Event-related potentials recorded from young and old adults during a memory retrieval task.

Six healthy old and 8 healthy young subjects each received a series of trials in a memory retrieval task devised by Sternberg (1966). On each trials, the subject received a memory set of 1-4 digits and was then shown a test digit. The subject's task was to press one of two response buttons indicating whether the test digit was a member of the memory set for that trial. Response time (RT) was found to be an increasing, linear function of the number of items held in memory. The slope of the RT function was a composite measure of the time necessary to process each additional item in memory while the intercept was a measure of stimulus encoding and response processes that do not depend on memory set size (see Sternberg 1966, 1969, 1975). We found that the latency of P3 to the test stimulus also increased with increases in memory set size, although the slope of the P3 was less than that for RT. We have suggested that the intercept of the P3 slope reflects the time it takes to encode the test stimulus before the evaluation of the stimulus starts, while the slope reflects the amount of time per digit needed to evaluate the set. We have suggested that the difference between RT and P3 slopes represents the additional time per digit which the subjects waits before making a response, due to low confidence occurring with more difficult task conditions (i.e., when set size = 4). We further suggest that the intercept of the RT-P3 latency slope is a reflection of pure response processes. Time estimates of these processes are made for young and old subjects.