Running missing scan: Perception of oldest member in serial presentations

Running missing scan: Perception of oldest member in serial presentations JAMES V. HINRICHS. University u[ lowa, lowa Cit)·, lowa 52240, and HERMAN BUSCHKE, Albert Finstein College o[ Medicine, tastehesfer Rd. and Morris Park Ave., Bronx, N. Y. 10461 The ability to seleet the current oldest item in a homogeneous series o[ items was examined in 250-item sequences with an "age" range o[ 1-15 items. Approximately 50% o[ fhe responses were correct, and fhe probability o[ a correet response inereased linearI)' with the age o[ the correct respO/;se. The mell1or)' requirements o[ the task were disel/ssed and compared with other cuvert retention tasks. Several recent studies have been concerned with evaluating human memory performance in situations where the retention measure is derived indirectly. If standard memory performance measures, such as recal! and recognition, can be considered overt measures of retention. then these indircct assessments might be termed "covert" retention. In an overt memory task, retention is tested by requiring the S to res pond with the item or items that have been presented previously. In a covert memory task, the S's task is to avoid responding with an item that has been presented. For example, in the missing scan (Buschke, 1963), the S is presented with N - I members of a well-known class of N items and asked to respond with the unpresented or missing item. With the guessing technique (Hinrichs, 1966), the S is required to respond be fore each item is presented, always attempting to respond with an item not yet presented. In both tasks, the evaluation of retention performance is indirect: The better the S successfully avoids responding with presented items, the better is his memory. One way to characterize the S's performance in covert retention tasks is to consider the S's problem as one of attempting to respond with the "oldest" item, that is, the least recent item presented. Then the S's problem rduces to one of attempting to judge the current oldest member of his response set. Conversely, the S avoids responding with the items judged to have been presented recently. Judgment of recency (JOR) has been studied as a task in discrimination (Y ntema & Trask, 1963). and as an absolute judgment task in short Iists (Peterson, Psychon. Sei., 1970. Vol. 19 (2) 1967) and in ~teady-state situations (Hinrichs & Buschke, 1968). In the steady-state paradigrn, a sm all set of items is repeatedly presented in a random order. As each successive presentation is made, the S is required to make a numerical estimate of the number of intervening items since its last presentation. When these judgments are compared with the actual distances, or lags, it is found that Ss overestimate the recency of short lags and underestimate the recency of longer lags in the range of lags from 2 to 15. The JOR task is doselv relat.:d to the disnimination 01' oldness r.:quircd in the covert memory tasks. In the present study, the same steady·state presentation paradigm used in the Hinrichs & Buschke (1968) experiment was employed to ask a different question. However, instead of numerical estimates of recency for each item, the S was required to respond with the item he perceived as being the current "oldest" item in the set available to him. This task more nearly resembles the retention requirements of covert memory tasks than does the JOR task. It is especially similar to the missing scan (Buschke, 1963) in that the S is required to respond with the least recent member of the set, i.e., the "least remembered" item. Unlike the missing scan paradigrn, however, the current procedure employs very long sequences and continuous judgments on the part of the S, hence the paradigm can be considered a "running missing scan." METHOO The materials and procedures used were identical in most respects to those used in Hinrichs & Buschke (1968), which should be consulted for further details. A summary of the method and discrepancies from the earlier study are reported here. Materials and Apparatus The -stimuli were eight letters of the alphabet: Q, R, S, T, W, X, Y, and Z. The letters were approximately evenly distributed within a sequence of 250 presentations so that the interpresentation intervals (lags) were approximately rectangularly distributed. Hence, each one of the 15 possible lags from I to 15 occurred approximately equally often. Eight sequences of 250 presentations were constructed and presented to each S, with the order of presentation counterbalanced across Ss. The stimuli were displayed on an Industrial Electronics Engineering Bina·View cel! programmed by a punched paper tape reader _ The programming and presentation equipment were isolated from the S in a soundproof chamber. The S viewed the display cell through the window of an adjoining soundproof chamber; his verbal responses were recorded by E. Subjects The eight Ss were paid for their participation. Each S was tested individually and served in three I-h sessions, with each session separated by at least I day. Procedure After each successive letter appeared, the S was instructed to respond with the letter that he believed to be the current "oldest" letter, i.e., the letter with the longest interval since its last presentation. Several examples were shown to the S, and a short practice session was administered to ensure S's understanding of the experimental requirements. The Hrst session then concluded with the presentation of two 250-letter sequences at a 3-sec presentation rate. In each of the other two sessions, three sequences were presented. The first 10 responses were removed from the response protocols, with the eight Ss producing 240 observations for each of the eight sequences for a total of 15,360 observations. RESULTS Of the 15,360 responses produced by the eight Ss, there were 7,421 instances in which the current oldest letter was the S's response and 7,939 instances in which it was not. Hence, the proportion of correct identißcation of the current oldest member of the stimulus set was .483. To evaluate further the quality of the Ss' performance, it is Hrst necessary to consider the relevant characteristics of the sequences presented to them. The interval since the last presentation of any particular item, i, was considered that item's age, Ai. At any given point in the sequence, after each of the items had been presented at least once, one of the items was the current oldest item. That is, at any point in the sequence there was one item, 0, in the sequence such that for any other item, i, in the set, Ao > Ai, 0 i. The range of possible values for Ao was 8 to 15; 15 was the maximum age possible as the distribution of items was constructed, and, because there were eight members of the set, any item that was older than all the others must have had an age of at least 8. F or the present set of items, the overall mean age, Le., the mean Ai of all items available to the S, was 5.74. The mean age of the oldest item, A (...truncated)