Computational mechanisms underlying thalamic regulation of prefrontal signal-to-noise ratio in decision making

Successful execution of complex decision-making tasks requires identification and processing of multiple sources of uncertainty, such as sensory uncertainty, mapping uncertainty and outcome uncertainty. The mediodorsal (MD) thalamus is a critical partner for the prefrontal cortex (PFC) in cognitive flexibility and resolving uncertainty in decision making [1,2]. Animal experiments have shown that the MD enhances prefrontal signal-to-noise ratio in decision making under uncertainty [3]. However, the computational mechanisms by which the MD enhances prefrontal activity in decision making under uncertainty remain unclear. In addition, how the newly discovered cellular diversity in MD contributes to such computations is unexplored. We developed computational models to dissect these mechanisms, and found that the inclusion of an MD-like feedforward module increased robustness to sensory noise and enhanced working memory maintenance in the recurrent excitatory-inhibitory PFC network performing a context-dependent decision-making task. The task uncertainty was represented by the coherence or congruence of sensory cues. We found that the PFC-MD model outperformed the PFC-alone model in task conditions with high sensory uncertainty. PFC excitatory units showed context-invariant rule tunings, whereas MD units showed modulation with respect to context and cue uncertainty. The MD enhanced working memory maintenance during the task delay. By imposing thalamocortical connectivity constraints, we found cell type-differential modulation in MD-to-PFC connectivity (more specifically, MD1PV and MD2VIP pathways for two target-specific MD subpopulations MD1 and MD2), which independently regulate signal amplification and noise suppression. Furthermore, our models made experimentally testable predictions that connect disrupted thalamocortical connectivity with prefrontal excitation-inhibition (E/I) imbalance and dysfunctional inhibitory cell types. In summary, our biologically realistic PFC-MD models replicate important behavioral and neurophysiological findings, reveal a key computational mechanism of context-invariant, cell-type specific regulation of sensory uncertainty, and provide insight into parsing cognitive deficits for mental disorders such as schizophrenia and autism spectral disorder.