Abstract
Object perception is remarkably stable, even though saccades move visual input across the retina several times per second. O’Herron and von der Heydt (2009) discovered that the assignment of a border in the classical receptive field (cRF) to a foreground object (border ownership, BOS) in visual cortical area V2 persists for more than a second even after the contextual information that determined BOS is deleted, a potential neural correlate of this stability. Multiple lines of evidence suggest that BOS selectivity depends on feedback from higher cortical areas. Object persistence neurons (OPNs) in higher areas with cRFs that are large enough to contain the contextual information may underlie persistence of BOS in V2 through feedback. A critical prediction for OPNs is that their response to an ambiguous border in the cRF depends on position: if a vertical border is positioned in the right part of the cRF, persistent activity should prefer a history of an object left of the border, and vice versa for a border positioned left in the cRF. This is in contrast to BOS for similar-sized objects in V2, which does not change with border position. Here we test whether macaque area V4, the main source of corticocortical feedback to V2, contains neurons that behave as predicted for OPNs. We recorded responses from units during fixation, while presenting a border at different positions in the cRF. BOS was initially defined by contextual cues indicating a square object, but then we removed these cues, rendering BOS ambiguous. Squares were sized such that they could fit within the cRF. Spiking responses were analyzed starting 200 ms after the removal of contextual cues. We find neurons in V4 that behave as predicted for OPNs. These data indicate that V4 contains neurons that show response hysteresis for objects within their cRF.