Reflections of Oneself: Neurocognitive Evidence for Dissociable Forms of Self-Referential Recollection

Abstract

Research links the medial prefrontal cortex (mPFC) with a number of social cognitive processes that involve reflecting on oneself and other people. Here, we investigated how mPFC might support the ability to recollect information about oneself and others relating to previous experiences. Participants judged whether they had previously related stimuli conceptually to themselves or someone else, or whether they or another agent had performed actions. We uncovered a functional distinction between dorsal and ventral mPFC subregions based on information retrieved from episodic long-term memory. The dorsal mPFC was generally activated when participants attempted to retrieve social information about themselves and others, regardless of whether this information concerned the conceptual or agentic self or other. In contrast, a role was discerned for ventral mPFC during conceptual but not agentic self-referential recollection, indicating specific involvement in retrieving memories related to self-concept rather than bodily self. A subsequent recognition test for new items that had been presented during the recollection task found that conceptual and agentic recollection attempts resulted in differential incidental encoding of new information. Thus, we reveal converging fMRI and behavioral evidence for distinct neurocognitive forms of self-referential recollection, highlighting that conceptual and bodily aspects of self-reflection can be dissociated.

Keywords: episodic retrieval; fMRI; medial PFC; self; social cognition.

Figure 1.

Stimuli examples in the study (left column) and test (right column) phases. In the study phase, a symbol at the top of the screen indicated whether the participant (a “plain” face) or the experimenter (a face resembling the experimenter) should speak the word out loud. A symbol at the bottom of the screen indicated whether participants should judge how well the word reflected themselves (pointing hand) or the US President Obama (the “Obama 2008” campaign logo). In the test phase, a question at the top of the screen indicated to participants whether they should remember who had spoken the word at study (Agentic recollection), remember who the word had been related to at study (Conceptual recollection), or make a nonepisodic Control judgment. Top left: a word spoken by the participant at study (Subject) that they also related to themselves (You). Top right: the same word tested with the Conceptual recollection question. Bottom left: a word spoken by the Experimenter at study that the participant related to Obama. Bottom right: a new word tested with the Agentic recollection question.

Figure 2.

fMRI activations in the mPFC associated with self/other recollection. Effects in A and B are thresholded at P < 0.001 (uncorrected), with a minimum cluster size of 10 voxels, and inclusively masked to display only activations within the mPFC region associated with social cognition in Amodio and Frith (2006). Effects in (C) are thresholded at P < 0.05 family-wise error corrected for the whole brain, with a minimum cluster size of 10 voxels. The percent signal change bar graphs (A and B) plot the mean difference between each displayed condition and the nonepisodic Control task extracted from the peak voxel in each mPFC cluster. (A) A dorsal mPFC region with a peak at [−12, 59, 25] showed enhanced activation for both old and new items in both recollection tasks, compared with the Control condition. (B) a ventral mPFC region with a peak at [−9, 53, 13] showed selective activation for old items that participants had processed in relation to their conceptual self during study, and only when the retrieval task required recollection of conceptual self/other information. (C) In a whole-brain analysis, general old > new effects (old > new collapsed across retrieval task; red) were associated with a very different activation pattern from episodic retrieval task effects (episodic tasks > Control task; green), except in the precuneus where the 2 effects overlapped.