UC Berkeley

In the first chapter, I investigate the role of prefrontal cortex activity in working memory. Here, I am motivated by inconsistencies in the neural substrates for working memory across studies, species, and recording techniques. For instance, non-human primate electrophysiology research finds that prefrontal circuitry maintains working memory representations, while human neuroimaging suggests that working memory content is instead stored in sensory cortices. These seemingly incompatible accounts for working memory are often confounded by differences in the amount of task training and stimulus exposure across studies, suggesting that long-term learning may influence the role of prefrontal function in working memory maintenance. To answer these questions, we longitudinally trained and scanned participants on a working memory task with complex stimuli. Then, we used multivariate analyses of functional neuroimaging (fMRI) data to test how representational structures of working memory activity patterns in prefrontal cortex change across intensive learning. We show that human prefrontal cortex develops stimulus- selective working memory responses with learning, more akin to results from electrophysiology studies. This approach uses a unique training and analysis framework to establish novel evidence for long-term memory influences on working memory maintenance.

In the second chapter, we investigate how working memory is constantly used to guide our moment to moment behaviors. This reliance on working memory can lead us to make mistakes, like saying aloud the wrong word in a conversation. Such “action slips” are common occurrences but especially pronounced in individuals with prefrontal lesions, who may often pour salt instead of sugar into one’s coffee, or mistakenly type “pizza” in an immediate texting conversation when thinking about your upcoming lunch (Lhermitte et al. 1986). To study this interaction between working memory and ongoing behavior, I implemented a dual-task experiment in which directional words must be held in working memory while more immediate, but unrelated, motor movements are performed. We show that motor behaviors unrelated to current working memory information are still influenced by one’s working memory content (Miller et al. 2020). We are currently testing these behaviors with predictions from cortico-striatal circuit models of working memory gating (e.g., O’Reilly and Frank, 2006) by using transcranial magnetic stimulation. By causally perturbing prefrontal functioning and cortico-striatal connectivity, can we alter when and how often working memory content influences our immediate actions?

In the final chapter, I outline how investigating human-specific neuroanatomical structures in frontal cortex is critical for a wider investigation of human cognition. The prefrontal cortex is disproportionately expanded in the human brain even relative to other advanced primates, and some structures such as tertiary sulci, small folds in the cerebral cortex, are often human-specific. I use multi-modal neuroimaging data to investigate relationships between microstructural and functional properties in human prefrontal cortex. We show that careful identification of often overlooked individual-level anatomical features (such as tertiary sulci) serve as a bridge between the microanatomical and functional properties of prefrontal cortex (Miller et al. 2021). Identifying these structures has implications for both individual-level prefrontal functioning and broader mappings between prefrontal anatomy, functioning, and cognitive domains. We propose that such careful investigations of individual-level neuroanatomy will help to generate structural-functional relationships in areas of cortex previously thought to have little or no consistent links between individual-level structure and function (Miller et al. 2021).