Figure 2.

Population receptive field modelling results. A) pRF model comparisons for AC1 and AC2 (left and right, respectively) and two control subjects at the respective measurement sites. The mean variance explained and 95% confidence intervals for fMRI responses in the right Calcarine sulcus are shown for four different pRF models. Only cortical locations where any model explains more than 50% of the variance are included the analysis. However the results are near-identical for any other or no threshold. Four different models were tested: the conventional pRF model containing 1 Gaussian, and three different pRF models with 2 Gaussians where the Gaussians are mirrored around the y-axis, fixation or x-axis. The pRF models are indicated by the x-axis’ cartoon representations. There is no difference in degrees of freedom in the models. In the right hemisphere of the achiasmic subjects, the model containing 2 pRFs mirrored around the y-axis explains most of the variance. In control subjects, the conventional model explains most of the variance in the data. Two example pRF model fits are shown in panels B and C. These panels show the fMRI data of AC2 (dotted line) fitted (solid line) with a conventional pRF using a single Gaussian (B) and a 2 Gaussian pRF model (C). The insets indicate the particular pRF model that is fitted to the data and the variance explained (r²). The conventional pRF model consistently misses certain time-series features that are captured by the 2 Gaussian pRF model (gray arrows). Next we compare the conventional pRF model to the pRF model consisting of 2 Gaussians mirrored on the y-axis by subtracting the variance explained of either model. The difference in percent variance explained of both models is shown on the cortical surfaces of AC2 (D) and a control subject (E). The data that is shown has at least a variance explained of 30% in any pRF model. The dashed white lines indicate the V1-V2 border (vertical meridian or Y-axis). In the subject without an optic chiasm the pRF model with two Gaussians mirrored on the Y-axis explains most of the variance within and beyond V1, whereas the in the control subject the conventional model explains most of the variance in the fMRI data. F) pRF size versus eccentricity in the right Calcarine sulcus of AC2 and 4 control subjects. In the control subjects pRF sizes increase with eccentricity (Dumoulin and Wandell, 2008). This relationship is also plotted for the subject without an optic chiasm for two pRF models (conventional model: open circles; pRF model consisting of two Gaussians mirrored on the Y-axis: closed circles). The pRF sizes of the conventional pRF model deviate from the known relationship between (p)RF size and eccentricity in humans and animals, but the pRF sizes of the novel two Gaussian pRF model are consistent with the known relationship as illustrated by control subjects. See also supplementary Figure S2.