Six male and six female undergraduates were presented with lists of up to six digits in length, each followed by a "probe" digit. They then searched the memorized lists for the presence or absence of the test item. Reaction times (RTs) were used to gain insight into the strategies employed in high-speed recognition scanning. This study showed parallel increases of RTs for both positive and negative responses as a function of list length (LL). However, the serial position of the probe digit produced no significant differences in average RTs. The experiment supports the indication of a serial exhaustive, as opposed to self-terminating search in high-speed scanning for "presence" in immediate memory.The problem of recall from recent memory can be viewed as a search process involving the rapid inspection and testing of those items in store for the critical characteristics necessary for the experimental task. This conceptualization leads to an analytic method of study that is similar to the techniques employed in visual search studies. Usually in visual search, the subject is first presented with a target item and then must search for the presence of that item in a later-presented list.Implicit in the paradigm of these search process experiments is the supposition that if the selection of the response requires the use of information that is in memory, the latency of the response will reveal something about the process by which the information is retrieved. Fitts and Posner (1967) reported that RTs to locate a probe item increase with the length of the list, whether the search is performed visually or in memory. A second more specific problem in the study of retrieval is the effect of serial position (of the probe) on the latency of response. It has been shown that RT is a function of serial position, whereby searching terminates and the recognition response is made as the target is encountered (Sternberg, 1969).Two alternative models have been designed to explain the processing necessary for a retrieval task. Townsend (1971) described the two processes that are labeled "serial" and "parallel." By parallel processing, it is meant that the items to be processed are examined concurrently. Although processing begins on all items simultaneously, it does not necessarily end simultaneously. Thus the variance in RT reflects the difference in that completion time. By serial processing, it is meant that the elements are processed one at a time, each being completed before the next is begun. It is necessary to differentiate between two types of serial serach (Sternberg, 1969). In self-terminating serial search, the test stimulus is compared successively to one item in memory after another, either until a match occurs (leading to a positive response) 01 until all comparisons have been completed without a match (leading to a negative response). In exhaustive serial serach, the test stimulus is compared successively to all of memorized items. Only then is a response made; positive if a match occurs, and negative if otherwise. Sinc...