Academic journal article Canadian Journal of Experimental Psychology

Order-Memory and Association-Memory

Academic journal article Canadian Journal of Experimental Psychology

Order-Memory and Association-Memory

Article excerpt

Many of the outstanding puzzling questions in verbal memory research concern how we remember not just which items (stimuli such as words) we studied, but so-called relational memory, memory for associations-pairings of stimuli-and memory for serial order-the precise sequence in which the items appeared (Murdock, 1974). It is easy to come up with examples of everyday situations that demand memory for pairings (e.g., names belonging to faces, Edmonton-Alberta, Toronto-Ontario, famous couples like Michelle-Barack) or sequences (e.g., speeches, recipes, dance choreography). An important research question is whether there is a deep theoretical relationship between association- and ordermemory. As an illustration, consider strategies people deliberately adopt to try to boost their memory functions. For example, to remember a particular pairing of words, such as Goat-Garden, an effective strategy is to make up a mental image that combines two words, such as a goat eating vegetables in a garden, known as interactive imagery (e.g., Bower, 1970b; Hockley & Cristi, 1996; Lowry, 1974; Paivio, 1969). To remember precisely ordered sequences of items, such as a list of unrelated words, or in the more famous, classical examples, concepts in a memorized speech or the seating order of dinner guests, there are three popular strategies: The method of loci, wherein one imagines a familiar environment such as one's own house, and places to-be-remembered items along an imagined path (Yates, 1966); the peg list method, wherein one first learns a standard list of "peg" words associated with numerals (1- bun, 2-shoe, etc.), and then associates each word in a new list to its corresponding peg word (e.g., Bugelski, 1968; Bugelski, Kidd, & Segmen, 1968; Bower, Reitman, 1972); and the link method, wherein one forms images combining each word in a list with the subsequent word (e.g., Roediger, 1980). There are plenty of things one might consider to be common between strategies used to learn sets of associations and serial-order information, such as the effectiveness of visual imagery. However, memory for associations and order have been, for the most part, investigated separately and even modelled separately, even by those who have formalized models that can handle both (e.g., Brown, Neath, & Chater, 2007; Humphreys, Bain, & Pike, 1989; Murdock, 1982). I suggest that this may be a bias specific to our recent tradition of research, that is not a particularly helpful conceptual distinction. Indeed, even considering the mnemonic strategies just mentioned: interactive images, used to support memory for associations, are also used in the link method, used to learn an ordered list of items. The peg-list method, used to learn ordered lists, is in fact built upon interactive images that combine a pair of words, one peg word with one target-list word. Similarly, the method of loci can be viewed like the peg-list method, as a way to learn a list of words by forming associations between each listword and a location or landmark in the memorized environment (Bower, 1970a). These parallels suggest that a goal of memory research should be to investigate association- and order-memory together and to model these paradigms using common mechanisms. Moreover, thinking beyond the lab, there are many real-life situations in which both association- and order information may be required simultaneously, so some kind of synergy, or at least, compatibility, between order- and association-memory must exist, particularly in complex memory behaviours like spatial navigation. It is unclear exactly how to model association- and order-memory together. I suggest that new experimental designs may hold the key to guiding theory in the direction of unified models of relational memory.

I first summarize a major, ongoing, unresolved debate about whether or not the representation of serial lists is based on interitem associations and argue that the polarized nature of this debate has exhausted its utility; rather, there is enough evidence for and against both classes of model, so we should consider hybrid mechanisms. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.