Memories affect mood: Evidence from covert experimental assignment to positive, neutral, and negative memory recall

Abstract

Memory recall has been proposed as a common and effective mood regulation strategy. Although several studies have presented results suggesting that recalling valenced memories affects subsequent mood, their designs allow for alternative interpretations of the observed effects. Two such alternatives include the reverse effect (mood effects on memory due to non-experimental assignment to memory recall condition) and demand characteristics of the experiment. We used covert experimental assignment to memory condition, asking subjects (N = 314; 56% female) to recall memories that were primarily positive, neutral, or negative. Results showed the expected effect on mood (p < .002), with reported mood worst in the negative memory condition, better in the neutral condition, and best in the positive condition. These results suggest that valenced memory recall does indeed exert an effect on mood, and may do so even without the individual’s awareness.

Introduction

One of the fundamental assumptions about the relation between thought and feeling is that memories and mood affect one another. Generally the effects are assumed to be reciprocal: An individual’s mood influences the valence of memories that the person recalls (see Bower, 1981, Eich and Metcalfe, 1989), and thinking about memories with a certain valence can affect a person’s mood. Numerous studies have demonstrated the former effect—that a person’s mood can influence the types of memories that the person recalls (e.g., Parrott and Sabini, 1990, Singer and Salovey, 1988).

The question of the reverse effect—that the memories a person chooses to think about can affect his or her subsequent mood—is an important one. Memory recall has been proposed as a common and effective (conscious or unconscious) mood-regulation strategy (e.g., Forgas, 2000, Joormann and Siemer, 2004, Josephson et al., 1996, McFarland and Buehler, 1997, Rusting and DeHart, 2000). One of the advantages of using memories to regulate one’s mood is that this strategy can be used in virtually any setting, which is not the case with many mood regulation behaviors that individuals commonly report using (e.g., going shopping, eating something, having sex; Thayer, Newman, & McClain, 1994). It is important, therefore, to determine whether a person can influence his or her mood by recalling specifically valenced memories.

The hypothesized effect of memory on mood is intuitively plausible; most everyone has experienced an improvement in mood after recalling a happy memory, or has felt sadness when recalling a memory involving hurt or loss. However, other relationships between memory and mood are possible. One possibility is simply the null hypothesis—that the valence of recalled memories has no effect on subsequent mood. Under this scenario, perceived effects of memories on mood might be nothing more than “hand waving”, with motivated attempts to improve mood being the real driver of the mood improvement, while the happy memories are simply epiphenomenal. Another possibility that is intuitively plausible is that memories actually affect moods in an incongruent direction, due to a comparison effect. For example, an individual who remembers the most recent time he or she had the flu might feel relieved that now he or she is well, and would experience an improvement in mood. Similar effects are plausible given the recall of happier times (e.g., the person’s wedding day), which might by comparison lead to a decrement in mood, since the person perceives him- or herself to be in a worse situation at present (e.g., experiencing marital conflict).

Distinguishing between these various hypotheses requires demonstrating that valenced memory recall causally influences moods, which is not an easy task. Simply ordering the collection of data in a logical way—by asking participants to report a memory and then having them report their mood—is not sufficient. If there is a correlation between memory valence and reported mood, it could be the case that the mood was primary and was responsible for the valence of the reported memory. This problem requires that participants be randomly assigned to recall positively- or negatively-valenced memories. In this way, prior mood cannot account for an observed association between the valence of recalled memory and the participant’s subsequent mood.

However, random assignment alone is not sufficient. If participants are aware that an experimental manipulation is intended to change their mood, it is impossible to rule out the effect of demand characteristics on reported mood (see, e.g., Buchwald et al., 1981, Polivy and Doyle, 1980). If participants are told, for example, to “think of a very happy memory”, it is likely that their subsequently reported mood will be contaminated with their perception that the experimenter is expecting them to report a positive mood. Similarly, the use of explicit mood induction procedures in many studies prior to asking participants to recall potentially mood-altering memories (e.g., Erber and Erber, 1994, Joormann and Siemer, 2004, Rusting and DeHart, 2000) can alert the participants that their moods are “supposed” to change in response to the experimental stimuli (see a meta-analysis by Westermann, Kordelia, Stahl, & Hesse, 1996, for a discussion of the influence of demand characteristics on the effectiveness of mood induction procedures). Such demand characteristics could lead to an erroneous rejection of the null hypothesis that recalled memory valence does not affect mood. Therefore, in addition to random assignment to memory condition, participants should not be made aware that the memories they are recalling have a specific, mood-relevant valence.

A large body of research is based on the assumption that recalling positively valenced memories is a strategy for regulating negative moods. For example, Forgas’s (2000) dual-process theory of mood regulation holds that individuals in negative moods are motivated to recall mood-incongruent information in order to maintain affective homeostasis; Boden and Baumeister (1997) suggested that the tendency by some individuals to recall more positive memories after watching an unpleasant film was an effort to regulate their moods; and Parrott and Sabini (1990) interpreted mood-incongruent memory recall as an attempt to regulate mood (see also Forgas and Ciarrochi, 2002, Joormann and Siemer, 2004, McFarland and Buehler, 1997, Parkinson and Totterdell, 1999, Thayer et al., 1994). Although many studies have reported effects of recalled memories on mood (e.g., Josephson et al., 1996, Rusting and DeHart, 2000, Setliff and Marmurek, 2002), their designs do not permit strong conclusions to be drawn about the direction of the effect. These design features represent not an insignificant methodological nuance, but rather a significant gap in the mood and memory literature.

The present study provides a more definitive answer to the question of whether recalling memories can influence individuals’ moods because to our knowledge it is the first study to satisfy both of the conditions outlined above: random assignment to memory valence, with participants blind to their assignment. These conditions were accomplished by asking participants to report memories that, unbeknownst to the participants, were either primarily negative, neutral, or positive, depending on which version of the questionnaire the participant received, followed by mood ratings that were supposedly unrelated to the memory questions. As such, this study addresses the unresolved issue of whether memories influence mood or whether apparent memory effects on mood are actually mood effects on memory, or result from demand characteristics of the experiments.