Striate cortical lesions affect deliberate decision and control of saccade: implication for blindsight

Abstract

Monkeys with unilateral lesions of the primary visual cortex (V1) can make saccades to visual stimuli in their contralateral ("affected") hemifield, but their sensitivity to luminance contrast is reduced. We examined whether the effects of V1 lesions were restricted to vision or included later stages of visual-oculomotor processing. Monkeys with unilateral V1 lesions were tested with a visually guided saccade task with stimuli in various spatial positions and of various luminance contrasts. Saccades to the stimuli in the affected hemifield were compared with those to the near-threshold stimuli in the normal hemifield so that the performances of localization were similar. Scatter in the end points of saccades to the affected hemifield was much larger than that of saccades to the near-threshold stimuli in the normal hemifield. Additional analysis revealed that this was because the initial directional error was not as sufficiently compensated as it was in the normal hemifield. The distribution of saccadic reaction times in the affected hemifield tended to be narrow. We modeled the distribution of saccadic reaction times by a modified diffusion model and obtained evidence that the decision threshold for initiation of saccades to the affected hemifield was lower than that for saccades to the normal hemifield. These results suggest that the geniculostriate pathway is crucial for on-line compensatory mechanisms of saccadic control and for decision processes. We propose that these results reflect deficits in deliberate control of visual-oculomotor processing after V1 lesions, which may parallel loss of visual awareness in human blindsight patients.

Figure 1.

Lesion site. A, Drawing of horizontal section of macaque brain. The left V1s are drawn in red. B–D, Magnetic resonance images of monkey brains before and after V1 lesion. Axial slices. B, Monkey A, 4 weeks preoperative. C, Monkey A, 1 week postoperative. D, Monkey T, 1 week postoperative. L, Left; R, right.

Figure 2.

Residual vision and error in the saccadic end points. A, Distribution of the saccadic end points in the visually guided saccade task. Colors of dots indicate the direction of the position of the saccadic targets. Monkey T. B, The distributions of the directional components of end points (in polar coordinates) are plotted for saccades to the normal (top) and affected (bottom) hemifields. The values were normalized so that the peak value for each curve was one. Monkey T. The bin size is 2°. Targets with 10° eccentricity. C, The discrimination index for each neighboring target pair (see Results) was separately plotted for the targets in the normal and affected hemifields in monkey T (left) and monkey A (right). Circles, Median values. D, Left, Histograms of directional errors (in polar coordinates) of the saccadic end points. Gray line, Trials with targets in the normal hemifield; black line, trials with targets in the affected hemifield. Right, Histograms of the eye position errors (in polar coordinates) in fixation during the calibration sessions. Gray line, Trials with fixation targets in the normal hemifield; black line, trials with fixation targets in the affected hemifield. The arrows indicate the median values. Top, Monkey T; bottom, monkey A. The bin size is 1°.

Figure 3.

Sensitivity to luminance contrast. A, Psychometric functions. Left, The success ratio was plotted against the luminance contrast. Gray line, Trials with targets in the normal hemifield; black line, trials with targets in the affected hemifield. Target eccentricity, 10°. Monkey T. Trials with different directions were merged for each hemifield. A dotted line indicates the success ratio 0.79, which corresponds to d′ = 2 in five-alternative forced choice in signal detection theory, used for defining the threshold for luminance contrast. Both plots were fitted with logistic functions with statistical significance (p < 0.01). Right, The discrimination index was plotted against the luminance contrast for the same data as in left. The discrimination indices of four pairs of neighboring targets were plotted against luminance contrasts. Gray lines, The normal hemifield; black lines, the affected hemifield. A dotted line indicates the chance level. B, The threshold for luminance contrast is displayed as a grayscale for each target position. Monkey T. C, Same as in B but for monkey A.

Figure 4.

Classification of trials. A, A schematic psychometric function. Trials were classified into three categories, based on the success ratio for each luminance contrast: (1) saccades to suprathreshold stimuli in the normal hemifield (Supra-threshold), (2) saccades to near-threshold stimuli in the normal hemifield (Near-threshold), and (3) saccades to targets with high luminance contrast (>0.8) in the affected hemifield (High luminance contrast). B, The discrimination index calculated for three trial categories. Symbols are the same as in Figure 2C.

Figure 5.

Scatter in the saccadic end points. A, The distribution of the saccadic end points in the three trial categories. Colors of dots indicate the direction of the position of the saccadic targets. For comparison, the figure for the affected hemifield is flipped horizontally. Trials with target eccentricity of 10°. Monkey T. B, Histograms of the directional components of the saccadic end points (in polar coordinates) in the three trial categories. Black line, The HWHH of the distribution. The bin size is 2°. C, The HWHH of each distribution is plotted across the three trial categories. The dotted lines denote individual values for each target position. The circles and error bars denote the means and SEs across target positions, respectively. **p < 0.01 in Tukey–Kramer's HSD test.

Figure 6.

Scatter in the initial direction of saccades. A, Examples of trajectories of the saccades to a target (10° in eccentricity and lower 60° in direction) in the three trial categories. For comparison, the figure for the affected hemifield is flipped horizontally. Green dots, The points calculated as the initial direction of the saccades. Magenta dots, The end points of the saccades. White circles, Possible target positions. B, Histograms of the directional component of the initial direction (green) in polar coordinates, plotted in the same manner as in Figure 5B. For comparison, the histograms of the directional component of the end points (magenta, the same data as Fig. 5B) were superimposed. Monkey T, Trials with eccentricity of 10°. The bin size is 2°. C, The mean values for the HWHHs of the initial direction (green) and for those of the saccadic end points (magenta, the same data as Fig. 5C). **p < 0.01 in Tukey–Kramer's HSD test. ^†p < 0.001 in Wilcoxon's signed ranks test with Bonferroni's correction. n.s., Not significant. D, The mean values for the compensation index (see Results) were plotted across the three trial categories. Mean and SE. *p < 0.05 in Tukey–Kramer's HSD test.

Figure 7.

Spatial pattern of saccade deficits. A, The HWHH of the end points is displayed as a grayscale for each target position, in the same manner as Figure 3, B and C. Monkey T. B, Same as in A but for monkey A. In one target (25° in eccentricity, upper 60° in direction, affected hemifield) of monkey A, the HWHH was not reliably calculated and the data were removed from analysis.

Figure 8.

Distributions of the saccadic reaction times. A, Trials were classified into five trial categories in the same manner as shown in Figure 4A, except that two additional categories were added. B, Histograms of the saccadic reaction times for the five trial categories. The bin size is 5 ms. Left column, Saccades to the normal hemifield. Right column, Saccades to the affected hemifield. Black line, Success trials; gray dotted line, error trials. Monkey T. C, Same as in B but for monkey A.

Figure 9.

A modified diffusion model. A, Illustration of the model. Top, The decision signal for the target accumulates with an accumulation rate μ and Gaussian noise σ (black line). The decision signals for the nontargets accumulate with an accumulation rate 0 and Gaussian noise σ (gray dotted lines). Middle, When the difference between the maximal signal and the second maximal signal exceeds the decision threshold θ, a saccade initiates. Td, Time of decision-related component; Tr, time of non-decision-related component. Bottom, By repeating this procedure, the distributions of the saccadic reaction times for the success trials and error trials were constructed. B, A sample simulation. The distributions of the simulated saccadic reaction times for the success trials (black line) and the error trials (gray, dotted line) are plotted for nine combinations of two parameters, the accumulation rate μ and the decision threshold θ. Tr is fixed here as 100 ms. The bin size is 5 ms.

Figure 10.

Fitting of the distribution of saccadic reaction times to the modified diffusion model. A, Histograms of saccadic reaction times of success trials (top) and error trials (bottom). Monkey T. The targets with high luminance contrast in the affected hemifield. Solid line, Observed data; dotted line, the best-fitted data (θ = 24, μ = 0.94, and Tr = 140 ms in this condition). To evaluate the goodness of fit, the data were subdivided into six bins (vertical lines) so that the ratio of the number of trials in the six bins was 1:2:2:2:2:1. B, Evaluation of the goodness of the fit. Each point represents a pair of the number of trials of observed data and that of fitted data in one bin of the histograms. For example, 12 data points were obtained from A, which have six pairs of behavioral data (solid line) and fitted data (dotted line) in success trials and another six pairs in error trials. The 12 data points were obtained from five trial categories of two monkeys. Thus, in total, 120 data points were plotted. log 10, Log10 scale. Circles, Monkey T; crosses, monkey A. C, The best-fitted values for the decision threshold θ are plotted against those for the accumulation rate μ. The decision threshold θ was held constant across the different trial categories within the hemifield. Gray, Saccades to the normal hemifield; black, saccades to the affected hemifield. Arrow, Condition plotted in A.

Figure 11.

A summary of the current study. A, A simplified scheme illustrating the potential circuits responsible for residual visual–oculomotor processing after V1 lesion. sSC, The superficial layer of the superior colliculus; dSC, the deeper layer of the superior colliculus. B, V1 lesions affected various stages of visual–oculomotor processing, not only visual processing but also the decision process and saccadic control, as indicated by open arrows.