Major sources of computational complexity in complex decision-making

What makes decision-making hard, and what determines the time it takes to make a decision? For simple decisions, the standard answer implicates neuronal noise: its presence makes decisions hard, and averaging it out comes at the cost of long reaction times. We argue that this explanation is unlikely to hold for complex decisions. Instead, complex decisions are constrained by two main factors: memory retrieval and value computation. Indeed, most decisions require retrieving relevant information from memory, and use it to compute the choice options’ values. For the large memories of vertebrate brains, both operations can be extremely complex even if the neural circuits implementing them are perfectly noiseless. Yet, the importance of these factors has not yet been fully recognized in systems neuroscience, which tends to focus on tasks in which values are retrieved from simple noisy look-up tables. The interrogation of more complex and realistic tasks, similar to the ones used in human research, might help bridge this gap.