Figure 1.
How changed population receptive fields may emerge from partially stimulating the visual field.
See also [8], [14]. If a stimulus (the checkered block) moves over a region of visual space that covers all neurons' receptive fields (top row), all neurons should respond and contribute to the pRF estimate (as indicated by red shading). If, in contrast, a stimulus moves over a more restricted region of visual space that covers a more limited number of neurons', only a subset of neurons will respond and contribute to the pRF estimate. This is true in the masked conditions we used (bottom row). Therefore, the pRF estimates can change as a result of a stimulus change even when the underlying neuronal receptive field properties remain constant. When a central mask is applied, it is also true that neuronal receptive fields contributing to the pRF estimate have, on average, more eccentric locations in the visual field than those that were silenced by masking the stimulus. This is why more eccentric pRF estimates emerge. Finally, as illustrated by the pair of plots on the right, larger pRF estimates (indicated by the arrowheads) also emerge because the active neurons (red circles) during the masked conditions have receptive field that are more likely to be larger. The diagonal line in the lower right plot corresponds to the sum of the receptive field size and location equating to the size of the mask.
Figure 2.
Illustration of the expanding ring stimuli in each experimental condition.
(a) Stimulus schematic of the full-field condition. The maximum stimulus radius was 15°. The bottom panels show how the stimulus changes over time. (b, c) Stimulus schematic of the 5° and 7.5° masked conditions, respectively (masks are shown in opaque red). Bottom panels indicate the resulting stimulus sequence. For clarity, only 5 of the 12 ring positions are shown.
Figure 3.
ROI locations on a flattened cortical surface of one of the participants.
(a) The polar angle maps are indicative to the locations of the early visual areas. (b) The eccentricity maps are indicative to the centrals and peripheral visual field representations. The occipital pole (OP) ROIs are located in V1 at the border between the calcarine sulcus and the occipital pole, a region that responds to stimuli presented to the center of the visual field. The calcarine sulcus (CS) ROI is also located in V1, but more anteriorly in the calcarine sulcus, which responds to more peripherally presented stimuli. Insets indicate the color maps that define the visual field representation. Solid black lines indicate the representation of the vertical meridian, and dashed black lines indicate the representation of the horizontal meridian. Solid white lines indicate the borders of the ROIs.
Figure 4.
Response distributions for the unmasked and masked conditions.
A. Response ratio (i.e., the number of responsive voxels per bin divided by the total number of responsive voxels) versus pRF eccentricity for the unmasked and masked conditions for the OP region-of interest. B. Response ratio versus pRF size for the unmasked and masked conditions in the OP region-of interest. C. Response ratio versus pRF location for the unmasked and masked conditions in the CS region-of interest. D. Response ratio versus pRF size for the unmasked and masked conditions in the CS region-of interest. Note that in the masked conditions the pRFs shift away from their original location in the OP regions (A) but not in the CS region (C). The pRFs in the OP regions are also larger for the masked conditions in the OP regions (B), but not in the CS region (D). Gray, blue and orange shadings indicate the jackknifed 95% confidence interval for the unmasked and the two masked conditions, respectively.
Figure 5.
Model fits to BOLD time-series are shown for a voxel in the OP region that explained more than 15% of the time-series variance in the unmasked (top) and both of the two masked conditions (middle and bottom). The BOLD time-series show increasingly broader peaks and more pronounced phase-shifts corresponding to an increased estimate of the pRF size and location, respectively. Note that the y-axes in the three panels have different scales.
Figure 6.
The effect of masking the central portion of the visual field on pRF eccentricity.
A. The population receptive field (pRF) eccentricity within the OP ROI derived for the 5.0° masked condition as function of the same measure derived from the unmasked condition is plotted on blue. The sum of pRF location and size for the 5.0° masked condition are also plotted as a function of pRF location for the unmasked condition (gray). B. The same plot as in A but for the 7.5° masked condition. The dashed black lines indicate the predicted result if voxels that responded in the same way in both conditions. The dashed red lines in each plot show the borders of the masks. Error-bars indicate the standard error. Note that the axes in the two panels have different scales.
Figure 7.
Response and expansion distributions across the different pRF sizes.
A. Response distribution of pRF sizes estimated for the unmasked conditions for voxels that responded in the 5.0° masked condition as well as the full-field condition. B. Response distribution of pRF sizes estimated for the unmasked conditions for voxels that responded in the 7.5° masked condition as well as the full-field condition. C. Mean pRF expansion for voxels that responded in the 5.0° masked condition as well as the full-field condition. D. Mean pRF expansion for voxels that responded in the 7.5° masked condition as well as the full-field condition. E. The net effect of the change in pRF size, as measured by the product of the mean change and the number of voxels per bin induced by the 5.0° mask. F. The net effect of the change in pRF size, as measured by the product of the mean change and the number of voxels per bin induced by the 7.5° mask. The error-bars for each bin are jackknife estimates of the 95% confidence interval.
Figure 8.
Effect of masking the central visual field for voxels that responded in all three conditions.
A. The pRF eccentricity within the OP ROI derived for the 5.0° (blue) and 7.5° (red) masked conditions as function of the same measure derived from the unmasked condition. B. The pRF size within the OP ROI derived for the 5.0° (blue) and 7.5° (red) masked conditions as function of the pRF size derived from the unmasked condition. The dashed black lines indicate the predicted result if voxels that responded in the same way in both conditions. The dashed blue and red lines represent the borders of the 5.0° and 7.5° masks, respectively. Error-bars indicate the standard error.
Figure 9.
The effect of modeling the presence of a central mask.
A. Response ratio (i.e, the number of responsive voxels per bin divided by the total number of responsive voxels) versus pRF eccentricity for the unmasked and masked conditions for the OP region-of interest. B. Response ratio versus pRF size for the unmasked and masked conditions in the OP region-of interest. Gray, blue and orange shadings indicate the jackknifed 95% confidence interval for the unmasked and the two masked conditions, respectively.
Figure 10.
An explanation of the pRF dynamics in terms of feedback signals from extrastriate cortex.
V1 neurons are presumed to have a receptive field (RF) center (dashed blue) that is measured by presenting high-contrast stimuli and commensurate with feed-forward connections from the lateral geniculate nucleus (LGN), a near surround (dashed green) that is measured by presenting low-contrast stimuli and commensurate with intra-areal V1 horizontal connections, and a far surround (dashed gray) that is commensurate with extrastriate feed-back connections [18], [20], [21]. We further assume that the pRF, which is measured by presenting high-contrast checkerboard stimuli, is an estimate of the RF center (dashed blue) when there is no mask. Normally, the response to stimulating the RF center is modulated by suppressive feed-back signals from the far periphery of the visual field [20]–[23]. However, when there is little or no stimulus contrast on the receptive field center, these feedback signals can also be excitatory [18], [20], [21], [24]. Under such circumstances, the far surround will be partially stimulated, which results in the skewed response indicated by the thick black curve. Fitting a Gaussian receptive field model to these responses will necessarily be shifted towards the fringe of the mask. It will also be larger than the RF center if the far surround extents sufficiently far into the periphery. Interestingly, this model also predicts that increasing the size of the mask (shifting the fringe to the right) results in a decrease of the pRF expansion. This is indeed what appears to happen when the mask size increases from 5.0° to 7.5° (see Figure 4B and corresponding text).