Academic journal article Canadian Journal of Experimental Psychology

Similarity-Guided Depth of Retrieval: Constraining at the Front End

Academic journal article Canadian Journal of Experimental Psychology

Similarity-Guided Depth of Retrieval: Constraining at the Front End

Article excerpt

Abstract

Lee Brooks has done important work to show that categorization often reflects reliance on specific instances rather than on an abstract representation. His work on the advantages of using a diagnostic hypothesis to search medical stimuli has demonstrated how constraining what one looks for influences clinical reasoning. Similarly, cognitive control can be accomplished by constraining memory retrieval in ways that influence interpretation of a memory probe. Here, we report two experiments in which the similarity of study items constrained how test items were interrogated for an immediate memory test and thereby produced differences in the depth of retrieval. A novel procedure that tests foil memory was used to diagnose differences in similarity-guided retrieval depth.

Lee Brooks forwarded the idea that judgments such as categorization of a stimulus are sometimes based on the similarity of that stimulus to a specific exemplar stored in memory, rather than to a prototype (Brooks, 1978, 1987; see also Jacoby & Brooks, 1984). More recently, Lee and his colleagues have expanded their investigation of "nonanalytic" cognition toward clinical applications, examining the influence that prior instances and the availability of a diagnostic hypothesis have on the reliability and accuracy of medical diagnoses (e.g., Brooks, LeBlanc, & Norman, 2000; Brooks, Norman, & Allen, 1991; Kulatunga-Moruzi, Brooks, & Norman, 2001; LeBlanc, Brooks, & Norman, 2002; Norman & Brooks, 1997).

From our perspective, this work highlights the importance of having front-end constraints on what information comes to mind. As an example, Lee's research has revealed that the generation of a diagnostic hypothesis can be of greater advantage when it is used to constrain, up front, the search for supporting features, rather than when it is synthesized later from having gathered unconstrained data (Norman, Brooks, Colle, & Hatala, 1999). Similarly, here we illustrate how "constraining at the front end" operates in answering a query about one's immediate memory, a notion we refer to as similarity-guided depth of retrieval.

Our approach contrasts sharply with traditional descriptions of recognition memory, such as global matching models (e.g., Gillund & Shiffrin, 1984), which emphasize the quantitative relationship between the strength or familiarity of a memory probe against some decision criterion: By this account, if a probe's familiarity exceeds criterion, it is judged as "old," otherwise, it is judged as "new." This perspective largely neglects the qualitative bases used in making recognition memory judgments, bases that, we argue, are critical for constraining retrieval. Specifically, we suggest that imposing constraints on what comes to mind during retrieval influences the bases by which "old" items are accepted and "new" items are rejected. As will be shown, predictions regarding the fate of memory for new items (foils) follow directly from this line of reasoning.

A recent experiment from our lab illustrates the notion of retrieval depth. Jacoby, Shimizu, Daniels, and Rhodes, (in press) manipulated levels of processing (Craik & Lockhart, 1972) such that participants made pleasantness (deep) judgments for words in one list and vowel (shallow) judgments for words in another list. During a second phase, participants were given a recognition memory test in which they were told correctly that "old" words were ones for which they had earlier made pleasantness judgments, and another recognition memory test for which they were told correctly that "old" words were ones for which they had earlier made vowel judgments. That is, participants were correctly informed regarding the source of old items for each of the tests. As expected, we found the levels of processing effect for these initial recognition memory tests, with higher recognition memory for pleasantness-judged words.

More important, we found evidence for differences in retrieval depth due to specifying the source of the earlier-presented old words. …

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.