Department of Neurobiology, The Weizmann Institute of Science, Rehovot 76100, Israel.
Curr Biol. 2012 Oct 9;22(19):1813-7. doi: 10.1016/j.cub.2012.07.059. Epub 2012 Aug 23.
Repeated performance of visual tasks leads to long-lasting increased sensitivity to the trained stimulus, a phenomenon termed perceptual learning. A ubiquitous property of visual learning is specificity: performance improvement obtained during training applies only for the trained stimulus features, which are thought to be encoded in sensory brain regions [1-3]. However, recent results show performance decrements with an increasing number of trials within a training session [4, 5]. This selective sensitivity reduction is thought to arise due to sensory adaptation [5, 6]. Here we show, using the standard texture discrimination task [7], that location specificity is a consequence of sensory adaptation; that is, it results from selective reduced sensitivity due to repeated stimulation. Observers practiced the texture task with the target presented at a fixed location within a background texture. To remove adaptation, we added task-irrelevant ("dummy") trials with the texture oriented 45° relative to the target's orientation, known to counteract adaptation [8]. The results indicate location specificity with the standard paradigm, but complete generalization to a new location when adaptation is removed. We suggest that adaptation interferes with invariant pattern-discrimination learning by inducing network-dependent changes in local visual representations.
重复进行视觉任务会导致对训练刺激的敏感性长期增加，这种现象称为感知学习。视觉学习的一个普遍特性是特异性：在训练过程中获得的性能提高仅适用于经过训练的刺激特征，这些特征被认为是在感觉大脑区域中编码的[1-3]。然而，最近的结果表明，在训练过程中，随着试验次数的增加，性能会下降[4,5]。这种选择性的敏感性降低被认为是由于感觉适应[5,6]引起的。在这里，我们使用标准的纹理辨别任务[7]表明，位置特异性是感觉适应的结果；也就是说，它是由于重复刺激导致的选择性敏感性降低而产生的。观察者在背景纹理中以固定位置练习纹理任务。为了消除适应，我们添加了与目标方向成 45°的纹理定向的无关任务（“虚拟”任务），已知这种任务可以抵消适应[8]。结果表明，在标准范式中具有位置特异性，但在消除适应时可以完全推广到新位置。我们认为，适应通过诱导局部视觉表示中的网络依赖变化来干扰不变模式辨别学习。
