Memory for an edge includes figure and ground assignment

Do memories of figures code the exclusive assignment of the bounding edge to one side or do they preserve evidence that both sides of the edge were assessed for figural status? A priming task adapted from Driver and Baylis (1996) examined shape memory following a single exposure to a novel figure. The prime was a small yellow figure displayed on the left or right side of a larger red ground with a jagged edge separating figure and ground regions. Observers viewed the prime, without making any response, and then made a speeded same-different discrimination regarding two probe shapes that were shown one above the other, either facing in the same direction as the figure prime (figure probes) or the opposite direction (ground probes). On experimental trials, at least one of the probe shapes had the same jagged edge as the prime (different response) or both did (same response). On control trials neither of the probe shapes had the prime's jagged edge. In Exp. 1, the prime was exposed for 180 ms, followed by a 500 ms blank screen, and then the probes. In Exp. 2, the prime was exposed for 128 ms, followed by a 128-ms mask and then the probes. In Exp. 3, half the experimental probes were mirror reversed. In all experiments, observers responded significantly slower to experimental ground probes than to control probes, and they tended to respond faster to experimental figure probes. This result shows that the memory for an edge includes more than the assignment of the edge to one side; it also includes memory for the abandoned assignment of the same edge to the opposite side. We refer to this as memory for the "edge complex," because in addition to the shape of the jagged edge, it includes information about both figure and ground. These results also reveal the dynamics of figure-ground assignment in the absence of shape familiarity, since they were evident following a single exposure to a novel shape. Figure-ground assignment When two adjacent regions share a contour, one has a definite shape (figure); the other is shapeless (ground). Traditional view of figure-ground assignment • object memories accessed for figures but not for grounds + + Ground Probes Figure Probes “Top” 1028 ms, 10% err “Bottom” 954 ms, 6% err Support for traditional view (Driver & Baylis, 1995, 1996) Task: Which probe shape has the same jagged contour as the colored display? Result: Figure probes matched more efficiently than ground probes. •Object memories play a role as do convexity, closure, etc. •Both side of an edge are processed An alternative view of figure-ground assignment • local edge competition, not global shape (Peterson, 2002) • cooperative interactions on each side influence the competition • a two-sided consequence shape seen on one side shape cues inhibited on the other side Parallel Interactive Model (PIM) (Peterson et al, 2000) The explicit contour matching task of Driver & Baylis • may not index figure-ground processing • without control trials does not permit slower responses to ground probes to be disentangled from faster responses to figure probes. A New Task Implicit Priming (rather than explicit memory) • Is a pair of probe shapes shown immediately after a prime the same as or different from each other? • Control shapes matching neither the figure or the ground were used. Experiment 1 Prime: 180 ms; ISI: 500 ms; probes on until response (see Driver & Baylis, 1995) 896 trials; N = 22 Experiment 2 Prime: 128 ms; Checkerboard Mask: 128 ms; N = 20 Data Analysis Correct RT for each condition was divided by proportion correct (pc) to arrive at an inverse efficiency score. Experimental scores were subtracted from Control scores to index priming. Positive differences indicate faster responses on experimental trials; negative differences indicate slower responses. P ri m in g S co re (R T /p c) Figure