Coding Position in Short-term Memory

Many formal models of short-term memory have recently been proposed (e.g. Anderson & Matessa, 1997; Brown, Preece, & Hulme, in press; Burgess & Hitch, 1992; Henson, 1998; Lee & Estes, 1981; Lewandowsky & Murdock, 1989; Murdock, 1995; Nairne, Neath, Serra, & Byun, 1997; Neath, 1993; Page & Norris, 1998). An important issue addressed by these models is the problem of serial order: that is, how people store and retrieve a novel sequence of items in the appropriate order. A popular solution to this problem is to assume that each item is coded for its position within a sequence. The present article discusses three different means of coding position. It is argued that the pattern of errors people make when they misrecall a sequence supports the hypothesis that position is coded relative to the start and the end of a sequence. Other evidence, however, suggests that positional coding is also sensitive to temporal factors. A new model is described that reconciles these two strands of evidence.