Connecting the past with the present: how do humans match an incoming visual display with visual memory?

Extensive cognitive research has been devoted to the sensitivity of the visual system to invariant statistical information. For example, many studies have shown that performance improves when a visual display is presented repeatedly. But what allows humans to connect the current visual input to previous memory? Is the connection made only when the entire incoming display matches with a previous memory, or can retrieval rely on an incomplete match between the input and a learned display? Using a visual search task, we show that (1) once a repeated display is learned, subjects can retrieve it even when an incoming display only matches it in 3-4 locations; (2) however, early during learning, repetition of a small proportion of a display is not enough to establish a strong memory trace for the repeated locations. We suggest that the retrieval of a well-established visual memory can proceed even if an incoming display partly matches the previous memory.