ABSTRACT
Decision making is an essential component of intelligent behavior. Whether interpreting sensory inputs, allocating cognitive resources, or planning actions, the brain constantly makes decisions based on noisy inputs and competing objectives. Understanding decision making at both the psychological and neurobiological levels is an important problem in neuroscience, and has received increasing theoretical attention in recent years. Of key interest is the way that uncertainty is represented and computed in cognitive processing. Uncertainty is a ubiquitous and critical component of neural computation. It arises from inherent stochasticity and nonstationarity in statistical relationships in the environment, incomplete knowledge about the state of the world (e.g., sensory receptor limitations, an observer’s solitary viewpoint), neuronal processing noise, and so on. Uncertainty complicates the task of constructing and maintaining an appropriate internal representation of the world. Despite the plethora of complications induced by such uncertainty, animals adeptly negotiate a complex and changeable world. A major challenge in neuroscience is to develop a formal theory of decision making that incorporates the various types of uncertainty, and link them to both animal behavior at the phenomenological level and neurobiology at the algorithmic level.
