Stimulus uncertainty and response compatibility in tactile reaction time.
The classic approach to distinguishing between effects of stimulus and response uncertainty was based upon the attempt to devise a task endowed with the stimulus uncertainty of choice RT but requiring no more response processing than simple RT (see Fig. 1). As is well known, Donders (1868), proposed for this purpose a selective reaction of Go/No-Go task (G/NG RT) in which only one of the predesignated stimuli was assigned an overt response. While this task indubitably requires the resolution of stimulus uncertainty, it also requires a decision about whether to respond. Hence, its contrast with SRT does not bear exclusively on stimulus analysis. Donders' inability to persuade his contemporaries that a selective reaction entailed no more response processing than the invariant reaction employed in the classic SRT task disabled his entire subtractivist programme. His view that the stimulus analysis required in his G/NG task was identical to that required in standard CRT has received less critical attention, yet it is also insecure. As Broadbent & Gregory (1962) pointed out, the decision not to respond to a stimulus from the No-Go set can be reached from a more cursory analysis than is required to decide which of several responses is appropriate in standard CRT.
In an alternative proposal, advanced by Wundt (1880), response invariance was reconciled with the introduction of stimulus uncertainty by employing a many-to-one stimulus-response (S-R) mapping, here termed convergent RT (CORT). In the binary case, illustrated in Fig. 1, the participant is instructed to execute the same response whichever stimulus is presented. However, Wundt reasoned that introduction of stimulus uncertainty in this way would be inconsequential if the participant were not obliged to attend to the stimulus variation. So, he added the provision that the participant be instructed not to respond to the mere sensory registration of the imperative stimulus, but to await instead the (putative) instant when the identity of the particular stimulus entered awareness. Fortunately, it appears that such phenomenological instructions are not necessary for experiments to detect a processing cost attaching to stimulus uncertainty. Rabbitt & Vyas (1979), in the only attempt in recent times to apply a convergent simple RT task to the Donders/Wundt problem, reported longer latencies in CoRT than in standard SRT tasks, even in the absence of Wundt's depth of processing instructions (for a review of other applications of convergent RT tasks, see Kornblum, 1973).
Rabbitt & Vyas (1979) had employed auditory stimuli varying in pitch, and visual stimuli varying in horizontal location to summon a single manual key-press, using 4:1 S-R convergence. While they did find reliable latency prolongation under these conditions of stimulus uncertainty, they failed to preclude alternative explanations of this effect. For instance, where more than one stimulus converges on a response, not all of the linkages can possess optimal S-R compatibility. Hence, there is a need to ensure that longer CoRT than SRT is not a product of that reduced compatibility. Rabbitt & Vyas failed to control compatibility. This might be thought to be of little consequence in tasks lacking any obvious component of response selection. However, as it transpires, this view is mistaken.
In the present study, the primary aim was to provide a more rigorous validation of the CoRT task as a means of detecting and manipulating stimulus uncertainty effects in the absence of any response choice requirement. In order to maximize control of compatibility, a tactile RT paradigm was employed in which finger-raising movements were summoned by a mechanical stimulus delivered to the fingertip. In a direct linkage condition, participants were required to move the finger stimulated, whereas oblique linkage involved movement of the finger contralateral to the one stimulated. The same two imperative stimuli were employed in four different RT tasks (Fig. 1). In CORT, as in standard SRT, every trial required the same response. However, whereas in SRT a single predesignated stimulus was used throughout a block of trials (either with direct or oblique response mapping), for the determination of CORT, G/NG RT and standard binary CRT, the two stimuli were randomly but equiprobably intermingled, creating stimulus uncertainty. Therefore, only in CoRT was S-R compatibility (in terms of the direct/oblique contrast) varied within a block. In all the other paradigms, compatibility varied between homogeneous blocks of trials.
The powerful manipulation of compatibility afforded by direct tactile elicitation of finger movements was also intended to maximize the likelihood of detecting reliable S-R compatibility effects in SRT and, if such effects were found, to determine whether they were increased by stimulus uncertainty. Traditionally, it has been assumed that S-R compatibility effects occur at a response selection stage and hence are confined to choice reactions. However, recently Kornblum & Lee (1995) have advocated a broader dimensional overlap conception of compatibility/congruity in terms of '... the degree to which a stimulus and a response set... are perceptually, structurally, or conceptually similar' (p. 875). Moreover, a few authors have reported effects of compatibility on SRT that, while being under 6 ms in magnitude, have nevertheless attained statistical significance (see Hasbroucq, Kornblum & Osman, 1988, for a review). A purely 'logical' appraisal of the demands of the SRT task might seem to suggest that neither the employment of multiple, convergent stimuli nor of low-compatibility responses should prolong latencies in this purported detection task. Since the response in SRT is invariant, stimulus discrimination might seem as unnecessary as response selection.
General methodology of the experiments
The RT apparatus consisted of a specially constructed platform with an array of four finger-stimulators/response-detectors for each hand (Fig. 2). This instrument was a development of one previously described in Frith & Done (1986). It allows up to four fingers of each hand to be stimulated and finger raising responses to be detected. However, in these experiments only the two index fingers received stimulation. Whatever the task, participants rested the tips of each finger on a pair of metal strips, separated by a small gap. The fingertip was positioned on the strips so as to bridge the gap, thereby closing a circuit. When the fingers were correctly positioned, circuit closure was detected by the passage of an electrical current, too weak to be sensed by the participant, through the circuit. The microcomputer continually monitored the state of this circuit, allowing a trial to be initiated only if closure were registered. The required finger-lifting responses could be detected as circuit-breaking and registered by the computer.
The tactile stimulus was delivered by protrusion of a round-tipped cylindrical metal rod (1 mm diameter) between the metal strips on which the fingertips rested. These rods were actuated with a specified force by a solenoid, under computer control. Stimuli were experienced by the participant as a sudden, punctiform impact. Only the two index fingers were ever stimulated. Stimulation and data acquisition were controlled by a BBC Master microcomputer in Expt 1. After this experiment the apparatus was overhauled and recalibrated in order to eliminate, as far as possible, any contribution to the measured RT made by constant electro-mechanical delays. This led to a substantial reduction in the absolute value of RTs without affecting the intra-participant error variance. The BBC microcomputer was also replaced by an IBM-compatible PC.
Participants sat in front of the RT apparatus, facing a white wall, and were required to perform with eyes closed in Expts 1 and 3. Participants were instructed to respond as fast and as accurately as possible according to the particular task instructions given at the beginning of each block. Trials commenced with an auditory warning that initiated a delay randomly varying between 0.75 s and 1.5 s, followed by delivery of the imperative stimulus. Either the participant's response or, in its absence, the expiry of 1 s terminated the trial and initiated 0.5 s inter-trial interval.
Design and analysis
Fully counterbalanced designs with repeated measures on the factors RT Task, Compatibility and Response Hand were employed throughout. The counterbalanced Hand factor involved a mirror image transformation of the S-R mappings. For example, in the G/NG task, if a participant started with the No-Go stimulus being delivered to the left finger, this was followed by the same number of trials in which the No-Go stimulus was delivered to the right finger. All the participants were right handed.
In these experiments, latency data are summarized in terms of the group mean of individual participants' median RT, in each condition. However, all quartile values were examined and effects are only reported if they held for upper and lower quartile as well as for the median. Parametric statistical analyses were conducted on the arithmetic means of the log-transformed latencies. In the first pass through the data, a General Linear Model analysis of variance (ANOVA) for repeated measures was used to detect main effects and interactions. Significant interactions and multi-level main effects were further analysed by post hoc tests for simple effects, using the SPSS MANOVA model. For additional comparisons of subsets of individual mean data, the t test and Wilcoxon test were used.
The first experiment compared the response latencies obtained using tactile stimulation in a 2:1 convergent RT task with those derived from standard SRT and binary CRT tasks, testing participants intensively over three sessions, with each session dedicated to one of the three main tasks. Rabbitt & Vyas (1979) had provided data only for mean RT in the initial and final blocks of trials. The decline in CoRT lag that they found between these two points seems to invite the dismissal of their effect as an ephemeral consequence of initial unfamiliarity with the demands of a novel task. The present experimenters therefore took care to investigate, in detail, the evolution of performance over a large number of trials.
Twelve right-handed volunteers (eight females and four males, with ages ranging from 20 to 27 years) were recruited from final year undergraduates in psychology at the University of Hertfordshire.
Participants underwent three sessions of testing on separate days. A session was entirely dedicated to one of the three RT tasks: SRT, CoRT or CRT. Allocation of tasks to sessions was counterbalanced across three Latin square subgroups. Compatibility and Response Hand were orthogonally counterbalanced within these subgroups. Participants were randomly divided into three groups of four. Each group completed the sessions and blocks in a different, counterbalanced order. All sessions comprised eight blocks of 76 trials.
Of the trials, 0.8% were rejected because the participant failed to respond within a preset 1 s interval. Two types of positive error were tabulated (see Table 1). Response latencies under 100 ms were rejected as anticipatory errors. Less than 0.2% of responses fell into this category, almost all in the SRT task. Responses were also deemed to be incorrect if both possible responses occurred within 400 ms or if the wrong response was made. The incidence of such errors was 2.5%, the preponderance occurring in CRT.
Table 1 displays the group mean latencies of direct and oblique responses in each of the three RT tasks. The latencies of correct responses were entered into a three-way ANOVA, which revealed significant main effects of Task (F(2,22) = 27.59, p [less than] .001), Compatibility (F(1,11) = 64.97, p [less than] .001 and Hand (F(1,11) = 14.97, p [less than] .001). Since the small effect of Hand (SRT: 3 ms; CORT: 17 ms; CRT: 15 ms) showed no tendency whatsoever to interact with any other variable, this is not discussed further. The only significant interaction was that between Task and Compatibility (F(2,22) = 26.71, p [less than] .001), reflecting the increased penalty of oblique mappings as the S-R uncertainty of a task increased. Much of this interaction is owing to the great magnitude of the Compatibility effect in CRT. Nevertheless, the interaction remained significant even when the CRT data were removed. Planned comparisons conducted separately on the high and low compatibility (HC/LC) data revealed that the tendency for CoRT latencies to be longer than standard SRT was significant in both cases (HC: F(1,11) = 5.99,p [less than] .05; LC: F(1,11) = 7.46; p [less than] .05), as was the tendency for CoRT to be briefer than CRT (HC: F(1,11) = 12.19, p [less than] .01; LC: F(1,11) = 81.05, p [less than] .001). An orthogonal comparison showed that even for the SRT task, wherein compatibility was a blocked variable, the latency of responses that issued from direct mappings were shorter than those resulting from oblique mapping (F(1,11) = 7.55, p [less than] .05).
[TABULAR DATA FOR TABLE 1 OMITTED]
Since the experimenters wished to preclude the possibility that any latency effects were owing to a tendency for participants to become ultra-cautious on trials immediately following an error, latencies were examined individually from three trials before to three after every error. Figure 3 displays the mean of these latencies trial-by-trial for each of the three RT tasks. It appears that latencies were relatively prolonged immediately after an Error (err) trial. While the peak magnitude of this effect was similar in all tasks, it scarcely lasted more than a single trial in SRT and CORT. In contrast, recovery in the CRT task did not generally appear to be complete until the third trial following an error (pre/post-Error comparison for CRT for (err+1) z = 2.90, p [less than] .05; for (err+2) z = 1.96, p = .05).
The CoRT task differs from the SRT task in that successive trials need not involve the same stimulus or S-R mapping. To determine whether the longer latency of CoRT was confined to trials involving change, all trials, except the first in a block, were divided into those involving stimulus repetition and those involving alternation. The same analysis was performed on CRT. It transpired that repetition trials were a non-significant 4 ms faster (t [less than] 1) than alternations in CoRT and a non-significant 17 ms (t(11) = 1.08, p [greater than] .10) slower in CRT.
A number of analyses were directed towards the detection of progressive trends in performance. First, the experimenters collapsed the latency data across the counterbalancing subgroups so as to extract sessions (days 1, 2, 3) as a factor. The lack of a consistent trend over sessions suggests that any benefit arising from experience acquired within a paradigm does not transfer successfully between paradigms. The experimenters therefore turned to the examination of latency trends within each of the six combinations of task and compatibility. The sequence of trials was divided into first and second halves. Within these halves, the mean latency of the first and last 10 trials (early/late) were calculated. These means were then entered into a mixed-model ANOVA with order in which the tasks were performed and order in which the compatibility conditions were performed as between-subjects factors and first/second half and early/late trials as within-subject factors. The only significant differences were that late trials showed briefer latencies than early trials (F(1,11) = 19.70, p[less than] .01). This trend increased with task uncertainty (F(2,22) = 3.98, p [less than] .05). These data can be inspected in Fig. 4, which reveals that improvement was almost entirely confined to the very first block of trials in a condition. After a brief initial reduction, latencies fell no further unless the condition involved both low compatibility (LC) and uncertainty. In particular, the final 10 latencies in a condition were significantly shorter than the overall average of the remaining trials only for LC CoRT and LC CRT. While the initial decline in latency was somewhat greater for CoRT than SRT, thereafter no further reduction of the difference ensued. The final 10 SRT trials were still a significant 25 ms (t(11) = 3.39, p [less than] .05) shorter than the final 10 CoRTs.
With regard to the impact on simple reactions of stimulus uncertainty and of response compatibility, the findings revealed an interesting symmetry. Despite the apparently inconsequential nature of the stimulus variation and of the oblique S-R mappings in tasks with a single invariant response, both of these manipulations turned out to lengthen RT. The effect here of stimulus uncertainty confirms and extends Rabbitt & Vyas's (1979) finding that the use of convergent S-R mappings to introduce seemingly inconsequential stimulus uncertainty into a simple reaction task prolongs RT, despite the apparent irrelevance of the stimulus variation. Moreover, the experimenters were able to exclude stimulus repetition per se, criterion shifts following an error and uncontrolled variation in S-R compatibility as the source of the longer latencies found in CoRT. The effect was evident for each quartile of the participants' RT distributions, albeit greater at Q3. Nor did the stimulus uncertainty effect appear to be an ephemeral phenomenon. Direct responses to tactile stimuli, whatever the task, became asymptotic after little more than 10 trials. Although the magnitude of this brief initial improvement was greater for CoRT than for SRT, a reliable SRT advantage was maintained without further diminution for the several hundred remaining trials.
The other main feature of these results was the finding of a reliable S-R compatibility effect for simple reactions. While some previous studies of standard SRT have obtained reliable compatibility effects, these have not generally exceeded a magnitude of 6 ms (see Bashore, 1981; Hasbroucq et al., 1998; Marzi, Bisacchi & Nicoletti, 1991, for reviews). This has led to such effects either being dismissed as the product of a bias towards publication of results that attain statistical significance, or to attribution of any compatibility effects found in SRT to a different basis from that invoked to explain the relatively large effects obtained in CRT. More recently, Hommel (1995, 1996, Expt 1) has demonstrated that response variation, in a task where a 100% valid precue entirely removes response uncertainty, is a sufficient condition for larger compatibility effects. Even in the standard SRT task of the present study, trials involving a direct S-R mapping exhibited latencies that were 20 ms shorter than those with an oblique mapping. Furthermore, the size of this compatibility effect was significantly increased by the introduction of stimulus uncertainty. When stimulus uncertainty was combined with Low Compatibility (LC CoRT), latencies were 63 ms greater than in standard SRT (see Table 1).
Hommel (1996) has also shown that employment of an intermittent secondary task can increase the magnitude of compatibility effects obtained in a primary task that lacks response uncertainty. However, such manoeuvres compromise the classical conception of SRT as the product of a single task, employing a single, invariant response. Moreover, both of the modifications of the standard SRT task made by Hommel seem likely to have increased the concurrent burden on working memory. Several authors (e.g. Frith & Done, 1986; Goodrich, Henderson, Allchin & Jeevaratnam, 1990) have shown that SRT tasks are particularly vulnerable to imposition of a secondary task. Such findings led Henderson & Dittrich (1998) to conclude that competing demands upon attention or working memory may interfere with strategic preparations that confer a special advantage on normal, simple reactions. In contrast, the present results show that compatibility effects at least as large as those found by Hommel can be obtained while adhering strictly to the classic concept of a simple reaction, provided that the manipulation of compatibility is as powerful as that available in the tactile stimulation RT paradigm.
In the tactile RT paradigm, oblique responses require involvement of both index fingers, even for simple reactions, since one finger must receive the stimulus and the other execute the response. So, the question can be posed of whether the compatibility effects obtained in the tasks using a single, invariant response are a peculiarity of this involvement of both index fingers. Admittedly, all that is required in the oblique SRT task is for one finger to be held in readiness for stimulation, and the other for responding. In the CoRT task, the change over of hand assignment required by the counterbalanced design did not take place until the completion of over 300 trials. Throughout that run, the passive finger was dedicated to registration of stimuli, whereas the response finger was required to executive the response when either finger was stimulated. Moreover, Hommel (1996) failed to obtain any effect on simple RT of positioning an irrelevant finger on an active key or inactive support. However, other evidence has been found for effects of hand posture on CRT. It appears that the well-established advantage in binary CRT of responding with one finger of either hand rather than two fingers of the same hand can be abolished simply by requiring the participant, in the bimanual task, to rest another finger of either hand on response keys. This effect appears to obtain despite the fact that these keys were never actually used (Alain, Taktak & Buckolz, 1993; Clark, 1982; Reeve & Proctor, 1988). It was therefore decided to investigate the role of bimanual hand posture in the next experiment. In order to release the passive hand from the need to be on the tactile platform with the index finger ready to receive a tactile stimulus, the experimenters simply transformed the tactile RT paradigm into a visual one.
Six right-handed undergraduate volunteers aged from 20 to 30, of whom four were female, participated in this study.
Design, tasks and procedure
Adaptation to the tactile stimulator to produce visual imperative signals was achieved by repositioning the participants' fingers on the metal strips so that the stimulator rod, when extruded, no longer touched the finger but could, instead, be detected as a visible event. This allowed comparison to the previous bimanual condition, in which both hands were positioned on the tactile platform in a potentially active mode, with a unimanual condition, in which the non-responding hand rested in the participant's lap. In direct-mapping trials, the probe always appeared just beyond the response finger. In oblique-mapping trials, it appeared just beyond the inactive contralateral finger when the bimanual hand posture was adopted. While it appeared at the same physical locus in the unimanual condition, in this case it was spatially far removed from the resting hand in the participant's lap. Otherwise, the design and procedure followed that of Expt 1, save for changes to the number of trials and counterbalancing required by the addition of the Hand Posture factor. Since the unimanual condition was only applicable to RT tasks involving a single overt response, a binary Go/No-Go task was substituted for the standard CRT task, retaining the SRT and CoRT tasks. The four repeated measures factors (RT Task, Response Side, Hand Posture and Compatibility) were administered in a counterbalanced order, derived from a Williams square. The three sessions devoted to a task each contained eight blocks of 80 trials, so the entire data set comprised 6 x 1920 trials.
Results and Discussion
Summed over types, the incidence of errors was less than 1.2%. The majority of these comprised anticipatory errors in the standard SRT task. With unimanual hand posture, the overall error rate was actually slightly higher than with the bimanual posture. The responses issuing from oblique mappings in the CoRT task, which should have been maximally exposed to any increased conflict, showed exactly the same error rate (0.5%) in the bimanual and unimanual conditions (see Table 2).
[TABULAR DATA FOR TABLE 2 OMITTED]
The latencies of correct responses were entered into an ANOVA for repeated measures. As before, significant effects were found for Task (F(2,10) = 32.28, p [less than] .001), Compatibility (F(2,10) = 18.16, p [less than] .01) and their interaction (F(2,10) = 5.30, p [less than] .05). A significant interaction was also found between RT Task and Response Side (F(2,10) = 5.27, p [less than] .05). However, the requirement to rest both hands upon the tactile RT platform yielded latencies that were actually slightly, though nonsignificantly, briefer overall than those obtained with only one hand on the tactile platform (Table 2).
Taken together, the latency and error data allowed the experimenters firmly to reject the notion that the bimanual set-up, which they had been obliged to employ in order to test oblique tactile reactions, introduced a degree of response uncertainty or conflict into the oblique versions of the SRT and CoRT tasks. All further analyses were conducted on data collapsed over the Hand Posture factor.
In order to examine the effect of stimulus uncertainty, separate planned comparisons were conducted of the latencies obtained in the SRT and CoRT tasks, under High Compatibility (HC) and Low Compatibility (LC) conditions. These revealed significant simple effects of stimulus uncertainty both for the LC (F(1,5) = 25.72, p [less than] .01) and HC (F(1,5) = 13.71; p [less than] .05) conditions. Likewise, the simple effect of compatibility was significant in the CoRT task (F(1,5) = 19.38, p [less than] .01). However, the effect of compatibility on SRT performance just failed to attain significance (F(1,5) = 5.24, p = .07), reflecting the fact that, as expected, the overall magnitude of the compatibility effect had diminished with the replacement of tactile by visual stimulation. Moreover, unlike the previous experiment, the compatibility effect found here was not reliably larger for CoRT than for standard SRT. A complex pattern of interactions involving Hand seemed to indicate that the response latencies of direct mappings were briefer when executed with the left index finger, whereas those for oblique mappings appeared, if anything, briefer when executed with the right index finger.
When the visual G/NG latencies from this experiment are compared to those obtained for tactile CRT in Expt 1 (see Table 1), it is evident that compatibility had a much greater impact on tactile CRT, where the participant has to decide which of two overt responses to execute, than on visual G/NG RT, where the decision is whether to execute a response. In such a comparison, modality is perfectly confounded with the nature of the response choice. The linkage employed in the present experiments between a tactile stimulus delivered vigorously to the tip of the finger and the response of raising that finger seems a peculiarly direct and natural one. It was therefore expected that the tactile paradigm would have near to maximal potency as a means of manipulating S-R compatibility, a view consistent with the magnitude of the compatibility effects obtained for simple reactions in Expt 1. The question arises, therefore, of whether the large difference between the compatibility effects obtained in the two choice tasks is at least partly owing to a general attenuation of compatibility effects with visual stimulation. The experimenters attempted to resolve this question in the next experiment.
The immediate aim of this experiment was to enable a controlled comparison of direct and oblique versions of the G/NG RT and standard CRT tasks, within the tactile paradigm. There was also a desire to have a comparable baseline of CoRT data.
The choice data were derived from four male and two female undergraduate volunteers ranging in age from 20 to 43. Baseline CoRT data were derived from a comparable group comprising two male and four female undergraduates, whose ages ranged from 21 to 36.
Apparatus and procedure
For reasons that lie outside of present concerns, stimulus intensity had been varied as a three-level, blocked variable in this experiment. Since the effects of intensity turned out to be very small and did not interact with other factors in the experiment, data exposition was simplified by confining attention to those blocks of trials which utilized the intermediate level of intensity. This was, in fact, the level employed in the previous experiments. Participants underwent testing on HC and LC CRT and G/NG RT tasks, in a fully counterbalanced, repeated measures design. Observations were made in two sessions per RT task, so the data set comprised 6 x 640 trials.
Since recalibration of the tactile apparatus had substantially reduced the latencies obtained in all studies performed after Expt 1, a fresh baseline was provided in the form of a new set of CoRT data obtained on a comparable group of undergraduates under the same conditions.
The incidence of anticipatory errors fell just short of 2% in CoRT, but was negligible in the G/NG task and zero in CRT. Conversely, the overall incidence of wrong responses was greatest (4.6%) in the CRT task, intermediate (1.8%) in the G/NG task and negligible (0.2%) in CoRT. Moreover, wrong responses, unlike anticipations, were more common in each task when compatibility was low. Of the wrong responses in the G/NG task, 95% were failures of response suppression on No-go trials.
The latencies of correct responses, in the two tasks involving response choice, were subjected to an ANOVA with RT Task, Compatibility and Response Side as repeated measures factors. Main effects were found for compatibility (F(1,5) = 24.36, p [less than] .01) and for Task (F(1,5) = 6.38, p [less than] .05), with G/NG latencies being more extended, overall, than CRT. A significant cross-over interaction between Compatibility and Task (F(2,10) = 35.64, p [less than] .01) showed that compatibility had a much greater impact upon CRT than on G/NG RT. Whereas, when the S-R mappings were oblique, latencies in the CRT task were marginally longer than in the G/NG task, when mappings were direct, CRT latencies were considerably shorter than in the equivalent G/NG task. Post hoc tests for simple effects revealed that only the latter effect was significant (F(1,5) = 45.13, p [less than] .001). Moreover, this effect was sustained even when each participant's data set was purged of blocks of trials containing one or more errors. These data are displayed in Fig. 5.
Comparison of the latencies for CoRT with those obtained in Expt 2 using visual stimulation (see Table 3) revealed that tactile CoRT (179 ms) was slightly shorter, overall, than visual CoRT (207 ms). This contrasted markedly with overall G/NG RT, which was much longer with tactile (322 ms) than with visual (261 ms) stimulation. However, this substantial prolongation of G/NG RT in the tactile paradigm was not accompanied by any increase in the magnitude of the compatibility effect, assessed in terms of the RT difference between oblique and direct mappings.
[TABULAR DATA FOR TABLE 3 OMITTED]
Two aspects of these data are particularly remarkable. First, G/NG response decisions are relatively slow when tactile stimuli rather than the visual stimuli of Expt 2 were used (see Table 3). Second, latencies in the standard CRT task were distinctly shorter than in G/NG RT when S-R mappings were direct. If these findings are taken in conjunction with the constant size of the direct/oblique contrast, in G/NG RT (37 ms with visual stimulation and 36 ms with tactile) a single composite account suggests itself, as follows.
Withholding a response when confronted with a No-Go stimulus in the G/NG RT task involves an S-R mapping that is inherently low. Furthermore, an overt response is likely to be particularly difficult to withhold when the stimulus has the immediacy of a mechanical deformation of the body surface and when it is delivered to a finger that has served as an effector in other conditions. Moreover, in the tactile paradigm, the linkage between stimulation of the fingertip and raising/withdrawal of that same finger seems likely to be especially prepotent. However, those very qualities which make for automaticity in ultra HC tactile CRT (e.g. Frith & Done, 1986; Goodrich, Henderson & Kennard, 1989; Leonard, 1959) impose an executive control requirement on the G/NG task which carries with it time costs when the prepotent response must be repressed, on No-Go trials. These difficulties in withholding a response to tactile No-Go stimuli result in the much longer latencies found here for Go responses when tactile rather than visual imperative stimuli are used. The modality differences here are considerably greater than those evinced by oblique mappings in the simple reactions of Expts 1 and 2.
The truncation of the upper range of possible compatibility values in G/NG tasks caused by the inherently low compatibility of No-Go mappings reduces the impact of any direct/oblique manipulations superimposed on the task, as compared to the effect of these same manipulations on standard CRT. Greater truncation in the tactile paradigm may also mask the potential for larger direct/oblique differences with tactile stimuli, evident in simple reactions. Such truncation may also be used to explain the unprecedented finding of significantly shorter CRT than G/NG latencies in the direct-mapping condition. This can be ascribed to severe limits on the speed of Go responses set by the low compatibility of the alternative No-Go mapping, with the consequent need to generate sufficient preparatory inhibition to prevent false positive responses. To speak of a high compatibility condition in the tactile G/NG task is probably misleading.
A referee has suggested that the seemingly shorter latency of CRT than G/NG RT, with direct S-R mappings, is merely the consequence of a speed/accuracy tradeoff, since the error rate was higher in the CRT task (3.8%) than the G/NG task (1.9%). It would be surprising if participants had to pay a price of fully 50 ms (the amount by which direct mapping G/NG RT exceeded direct CRT) in order to reduce their CRT error rates by this amount. However, in the absence of systematic manipulations of the participant's criterion, this can only be a matter of more or less plausible conjecture. In the experimenters' experience, participants very rarely make errors on the single, overt response of the G/NG task. In fact, all of the errors in the direct-mapping condition consisted of false positives on No-Go trials (the error rate for this condition being therefore almost identical with the overall CRT error rate).
Donders (1868) maintained that RT would invariably be shorter when participants were required to respond selectively to one of a pair of stimuli than when they had to execute a different response to each. His claim was widely disputed on various conceptual grounds, but Broadbent & Gregory (1962) seem to have been the first to mount an experimental challenge to his thesis. They used word naming and vibrotactile finger-stimulation paradigms to compare G/NG RT to standard binary CRT, under conditions of high and low compatibility. They reported that the G/NG advantage was confined to the LC conditions, HC leading to a very slight advantage for CRT that did not approach significance. Subsequent investigators, working with less potent manipulations of compatibility than those employed here, have tended to confirm the conclusion that reliably shorter latencies in G/NG as opposed to standard CRT tasks are confined to low compatibility tasks (e.g. Callan, Klisz & Parsons, 1974; Hackley, Schaffer & Miller, 1990; Hommel, 1996). Such a finding is very difficult to reconcile with Donders' assertion that the tasks differ only in the lack of a response choice element in the G/NG task. Yet, apparently, Donders' view still has its adherents. Berlucchi, Crea, DiStefano & Tassinari (1977) argued that their (somewhat unusual) failure to find a compatibility effect in G/NG RT was consistent with their supposition that compatibility effects were absent in SRT, on the grounds that neither task required a choice among responses. Hommel (1996, p. 550) seems to adopt a similar stance on the requirements of the decision whether to respond in the G/NG task. He writes: 'As the response is constant throughout a particular block there is no response selection process.' Hommel goes on to propose that performance in the G/NG task draws on two strategies: an SRT-like strategy involving a high degree of response preparation, and a CRT-like strategy in which participants delay response preparation until they have identified the Go signal. This two-strategies account commits Hommel to the view that G/NG RT must fall within the range set by performance on comparable SRT and CRT tasks, a prediction which cannot be reconciled with G/NG latencies that are reliably longer than binary CRT latencies.
Composition of tasks from component S-R mappings
The design of the matched, binary CoRT, G/NG and CRT tasks used in these experiments and diagrammatically represented in Fig. 1 is such that all utilize the same two, equiprobable stimuli, either of which can be presented on any trial. In the instructions to participants, each stimulus is assigned a mapping rule. Two subsets of task can be distinguished, according to whether they share a direct or an oblique mapping as the common constituent of the set. (In the diagrams of Fig. 1, the common constituent is the leftward mapping shared by all members of the set.) Within these direct and oblique subsets, the task is defined by the nature of the response required to the alternative stimulus. The nature of this mapping also exercises a major effect on the speed with which the common constituent can be executed (e.g. De Jong, 1995; Duncan, 1977, 1978). For example, at least for tactile stimuli, the latency of raising the left finger when it was stimulated appeared to be slightly greater when the alternative stimulus was a No-Go one (direct G/NG RT), than when it required a direct mapping (direct standard CRT). A more complete design would have included a CRT task comprising one direct and one oblique but non-convergent mapping. This would have enabled the experiments to determine the relative demands of oblique and No-Go mappings as alternatives to the direct common constituent. However, since at least one more response finger would have had to be conscripted for tasks involving oblique but non-convergent mappings, several more control conditions would have been needed, making for an extravagant design. Note that the comparison of CoRT with either of the choice tasks is uninformative, despite the fact that the alternative mapping is an oblique one, because the convergent mapping allows the task to be dealt with in terms of a single, unconditional response rule (see below). It appears that tasks involving mappings, one of which is high and the other low in compatibility, yield disproportionately long latencies, prolongation being greater for the HC mapping (Proctor, Lu, Wang & Dutta, 1995; Stoffels, 1996; Van Duren & Sanders, 1988).
The vigilant reader will have noted that throughout this study the experimenters have employed repeated measures designs. Could it be that compatibility effects are only evident in simple reactions after the participant has experienced both index fingers in an executive role in a choice task? With this in mind, the experimenters examined the counterbalancing of order of testing, with particular regard to the magnitude of the compatibility effect in simple reactions. No systematic order effects were evident.
Stimulus uncertainty in simple reactions
The primary objective in this study was to validate the convergent simple reaction task as a means of detecting and manipulating stimulus uncertainty effects in the absence of any variation of the required response. The comparison of standard SRT with CoRT in Expts 1 and 2 allowed the experimenters to demonstrate a processing cost attached to stimulus uncertainty in the form of longer CoRT than SRT latencies. In order to preclude uncontrolled variation in S-R compatibility as a possible cause of this CoRT lag, a tactile RT paradigm was adapted that allowed exceptionally powerful manipulations of compatibility. By these means the experimenters were not only able to resolve ambiguities in the outcome of Rabbitt & Vyas's (1979) pioneering study of Convergent RT, but also to show that simple reactions were indeed sensitive to variations of compatibility. Moreover, these compatibility effects were larger when stimulus uncertainty was combined with low compatibility (i.e. in oblique-mapping CoRT). This interaction provided further reassurance of the psychological reality of the stimulus uncertainty effect. Taken together, these results seem to have general implications for understanding simple reactions. The demonstration that RT tasks which lack response uncertainty (or even predictable response variation) are sensitive both to stimulus uncertainty and to S-R compatibility has as its principal casualty the minimalistic view of simple reactions as served by a cognitively impoverished subsystem that is impervious to strategic considerations (see Henderson & Dittrich, 1998).
What processes give rise to the delayed responding found under conditions of stimulus uncertainty? The most obvious possibility is that knowing in advance the spatial location where a target will appear somehow assists in its detection (Posner, 1980), perhaps because spatially selective attention can improve the signal to noise ratio. In the present experiments, participants were required to keep their eyes closed during performance of the tactile task, so fixation of the stimulus or response finger (Bradshaw, Howard, Pierson, Phillips & Bradshaw, 1992) cannot have played a role. Comparison of SRT with CoRT using a task in which the locus of the stimulus source does not vary should allow one to determine whether it is predictability of the stimulus' identity or merely of its location that forms the essential basis of the advantage of single-stimulus tasks.
It should not be assumed without question that the processes which set CoRT apart from SRT exhaust all possible costs of stimulus uncertainty. Wundt was probably correct in supposing that the extraction of sufficient information to serve response selection in choice tasks requires more extensive stimulus analysis/discrimination than is involved in the basic form of CoRT task. The present working assumption must therefore be that the CoRT task affords useful leverage on the role of stimulus uncertainty in the SRT advantage, but does not capture it in its entirety. Nevertheless, it would be interesting to determine whether participants who exhibit a neuropsychological impairment of the SRT advantage (e.g. Bloxham, Mindel & Frith, 1984; Botzel, Mayer, Oertel & Paulus, 1995; see also Henderson & Dittrich, 1998 for elaboration) fail to show any impairment of CoRT. Such an outcome would suggest that their deficit lay in the ability to utilize stimulus predictability in order to expedite stimulus detection. For analogous reasons, it would be worth investigating whether the greater sensitivity of SRT than CRT to interference from a secondary task holds equally for CoRT.
The tactile RT paradigm
The neglect of tactile effector cuing as a means of maximizing S-R compatibility in keying tasks is puzzling, since its primacy among skeleto-motor channels seems beyond dispute. In the present study, stimulus modality turned out to be a major determinant of RT. The prolongation of RT by oblique mappings was greater in both SRT and CoRT when tactile rather than visual stimuli were used, but no different in the G/NG RT task. Overall latency in the G/NG task was, however, much greater when the stimuli were tactile rather than visual, despite the fact that tactile CoRT latencies had actually been briefer than visual ones (see Table 3).
Most extant models of compatibility (e.g. Fitts & Seeger, 1953; Hasbroucq & Guiard, 1991; Hasbroucq, Guiard & Ottomani, 1990; Proctor, Reeve & Weeks, 1990; Simon, 1990) have simply not been designed to accommodate the possibility of compatibility effects in tasks lacking an explicit component of response choice. Nor have they even been constructed to supply an account of the G/NG task. These two sets of problems are discussed in turn.
Compatibility effects in tasks lacking response uncertainty. Where simple reactions have been shown to be sensitive to compatibility, the source of the effect must surely be automatic processing of ramifications of the stimulus that are strictly irrelevant to the task. This should not be surprising. After all, most stimulus analysis is conducted without volitional control and it is not difficult to find examples of participants' inability to attend exclusively to a set of attributes that seem both elementary and clearly defined. In the domain of printed words, a pair of examples can be found which possess a pleasing contrast. Linguistic content and physical format can act either as target or as distractor variable, depending on whether one is the victim of the Simon effect or Stroop interference (Simon, 1990; Stroop, 1935). However, such demonstrations have almost invariably been drawn from tasks possessing response uncertainty. Tasks in which the response never varies might seem to be quite another matter.
In the present study, only tactile stimuli yielded a reliable compatibility effect when neither stimulus nor response varied (oblique SRT). Moreover, this effect was larger in a CoRT task that invites the application of a rule taking the form 'whatever stimulus arrives, react with this response'. Despite the applicability of this rule, binary uncertainty concerning the finger to receive stimulation extended response latency. This must reflect a limitation in one's ability to attend selectively to one response.
Latencies were further extended when the stimulus was one habitually associated with a response other than that assigned in the laboratory. It is suggested above that the SRT/CoRT contrast is best viewed as attesting to participants' ability to benefit in detection speed from selective attention to the location or form of an entirely predictable imperative stimulus. In contrast, the oblique-mapping effect stems from the participant's inability to focus exclusively on the unvarying response required, to the exclusion of the more nature one. It is this inability to ignore irrelevancies rather than the inability to prevent their initial activation that merits enquiry.
The experimenters share with Hommel (1996) the view that responses specified by oblique mappings must compete with unwanted direct responses at a late stage of processing. The process of conflict resolution can be modelled either in terms of relaxation brought about by the interplay of excitation and inhibition through lateral connections between competing response units, or by a verification operation that is engaged when independent, parallel pathways yield inconsistent outputs.
Kornblum, Hasbroucq & Osman (1990) postulated a categorical distinction between Automatic Direct Activation (ADA) of response units that stand in some sort of correspondence to each other and Stimulus-Response Translation, which serves response selection in the absence of correspondence. ADA was held responsible for the conflict that occurs on oblique-mapping trials between the oblique and direct solutions. The amount of activation was assumed to be proportional to the degree of physical or conceptual similarity between stimulus and response, a quantity for which Kornblum et al. (1990) reserved the term dimensional overlap (see Kornblum & Lee, 1995, for elaboration). Subsequent work by Proctor & Wang (1997) has shown that dimensional overlap is not a unitary concept, as supposed by Kornblum et al. (1990). The model advanced by Kornblum and his colleagues has several interesting properties, many of which serve to illustrate interesting parallels between print to sound translation and the present RT problem domain. However, these lie outside the present concern.
A major problem for any model that incorporates an ADA mechanism is the specificity of response activation. Spatial congruity may result in the activation of a left field response, but how does the particular movement comprising, for instance, finger-raising come to be activated? The specification of this particular response derives from the laboratory task instructions rather than the state of the organism. Indeed, the degree of constraint with which an imperative stimulus specifies the precise form of a prepotent response shows considerable natural variation. At one extreme, the saccadic eye movement attracted by a novel target has its kinematics fully determined by the retinal error computation. At the other, imperative signals distinguished only by hemifield may specify nothing more than the hemifield in which the response must lie. The directness of ADA in the Kornblum et al. (1990) model inheres in its activation of the motor programme for the response rather than a more preliminary representation, so this issue must be faced.
An alternative formulation by Stoffels (1996) restricts the ADA pathway to situations where congruence holds fully and for all trials. Important support comes from data by Van Duren & Sanders (1988) showing that where congruent and incongruent mappings are intermingled, RT for the congruent mappings suffers greater prolongation. Interestingly, the response delay fails to materialize in cases in which the mapping rules are not specified well in advance of response initiation (De Jong, 1995; Shaffer, 1965).
Compatibility and Go/No-Go trials. If the available models of compatibility neglect simple reactions, they are certainly no more hospitable to the G/NG RT task, which the experimenters take to be inherently a mixed compatibility task. One of the difficulties of interpreting G/NG RT consists in knowing how No-Go responses are internally represented and how response decisions are reached. Indeed, to speak of 'No-Go' responses will seem alien to some. In addition to the task's sensitivity to modality, it was found that latencies were actually substantially shorter in the direct-mapping version of tactile CRT than in the equivalent G/NG task, albeit accompanied by a slightly higher error rate. The only other comparison of tactile G/NG and CRT known to the present experimenters (Broadbent & Gregory, 1962) failed to detect a CRT advantage, but differed in important procedural details (e.g. vibrotactile stimuli, method of response detection, timing in centiseconds).
In Henderson & Dittrich (1998) the present experimenters lament the general neglect of preparatory strategies in discussions of RT performance. Inhibitory strategies and the strategic deployment of selective attention seem likely to play a particularly important role in G/NG RT. As previously stated, it is assumed that participants in the CoRT task adopt a response set (or rule) of the form 'make this response whatever stimulus arrives'. The sudden unheralded introduction of a novel stimulus might be informative in this regard. The direct version of the G/NG task seems, on the other hand, more amenable to a strategy comprising selective attention to the Go stimulus. This should attenuate any tendency towards false positive reactions to the No-Go stimulus. Such a strategy, however, is of no avail in the oblique version, where it would increase the likelihood of unwanted direct responses to a Go stimulus requiring an oblique response.
In tasks amenable to an unconditional response-inhibition rule, preparatory deployment of inhibition is likely to be effective to the extent that it is selective. However, the experimenters' intuition is that preparatory inhibition of unwanted responses (whether an oblique response is called for or no response at all) is not entirely selective, but tends to spill over to other responses, producing a general retardation of RT.
Although tactile G/NG RT was protracted relative to visual G/NG RT, the error rates obtained, even in the tactile version of this task, did not approach the rates widely reported for saccadic eye movements and for word-naming tasks. This point is most vividly illustrated by the (anticipatory) error rate found in normal participants' execution of delayed saccades. In this task, participants are required to suppress the natural tendency to generate a quasi-reflexive saccade towards a novel target. Rather, they are to await a prearranged Go signal which follows the target after a brief delay (when spatial memory is at issue, after the target has been extinguished). What is remarkable about this task, which often yields mean error rates of 10-15% in normal participants (e.g. Crawford, Henderson & Kennard, 1996), is that - translated into the terms of the G/NG RT task - it consists entirely of No-Go trials. The experimenters are reasonably confident of their ability to avoid false positive errors entirely in a finger-raising version of such a task.
In summary, tactile effector-cuing exhibits some, but not all, of the criteria one might propose for automacity. Participants reach asymptotic latencies after relatively few trials [ILLUSTRATION FOR FIGURE 4 OMITTED]. By that time they have usually also begun to show insensitivity of CRT to size of the choice set (e.g. Leonard, 1959). Tactile CRT, moreover, scarcely competes for attention with a continuous secondary task like shadowing or reading (Frith & Done, 1986; Goodrich et al., 1989; Goodrich et al., 1990). On the other hand, the tactile G/NG RT task does not yield error rates approaching those of lexical transcoding tasks or reflexive saccades. Furthermore, withdrawal of a finger from contact with the stimulation device, while probably the most prepotent response, still falls short of the spontaneous attraction of foveation by a novel visual target. People engage spontaneously in saccades, just as they assimilate the contents of printed displays without invitation or conscious deliberation. In contrast, experimental participants usually require instruction in the required action before finger-raising responses can be reliably evinced by tactile stimulation. This serves as a reminder that the reflexive saccade is part of a complete and relatively encapsulated sub-system that serves the needs of the duplex retina without requiring or even permitting much deliberate intervention. The withdrawal of a finger from a tactile stimulus, in contrast, while involving a relatively direct stimulus-response linkage, is but a component of a much larger, multimodal system that directs and coordinates the hands in the service of voluntary and purposive action.
The experimenters are greatly indebted to C. D. Frith for the loan of his tactile stimulation apparatus and for many stimulating discussions, and to M. Hawken and J. Everitt for engineering and software support. Helpful comments on early drafts of this manuscript were received from S. E. Henderson and S. Monsell; J. Simpson, A. Ward and A. Kulkarni assisted with the collection and analysis of data.
Alain, C., Taktak, K. & Buckolz, E. (1993). Same-hand and different-hand finger pairings in two-choice reaction time: Presence or absence of response competition. Journal of Motor Behavior, 25, 45-51.
Bashore, T.R. (1981). Vocal and manual RT estimates of inter-hemispheric transmission time. Psychological Bulletin, 89, 352-368.
Berlucchi, G., Crea, F., DiStefano, M. & Tassinari, G. (1977). Influence of spatial stimulus-response compatibility on reaction time of ipsilateral and contralateral hand to lateralized light stimuli. Journal of Experimental Psychology: Human Perception and Performance, 3, 505-517.
Bloxham, C. A., Mindel, T. A. & Frith, C. D. (1984). Initiation and execution of predictable and unpredictable movements in Parkinson's Disease. Brain, 107, 371-384.
Botzel, K., Mayer, M., Oertel, W. H. & Paulus, W. (1995). Frontal and parietal premovement slow brain potentials in Parkinson's disease and aging. Movement Disorders, 10, 85-91.
Bradshaw, J. L., Howard, M. J., Pierson, J. M., Phillips, J. & Bradshaw, J. A. (1992). Effects of expectancy and attention in vibrotactile choice reaction time tasks. Quarterly Journal of Experimental Psychology, 44A, 509-528.
Broadbent, D. E. & Gregory, M. (1962). Donder's B- and C-reactions and S-R compatibility. Journal of Experimental Psychology, 63, 575-578.
Callan, J., Klisz, D. & Parsons, O. S. (1974). Strength of auditory stimulus-response compatibility as a function of task complexity. Journal of Experimental Psychology, 102, 1039-1045.
Clark, J. E. (1982). Developmental differences in response processing. Journal of Motor Behaviour, 14, 247-254.
Crawford, T. J., Henderson, L. & Kennard, C. (1996). Neuropsychology of eye movements abnormalities in schizophrenia. In C. Pantelis, H. E. Nelson & T. R. E. Barnes (Eds), The neuropsychology of schizophrenia, pp. 259-278. Chichester: Wiley.
De Jong, R. (1995). Strategical determinants of compatibility effects with task uncertainty. Acta Psychological, 88, 187-207.
Donders, F. C. (1868). On the speed of mental processes. Acta Psychologica, 30, 412-431.
Duncan, J. (1977). Response selection errors in spatial choice reaction tasks. Quarterly Journal of Psychology, 29, 415-423.
Duncan, J. (1978). Response selection in spatial choice reaction. Further evidence against associative models. Quarterly Journal of Psychology, 30, 429-440.
Fitts, P. M. & Seeger, C. M. (1953). S-R compatibility: Spatial characteristics of stimulus and response codes. Journal of Experimental Psychology, 46, 199-210.
Frith, C. D. & Done, D. J. (1986). Routes to action in reaction time tasks. Psychological Research, 48, 169-177.
Goodrich, S., Henderson, L., Allchin, N. & Jeevaratnam, A. (1990). On the peculiarity of simple reaction time. Quarterly Journal of Experimental Psychology, 42A, 763-773.
Goodrich, S., Henderson, L. & Kennard, C. (1989). On the existence of an attention demanding process peculiar to simple reaction time: Converging evidence from Parkinson's Disease. Cognitive Neuropsychology, 6, 309-331.
Hackley, S. A., Schaffer, R. & Miller, J. (1990). Preparation for Donders' type b and c reaction tasks. Acta Psychological, 74, 15-33.
Hasbroucq, T. & Guiard, Y. (1991). Stimulus-response compatibility and the Simon effect: Toward a conceptual classification. Journal of Experimental Psychology: Human Perception and Performance, 5, 246-266.
Hasbroucq, T., Guiard, Y. & Ottomani, L. (1990). Principles of response determination: The list-rule model of S-R compatibility. Bulletin of the Psychonomic Society, 28, 327-330.
Hasbroucq, T., Kornblum, S. & Osman, A. (1988). A new look at reaction time estimates of interhemispheric transmission time. Cahiers de Psychologie Cognitive/European Bulletin of Cognitive Psychology, 8, 207-221.
Henderson, L. & Dittrich, W. H. (1998). Preparing to react in the absence of uncertainty: I. New perspectives on simple reaction. British Journal of Psychology, 89, 531-554.
Hommel, B. (1995). Stimulus-response compatibility and the Simon effect: Toward an empirical clarification. Journal of Experimental Psychology: Human Perception and Performance, 21, 764-775.
Hommel, B. (1996). S-R compatibility effects without response uncertainty. Quarterly Journal of Experimental Psychology, 49A, 546-571.
Kornblum, S. (1973). Sequential effects in choice reaction time: A tutorial review. In S. Kornblum (Ed.), Attention and performance, vol. 4, pp. 259-288. New York: Academic Press.
Kornblum, S., Hasbroucq, T. & Osman, A. (1990). Dimensional overlap: Cognitive basis for stimulus-response compatibility - A model and taxonomy. Psychological Review, 97, 253-270.
Kornblum, S. & Lee, J.-W. (1995). Stimulus-response compatibility with relevant and irrelevant stimulus dimensions that do and do not overlap with the response. Journal of Experimental Psychology: Human Perception and Performance, 2, 855-875.
Leonard, J. A. (1959). Tactual choice reactions. Quarterly Journal of Experimental Psychology, 11, 76-83.
Marzi, C. A., Bisiacchi, P. & Nicoletti, R. (1991). Is interhemispheric-transfer of visuomotor information asymmetric - evidence from a meta-analysis. Neuropsychologia, 29, 1163-1177.
Posner, M. I. (1978). Chronometric explorations of the mind. NJ: Erlbaum.
Posner, M. I. (1980). Orienting of attention. Quarterly Journal of Psychology, 32, 3-25.
Proctor, R. W., Lu, C.-H., Wang, H. & Dutta, A. (1995). Activation of response codes by relevant and irrelevant stimulus information. Acta Psychologica, 90, 275-286.
Proctor, R. W., Reeve, T. G. & Weeks, D. J. (1990). A triphasic approach to the acquisition of response selection skill. In G. H. Bower (Ed), The psychology of learning and motivation, pp. 207-240. New York: Academic Press.
Proctor, R. W. & Wang, H. (1997). Differentiating types of set-level compatibility. In B. Hommel & W. Prinz (Eds), Theoretical issues in stimulus-response compatibility, pp. 11-37. Amsterdam: Elsevier.
Rabbitt, P. M. A. & Vyas, S. M. (1979). Wundt's 'D Reaction' rehabilitated as a new technique to study perceptual set. Psychological Studies, 24, 119-131.
Reeve, T. G. & Proctor, W. (1988). Determinants of two choice reaction time patterns for same-hand and different-hand finger pairings. Journal of Motor Behavior, 20, 317-340.
Rosenbaum, D. A. (1991). Human motor control. New York: Academic Press.
Shaffer, L. H. (1965). Choice reaction with variable S-R mapping. Journal of Experimental Psychology, 70, 284-288.
Shallice, T. (1988). From neuropsychology to mental structure. Cambridge: Cambridge University Press.
Simon, J. R. (1990). The effects of an irrelevant directional cue on human information processing. In R. W. Proctor & T. G. Reeve (Eds), Stimulus-response compatibility: An integrated perspective, pp. 31-86. Amsterdam: North-Holland.
Stoffels, E.-J. (1996). Inhibition of concurrent processes in letter and orientation discriminations. Acta Psychologica, 91, 153-173.
Stroop, J. R. (1935). Studies of inference in serial verbal reactions. Journal of Experimental Psychology, 18, 643-662.
Van Duren, L. & Sanders, A. F. (1988). On the robustness of the additive factors stage structure in blocked and mixed choice reaction designs. Acta Psychologica, 69, 83-93.
Woodworth, R. S. (1938). Experimental psychology. London: Methuen.
Wundt, W. (1880). Grundriss der physiologischen Psychologie, 2nd ed. Leipzig: Engelmann.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Preparing to React in the Absence of Uncertainty, part 2|
|Author:||Dittrich, Winand H.; Henderson, Leslie|
|Publication:||British Journal of Psychology|
|Date:||Aug 1, 1999|
|Previous Article:||On the relationship between time management and time estimation.|
|Next Article:||Role of memorization conditions in the haptic processing of orientations and the 'oblique effect.'.|