We tested the statistical significance of the results by performing a group ANOVA with the factors gaze position (up/center/down), polar angle (32 levels), and meridian (horizontal/vertical), and treating subjects as a random effect (Fig.
4, see caption for details). Figure
4 shows the average time course across all six subjects, extracted from the individually-defined ROIs in each subject. In Fig.
4 meridians were defined using the same individual ROIs identified above (see Fig.
2). The group-level ANOVA yielded a significant three-way interaction between gaze position, polar angle, and meridian in all eight areas (left and right, dorsal and ventral V1 and V2;
F(62, 480) > 1.43;
p < 0.05, Bonferroni corrected). Post-hoc tests showed that up/center/down × polar angle interactions were significant along the vertical meridian (V1 dorsal, right hemisphere, vertical meridian, region 4,
F(62, 480) = 1.370,
p < 0.05; V2 dorsal, right hemisphere, vertical meridian, region 8,
F(62, 480) = 3.493,
p < 0.001, V1 ventral, right hemisphere, vertical meridian, region 12,
F(62, 480) = 2.149,
p < 0.001, V2 ventral, right hemisphere, vertical meridian, region 16,
F(62, 480) = 1.430,
p < 0.05; V1 dorsal, left hemisphere, vertical meridian, region 1,
F(62, 480) = 1.859,
p < 0.05; V2 dorsal, left hemisphere, vertical meridian, region 5,
F(62, 480) = 5.193,
p < 0.001, V1 ventral, left hemisphere, vertical meridian, region 9,
F(62, 480) = 2.455,
p < 0.001, V2 ventral, left hemisphere, vertical meridian, region 13,
F(62, 480) = 2.004,
p < 0.001) but not along the horizontal meridian in all areas, with one exception: the right region in V1 dorsal (in the right hemisphere) along the horizontal meridian also showed a significant effect (region 3,
F(62, 480) = 2.455,
p < 0.001). The results match what was observed in the individuals: enhanced response to wedges in gaze conditions that positioned the wedge nearer the straight-ahead direction (in head-centered coordinates).