Conference on Cognitive Computational Neuroscience

A fundamental feature of decision making is the influence of experimental history. In two independent datasets we show that previous responses and stimuli can shape subsequent decision making, a process that depends on decision confidence. In the drift diffusion model these effects were explained by a bias on the evidence accumulation process. Simulations suggest that this bias could be caused by either via sensory adaptation and reinforcement learning, or a via a randomly fluctuating decision criterion. By directly estimating these fluctuations in decision criterion from behavioral data, it will be possible to dissociate between both mechanisms.