Abstract
Repeated visual stimuli elicit reduced neural responses compared to novel stimuli in various brain regions (repetition attenuation). This effect has become a powerful tool in functional magnetic resonance imaging (fMRI) studies, allowing researchers to investigate the stimulus-specific neuronal representations underlying perception and cognition. Repetition attenuation is also commonly associated with behavioral priming, whereby response accuracy and speed increase with repetition. This raises the possibility that repetition attenuation reflects decreased processing time, rather than (or in addition to) changes in the response properties of individual selective neurons. Here we report a full dissociation between repetition attenuation and behavioral performance by varying the task performed on identical visual stimuli. Observers were presented with pairs of scene photographs that overlapped either 83% (very similar) or 50% (less similar). Filler trials where the photographs overlapped 0% and 100% were also included. In the scene task, observers judged whether the two photographs came from the same scene, and in the image task, they judged whether the two photographs were identical pixel for pixel. The two tasks produced opposite patterns of behavioral performance: responses were faster and more accurate for the very similar pairs relative to the less similar pairs in the scene task, and vice versa in the image task. However, in the parahippocampal place area (PPA), a scene-selective region of ventral cortex, identical repetition attenuation was observed in both tasks: lower neural responses for the very similar pairs relative to the less similar pairs. While the PPA was impervious to task modulation, two frontal regions showed the same crossover interaction as behavior, consistent with their role in decision-making. Thus, even though repetition attenuation and priming are often correlated, they can be dissociated, suggesting that attenuation in ventral visual areas reflects stimulus-specific processing independent of task demands.
Supported by a foreign Natural Sciences and Engineering Research Council of Canada Post-Graduate Scholarship to NBTB, NSF grant 0518138 to YX, and NIH grant EY014193 to MMC.