Predictions and incongruency in object recognition: A cognitive neuroscience perspective

The view presented here is that visual object recognition is the product of interaction between perception and cognition, a process budding from memories of past events. Our brain is proactive, in that it is continuously aiming to predict how the future will unfold, and this inherent ability allows us to function optimally in our environment. Our memory serves as a database from which we can form analogies from our past experiences, and apply them to the current moment. We are able to recognize an object through top-down and bottom-up processing pathways, which integrate to facilitate successful and timely object recognition. Specifically, it is argued that when we encounter an object our brain asks "what is this like" and therefore draws on years of experience, stored in memory, to generate predictions that directly facilitate perception. These feed-forward and feed-back systems tune our perceptive and cognitive faculties based on a number of factors: predictive astuteness, context, personal relevance of the given event, and the degree to which its potential rarity differs from our original expectations. We discuss both computational and theoretical models of object recognition, and review evidence to support the theory that we do not merely process incoming information serially, and that during our attempts to interpret the world around us, perception relies on existing knowledge as much as it does on incoming information.