Integrated Ising model with global inhibition for decision-making.

What is it about?

We propose a new model for the neuronal circuit that makes decisions in the brain. Each option is represented by a group of neurons in non-firing/firing states, which interact with each other. When an integrator of these relative firings reaches a threshold, a decision is made. The model has an order/disorder phase transition, as function of the noise (equivalent to temperature) and global inhibition. We find that near the tricritical point, where the dynamics is of run-and-tumble type, the model best fits experimental data from both perceptual and reinforced learning evidence. Within this critical region the model predicts advantageous properties, such as tradeoff between speed and accuracy and maximal increase in accuracy per small increase in global inhibition.

Featured Image

Why is it important?

The model makes an advance in our modelling of the decision making process in the brain, and for the first time proposes that this process is performed by positioning the circuit in a critical regime near a phase transition. The model presents a new framework for this fundamental process in our brains.

Perspectives