作者: J. Gold , P. J. Bennett , A. B. Sekuler
DOI: 10.1038/46027
关键词: Perceptual discrimination 、 Signal strength 、 Discriminator 、 Computer science 、 Cognitive psychology 、 Internal noise 、 Response characteristics 、 Perceptual learning 、 Sensory system 、 Learning theory
摘要: Perceptual discrimination improves with practice. This ‘perceptual learning’ is often specific to the stimuli presented during training1,2,3,4,5, indicating that practice may alter response characteristics of cortical sensory neurons6,7. Although much known about how learning modifies circuits8, it remains unclear these changes relate behaviour. Different theories assume by enhancing signal1,9,10, diminishing internal noise11,12 or both13. Here, distinguish among alternatives, we fashioned sets faces and textures whose signal strength could be varied, trained observers identify patterns embedded in noise. Performance increased up 400% across several sessions over days. Comparisons human performance an ideal discriminator showed efficiency which encoded task-relevant information. Observer consistency, measured a double-pass technique identical are shown twice each experimental session14,15, did not change training, showing had no effect on These results indicate perceptual enhance strength, provide important constraints for learning.