Abstract

Brains allow us to change how we behave. We can adopt new strategies to forage for food, learn new motor skills to reach new places, and remember all those occasions on which we failed in our endeavours to improve our odds of success in the future. Ongoing changes in how the brain assimilates and represents information make it a challenging operating environment, both for clinical interventions like brain-machine interfaces, and for those systems internal to the brain that aim to produce consistent behaviour. This talk presents an overview of three projects connected to this problem. We outline the design of a brain-machine interface for control of navigation through a virtual maze and describe how the use of this interface interacts with and alters neural representations within CA1 hippocampus, an extremely plastic brain region associated with episodic memory and spatial memory. Next, we present an information-theoretic argument for why reconfigurations of neural populations should proceed by incrementally applying dramatic changes to a small number of neurons at a time, rather than by gradually changing all neurons at once. Finally, we discuss the feasibility of measuring ongoing changes in human behaviour and problem-solving through a massively-online field experiment, and present some preliminary results that mirror the statistics of ongoing changes in neural representations.