Song Yuning, Baker Curtis L
McGill Vision Research Unit, Department of Ophthalmology, McGill University, Montréal, Québec, Canada.
Vis Neurosci. 2006 Mar-Apr;23(2):181-99. doi: 10.1017/S0952523806232036.
The discontinuities of phase-shifted abutting line gratings give rise to perception of an "illusory contour" (IC) along the line terminations. Neuronal responses to such ICs have been interpreted as evidence for a specialized visual mechanism, since such responses cannot be predicted from conventional linear receptive fields. However, when the spatial scale of the component gratings (carriers) is large compared to the neuron's luminance passband, these IC responses might be evoked simply by the luminance edges at the line terminations. Thus by presenting abutting gratings at a series of carrier spatial scales to cat A18 neurons, we were able to distinguish genuine nonlinear responses from those due to luminance edges. Around half of the neurons (both simple and complex types) showed a bimodal response pattern to abutting gratings: one peak at a low carrier spatial frequency range that overlapped with the luminance passband, and a second distinct peak at much higher frequencies beyond the neuron's grating resolution. For those bimodally responding neurons, the low-frequency responses were sensitive to carrier phase, but the high-frequency responses were phase-invariant. Thus the responses at low carrier spatial frequencies could be understood via a linear model, while the higher frequency responses represented genuine nonlinear IC processing. IC responsive neurons also demonstrated somewhat lower spatial preference to the periodic contours (envelopes) compared to gratings, but the optimal orientation and motion direction for both were quite similar. The nonlinear responses to ICs could be explained by the same energy mechanism underlying responses to second-order stimuli such as contrast-modulated gratings. Similar neuronal preferences for ICs and for gratings may contribute to the form-cue invariant perception of moving contours.
相移邻接线光栅的不连续性会导致在线条终端处产生“虚幻轮廓”（IC）的感知。对这种IC的神经元反应被解释为一种特殊视觉机制的证据，因为这种反应无法从传统的线性感受野预测出来。然而，当组成光栅（载体）的空间尺度与神经元的亮度通带相比很大时，这些IC反应可能仅仅由线条终端处的亮度边缘诱发。因此，通过向猫A18神经元呈现一系列载体空间尺度的邻接光栅，我们能够区分真正的非线性反应和由亮度边缘引起的反应。大约一半的神经元（简单型和复杂型）对邻接光栅表现出双峰反应模式：一个峰值出现在与亮度通带重叠的低载体空间频率范围内，另一个明显的峰值出现在远高于神经元光栅分辨率的更高频率处。对于那些双峰反应的神经元，低频反应对载体相位敏感，但高频反应是相位不变的。因此，低载体空间频率处的反应可以通过线性模型来理解，而高频反应代表真正的非线性IC处理。与光栅相比，IC反应神经元对周期性轮廓（包络）的空间偏好也略低，但两者的最佳方向和运动方向非常相似。对IC的非线性反应可以用与对二阶刺激（如对比度调制光栅）反应相同的能量机制来解释。对IC和光栅的类似神经元偏好可能有助于对移动轮廓的形状线索不变感知。
