Ghose Geoffrey M, Yang Tianming, Maunsell John H R
Division of Neuroscience and Howard Hughes Medical Institute, Baylor College of Medicine, Houston, Texas 77030, USA.
J Neurophysiol. 2002 Apr;87(4):1867-88. doi: 10.1152/jn.00690.2001.
Performance in visual discrimination tasks improves with practice. Although the psychophysical parameters of these improvements have suggested the involvement of early areas in visual cortex, there has been little direct study of the physiological correlates of such perceptual learning at the level of individual neurons. To examine how neuronal response properties in the early visual system may change with practice, we trained monkeys for more than 6 mo in an orientation discrimination task in which behaviorally relevant stimuli were restricted to a particular retinal location and oriented around a specific orientation. During training the monkeys' discrimination thresholds gradually improved to much better than those of naive monkeys or humans. Although this improvement was specific to the trained orientation, it showed little retinotopic specificity. The receptive field properties of single neurons from regions representing the trained location and a location in the opposite visual hemifield were measured in V1 and V2. In most respects the receptive field properties in the representations of the trained and untrained regions were indistinguishable. However, in the regions of V1 and V2 representing the trained location, there were slightly fewer neurons whose optimal orientation was near the trained orientation. This resulted in a small but significant decrease in the V1 population response to the trained orientation at the trained location. Consequently, the observed neuronal populations did not exhibit any orientation-specific biases sufficient to explain the orientation specificity of the behavioral improvement. Pooling models suggest that the behavioral improvement was accomplished with a task-dependent and orientation-selective pooling of unaltered signals from early visual neurons. These data suggest that, even for training with stimuli suited to the selectivities found in early areas of visual cortex, behavioral improvements can occur in the absence of pronounced changes in the physiology of those areas.
视觉辨别任务的表现会随着练习而提高。尽管这些改善的心理物理学参数表明早期视觉皮层区域参与其中，但对于单个神经元水平上这种知觉学习的生理相关性，几乎没有直接研究。为了研究早期视觉系统中的神经元反应特性如何随练习而变化，我们让猴子在一项方向辨别任务中训练了6个多月，在该任务中，行为相关刺激被限制在特定的视网膜位置，并围绕特定方向定向。在训练过程中，猴子的辨别阈值逐渐提高，比未训练的猴子或人类的阈值要好得多。尽管这种改善特定于训练的方向，但几乎没有视网膜定位特异性。在V1和V2中测量了来自代表训练位置和对侧视觉半视野中一个位置的区域的单个神经元的感受野特性。在大多数方面，训练区域和未训练区域的表征中的感受野特性难以区分。然而，在V1和V2中代表训练位置的区域，其最佳方向接近训练方向的神经元略少。这导致在训练位置对训练方向的V1群体反应出现微小但显著的下降。因此，观察到的神经元群体没有表现出足以解释行为改善的方向特异性偏差。汇总模型表明，行为改善是通过对早期视觉神经元未改变信号进行任务依赖和方向选择性汇总来实现的。这些数据表明，即使是用适合视觉皮层早期区域发现的选择性的刺激进行训练，在这些区域的生理学没有明显变化的情况下，行为也可以得到改善。
