Electronic game used to investigate how the human brain responds after a prediction error

The work was conducted at the University of São Paulo’s Research, Innovation and Dissemination Center in Neuromathematics; the results were published in Scientific Reports.

Agência FAPESP* – Understanding how the human brain works and learns has been one of the main goals of scientists associated with the Research, Innovation and Dissemination Center in Neuromathematics (NeuroMat) – a FAPESP Research, Innovation and Dissemination Center (RIDC) based at the University of São Paulo (USP).

One of NeuroMat’s lines of research, called “Statistical Brain,” has been conducting experiments to test how mathematical models can be useful for representing the interaction between billions of neurons and predicting behavior.

In a recent study published in the journal Scientific Reports, the group used an electronic game to investigate the impact of (prediction) error on the brain’s response time. The goal was to understand how an incorrect decision affects the time it takes a person to process and respond to environmental stimuli after the decision.

In the Goalkeeper’s Game, the player assumes the role of goalkeeper in a penalty shootout and must predict where the shooter will kick the ball. The direction of the shot varies according to a statistical sequence called a “stochastic chain with variable length memory,” in which the direction of the next shot is influenced by previous shots, and some of these directions are the product of random experiments, as in a game of dice. Therefore, the player must understand the structure of the sequence in order to save as many penalty kicks as possible. For example, sometimes it is enough to remember the penalty shooter’s last two choices to predict the next shot, while in other cases it is necessary to remember a larger number of shots.

It was observed that the response time of the participants in the electronic game varied depending on their success or failure in defending against the penalty kick. The results indicate that when errors are frequent, the response time becomes slower after correct decisions.

However, when errors are rare, the response time after incorrect choices is slower. Thus, it is concluded that the functioning of the response time during the game depends not only on the results of the choices but also on the predictability of the sequence of stimuli provided.

“These results provide insights into the statistical models that operate within the brain based on the investigation of simple aspects of behavior, such as response time. As we progress in this investigation, we’re discovering the building blocks of the models and how they’re connected. This paves the way for a better understanding of learning and its dysfunctions,” says Paulo Passos, NeuroMat researcher and FAPESP scholarship holder.

The article “Response times are affected by mispredictions in a stochastic game” can be read here: www.nature.com/articles/s41598-024-58203-7.

* With information from NeuroMat.