From structured knowledge to flexible behaviour: The importance of prefrontal dynamics

Abstract:

Intelligent behaviour requires us to translate structured knowledge about the world into flexible, goal-directed actions. The prefrontal cortex (PFC) has long been implicated in this transformation, inspiring decades of mechanistic studies. When it’s lesioned, people often retain structured knowledge about the world, but fail to organise their behaviour based on this knowledge. Our group seeks to understand this projection of world structure onto behaviour on two fronts. First, we investigate the basic behavioural coordinate system used in the PFC. We find that medial prefrontal dynamics are strongly paced by progress to goal. These goal progress sequences are akin to a coordinate system for behaviour, effectively pacing prefrontal dynamics by “where an animal is” in a set of goal-directed actions. Our results show that this differs drastically from activity in the hippocampal formation which is paced by dynamics that track “where am I in the world”. Using a combination of cross-circuit silicon probe recordings and optogenetic manipulations we show how inputs from world models in the hippocampus are used to compute goal progress dynamics in the PFC. Second, we investigate how rich structures in the world influence the structure of internally generated behaviours. Our findings support a framework of duplicated and structured representations (DSRs), where the abstract structure of the world is embedded in connections between prefrontal neurons encoding conjunctions of action and context. This allows flexibly organising behaviour using local network dynamics - supporting rapid inferences in novel situations. Our findings align with an extensive body of empirical findings showing that DSRs explain both the form and function of prefrontal representations across behaviours as diverse as sequence working memory, planning, and value-guided decisions. Ongoing research in our group, including collaborations on human fMRI and RNN simulations, is beginning to reveal how such world-behaviour transformations could succeed and fail across a range of behaviours.

Bio:

I completed my undergraduate studies in Biochemistry at the University of Oxford before transitioning to Neuroscience for my MSc at UCL. My growing interest in cognitive neuroscience led me to study synaptic mechanisms of memory formation during my PhD with Ole Paulsen at the University of Cambridge. I then returned to Oxford as a postdoctoral scientist, initially working with David Dupret on hippocampal neuronal dynamics underlying memory, and later with Tim Behrens on the algorithms driving abstraction in the medial prefrontal cortex. In 2025 I started an independent position in the department of Experimental Psychology, leading the Mind Mechanisms group. Our group aims to understand how structured knowledge is encoded in the brain and used to guide flexible behaviour.