Salience Coding in the Basal Forebrain and the Heterogeneous Underpinnings Underlying Novelty Computations
Date of Award
Doctor of Philosophy (PhD)
Humans and animals are consistently learning from the environment by interacting with it and getting feedback from their actions. In the environment, some objects are more important than others, because they are associated with reward, uncertainty, surprise, or novelty etc. These objects are salient to the animal. Salient objects attract attention and orientation, increase arousal, facilitate learning and memory, and affect reinforcement learning and credit assignment. However, the neural basis to support these effects is still not fully understood.We first studied how the basal forebrain, one of the principal sources of modulation of the neocortex, encodes salience events. We found two types of neurons that process salient events in distinct manners: one with phasic burst activity to cues predicting salient events and one with ramping activity anticipating such events. Bursting neurons respond to reward itself and cues that predict the magnitude, probability, and timing of reward. However, they do not have a selective response to reward omission. Thus, bursting neurons signal surprise associated with external events, which is different from the reward prediction error signaled by the midbrain dopamine neurons. Furthermore, they discriminate fully expected novel visual objects from familiar objects and respond to object-sequence violations. In contrast, ramping neurons predict the timing of many salient, novel, and surprising events. Their ramping activity is highly sensitive to the subjects' confidence in event timing and on average encodes the subjects' surprise after unexpected events occur. These data suggest that the primate BF contains mechanisms to anticipate the timing of a diverse set of salient external events (via tonic ramping activity) and to rapidly deploy cognitive resources when these events occur (via phasic bursting activity). Then we sailed out to study one special salience signal – Novelty. The basal forebrain responds to novelty, but the neuronal mechanisms of novelty detection remain unclear. Prominent theories propose that novelty is either derived from the computation of recency or is a form of sensory surprise. Here, we used high-channel electrophysiology in primates to show that, in many prefrontal, temporal, and subcortical brain areas, object novelty sensitivity is related to both computations of recency (the sensitivity to how long ago a stimulus was experienced) and sensory surprise (violation of predictions about incoming sensory information). Also, we studied neuronal novelty-to-familiarity transformations during learning across many days and found a diversity of timescales in neurons' learning rates and between-session forgetting rates within and across brain regions that is well suited to support flexible behavior and learning in response to novelty. These findings show that novelty sensitivity arises on multiple timescales across single neurons due to diverse related computations of sensory surprise and recency, and shed light on the logic and computational underpinnings of novelty detection in the primate brain.