Understanding the dynamics of correct and error responses in free recall: Evidence from externalized free recall

Abstract

A great deal of work has been done in which the dynamics of free recall have been examined in order to understand how individuals correctly recall items stored in their memory systems. Less work, however, has been done in which incorrect responses and the dynamics of correct and incorrect recalls have been examined. That is, although there is a burgeoning literature on false recalls in lists of semantically associated word lists (Deese, 1959;Roediger & McDermott, 1995), less work has been done in which the systematic effects of intrusion errors (items that were not presented on the current list) in standard free recall tasks have been examined. This is largely because these types of errors are generally quite rare, making analysis difficult. The aim of the present study was to examine the dynamics of correct and incorrect responses in free recall in order to better understand how individuals probe their memory systems and edit their responses in order to retrieve information from the recent past. Generate-Edit Models and Context RetrievalMany models of free recall can be classified as generate-edit models (Anderson & Bower, 1972;Kintsch, 1970;Metcalfe & Murdock, 1981;Watkins & Gardiner, 1979). For instance, in the search of associative memory model (SAM; Raaijmakers & Shiffrin, 1980) and its variants (Kimball, Smith, & Kahana, 2007;Mensink & Raaijmakers, 1988;Sirotin, Kimball, & Kahana, 2005), items are first generated (sampled) on the basis of the match between a general list cue and items stored in memory.Once an item has been generated, it is subjected to an editing and monitoring process that checks to see whether the generated item is correct. If the item is deemed correct, it is recalled. If the item is deemed incorrect, it is not recalled. Thus, recall performance is determined, in part, by two stages: a generation stage and an editing stage. Importantly, this type of model suggests that both correct and error responses are generated, but only correct responses (for the most part) are actually recalled because of a highly efficient editing process. This basic type of model naturally accounts for many recall findings and has been used to make novel predictions regarding recall performance.In generate-edit models of this type, it is assumed that retrieval starts with a general context cue to probe the memory system. This cue is determined, in part, by the retrieval question and the retrieval plan that an individual develops. In standard free recall tasks, the retrieval question would be "What items were on the last presented list?" At the start of recall, it is assumed that individuals use this general list context cue to delimit possible items in memory into a search set composed of list items and possibly extralist items. Items are then generated (or sampled) on the basis of the match between information specified in the cue and information stored in the representations (encoding specificity; Tulving & Thomson, 1973). Importantly, less distinct contextual cues will tend to activate many items, leading to ...

Temporal clustering and sequencing in short-term memory and episodic memory.

“…, 2001Davelaar et al, 2005) or region of memory (Oberauer, 2002(Oberauer, , 2009) so they can be easily or directly accessed. The interpretation offered by Unsworth and Engle (2007a) for their results (see also Unsworth, 2009;Unsworth, Brewer, & Spillers, 2010;Unsworth & Engle, 2007b) fits well with current dual-store theories of memory. Working memory is measured and analyzed with a vadety of tasks, including simple span, complex span, mnning memory span, the n-back task, and visual change detection (see A. R. A.…”

“…In particular, two independent random deviates were drawn from a standard normal distribution, and the sum of these was the designated WMC for a particular individual. Accordingly, the model produces independent correlations between WMC and the measures of primary and secondary memory and additionally matches the data in producing a stronger partial correlation between SM and WMC than between PM and WMC (see Unsworth, Spillers, & Brewer, 2010). The first deviate was scaled such that (J), was sampled from a normal distribution with \i.…”

“…One potential challenge to this account is offered by recent data from Unsworth, Spillers, and Brewer (2010). One potential challenge to this account is offered by recent data from Unsworth, Spillers, and Brewer (2010).…”

“…To determine whether the model could account for the challenging pattem of data presented by Unsworth, Spillers, and Brewer (2010), I repeated Simulation 11, with a few changes to capture features of their design and data. First, individual differences were simulated from a continuous distribution.…”

The developmental influence of primary memory capacity on working memory and academic achievement.

“…Secondary memory (which may be seen as a complement to long-term memory) is, on the other hand, characterized by recall that is not spontaneously serial ordered and is likely to be the product of a more controlled or probed search of items not within the current focus of attention. Shelton, Elliott, Matthews, Hill, and Gouvier (2010) have used structural equation modeling to show that primary memory, secondary memory, and working memory are indeed separable latent factors, which contribute differently to variation in fluid intelligence (see also Unsworth & Engle, 2007 ; Unsworth, Brewer, & Spillers, 2010 ).…”