We compared performances in patients and controls. Because we did not observe significant differences between the thresholds measured in the control group for the two viewing conditions (with and without a simulated scotoma; see above), here, we used only the data collected with a simulated scotoma.
Figure 5 shows the distributions of these thresholds for translational (in red), rotational (in green), and radial (in blue) patterns. Statistical analyses consisted of a two-way ANOVA with the optic flow pattern (translational, rotational, or radial) as the within-subject factor and the group (patient or control) as the between-subject factor.
In agreement with the results observed in the previous section, this ANOVA led to a significant effect of the optic flow pattern, F(2, 44) = 14.29, P < 0.001. Here as well, post hoc t-tests indicated that motion coherence thresholds for the translational pattern (mean ± SD, 48.05 ± 27.14) were significantly higher than those for the rotational (22.84 ± 14.03, P < 0.001) and radial (29.23 ± 20.3, P = 0.009) patterns. No differences were observed between rotational and radial patterns (P = 0.897). The ANOVA did not lead to a group effect, F(1, 22) = 1.19, P = 0.287, but revealed a significant interaction between groups and patterns, F(2, 44) = 3.28, P = 0.047, although this effect was statistically smaller than the pattern effect. Post hoc pairwise t-tests corrected for multiple comparisons (Bonferroni) indicated that there were no significant differences between thresholds measured in the two groups for translational (P = 0.21), rotational (P = 0.887), and radial (P = 0.964) patterns. Paired t-tests also revealed that threshold differences between translational patterns, on the one hand, and the rotational and radial patterns, on the other hand, were significant in patients but not in controls (patients: respectively, P = 0.001 and P = 0.014; mean ± SD: translational, 58.7 ± 30.05; rotational, 22.24 ± 15.76; radial, 29.43 ± 25.5; controls: respectively, P = 1.0 and P = 1.0; mean ± SD: translational, 37.39 ± 20.39; rotational, 23.44 ± 12.74; radial, 29.04 ± 14.54).
In order to control whether the extreme data points measured in our experiment (i.e., the thresholds measured in two patients for rotational and radial optic flow patterns in
Fig. 5) had an impact on our results, we reproduced the statistical analyses described above without these patients (and their associated controls) and found the same effects.
Altogether, these results suggest that in patients with MD, the processing of rotational, radial, and translational patterns of optic flow is preserved. Their robustness was controlled by reproducing the analyses using the threshold measured in control participants without a simulated scotoma. Effects remain unchanged in this case (see
Supplementary Text 3 for more details).