Neural circuits for adaptively biasing decision making

PROJECT SUMMARY (ABSTRACT) Our long-term goal is to elucidate the systems and circuits, within and across brain regions, responsible for decision making. The objective of this proposal is to examine the neural circuit basis for how, in the face of sensory uncertainty, perceptual decisions are adaptively biased to the most valuable option, as instructed by the recent history of choices and their outcomes. We address this question by focusing on the superior colliculus (SC), a key midbrain node in the network of brain regions responsible for selecting targets for movement (i.e., “spatial choice”), an important form of decision making amenable to circuit-level interrogation. The SC integrates input from numerous brain regions, has extensive intrinsic circuitry capable of integrating priors with sensory evidence as required by Bayesian frameworks, and outputs pre-motor orienting commands to downstream motor nuclei. Given its role as a functional hub for spatial choice, as well as its known cell types and circuitry, the SC is ideal for examining the neural circuit basis for adaptive decision making and how decisions are influenced by priors. In particular, the SC receives a robust input – of unknown function – from the mesencephalic pedunculopontine tegmental nucleus (PPTg), which we have shown represents recent spatial choices and outcomes. In a set of experiments in behaving mice, we test the overarching hypothesis that the SC adaptively biases spatial choice by integrating priors represented by PPTg input. We use a variant of an established spatial choice task for mice in which the dominant component of an odor mixture presented at a central port cues whether reward is available at the left or right reward port, and in which the left/right reward ratio changes across blocks of trials. On trials in which the odor mixture provides only weak evidence about reward location, decision making can be optimized by biasing choices towards the port yielding the larger reward. In Aim 1, we test the hypothesis that pre-motor SC output biases choice towards the most valuable target, by recording and perturbing the activity of genetically defined pre-motor SC neurons during the task. In Aim 2, we test the hypothesis that PPTg input transmits representations of priors to the SC that instruct adaptive choice bias in pre-motor SC neurons. We examine whether SC neurons – particularly inhibitory commissural neurons well positioned to mediating competition between left and right choices – represent choices and outcomes of previous trials, as we have seen in the PPTg. We then determine whether representations of priors and choice bias in the SC depend on PPTg input, by perturbing PPTg activity and recording SC activity during the task. If successful, the overall impact of our proposal will be the elucidation of a key circuit mechanism for how decisions are optimized by priors, a critical nervous system function. In addition, our proposal will enable future research into how genetically-defined and projection-specific neural circuits contribute to decision making, offering insight into basic brain function and neurological conditions in which decision making is impaired.