Modelling selective processing in vision and action

The selective action network is an alternative to traditional accounts suggesting that visual processing is independent of action. The model proposes simultaneous influence among systems processing spatial and nonspatial visual attributes of objects, spatial and non-spatial components of action, and intentions or goals. Objects and actions compete within processing systems, but between systems, consistent states support one another. Plans for action bias selective visual processing in much the same way that selection can be biased towards particular locations or particular object features. The end result is that biases induced in one system, whether as a result of stimulation from the environment, strategic goals, or damage, propagate gradually and broadly throughout the entire network, so that a single object and its implications for behaviour come to dominate across multiple systems of perception and action.