On the locus of dual-task interference: Is there a bottleneck at the stimulus classification stage?

Recent studies have provided evidence that dual-task interference is typically caused by a single-channel bottleneck, but the processing locus of the bottleneck has yet to be pinned down. A bottleneck locus at the response-selection stage is widely advocated, but an earlier locus would be consistent with most previous evidence. Four new experiments used the “locus of slack” method to investigate whether the stages postponed by the central bottleneck include stimulus classification, a very late stage of perceptual processing. The experiments varied stimulus classification difficulty for two different analogueue perceptual judgements. Experiment 1 found only modest absorption into slack for the difficulty of a spatial position judgement. Experiments 2-4 found virtually no absorption into slack for the difficulty of a box-width judgement. These results support a bottleneck locus beginning at or before the stage of stimulus classification and hence prior to the stage of response selection. Other evidence, however, leaves no doubt that response selection is also subject to bottleneck postponement. Two architectures are discussed that can account parsimoniously for both old and new results. One posits a single bottleneck resulting from a unified CPU-like central processor; the other posits multiple bottlenecks resulting from multiple processors accomplishing different substages of central processing.     

The nature of phoneme representation in spoken word recognition

Four experiments used the psychological refractory period logic to examine whether integration of multiple sources of phonemic information has a decisional locus. All experiments made use of a dual-task paradigm in which participants made forced-choice color categorization (Task 1) and phoneme categorization (Task 2) decisions at varying stimulus onset asynchronies. In Experiment 1, Task 2 difficulty was manipulated using words containing matching or mismatching coarticulatory cues to the final consonant. The results showed that difficulty and onset asynchrony combined in an underadditive way, suggesting that the phonemic mismatch was resolved prior to a central decisional bottleneck. Similar results were found in Experiment 2 using nonwords. In Experiment 3, the manipulation of task difficulty involved lexical status, which once again revealed an underadditive pattern of response times. Finally, Experiment 4 compared this prebottleneck variable with a decisional variable: response key bias. The latter showed an additive pattern of responses. The experiments show that resolution of phonemic ambiguity can take advantage of cognitive slack time at short asynchronies, indicating that phonemic integration takes place at a relatively early stage of spoken word recognition.     

Information continuity across the response selection bottleneck: Early parallel Task 2 response activation contributes to overt Task 2 performance

Several studies of dual-task performance have demonstrated Task 2 to Task 1 response priming (backward compatibility effects), indicating some degree of parallel response computation for concurrent tasks and suggesting that the well-established response selection bottleneck (RSB) model may be incomplete. However, the RSB might be considered to remain informationally intact if this early parallel Task 2 response information does not persist across the attentional shift between tasks to contribute to overt Task 2 performance. We used an adapted psychological refractory period paradigm with an additional early transient Task 2 stimulus to examine whether response information generated for Task 2 in parallel with overt Task 1 response selection could persist across the bottleneck to influence eventual overt Task 2 performance. After controlling for potential indirect effects of Task 1 processing stage variability propagating onto Task 2 reaction time via locus of slack effects, we observed reliable and consistent effects of early Task 2 response information facilitating Task 2 reaction times. These effects were observed only when the responses to both tasks of the dual-task pair were compatible, under both univalent and bivalent response mappings across tasks. These findings may represent evidence of a variably sensitive response gating or suppression mechanism in dual-task performance and support the idea that backward response compatibility effects represent transient informational influences on central response codes, rather than later postbottleneck response execution processes.     

注意力资源限制与双任务的相互干扰机制

采用心理不应期研究范式, 两个反应时实验检测了注意力资源分配的特征以及双任务的相互干扰机制。每次实验中, 要求被试快速、相继对高低音辨别任务(T1)和Stroop任务(T2)作出选择性反应, T1和T2间采用6种不同的时间间隔(SOA), 以系统考察不同SOA条件下两个任务的反应时走势。结果发现：(1) 在重叠的双任务情境中, T1的中枢加工导致在T2上出现显著的PRP效应, T2的中枢反应选择对T1的反应选择和反应执行加工同样产生显著的影响。SOA以及T2的难度与复杂度实质性地影响了T1的反应选择和反应执行加工。(2) 当两个任务同时需要进行中枢反应选择加工时, 一个任务占用更多的注意资源将导致另一任务获得较少的注意资源, 注意资源量的多寡直接决定了该任务的加工效率。(3) 两个任务的加工相互影响、相互制约, 这种制约机制不仅仅存在于中枢反应选择阶段, 在反应执行阶段仍然存在。     

双任务情境下心理旋转的并行加工机制

采用心理不应期研究范式,三个反应时实验检测了心理旋转任务和其他认知操作任务能否并行加工的问题。在每个实验中要求被试快速、系列地完成对高低音的辨别任务(T1)和不同旋转角度的正反像辨别任务(T2),T1和T2呈现的时间间隔运用变化的SOA。结果发现：(1)T1的反应选择对T2的反应选择产生了很大的影响,在T2上PRP效应显著。心理旋转的操作成绩随着SOA的缩短而降低。(2)在T1上同样存在随着SOA缩短,反应时增加,正确率下降的趋势。T2的反应选择对T1的反应选择同样产生了显著的影响。(3)T2的反应选择对T1的中枢加工产生了相应的影响,表明当T1的反应选择占据中枢瓶颈时,心理旋转任务和其他认知操作任务在中枢瓶颈中并行得到了有效的加工     

Determinants of central processing order in psychological refractory period paradigms: central arrival times, detection times, or preparation?

Three psychological refractory period (PRP) experiments were conducted to assess the effect of central arrival times at the bottleneck on task order scheduling. In Experiment 1, a visual first task (plus-minus symbol discrimination) was combined with an auditory second task (left-right tone judgement) in a standard PRP paradigm with constant task order. In Experiment 2, the order of the tasks varied unpredictably. In Experiment 3, visual-auditory dual-task trials were randomly mixed with single-task trials. To dissociate central arrival times from stimulus detection times, the perceptual stage of the visual task was extended using stimulus degradation. Most importantly, no evidence for a first-come, first-served principle at the central bottleneck was found with the employed paradigms. Instead, the results indicated that preparation (Experiment 1) and the detection times of the stimuli (Experiments 2 and 3) were the main determinants of central processing order in the present study. In the light of previous research, the results indicate that central processing order can be influenced by various factors. The interplay between these factors seems to depend highly on the conditions and requirements of the employed experimental paradigm.     

Processing order in dual-task situations: The “first-come,first-served” principle and the impact of task order instructions

When two overlapping tasks are processed, they hit a bottleneck at a central processing stage that prevents simultaneous processing of the two tasks. Thus far, however, the factors determining the processing order of the tasks at the bottleneck are unknown. The present study was designed to (re)investigate whether the arrival times of the two tasks at the central bottleneck are a key determinant of the processing order (cf. Sigman & Dehaene, 2006). To this end, we implemented a visual–auditory dual task with a random stimulus order, in which we manipulated arrival time by prolonging the initial, perceptual processing stage (stimulus analysis) of the visual task and compared the effects of this manipulation with those of one impacting the central bottleneck stage of the visual task. Additionally, we implemented two instruction conditions: Participants were told to respond either in the order of stimulus presentation or in the order they preferred. The manipulation of the visual perception stage led to an increase in task response reversals (i.e., the response order was different from the order of stimulus presentation), whereas there was no such increase when the bottleneck stage was manipulated. This pattern provides conclusive evidence that the processing order at the bottleneck is (at least in part) determined by the arrival times of the tasks at that point. Reaction time differences between the two instruction conditions indicated that additional control processes are engaged in determining task processing order when the participants are expressly told to respond in the order of stimulus presentation.     

Multiple bottlenecks in information processing? An electrophysiological examination

When two stimuli are to be processed in rapid succession, reaction time (RT) to the second stimulus is delayed. The slowing of RT has been attributed to a single processing bottleneck at response selection (RS) or to a central bottleneck following the initiation of the first response. The hypothesis of a response initiation bottleneck is mainly based on reports of underadditive interactions between stimulus onset asynchrony (SOA) and the number of stimulus—response alternatives (simple vs. two-choice response). The present study tested the hypothesis of a response initiation bottleneck by recording the lateralized readiness potential (LRP), a brain wave, emerging during or immediately following RS. The LRP findings were consistent with a central bottleneck but did not support the late bottleneck hypothesis. Instead, the LRP provided direct evidence that the underadditive interaction of number of alternatives and SOA is due to an increase of response anticipations in the simple response condition.     

Testing the predictions of the central capacity sharing model

The divergent predictions of 2 models of dual-task performance are investigated. The central bottleneck and central capacity sharing models argue that a central stage of information processing is capacity limited, whereas stages before and after are capacity free. The models disagree about the nature of this central capacity limitation. The central bottleneck model claims that central processing acts on only 1 task at a time and, therefore, constitutes a bottleneck that processes tasks serially. The central capacity sharing model postulates that the central stage is a limited-capacity parallel processor that divides resources among to-be-performed tasks. As a result of this difference, in the psychological refractory period paradigm, the central capacity sharing model predicts that lengthening Task 2 precentral processing will improve Task 1 performance at short stimulus onset asynchronies, whereas the central bottleneck model does not. Results of 2 experiments confirm the prediction of the central capacity sharing model.     

Time pressure effects on information processing in overlapping tasks: evidence from the lateralized readiness potential

Information processing is impaired when two tasks are performed concurrently. The interference between the tasks is commonly attributed to structural bottlenecks or strategic scheduling of information processing. The present experiment investigated the effects of time pressure for the second of two responses on information processing in overlapping tasks by recording the lateralized readiness potential (LRP). Time pressure shortened the latency of the second response by diminishing the time devoted to motoric processing. In addition, task interference decreased under time pressure, which is probably due to the relatively early availability of the central bottleneck stage rather than to increased overlap of central stages. The LRP also provided direct evidence for an additional bottleneck following response selection, possibly due to supramodal refractoriness of response initiation.     

Making two responses to a single object: implications for the central attentional bottleneck.

Experiments with two stimuli and two responses have revealed a central attentional bottleneck and pointed to response selection as its primary locus; however, little has been said about the underlying reasons for this bottleneck. Here we explore these reasons. In the first three experiments, Ss made two separate responses to different aspects of the same object. Interference between selection of the responses persisted, ruling out the possibility that the dual-task bottleneck is caused by the input to the response-selection mechanism being limited to one object at a time. The next four experiments examined what happens when two responses are made to the same attribute of a single object. These experiments show that only one response selection occurred. Hence, the central mechanism is not limited to picking one motor action at a time. Several possible theories about the nature of the bottleneck are discussed.     

Attentional limits in memory retrieval-revisited

Carrier and Pashler (1995) concluded-based on locus-of-slack dual-task methodology-that memory retrieval was subject to a central bottleneck. However, this conclusion conflicts with evidence from other lines of research suggesting that memory retrieval proceeds autonomously, in parallel with many other mental processes. In the present experiments we explored the possibility that Carrier and Pashler's conclusions were distorted by use of an experimental method unfavorable to parallel memory retrieval. New locus-of-slack experiments were performed that encouraged parallel memory retrieval strategies with instructions and feedback, along with the use of "preferred" stimulus-response modality mappings. Results from two psychological refractory period experiments showed that the effect of Task 2 recognition difficulty was consistently absorbed into cognitive slack, with both word and picture recognition. We conclude that the memory retrieval stage of recognition tasks can proceed in parallel with central operations of another task, at least under favorable conditions. Our new findings bring results from dual-task locus-of-slack methodology into agreement with other evidence that memory retrieval is not subject to severe, generic central resource limitations.     

Electrodermal responses to sources of dual-task interference

There is a response selection bottleneck that is responsible for dual-task interference. How the response selection bottleneck operates was addressed in three dual-task experiments. The overlap between two tasks (as indexed by the stimulus onset asynchrony [SOA]) was systematically manipulated, and both reaction time and electrodermal activity were measured. In addition, each experiment also manipulated some aspect of the difficulty of either task. Both increasing task overlap by reducing SOA and increasing the difficulty of either task lengthened reaction times. Electrodermal response was strongly affected by task difficulty but was only weakly affected by SOA, and in a different manner from reaction time. A fourth experiment found that the subjectively perceived difficulty of a dual-task trial was affected both by task difficulty and by SOA, but in different ways than electrodermal activity. Overall, the results were not consistent with a response selection bottleneck that involves processes of voluntary, executive attention. Instead, the results converge with findings from neural network modeling to suggest that the delay of one task while another is being processed reflects the operation of a routing mechanism that can process only one stream of information for action at a time and of a passive, structural store that temporarily holds information for the delayed task. The results suggest that conventional blocked or event-related neuroimaging designs may be inadequate to identify the mechanism of operation of the response selection bottleneck.     

Localizing practice effects in dual-task performance

Practice effects on dual-task processing are of interest in current research because they may reveal the scope and limits of parallel task processing. Here we used onsets of the lateralized readiness potential (LRP), a time marker for the termination of response selection, to assess processing changes after five consecutive dual-task sessions with three stimulus onset asynchronies (SOAs) and priority on Task 1. Practice reduced reaction times in both tasks and the interference between tasks. As indicated by the LRP, the reduction of dual-task costs can be explained most parsimoniously by a shortening of the temporal demands of central bottleneck stages, without assuming parallel processing. However, the LRP also revealed a hitherto unreported early activation over the parietal scalp after practice in the short SOA condition, possibly indicating the isolation of stimulus–response translation from other central processing stages. In addition, further evidence was obtained from the LRP for a late motoric bottleneck, which is robust against practice.     

Dual-task performance with ideomotor-compatible tasks: is the central processing bottleneck intact, bypassed, or shifted in locus?

The present study examined whether the central bottleneck, assumed to be primarily responsible for the psychological refractory period (PRP) effect, is intact, bypassed, or shifted in locus with ideomotor (IM)-compatible tasks. In 4 experiments, factorial combinations of IM- and non-IM-compatible tasks were used for Task 1 and Task 2. All experiments showed substantial PRP effects, with a strong dependency between Task 1 and Task 2 response times. These findings, along with model-based simulations, indicate that the processing bottleneck was not bypassed, even with two IM-compatible tasks. Nevertheless, systematic changes in the PRP and correspondence effects across experiments suggest that IM compatibility shifted the locus of the bottleneck. The findings favor an engage-bottleneck-later hypothesis, whereby parallelism between tasks occurs deeper into the processing stream for IM- than for non-IM-compatible tasks, without the bottleneck being actually eliminated.     

Viewer perspective affects central bottleneck requirements in spatial translation tasks

A psychological refractory period (PRP) approach and the locus of slack logic were applied to examine the novel question of whether spatial translation processes can begin before the central bottleneck when effector or noneffector stimuli are processed from an egocentric (viewer-centered) perspective. In single tasks, trials requiring spatial translations were considerably slower than trials without translations (Experiment 1). Dual tasks consisted of tone discriminations (Task 1) and spatial translations (Task 2) using PRP methods with different manipulations on perceptual and response demands. When a viewer-centered perspective was used, the effect of spatial translation was reduced at short compared with long stimulus onset asynchronies (SOAs) when the potential for code overlap between tasks was removed (Experiments 2, 3, and 4); this finding supports the view that translation processes can begin before the central bottleneck. When an allocentric (non-viewer-centered) perspective was used (Experiment 5), the slowing associated with spatial translation was additive with SOA, suggesting that the processes of spatial translation cannot begin before the bottleneck. These findings highlight the importance of viewer perspective on central bottleneck requirements. Findings are further discussed in relation to the dorsal-ventral model of action and perception.     

Decision and Response in Dual-Task Interference

Experiments with two stimuli (S₁and S₂) and two responses suggest the existence of a stage of processing that cannot be shared between two concurrent tasks. Widespread support has been found for the hypothesis that response selection for Task₂is postponed when the S₁to S₂stimulus onset asynchrony (SOA) is short (Pashler, 1994a). At short SOAs, manipulations which impact Task₂processing prior to response selection (e.g., degradation of stimulus quality) have little effect on Task₂response times (RTs). On the other hand, manipulations which are thought to impact response selection or execution (e.g., Stroop interference) always impact Task₂RTs. There is, however, one particularly compelling demonstration that appears to be inconsistent with the response selection bottleneck hypothesis: Karlin and Kestenbaum (1968) report that the RT difference between detection (i.e., 1-choice) and 2-choice discrimination dramatically decreases with decreasing SOA. Given that the primary difference between detection and discrimination is believed to be at response selection, their result may indicate a processing bottleneck at response execution (Keele, 1973). We fail to replicate the Karlin and Kestenbaum result in two substantive replications of Karlin and Kestenbaum's tasks and procedures. In the single experiment in which Karlin and Kestenbaum's result is replicated, a simple response execution bottleneck account is ruled out by the stability of the difference between 2-choice and 3-choice discrimination times across SOA. Two additional experiments demonstrate that response preparation and task strategy do not substantially contribute to the attenuation of response selection-level effects with decreasing SOA.     

Repetition blindness is immune to the central bottleneck

Theattentional blink (AB) andrepetition blindness (RB) phenomena refer to subjects’ impaired ability to detect the second of two different (AB) or identical (RB) target stimuli in a rapid serial visual presentation stream if they appear within 500 msec of one another. Despite the fact that the AB reveals a failure of conscious visual perception, it is at least partly due to limitations at central stages of information processing. Do all attentional limits to conscious perception have their locus at this central bottleneck? To address this question, here we investigated whether RB is affected by online response selection, a cognitive operation that requires central processing. The results indicate that, unlike the AB, RB does not result from central resource limitations. Evidently, temporal attentional limits to conscious perception can occur at multiple stages of information processing.     

An additive factors analysis of the effect(s) of location cues associated with auditory stimuli on stages of information processing

The additive factors method (AFM) was used as a tool for assessing the locus (or loci) of the detrimental effect of auditory location cues in the chain of (visual) information processing. In the first experiment the location variable was factorially combined with response specificity, which is assumed to affect the response adjustment stage. A second experiment was performed in which movement amplitude, assumed to affect the response programming stage, was manipulated in addition to the location variable and a different variety of response specificity. Finally, the location variable was combined with relative S-R frequency, which is also assumed to affect the response programming stage, in a third experiment. The results of these experiments showed additive effects of the location variable with motor variables. The remaining two experiments were designed to assess the effects of location cues on response selection. In these experiments the location variable was combined with the number of response alternatives. Response speed decreased with an increase in the number of response alternatives. However, the effects of the location variable and number of response alternatives were additive. According to the additive factor logic, then, the results of experiments 1, 2 and 3 seem to indicate that the locus of interference of the location cues is not in the later response stages of the reaction process. The results of the last two experiments were interpreted to suggest that the effects of location cues and the number of response alternatives affect either different processes within the response selection stage or affect different process stages. It was concluded that the latter alternative explains most of the data currently available and that the stimulus identification stage is the most likely candidate for the locus of the location effect.     

The attentional blink is susceptible to concurrent perceptual processing demands

In a rapid serial visual presentation stream processing of a first target (T1) impairs detection or identification of a second target (T2) that appears within 500 ms after T1. This effect characterizes the so-called attentional blink (AB). To evaluate contemporary information-processing accounts of the AB phenomenon in terms of the underlying processing mechanisms the present study examined the potential influence of Task 1 difficulty on the AB effect. To this end, T1 contrast and T1 response requirements were systematically varied across four experiments. Experiment 1 ruled out a mere sensory basis of the contrast manipulation on T2 performance. When only T2 had to be reported (Experiment 2) an AB effect occurred that was slightly modulated by T1 contrast. When report of both T1 and T2 was required in a standard AB task (Experiment 3), the magnitude of the AB depended to a larger extent on stimulus contrast, and it increased further when speeded T1 choice responses were additionally required (Experiment 4). On the basis of the present impact of Task 1 difficulty on the AB effect we conclude that processing limitations cause the AB phenomenon. We discuss such limitations in terms of perceptual (T1 consolidation) and central (response selection) bottleneck processes.