Abstract
Background. Individuals get better with practice on a variety of perceptual and cognitive tasks, and this learning is thought to reflect plasticity in underlying neural mechanisms. Is perceptual learning of biological motion reflected in the neural activity of the brain areas underlying perception of these displays, namely STSp and FFA? We trained naïve observers to discriminate biological from scrambled motion in displays masked with noise dots and measured neural activity associated with the learning using fMRI. Methods. Naïve observers viewed point-light biological and scrambled motion animations superimposed in noise dots. The number of noise dots was adjusted such that prior to training the observers performed at near chance levels for discriminating the two kinds of animations. Brain activity was measured while observers performed a 1-back task on these masked displays. Next observers received daily training sessions in which the difficulty of the task was continuously adjusted as observers improved Following this training, brain activity was again measured while observers viewed the masked animations. Results. D-prime measures prior and subsequent to training indicate that observers became more proficient at discriminating the biological from scrambled animations in the masked displays, with the degree of improvement varying among observers. Prior to learning, neural activity in STSp and FFA were suppressed in the masked condition relative to the unmasked animations, but after learning, BOLD signal levels were equivalent during the masked and unmasked conditions. These results were replicated when the same observers viewed a novel set of biological motion animations not seen during the training sessions. The degree of learning was positively correlated with the changes in BOLD signals. Thus observers become more sensitive to biological motion with learning, and this change in sensitivity is reflected in the amplitude of the neural signals in STSp and FFA.