Attention, uncertainty, and free-energy.

Abstract in brief

• This paper suggests that attention works by inferring how uncertain or precise sensory information is during perception, using neuronal simulations to show that brain activity encodes probabilistic models that optimize free energy, which is related to reducing surprise.
• The researchers demonstrate that when the brain adjusts the precision of sensory data based on the current state, it can explain various attention-related behaviors, such as how we focus, ignore, or switch between different stimuli, as seen in experiments like the Posner paradigm.
• The simulations reveal that biased competition for neural resources naturally arises from Bayesian inference, leading to phenomena like attention bias, faster response times, and electrophysiological patterns consistent with how attention modulates sensory processing.