Perceptual and cognitive spatial learning

Abstract

Ss were taught novel mappings between visual space and motor space with either a variant on a prism adaptation paradigm (Experiments 1 and 2) or a nonperceptual cognitive task (Experiments 3 and 4). First, discrimination training specified that 1 visual location required a new pointing response but another location did not. This led to unusual generalization unlike typical generalization decrement. Second, training at 9 locations specified that 1 location required a new response but that the remaining 8 did not. This simple isolation mapping was unlearnable and instead a flat function fit through all of space. In contrast, for the cognitive paradigm, not only was isolation of one region of space easily learned, it was the preferred pattern of generalization. Implications for perceptual learning, as well as the qualitative distinctions between perceptual and cognitive learning, are discussed.