Deciding on one option or another is a matter of consensus among neurons

Confidence in a decision is a fundamental characteristic of human beings’ subjective experience. It allows us to estimate the results of taking a decision and, for this reason, is linked to highly complex cognitive functions such as consciousness.

A study published on 24 February in the journal Scientific Reports proposes a new computational model that explains the mechanism that allows calculating the confidence in a decision.

The study involved Andrea Insabato and Gustavo Deco (ICREA), researcher and director of the Center for Brain and Cognition (CBC), respectively, of the Department of Information and Communication Technologies (DTIC) at UPF, together with researchers from the universities of Torcuato di Tella and Buenos Aires (Argentina) and Columbia (USA).

Neuronal populations cast their vote

In the specialist literature, the classic models assumed that, when having to decide between two alternatives, two populations of neurons were involved, each of them integrated the sensory evidence received and the action was decided by one of the two possible alternatives. Then, in the classic model, confidence in a decision is coded on the basis of the activity of the two neural populations involved, that is to say according to the balance of the integrated sensory evidence.

However, recent discoveries in the influence of sensory evidence on confidence in decisions contradict the classic models. The authors of the study published in Scientific Reports propose a new model that can explain these new findings. Unlike the classic paradigm, in this new model many neuronal populations would come into play that would integrate the sensory evidence almost independently from each other.

A model in which “the final decision is calculated on the basis of a process of ‘voting’ among the neuronal populations involved, as in the decisions taken collectively by humans (e.g., in democratic voting), so that the option that obtains more than 50% of the votes is the one chosen”, explained Andrea Insabato, co-author of the study.

“Thus, confidence in decisions is calculated on the basis of the variability in the activity of the neuronal populations involved”, the study authors conclude.

Reference work:

Luciano Paz, Andrea Insabato, Ariel Zylberberg, Gustavo Deco, Mariano Sigman (2016), “Confidence through consensus: a neural mechanism for uncertainty monitoring”, Scientific Reports 6, Article number: 21830, doi:10.1038/srep21830.