Abstract
The current research investigated whether visual memory and visual percepts can interact such that information from each source can be integrated to form a single representation. Two dot arrays were serially presented within a grid leaving one space empty which subjects identified. Past research has shown that when the delay is very short (<100 ms), the arrays are perceptually integrated and performance is excellent. Our goal was to determine if integration can also occur at longer delays where perceptual integration is not possible. Specifically, we investigated whether subjects could generate a visual memory of Array 1, retain it across the delay separating the arrays, and combine it with information obtained from Array 2. To test this, array separation ranged from 0 to 5000 ms. After a drop in accuracy from 0 to 100 ms due to backward masking, performance improved, reaching asymptote after 1500 ms as a memory for Array 1 was generated which could be combined with Array 2. Strikingly, accuracy after 1500 ms approached that when the arrays were perceptually integrated and reaction times indicated memories and percepts were integrated at the same speed that two percepts are integrated. The insertion of a probe-discrimination task showed that during the delay subjects assigned spatial attention / memory to the locations occupied by Array 1 as responses to the probe were faster when it appeared at a location previously occupied by Array 1 compared to when it appeared elsewhere. This assignment of attention or memory, however, was not strictly bound to the arrays' physical or spatial properties. Subjects were able to identify the location of the missing dot even when the arrays were presented at different locations and sizes. Thus, the memory trace of Array 1 can be transformed during the delay between the arrays to match the expected properties of Array 2. Implications of a memory-percept integration process on theories of visual memory will be discussed.