Trait self-esteem and neural activities related to self-evaluation and social feedback

Self-esteem has been associated with neural responses to self-reflection and attitude toward social feedback but in different brain regions. The distinct associations might arise from different tasks or task-related attitudes in the previous studies. The current study aimed to clarify these by investigating the association between self-esteem and neural responses to evaluation of one’s own personality traits and of others’ opinion about one’s own personality traits. We scanned 25 college students using functional MRI during evaluation of oneself or evaluation of social feedback. Trait self-esteem was measured using the Rosenberg self-esteem scale after scanning. Whole-brain regression analyses revealed that trait self-esteem was associated with the bilateral orbitofrontal activity during evaluation of one’s own positive traits but with activities in the medial prefrontal cortex, posterior cingulate, and occipital cortices during evaluation of positive social feedback. Our findings suggest that trait self-esteem modulates the degree of both affective processes in the orbitofrontal cortex during self-reflection and cognitive processes in the medial prefrontal cortex during evaluation of social feedback.

effect of different cultural samples is also possible given the substantial evidence of cultural influences on brain activity involved in multiple cognitive and affective processes 13,14 . To clarify these, the present study recruited the same cultural sample (i.e., Chinese) and employed the same evaluation task. We tested whether and how one's self-esteem is associated with the neural activity during evaluations of one's own personality trait and evaluation of others' opinion about one's own personality its. During fMRI scanning, participants were asked to reflect on the self or a celebrity and to reflect on social feedback to the self or to a celebrity by responding on a 4-point scale. Neural activity underlying the reflection tasks was estimated by contrasting reflection on the self versus other or by contrasting reflection on social feedback about the self or a celebrity. Neural activity related to attitude was defined by regressing brain activity to participants' responses during self-reflection and during judgments on social feedback. This design allowed us to examine whether trait self-esteem can be associated with neural activity related to both self-evaluation and social feedback. Moreover, as trait self-esteem was defined as the tendency to evaluate oneself positively rather than negatively 15 , we were also interested in the associations between self-esteem and the neural activity related to both evaluation of positive traits of the self and attitudes toward the positive traits of the self.
Self-esteem is considered to be a relatively enduring characteristic that possesses both affective and cognitive components 16 . The way people represent themselves in comparison with others, and the role of affective processes in such representations are matters of significant interest to a social cognitive and affective neuroscience of trait self-esteem 11 . On the one hand, research on self-esteem that focused on associated intrapersonal experiences (self-evaluation) defined self-esteem as one's feeling about the self 17 . Self-esteem, at its roots, is an affective, top-down, internal experience. That is, people feel either good or bad about the self (affect), which then guides evaluation (cognition), and drives action (behavior) 16 . On the other hand, there is also a long history of accounting for self-esteem's interpersonal influences 18 . From this perspective, people processed others' reactions of inclusion and exclusion in an automatic cognitive mechanism 5 . In other words, how an individual believes that he or she is viewed in the eyes of others is critical in forming of self-esteem. As outlined above, we assumed that neural activity in brain regions involved in affective process would be associated with self-esteem during the self-evaluation process, whereas neural activity in brain regions involved in cognitive process would be associated with self-esteem during the social feedback process.

Materials and Methods
Participants. Twenty-nine healthy college students (15 males) participated in this study as paid volunteers (mean age ± SD = 20.1 ± 1.4). All reported no history of psychiatric or neurological disorders, significant physical illness, head injury, or alcohol/drug use, were recruited from Southwest University (Chongqing, China). Four participants were excluded from data analysis due to excessive head motion during scanning and 25 participants (13 males) were included in fMRI data analysis [mean age ± SD = 20.1 ± 1.3]. All participants gave informed consent prior to scanning. The experiments were carried out in accordance with relevant guidelines and regulations. The protocol for this study was approved by the Ethics Committee of Southwest University.
Questionnaire. Before scanning, participants completed the Rosenberg self-esteem scale (RSE) 7 to assess their overall evaluation of self-worth. The RSE consists of 10 items such as 'I feel I do not have much to be proud of ' or 'On the whole, I am satisfied with myself ' , which are coded on a 4-point scale ranging from 1 (strongly disagree) to 4 (strongly agree). Negative items were reversed scored.

Stimulus and Procedure.
To create a context of social feedback for scanning, we first asked 120 students to evaluate five of their classmates as well as a celebrity (i.e., Xiang Liu, a well-known Chinese athlete) on 10 personality-trait adjectives (1 = strongly disagree; 4 = strongly agree). Twenty-nine of these students were then randomly recruited for fMRI scanning. 336 personality-trait adjectives (half positive and half negative) were selected from an established personality trait adjective pool 19 for fMRI scanning. Each adjective consists of two Chinese characters. Stimuli were presented through a projector onto a rear-projection screen located at the subject's head.
A mixed blocked/event-related design was used in 7 functional scans. Each scan consisted of 8 blocks of trials and 2 successive blocks were intervened by a 10-s grey screen. Each block consisted of 6 trials. On each trial a black trait adjective was presented for 4 s at the center of a grey screen, which was followed by a fixation with a duration of 2, 4, or 6 s. Experimental procedure and experimental conditions used in the fMRI study are listed in Fig. 1. There were 4 different tasks and participants performed one task in each block of trials. The self-evaluation task asked participants to evaluate "Does this adjective describe the self?" by pressing on of 4 buttons (1 = not at all like me, 4 = most like me). The celebrity-evaluation task asked participants to indicate "Does this adjective describe Xiang Liu?" During the task of evaluating feedback on the self, participants were told that they were presented with a number of trait adjectives that were used by their classmates to describe them. They had to indicate "Do you agree with others' evaluation on the self?" by a button press (1 = strongly disagree, 4 = strongly agree) on each trait adjective. During the task of evaluating others' feedback on the celebrity, participants were told that the celebrity had been described by their classmates using a number of trait adjective and they had to indicate "Do you agree with others' evaluation on Xiang Liu?" by a button press (1 = strongly disagree, 4 = strongly agree) on each trait adjective. There were 42 positive and 42 negative adjectives for each task. fMRI Data Analysis. Data were analyzed using Brain Voyager QX v2.3 (Brain Innovation, The Netherlands).
Functional scans were realigned within and across runs to correct for head motion, and co-registered with each participant's anatomical data. Functional data were then normalized into standard stereotactic Talairach space, resliced into a voxel size of 3 × 3 × 3 mm 3 and smoothed with an 8 mm Gaussian kernel to increase signal-to-noise ratio. Event-related effects were estimated using the general linear model and employing a canonical hemodynamic response function convolved with the experimental design. Fixed effect analyses were first performed to estimate effect at each voxel and to compare regionally specific effects in individual participants using linear contrast. Group analyses were then conducted using random-effects models to enable population inferences. Inference of statistical significance used uncorrected p value p < 0.005, in regions encompassing at least 20 voxels 20 .
Modeling of self-related contrasts during the self-evaluation task. Brain activations associated with evaluation of one's own traits was estimated by contrasting (EPS + ENS) versus (EPC + ENC). The contrast of (EPS−EPC) versus (ENS-ENC) was calculated to define brain regions involved in evaluation of positive traits of the self. Moreover, to identify whether participants' trait self-esteem can modulate their brain activations related to evaluation of one's own traits, self-esteem scores derived from the RSE questionnaire were entered as a regressor in a whole-brain regression analysis to assess its associations with the contrast value of (EPS + ENS) versus (EPC + ENC) or (EPS-EPC) versus (ENS-ENC), respectively.
Further, brain activations related to participants' attitude about the self were estimated by regressing participants' rating of each trait adjective on a 4-point scale (1 = strongly disagree, 4 = strongly agree). Brain activations that showed linear relationships with increasing rating in evaluating traits of the self were calculated. The contrasts of (EPS + ENS) versus (EPC + ENC) or (EPS-EPC) versus (ENS-ENC) were conducted to assess brain activations related to attitudes toward the self or attitudes toward the positive traits of the self, respectively. Moreover, to identify whether people's trait self-esteem could modulate the brain regions that showed a linear relationship with increasing rating in evaluating traits of the self or positive traits of the self, the self-esteem scores derived from the RSE questionnaire were entered as a regressor in a whole-brain regression analysis to assess its associations with the contrast value of (EPS + ENS) versus (EPC + ENC) or (EPS-EPC) versus (ENS-ENC), respectively.
Modeling of self-related contrasts during the social feedback task. Brain activations related to evaluation of others' feedback on the self was estimated by contrasting (EPFS + ENFS) versus (EPFC + ENFC). The contrast of (EPFS-EPFC) versus (ENFS-ENFC) was calculated to define brain regions engaged in evaluation of others' positive feedback on the self. Moreover, to identify whether participants' trait self-esteem can modulate their brain activations involved in evaluation of social feedback on the self or positive social feedback on the self, a wholebrain regression analysis of the contrast value of (EPFS + ENFS) versus (EPFC + ENFC) or the contrast value of (EPFS-EPFC) versus (ENFS-ENFC) were conducted with self-esteem score as a regressor.
Further, brain activations related to participants' attitude about social feedback were estimated by regressing participants' rating of each trait adjective on a 4-point scale (1 = strongly disagree, 4 = strongly agree). The contrast of (EPFS + ENFS) versus (EPFC + ENFC) was then conducted to assess brain activations related to attitudes toward the social feedback on the self. In addition the contrast of (EPFS-EPFC) versus (ENFS-ENFC) was used in the regression analyses to examine brain activations related to attitude toward positive social feedback on the self. Moreover, to identify whether people's trait self-esteem could modulate their brain regions that showed a linear relationship with increasing rating in evaluating other's feedback on the self or other's positive feedback on the self, we regressed self-esteem score with the contrast value of (EPFS + ENFS) versus (EPFC + ENFC) or the contrast value of (EPFS-EPFC) versus (ENFS-ENFC), respectively.

Results
Behavioral Performances. Behavioral data from four participants who had been excluded from fMRI data analyses due to head motion were also excluded from the behavioral data analyses. A 2 (Task: self-evaluation versus social feedback) × 2 (Target: self versus celebrity) × 2 (Valence: positive versus negative trait adjectives) repeated measures analysis of variance (ANOVA) showed a significant interaction of Task × Target on response speed, F (1, 24) = 7.55, p < 0.05. Post hoc analyses revealed that participants responded faster during evaluation of the self compared to the celebrity (p < 0.01), whereas response speeds did not differ between evaluations of the social feedback on the self and the celebrity (p = 0.51) ( Fig. 2A).
Judgments were then collapsed into high (3 and 4 responses) and low (1 and 2 responses) self-relevance categories and considered into a 2 (Task: self-evaluation versus social feedback) × 2 (Target: self versus celebrity) × 2 (Valence: positive versus negative trait adjectives) × 2 (Response: high versus low) ANOVA. Results revealed a significant Response × Valence interaction, F (1, 24) = 87.44, p < 0.01, due to that participants were more likely to endorse positive information as high in self-relevance and negative information as low in self-relevance (p < 0.05). Results also revealed a trend toward four way interaction effect, F (1, 24) = 3.92, p = 0.059. Separate analyses showed a significant Target × Response interaction during the self-evaluation task, F (1, 24) = 13.66, p < 0.01, suggesting that participants made significantly more high response than low response to self (p < 0.05), and that they made significantly more low response than high response to celebrity (p < 0.05, Fig. 2B). However, such pattern failed to present during the social feedback task, F (1, 24) = 1.73, p > 0.1 (see Fig. 2C).
Further, a whole-brain regression analysis of the neural activity in response to rating one's own traits compared to those of the celebrity did not show any significant activation. A whole-brain regression analyses of attitude-related neural activity with self-esteem rating score as a regressor revealed significant activations in the bilateral OFC in responses to evaluation of positive traits of the self versus the celebrity (Left: − 28/60/− 2, r = 0.73, k = 58; Right: 40/61/3, r = 0.69, k = 32) (Fig. 3 ).  (Table 2). However, people's trait self-esteem did not correlate with the neural activity related to social feedback on oneself.   Table 1. Association between self-esteem and the neural activity related to the self during the selfevaluation task.
Further, a whole-brain regression analysis of the neural activity in response to the evaluation of social feedback to the self versus the celebrity revealed a significant activation in the right caudate (22/− 14/29, t = 3.49, k = 20). Meanwhile, a whole-brain regression analyses of attitude-related neural activity with self-esteem rating score as a regressor revealed significant activations in the ventral medial prefrontal cortex (mPFC: 9/53/3, r = 0.66, k = 25), PCC (− 36/31/24, r = 0.63, k = 24) and occipital cortex (6/− 92/− 5, r = 0.63, k = 73) in responses to evaluation of positive social feedback to the self versus the celebrity (Fig. 4).

Discussion
There has been less than perfect agreement within the psychological literature on the nature of self-esteem in terms of intrapersonal versus interpersonal perspectives 1,4 , and affective versus cognitive processes 16,[21][22][23] . The current work examined whether and how neural activity related to self-evaluation and social feedback can be related to one's trait self-esteem and whether and how trait self-esteem can be associated with the neural activity related to both task and attitude. Consistent with our first hypothesis, people's trait self-esteem was positively correlated with the intrapersonal processing in OFC which has been shown to support affective processes; and consistent with our second hypothesis, their trait self-esteem was positively correlated with the interpersonal  processing in mPFC/PCC which supports cognitive processes. Moreover, our fMRI results suggested that trait self-esteem predicted the task-related neural activity in the middle frontal gyrus, inferior temporal gyrus and middle temporal gyrus in response to evaluation of one's own positive traits compared to those of a celebrity. Interestingly, one's self-esteem was positively associated with the affective-related neural activity in bilateral OFC, which was involved in evaluation of positive traits of the self. The orbitofrontal cortex (OFC) is an important part of the network involved in emotional processing because of its neuroanatomical connectivity with affective regions such as the amygdala, cingulate cortex, and insula [24][25][26] . Some studies have even suggested that OFC can be viewed as part of a global workspace for evaluating the affective valence of stimuli 27,28 . Numerous studies have shown OFC activations during affective processing, such as when receiving pleasant and painful touches 29 . OFC activation was also correlated with the amount of money received/lost in a probabilistic visual association task 30 . Damage to the OFC in humans may preclude the generation of helpful emotional information 31 , which may be associated with impairments in emotional and social behavior characterized by social inappropriateness and irresponsibility. Self-esteem is an affectively laden self-evaluation from the intrapersonal view 5 and at its core, self-esteem refers to how we feel about ourselves and is inherently rooted in affective processes from the affective model of self-esteem 17,32 . Rather than being based solely on cognitive self-evaluations, self-esteem involves affective processes that may or may not be related to specific, conscious self-evaluation 5 . Therefore, the activation of OFC may be also involved in affective processing and was associated with people's self-esteem during the self-evaluation task.
Our study also showed evidence that trait self-esteem can be also positively related to the cognitive-related neural activity in the medial prefrontal/posterior cingulate cortex during evaluation of positive social feedback about the self. Accumulating data suggests that conceiving a viewpoint of others (theory of mind), as a related form of self-projection, involves brain networks associated with the cognitive processing, including frontal lobe systems that are traditionally associated with planning, as well as medial temporal-parietal lobe systems that are associated with memory 33 . The sociometer theory proposes that self-esteem is essentially a psychological meter, or gauge, that monitors the quality of people's relationships with others 34 . It is a person's internal, subjective index or marker regarding the degree to which the individual is being included versus excluded by other people 4 . Thus self-esteem encompasses a cognitive processing in monitoring the relationship with others, from the interpersonal perspective. Moreover, trait self-esteem was also associated with activities in the occipital cortices during evaluation of positive social feedback, which was also demonstrated in a previous study showing that higher levels of state self-esteem were associated with greater activity in occipital cortex 9 .
Despite our assumption that the 'intrapersonal' and 'interpersonal' aspect of self-esteem are more affective and cognitive, respectively, we do not preclude the possibility that affective and cognitive processes are more or less present in both aspects. The only regions that associated with self-esteem were OFC and mPFC/PCC during the intrapersonal and interpersonal processes, respectively. However, other brain regions may also be involved by means of, e.g., connectivity, that was not captured in the present study. Furthermore, people's trait self-esteem was associated with the neural activity of regions in response to positive self-reflection and social feedback. Regions that previously found to be associated with emotional self-reflection, such as anterior insula and ACC 35,36 , did not associate with self-esteem. One possibility is that negative self-reflection and feedback-evaluation are complicated by emotion-regulation (e.g., distraction or reappraisal) and defensive mechanisms (e.g. denial), thus less directly linked to trait self-esteem.
In addition, both ACC and PCC exhibiting high levels of activity in response to evaluation of positive social feedback about the self is remarkably similar to the default mode network (DMN), in which the total cerebral mean blood flow and oxygen uptake remain constant from a restful state to an active state 37,38 . Previous studies suggested that the convergence of brain regions between the DMN and that which is activated during a cognitive state raises the possibility that default modes of cognition are characterized by a shift from perceiving the external world to internal modes of cognition that simulate worlds that are separate from the one being directly experienced 33 . For instance, due to the overlap in activity between regions that are involved in self-relatedness processing and DMN regions 39 , some speak of a so-called 'default self ' arguing that the self may be more or less identical with the resting state activity observed in DMN regions 37,40,41 . In the current study, since higher trait self-esteem was associated with significant activations in the cortical midline structures in responses to evaluation of positive social feedback about the self, we suggest that the constitution of the positive social feedback to the self in high self-esteem participants may rely on the internal resting state activity of the brain. In other words, one may then assume that processing positive social feedback in high self-esteem participants already occurs in those psychological processes associated with the brain's resting state activity.
Eisenberg et al. (2011) reported greater mPFC activation in response to negative feedback words in individuals with lower state self-esteem. In the present study, higher trait self-esteem was associated with more mPFC activation in response to positive social feedback traits. Given the mPFC activation is associated with reflection upon one's personal traits and encoding self-relevant information 42,43 , one may speculate that the mPFC activity may be more strongly related to the information that is being more deeply processed, independent of the valence of that information. Furthermore, the mPFC activity in response to the social feedback can predict one's selfesteem, independent of trait or state self-esteem.
Our fMRI results also showed that task-related neural activity in memory-related brain region such as the middle frontal gyrus, inferior frontal gyrus, precuneus, cuneus, parahippocampal cortex, middle temporal gyrus, superior temporal gyrus and middle occipital gyrus in response to evaluation of one's own traits compared to those of a celebrity could be positively related to one's trait self-esteem. It further showed a positive correlation between one's trait self-esteem and the neural activity in the middle frontal gyrus, inferior temporal gyrus and middle temporal gyrus related to evaluation of one's own positive traits. Given that the self is well-developed and often used construct that promotes elaboration and organization of encoded information 44 , personality traits processed with reference to the self are better remembered than information that is processed in other ways 44 . Thus, the fMRI results in the current work are consistent with the findings of behavioral studies by showing that neural activities in the memory related brain regions such as the middle frontal gyrus, inferior frontal gyrus, precuneus, parahippocampal cortex, and middle temporal gyrus are associated with trait self-esteem.
Several limitations of this study should be noted. First, state self-esteem could reflect the feelings about the self at any moment in time 45 . Participants in the past sociometer studies rated how they felt in response to seeing each feedback word 4,9 . A future study could explore whether state self-esteem could be associated with both the cognitive and affective brain network during explicit self-evaluation, and to further investigate whether the cognitive network and the affective network of the brain underlying people's perception of how they are evaluated by others would be similarly associated with state self-esteem. Second, Similar to most of the previous fMRI studies of self-evaluation, we used the celebrity-evaluation task as a control condition to rule out the impact of general evaluative processing and the impact of diversity in response selection. However, a participant might be more familiar with the self than the celebrity. This raised the issue of to what degree our results can be explained by the difference in familiarity. The previous studies have shown that coding of familiar others activated the dorsal MPFC and posterior cingulate 41 . Our results showed, however, that distinct associations between self-esteem and brain activity in responses to evaluation of the self and social feedback were evident in the OFC and medial prefrontal cortex. Although the brain activity in these brain regions was less likely to reflect the effect of familiarity, the posterior cingulate activity might be attributed to greater familiarity with the self than the celebrity. This should be clarified in the future research. Third, people learn their self-knowledge of personality through looking inward (e.g., introspection) and looking outward (e.g., feedback) 46 . Others (e.g., friends) know more than the self about aspects of personality that are observable (e.g., funny) and highly evaluative (e.g., attractive), while the self is more accurate than others for traits low in observability (e.g., neuroticism) 47,48 . It is likely that one's trait self-esteem can be associated with both social feedbacks from others and self-evaluation on traits that are high in observability or low in observability. Future research can classify the personality traits into different categories to investigate this interesting topic.
In summary, our neuroimaging findings indicate associations between trait self-esteem and neural activity related to reflection on oneself and evaluation of social feedback in brain regions that are associated with affective and cognitive processes, respectively. In addition, trait self-esteem can also modulate the neural activity related to evaluation tasks on the self (but not social feedback) in a brain network that is associated with memory processes.