The nature of representation in cognitive control

Cognitive control broadly refers to those processes which adaptively coordinate behaviour in service of a goal. To achieve control, the brain must resolve conflicting information and competing cognitive demands, even when doing so runs counter to more dominant, or prepotent impulses. Explaining this property in the context of the brain has long posed a general problem to researchers. Mechanisms of control have been typically posed as intentional processes and are thus subject to anthropomorphism-styled as a brain within the brain. It is difficult to imagine how neural circuits can achieve this. Classical cognitive science has often been criticised for invoking these 'homunculi' to account for control-related processing.Contemporary neuroscientific and computational literature provides an opportunity to resolve these homuncular accounts. Neural network function provides a plausible means of representing information in the brain. Viewed through the lens of network dynamics, certain structural and functional specialisations characterising control-related phenomena can be grounded in neurally plausible properties of the brain. I pay particular attention to how the circuit organisation of the neocortex may contribute to cognitive control mechanisms.I show that such a structurewould achieve a high level of cognitive control as an emergent property of network function, without the need to invoke homuncular mechanisms.