GABAERGIC INHIBITION SHAPES ACCUMULATION DYNAMICS AND CHOICE BIASES DURING PROTRACTED DECISIONS

The computational and neural mechanisms underlying evidence accumulation for perceptual decisions are reasonably well understood. By contrast, much less is known about the mechanisms guiding more complex decisions that are based on abstract (e.g., numerical) information and operating across longer timescales.
Here we used a protracted decision-making task (trials of 5-10s) in which participants compared the mean values of two simultaneously presented streams of numbers. To unravel the underlying neural circuit mechanisms, we combined this task with source-level magnetoencephalography (MEG) and the placebo-controlled pharmacological increase of GABAergic transmission (via lorazepam).
Human participants (N=40) exhibited analogues of “irrational” choice biases, typically reported in complex decisions. Simulations showed that those can emerge from a prototypical circuit model of decision-making with attractor dynamics. Pharmacological boost of GABAergic neuronal inhibition amplified choice biases and produced correlated changes in choice behaviour and cortical dynamics in several regions of posterior parietal cortex, which encoded both instantaneous numerical evidence and the decision variable (accumulated evidence). These changes were in line with boosting inhibition in the model. In contrast, the brain-behaviour correlations of pharmacological effects reversed in a set of areas in ventral prefrontal cortex. This result is consistent with a “disinhibition” of parietal decision circuits under lorazepam, due to release of inhibitory control signals from ventral prefrontal cortex.
Our findings shed light on the mechanisms of protracted decision computations and of choice biases typically reported in behavioural economics research, while also calling for an extension of the canonical cortical decision circuit with a prefrontal control mechanism.