University of Groningen
Not enough time or not enough attention? Speed, error and self-maintained control in the Sustained Attention to Response Test (SART)
Manly, Tom; Davison, Bruce; Heutink, Joost; Galloway, Maria; Robertson, Ian H.
Published in:
Clinical Neuropsychological Assessment : an international journal for research & clinical practice
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Final author's version (accepted by publisher, after peer review)
Publication date:
2000
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Manly, T., Davison, B., Heutink, J., Galloway, M., & Robertson, I. H. (2000). Not enough time or not enough attention? Speed, error and self-maintained control in the Sustained Attention to Response Test (SART).
Clinical Neuropsychological Assessment : an international journal for research & clinical practice, 3, 167- 177.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license.
More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne- amendment.
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Not enough time or not enough attention? Speed, error and self-maintained control in the Sustained Attention to Response Test (SART)
Article · January 2000
CITATIONS
59
READS
2,147 8 authors, including:
Some of the authors of this publication are also working on these related projects:
Visual impairments in patients with Parkinson's diseaseView project
Safe cycling (Veilig op de Fiets)View project Tom Manly
Medical Research Council (UK) 123PUBLICATIONS 6,733CITATIONS
SEE PROFILE
Joost Heutink University of Groningen 73PUBLICATIONS 527CITATIONS
SEE PROFILE
Maria Galloway
MRC Cognition and Brain Sciences Unit 2PUBLICATIONS 531CITATIONS
SEE PROFILE
Ian H Robertson Trinity College Dublin
435PUBLICATIONS 23,111CITATIONS SEE PROFILE
All content following this page was uploaded by Joost Heutink on 22 December 2013.
The user has requested enhancement of the downloaded file.
Not enough time or not enough attention?: Speed, error and self- maintained control in the Sustained Attention to Response Test (SART).
A final version of this manuscript appears in Clinical Neuropsychological Assessment (Manly, T., Davison, B., Heutink, J., Galloway, M., & Robertson, I.
(2000). Not enough time or not enough attention?: Speed, error and self-
maintained control in the Sustained Attention to Response Test (SART). Clinical Neuropsychological Assessment, 3, 167-177.)
Tom Manly1, Bruce Davison12, Joost Heutink1, Maria Galloway1 and Ian H Robertson3
1. MRC Cognition and Brain Sciences Unit, 15 Chaucer Road Cambridge UK 2. Department of Psychology, Aston University, Birmingham, UK
43. Department of Psychology, Trinity College Dublin
Manly, T., Davison, B., Heutink, J., Galloway, M., & Robertson, I. (2000). Not enough time or not enough attention?: Speed, error and self-maintained control in the Sustained Attention to Response Test (SART). Clinical Neuropsychological Assessment, 3, 167- 177.
Address of correspondence
Tom Manly, MRC Cognition and Brain Sciences Unit, 15 Chaucer Road Cambridge UK. tom.manly@mrc-cbu.cam.ac.uk
Abstract.
Traumatic brain injuries are associated with an increased frequency of 'absentminded' slips of action. As such errors tend to occur in routine situations,
assessment in the clinic or laboratory has proved difficult. Recently, however, Robertson and colleagues described a simple and reliable computerised measure, the Sustained Attention to Response Test (SART) that was predictive of such error propensity in both head injured and control groups. In the SART, participants are asked to respond to frequent 'go' stimuli but maintain a readiness to withhold a response to rare and unpredictable no-go trials. Here, data from 109 healthy participants is pooled to allow more detailed analysis of the relationships between speed of ' go' responses and accuracy on 'no-go' trials. The results show that individual differences in response speed are related to error rates, but that variability within individual's reaction times (RT) is also a strong predictor. To clarify whether such variability is itself occasioned by lapsing attention, a 'response locked' version of SART task was developed to reduce both individual differences and within-subject variability in RT. While both aims were successful, no significant change in error rates was observed and the performance on the modified task was strongly related to standard SART performance. The results suggest that a significant component in RT variability is related to the same pattern of lapsing attention that underpins errors and that reducing these differences 'at source' may do little do undermine the sensitivity of the task.
Introduction
An important aim of neuropsychological assessment is to predict problems that patients may experience in everyday life. In 1997, Robertson and colleagues first reported on a computerised measure, the Sustained Attention to Response Test (SART), that proved to be predictive of everyday attentional lapse frequency both in traumatically brain injured and healthy participants (Robertson, Manly, Andrade, Baddeley, & Yiend, 1997).
In the SART, single digits are presented on a computer screen at a steady, invariant rate. Participants are asked to press a single response key as soon as possible after each digits appearance. The exception is a nominated no-go target to which no response should be made. Due to the rhythmic nature of the task and the low probability of the no- go target appearing, the SART was designed to rapidly induce a rather inattentive,
‘absentminded’ response style. In order to successfully withhold the response to the no- go target, it was argued that participants must endogenously resist this tendency and self- maintain a more active, controlled stance to the task. In this manner, the error score (pressing for no-go targets) would reflect the efficiency of this endogenous control.
In previously reported studies, differences in reaction times (RTs) to the frequent
‘go’ stimuli of the SART have not accounted for the group differences in error rates between traumatic brain injured groups and controls, nor between healthy participant groups defined by high or low frequency of everyday attentional slips (Robertson, et al., 1997; Manly, Robertson, Galloway, & Hawkins, 1999). However, speeding within individuals’ reaction times (RTs) as they performed was found to be significantly predictive of subsequent error. This speed-accuracy ‘trade-off’ makes unambiguous interpretation of error rates purely in terms of poorly maintained attention difficult.
The SART instructions are to respond as quickly as possible to ‘go’ trials while making as few errors - pressing on no-go trials - as possible. Accordingly, it is possible that the speed-accuracy characteristics arise from participants strategically ‘hunting’ for the optimal trade-off point, the faster boundaries of their search being marked by errors of commission. In this case, errors could occur despite adequately maintained attention to the task. Alternatively, speeding in responses could result from transient inattention to one’s actions - effectively allowing responses to be ‘driven’ by the task. Without
interceding processing of the relevance of the stimulus for the response, responses could become triggered by the onset of the trial - or given the regular pacing - by the
expectation of trial onset and, hence, become faster. Of course, these possibilities are not mutually exclusive.
While there is electrophysiological evidence to support the argument that errors on the SART predominantly occur in the context of poorly maintained attention1, in any individual case disambiguation of the possibilities outlined above is difficult. If RT differences were clearly a confound in the assessment of ‘attention’, then they could be controlled for using normative data to weight both speed and accuracy components. If, however, the RT characteristics were predominantly a consequence of poor attention to action, such statistical control would risk throwing out the baby with the bath water and undermine the ecological sensitivity of the task.
1 In a recent study (Manly et al, 2000) we found that a reduced amplitude of an ERP component previously related to attention, the P300, was indeed predictive of subsequent errors on the task and that this effect was independent of reaction time.
In the first analysis reported here, data from 109 healthy participants who had performed the SART for the first time were pooled to allow closer examination of the reaction time and error characteristics. This analysis also allows the effects of age and gender to be considered over a relatively large sample. In the subsequently presented novel experimental study, the effect of an intervention designed to stabilise response speed within the SART was investigated.
Study 1:
Method
Participants
Data on SART performance from 109 neurologically healthy adults between the ages of 18 and 68 (mean = 36.05, SD 11.87) were included in the analysis. Eighty of the participants had contributed to previously reported studies (Robertson, et al., 1997;
Manly, et al., 1999), 29 completed the SART as part of ongoing projects.
Results were included in the analysis if:
1. The participant was neurologically healthy.
2. The results were from their first exposure to the SART
3. They performed the ‘standard’ version of the task comprising 225 trials containing 25 no-go targets, preceded by 18 practice trials containing 2 targets (see below).
4. The trials were continuous (i.e. not interspersed with other experimental conditions, nor completed after other experimental conditions using similar tasks).
5. The instructions were to “press for each digit as quickly as possible with the exception of the digit 3. Try and press as quickly as possible while making as few errors (pressing for a 3) as possible.”
6. The test was conducted in a quiet office under conventional testing conditions.
These criterion yielded approximately equal numbers of people within the age- bands 18-29 years, 30-39 years and 40-49 years (36, 30 and 31 respectively). Adults over the age of 50 were more poorly represented (n = 11). Women predominated within the group (female n = 75, male n = 34).
The Sustained Attention to Response Task (SART)
In each trial of the SART, a single digit (1-9) was presented in the centre of a laptop computer screen for 250 msec followed by a 900 ms mask (a circle with a diagonal cross). The digits appeared in 48 point, 72 point, 94 point, 100 point or 120 point (selected randomly by the program). Each digit appeared with equal frequency within a random sequence.
Following 18 practice trials including 2 no-go targets, 225 continuous test trials were presented including 25 no-go targets. Participants were instructed to press the mouse key with the index finger of their preferred hand as quickly as possible for each number that they saw, with the exception of the digit 3. The probability of a no-go target appearing on any one trial was 1/8 (0.11). The instructions to take into account both speed and accuracy was emphasised.
Results.
Errors of commission (pressing on a no-go trial).
The 109 participants made a mean of 6.36 errors of commission (SD = 4.36; range
= 0-19). There was no significant difference between the error propensity of male of female participants (mean errors of commission made by women = 5.89 (SD = 4.07), mean errors of commission made by men = 7.38 (SD = 4.86); F(1,107) = 2.771, p = 0.1)
Reaction times to go trials.
The participants responded to go targets at a mean of 375 ms following the presentation of the digit (SD 65). There was no significant RT difference between men and women (mean reaction times for women = 375 ms (SD 68.6), mean reaction times for men = 376 ms (SD = 58.4); F(1,107) = 0.005, P = 0.941)
The relationship between reaction time and errors: Between subject analysis.
The number of errors of commission (pressing for a no-go target) made by participants were significantly related to individual go trial reaction times, averaged across all of the go trials in the task (Pearson r = -0.49, P <0.001; see also Figure 1 below). Individuals who responded more quickly were more likely to make more errors.
Figure 1: Scatterplot of errors of commission against reaction time to non- targets for 109 neurologically healthy subjects performing the SART.
Reaction times Errors of commission
0 2 4 6 8 10 12 14 16 18 20
250 350 450 550 650
Within-subject analysis.
In order to examine the effect of changes in reaction times during performance, RTs for the following categories of trials were first averaged for each participant and then entered into the described repeated measures ANOVAs.
1. Go trials with responses that occurred immediately before a correct no-go trial (where a response was successfully withheld).
2. Go trials with responses that occurred immediately before an error of commission (where a response was not successfully withheld).
3. Go trials with responses that occurred immediately after a correct no-go trial.
4. Go trials with responses that occurred immediately after an error of commission on a no-go trial.
The mean number of observations per participant in each of the categories were as follows; 1 = 17.6 (SD 4.34); 2 = 6.0 (SD 4.03); 3 = 17.30 (SD 4.32); 4 = 6.07 (SD 3.95).
Participants who made no errors of commission clearly do not contribute to the error trial analysis. The original data files for 4 of the participants were no longer available.
Prior to a correct no-go trial, the mean RT was 384 ms (SD 68.4). Prior to an error of commission, responses were significantly faster at 333 ms (SD 63.7; F (1, 99) = 37.75, P < 0.001).
Following an error of commission, RTs showed a significant increase relative to the pre-error go trial (364. msec; SD 95.17; F(1,99) = 8.8, P < 0.01).
After a correct no go trial, RTs showed a speeding effect (RT on go trial prior to a correct no-go trial = 384 msec (SD 68.37), RT on go trial after a correct no-go trial = 351 msec (SD 60.61); F (1, 104) = 55.3, P < 0.001).
In summary, speeding in responses was predictive of subsequent errors of
commission for individuals. Following errors of commission, participants had a tendency to slow down. Following a correctly withheld response to a no-go trial, participants had a tendency to speed up. The mean RT values for the four different types of trial are
illustrated in figure 2, below.
Figure 2: Speed-accuracy characteristics in the SART. The upper panel shows the trial type (highlighted) contributing to the RT data relative to the 3 no-go trial. Relatively
fast responses are predictive of subsequent error. Following an error, participants tend to slow their responses. Following a correct no-go trial, a modest speeding in responses
occurs. Units are milliseconds.
p r e - er r o r post erro r pr e -c or r ec t post - cor rect
2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0 5 0 0
Variability in response speed.
The standard deviation associated with each individual’s mean reaction time reflects the variability in their response speed. In this sample, the Pearson's correlation between participants' standard deviations in reaction times and errors of commission was 0.24 (P < 0.05) - the more variable their response times, the more errors they were likely to make. However, the size of a standard deviation is generally related to the absolute magnitude of reaction times (in this group the correlation is r = 0.44, P<0.001). In order to examine whether variability was related to error propensity independently of the previously reported reaction time relationship, a partial correlation was performed between RT standard deviations and errors of commission, with participants' mean reaction time partialled out. There was a significant positive relationship (r = 0.59, P<0.001). The increased size of the relationship compared with the standard correlation suggests that the variability and mean reaction time effects in fact work in rather opposite directions. A further method for examining this question is to use coefficients of
variation, in which participants' standard deviations are first divided by their mean reaction time (Howell, 1997). The Pearson correlation between individual coefficients of variation and errors of commission was 0.58, P < 0.001).
Errors of omission
Errors of omission are relatively rare within SART performance, being made on a mean of only 1.06 trials (0.5%; SD = 3.41) in this group. The number of such omission errors are not related to the number of errors of commission made (r = 0.15, P = 0.13).
The effect of age on performance
The correlation between age and errors of commission did not reach statistical significance, although there was a modest trend for performance to be slightly improved
in older participants (r = -0.18. P = 0.062). There is an extensive literature on slowing of response speed across many tasks as people age (see Cerella & Hale, 1994; Salthouse, 1996 for reviews). The SART is, however, an unusual reaction time task in that the onsets of stimuli can be anticipated. Despite this, reaction times modestly, although significantly, increased with age (r = 0.36, P < 0.001). If this RT slowing is controlled for in a partial correlation of age with errors of commission (mean RT partialled out), the trend towards accuracy gains with age is abolished (r = -0.0002, P = 0.998). However, partialling age out of the relationship between mean reaction times and errors, does little to weaken this relationship (r = -0.46, P < 0.001).
Discussion
There is a significant relationship between participants mean speed of response to frequently presented go stimuli and a propensity to make errors on no-go trials in the SART. There is also a significant relationship between the variability within individual participants’ response times and error rates. These two observations are independent and appear to operate in contrary directions. Short reaction times are associated with
increased errors. Long reaction times are associated with fewer errors and increased standard deviations. Despite this, increased standard deviations are associated with higher error rates - indeed the relationship is stronger than that for mean reaction time.
‘Speed-accuracy trade-off’ does not form an account of these relationships, it simply describes their direction. The question - as posed in the introduction - as to whether this variability is itself a useful measure of fluctuating attention to action or whether it is an unhelpful confound remains. An alternative to statistically controlling for speed differences between participants, and one that is potentially more informative as to
underlying processes, is to prevent those differences emerging in the first place. This is explored in the following experiment.
In Experiment 1 the SART was manipulated to include a brief auditory tone at a fixed point within each SART trial. By asking participants to synchronise their responses to the tone on go trials we attempted to achieve the following; a) to reduce individual differences in reaction times; b) to reduce within-subject variability in reaction times; c) to maintain reaction times at a point within the trial that, under conventional conditions, is associated with correctly withheld responses on no-go trials.
By comparing performance on this modification with the standard SART rather different predictions could be made depending upon the predominant source of the speed- accuracy relationships described above.
If the speed-accuracy relationships in the standard task is predominantly caused by participants strategically (and attentively) trying to balance the instructions for speed and accuracy, then preventing this adaptive ‘roving’ should lead to both fewer errors overall and less differentiation between participants. In effect, one aspect of the participant’s task is taken care of and they can simply concentrate on withholding responses at the
appropriate moment - which given the latency of the tone and the general success at that onset-response delay - should be relatively easy.
The alternative argument is as follows. In the standard SART, waning attentional control results in both increased errors of commission and reduced response speed, as the onset of the trial is allowed to act effectively as an exogenous response trigger. In the modified version, although the response trigger is transferred to a later point in the trial, poor attentional control over action should still result in errors of commission. The same pattern of speeding and slowing that accompanies such inattention in the standard SART should not, however, be apparent.
Experiment 1
Method
Participants.
Thirty neurologically healthy participants, recruited from the MRC Cognition and Brain Sciences Unit subject panel, volunteered to take part in this study. None of the participants had contributed to the previous analysis. The group comprised 18 women and 12 men and were of mean age 46.5 years (SD 18.8)
Measures
The Sustained Attention to Response Test
The task was essentially as described above with the exception that 270 test trials (including 30 targets - rather than 225 trials with 25 targets) were administered. Trial timing and no-go target probability were identical to the previously described measure.
‘Response-locked’ SART.
This measure was identical to the version of the SART described above with the exception that a brief tone (50 msec, 587.3 Hz sine tone) was presented at a fixed interval after trial onset within each trial. The aim was that, if subjects responded in time with the tone, the delay between digit/trial onset and that response would be at a level that, on average, is associated with accurate performance in the standard SART. In piloting we found that initiating the tone presentation at 100 msec after the trial onset produced reaction times within the appropriate range (approximately 380 msec after trial onset).
Two hundred and seventy test trials were run including 30 no-go trials. The tone was presented on both go and no-go trials.
Procedure
The instructions for the ‘standard’ condition were, as previously described, to
“Press for each number you see with the exception of the digit 3. If you see a 3 don’t press, simply wait for the next number. Try and press as quickly as possible while making as few errors as possible.”
The instructions for the response locked condition were as follows.
“Press for each number you see with the exception of the digit 3. If you see a 3, don’t press, simply wait for the next number. You will also hear a tone presented with each number. Try and make your response in time with the bleep, rather than before or after it. Please try to keep up with the rhythm whilst at the same time trying to make as few errors (pressing for 3) as possible.”
Condition order was balanced across participants.
Results
Stability of reaction times - within subjects.
All correct reaction times to non-targets were considered from the standard SART condition and the response-locked condition. To consider whether the response locking had reduced the amount of variance seen in individuals’ reaction times the standard deviations associated with each participant’s mean were entered into a repeated measures ANOVA, with condition as the factor. The results confirmed the hypothesis. Standard deviation of reaction times in the response locked condition were significantly reduced relative to those in the standard SART condition (F(1,29) = 8.80, P <0.01; Standard SART mean standard deviation = 80 msec (SD 35); Response Locked SART mean standard deviation = 62 msec (SD 18)). The fact that mean reaction times did not
significantly differ between the conditions (F(1,29) = 1.92, P = 0.18) suggests that this reduction in variability was not simply mediated by faster reaction times overall.
Stability of reaction times - between subjects.
The aim of the response-locked condition was not simply to reduce variability within each participants' responses but also to reduce variability between participants.
This was tested using a procedure devised by Pitman (1939, in Howell, 1997) to test for the heterogeneity of variance of non-independent samples. Taking the variance in mean reaction times for the two conditions, this analysis revealed that this aim was successful (t = 4.95, P < 0.001). Participants were much less likely to differ from one another in their mean time of responses to go trials under the response locked condition.
The relationship between reaction times and errors.
The auditory response trigger for the response locked condition was set in piloting in order to achieve approximately equal mean reaction times with those that, on average, preceded correct withholding of responses to targets in the standard SART. In this group, correct no-go trials in the standard SART condition were preceded by go responses at a mean of 346 msec (SD 78). The mean response times in the response locked condition were 344 msec (SD 35). The manipulation was therefore successful in retarding responses to a post-trial onset delay that is normally associated with successfully withheld responses on no-go trials.
If the response locking had been successful, it would be expected that the
relationship between response speed and error rates that is apparent in the standard SART would be abolished in this modification. This was the case. The correlation between reaction times and errors of commission in the standard SART was r = -0.42 (P<0.05), that is broadly equivalent and in the same direction as that observed in the large group
analysis described above. For the response locked condition the correlation failed to reach statistical significance r= -0.19 (P= 0.33).
The aims of the manipulation were therefore broadly met. Providing a response cue within each trial significantly reduced the variability of within each participant's
responses, reduced the variability in response times between participants, ‘held’
responses at a level where errors would be less likely within the standard SART, and eroded the relationship between errors and reaction times. These conditions allow the differential predictions to be examined.
Error rates under the two conditions.
It was hypothesised that if lapsing attentional control was the strongest predictor of errors on the task, then errors under the two conditions may be broadly equivalent. If, alternatively, speed characteristics were the most important, then errors should be reduced within the response locked condition together with the sensitivity of the task to individual differences (in that these are predominantly the result of speed differences).
Participants made a mean of 8.30 (SD 5.11) errors on the 30 targets of the standard SART condition. Although slightly fewer errors were made in the response locked condition (6.77 (SD 4.80), the difference did not reach statistical significance (F(1,29) = 3.16, P = 0.09). More importantly the propensity of certain subjects to make more or less errors was significantly related in the two tasks (Pearson correlation of errors of
commission in the standard SART and in the response locked condition, r = 0.55, P <
0.01).
Discussion
The results of this study show that;
1. Asking subjects to respond in time to an auditory cue rather than as quickly as possible to the onset of the digit was successful in significantly reducing response speed variability - both within and between participants, ‘holding’ responses at an interval associated with accurate performance in the standard task, and breaking the relationship between response times and errors.
2. Despite the successful manipulation of response speed, error rates did not significantly change, and individual error propensity in one task was predictive of error propensity in the other.
While the two hypotheses outlined are neither mutually exclusive nor the only possible interpretations of task performance, the results are consistent with ‘attentional’
control over action being an important determinant of accuracy in the SART task. If speed factors, however determined, were a sufficient account for errors, then it would be expected that error rates would have declined and that the correlation in performance between the two tasks would be small. The observation that the same subjects show similar error propensities, despite the constraints on speed in the experimental condition, suggests that it is some other aspect of their ‘approach’ which determines error rates.
The results further suggest that the relationship between errors and variability in response speed seen in the standard SART might plausibly arise via a third ‘hidden’
factor - such as might be termed attention to task. In the standard version of the task, lapsing attentional control would lead to higher error rates and speed variability (controlled at some points, task driven at others). In the response locked condition, the task cue serves to reduce variability but this does not, of itself, reduce error rates nor change individual propensity to error.
Influential views of the response inhibition process have framed it in terms of a
‘race’ between independent ‘go’ and ‘stop’ functions (Logan, Schachar, & Tannock, 1997). In the Stop-Signal paradigm for example, participants are presented with a go signal that may or may not be followed by a subsequent stop signal. The probability of successfully inhibiting a response, if a stop signal is presented, can be manipulated by increasing or decreasing the delay between the two. The longer the delay after the go signal, the less probable it is that individuals will be able to withhold their response. By examination of the reaction times to ‘unstopped’ go signals, the speed of the inhibition function can thus be determined (Logan, et al., 1997).
The ‘response locked’ SART condition described above is unusual in this respect as it can be seen as a ‘stop-signal’ paradigm in which the stop-signal (the presentation of a 3) actually precedes the pre-potent go signal (the tone in the trial). Even under these conditions, where the stop processes should be given a considerable head-start in the race, little difference in the capacity of participants to withhold responses is observed. In line with the original aims of the SART, it is suggested that the failure to maintain control over this external ‘driving’ of responses is a principle determinant of the error characteristics of the task. In other words, that the dynamics of the go -stop race are determined to an extent by pre-existing attentional control.
While the results presented here, taken together with the results of a study on ERP characteristics of the SART reported elsewhere ((Manly, Datta, Heutink, Hawkins, Cusack, Rorden, et al., 2000), suggest that the standard SART predominantly acts as a measure of self-maintained attentional control, the influence of speed determined by other factors cannot be ruled out within an individual case. If, for example, a patient gives undue emphasis to the speed aspect of the instructions, they may make a high number of errors despite reasonable attention to what they are doing. The response
locked version of the task may therefore offer a useful and more clearly interpretable alternative to the standard task.
References
Cerella, J., & Hale, S. (1994). The rise and fall of information-processing rates over the life span. Acta Psychologica, 86, 109-197.
Howell, D.C. (1997). Statistical Methods for Psychology. (4th ed.) Belmont, CA:
Duxbury Press.
Logan, G. D., Schachar, R. J., & Tannock, R. (1997). Impulsivity And Inhibitory Control. Psychological Science, 8(1), 60-64.
Manly, T., Datta, A., Heutink, J., Hawkins, K., Cusack, R., Rorden, C., &
Robertson, I. H. (2000). An electrophysiological predictor of imminent action error in humans. Journal of Cognitive Neuroscience, ss.xx
Manly, T., Robertson, I. H., Galloway, M., & Hawkins, K. (1999). The absent mind: Further investigations of sustained attention to response. Neuropsychologia, 37, 661-670.
Pitman, E.J.G. (1932) A note on normal correlation. Biometrika, 21, 9-12 Robertson, I. H., Manly, T., Andrade, J., Baddeley, B. T., & Yiend, J. (1997).
'Oops!': Performance correlates of everyday attentional failures in traumatic brain injured and normal subjects. Neuropsychologia, 35(6), 747-758.
Salthouse, T. A. (1996). The processing-speed theory of adult age differences in cognition. Psychological Review, 103,, 403-428.
Acknowledgements
The authors are grateful to Ian Nimmo-Smith for statistical advice, to Julia Darling for her careful preparation of the manuscript, and to two anonymous reviewers for their thoughtful comments and suggestions. This research was generously supported by the UK Medical Research Council.
View publication stats View publication stats
