How Your Brain Really Makes Choices

Princeton neuroscientists have created a new mathematical model that sheds light on how the brain interprets auditory and visual stimuli when making decisions. The work, which was published in Nature Neuroscience on February 10, 2025, may further our knowledge of neurological conditions like Alzheimer's and improve artificial intelligence systems like self-driving cars and digital assistants.

In order to make wise decisions, the brain concurrently processes several sensory inputs, including sirens and traffic lights. Higher cognition and decision-making are controlled by the prefrontal cortex, which is situated behind the eyes. Understanding the larger neuronal mechanisms at play is difficult because prior research shows that individual brain cells in this area react to sensory stimuli in complex ways.

Neural circuits have been studied using traditional mathematical models, such as recurrent neural networks (RNNs), however these models are challenging to interpret due to their dense connections. The latent circuit model, developed by Princeton academics Christopher Langdon, Ph.D., and Tatiana Engel, Ph.D., in response to these constraints, streamlines decision-making analysis by concentrating on a small number of important nerve cells rather than the complete neural network.

Researchers used their model on RNNs that had been trained on a decision-making task in order to test it. Following the display of a shape on a screen, participants were required to determine the colour or motion of a moving grid by using the shape cue. According to the model, brain cells that processed shape repressed those that processed colour when motion was the main focus, and vice versa. This made it easier to comprehend how brain circuits rank information.

By altering the connections between various cerebral nodes, the latent circuit model enables researchers to forecast shifts in decision-making. Task performance decreased predictably when certain connections were eliminated. This method offers a fresh perspective on how to examine brain activity and comprehend how neural connections impact behaviour.

The results could have important ramifications for mental and neurological disorders including depression and ADHD that influence judgement. Furthermore, by improving decision-making in AI-driven systems like virtual assistants and self-driving cars, the model may advance artificial intelligence. Subsequent studies will concentrate on using the model for other decision-making tasks in order to confirm its efficacy in various contexts.

Source: February 10, 2025, Princeton University, Nature Neuroscience.