The long-term goal of our investigations is to understand how neural circuits in the frontal cortex support decision- making. Orbitofrontal cortex (OFC) plays a key role in decision-making under uncertainty and is thought to enable humans and other animals to make predictions about outcomes more accurately. The objective of this proposal is to determine the algorithms responsible for confidence-guided behaviors and their neural basis in the mammalian brain. Previous work from our laboratory has shown that `decision confidence', a cognitive variable, is encoded in single OFC neurons and OFC inactivation specifically impairs a confidence-guided behavior, time investment. The proposed experiments are designed to determine the computational and neural algorithms responsible for two confidence-guided behaviors: time investment and choice strategy updating (learning). Our central hypothesis is that OFC generates an abstract representation of decision confidence, independent of sensory evidence, that supports multiple confidence-guided behaviors. To test this idea, we have designed a quantitative psychophysical task for rats, adapted from human and primate work, that enable behavioral readouts of confidence, as a post-decision temporal wager. Simply, after each perceptual decision (olfactory or auditory) rats invest time waiting for a delayed, uncertain reward. These graded- duration time investments serve as a behavioral report of confidence that the perceptual decision just made will result in the success of the perceptual decision. First, we will develop computational algorithms to explain confidence-guided time investment and learning, and second we will identify neural substrates of these algorithms in the OFC. Third, we will determine if OFC representations are sensory-modality and behavioral-output general and test the causal role of OFC using inactivations. Finally, we will map a sensory route for auditory information to inform OFC representations and test the hypothesis that auditory cortex lesions lead to deaf-hearing, the ability to discriminate sounds without perceptual confidence. These contributions are significant, in our opinion, because they will provide critical missing information about the algorithmic and neural foundations of decision confidence, a key cognitive variable. Our approach is innovative, chiefly because we have developed a computational and behavioral framework to study confidence in rats. Beyond these mechanistic studies, the proposed work will advance knowledge about the frontal cortical logic of cognitive variable representations and inform an improved framework for understanding how impairments in a single brain area can lead to a wide range of psychiatric disorders, as seen in depression, obsessive- compulsive and psychotic disorders.
This proposal aims to identify the computational algorithms of two confidence-guided behaviors, time investment and choice strategy updating, and the neural processes underlying these. We will use a combination of electrophysiology, viral tracing, optogenetics in a quantitative confidence-reporting behavior in rats, interpreted with computational modeling and statistical analysis. We aim to test our central hypothesis that OFC generates an abstract representation of decision confidence that supports multiple confidence-guided behavioral processes.
