Understanding spatial representations in rat orbitofrontal cortex

Project summary

Growing evidence suggests that adaptive value representations in the brain are critical to flexible decision making. Inflexible

decisions are a hallmark of a range of neuropsychiatric disorders. Thus, understanding the typical neural substrates of value representation may provide critical insights into the patterns of disordered decisions that characterize pathological

conditions. Convergent evidence over the past decades points to the orbitofrontal cortex (OFC) as a critical locus for value

representation and adaptive decisions. Evidence for neural encoding of the value of choice options is abundant, but it is less

well appreciated that such representations are also intertwined with spatial information the reflects the location where the

outcomes in question will be delivered. The utility—if any—of these pervasive OFC spatial representations is unknown.

This is at least partly due to the fact that most studies of OFC’s role in decision making involve subjects that make their

choices with minimal, if any, movement. We hypothesize that OFC spatial representations are a neural substrate for spatial

credit assignment, the process by which organisms learn to associate locations with the value of outcomes those locations

predict. Drawing inspiration from decision problems that foraging animals face in natural settings, we developed two new

behavioral tasks for rats that require spatial credit assignment. In the patch-leaving task rats shuttle between two foraging

patches where food pellets are delivered at different rates. The longer rats remain in one foraging patch, the lower the rate

of reward becomes. Switching between patches resets the reward rate to its maximal level, but incurs a time penalty in the

form of delay, during which food is not available. Adaptive decision making requires rats to associate information about

reward rate with each patch. In the value-map task, rats are trained to approach visual stimuli projected on to the floor of an

open-field arena, and are reinforced probabilistically for successfully completing trials. Reward probability is determined by an uncued underlying probability map with localized regions where reward is more or less likely to be delivered.

Choosing correctly on choice trials requires learning the underlying probability structure of the task, and attributing higher

or lower value to locations in the arena with increased or decreased reward probability. Preliminary data demonstrate that

rats successfully perform both of these tasks. Neural recordings in lateral OFC identified two classes of spatially-responsive neuron, one of which encodes space coarsely, over a large scale, and the other of which encodes space over a much finer

scale exhibiting discrete firing fields at particular locations. We hypothesize that these spatial cells are recruited to represent

value over large spatial scales (as required by the patch-leaving task) or finer spatial scales (as required by the value-map task). Combining electrophysiology, modelling, and chemogenetic manipulations, we will quantify spatial representations

across the medial-lateral axis of OFC, assess how OFC spatial representations are modified by spatial value learning, and

test the role of cortio-hippocampal interactions in these processes. This work will provide the first detailed characterization of the prevalence, form, and dynamics of medial and lateral OFC spatial responses, and take the first steps towards understanding their role in value-based decision making.