In defense of functional independence: Violations of assumptions underlying the process-dissociation procedure?

Rotman Research Institute of Baycrest Centre,University of TorontoT. Curron and D. L. Hintzman (1995) claim to have shown that the independence assumptionunderlying the process-dissociation procedure (L. L. Jacoby, 1991) is not justified. Theyargued that correlations between processes at the level of items can result in an underestima-tion of automatic processes large enough to produce artifactual dissociations between processestimates. In contrast, the authors show that the effects of extremely high correlations at thelevel of items are likely to be trivial, and not differential across conditions. Curran andHintzman's dissociations probably reflect violations of boundary conditions for use of theprocess-dissociation procedure, rather than violations of independence.It is important to distinguish between automatic andconsciously controlled memory processes. For example,although amnesic patients are often unable to consciouslyremember previously presented words on direct memorytests, such as recall or recognition tests, they use the wordson indirect memory tests, such as stem- or fragment-completion tests, more often than would be expected bychance (Moscovitch, Vriezen, & Gottstein, 1993). Similardissociations are found in people with normally functioningmemory (Roediger & McDermott, 1993). Comparing directand indirect memory tests has significantly advanced ourunderstanding of automatic and controlled processes. How-ever, performance rarely reflects only one process acting inisolation; that is, controlled processes often influence perfor-mance on indirect memory tests (Holender, 1986; Toth,Reingold, & Jacoby, 1994), and automatic processes affectperformance on direct memory tests (Jacoby, Toth, Y Jeffrey P.loth, Rotman Research Institute of Baycrest Centre, University ofToronto, Toronto, Ontario, Canada.Preparation of this article was supported by Natural Sciences andEngineering Research Council Operating Grant OGP0000281 andby National Science Foundation Grant SBR-9596209. We thankC. M. Kelley and B. Spellman for their helpful comments.Correspondence concerning this article should be addressed toLarry L. Jacoby, Department of Psychology, McMaster University,1280 Main Street West, Hamilton, Ontario L8S 4K1, Canada.Electronic mail may be sent via Internet to jacoby @mcmaster.ca.motel). For an inclusion test, subjects were instructed to usethe stem as a cue to recall an old word or, if they could not doso, to complete the stem with the first word that came tomind. An inclusion test is like a standard test of cued recallwith instructions to guess when recollection fails. Subjectscould complete a stem with an old word either because theyrecollected the old word, with a probability of R, or becausethe old word came automatically to mind, with a probabilityof A. If these two bases for responding are independent, theninclusion performance equals R + A - RA. For an exclusiontest, subjects were instructed to use the stem as a cue torecall an old word but not to use recalled words to completethe stems. That is, subjects were told to exclude old wordsand to complete stems only with new words. In thiscondition, subjects would complete a stem with an old wordonly if the word came automatically to mind withoutrecollection of its prior presentation: A{\ - R) = A — RA.The difference between the inclusion and exclusion testsprovides an estimate of the probability of recollection.Given that estimate, one can compute the probability of anold word automatically coming to mind: A = Exclusion/(1 — R). When these equations were applied to the data fromJacoby et al., results showed that dividing attention signifi-cantly reduced estimates of recollection (.25 vs. .00) but leftautomatic influences almost invariant (.47 vs. .46). That is,the estimates showed a process dissociation similar to thetask dissociations found between direct and indirect memorytests (Koriat & Feuerstein, 1976; Parkin, Reid, & Russo,1990).Table 1 summarizes results from experiments that haveused the process-dissociation procedure to examine theeffects of attention and of presentation duration on R and A.'