Interference in Working Memory is not Displacement nor Competition - It is Limited Destruction

Free recall of 499 Russian college students was measured using the Tarnow Unchunkable Test (Tarnow, 2014) consisting of sets of 3 and 4 double digit items. Most students can remember 3 items but not 4 items and when the 4th item is added the total recall decreases (Ershova & Tarnow, 2016a). Here we describe the interference that results when adding the fourth item. First, we find that interference affects the items differently, evidence that working memory does not consist of identical “slots”; primacy is found to be an important stabilizer. We model the four item experiment as a superposition of the three item result and a perfectly recalled 1st or 4th item and find that the 4th position is affected 2.5 times as much as is the 1st position. Second, contrary to the displacement/competition theory, recall correlations of the added item with the old items (apparently reported for the first time in a free recall experiment) are typically positive. Third these correlations decay exponentially with item-item presentation distance and are symmetric with respect to time reversal. Small negative recall correlations only appear for subjects with the smallest working memory capacities. Third, also contrary to displacement/competition theory, the fourth item is the least likely to be recalled, thus there is not much need for it to displace the other items. This creates a paradox: while displaying the N+1 item decreases the probability of recall of the N items, actually recalling the N+1 item is positively correlated with recalling the other N items: the N+1 item destroys some of the underlying memory system and then functions as a gauge of its own destruction.

interference, working memory, item-item correlations, serial position effect