Kodaka Y, Sheliga B M, FitzGibbon E J, Miles F A
Laboratory of Sensorimotor Research, National Eye Institute, National Institutes of Health, Building 49, Room 2A50, 49 Convent Drive, Bethesda, MD 20892-4435, USA.
Vision Res. 2007 Sep;47(20):2637-60. doi: 10.1016/j.visres.2007.06.013. Epub 2007 Aug 15.
Radial optic flow applied to large random dot patterns is known to elicit horizontal vergence eye movements at short latency, expansion causing convergence and contraction causing divergence: the Radial Flow Vergence Response (RFVR). We elicited RFVRs in human subjects by applying radial motion to concentric circular patterns whose radial luminance modulation was that of a square wave lacking the fundamental: the missing fundamental (mf) stimulus. The radial motion consisted of successive 1/4-wavelength steps, so that the overall pattern and the 4n+1 harmonics (where n=integer) underwent radial expansion (or contraction), whereas the 4n-1 harmonics--including the strongest Fourier component (the 3rd harmonic)--underwent the opposite radial motion. Radial motion commenced only after the subject had fixated the center of the pattern. The initial RFVRs were always in the direction of the 3rd harmonic, e.g., expansion of the mf pattern causing divergence. Thus, the earliest RFVRs were strongly dependent on the motion of the major Fourier component, consistent with early spatio-temporal filtering prior to motion detection, as in the well-known energy model of motion analysis. If the radial mf stimulus was reduced to just two competing harmonics--the 3rd and 5th--the initial RFVRs showed a nonlinear dependence on their relative contrasts: when the two harmonics differed in contrast by more than about an octave then the one with the higher contrast completely dominated the RFVRs and the one with lower contrast lost its influence: winner-take-all. We suggest that these nonlinear interactions result from mutual inhibition between the mechanisms sensing the motion of the different competing harmonics. If single radial-flow steps were used, a brief inter-stimulus interval resulted in reversed RFVRs, consistent with the idea that the motion detectors mediating these responses receive a visual input whose temporal impulse response function is strongly biphasic. Lastly, all of these characteristics of the RFVR, which we attribute to the early cortical processing of visual motion, are known to be shared by the Ocular Following Response (OFR)--a conjugate tracking (version) response elicited at short-latency by linear motion-and even the quantitative details are generally very similar. Thus, although the RFVR and OFR respond to very different patterns of global motion-radial vs. linear-they have very similar local spatiotemporal properties as though mediated by the same low-level, local-motion detectors, which we suggest are in the striate cortex.
已知将径向光流应用于大型随机点图案会在短潜伏期引发水平聚散眼球运动,扩张导致会聚,收缩导致发散:即径向流聚散反应(RFVR)。我们通过对同心圆形图案施加径向运动来诱发人类受试者的RFVR,该同心圆形图案的径向亮度调制为缺少基波的方波:即缺失基波(mf)刺激。径向运动由连续的1/4波长步长组成,因此整体图案和4n + 1谐波(其中n为整数)经历径向扩张(或收缩),而4n - 1谐波 - 包括最强的傅里叶分量(第三谐波) - 经历相反的径向运动。径向运动仅在受试者注视图案中心后才开始。最初的RFVR总是朝着第三谐波的方向,例如,mf图案的扩张导致发散。因此,最早的RFVR强烈依赖于主要傅里叶分量的运动,这与运动检测之前的早期时空滤波一致,就像在著名的运动分析能量模型中一样。如果将径向mf刺激减少到仅两个相互竞争的谐波 - 第三谐波和第五谐波 - 最初的RFVR显示出对它们相对对比度的非线性依赖性:当两个谐波的对比度相差超过大约一个八度时,对比度较高的那个完全主导RFVR,而对比度较低的那个失去其影响:胜者全得。我们认为这些非线性相互作用是由感知不同竞争谐波运动的机制之间的相互抑制引起的。如果使用单个径向流步长,短暂的刺激间隔会导致RFVR反转,这与介导这些反应的运动探测器接收视觉输入的观点一致,其时间脉冲响应函数是强烈双相的。最后,我们将RFVR的所有这些特征归因于视觉运动的早期皮质处理,已知眼跟踪反应(OFR) - 一种由线性运动在短潜伏期引发的共轭跟踪(版本)反应 - 也具有这些特征,甚至定量细节通常也非常相似。因此,尽管RFVR和OFR对非常不同的全局运动模式 - 径向与线性 - 做出反应,但它们具有非常相似的局部时空特性,就好像是由相同的低级局部运动探测器介导的,我们认为这些探测器位于纹状皮层。