Process dissociation, cognitive architecture, and response time: Comments on Lindsay and Jacoby (1994).

Abstract

), both of those assumptions were violated. It is argued that these violations were due to (a) an oversimplification of the processing architecture that ignores common stages such as guessing and response selection, (b) an assumption that the more automatic process (word reading) dominates over the intended process (color naming) in determining responses, and (c) an assumption that switching from the more common speeded response instruction (measuring speed) to a deadline response instruction (measuring, accuracy) does not change processing. General implications for applying process dissociation to dynamic tasks are discussed.Process dissociation is a method for determining how task accuracy is influenced by cognitive processes that are capable of either competing or working in conceit (Jacoby, 1991). It has been used primarily to show relative contributions of implicit and explicit memory in different tasks, but it has the potential to be used for other issues as well. Recently, Lindsay and Jacoby (1994) extended process dissociation to a visual identification task, the Stroop task, which appears to have underlying processes that sometimes compete and sometimes work in concert. The process dissociation method puts few explicit constraints on processing. But when we considered in detail how processing would be constrained by the assumptions of the method, the results are problematic, partly in ways addressed by earlier criticisms and partly in ways unique to the new domain. In particular, when one tries to reconcile this work with earlier work on understanding processing architecture of the Stroop task, three issues appear problematic: the assumption of processing independence, the side effects of applying deadlines to a dynamic process, and the characterization of automatic processing.

Item-specific control of automatic processes: Stroop process dissociations

“…In accordance with the independenceassumption,the proportioncongruent manipulation dramatically affected W but had no effect on C (for a discussion of assumptions underlying the application of the PD procedure to Stroop tasks, see Hillstrom &Logan, 1997, andTrainham et al, 1997).…”

Advance task preparation reduces task error rate in the cuing task-switching paradigm

“…This observation suggests that the reduction of RT switching cost mirrors advance preparation, although perhaps not as closely as the reduction of T switching cost. One possible explanation for this coupling is that participants slow down under conditions in which they are likely to err (see Hillstrom &Logan, 1997, andLogan, 1994, for discussions of this control strategy). We also observed a similar pattern in the R switching cost, but we will defer a discussion of this aspect until after the entire empirical picture has been presented.…”

“…The equations that represent these relations are presented in Table 1. The present model is a multinomial processing tree (MPT; Batchelder & Riefer, 1999;Riefer & Batchelder, 1988; see also Lindsay & Jacoby, 1994, and the related criticism by Hillstrom & Logan, 1997, which led us to adopt the MPT).…”

A Dynamic Approach to Recognition Memory

“…However, this procedure will nonetheless produce high estimates of the contribution of recollection even when the data are simulated from a single-process model where it is known that recollection did not generate the data, and when a model with recollection is used to generate the data, the process dissociation estimates of recollection's contributions are incorrect (Ratcliff, Van Zandt, & McKoon, 1995). The process dissociation procedure also relies on the strong assumption that the two processes are stochastically independent of one another; when the assumption is violated, estimates obtained from this procedure are uninterpretable (Curran & Hintzman, 1995;Hillstrom & Logan, 1997). Indeed, it has long been established that there is a correlation between the probability of correct recall and that of correct recognition (Tulving & Wiseman, 1975;Kahana, Rizzuto, & Schneider, 2005), strongly arguing against the process dissociation assumption of independence (item and source memory performance are also correlated; Starns, Rotello, & Hautus, 2014).…”