Plain wording

A classifier trained on foveal V1 during feedback still identifies the same images when they are shown directly at the fovea.

slug cross-decoding-experimental-to-controlrole empiricalpanel fig2B

Feedback into foveal V1 carries the shape of the saccade target but not its semantic category.

slug decoding-shape-sensitive-not-semanticrole synthesispanel fig3

Frontal eye fields and lateral occipital cortex trend the expected way but fail correction, leaving IPS as the only region that holds.

slug fef-lo-nonsignificant-after-correctionrole controlpanel fig4B

The saccade target is decoded far less accurately from foveal feedback than from direct stimulation of the fovea by the same image.

slug foveal-feedback-below-direct-stimulationrole empiricalpanel fig2A

Which peripheral target a person is about to look at can be decoded from foveal V1, even though the target vanishes before fixation.

slug foveal-v1-decodes-peripheral-saccade-targetrole empiricalpanel fig2A

Foveal feedback to early visual cortex carries the target's shape, not its semantic category.

slug hypothesis-feedback-carries-shape-not-categoryrole hypothesispanel hypothesis

Foveal V1's information about peripheral targets is genuine top-down feedback, not leakage from large peripheral receptive fields.

slug hypothesis-feedback-not-spilloverrole hypothesispanel hypothesis

Foveal feedback during saccade preparation uses the same population code in V1 as an image shown directly at the fovea.

slug hypothesis-shared-representational-formatrole hypothesispanel hypothesis

Williams and colleagues found that foveal V1 carries information about peripheral stimuli, which others attribute to receptive-field spillover.

slug interprets-williams-2008-peripheral-spilloverrole literature-contextpanel hypothesis (spillover alternative), fig2B (U-shape test)

Intraparietal sulcus activity tracks foveal decoding more closely than peripheral decoding, making it a candidate source of the feedback.

slug ips-candidate-driver-foveal-feedbackrole synthesispanel fig4B

When the image is shown directly at the fovea, intraparietal sulcus activity tracks foveal decoding in the opposite direction.

slug ips-foveal-effect-reverses-in-controlrole controlpanel fig4-figure-supplement-1

In lateral occipital cortex the pattern reverses: decoding across shape falls while decoding across semantic category holds up.

slug lo-shows-reversed-specificityrole empiricalpanel fig3B

Peripheral decoding was unrelated to activity in all three regions where foveal decoding was related to each, ruling out global fluctuations and signal quality.

slug parametric-effects-not-global-fluctuationsrole interpretationpanel fig4s1

The analysis linking intraparietal, frontal eye field and lateral occipital activity to foveal decoding was exploratory, not preregistered.

slug parametric-modulation-exploratory-not-preregisteredrole scopepanel methods

A classifier trained on foveal V1 during feedback should still identify the same images shown directly at the fovea.

slug prediction-cross-decoding-generalizesrole predictionpanel hypothesis

Lateral occipital cortex should show the opposite profile to foveal V1, losing shape decoding while keeping category decoding.

slug prediction-lo-inverse-patternrole predictionpanel hypothesis

Decoding accuracy across early visual cortex should dip at parafoveal eccentricities and rise again at the fovea, not decay steadily.

slug prediction-u-shape-eccentricityrole predictionpanel hypothesis

In foveal V1, decoding across semantic category should fall under feedback while decoding across shape should hold up.

slug prediction-v1-category-drops-shape-preservedrole predictionpanel hypothesis

The decoding pipeline, regions of interest and statistical tests were all specified in advance in a preregistered plan.

slug preregistered-design-validates-mvparole methodologicalpanel methods

The preregistration reached OSF only after the manuscript was submitted, so its timing rests on website timestamps the authors cite.

slug preregistration-submitted-after-manuscriptrole scopepanel methods

The gaze-contingent display kept the peripheral target out of the central fovea on all but a fraction of a percent of saccades.

slug target-excluded-fovea-in-99pct-saccadesrole methodologicalpanel fig1C

Decoding accuracy in V1, V2 and V3 dips at parafoveal eccentricities and rises again at the fovea rather than decaying steadily.

slug u-shaped-eccentricity-rejects-spilloverrole controlpanel fig2B

In foveal V1 under feedback, decoding across semantic category falls while decoding across shape stays high.

slug v1-category-decoding-drops-in-feedbackrole empiricalpanel fig3B

Foveal V2 and V3 show the same shape-sensitive, category-insensitive feedback as V1.

slug v2-v3-generalize-shape-not-categoryrole empiricalpanel fig3-figure-supplement-1