Linking associative and serial list memory: Pairs versus triples.

Abstract

Paired associates and serial list memory are typically investigated separately. An "isolation principle" (J. B. Caplan, 2005) was proposed to explain behavior in both paradigms by using a single model, in which serial list and paired associates memory differ only in how isolated pairs of items are from interference from other studied items. In the present study, 2 experiments identify a critical dissociation between the 2 paradigms, challenging this unified account. Specifically, forward and backward probes were highly correlated for pairs and less so for short lists (triples). The authors asked whether the isolation principle could quantitatively accommodate this type of dissociation. A simulation confirmed that a single model incorporating the isolation principle can adequately explain this and other dissociations, supporting the common processes view.Paired associates and serial list memory paradigms have much in common. Memory for associations tests whether participants can remember that two items are linked (usually by having been presented at adjacent times during a study episode). Memory for serial lists also requires participants to recall the relationships among items, but serial lists include more than two items. One important difference is that with a pair of items, AB, order is irrelevant for many types of memory probes. For example, consider cued recall. When presented with one item of the pair, the participant need only recall the other item, regardless of whether it was the first or second item. In contrast, when presented with a single item from a list of three items, ABC, the participant may require some order information to recall the actual target item and rule out the third, incorrect item. Memory investigators have generally asked questions about associative memory and serial list memory in separate studies, underlining their differences. But models that account for a broad range of data have more explanatory value. If it could be shown that there are common mechanisms underlying behavior in both associative and serial list memory, then researchers could not only explain behavior in those two paradigms, but also potentially explain behavior in related and hybrid paradigms. By investigating associative and serial list memory together, we identify how performance on the two paradigms is similar and how it differs (Experiments 1 and 2), and test whether a single model can account for both the similarities and differences in behavior by using identical model processes (Simulation).The notion that associative and serial list memory are closely related dates back to Ebbinghaus's (1885Ebbinghaus's ( /1913) chaining model. In chaining, the participant learns a list by encoding item-to-item associations between neighboring items in the list (and sometimes between pairs of items separated by a lag): ͕A-B, B-C, C-D, D-E, E-F, . . .͖,where letters denote items (e.g., nouns) and the dash symbol denotes a learned association. The participant could retrieve the list by starting at the first item, A, using ...

The influence of item properties on association-memory

“…The extremely high ''Same" correlation (testing test-retest reliability) is typical of paired associate experiments (Caplan, 2005;Caplan et al, 2006;Rizzuto & Kahana, 2000;Rizzuto & Kahana, 2001). Following Caplan (2005), we mediansplit the participants based on their average ''Same" correlation and for the low-same-correlation group we still fail to find a significant interaction [p > :1 for both sessions].…”

“…Holistic learning would be observed as a high correlation if forward and backward probes measure the same underlying associative strength. Indeed, such a high forward-backward correlation has been observed in several studies of verbal paired-associates learning (e.g., Caplan, Glaholt, & McIntosh, 2006;Kahana, 2002;Rehani & Caplan, in preparation;Rizzuto & Kahana, 2000;Rizzuto & Kahana, 2001) and object-location learning (Sommer, Rose, & Büchel, 2007).…”

“…In previous research, Caplan (2005) and Caplan et al (2006) partly dissociated these two measures in serial lists. Rehani and Caplan (in preparation) found symmetry in mean performance of pairs with reduced correlations.…”

“…Furthermore, a study by Poirier and Saint-Aubin (1996) found that in immediate serial recall, word frequency enhanced item-memory (specifically item retrieval), but had no effect on memory for order. Kahana and Caplan (2002) suggested that asymmetric performance in cued recall of serial lists distinguished serial list-learning from paired-associate learning, challenging attempts to model both association-and serial-list-learning using the same model mechanisms (Caplan, 2004;Caplan, 2005;Caplan et al, 2006;Lewandowsky & Murdock, 1989).…”

Temporal associative processes revealed by intrusions in paired-associate recall

“…According to this view, associations are learned by incorporating the representations of the constituent items into a new holistic representation. Formalized in computational models, this view predicts that the strengths of forward and backward associations are approximately equal and highly correlated (Caplan, Glaholt, & McIntosh, 2006;Kahana, 2002;Rizzuto & Kahana, 2001;Sommer, Rose, & Büchel, 2007).…”