In the psychological refractory period (PRP) paradigm, two unmasked targets are presented, each of which requires a speeded response. Response times to the second target (T2) are slowed when T2 is presented shortly after the first target (T1). Electrophysiological studies have previously shown that the P3 event-related potential component is not delayed during T2 response slowing in the PRP paradigm, but that the lateralized readiness potential is delayed, which suggests a bottleneck on response selection operations but not on stimulus identification. Recently, researchers (Arnell & Duncan, 2002; Jolicoeur & Dell'Acqua, 1999) observed T2 response slowing in an encoding-speeded response (ESR) paradigm where T2 followed a masked T1 that required identification but not a speeded response. T2 response slowing in the ESR paradigm is often indistinguishable from that in the PRP paradigm, prompting some researchers to postulate a common processing bottleneck for the two paradigms. With the use of the ESR paradigm, we observed T2 response slowing and, in contrast to the PRP paradigm, we also observed corresponding P3 delays. The results suggest that dissociable bottlenecks underlie the dual-task costs from the two paradigms.
When people attend to two targets presented within half a second of each other, performance on the second target (T2) is typically impaired or delayed. In the psychological refractory period (PRP) paradigm, two unmasked targets are presented at variable target-target stimulus onset asynchronies (SOAs), with each target requiring a speeded response. The target stimuli and tasks are typically very simple, yet results show that response times (RTs) to T2 are often lengthened dramatically when T2 is presented soon after the first target (T1), as compared with when the targets are further apart in time (for a review, see Pashler, 1994). Historically, one popular view has been that the PRP reflects a bottleneck on response selection operations (see, e.g., Pashler, 1994; WeIford, 1952), in which the mapping from stimulus identity to the required response can proceed for only one stimulus at a time (for a contradictory model, see Schumacher etal., 1999).
Recently, T2 response slowing at short SOAs has also been observed in an encoding-speeded response (ESR) paradigm, in which T1 is masked and requires identification for a later response but does not require a speeded response (see, e.g., Arnell & Duncan, 2002; Jolicoeur & Dell'Acqua, 1998,1999). For example, if T1 is a masked visual digit and T2 is an unmasked auditory tone, participants may be asked to report the pitch of the T2 tone as soon as possible after its presentation and, at the end of the trial, to report the identity of the T1 digit in an unspeeded manner. The presence of T2 response slowing in the ESR paradigm has led some researchers to postulate that conscious stimulus consolidation and response selection require, at least in part, the same limited capacity processing stages and, therefore, result from the same processing bottleneck (Jolicoeur, 1999; Jolicoeur & Dell'Acqua, 1998, 1999). This theory is in contrast to the more traditional view of separate processing limitations for stimulus identification and response selection (Pashler, 1989). Although response slowing appears to be similar in the two paradigms, it is unclear whether the T2 response slowing observed in the PRP paradigm and that observed in the ESR paradigm result from bottlenecks at the same stage of processing or at different stages.
EVENT-RELATED POTENTIAL INVESTIGATIONS
For researchers favoring bottleneck models, the results from electrophysiology experiments have been useful in constraining the locus of the proposed bottleneck in the PRP paradigm. The P3 (or P300) event-related potential (ERP) component is sensitive to stimulus identification and categorization manipulations, and it has been suggested by some researchers (e.g., Donchin, 1981) to reflect consolidation into working memory. …