Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2010 Sep 8;30(36):12198-209.
doi: 10.1523/JNEUROSCI.3055-10.2010.

Visual Motion Processing by Neurons in Area MT of Macaque Monkeys With Experimental Amblyopia

Affiliations
Free PMC article

Visual Motion Processing by Neurons in Area MT of Macaque Monkeys With Experimental Amblyopia

Yasmine El-Shamayleh et al. J Neurosci. .
Free PMC article

Abstract

Early experience affects the development of the visual system. Ocular misalignment or unilateral blur often causes amblyopia, a disorder that has become a standard for understanding developmental plasticity. Neurophysiological studies of amblyopia have focused almost entirely on the first stage of cortical processing in striate cortex. Here we provide the first extensive study of how amblyopia affects extrastriate cortex in nonhuman primates. We studied macaque monkeys (Macaca nemestrina) for which we have detailed psychophysical data, directly comparing physiological findings to perceptual capabilities. Because these subjects showed deficits in motion discrimination, we focused on area MT/V5, which plays a central role in motion processing. Most neurons in normal MT respond equally to visual stimuli presented through either eye; most recorded in amblyopes strongly preferred stimulation of the nonamblyopic (fellow) eye. The pooled responses of neurons driven by the amblyopic eye showed reduced sensitivity to coherent motion and preferred higher speeds, in agreement with behavioral measurements. MT neurons were more limited in their capacity to integrate motion information over time than expected from behavioral performance; neurons driven by the amblyopic eye had even shorter integration times than those driven by the fellow eye. We conclude that some, but not all, of the motion sensitivity deficits associated with amblyopia can be explained by abnormal development of MT.

Figures

Figure 1.
Figure 1.
Responses of an example neuron recorded in MT of an amblyopic monkey. We presented random dot fields through the fellow (blue) and amblyopic (red) eye on separate runs. A, Tuning for the direction of motion of coherent dot fields; colored circles near the origin show baseline firing. B, Tuning for the speed of coherent dot fields moving in the preferred direction. C, Responses to dot fields of varying coherence moving in the preferred (open symbols) and antipreferred (filled symbols) directions. Dashed lines in B and C show baseline firing. D, Neurometric functions derived from data in C. Solid lines are fits to the data; the vertical tick through each function represents coherence threshold (at 82% correct).
Figure 2.
Figure 2.
Distributions of eye dominance in MT. A, Data from visually normal controls (from Kiorpes et al., 1996). B, Data from two amblyopic anisometropes. C, Data from two strabismics. Eye dominance categories are transformed so that 1 and 7 represent neurons driven exclusively by the amblyopic (AE) and fellow (FE) eye. Eye dominance in amblyopic MT showed reduced binocularity; fewer neurons were responsive to the amblyopic eye compared to the fellow eye.
Figure 3.
Figure 3.
Distributions of visual response properties. Data from anisometropes (A, C, E) and strabismics (B, D, F) are shown. A, B, Peak firing rate to coherent dot fields. C, D, Distributions of direction index measured with coherent dot fields. E, F, Distributions of pattern index measured with grating and plaid stimuli; positive values indicate pattern direction selectivity; negative values indicate component selectivity (Smith et al., 2005). For distributions of peak response and direction index, arrows represent geometric mean values; for distributions of pattern index, arrows represent arithmetic mean values. Amblyopic and fellow eye neurons did not differ in peak response or pattern selectivity, but neurons tested through the amblyopic eye tended to be less direction selective (Table 1).
Figure 4.
Figure 4.
Behavioral motion sensitivity of amblyopic monkeys. Behavioral sensitivity to coherent motion in random dot fields for the amblyopic monkeys used in this study. A, B, Data for the two anisometropes. C, D, Data for the two strabismics. Peak sensitivity (inverse of coherence threshold) of amblyopic eyes was reduced and shifted toward larger dot displacements (bottom abscissa) and higher speeds (top abscissa) compared with fellow eyes. Some of these data (A and D) were published previously [Kiorpes et al. (2006), their Fig. 1].
Figure 5.
Figure 5.
Coherence thresholds of single units and neuronal populations. A, Distributions of coherence threshold for single units recorded in anisometropes. B, Similar data recorded in strabismics. Insensitive cells that did not reach criterion performance even at 100% coherence are labeled “U” for “unclassified”; arrows represent median values. Amblyopic eye distributions were shifted to higher thresholds; this was significant for anisometropes (Table 1). C–F, Neurometric functions computed from pooling analyses of direction discrimination. Solid lines are fits to the data; vertical ticks indicate coherence thresholds. For all subjects, amblyopic eye functions were shifted to the right of fellow eye functions, indicating elevated coherence thresholds (Table 2).
Figure 6.
Figure 6.
Comparison of population coherence thresholds for motion discrimination and detection. Detection thresholds are plotted against discrimination thresholds; both were computed from population analyses. The dashed line represents the identity diagonal. Detection and discrimination coherence thresholds were well matched, suggesting that interocular differences in coherence thresholds observed in Figure 5 C–F were not caused by reduced directional selectivity in amblyopic eye neurons.
Figure 7.
Figure 7.
Speed tuning of single units and neuronal populations. A, Distributions of preferred speed for single units recorded in anisometropes in. B, Similar data recorded in strabismics. Arrows represent geometric mean preferred speeds. C–F, Speed tuning for each amblyope, computed from pooling analyses; data were normalized to the peak amplitude of the fellow eye. Amblyopic eye data had higher preferred speeds compared with the fellow eye (Tables 1, 2).
Figure 8.
Figure 8.
Spatial frequency tuning of single units and neuronal populations. A, Distributions of preferred spatial frequency for single units recorded in anisometropes. B, Similar data recorded in strabismics. Arrows represent geometric mean preferred frequencies. C–F, Spatial frequency tuning for each amblyope, computed from pooling analyses; data were normalized to the peak amplitude of the fellow eye. Amblyopic eye data had lower preferred spatial frequencies compared to the fellow eye (Tables 1, 2).
Figure 9.
Figure 9.
Temporal frequency tuning of single units and neuronal populations. A, Distributions of preferred temporal frequency for single units recorded in anisometropes B, Similar data recorded in strabismics. Arrows represent geometric mean preferred frequencies. C–F, Temporal frequency tuning for each amblyope, computed from pooling analyses; data were normalized to the peak amplitude of the fellow eye. Amblyopic and fellow eye data had similar preferred drift rates (Tables 1 and 2).
Figure 10.
Figure 10.
Comparison of measured and predicted preferred speeds for neuronal populations. For each eye, we plot the measured preferred speed (ordinate, derived from Fig. 7 C) against the predicted preferred speed (abscissa, derived from the optimal spatial and temporal frequencies in Figs. 8 and 9); both were computed from population analyses. The dashed line represents the identity diagonal. Measured and predicted preferred speeds were well matched, suggesting that amblyopic shifts in preferred speed were caused by changes in the spatiotemporal properties of MT neurons.
Figure 11.
Figure 11.
Temporal integration. A, Behavioral motion sensitivity of an example amblyopic subject. Performance through the amblyopic eye fell gradually with increasing temporal offset (Δt); fellow eye performance was stable. B, Responses of an example binocular MT neuron to coherent dot fields of optimal direction and speed. Responses fell rapidly with increasing Δt, independent of the viewing eye. The intersection of the two fitted lines (arrows) determines integration time (see Materials and Methods). Baseline firing is given by the horizontal portion of each fit; this fitted measure is similar to other measurements of baseline firing. C, D, Population responses for each anisometropic subject. E, F, Population responses for each strabismic subject. Data were normalized to the peak amplitude of the fellow eye. G, Comparison of behavioral (left ordinate, open symbols) and physiological (right ordinate, filled symbols) measures of integration time. Treatment conditions are indicated on the abscissa (FE, fellow eye; AE, amblyopic eye). MT neuron integration times were notably shorter than behavioral integration times, and shorter for amblyopic eyes than fellow eyes.
Figure 12.
Figure 12.
Comparison of neuronal and behavioral motion deficits. A, Interocular ratios of coherence thresholds measured physiologically from neuronal populations (ordinate, derived from Fig. 5 C–F) and behaviorally (abscissa, derived from Fig. 4). Physiological and behavioral shifts in coherence threshold were well matched. B, Interocular ratios of preferred speed measured physiologically from neuronal populations (ordinate, derived from Fig. 7 C–F) and behaviorally (abscissa, derived from Fig. 4). Physiological shifts in speed preference were correlated with behavioral changes but were more modest in magnitude.

Similar articles

See all similar articles

Cited by 28 articles

See all "Cited by" articles

Publication types

Feedback