UCLA

The social world is incredibly complex and the ability to keep track of various pieces of social information at once is imperative for success as a social species. Yet, how humans manage social information in mind has to date remained a mystery. On the one hand, psychological models of working memory, or the ability to maintain and manipulate information in mind, suggest that managing social information in mind would rely on generic working memory processes. However, recent research in social neuroscience questions this possibility, as the same neural system, the medial frontoparietal system, that supports basic social cognitive processing disengages during working memory (McKiernan et al., 2003). In fact, a failure to decrease the medial frontoparietal system during working memory actually interferes with working memory performance (Anticevic et al., 2010). However, prior to this dissertation, no research had examined the neural mechanisms supporting the ability to maintain and manipulate social cognitive information in mind, or engage in `social working memory' (SWM). Thus, the goal of the present dissertation was to examine, for the first time, the extent to which SWM and non-social, cognitive working memory (CWM) rely on the same or different underlying neural mechanisms.

Paper 1 employed a novel SWM paradigm to examine which neural regions support SWM. In a within-subject design, participants (N=16) completed SWM trials that required them to consider two, three, or four of their friends along a trait dimension during a delay period while undergoing functional magnetic resonance imaging (fMRI). Linear increases in neural activity as a function of the number of friends considered (a response pattern characteristic of working memory systems (Rypma et al., 1999)) were found in two neurocognitive networks: the lateral frontoparietal system associated with CWM and the medial frontoparietal system previously associated with basic forms of social cognitive processing (Meyer et al., Evidence for social working memory from a parametric functional MRI study, Proceedings of the National Academy of Sciences, 2012). Thus, SWM and CWM may both rely on the lateral frontoparietal system supporting generic working memory processes, but only SWM may specifically recruit the medial frontoparietal system. Results from Paper 1 are discussed in terms of updating theories of effortful social cognition, medial frontoparietal system function, and implications for clinical populations with dual and/or differential impairments in social cognition and working memory.

To confirm whether there may be a full dissociation in the medial frontoparietal system across SWM and CWM, Paper 2 compared, within a new sample of subjects (N=25), neural responses to the SWM paradigm (thinking about 2, 3, or 4 friends along a trait dimension during a delay period) to a difficulty-matched CWM paradigm (alphabetizing 2, 3, or 4 friends' names during a delay period). Results showed that the medial frontoparietal system differentially responds to SWM and CWM: these regions linearly increased as a function of the number of friends considered along trait dimensions in working memory, but linearly decreased as a function of the number of friends' names alphabetized in working memory. Thus, the medial frontoparietal system may uniquely support SWM processes, despite interfering with CWM processes. Moreover, linear increases as a function of the number of friends considered during SWM in the medial frontoparietal system, but not lateral frontoparietal system associated with generic forms of working memory, predicted individual differences in experimentally measured perspective-taking ability. These results suggest that working memory properties in the medial frontoparietal system track with social cognitive ability and may serve as a useful biomarker for social cognitive ability in clinical populations with deficits in social cognition and perspective-taking, such as schizophrenia, autism spectrum disorder (ASD), and/or social anxiety.

Inspired by these neural results, Paper 3 examines the extent to which, just as CWM training can improve performance on tasks measuring cognitive ability (e.g., math and reading ability; Chein et al., 2010; Holmes et al., 2009), SWM training may improve social cognitive performance, and the extent to which such transfer effects may be unique to SWM (versus CWM) training. 57 participants were randomly assigned to complete SWM or CWM training. Both training interventions improved SWM, as well as perspective-taking performance. Thus, social cognition may be plastic and bolstered by working memory training. However, SWM training (vs. CWM training) uniquely related to the social cognitive gains in perspective-taking. Moreover, individuals with the most autistic traits, an index of social competence, showed the most gains in SWM only if they had undergone SWM (but not CWM) training. Consistent with the brain imaging results from Papers 1 and 2, results from Paper 3 suggest that SWM and CWM training may improve social cognition through at least partially different underlying mechanisms.

Together, these studies suggest that SWM relies on at least partially unique neurocognitive mechanisms than CWM and that these unique mechanisms may be trained to bolster social cognitive ability. Results are discussed in terms of advancing theories of social cognition and working memory, as well understanding the etiology of various clinical disorders associated with social cognitive deficits.