Neural Correlates of Self

As part of the growing interest of understanding the neural basis of self-representation, we examined the self-reference effect in memory, in which knowledge about the self is typically remembered better than other types of semantic information. This line of research has been supported by grants from the National Science Foundation and the National Institute of Mental Health and has helped establish an important role for medial prefrontal cortex (MPFC) in processing stimulus materials with reference to self.

We conducted a series of studies demonstrating that (a) a region of medial prefrontal cortex (MPFC) was selectively engaged during self-referential processing (Kelley et al., 2002), (b) that MPFC activity is specific to self-referential processing and not to processing information about familiar and personally-known others (Heatherton et al., 2006), (c) that the memorial advantage afforded to self-knowledge appears to depend on the additional recruitment of MPFC (Macrae et al., 2004), and (d) that MPFC activity further indexes the self-relevance of information; MPFC activity increases as material is judged to be more self-descriptive (Moran et al., 2006; Moran et al., 2009).

A particular challenge, however, in linking MPFC activity to self-relevance is that self-relevance is often confounded with emotional valence. Indeed, when considering the personal relevance of information, individuals are more likely to endorse positive information as self-descriptive (Taylor & Brown, 1988). Thus, cortical activity during self-reflection may index the self-relevance of the material, the valence of the material, or an interaction of the two.  Moran et al., 2006 was designed to disentangle these possibilities and revealed a functional dissociation between MPFC and an adjacent region of the ventral anterior cingulate cortex (vACC). MPFC responded preferentially to material that was self-relevant, regardless of valence, whereas vACC responded preferentially to valence information, but only when this material was judged to be personally relevant. These findings suggest that distinct neural circuits in adjacent regions of prefrontal cortex subserve cognitive and emotional aspects of self-reflection. Whereas MPFC signifies the personal relevance of information, the emotional impact of self-relevant information may be distinguished in adjacent vACC.  This finding fits nicely with clinical research showing that the vACC is implicated in emotional disorders such as depression and PTSD (Drevets, et al., 1997; George et al., 1997).  In a particularly striking study, Mayberg and colleagues (2005) demonstrated that deep brain stimulation in vACC was effective in alleviating depression in treatment-resistant patients.

Given the fundamental importance of social feedback to a healthy sense of self, we became especially interested in whether vACC would similarly index peer acceptance and rejection, and whether such activity would be more pronounced in people with low versus high self-esteem. To explore these possibilities, we conducted two fMRI studies to measure neural activity while subjects viewed a series of unfamiliar faces (Somerville, Heatherton, and Kelley, 2006).  Several weeks prior to scanning, subjects were photographed and led to believe that individuals at other institutions would be forming impressions of them during this interim period. In the scanner, subjects viewed faces and were asked to either form a first impression of these ‘participants’ from other universities (i.e., “Do you think you would like this person?”) or to predict whether the ‘participant’ would accept or reject them (i.e., “Do you think this person would like you?”).  In both studies, subjects were given false feedback for some of the faces indicating that the person in the photo had previously formed a negative or positive first impression of the subject. This approach permitted a factorial analysis that examined neural responses specific to feedback as a function of expectancy violation (i.e., when feedback matched vs. did not match subjects’ first impressions or predictions) and social feedback (i.e., when feedback was negative vs. positive). Results revealed a double dissociation between dorsal and ventral ACC regions. Whereas a dACC region was sensitive to expectancy violations (incongruent > congruent), vACC was sensitive to social feedback (accepted > rejected).

Of course, people differ in how they view themselves, with some people generally favorable (i.e., those with high self-esteem) and others generally more negative (low self-esteem).  Importantly, this study also allowed us to assess the effect of self-esteem on response to social rejection (Somerville, Heatherton, & Kelley, 2010). There was a significant interaction, such that the strongest responses occurred for those with low self-esteem.

Our current and future work in this line of research investigates the extent to which the self is stable or malleable.  On the one hand, there is clear evidence for the stability of personality dispositions that shape the sense of self, and these change very little after early adulthood (Heatherton & Weinberger, 1994).  That is, if you have high self-esteem today, you will probably have high self-esteem tomorrow.  From this perspective, self-esteem is stable because it slowly builds over time through personal experiences, such as repeatedly succeeding at various tasks or continually being valued by significant others.  We are using recently developed applications of network analysis to assess resting-state functional connectivity (rs-fcMRI) and its relation to individual differences in personality. Network-based rs-fcMRI allows for the examination of functional coupling of brain networks, patterns of statistical coherence across brain regions that arise throughout development, in a manner that permits assessment of a network’s integrity (i.e., strength of connections between nodes in the network). When subjects are not performing an explicit task, coherent activity within several separable and reproducible brain networks can be identified. A key element of emphasis here is the discovery of functional brain networks— sets of brain regions whose spontaneous activity correlates at rest (i.e., in the absence of explicit task-constraints). Reproducibility of functional networks across individuals and laboratories suggests that resting state connectivity has a common architecture, and yet there are also marked individual differences in the strength of connectivity within networks that are likely related to important behavioral outcomes. One of these networks is the default network, which includes regions of the MPFC, PCC, and vACC.  Preliminary findings from a database of resting-state scans on more than 500 subjects collected in our laboratory demonstrate that rs-fcMRI within default network can accurately predict individual differences in self-esteem.

On the other hand, it is well known that behavior can also be influenced temporarily by situational factors that lead to momentary changes in self-evaluation.  Thus, many theorists have distinguished between long-lasting traits and temporary states, such as between trait self-esteem and state self-esteem (Heatherton & Polivy, 1991).  Around a stable baseline are fluctuations; although we might generally feel good about ourselves, there are times when we may experience self-doubt and even dislike.  Additional ongoing studies investigate whether temporary fluctuations in state self-esteem engage the same neural architecture as the more enduring trait self-esteem system.  By inducing a state of lowered self-esteem and using hybrid state-item fMRI paradigms, we will be able to examine an analogue of negative mental states that accompany many mental health disorders, such as depression.  We have successfully used similar state-item paradigms in studies of the effect of negative mood on reward-related activity to food cues in dieters (Wagner et al., 2012) and in differentiating transient and sustained neural signals underlying anxious emotion (Somerville et al., 2013).