Spatial sampling in human visual cortex is modulated by both spatial and feature-based attention

  1. Daniel Marten van Es  Is a corresponding author
  2. Jan Theeuwes
  3. Tomas Knapen  Is a corresponding author
  1. Vrije Universiteit Amsterdam, The Netherlands
  2. Royal Academy of Sciences, The Netherlands
10 figures and 2 additional files

Figures

Experimental design and pRF fitting procedure.

(A) Experimental design. Throughout a bar pass participants reported either changes in color (Attend Color) or temporal frequency (Attend TF) of Gabor elements within the moving bar stimulus, or …

https://doi.org/10.7554/eLife.36928.003
Figure 2 with 2 supplements
ROI definition.

(A) Attend Fixation pRF polar angle maps for an example participant with retinotopic areas up to the intra-parietal sulcus defined by hand. (B) Attend Fixation pRF size as a function of eccentricity …

https://doi.org/10.7554/eLife.36928.004
Figure 2—figure supplement 1
Eccentricity-size relations for all statistical methods.

This figure shows that eccentricity-size relations are positive in all ROIs for both the 'super subject' method (Supplementary file 1 - Table 1) and the 'over subjects' method (Supplementary file 1

https://doi.org/10.7554/eLife.36928.005
Figure 2—figure supplement 2
Explained variance and voxel count for all statistical methods.

The violin plots depict the distribution of explained variance across voxels for the different ROIs and for all subjects.The horizontal line at 0.10 indicates the exclusion criterion (voxels with …

https://doi.org/10.7554/eLife.36928.006
Figure 3 with 3 supplements
Effect of spatial attention on pRF position.

(A) Plotting strategy. For pRF shift visualizations, all pRF positions are mirrored into one quadrant of the visual field. Then, vectors representing the shift of pRF centers between conditions were …

https://doi.org/10.7554/eLife.36928.007
Figure 3—figure supplement 1
pRF shift plots for all statistical methods.

Shift vectors run from the Attend Fixation to the Attend Stimulus pRF location. This figure indicates that the shift patterns seen in the 'super subject' are almost identical to the 'over subjects' …

https://doi.org/10.7554/eLife.36928.008
Figure 3—figure supplement 2
pRF shift directions for all statistical methods.

The 'over subjects' panel replicates the 'super subject' results, namely that shifts are best explained by eccentricity changes, followed by changes in pRF x and finally in pRF y in all ROIs (see Sup…

https://doi.org/10.7554/eLife.36928.009
Figure 3—figure supplement 3
pRF x, y and eccentricity position shifts as a function of polar angle for all statistical methods.

The data from all statistical methods closely matches the radial shift direction hypothesis, showing strongest pRF x shifts close to the horizontal meridian, strongest pRF y shifts close to the …

https://doi.org/10.7554/eLife.36928.010
Figure 4 with 2 supplements
Effect of spatial attention on pRF eccentricity and size.

Difference between Attend Stimulus and Attend Fixation pRF eccentricity (A) and size (B) as a function of Attend Fixation eccentricity. Overall, parafoveal pRFs tend to shift away from the fovea and …

https://doi.org/10.7554/eLife.36928.011
Figure 4—figure supplement 1
Effect of spatial attention on pRF eccentricity and size for all statistical methods.

Difference between Attend Stimulus and Attend Fixation pRF eccentricity and size as a function of Attend Fixation eccentricity (see Supplementary file 1 -Tables 4-7 and 16-23). The data show a …

https://doi.org/10.7554/eLife.36928.012
Figure 4—figure supplement 2
Correlation between pRF eccentricity and size changes for all statistical methods.

In the 'over subjects' method, we find that pRF eccentricity and size changes are significant in all ROIs except IPSO (although p = 0.070). In the 'per subject method', we find such a significant …

https://doi.org/10.7554/eLife.36928.013
Figure 5 with 2 supplements
Capturing spatial attention in attentional gain field model.

(A) Observed (black) and predicted (color) pRF shifts. (B) Observed and predicted changes in pRF eccentricity (the main pRF shift direction) as a function of eccentricity. Markers depict data and …

https://doi.org/10.7554/eLife.36928.014
Figure 5—figure supplement 1
Attentional gain field modeling results for each subject - shift vectors.

Arrows depict observed (black) and predicted (color) pRF shifts. This shows that the model closely captured the data in all individual subjects.

https://doi.org/10.7554/eLife.36928.015
Figure 5—figure supplement 2
Attentional gain field modeling results for each subject - changes in pRF eccentricity.

Observed (markers) and predicted (solid lines) pRF eccentricity difference between Attend Fixation and Attend Stimulus conditions for each subject. This shows that although individual subjects show …

https://doi.org/10.7554/eLife.36928.016
Figure 6 with 4 supplements
Feature-based attentional modulation of pRF changes.

(A) Differences in pRF eccentricity and size relative to the Attend Fixation condition, for both the Attend Color and Attend TF conditions. The changes in both eccentricity and size are more …

https://doi.org/10.7554/eLife.36928.017
Figure 6—figure supplement 1
Feature-based attentional modulation of pRF parameters across eccentricity for each ROI.

The bottom-right panel shows the feature AMI for all ROIs together for the 'super subject'. This shows that feature AMI is generally positive along eccentricity across ROIs. In addition, differences …

https://doi.org/10.7554/eLife.36928.018
Figure 6—figure supplement 2
Feature AMI compared to feature preference across ROIs with different ways of computing the FAMI.

In (A), FAMI only includes the change in pRF size, whereas in (B), FAMI only includes the change in pRF eccentricity. This reveals comparable results to use of the combined FAMI pRF size and …

https://doi.org/10.7554/eLife.36928.019
Figure 6—figure supplement 3
Feature-based attentional modulation of pRF size and eccentricity changes in the combined ROI for all statistical methods.

Differences in pRF eccentricity and size relative to the Attend Fixation condition, for both the Attend Color and Attend TF conditions separately.The changes in both eccentricity and size are more …

https://doi.org/10.7554/eLife.36928.020
Figure 6—figure supplement 4
Feature AMI compared to feature preference for each ROI, for all statistical methods.

The y-axis displays feature AMI, which increases when pRF modulations (size and eccentricity changes combined, see Materials and methods) are greater when attending color compared to TF. The x-axis …

https://doi.org/10.7554/eLife.36928.021
Figure 7 with 1 supplement
Feature preference and eccentricity.

Preference to color compared to TF is greatest near the fovea. Error bars denote 95% CI over voxels. Also see Figure 7—figure supplement 1.

https://doi.org/10.7554/eLife.36928.022
Figure 7—figure supplement 1
Color compared to TF preference versus eccentricity correlations for all statistical methods.

The 'super subject' method shows negative Spearman correlations in all ROIs except VO and LO (see Supplementary file 1 -Table 10). The 'over subjects' method shows that this correlation is negative …

https://doi.org/10.7554/eLife.36928.023
Feature-based attentional modulation of attentional gain field sizes.

Results are shown for the combined ROI and for the median across ROIs. Positive values here indicate smaller attentional gain fields when attending color compared to temporal frequency in the bar. …

https://doi.org/10.7554/eLife.36928.024
Figure 9 with 1 supplement
Task and fixation performance.

(A) Behavioral accuracy per attention condition and per bar stimulus eccentricity bin. Horizontal line denotes Quest target of 83%; chance level was 50%. (B) Median (top panel) and standard …

https://doi.org/10.7554/eLife.36928.025
Figure 9—figure supplement 1
Ratio of Gabor elements of either feature value used by the Quest procedure to equate difficulty.

In the Attend Color condition, this reflected the ratio of elements that were either blue/yellow or green/magenta. In the Attend TF condition this reflected the ratio of elements that were either …

https://doi.org/10.7554/eLife.36928.026
Schematic overview of modeling procedure.

(A) The Stimulus Drive (SD) was estimated by dividing the measured Attend Fixation pRF by an AF at fixation (AFfix). (B) Attention toward the bar stimulus at a given timepoint t was modeled as the …

https://doi.org/10.7554/eLife.36928.027

Additional files

Download links