Aaron C. Koralek received an A.B. in molecular biology and a certificate in neuroscience from Princeton University in 2006. He received a Ph.D. in neuroscience with a designated emphasis in computational science and engineering from the University of California, Berkeley in 2014 under the supervision of Dr. Jose M. Carmena. He is currently a postdoctoral researcher at the Champalimaud Neuroscience Programme under the supervision of Dr. Rui M. Costa. His research combines novel behavioral paradigms with state-of-the-art electrophysiological and imaging techniques to investigate the neural bases of skill learning and action selection. His work also applies a range of computational methods to investigate functional interactions and evolving dynamics in large populations of simultaneously recorded neurons over the course of learning.
“The role of dopaminergic network dynamics in behavioral exploration”
Imagine your commute home from work. You have your usual highway route that gets you home in time for dinner. But one day, a construction project starts that closes several highway lanes, greatly increasing your travel time. Do you continue to use this route, or explore new ones? Most people would at least try out the latter. This is just one example, but we constantly are faced with whether to exploit past actions with a known outcome or whether to explore novel actions whose outcomes may be better. Trying out novel actions is especially important in a changing environment, where a previous action may no longer lead to the same outcome. So how does the brain make these complex decisions? It turns out that this balance between exploitation of past actions and exploration of new ones may be controlled by neurons in deep structures of the brain that control the release of the neurotransmitter dopamine. But how do these cells change their activity during behavioral changes adopted in a new environment? To study this question, we will train mice to choose between a number of different actions that have different probabilities of having a favorable outcome. For example, a mouse will have to choose which of several small holes in a container lead to a reward, and each hole will have its own probability of reward. For example, one hole might result in a reward 80% of the time while another hole will result in a reward 20% of the time. There will also be a few different contexts, mimicking stable and changing—i.e., unstable—environments. In the unstable condition, the probability of reward in each hole will randomly change from 20% to 80% or vice versa. The mouse will be forced to adapt to these new conditions, and what exactly happens in the dopamine neurons during this adaption is what we study. Using advanced recording technologies, we will monitor the activity of hundreds of dopamine neurons at once over the course of several days to see how their activity relates to the mouse’s behavioral choices in stable and unstable environments. We will then employ techniques to alter the activity of dopamine neurons, and observe how our manipulation of their activity affects the mouse’s behavior. These experiments will provide insight into how dopamine neurons influence choices, and, since it is well-established that failures in the dopamine system are related to Parkinson’s disease, we expect our results will also shed light on this debilitating disease.