1.1. Love and the commitment problem 

Frank, 1988, Frank, 2001).) and Hirshleifer (1987) proposed that emotions, such as vengefulness, outrage, empathy, affection, and love, act as commitment devices by leading people to forego speciously attractive immediate rewards in favor of strategies that yield greater fitness benefits over the long-term. Vengefulness, while immediately costly, can lead to behaviors that shape one's reputation as a person not to be scorned. Empathy, affection, and love can lead to immediate sacrifices that serve to promote long-term relationships. The recurrent experiences of these emotions constitute immediate rewards or punishment that can facilitate commitment in spite of the fact that their ultimate benefits may occur in the distant future (Frank, 1988).

Love, according to the commitment device theory, should help an individual select a mate and to bond exclusively to his or her partner (at least for a time). Indeed, people in love often believe that they have found their one true soul mate in a world of billions of possibilities, and hence, the experience of love appears to help them genuinely foreclose other options. In the current research, we conduct the first experimental examination of the possible relationship between experiences of love and reduced temptation of attractive alternatives.

1.2. The derogation of alternatives 

In psychology, there is an often-cited literature suggesting that entering relationships leads people to “derogate” their alternatives. These studies have shown that, relative to people not in relationships, those in relationships spend less time visually attending to members of the opposite sex (Miller, 1997), and they rate opposite-sex others as less attractive (Johnson and Rusbult, 1989, Lydon et al., 1999, Simpson et al., 1990).

This work is consistent with the hypothesis that love diminishes attraction to alternatives and promotes commitment to a single partner, but for two central reasons, it is far from definitive. First, the conclusions one can draw from these studies are limited because nearly all of them are quasi-experimental in nature, comparing individuals in or out of relationships or dividing participants by level of expressed relationship commitment. A simple and much less interesting alternative explanation for the findings, for example, is that participants involved in relationships realize that it is socially undesirable to state that alternative mates are attractive. In the single study that manipulated commitment (Johnson & Rusbult, 1989), participants imagined themselves in a hypothetical situation in which they were explicitly instructed—in the low-commitment condition—to actively search for someone else to date, potentially causing participants to report that the alternative was more attractive because of an acquiescence bias. Second, existing research on attraction to alternatives has only examined relationship status and subjective feelings of commitment, rather than directly examining the experience of love. According to the commitment device theory, it is the actual experience of the emotion—the feeling of love—that promotes long-term bonds by acting as an internal incentive system (Frank, 1988).

The central goal of our study was to address these weaknesses by conducting an experimental manipulation of love in the laboratory. To do this, we chose to make use of a subtle psychological phenomenon, the rebound effect.

1.3. The rebound effect following thought suppression 

Previous work has documented that the active suppression of a thought can lead to an ironic rebound effect (Wegner, Schneider, Carter, & White, 1987), such that one experiences more of the thought after thought suppression than if one had not attempted to suppress the thought at all. This effect may be especially potent when thoughts are exciting, secretive, and relevant to relationships. For example Wegner and colleagues had participants suppress thoughts about sex and found that the process of suppression led to increase in physiological arousal (i.e., skin conductance level) as compared to suppressing nonexciting thoughts (i.e., about dancing) (Wegner, Shortt, Blake, & Page, 1990, also see Wegner & Gold, 1995).

We had individuals suppress thoughts of an attractive alternative while they were induced to feel different emotional states. We predicted that the experience of love, but not other closely-related emotional states, would render the alternative less tempting, and thus, it would reduce or eliminate the rebound effect.

This methodology has three major advantages. First, in addition to eliciting the experience of love, we could induce other closely related emotion states and compare their effects with those of love. We used sexual desire as a comparison. Although romantic love and sexual desire often co-occur, researchers have hypothesized that sexual desire promotes sexual initiation rather than the maintenance of established bonds (Buss, 2003, Diamond, 2003, Fisher et al., 2002, Gonzaga et al., 2001, Gonzaga et al., 2006).

The second advantage of this methodology is that it allowed us to make a strong test of this comparison by eliciting love and sexual desire for the same target, the current romantic partner, and then investigating the divergent effects of these emotional states on participant's ability to suppress thoughts of an attractive alternative. If there are differences in participants' ability to suppress the thought of an attractive alternative under these conditions, we felt we could be confident that the differences are caused by the emotional states themselves.

Finally, the rebound effect is a subtle and nonobvious phenomenon, making it virtually impossible for participants to bias their results to confirm our expectations. This is important because, in past studies on the attractiveness of alternatives, participants in relationships may have believed that they should avoid indicating that they found alternatives attractive and biased their responses accordingly.

2.1. Participants 

One hundred twenty undergraduate students, 72 women and 48 men [mean age=19.85 years; S.D.=2.53, Caucasian (38.2%), Asian (34.2%), Latino (8.3%) African American (2.5%), others (15.8%), from a public university participated for research credit or $7. All participants were in a monogamous heterosexual dating relationship (mean length, 34.5 months; S.D.=32.1).

2.2. Materials: photographs 

The authors selected photographs of potential alternatives from public Web sites. Those pictured approximated the age range and ethnic breakdown of the university and were rated as quite attractive by the participants (M=6.98, S.D.=0.70).

2.3. Procedure 

Participants completed the procedure individually with a same-sex experimenter who was blind to the hypotheses of the study. After participants completed consent forms and demographic, relationship, and personality measures, they were led to a table with photographs of attractive opposite-sex individuals. They selected the photograph of the individual they found most attractive (i.e., the alternative). The remaining photographs were covered.

Participants reported how attractive they found the alternative on a scale of 1 (least attractive person I have ever seen) to 9 (most attractive person I have ever seen). Then, while alone, participants wrote two 5-min essays on the alternative. The first essay was on why the alternative was attractive. The second essay was on what would constitute a perfect first meeting with the alternative. After writing, each essay participants reported how much they felt 15 different emotions while writing each essay. The photograph was then removed.

2.3.2. Suppression manipulation 

Participants were also randomly assigned to either (a) suppress the thought of the alternative during the third essay and express the thought of the alternative during the fourth essay (rebound condition) or (b) express the thought of the alternative during the third essay and suppress the thought of the alternative during the fourth essay (non-rebound condition). While writing, participants placed a check mark on the page each time they thought of the alternative (Kelly & Kahn, 1994). The rebound effect was indexed by the number of checkmarks in the express essay after they had suppressed the thought of the alternative (Macrae et al., 1994). Fig. 1 displays the timing and content of the experimental session.

Fig. 1. The timeline and content of the experimental session.

This 3×2 between-groups design had three emotion conditions (love, sexual desire, control) and two rebound conditions (rebound, nonrebound). Participants reported how much they felt 15 different emotions (including love and desire) during each essay on a scale of 0 (none at all) to 8 (an extreme amount) after each essay. They then completed the Investment Model Scale (Rusbult, Martz, & Agnew, 1998), a seven-item measure of commitment (α=.90). After this, participants were given time to recall details about the photograph.

3.1. Manipulation check 

Participant's ratings of how much love they felt while writing the final two essays were combined. We ran between-subjects analyses of variance (ANOVAs) to check that reports of love were highest in the love condition. Writing about experiences of love for a romantic partner produced higher levels of love (M=6.12, S.D.=2.25) than in the control condition [M=3.66, S.D.=2.21, F(1,78)=24.31, p<.001, η2=.238], and desire condition [M=5.29, S.D.=2.10, F (1,78)=2.96, p<.05, η2=.037] (both one-tailed tests).

3.2. Sex differences 

Sex differences in the effects are noted when they occurred. In all other analyses, no sex differences were found.

3.3. Does love positively relate to subjective commitment? 

We first related reports of love experienced during the procedure to our measure of commitment taken after the end of the study. As a point of comparison, we also related reports of desire to measures of subjective commitment. We predicted commitment from mean reports of love and desire across the third and fourth essays separately for the participants in the love and sexual desire conditions. The control condition was excluded because reports of love and desire were not focused on the romantic partner. Because love and desire were significantly correlated [r(80)=.48, p<.001], we predicted subjective commitment with each emotion independently and also when both were entered simultaneously to show their independent effects. Reports of love positively correlated to subjective commitment using both zero-order correlations [r(79)=.52, p<.001, r2=.27] and beta weights when love and desire were entered into a regression together [β=.60, t(76), p<.001]. Reports of desire did not relate to measures of subjective commitment [zero-order r(79)=.12, ns, beta weight β=−.17, t(76)=−1.52, ns].

3.4. Does love predict diminished thoughts of attractive others? 

We constructed two measures of thoughts of attractive others. These were the number of thoughts individuals had about the attractive alternative and memory for the attractive alternative at the end of the study. Relative to those in other conditions, we predicted that participants in the love-and-rebound condition would have fewer thoughts of the alternative in the express essay and remember less about the alternative at the end of the study.

3.5. Does suppression while feeling love reduce the rebound of thoughts of an attractive alternative? 

In order to estimate the number of times participants had thoughts of the attractive alternative while writing the essay, we asked them to write checkmarks in the essay each time these thoughts occurred (see instructions above). We used the total number of checkmarks as the estimate of the number of thoughts of the alternative that occurred to the participant (e.g., Kelly & Kahn, 1994). Because the distribution of checkmarks was positively skewed we used a square root transformation to normalize checkmark data (pretransformation skew=2.48, posttransformation=0.52; Fraley & Shaver, 1997). In the tables, we present raw numbers for ease of interpretation.

Table 1 shows the mean number of checkmarks by experimental condition and sex of the participant. We first tested to see if love reduced the rebound effect relative to the desire and control condition. This hypothesis held that a significant difference would emerge between the love and desire conditions and the love and control conditions for checkmarks made in express essay in the rebound condition, but not the nonrebound condition. In line with Keppel and Zedeck (1989), we moved directly to our planned comparisons to test these predictions.

Table 1. Means and standard deviations of the number of checkmarks by condition and sex

		Emotion condition
		Love	Desire	Control
		M (S.D.)	M (S.D.)	M (S.D.)
	Rebound condition
	Express essay	0.60 (0.88)	2.10 (2.69)	3.90 (4.48)
	Suppress essay	0.55 (0.94)	1.00 (1.26)	3.25 (3.06)
	Non-rebound condition
	Express essay	2.40 (2.50)	1.45 (1.23)	3.55 (3.59)
	Suppress essay	0.90 (1.17)	0.55 (1.10)	1.85 (2.08)

The critical comparisons are shown in bold. M, mean.

Our planned comparisons, tested with ANOVAs, confirmed our hypotheses. As expected, participants in the love condition had fewer thoughts of the alternative than those in the sexual desire condition [F(1,38)=6.33, p<.05, η2=.143] and the control condition [F(1, 38)=17.87, p<.001, η2=.320]. Participants in the sexual desire condition had marginally fewer thoughts of the alternative than those in the control condition [F(1,38)=2.89, p<.10, η2=.071]. The overall emotion effect in the rebound condition for number of thoughts of the alternative in the express essay was significant [F(2,57)=8.56, p<.001, η2=.231].

In the nonrebound condition, there were no significant differences between the three different emotion conditions [overall effect F(2, 57)=1.77, ns]. The omnibus 3×2 ANOVA with emotion condition and rebound condition as independent variables showed a significant interaction [F (2, 114)=3.42, p<.05, η2=.057] as well as a main effect for emotion condition [F(2,114)=7.12, p<.01, η2=.111].

We then investigated the correlation between thoughts of the alternative and mean reports of love across the third and fourth essays for the participants in the love and sexual desire conditions. Here, we predicted that the more love a participant felt (regardless of emotion condition), the fewer thoughts they would have about the alternative. We related both love and desire, our comparison emotion, to the number of thoughts participants had of the alternative. We also entered both emotions into a regression equation predicting the number of thoughts of the alternative simultaneously to show their independent effects. Love reports related to having fewer thoughts of the attractive alternative, zero-order correlation [r(80)=−.28, p<.05, r2=.078, beta weight β=-.34, t (77)=−2.77, p<.01]. Reports of desire did not relate to thoughts of the attractive alternatives [zero-order correlation r (80)=−.04, ns, beta weight β=.12, t (76)=1.00, ns]. There was one significant sex difference; after controlling for love, desire was positively related to thoughts of the alternative in men [β=.45, t (29)=2.07, p<.05] but not women [β=−.01, t (45)=−0.07, ns; difference Z (78)=2.04, p<.05].

3.6. Does suppression while feeling love reduce memory for attractiveness-related details of the alternative? 

The next set of analyses parallel to those for the checkmark data. Here, we investigated how love and suppression affected memory for details of the attractive alternative. We reasoned that if feeling love and suppressing the thought of the alternative reduced the rebound effect, leading to fewer thoughts, then participants would have a more difficult time remembering details about the alternative. Moreover, we reasoned that if love acts as a commitment device, it would help to suppress relationship-threatening thoughts (i.e., of how attractive an alternative was) more than other types of thoughts. Thus, we predicted that this effect should occur primarily for details about how attractive the alternative was.

Specifically, we predicted that for participants in the rebound condition those in the love condition would remember fewer attractiveness-related details than those in the desire or control conditions. These differences would not occur for attractiveness-unrelated details.

This was the case. We used two planned comparison ANOVAs to test the difference between the love and desire as well as the love and control conditions for attractiveness-related details. As predicted, there were significant differences between the love [M=4.58, S.D.=2.81] and desire conditions [M=6.42, S.D.=2.23, F(1,37)=5.20, p<.05, η2=.123], as well as between the love and control conditions [M=6.38, S.D.=2.55, F(1,37)=4.38, p<.05, η2=.106]. These means are shown in Table 2.

Table 2. Means and standard deviations of the number of checkmarks by condition and sex

		Emotion condition
		Love	Desire	Control
		M (S.D.)	M (S.D.)	M (S.D.)
	Rebound condition
	Attraction-related	4.58 (2.80)	6.43 (2.23)	6.38 (2.55)
	Nonattraction-related	1.89 (1.80)	2.73 (2.90)	2.08 (1.58)
	Nonrebound condition
	Attraction-related	6.33 (3.00)	6.15 (2.96)	7.55 (2.70)
	Nonattraction-related	1.44 (1.54)	1.70 (1.45	1.95 (1.38)

The critical comparisons are shown in bold.

The main effect of emotion for the number of attractiveness-related details occurred only for participants in the rebound condition [F(2,56)=3.32, p<.05, η2=.106]. The effect was not significant in the nonrebound condition [F(2,55)=1.38, ns]. The omnibus ANOVA with rebound condition and emotion condition was not significant [F(2,111)=1.44, ns].

There were no significant differences or close to significant differences when testing for differences in the number of attractiveness-unrelated details. As we expected, love decreased participants' ability to remember attractiveness-related details in the rebound condition.

3.7. Do thoughts of attractive others relate to subjective commitment? 

The final analyses tested the number of thoughts of attractive alternatives related to measures of subjective commitment. Here, we correlated the number of thoughts participants had for the alternative and the number of details (both attractiveness related and attractiveness unrelated) to measures of subjective commitment in the entire sample. There was a significant negative correlation between the number of thoughts of the alternative and measures of subjective commitment [r (119)=−.37, p<.001, r2=.137]. There was also a negative correlation between remembering attractiveness-related details about the alternative and commitment [r (17)=−.20, p<.05, r2=.04]. There was no correlation between remembering attractiveness-unrelated details about the alternative and commitment [r(117)=.08, ns]. The difference between the correlations of attractiveness-related details and commitment and attractiveness-unrelated details and commitment was significant [Z(117)=3.02, p<.01]. There was one significant sex difference: in women, there was a negative correlation between subjective commitment and memory for attractiveness-related details [r (70)=-.32, p<.01] but no correlation in men [r(47)=.03, ns, difference Z(117)=1.86, p<.05].

4.1. Limitations of the current studies 

There are limitations to these findings. First, a legitimate question is whether our laboratory manipulation reflects events that occur in the natural world. In a sample of college students, we found that over 33% had attempted to suppress intrusive thoughts of attractive alternatives. Thus, at least among young adults, suppressing thoughts of attractive others seems to be a common practice (those interested in these data may contact the authors for more details). Whether or not people engage in thought suppression as a way of shoring up their relationships, our findings are clearly noteworthy from a theoretical standpoint. They are the first results to demonstrate that love diminishes the temptation of alternatives and to link this phenomenon to subjective commitment.

Second, a question not addressed by our study is whether the effects we have documented are sufficiently strong to maintain commitment in relationships. We do not claim that any single instance of love helps an individual to suppress thoughts of others or that increasing commitment would have a major, longstanding impact on an ongoing relationship. Rather, we believe that repeated instances of love, aggregated over time, help to maintain commitment. This proposal could be tested in future work.

Third, there were strong correlations between love and sexual desire, raising the question of why emotions we claim to be distinct relate so strongly to each other. Diamond, 2003, Diamond, 2004) has proposed that love and sexual desire will elicit each other in many but not all circumstances. This elicitation is most likely to occur when a relationship embodies both commitment and reproductive functions. For example, if a partner has qualities that are valuable in both a long-term partner (Buss, 1989, Buss and Schmitt, 1993) and a sexual partner (Gangestad and Simpson, 2000, Miller, 2000, Singh, 1993), love and desire should co-occur, and the experiences of love and desire positively correlate. There are also circumstances in which one would not expect these two emotions to be strongly related (e.g., one-night stands, getting support from a partner during a long-term illness, engaging in cooperative child care).

There have been proposals that emotions like love are distinct and relatively independent modules that have evolved to address longstanding evolutionary challenges (Kenrick, 2006; also see Keltner & Haidt, 2001, for a related view). Other work has shown that love coordinates a number of different physiological, motivational, and behavioral systems to maintain commitment to a romantic partner (Ellis and Malamuth, 2000, Gonzaga et al., 2001, Gonzaga et al., 2006). In a related vein, for men there is a strong link between short-term mating motives, sexual arousal, and over-perceiving sexual arousal in the faces of attractive women (Maner et al., 2005). Viewing emotions as evolved responses to adaptive challenges is likely to be empirically fruitful.

4.2. Love, commitment, and thought suppression 

How does the momentary experience of love facilitate the suppression of thoughts of attractive alternatives? Previous work has shown that individuals can successfully suppress particularly threatening or emotionally evocative thoughts (Fraley and Shaver, 1997, Kelly and Kahn, 1994, Wegner and Gold, 1995). These researchers have theorized that some thoughts are so disturbing that they force the individual to invest large amounts of psychological resources to suppress the thought. In time, these individuals practice suppression so much that they are eventually successful (Fraley & Shaver, 1997).

We believe that it is unlikely that individuals invested the psychological resources necessary to become adept at successfully suppressing thoughts of attractive alternatives, which would relegate the effects of love we observed to a practice effect. It may also be that love facilitates suppression by refocusing attention on the partner, thereby reducing the opportunity to think about the attractive alternative. This explanation seems unlikely considering the differences between the love and sexual desire conditions and the fact that both conditions elicited the emotion in reference to the partner. As sexual desire was targeted towards the partner, it would seemingly focus more of the participant's attention on their partner.

For individuals in long-term relationships they value, thinking of an attractive other is likely emotionally evocative and potentially threatening. Experiencing love—and its promotion of commitment to a partner—may cause an individual to increase their efforts to suppress the thought of the other, leading to successful suppression. This does not preclude the possibility that, in the long term, these thoughts may still rebound. It is unlikely that individuals will permanently foreclose their romantic alternatives—few commitments are truly final, and relationships change over time. Just as burnt bridges can be rebuilt, alternatives foreclosed today may re-emerge, and if love does not persist, the cooperative alliance may dissolve. However, if love is a commitment device, the suppression of these thoughts should remain in place as long as the relationship is valued and elicits the experience of love.

4.3. Conclusions 

Our studies add to the growing list of work investigating function-specific effects of emotion (Buss et al., 1992, Fessler et al., 2004, Gonzaga et al., 2001, Gonzaga et al., 2006, Lerner and Keltner, 2001). We supported the hypothesis that love, but not sexual desire, is a commitment device, helping an individual suppress thoughts of an attractive other. Challenges to the relationship are thereby thwarted, and commitment is reinforced. These findings suggest that love is the prescient emotion that helps partners stay committed even when tempted by romantic alternatives.