Publications

Our sensory inputs are partial, noisy, and ambiguous, but they are nevertheless sufficient for surviving in natural environments. One reason for our ability to use these far-from perfect sensory inputs for guiding behavior is the fact that sensory stimuli (and the associated motor acts) are highly structured, so that future events can be predicted, with a fair amount of success, from past events. In fact, using models for making predictions and matching those predictions against the incoming stimulus stream seems to be a useful strategy for many real-world recognition (e.g. Berkes et al., 2009) and motor control (e.g. Davidson and Wolpert, 2005) tasks. The collection of papers presented here illustrates some of the depth and variety of current research into predictive information processing, highlighting basic phenomenology, clinical applications, and conceptualizations of the myriad ways in which predictions are used and expressed by the human brain…