Recent research by Princeton University has revealed the tricky process underlying how your brain reaches key decisions even amidst confusion. In this paper presented in Nature Neuroscience, several researchers presented vital revelations regarding the integration of several senses—visual or audio input, for instance—in reaching any specific decision.
The findings have major implications for the improvement of brain function in neurological disorders, including Alzheimer’s disease, and the improvement of the performance of artificial systems ranging from voice assistants like Alexa to self-driving cars.
The prefrontal cortex is the part of the brain sitting right behind the eyes and was known as the seat of higher cognitive function and integral to making decisions. Previous studies have touched on the complex responses of single brain cells within this area in decision-making processes, but the way these neurons process sensory input—such as the meaning of traffic lights to cross the street or not—has remained elusive.
To surmount this challenge, several researchers have employed RNNs-a mathematical framework that purports to explain neural dynamics-but due to their complexity, the interpretability of these neural networks is shrouded with difficulty.
In this seminal experiment, scientists heralded a mathematical model that provided much-needed insights into decision making. By providing a “low-dimensional” understanding, they reduced the complexity concerning individual cell performance and the generalized performance of the brain.
The hypothesis was tested with recurrent neural networks engaged in a decision-making task where participants—humans, monkeys, or computers—observed a shape (either a square or triangle) followed by a moving grid. They were tasked with identifying either the color or motion based on the shape presented. The results indicated that cells in the prefrontal cortex responsible for processing shapes suppressed the activity of neighboring cells focused on color, and vice versa.
This study outlines how neural circuits work within large-scale networks. For example, changing the strength of connections between latent nodes systematically influenced task performance and reinforced the promise of uncovering simple mechanisms in the midst of complexity underlying decision-making.
In the future, this model can help further the understanding of disorders in mental health and also can be applied to many other types of decision-making tasks within a laboratory.
