T hree experiments examined the shor t-ter m retention of order in a modi® ed Brown±Peter son task. O ur intent was to examine the loss of order memory, unconfounded by item m em or y, under conditions in wh ich interference from prior trials is kept low. In previous work on the shor t-ter m for getting of order, experimenter s have tended to repeat the same items across trials or to draw from a restricted set; in our experiments, we changed the to-be-recalled items from trial to trial and used reconstr uction as the retention measure. In all three exp eriments, ver y little forgetting was obtained across retention intervals that have traditionally produced dr amatic and systematic loss. Ou r results are reminiscent of those obtained in the Brown± Peterson task wh en per for mance is assessed after only the ® r st experimental trial.In the typ ical Brown±Peterson experim ent (Brown, 1958; Peterson & Peterson, 1959), su bjects are presented with a shor t list of item s (e.g. th ree words or a conson an t trigram), followed by a distractor-® lled retention interval of varyin g d ur ation . At th e p oint of recall, the task is to rem ember th e list items in their correct serial order. U nder these conditions, forgetting is often dramatic. In one condition, for exam ple, Peterson and Peter son (1959) found that subjects respon d ed correctly over 70% of the tim e following 3 sec of distraction. After 18 sec, the percentage of correct responses had drop ped to 15% .One advan tage of requiring serial recall is that separate estimates of forgetting can be obtained for recall based on an ord ered or free-scoring criterion. D irect comparisons of these measu res reveal that ordered recall shows much more forgetting than does free recall, in wh ich items are scored with ou t regard to or igin al serial position (e.g. M arsh, Sebrechts, H icks, & L an dau, 1997;M uter, 1980; Sebrechts, M arsh, & S eam on , 1989 Sebrechts, M arsh, & S eam on , ). M uter (1980 found, for example, little difference between or dered and unordered scoring on immediate tests, bu t there were advantages for unordered scoring of 20±30% at retention intervals of 2, 4, and 8 sec. O ne might in terpret these results as suggesting that order in for m ation is lost rapidly from mem ory, bu t serial recall has the disadvan tage of confounding mem ory for order with m emor y for the item s themselves. A scoring criterion based on ord ered recall therefore measu res not on ly the loss of or der infor m ation , bu t the loss of item in for m ation as well.A m or e accurate assessm ent of the rate at wh ich ord er infor mation is lost comes from experiments using order reconstruction as the retention measu re. In reconstruction tasks,
T H E QUART ERLY JOU RN AL O F EXPERIM EN T AL PS YCH OL OG Y, 1999, 52A (1), 241±251Requ ests for rep rints and cor respondence sh ou ld be sent to James S. N air ne, D epartment of Psychological Sciences, P urd ue U n iver sity, West L afayette, IN 47907±1364, U.S.A.