Abstract
Contour integration relies on neural mechanisms sensitive to contour smoothness, such as the association field (AF) proposed by Field et al. (1993). We conducted three experiments to test whether or not the sensitivity of the AF to contour smoothness elicits similar effects for a contour detection (CD) experiment and two novel contour camouflage experiments (CC1 and CC2). In the CD task, background elements were successively removed until a target contour became visible. In the CC1 task, a contour appeared against a static background of constant density and remained visible for up to a few seconds and then became camouflaged. In the CC2 task, a contour became camouflaged as the density of a dynamic background was increased. In all three experiments we observed sensitivity to parametric manipulation of contour smoothness. Importantly, individual observers' sensitivities to contour smoothness were strongly correlated across all three experiments, which suggest the involvement of a common AF mechanism. In addition to correlating smoothness sensitivity, we examined prospective correlations for task-specific bias (i.e. willingness to report that a contour had become visible or disappeared) across the experiments. Interestingly, we found that this bias only correlated between the CC1 and CC2 experiments in which contours became camouflaged. Taken together, our findings suggest that AF sensitivity to image-based influences (contour smoothness) is not task-specific, and it can be dissociated from top-down effects on the AF hysteresis related to task-specific bias.
Meeting abstract presented at VSS 2016