When I think about me and simulate you: Medial rostral prefrontal cortex and self-referential processes

Abstract

While neuroimaging studies implicate medial rostral prefrontal cortex (mrPFC) in self-referential processing, simulation accounts of social cognition suggest that this region also supports thinking about other people. This study tested the prediction that mrPFC might be involved in appraising the personality traits of another person to the degree that this person is perceived as similar to oneself. We also examined whether recruiting common processes for thinking about oneself and others might impact on subsequent memory for those judgments. Functional MRI was used while two factors were crossed: (i) the requirement to engage in personality trait or episodic source memory judgments and (ii) the reference for these judgments (i.e., oneself or a friend). The results link haemodynamic changes in mrPFC to both personality judgments about oneself and subsequent episodic memory retrieval of these judgments. The degree to which BOLD signal in this region was also associated with thinking about others correlated with perceived similarity in both tasks, thus corroborating simulation accounts. Moreover, participants who perceived themselves as having similar traits to their friends tended to be poorer at remembering whether they had made trait judgments in reference to themselves or their friend. This behavioral effect was reflected in the BOLD signal in mrPFC: there was a positive correlation between signal change for self versus friend judgments and subsequent memory for the reference of such judgments. The results suggest that investigations of mrPFC activity in the context of self/other judgments should take into account this psychological similarity effect.

Introduction

An important feature of human experience is to conceive of the self as a more or less coherent entity that persists across time (Gallagher, 2000, Conway and Pleydell-Pearce, 2000). Therefore, one is capable of attributing personality traits to oneself or to identify recollected episodes as memories of one's past. These are examples of self-referential processing. That is, in both cases, the processing subject (i.e., the self) refers to itself.

In recent years, the medial rostral prefrontal cortex (mrPFC; approximating medial aspects of Area 10) has been implicated in a variety of tasks thought to require self-referential processing (Northoff et al., 2006). Some of these asked participants to judge whether personality trait words are descriptive of oneself, and mrPFC activation was stronger when traits were deemed to be more self-relevant (Macrae et al., 2004, Moran et al., 2006). Furthermore, this region tends to be more strongly activated during such judgments about oneself than about either close friends (D'Argembeau et al., 2007, Heatherton et al., 2006) or famous others (Kelley et al., 2002, D'Argembeau et al., 2005).

Though these studies compellingly link mrPFC with self-referential processing, this region is not exclusively recruited while thinking about oneself. In fact, there is some evidence for comparable activation during trait attributions about either oneself or other people (Ochsner et al., 2005, Schmitz et al., 2004). Therefore, a goal of the present study was to characterize the degree to which mrPFC might also be involved in appraising others. Specifically, predictions from simulation theory were tested: this account posits that one can make inferences about others by referring to knowledge about oneself (Gallese and Goldman, 1998, Frith and Frith, 1999, Mitchell et al., 2005). For example, one could use one's thoughts and feelings to extrapolate what another person might think and feel. Such a strategy, however, is only valid in so far as the self is a good model of the other person. That is, one has to assume that this person is reasonably similar to oneself. In such cases, self-referential processes subserved by mrPFC might also be involved in trait judgments about others (Mitchell et al., 2005). Consistent with this idea, Mitchell et al. (2006) reported more similar activation of this region during mental state inferences about oneself and an unknown person, if this unknown person was thought of as more politically like-minded.

The present study tested whether mrPFC is also more strongly engaged during trait judgments about personally known others in cases of greater perceived similarity. Moreover, we hypothesized that recruiting self-referential processes for similar others may impact on performance on a subsequent episodic source memory task: if trait attributions about the self and a similar friend are subserved by common processes, these shared encoding operations might yield overlapping memory representations for both judgment types (Mashek et al., 2003, Tulving and Thomson, 1973). Hence, remembering the reference of the judgments was expected to be more difficult in case of greater similarity.

The episodic memory task further served the second goal of this study: an integral part of remembering an episode from one's past is the feeling that this memory refers to oneself (Tulving, 2002). Yet, this sense can differ between conditions, and mrPFC has been especially associated with the recollection of memories that entail a pronounced self-referential component (Cabeza et al., 2004, Simons et al., 2008). For example, Hassabis et al. (2007) contrasted recollection of autobiographical experiences and of previously imagined, fictional experiences. Whereas both conditions similarly recruited a widespread network, recollection of autobiographical episodes yielded stronger mrPFC activation. The authors argued that both tasks involved a common process of scene construction. Selective mrPFC recruitment for autobiographical events, in contrast, was proposed to reflect greater self-referential processing of these real-life memories.

To date, the link between activation in this region during episodic memory retrieval and self-referential processing is primarily supported by studies such as those investigating trait judgments (but see also Summerfield et al., 2009). The present study allowed for a more direct test of this account: if mrPFC activation reflects the great self-referential component of recalled memories, this area should also be involved in the successful retrieval of trait judgments about oneself. Moreover, consistent with simulation theory, this region should further be recruited during the successful retrieval of trait judgments about others to the degree that the other person is perceived as similar.

The only two previous neuroimaging studies on the retrieval of self-referential trait judgments indeed observed mrPFC activation (Fossati et al., 2004, Lou et al., 2004). In both cases, however, the task used to probe memory did not require any recollection of the actual study episodes (Yonelinas, 2001). Thus, they did not assess whether recruitment of mrPFC was specifically associated with the successful retrieval of self-related information.

Taken together, the present study investigated the contribution of mrPFC to self and other trait judgments and to their subsequent retrieval from memory. Two factors were crossed: (i) the requirement to engage in personality trait or episodic memory judgments and (ii) the reference for these judgments (i.e., oneself or a friend). Specifically, participants alternated between study and test phases, while being scanned by functional magnetic resonance imaging (fMRI). During study, they judged the degree to which personality trait words were descriptive of either themselves (self study condition) or a close friend (other study condition). In a non-semantic control condition, they specified the number of the words' syllables (syllables study condition). During test, traits were presented that comprised words studied in all conditions plus new words. In the self test condition, participants indicated whether they had judged themselves with these traits. Thus, this condition required retrieval of self-referential information. In the other test condition, in contrast, they indicated whether they had previously made a judgment with reference to their friend. Accordingly, in the syllables test condition participants decided whether they had performed the syllables task. Finally, the perceived similarity to the friend was assessed.

This design addressed the two hypotheses. First, if self-referential processes are more strongly engaged for trait judgments about more similar others, three findings were predicted: foremost, any difference in mrPFC activation between the self and other study condition should be smaller for those participants who perceive their respective friend as more alike. Furthermore, due to shared encoding operations, it should be more difficult to correctly remember the reference of trait judgments about oneself and a more similar other person. Finally, areas specifically implicated in retrieving self-referential information should be more strongly recruited while retrieving judgments about others in cases of greater similarity. This prediction is further specified by the second hypothesis: mrPFC activation is expected for successful retrieval of self-referential information.