6.32, p = .001).

To further explore the underlying mechanism for the observed lay beliefs, we focused only on those participants who believed in either long-term assimilation or long-term contrast. Because of the small size of the group of participants who believed in contrast for a positive mood ($$mathtex$$ $n=8$ $$mathtex$$ ), we focused our analysis on the remaining three groups: positive mood assimilation, negative mood assimilation, and negative mood contrast. We compared participants’ agreement with the three statements related to lasting mood effects. We found that participants who believe in assimilation tend to agree more with the statement that one would associate feelings with the poster 5 days later than participants who believed in contrast (Massimposmood = 4.71 vs. Mcontrnegmood = 3.8; t(136) = 4.0, p < .001; and Massimnegmood = 5.0 vs. Mcontrnegmood = 3.8; t(136) = 4.6, p < .001). Similarly, the assimilation groups tend to agree more with the statement that mood would spread to the poster than participants who believed in contrast (Massimposmood = 4.47 vs. Mcontrnegmood = 3.97; F(1, 138) = 2.23, p = .028; and Massimnegmood = 5.0 vs. Mcontrnegmood = 3.97; F(1, 138) = 3.8, p < .001). However, participants who believed in assimilation were less likely to agree that feelings about the painting would be compared to one’s current mood (Massimposmood = 3.17 vs. Mcontrnegmood = 3.64; F(1, 133) = 1.96, p = .053; and Massimnegmood = 2.97 vs. Mcontrnegmood = 3.64; F(1, 133) = 2.47, p < .05).

These results suggest that there is a similarity in lay beliefs for participants in the positive and negative mood conditions who hold a view that incidental mood has an assimilating effect on judgments. These participants think that incidental mood will spread to the object they are evaluating and will generate mood-congruent associations that affect judgments even after a delay. However, participants who believe in contrast are more likely to think that they would compare their feelings toward the target object with the mood they are experiencing, thus leading to an increase in evaluation with an incidental negative mood. Moreover, they believe that this increase in evaluation due to comparison processes will be reflected in their delayed judgments. These findings suggest that participants are more likely to believe that a positive mood will spread to unrelated objects than a negative mood. Conversely, participants are more likely to believe that a negative mood will be compared to feelings arising from an object than a positive mood.

Across studies 2a and 2b, a substantial fraction of participants believed that a negative mood would produce contrast in their delayed judgments. This finding is particularly interesting considering that there was no contrast in either immediate or delayed judgments. While the fraction of participants believing in contrast was quite similar across these two studies, in study 2a the average belief was reliably in the direction of contrast, whereas in study 2b the average belief was very close to a null effect. These results suggest that average beliefs about a negative mood may vary from situation to situation, but there is substantially more belief in contrast for a negative mood than for a positive mood.

Study 3a: “Correcting” for Perceived Effects of Positive Mood on Memory-Based Judgments

Studies 2a and 2b show that participants believe that their memory-based judgments are biased by incidental mood after a delay even though our results suggest that no such effect exists. Consistent with previous research (Wegener and Petty 1995), we expect that participants will adjust their context-free memory-based judgments when reminded about the context of their experience to correct for the perceived influence of mood. We also predict that participants in the positive mood condition will be more likely to adjust in a direction consistent with assimilation than the participants in the negative mood condition because belief in assimilation is not predominant for a negative mood. Ironically, any adjustment of delayed (context-free) judgments would bias rather than correct them.

In study 3a, we test whether individuals will use their lay beliefs about positive mood to adjust memory-based judgments downward. Since these lay beliefs do not account for the fact that the delay between experience and judgment has already removed the influence of mood, we expect participants to adjust their context-free judgment and introduce a downward bias.

Method

Fifty-six participants were randomly assigned to one of two conditions. We replicated the procedure of study 1b’s no real-time evaluation positive mood condition with one addition. At time 2 (5 days later) before completing the memory-based judgment, half of the participants were reminded that they read a story at time 1. Specifically, they were told: “Last time you completed several studies. The first one asked you to read a story about foals. The second one showed a painting.” The reminder was presented as though it were meant to aid participants’ recall of the survey conducted at time 1.

Results and Discussion

Memory-based ratings in the reminder condition (M = 3.38) were significantly lower than in the no reminder condition (M = 4.46; t(55) = 2.2, p < .05), suggesting that participants corrected for the perceived influence of incidental mood on their memory-based evaluations by adjusting them downward. This correction was consistent with the lay theories measured in studies 2a and 2b. The results of study 1b suggest that participants in the no reminder condition in this study made context-free judgments. Therefore, the reminder seemed to have the ironic effect of inducing participants to bias an otherwise unbiased judgment in an attempt to correct that judgment.

Study 3b: “Correcting” for Perceived Effects of Negative Mood on Memory-Based Judgments

In our next study we further tested this correction mechanism by looking at corrections for negative mood. The results of study 3a suggest that a reminder about the context prompted participants to apply their lay theories. It is also possible, however, that a reminder helped participants better remember the experience and their feelings arising from the mood manipulation. Hence, participants may recall that the painting seemed particularly good after the story and correct for the influence of mood on their evaluations. Since people hold correct lay theories for the real-time effect of a positive mood, we cannot tell whether the results of study 3a are due to better memory of the experience or the use of lay theories. In the case of a negative mood, however, lay theories and actual incidental mood effects differ: actual real-time experiences included assimilation, yet many participants hold lay theories involving contrast. Therefore, testing the effect of a reminder on negative mood can tease apart these two accounts. If participants do not adjust their judgments upward after being reminded of an earlier negative mood, this would strongly suggest that their adjustments are based on lay beliefs rather than on memories of any real-time effects.

Method

One hundred four participants were assigned to one of two conditions. We replicated the procedure of study 1a’s no real-time evaluation negative mood condition with one addition. At time 2 (5 days later) before completing the memory-based judgment, half of the participants were reminded that they read a story at time 1.

Results and Discussion

The reminder caused participants to adjust their ratings downward, such that memory-based evaluations in the reminder condition (3.77) were lower than those in the no reminder condition (4.51; t(102) = 2.09, p < .05). Since participants corrected downward for negative mood, the correction presumes a contrast effect, suggesting that participants’ adjustments are based on lay theories rather than on enhanced memory of past affect.

Studies 1a and 1b showed that incidental mood generally does not affect judgments made after a delay; therefore, participants in studies 3a and 3b should not be correcting for the influence of mood, as they already hold context-free memory of the experience. However, because lay theories do not incorporate the fact that mood effects dissipate over time, participants corrected their delayed judgments. Thus, ironically, by correcting for the influence of mood, they are in fact introducing a context-related bias into their unbiased judgments. This suggests that when the context of an experience is salient while making a memory-based judgment, that judgment may be biased by beliefs about the context, even if the judgment was not biased prior to the activation of those beliefs.

General Discussion

Conclusions and Theoretical Implications

In a series of studies we examined the effect of incidental affect on memory-based judgments. We proposed that lay theories would play an important role when the context of a prior experience is recalled during later judgment. In the first two studies we found that in the absence of a reminder about the context of the experience, incidental mood dissipated following a delay. However, we also found that real-time evaluations extend mood effects into delayed judgments. All observed mood effects were assimilative.

Next, we examined lay theories about the effect of incidental mood on delayed judgments. In two studies we showed that there is a mismatch between lay beliefs and actual effects. Consumers do not intuit the impact of delay on mood effects and believe that their judgments will be equally biased by incidental affect immediately and several days later. Further, they do not intuit the role of real-time evaluations in extending mood effects. They believe that delayed judgments will be equally biased following experiences with and without real-time evaluations. While participants predominantly expected assimilation for positive mood, a substantial portion of participants thought that negative mood would lead to contrast effects. These beliefs are based on a stronger belief that comparisons will occur between current mood and judgment targets for negative mood than for positive mood.

In the final two studies we found that when reminded about the context of an experience, people tend to rely on their lay theories about incidental mood effects and correct for the perceived bias in their memory-based judgments. Because participants falsely believe that mood influences delayed judgments, they adjust these judgments in accordance with these beliefs, thereby introducing rather than attenuating bias in these judgments.

Future Research

Our investigation of lay beliefs about long-term mood effects uncovered marked differences between those beliefs and observed effects. While we consistently observed assimilation in real time and no effect following a delay for both positive and negative mood, studies 2a and 2b revealed that a substantial fraction of participants believed in contrast effects for negative mood. To further investigate when beliefs will shift between assimilation and contrast, we ran a study similar to study 2a, but varying the quality of the painting. Participants were told that the rating of the painting in a neutral mood was either bad (3 out of 7), average (4), or good (5). The results showed that for bad and average paintings, participants reliably predicted that being in a negative mood would produce a contrast effect (Mbad = .97; t(29) = 2.96, p = .006; Maverage = .55; t(46) = 2.92, p = .005 tested against zero). However, for more pleasant paintings, participants predicted assimilation (Mgood = −.59; t(21) = 2.2, p = .056). For a good mood, assimilation was expected regardless of the quality of the target (Mbad = 1.58; t(42) = 10.83, p < .001; Maverage= .76; t(38) = 4.09, p < .001; Mgood = .23; t(34) = 1.06, NS), although this belief was less strong as the painting got better. These results suggest that beliefs may shift between assimilation and contrast depending on the match between the valence of the ambient affect and the judgment target. When the valences are similar, there seems to be more of a tendency toward contrast and, as they become more different, a greater tendency toward assimilation. Future research could further explore this trend and more generally examine the drivers of lay beliefs in contrast and assimilation effects over time to better understand when lay theories will diverge from actual effects, in both direction and duration.

An important issue that has not been explored in the present research is whether immediate judgments have to be explicit to carry over the effect of context to memory-based judgments. If you see a piece of art and think to yourself, “nice painting,” is this the same as explicitly reporting to someone else, “I like this painting!”? An interesting question for future research is exactly what types of real-time evaluations are necessary to influence later evaluations.