Embodied self-other overlap in romantic love: Cerebral regions known to be involved in the processing of self-related information are activated by the presentation of the beloved’s name or face

Embodied self-other overlap in romantic love: a review and integrative perspective. Virginie Quintard, Stéphane Jouffe, Bernhard Hommel, Cédric A. Bouquet. Psychological Research, February 15 2020. https://link.springer.com/article/10.1007/s00426-020-01301-8

Abstract: Romantic love has long intrigued scientists in various disciplines. Social-cognitive research has provided ample evidence for overlapping mental representations of self and romantic partner. This overlap between self and romantic partner would contribute to the experience of love and has been found to be a predictor of relationship quality. Self-partner overlap has been mainly documented at the level of conceptual or narrative self, with studies showing confusion between one’s own and partner’s identity aspects, perspectives, and outcomes. But the self is not restricted to abstract, conceptual representations but also involves body-related representations, which, research has revealed, are linked to social-cognitive processes. In this article, we review the emerging evidence that romantic love involves not only a blurring of conceptual selves but also a reduction of the distinction between self and romantic partner at a bodily level. We discuss the potential function(s) of self-other overlap in romantic relationship at the level of body-related representations and consider possible mechanisms. We conclude with possible future directions to further investigate how romantic love engages embodied self-other representations involved in social interactions.

TEC: the Theory of Event Coding
ASL: the Associative Sequence Learning


Imitation

According to TEC and the ASL model, sensorimotor learning plays an important role in the development of shared bodily representations. Importantly, while both theoretical views suggest that the main source of learning is through execution and perception of one’s own movement, the perception of others’ movements may also contribute to the development of perception–action links (Heyes & Ray 2000; Hommel, 2018). These links can be created through the establishment of associations between self-produced and others’ perceived actions, during the experience of synchronous action or when being imitated (accounting for example for mirroring of opaque actions such as facial expression—self-produced actions that one cannot see, except in a mirror). Therefore, these models predict that perception of others’ actions that often co-occur with self-performed action is more likely to trigger a perception-compatible action than other possible actions. Accordingly, people spending more time together and sharing activities, like romantic partners, should show more imitation of each other’s actions. Consistent with this proposal, in a now famous study, it was revealed that romantic partners tend to become physically similar after twenty years of marriage (Zajonc, Adelmann, Murphy, & Niedenthal, 1987). Such similarities may be explained by reciprocal imitation of the partner’s facial and other bodily expressions over time, leading partnered individuals to incorporate the bodily expressions of the other in their own body representation. A recent study by Maister and Tsakiris (2016) has more directly investigated automatic imitation (as a behavioral index of embodied self-other overlap) in romantic relationships. In their study, romantically involved participants were instructed to open or close their mouth depending on the color of a target dot while viewing pictures of their partner vs. a friend who produced either the same or a diferent action. The interference between observed and executed action (automatic imitation) was found to be larger when participants were exposed to pictures of their romantic partner’s actions compared to pictures of their friend’s actions. This may indicate greater overlap in embodied representations between romantic partners as compared to friends, even though attentional efects (more attention being drawn to pictures of the romantic partner) cannot be excluded. Increased automatic imitation of the beloved may have important consequences, as it potentially implies greater reciprocal imitation between romantic partners. Indeed, given the well documented positive outcomes of behavioral mimicry (such as liking, trust, and closeness; Chartrand & Lakin, 2013), increased imitation of the beloved may play a signifcant role in romantic relationships by fostering liking and connection between lovers. In line with this, studies based on naturalistic observation of interacting individuals have demonstrated that the amount of rapport individuals feel with each other is correlated with the assessment of the amount of posture sharing during their interactions (Lafrance, 1979; Lafrance & Broadbent, 1976). It is interesting to note that people involved in a romantic relationship are less prompt to imitate an attractive alternative, especially when passionately in love (Karremans & Verwijmeren, 2008). From this reduced mimicry of attractive alternatives, which might contribute to relationship maintenance, it is tempting to conclude that imitation has a special status within the romantic relationship.


Overlapping (conceptual) selves in romantic love

Social-psychological research has provided various conceptualizations and models of romantic love (e.g., Berscheid & Walster, 1978; Sternberg, 1986). Among these, the selfexpansion model of love is a process-oriented model conceiving romantic love as the cognitive inclusion of the (cognitive representation of) romantic partner in the self (Aron & Aron, 1996; Aron, Aron, & Norman, 2004). This model frst posits a central human motivation to extend the self, in the sense that people seek to enhance their perspectives, resources, and identities, to improve their potential efcacy (i.e. their ability to accomplish goals) (Aron & Aron, 1986; Maslow, 1967). Second, close relationships, especially romantic relationships, are assumed to provide self-expansion through the inclusion of other in the self, a process by which partners become closer and develop overlapping self representations (Aron & Aron, 1996; Aron, Mashek, & Aron, 2004). Accordingly, the self is expanded in romantic relationships by including aspects of the other in the self: the partner’s resources, perspectives, and characteristics are to some extent treated as one’s own (Aron & Aron, 1996). The experience of self-expansion is rewarding and positive, in that it broadens one’s own potentialities. It is assumed that the exhilaration associated with romantic love would be related to the experience of such self-expansion through the inclusion of partner’s characteristics into one’s self content (Aron, Norman, Aron, McKenna, & Heyman, 2000). Consistent with this view, self-expansion has been linked to positive outcomes in romantic relationships, such as admiration for the romantic partner, greater levels of satisfaction, commitment, and passion (Aron et al., 2000; Fivecoat, Tomlinson, Aron, & Caprariello, 2015; Mattingly, Lewandowski, & McIntyre, 2014; Reissman, Aron, & Bergen, 1993; Schindler, Paech, & Löwenbrück, 2015). Moreover, to the extent romantic partner’s perspectives and identities are experienced as one’s own, these partner’s characteristics thus turn out to be cognitively linked to one’s sense of self, leading to an overlap of representations of self and other (Aron & Aron, 1986). Being in love thus entails a reduced distinction, or a greater confusion, between the self and the romantic partner (Aron et al., 1991; Mashek et al., 2003; Rusbult, Martz, & Agnew, 1998). In line with the idea of self-expansion through the inclusion of the romantic partner in the self, people involved in a romantic relationship report more and wider domains (i.e. an increased diversity in self-descriptive terms) in the contents of the self-concept than singles (Aron, Paris, & Aron, 1995). This result attests to an expansion of the self-concept which is thought to arise from the inclusion of romantic partner’s characteristics into one’s own self-representation. Furthermore, consistent with the view that love involves overlapping representations of self and partner, research based on the Inclusion of the Other in the Self (IOS) scale, a pictorial measure supposed to capture self-other overlap (Aron, Aron, & Smollan, 1992), showed that participants report more overlap with their romantic partner compared to with a close friend or a family member (Acevedo, Aron, Fisher, & Brown, 2012; Quintard, Joufre, Croizet, & Bouquet, 2018). Importantly, comforting the idea that self-other overlap plays a significant role in romantic relationships, IOS scores are predictors of relationship stability over 3 months, the degree of overlap being negatively related to the likelihood of relationship dissolution (Le, Dove, Agnew, Korn, & Mutso, 2010). A key prediction of the self-expansion model is that overlapping representations of self and romantic partner elicit selfother confusion. Consistent with this prediction, studies have shown confusion between partner’s and one’s own traits (Aron et al., 1991; Mashek et al., 2003), interests, or attitudes (Aron, Steele, Kashdan, & Perez, 2006). For example, Aron et al. (1991) found slowed response times in a “me/not me” decision task (i.e., does the trait describe me?) when romantically involved participants had to evaluate traits that were relevant only for self or partner, compared to shared traits, suggesting blurred self-other boundaries in romantic love (see also Smith, Coats, & Walling, 1999). Also, the typical drop of self-esteem observed when one experiences upward social comparison is no longer present when people are outperformed by their romantic partner, especially for those reporting a high level of closeness (Lockwood, Dolderman, Sadler, & Gerchak, 2004). Thus, people do seem to experience their partner’s outcomes as their own. Results from neuroimaging studies also suggest overlapping representations of self and romantic partner. Cerebral regions known to be involved in the processing of self-related information, such as anterior cingulate cortex, fusiform and angular gyri, are activated by the presentation of the beloved’s name or face (Aron et al., 2005; Ortigue, Bianchi-Demicheli, Hamilton, & Grafton, 2007). Moreover, these cerebral activations triggered by the evocation of the beloved are positively correlated with the IOS-scores reported by participants with respect to their partner (Acevedo et al., 2012). To sum up, several lines of evidence show a blurring of the distinction between self and romantic partner, in line with the model of romantic love as an inclusion of the partner in the self. A large part of this work dealt with individuals’ traits, interests or attitudes (Aron et al., 1991, 2006; Mashek et al., 2003), that is, abstract, conceptual forms of self-representation (Gallagher, 2000; Smith, 2008). However, romantic, intimate relationships involve embodied cues such as touch, physical proximity, and shared bodily experiences (Fiske, 2004). Moreover, as speculated in early formulations of the self-expansion model of love, “in close relationships one’s body also behaves as if it is the other’s body” (Aron & Aron, 1996, p. 50), implying that the inclusion of the romantic partner in the self encompasses the body. Self-partner processing at a bodily level may be an important aspect to consider in romantic relationships in the light of empirical and theoretical research suggesting that self-other representations and social relationships are grounded in sensorimotor processing and body-related representations (e.g. Barsalou, 2010; Smith, 2008). In the next section, we provide an overview of this line of work, focusing on self-other overlap at a bodily level and its potential role in social cognition. Then, we explain why it is relevant to approach romantic love from this perspective, addressing why romantic love would affect the overlap between self and romantic partner at the bodily level.


Functional mechanisms underlying self‑other confusion

Theoretical frameworks have successfully predicted that romantic love increases self-other overlap and the sense that representations of oneself and representations of the beloved one becomes less discriminable. While the social effects and implications of this increase in representational overlap are varied and substantial, it is not yet well understood how it actually works. On the basis of ASL and TEC frameworks, we argued above that shared bodily experiences with the romantic partner may partly explain stronger bodily self-other overlap, in that it would promote the development of sensorimotor links between, or integration of, self and partner actions. But romantic love does not only, or necessarily, involve shared bodily experiences, it also entails a strong affective component and social evaluation and attitudes, which are further potentially important modulatory sources of self-other overlap. Although models acknowledge that action–perception links can be modulated by factors such as social distance or attitudes, through top-down modulation (e.g. Heyes, 2011; Chartrand & Lakin, 2013; Wang & Hamilton, 2012), they remain relatively silent regarding the cognitive mechanisms involved. However, there are frst indications that romantic love might be systematically related to, and perhaps trigger a particular cognitive-control style that favors the integration of representations, be they social or not, over discrimination. The general idea that interpersonal relations might be systematically associated with particular styles of information processing has received quite some support in recent years. For instance, high (vs. low) social power has been suggested to lead to more abstract information processing (Smith & Trope, 2006) and, conversely, abstract thinking has been shown to raise one’s subjective sense of power (Smith, Wigboldus, & Dijksterhuis, 2008). More specifcally, it has been reported that love priming (via imagination instructions), as compared to sexual priming, promotes a global, integrative processing of information in a classical local/ global processing task (Förster, Özelsel, & Epstude, 2010; see also Förster, 2009). The idea that romantic love is associated with a global processing bias, accompanied by reduced attentional selectivity (de Fockert, Caparos, Linnell, & Davidof, 2011), is consistent with the outcomes of a recent study investigating the link between romantic love and sensitivity to irrelevant information (van Steenbergen, Langeslag, Band, & Hommel, 2014). Romantically involved participants were frst required to imagine or write about a romantic event and listen to their favorite love-related music. Then they completed two conflict tasks (a Stroop and a flanker task) indexing the ability to regulate interference from irrelevant information according to situational demands. The results showed a positive association between the intensity of passionate love (as reported on the Passionate Love Scale) and the degree of interference control: more intense loving was associated with reduced selectivity, leading to a stronger impact of irrelevant information. In other words, romantic love is accompanied by a global/integrative mode of cognitive processing, which would indeed be expected to reduce self-other discrimination. This observation raises the questions (1) how the hypothetical global/integrative mode reduces discrimination between self- and other-representations, (2) why this reduction covaries with the kinds of efects and behaviors that were found to accompany romantic love, and (3) why this mode is sensitive to romantic love. Even though none of these questions has been investigated in the context of romantic love already, available fndings suggest a preliminary scenario. With respect to the frst question, general models of cognitive control have suggested that adaptive behavior requires a dynamic balance between two conficting cognitive control states, persistence/selectivity and flexibility/integration (Cools & D’Esposito, 2011; Goschke, 2003; Hommel & Wiers, 2017)—a process that has been called metacontrol (Hommel, 2015). According to a recent formulation of this view (Hommel & Wiers, 2017), a metacontrol bias toward persistence/selectivity strengthens the top-down infuence of the current goal, which focuses the system on relevant information and creates a strongly selective processing state reinforcing mutual competition between alternative representations. Conversely, a metacontrol bias toward fexibility/ integration is characterized by the weak top-down influence of the current goal and weak mutual competition between alternative representations, which widens the focus and creates a more integrative processing mode. Furthermore, the individual pattern of persistence/flexibility tradeof would emerge from an interaction between various factors known to bias cognitive control, such as genetic predisposition (Colzato, Waszak, Nieuwenhuis, Posthuma, & Hommel, 2010), cultural learning (Hommel & Colzato, 2017), task constraints (Bonnin, Gaonac’h, & Bouquet, 2011; Mekern, Sjoerds & Hommel, 2019), and afect (Dreisbach & Goschke, 2004). Various fndings have provided evidence that participants biased towards persistence/selectivity outperform others in tasks that require the exclusion of irrelevant information but perform more poorly than others in tasks that require the conjoint processing of different kinds of information, while participants biased towards fexibility/ integration show the exact opposite pattern (for reviews, see Hommel, 2015; Hommel & Colzato, 2017). From this theoretical perspective, it would make sense to assume that romantic love induces a bias towards flexibility/integration.
According to this metacontrol approach, and now we turn to the second question, romantic love should reduce the impact of the current goal on information processing, which is consistent with the observation that viewing a picture of the romantic partner is associated with a deactivation of brain areas involved in the representation of task intentions (Bartels & Zeki, 2000, 2004; Zeki & Romaya, 2010). The approach would also predict more confict between alternative representations, given that mutual inhibition is reduced, which accounts for van Steenbergen et al.’s (2014) observation that the intensity of romantic love is accompanied by a loss of conflict control. Converging evidence comes from studies in which metacontrol biases towards flexibility/integration were experimentally induced by having participants engage in a divergent thinking task, which is taxing people’s flexibility (Guilford, 1967). This manipulation has been found to evoke behavior that is very similar to that evoked by romantic love: it promotes interpersonal trust (Sellaro, Hommel, de Kwaadsteniet, van de Groep, & Colzato, 2014) and the integration of the others’ actions into one’s own task representation (Colzato, van den Wildenberg, & Hommel, 2013). Hence, romantic love may generate a similar bias towards flexibility/integration, which would account for the blurring of boundaries between the self and the romantic partner, at both conceptual and bodily levels. But why would romantic love do this? This brings us to our third question. Metacontrol biases have been shown to depend on genetic predisposition and cultural molding—two factors that are rather permanent and stable—but also on situational factors (Hommel & Colzato, 2017). The best-investigated situational factor is mood, which in the case of positive mood has been demonstrated to promote metacontrol fexibility at the cost of persistence (Dreisbach, 2006; Dreisbach & Goschke, 2004). This is interesting for our purposes for no less than four interconnected reasons. First, positive mood has also been consistently found to improve divergent thinking (Baas, de Dreu & Nijstad, 2008), the task that apparently induces similar kinds of behavior than romantic love does. Second, inducing positive mood was found to reduce interference control in confict tasks in similar ways than romantic love does (van Steenbergen, Band, & Hommel, 2010). Third, both positive mood (Akbari Chermahini & Hommel, 2012; Dreisbach et al., 2005) and divergent thinking (Akbari Chermahini & Hommel, 2010, 2012) have been shown to rely on (presumably striatal) dopamine, the neurotransmitter that is assumed to underlie metacontrol (Cools & D’Esposito, 2011; Hommel & Colzato, 2017). And, fourth, romantic love has been linked to dopaminergic transmission (Fisher et al., 2006). Taken altogether, this picture implies that engaging in romantic love and similar positive emotions are neurally represented as and/or accompanied by tonic increases of (presumably striatal) dopamine. Given that metacontrol biases are assumed to emerge from the interaction of frontal and striatal dopaminergic activity, and that increases of striatal dopamine are related to a stronger bias towards flexibility (Hommel & Colzato, 2017), this means that romantic love and flexibility biases are sharing a neuromodular mechanism that is known to generate behavior that has been observed in romantic lovers


Conclusion and future directions

In this paper, we combined theories and empirical work from contemporary social cognition, experimental psychology and neurosciences, to address the outstanding question of self-other processing in the bodily domain within romantic relationships. Extending previous research on conceptual self-other representations, the reviewed empirical findings suggest a prominent overlap between self and partner at a bodily level in romantic love. Moreover, this bodily overlap has been repeatedly found to be related to the intensity of romantic feelings. Thus, the two forms of selfhood—bodily self and conceptual self—seem to be engaged in the creation of the unique, intimate link between lovers. Building on models of perception–action links [the Theory of Event Coding (TEC) and the Associative Sequence Learning (ASL)], we propose an integrative view of causes and consequences of embodied self-other overlap in romantic relationships. We suggest that the ability to share partner’s bodily states facilitates interactions and promotes behaviors strengthening the afective bond between self and partner. Hence, this view highlights the key role of shared bodily states in social functioning as embodied cues of connectedness (Smith, 2008). Furthermore, an important and original aspect of our proposal is that it articulates the role sensorimotor and afective experiences in self-other processing. We argue that bodily experiences shared with the romantic partner and afective states may play a role in promoting the integration of self and partner bodily representations. Our proposal, which is rooted in TEC, is partly consistent with other models suggesting that action–perception links acquired through sensorimotor experience may be suficient to explain the development of social cognitive abilities such as imitation and/or empathy (e.g. Brass & Heyes, 2005; Heyes, 2018; Keysers & Gazzola, 2009). Here we suggest that to fully account for the impact of these links on social cognition, it is necessary to assume regulatory processes (metacontrol states) biasing the discrimination between self- and other-representations. By the same token, these regulatory processes allowed us to explain how afective states associated with romantic love may contribute to the reduction of bodily boundaries between self and romantic partner.
Yet, the potential role of self-partner bodily merging and the underlying mechanisms are far from being understood. Future directions can be identified to further investigate how romantic love engages embodied self-other representations involved in social interactions. Conceptual self-other overlap (as indexed for instance by IOS-scores) in the romantic context has been found to correlate with intimacy, relationship commitment and satisfaction, and to be a predictor of long term relationship stability and quality (Agnew et al., 1998; Aron et al., 1991; Le et al., 2010). Much less is known however regarding selfother overlap in the bodily domain. Future research should explore the links between embodied self-other overlap and diferent aspects of the relationship (satisfaction, self-disclosure, emotional expression, level of intimacy…). Another valuable direction would be to test how bodily self-other overlap with the romantic partner, as indexed for instance by behavioral mimicry, in the early stage of the relationship is a predictor of future relationship outcomes. More broadly, further work should examine the association between embodied self-other overlap and social cognition processes associated with the beloved. Studies already found a positive association between the intensity of romantic feelings and self-partner overlap at a bodily level (Ortigue et al., 2010; Quintard et al., 2018). This brings about the exciting question of the causal role of blurred self-other bodily boundaries to romantic feelings or attraction. A promising direction would be to test whether imitation of the beloved or procedures creating confusion between one’s own and other’s body (such as interpersonal multisensory stimulation; Tajadura-Jiménez & Tsakiris, 2014) can afect self-partner relationship. Another interesting future direction is to explore the impact of reduced bodily boundaries between self and romantic partner on the representation of peripersonal space (i.e., the space within reach). This space, which is crucial for our interaction with objects and others, has been proven to be very plastic. It is expanded in presence of another person—especially if the person is cooperative (Teneggi, Canzoneri, di Pellegrino, & Serino, 2013) and it is updated following physical changes in one’s own body (Cardini, Fatemi-Ghomi, Gajewska-Knapik, Gooch, & Aspell, 2019). Interestingly, recent work using interpersonal multisensory stimulation has demonstrated that experimentally induced reduction of self-other bodily boundaries modifes the representation of the other’s peripersonal space—in the sense of a remapping onto one’s own space (Maister, Cardini, Zamariola, Serino, & Tsakiris, 2015). Such a modification of the representation of the other’s peripersonal space may afect the way one processes events related to the other and his/her behaviors. This calls for future investigations of whether peripersonal space boundaries between romantically involved individuals may be modified and how this may relate to reduced bodily boundaries between self and romantic partner. A frst step to address this fascinating question may be to test whether the presence of the romantic partner modifies our representations of peripersonal spaces differently from other individuals. An important, yet relatively unexplored, possible efect of self-other overlap is a transfer of (usually) positive self-evaluations to the other. This kind of transfer would explain why physically touching objects triggers ownership efects (i.e. the fact that we value more objects that we own; Beggan, 1992): owned/touched objects would be valued because of their association with the (positive) self. An outstanding question is to what extent this transfer applies to one’s romantic partner, and whether self-other bodily merging may sustain such an effect. Finally, as stressed earlier, only a few studies have been devoted to the examination of basic, domain-general cognitive processes in the context of romantic love. Cognitive skills such as self-regulation and self-control (i.e. cognitive skills related to cognitive control) have been connected with relationship maintenance (Finkel & Campbell, 2001; Ritter, Karremans, & van Schie, 2010). Recent work also suggests that romantic love involves particular cognitive control states (van Steenbergen et al., 2014). Identifying potential modes of cognitive processing associated with romantic love and specifying their links with both conceptual and embodied self-other overlap is thus another crucial direction for research on romantic cognition. We hope considering these future directions will shed some new light on cognition and behaviors associated with romantic love.