Search Papers | Poster Sessions | All Posters

Poster B11 in Poster Session B - Thursday, August 8, 2024, 1:30 – 3:30 pm, Johnson Ice Rink

Reward probability encoding in human neurons

Alexander Price¹, Rhiannon Cowan¹, Tyler Davis¹, Shervin Rahimpour¹, Ben Shofty¹, Matthew Botvinick², Timothy Muller³, Steven Kennerley⁴, John Rolston⁵, Elliot Smith¹ (); ¹University of Utah, ²Deepmind, ³Univeristy College London, ⁴University of Oxford, ⁵Harvard University

People routinely make decisions with uncertain outcomes. Economists have defined this uncertainty about a possible outcome as risk. Previous studies have found subcortical neural computations related to reward and risk. The prefrontal cortex has also been shown to play a role in these computations. Here we present evidence of computations underlying reward and risk in human prefrontal and temporal cortices. These representations may be signatures of Distributional Reinforcement Learning, a framework by which the brain makes value-oriented predictions and updates those predictions to make decisions.

Keywords: distributional reinforcement learning value-based decisions risk processing human neuronal computation 

View Paper PDF