The ability to understand the emotions of others is crucial to successfully function in social interactions. Due to a lack of direct access to other people's mind, we have to rely on indirect sources of information to infer how others feel. For example, we could use information about the context the other person is exposed to, or read their emotional expressions. While these external cues about the other can be valuable means to emotion understanding, research in social cognition has shown that one’s own affective state is also used to guide judgments about others’ emotional experiences (Silani, Lamm, Ruff, & Singer, 2013
; Steinbeis and Singer, 2014
). That is, people tend to project their own emotions when inferring what other people feel, a process known as emotional egocentricity. Self-projection can be an efficient heuristic, especially when our experiences are similar to the others. However, it can also lead to inaccurate emotion attributions unless egocentric inferences are adjusted to account for dissimilarities between oneself and the other person (Mitchell, 2009
Emotional egocentricity has been studied using tasks in which participants are asked to make emotion judgments about themselves and about another person while being simultaneously exposed to either affectively congruent stimulation (e.g., both were touched by a pleasant material) or affectively incongruent stimulation (e.g., the participant received pleasant touch, while the target was touched by an unpleasant material; Silani et al., 2013
). A tendency to project the own emotions onto others is typically indicated by emotion judgments that are biased towards the participants’ own affective states, particularly in incongruent conditions. Egocentric biases have consistently been observed in similar perspective-taking paradigms that used monetary reward and punishment (Steinbeis and Singer, 2014
), as well as visuogustatory (Hoffmann, Singer, & Steinbeis, 2015
) and audiovisual (von Mohr, Finotti, Ambroziak, & Tsakiris, 2019
) stimulation, to induce congruent and incongruent affective states to the participant and the target.
A common feature of the existing emotional egocentricity tasks is that participants are not able to see the target's reactions to the affective stimulations, so emotion judgments are exclusively based on information about the type of stimulation the other is exposed to. Under these conditions, social cognitive processes such as perspective-taking may be activated to infer the other’s emotion. In particular, performance in these paradigms has been taken as an indicator of the participants’ self-other distinction abilities, as egocentric biases in this context are thought to reflect a failure to distinguish the representation of one’s own affective states from that of the other (Silani et al., 2013
; Hoffmann, Koehne, Steinbeis, Dziobek, & Singer, 2016
; Tomova, von Dawans, Heinrichs, Silani, & Lamm, 2014
In daily life situations, however, we can often rely on more basic abilities that do not require perspective-taking processes to understand what others are feeling, such as emotion perception. Imagine, for example, that you give a present to a friend. You will probably first judge whether they liked it or not based on interpreting their emotional reactions when they unwrap it. Indeed, the accurate reading of emotional signals such as facial expressions (Lindner and Rosén, 2006
), body postures (de Gelder, de Borst, & Watson, 2015
) or speech prosody (Golan, Baron-Cohen, Hill, & Rutherford, 2007
) has been shown to be key for understanding the affective states of others. Until now, however, little attention has been placed in studying egocentricity during perception-based emotion attribution.
Previous research on the influence of mood on emotion perception provide a first indication of egocentric biases when reading other’s emotional states. Studies inducing positive and negative affective states to participants have shown that emotional facial expressions are more easily recognized when they are congruent with the participant’s induced mood (Lee, Ng, Tang, & Chan, 2008
; Niedenthal, Halberstadt, Margolin, & Innes-Ker, 2000
; Qiao-Tasserit, Quesada, Antico, Bavelier, Vuilleumier, & Pichon, 2017
; Schmid and Schmid Mast, 2010
). These mood-congruency effects have been often contextualized under general cognitive theories of affect congruence, according to which affective states activate linked memory representations and facilitate the encoding and processing of affectively-congruent information (Forgas, 2017
). However, mood-congruency effects could also reflect emotional egocentricity: they may be the result of an over-attribution of the own affective states to others. In line with this interpretation, biases in emotion perception seem to be stronger when the emotion experienced by the participant and the emotion expressed by the target are incongruent (Schmid and Schmid Mast, 2010
). Compared to classic emotional egocentricity paradigms, however, these biases may stem from more implicit and unconscious processes of self-projection, rather than reflecting self-other distinction abilities.
The current study sought to revisit mood-congruent biases in emotion perception as a measure of emotional egocentricity. First, we developed a novel approach to estimate the degree to which the own affective states bias judgments of emotional facial expressions. Using a combination of brief autobiographical recall and audiovisual stimuli, we induced happy, neutral and sad transient states to the participants. After each emotion induction, participants completed a short emotion perception task in which they made binary decisions (“happy” or “sad”?) about the expression of faces displaying a mixture of happiness and sadness. We hypothesized that emotional judgments would be biased by the participants’ affective states, such that they would more likely judge the ambiguous faces as happy when feeling happy than when being sad.
Second, we predicted that the magnitude of egocentric biases during the emotion judgments would be related to the participants’ disposition to consider and react to other people’s experiences. In particular, we examined associations with two components of dispositional empathy measured with the Interpersonal Reactivity Index (IRI; Davis, 1980
). On the one hand, the empathic concern scale taps into affective empathy and measures the tendency to react with feelings of sympathy and concern for unfortunate others (Davis, 1980
). On the other hand, the perspective-taking scale assesses the tendency to adopt the point of view of another person, a facet of cognitive empathy (Davis, 1980
). An association between mood-congruent biases and dispositional empathy would be an indication that the mood effects on emotion perception are related to processes of social cognition. Finally, we explored associations with autistic traits, as stronger egocentricity during cognitive mentalizing (Bradford, Hukker, Smith, & Ferguson, 2018
; Pearson, Ropar, & de Hamilton, 2013
) and deficits in emotion recognition (Uljarevic and Hamilton, 2013
) are commonly observed in autism spectrum conditions (ASC).
Research in social cognition has identified that one’s own emotional experiences are an important source of information to understand how another person is feeling. Previous studies detected egocentric biases when people make inferences about someone’s affective state based on information about the other’s context (Hoffmann et al., 2015
; Silani et al., 2013
; Steinbeis and Singer, 2014
). The goal of the current study was to investigate whether emotional egocentricity also occurs when affective inferences rely on reading the person's emotional expressions.
Using a combination of brief emotion induction blocks with psychophysical measures of emotion perception, we were able to detect the occurrence of egocentric biases when participants judged ambiguous emotional faces. As hypothesized, facial expressions were more readily classified as happy when participants reported feeling happy as compared to sad. These results are indicative of a tendency to project the own affective states when making inferences about others’ emotions. Moreover, we found an association between perspective-taking and the extent to which the own mood influenced the emotion judgments, which provides evidence that these egocentric biases are related to social cognitive abilities.
Our results replicate the mood-congruent biases in emotion perception documented in the literature. One limitation of previous studies is that they could not disentangle whether the observed effects were due to the influence of the affective state experienced by the participant, or due to more general framing or priming effects, whereby exposure to affective stimuli may increase the readiness to process cues of the same valence. Evidence for perceptual biases provided in previous emotion induction studies consisted on comparing emotion recognition between groups exposed to induction of different affective states (Lawrie, Jackson, & Phillips, 2019
; Lee et al., 2008
; Niedenthal et al., 2000
, Niedenthal, Brauer, Halberstadt, & Innes-Ker, 2001
; Schmid and Schmid Mast, 2010
). Although most studies included manipulation checks to demonstrate that groups differed in the experienced mood, statistical analyses could not rule out the possibility that the perceptual biases were caused by the mere exposure to an affective context, beyond whether or not this elicited an affective state to the participant. For example, Aguado, Martínez-García, Solís-Olce, Dieguez-Risco, & Hinojosa (2018
) showed that perception of emotional expressions is enhanced when faces are introduced by a statement describing events that elicit affectively congruent emotions (e.g., angry faces are recognized faster when primed with the sentence "He notices someone has vandalized his car."). Arguably, this manipulation did not lead to significant changes in the participant's affective state, yet it elicited similar perceptual biases to those in mood induction studies.
Our study addressed this limitation by additionally using a statistical model that tested whether the participant's reported mood, instead of the emotion condition, predicted the PSEs. This statistical analysis allowed us to account for the inter-subject variability in mood ratings within conditions, which increased the power to detect any influences of affective state. The observed relationship between affective state and emotion perception (i.e., the happier the participant was, the more likely they perceived the faces as happy) provides a more direct support to the existence of mood-congruency effects, and strengthens the results from the ANOVA analyses, which despite of showing an overall effect of emotion condition on PSEs, post-hoc pairwise comparisons between conditions did not reach statistical significance.
The observation of egocentric biases extends previous literature on emotional egocentricity by showing that self-projection also occurs during perception-based emotion attribution. Relying on self-knowledge for understanding other's mental states can be an efficient heuristic, especially when only limited information about the other is available (Ready, Clark, Watson, & Westerhouse, 2000
). Given that in our study participants had to make quick judgments about ambiguous emotional expressions, the own affect may have been used to guide their decisions. In fact, judgments of ambiguous faces, rather than clearer emotional expressions, have been shown to be more influenced by the participant's affective state (Cavanagh and Geisler, 2006
). Future studies should assess to what extend individuals attribute their own affective states when making emotion judgments in more naturalistic situations in which additional contextual information and more time to correct for egocentric projections are available.
Importantly, the processes underlying egocentric biases in this study may be distinct to those previously observed in classic emotional egocentricity paradigms. In those tasks, participants are asked to make emotion judgments about a person based on information about the affective stimulation the other is exposed to, while at the same time receiving an affectively (in-)congruent stimulation themselves (e.g., Silani et al., 2013
; von Mohr et al., 2019
). To unbiasedly infer the other’s feelings, participants need to disengage from their own experience and adopt the other’s perspective, a process that relies on self-other distinction abilities (Lamm, Bukowski, & Silani, 2016
; Steinbeis, 2016
). Under those conditions, emotional egocentric biases have been interpreted as a failure to differentiate between the representations of one’s own affective states and those of the other (Hoffmann et al., 2016
; Silani et al., 2013
; Tomova et al., 2014
In our paradigm, participants were not primed to think about their own affective state while judging the other’s emotional expressions, nor were asked to switch between self and other processing. In this context, the intrusion of the self-affect was more implicit and unconscious, and participants may have not actively tried to inhibit the influence of their own state nor tease apart the self- and other-representations as in previous emotional egocentricity tasks. Therefore, while our results support the idea that the own experience is recruited when reading others’ emotions, egocentric biases here should not be interpreted as an index of the participants’ self-other distinction abilities.
Instead, egocentric judgments during emotion reading could be related to the participants’ general disposition to shift attention towards the other’s experience during social interactions. Specifically, we found a small but significant correlation between emotional egocentricity and individual differences in the perspective-taking scale of the IRI, such that individuals with a higher predisposition to adopt the point of view of others were less influenced by their own affective states. This goes in line with results of a recent meta-analytic study that found a positive association between dispositional perspective-taking and emotion recognition accuracy (Israelashvili, Sauter, & Fischer, 2019
). A higher tendency to engage in perspective-taking may lead to more attention deployed to the other during processes of mental state inference, thus minimizing the influence of the own affect and facilitating a more accurate representation of the other’s experience. Supporting this hypothesis, perspective-taking tendencies have been related to the extent to which participants focus on the other person’s perspective relative to their own in a visual perspective-taking task (Bukowski and Samson, 2017
). Egocentric biases in our study may have thus partly resulted from a lower disposition to amplify the representation of the other, rather than a failure in inhibiting the self-representation.
While our results suggest that components of cognitive empathy are linked to the tendency to project one’s own emotions onto others, we did not find evidence for an involvement of affective empathy. Specifically, individual differences in empathic concern did not significantly correlate with the degree of emotional egocentricity. Though conclusions from null results should be drawn cautiously, our finding parallels previous studies in which emotional egocentricity was also not associated with empathic concern (Hoffmann et al., 2016
) nor with other factors related to affect sharing such as alexithymia (von Mohr et al., 2019
). Instead, emotional empathy seems to be more related to altercentric biases, that is, to the influence of the other’s emotions on the judgment of our own affective states (Hoffmann et al., 2016
Finally, we did not find evidence for an association between emotional egocentricity and autistic traits. Even though stronger egocentric biases have been reported in ASC (Bradford et al., 2018
; Pearson et al., 2013
), as well as in individuals from the general population with high autistic traits (Brunyé et al., 2012
), these were mainly detected in cognitive mentalizing tasks. The only study known to us that specifically investigated emotional egocentricity did not find differences in the magnitude of egocentric biases between individuals with ASC and controls (Hoffmann et al., 2016
). Taken together, these findings point to a dissociation in the impact of autistic traits on cognitive and affective mentalizing: while people with high autistic traits may show difficulties when inferring others’ knowledge and beliefs, autistic traits do not seem to significantly impact the capacity to overcome self-projections during emotion inferences.
Some limitations of the study should be mentioned. First, only face morphs of one female identity were used in the emotion perception task. Previous research has shown that factors such as liking or the perceived similarity with the other influence the degree to which people project their own mental states (Davis, 2017
; O’Brien and Ellsworth, 2012
). As such, one could expect that female participants in our study would have shown stronger egocentric biases than male participants as they were rating same-gender targets. Our exploratory analysis did not show significant gender differences in the influence of affective state on emotion judgments. Nevertheless, given our sample size, this result should be interpreted with caution, as we had limited power to test the moderation by gender. Moreover, due to our task design, we could not have distinguished whether potential gender effects would reflect in-group biases, different motivational attitudes towards the target, or general gender differences in emotion attribution. By including more diversity of face identities and controlling for the perceived similarity with the target, future studies should address this issue and improve generalizability of the findings.
Second, participants made binary emotion choices in the emotion perception task, while some of the morphs may have actually been perceived neither as happy nor sad. The 2-alternative forced-choice design was chosen to increase the likelihood to detect egocentricity effects and limit the task duration. To increase ecological validity, new emotional egocentricity paradigms could offer a wider range of emotion options. In addition, future task designs would benefit from adding a control condition in which participants make non-emotion judgments about the target while being under the emotion induction effects. This would allow to draw more definite conclusions about the specificity of the mood effects in relation to emotion reading vs. other mood-congruent perceptual biases unrelated to social cognitive processes.
Finally, although on average the emotion induction procedure worked, there was variability in the effectiveness of the manipulation, with some participants not reporting the expected mood in each condition. Moreover, even though the assessment of PSEs was completed within a short period of time (approx. 1.5 min) right after each emotion induction, we cannot rule out that the strength of the induced affect decreased throughout the task. Attenuated affective self-representations may have reduced the chance of observing a bias. In future work, the use personalized and longer-lasting forms of emotion induction could facilitate the detection of mood-congruency effects.
Notwithstanding these limitations, our study implemented a novel approach to quantify egocentric biases during emotion attribution. The adaptive psychophysical task allowed us to detect subtle changes in the tendency to perceive emotional faces as happy when participants were in different affective states. Unlike some of the previous mood-congruence studies (e.g., Harris et al., 2016
; Lee et al., 2008
; Niedenthal et al., 2000
; Schmid and Schmid Mast, 2010
), we used a within-subject design, which gave us the possibility to estimate individual bias scores and explore associations with socioemotional traits. The simplicity of the paradigm makes it appropriate to be used in emotional egocentricity research with clinical samples.
In conclusion, the current study established the existence of egocentric biases when reading facial expressions of emotion. We showed that individuals are more likely to perceive ambiguous happy-sad face morphs as happy when they are feeling happy as compared to when they are sad. More importantly, the magnitude of the egocentric bias was associated with perspective-taking tendencies, which suggests that socio-cognitive processes may underlie mood-congruency biases in emotion perception. Our study extends the literature on emotional egocentricity by showing that self-projection also occurs when we rely on the other’s emotional expressions for understanding their affective state.