We used a 300-msec delay between stimulus
offset and postcue here to reduce the possibility of observers relying on high-fidelity, short-lived
iconic memory. At shorter delays, observers do appear to use such memory for certain
attentionally demanding tasks. For instance, Narasimhan, Tripathy, and Barrett (2009) saw
thresholds rise sharply in a trajectory-deviation detection task when a delay
of as little as 100 msec was introduced between the two halves of the trajectory that were
presented for comparison. If this memory survived our 300-msec delay in some weak form and
observers used it, the memory trace must have been fragile enough to be affected by a bounce
event in the nonqueried target.