Gochin P M
Department of Psychology, Princeton University, New Jersey 08540.
Cereb Cortex. 1994 Sep-Oct;4(5):532-43. doi: 10.1093/cercor/4.5.532.
The physiological properties of neurons in inferior temporal (IT) cortex of the macaque monkey suggest that this cortical area plays a major role in visual pattern recognition. Based on the properties of IT, and one of its major sources of input, V4, a model is proposed that can account for some of the shape recognition properties of IT neurons including selectivity for complex visual stimuli and tolerance to the size and location of the stimuli. The model is composed of three components. First, stimulus location tolerance is modeled after the complex-cell-like properties observed in some V4 neurons. The second component of the model is an attentionally controlled scaling mechanism that facilitates size-invariant shape recognition. The transition from edge orientation-selective neurons in V4 to neurons with more complicated stimulus preference in IT is explained by the third component of the model, a competitive learning mechanism. Single-unit analysis of receptive field properties, stimulus selectivity, and stimulus size and position tolerance was performed on "neurons" from the simulation. Comparison of results from the simulation and a study of actual IT neurons shows that the set of mechanisms incorporated into the simulation is sufficient to emulate the physiological data.
猕猴颞下（IT）皮质中神经元的生理特性表明，该皮质区域在视觉模式识别中起主要作用。基于IT的特性及其主要输入源之一V4，提出了一个模型，该模型可以解释IT神经元的一些形状识别特性，包括对复杂视觉刺激的选择性以及对刺激大小和位置的耐受性。该模型由三个部分组成。首先，刺激位置耐受性是根据在一些V4神经元中观察到的类似复杂细胞的特性建模的。模型的第二个组成部分是一个受注意力控制的缩放机制，它有助于大小不变的形状识别。模型的第三个组成部分，即竞争学习机制，解释了从V4中边缘方向选择性神经元到IT中具有更复杂刺激偏好的神经元的转变。对模拟中的“神经元”进行了感受野特性、刺激选择性以及刺激大小和位置耐受性的单细胞分析。模拟结果与对实际IT神经元的研究结果比较表明，模拟中纳入的一组机制足以模拟生理数据。
