Multi-region neural dynamics of internally driven timing decisions, at spiking resolution

About This Grant

PROJECT SUMMARY Animals depend critically on their capacity to choose the appropriate time to act. In the absence of explicit cues, these decisions are thought to depend on internal deliberation that dictates action timing to maximize outcomes. A broad range of previous observations suggest that internally driven timing decisions involve a distributed network of brain regions. However, it remains unclear how these regions interact during decision-making; mechanistic understanding is very limited and hypothesized interactions between brain regions remain untested. This critical knowledge gap stems from two main factors. First, the absence of explicit cues makes decision timing unpredictable, which historically has presented problems for experimental design. Second, previous methods for measuring and perturbing neural activity, and for quantifying the dependence relations between neural activity patterns, have been ill-suited for probing interactions between brain areas on the relevant timescales. To overcome these barriers, our collaboration combines a novel behavioral paradigm, multi-region spike- resolution neural recording, rapid neural activity perturbation, and a range of model-based computational approaches. In recent work, our analysis of activity on individual decision trials has demonstrated the prominent involvement of a deterministic process, in contrast to recent models that emphasize stochasticity. We have also developed an approach for analyzing multi-region recordings that has revealed a modular structure in the influence of several frontal cortical regions on the striatum, an influence thought to be central to timing decisions. Our preliminary results for this proposal point to a revised decision model that involves an urgency-like signal and a source of unpredictability distinct from that of prevailing models. We have also begun to examine interactions between prefrontal and somatomotor circuits that are also thought to be central to timing decisions. Here we have found evidence of an interaction mediated primarily by corticocortical connections, and one that has a modular structure. Our proposed work would build on these results to test our new model and identify its neural substrates (Aim 1), and test existing ideas about prefrontal-somatomotor interactions (Aim 2). Here we will use multi-region Neuropixels recordings, and an approach we have recently demonstrated for fast optogenetic silencing during internally driven timing decisions. We will compare our new model to others using rigorous statistical methods for model selection. We will identify neural substrates by analyzing best-fit models and perturbation results. We will also continue to develop our new methods for quantifying interregional interactions from activity recordings. Collectively, our work will quantify relevant interactions between brain regions that could not be resolved with previous approaches, leading to improved models. This will provide a new foundation for understanding the neural mechanisms of a basic aspect of natural behavior with relevance to cognition.