Beyond the circuit architecture: attractor dynamics reveals the mechanism of improved performance in decision-making and working memory

Understanding the underlying mechanisms of cognitive functions such as decision-making(DM) and working memory(WM) is always one of the most essential concerns in modern neuroscience.Recent experimental and modelling works suggest that decision-making is supported by the selective subnetwork of inhibitory neurons, rejecting the previously proposed circuit mechanisms assuming a single non-selective pool of inhibitory neurons. The mechanism underlying decision-making and working memory functions based on such circuit architecture is still unclear. Here we applied a general non-equilibrium landscape and flux approach to a biophysically based model that can perform the decision-making and working memory functions. The quantified attractor landscapes reveal that the accuracy in decision-making can be improved due to the stronger resting state in the circuit architecture with selective inhibition, while robustness of working memory against distractors is weakened, which implies a trade-off between DM and WM. We found that the presence of a ramping non-selective input during the delay period of the decision-making tasks can serve as a cost-effective mechanism of temporal gating of distractors. This temporal gating mechanism, combined with the selective-inhibition circuit architecture, can support a dynamical modulation for emphasizing the robustness or the flexibility to incoming stimuli in working memory tasks according to the cognitive task demands. These mechanisms can also achieve an optimal balance in the trade-off between DM and WM. Our approach can provide a global and physical quantification which helps to uncover the underlying mechanisms of various biological functions beyond the circuit architectures.