Proactive interference in immediate serial recall

Anticipatory access to group-level information in working memory

“…One explanation is that this advantage arises by giving a head-start to a scanning process. Other data from the probed recall paradigm (e.g., Farrell & Lelièvre, 2009b; Hendrikx, 1984a; Sanders & Willemsen, 1978) suggest that in probed recall, the majority of serial positions are accessed by sequentially scanning from the beginning of the list. In the case of grouped lists, Hendrikx (1984a) suggested that grouping provides strong cues to the first item in each group, and that forward scanning is used to access other items (with the exception of the last item on the list, and occasionally the second-to-last item).…”

A revised model of short-term memory and long-term learning of verbal sequences

“…Factors that increase the occurrence of partially matching lists will increase the occurrence of position-specific intrusions (e.g., previous presentations of similar lists, or the Hebb repetition paradigm itself, given the high rate of errors in supraspan lists). Accordingly, it is possible that the observed build up of proactive interference over the first few trials of a block (Keppel & Underwood, 1962;Sanders & Willemsen, 1978), or over longer 1 We increased h in Eqs. (1) and (2) of the 1999 model from 0 to 0.5 for context-item connections, the same value as for the input phoneme to item connections, see Appendix.…”

“…We note that these errors will tend to be towards the end of the list (since there must be a good match between the starts of the previous and current lists for re-use to occur). It follows that the marked primacy gradient seen in longer lists may be due in part to the build up of proactive interference within an experiment (Keppel & Underwood, 1962;Sanders & Willemsen, 1978), and over longer timescales (Lustig & Hasher, 2002;Underwood, 1957). In addition, the frequency of errors due to prior learning will depend on the characteristics of the vocabulary of items from which lists are formed (e.g., digits, letters, phonemically similar letters), and might explain some aspects of performance in experiments using lists of mixed composition (see e.g., Farrell & Lewandowsky, 2002) An important question for the model concerns the total number of context sets that might be required.…”

Immunity to proactive interference is not a property of the focus of attention in working memory

“…Immediate tests of subspan lists are largely immune to PI when correct recall is the dependent measure (Dempster & Cooney, 1982;Humphreys & Tehan, 1992;Sanders & Willemsen, 1978;Tehan & Humphreys, 1995) or when reaction time is used in recognition paradigms (Cowan, Johnson, & Saults, 2005;Halford, Maybery, & Bain, 1988;Wickens et al, 1981). Two studies in particular are relevant to the linkage between capacity and PI.…”