Dose-related effect of sevoflurane sedation on higher control of eye movements and decision making

S. A. R. Nouraei1, N. de Pennington1, J. G. Jones1 and R. H. S. Carpenter*,2

1 University Department of Anaesthesia, Level 4, Addenbrooke’s Hospital, Cambridge UK. 2 Physiological Laboratory, University of Cambridge, Cambridge CB2 3EG, UK

Corresponding author. E-mail: rhsc1@cam.ac.uk

Accepted for publication: March 29, 2003


    Abstract
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendix: a brief summary...
 References
 
Background. Saccadic latency may provide an objective method to assess sedative doses of anaesthetic on cortical oculomotor mechanisms and decision making.

Methods. We tested the effects of random doses of 0, 0.1, 0.2 and 0.3 MAC sevoflurane in six subjects, in a double-blind study using two measures of behavioural impairment: saccadic latency and stop signal reaction time (SSRT) in a countermanding task.

Results. Saccadic latency and SSRT both increased with increasing doses of sevoflurane. In both measures, reciprocal reaction time was linearly related to dose in each subject: all but two of the twelve regression coefficients were statistically significant (P<0.05). In one subject, SSRT was significantly more sensitive than simple latency (P<0.05); for the others there was no significant difference.

Conclusion. Measurements of this kind could potentially provide estimates of cortical effects of sevoflurane sedation, and give a clinically useful measure of cognitive fitness.

Br J Anaesth 2003; 91: 175–83

Keywords: anaesthetics volatile, sevoflurane; complications, saccadic latency; recovery; sedation


    Introduction
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendix: a brief summary...
 References
 
To assess patient recovery from anaesthesia we need to assess the effects of anaesthetic agents on neural function. At present the decision to discharge patients is largely clinical, but the widespread use of ambulatory surgery calls for an objective assessment of postoperative anaesthetic recovery, including the patient’s capacity to make decisions. Saccadic eye movements may provide such a measure. Saccades – the rapid conjugate eye movements that shift the gaze to a target of interest – can be measured easily, accurately and non-invasively in a manner that is unusual among neurological tests14 and can be detected readily by automatic recording systems, to obtain large data sets quickly.5 6

The neural mechanisms controlling saccades extend from the brainstem, which determines the amplitude and velocity of movements, to the cortical frontal eye fields, which are involved in the decisions to initiate movements and timing of saccades (latency), and in decisions to withhold or change a previously planned movement (countermanding).

In principle, saccadic eye movements could be used to indicate anaesthetic effects at two levels: i) the brainstem, containing the neural circuits that control saccadic velocity and amplitude; this can be measured by peak saccadic velocity (PSV); ii) the cortex, including frontal cortical areas implicated in decision making and saccadic initiation;7 8 this can be measured by either saccadic latency or the stop signal reaction time (SSRT).9 10 The latter uses a countermanding task to assess the even higher cortical ability to change or stop an already planned action.

PSV has been used widely to measure brainstem depression by anaesthetics.1113 In the 0.05–0.1 MAC range, there is a dose-related effect, allowing a sensitive index of sedation.1417 Latency and similar measures of cortical decision-making function do not indicate the concentration of anaesthetic agent per se, but the actual effect on cognitive fitness, which in the end is what matters. For example, a patient who ignored medical advice and drove a car home after a general anaesthetic might need to apply the brakes suddenly – a decision – in response to a child stepping onto the road. In contrast to saccadic velocity, the effects of anaesthetics on saccadic latency and SSRT are not well studied. Both latency and SSRT18 were increased by isoflurane 0.15% (0.1 MAC) in one study; in another, saccadic velocity and latency both changed 1 h after midazolam.19 A third study suggested that latency is more sensitive to sedation than saccadic velocity: 4 h after diazepam or thiopental, only latency was increased, PSV having returned to the control value.20 However, the latter two studies used about 30 trials to measure latency,19 20 an insufficient sample given the intrinsic variability of oculomotor latency.

We measured the dose response of saccadic latency and SSRT to sub-anaesthetic sevoflurane in order to assess the effects on cortical decision making, using enough trials (2000 for each subject) to carry out distributional and other kinds of quantitative analysis.


    Methods
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendix: a brief summary...
 References
 
Subjects
With the approval of the Addenbrooke’s Hospital Local Ethics Committee and the informed consent of the participants, six subjects (five males and one female) aged 20–62 yr and with no known history of neurological or oculomotor diseases took part in the study.

Oculometry
We measured eye movements using a head-mounted scleral reflection infrared oculometer2 in the form of goggles having a range of ±30° and symmetrically linear to 1% over about ±10°. Output was sampled at 10-ms intervals with a PC-based saccadic analysis system (Saccadic Programming and Instrumentation Computer)6 which displayed and stored eye-movement data, controlled the stimuli and detected saccades in real time with an algorithm using acceleration, speed and position. The computer ignored saccadic latencies of less than 50 ms. The traces were inspected after each session and records with errors caused by blinks, head movements or other artefacts (typically less than 2% of the total) were eliminated from further analysis.

Stimuli
The stimuli were three red spots projected in the horizontal plane at different times onto a screen approximately 2 m from the subject by low-voltage laser diodes mounted on the goggles, which also housed the oculometer. Because the target positions were thus essentially fixed relative to the head, for moderate head movements the subject’s head did not have to be immobilized, evidently a clinical advantage. The spots subtended a diameter of 26 min arc and were positioned horizontally 12° apart with a luminance of 13.2 cd m–2.

Saccadic latency and the countermanding saccadic task
An experimental run consisted of a sequence of consecutive target presentations or trials. Most trials were control trials, and began with presentation of the central target, which the subject fixated. After a random interval of 0.5–1.5 s (to discourage prediction), the target then jumped 12° unpredictably to the left or right, and the subject tracked it with a saccade. Detection by the computer of this saccade ended the trial, and the next trial started after 100 ms. However, in a certain proportion of trials (stop trials), the central fixation spot reappeared 50 ms (the stop-signal delay) after the presentation of the lateral target, indicating to the subject not to make a saccade (Fig. 1). Control and stop trials were randomly interleaved in the ratio 4:1. Two hundred trials were presented for each dose of sevoflurane. In control trials the subject always made a saccade to fixate the peripheral target, whereas in stop trials the movement was sometimes inhibited successfully, sometimes not. The proportion, P, of stop trials on which a saccade was made, despite the countermanding signal, was recorded. Subjects became accustomed to the protocol quite quickly, with latencies becoming stable after some 30 trials; results from these learning trials were discarded.



View larger version (17K):
[in this window]
[in a new window]
 
Fig 1 The countermanding paradigm. In a control trial (top), at the start of the trial a fixation target (black square) is presented. The dotted circle represents the position of the subject’s gaze. After a random delay, the target jumps unpredictably to the left or right, and the subject makes a saccade to track it after a delay, the saccadic latency. In a stop trial (below), fixation is followed by a target jump as in a control trial, but the fixation light reappears after a fixed period (the stop-signal delay), signalling to the subject that no saccade should be made. On some trials subjects successfully inhibit the saccade, on others they do not.

 
For each distribution of control trial latencies, we used a Kolmogorov–Smirnov measure21 to determine the best-fit values of the parameters of the Linear Approach to Threshold with Ergodic Rate (LATER) model (see Appendix).8 22 Reaction times generally obey a recinormal distribution: that is, the reciprocal of latency is normally distributed. Provided this is so, then given the best-fit estimates of the parameters of the distribution, and knowing the proportion P of saccades that fail to be inhibited during stop-signal trials, one may estimate the SSRT.21 23 The procedure is based on the following deduction: if saccades are generated in a proportion P of stop-signal trials, then P must also equal the proportion of the distribution of saccadic latencies which has already occurred by the average time, T, of completion of the countermanding process. Thus to estimate T we first measure P from the data, then find the latency corresponding to this value from the best-fit recinormal line on the reciprobit plot. The SSRT is then given by the difference between T and the stop-signal delay, which in this study was always 50 ms:

SSRT = T – 50.(1)

The SSRTs calculated in this way are in effect mean values whose SEs can be obtained from an estimate of the SE of the corresponding value of P, which in turn can be calculated assuming a binomial distribution of uninhibited saccades. From this the corresponding fiducial limits for the SSRT can be derived using the cumulative distributions of latencies, providing a value for the SE at time T. This value also represents the SE of the SSRT since it is related to T by a constant difference (see Equation 1). The statistical significance of the difference between the two median latencies or mean SSRT values obtained for a subject in a particular pair of conditions was assessed using a standard comparison of means test to compute a t statistic.24

Anaesthetic system
The subjects were seated during the experiments. They breathed either pure oxygen or a mixture of sevoflurane and oxygen, delivered by a standard Boyle’s anaesthetic machine through a Bain system attached to a physiological mouthpiece. Each subject wore a nose clip with end-tidal carbon dioxide (FE'CO2) and sevoflurane concentration sampled at the mouthpiece using a Datex Capnomac Ultima (Datex Instrumentarium, Helsinki, Finland). FE'CO2 was 4.5–5 kPa. The machine was calibrated using the manufacturer’s known gas samples, as described previously.25 Pilot studies using the same study plan as the main study showed that breathing pure oxygen rather than air did not have a confounding effect. For each subject, end-tidal concentrations of sevoflurane which corresponded to 0, 0.1, 0.2 and 0.3 MAC when corrected for the age of the subject were used.26 When doses were changed the end-tidal sevoflurane was equilibrated without over-pressure, using an equilibration period of 10 min.

Experimental protocol
Each subject completed a total of 10 runs each of 200 trials in two study days. On each of the 10 runs each subject breathed one of the four doses of sevoflurane, so that each dose was breathed on two occasions, with two further zero-dose runs to make up the ten runs. The order of administration of doses was essentially random, but the first dose was always 0 MAC to establish a baseline. Subsequent experimental doses were always the same as or greater than the previous one, to avoid the effect of residual anaesthetic levels on subsequent blocks (data from the two extra 0-MAC trials were always discarded). For example, on the first day a subject might breathe 0 then 0.1 then 0.2 then 0.3 then 0.3 MAC again, and on the next day 0 then 0 then 0.1 then 0.2 then 0 MAC, the data from the final trial being discarded.

Both the subject and the experimenter were masked to the anaesthetic dose. Sevoflurane was chosen because in low doses, and when the subject is wearing a nose clip, it cannot be smelled or tasted. The same 10-min period of equilibration was used even when no anaesthetic was actually being administered, so that the subjects could not guess whether they were receiving the anaesthetic. They showed no signs of either sedation or excitement.

To ensure that the experimenters did not know the dose being given, the instruction cards for the anaesthetist were prepared before the experiments and were sealed in envelopes. Each run of 200 trials, corresponding to one distinct dose, was stored in one computer file with a randomised name; only after the analysis was complete were the doses of anaesthetic matched to each record.

Statistical analysis
Cumulative latency distributions should conform to a straight line on a reciprobit plot8 (Fig. 2). Comparisons of subject’s distributions with the expected recinormals were made using the Kolmogorov–Smirnov statistic.27 Median saccadic latency and SE were calculated for each subject and condition, as were SSRT and its SE using the method explained above, using the percentage of saccades that were not inhibited in the stop trials. We used Origin (Microcal Software Inc., Northampton, MA, USA) to correlate the dose dependence of saccadic latency and SSRT for each subject.



View larger version (15K):
[in this window]
[in a new window]
 
Fig 2 Typical reciprobit plot of saccadic latency data from one subject (A) for control trials at four different sevoflurane doses. Cumulative histograms (i.e. plots of the proportion of trials on which a saccade has occurred by a particular time) are plotted on a probit scale as a function of reciprocal latency (using a reciprocal scale labelled for ease of interpretation with latencies rather than their reciprocals: longer latencies lie to the right). If the LATER model is correct, the data should then lie on a straight line. The lines are best-fit straight lines, subject to the constraint of parallelism. Note that the use of a probit scale tends to throw the tails of the distribution into prominence: the deviations at longer latencies in some cases are not statistically significant.

 

    Results
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendix: a brief summary...
 References
 
Saccadic latency
Figure 3 shows plots of latency in the control trials for each of the subjects and for the different doses, together with their associated regression lines. In these plots, reciprocal reaction times are plotted, since the rate of increase of the underlying LATER decision signal, rather than reaction time itself, seems to be the more fundamental variable, suggested by the fact that the former is generally normally distributed whereas the latter is not. No subject’s latency distribution differed significantly from a recinormal (Kolmogorov–Smirnov P=0.05). The normal distribution also makes statistical analysis more straightforward. The median latencies and changes in latency are similar to those reported for isoflurane 0.15% in a previous study.18



View larger version (16K):
[in this window]
[in a new window]
 
Fig 3 Reciprocal median simple latency (±1 SE) plotted as a function of sevoflurane dose (MAC) for each subject, with fitted regression lines. Note that the latency axes are chosen to be appropriate for each subject.

 
SSRT
Figure 4 shows similar plots of SSRT, again using reciprocal latency scales and showing regression lines. Table 1 summarizes the parameters derived for each subject from the analysis (slopes and r values).



View larger version (15K):
[in this window]
[in a new window]
 
Fig 4 Reciprocal median stop-signal reaction time (±1 SE) plotted as a function of sevoflurane dose (MAC) for each subject, with fitted regression lines. Note that the SSRT axes are chosen to be appropriate for each subject.

 

View this table:
[in this window]
[in a new window]
 
Table 1 Slopes of the reciprocal latency vs MAC and reciprocal stop signal reaction time (SSRT) vs MAC in the six subjects; r, correlation coefficient; P, two-tailed significance of r. Note that the units for slope are s–1 per MAC
 
Both latency and SSRT increase with increasing levels of anesthetic. Correlation analysis yields correlation coefficients between dose and latency and between dose and SSRT of 0.76–0.99 (the slopes and r values for each subject are summarized in Table 1). Comparison of the slopes for latency and SSRT showed no significant difference (P=0.05), except for one subject for whom the slope for SSRT was steeper.


    Discussion
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendix: a brief summary...
 References
 
There was a monotonic increase in simple latency and in SSRT in all six subjects with increasing dose of sevoflurane in the sedative range 0–0.3 MAC. The reciprobit curves of saccadic latency distribution fitted the LATER model under all conditions, and sevoflurane moved these curves to the right with constant slope in a parallel manner (as shown in Fig. 2). In terms of the LATER model (see Appendix and elsewhere10 22 28), this implies that the movement of the curve was caused by a change in mean rate of rise of the decision signal, rather than an alteration in the threshold at which a saccade is generated. This, in turn, suggests that the anaesthetic effect is to reduce the rate at which information reaches the decision system. However, although the data appear to have a parallel shift, the data obtained in this study were too small to allow statistical verification of the shift in the sense of excluding other kinds of transformation. In other studies where this kind of analysis has been successfully carried out, several thousand trials were required.

The SSRT from the countermanding paradigm9 10 23 24 measures an individual’s capacity to stop an already planned action and implies a much more complex series of neuronal processes, almost certainly in the cortex. This suggested that the SSRT might be more susceptible to the effects of anaesthetic sedation than saccadic latency, and perhaps be a clinically relevant measure of whether patients can be discharged safely after day-care general anaesthesia. In terms of LATER, countermanding tasks can be considered a race between competing signals, one representing the ‘go’ signal and one representing ‘stop’. If both are subject to random variation of rate, then on some trials the stop signal may reach threshold before the go signal, and therefore inhibit the response, whereas in other trials it may not. This kind of competition can be observed in the activity of cortical neurones in monkeys carrying out countermanding tasks.10 The go signal activates gaze-shift neurones which progressively increase their activity in ramp-like fashion (LATER model) and at the same time the gaze-hold neurones show a comparable but progressive reduction in activity. In the absence of a stop signal, gaze-shift activation increases until the threshold for activating a saccade is reached and the eyes move. However, when the stop signal appears 50 ms after the go signal there is a much more rapid increase in activation of the gaze-hold neurones and the gaze-shift activation falls back to baseline.

Curiously, a number of our subjects commented that the SSRT task seemed easier to perform during anaesthetic administration. Only one subject showed a significant difference between the slopes of the latency–dose curve and the SSRT–dose curve, with SSRT being more sensitive. Nevertheless all these data show a correlation between anaesthetic dose and saccadic latency and with SSRT. This implies a dose-dependent effect of the drug on these two variables, suggesting that these can be used to assess functional recovery from anaesthesia.

The increase in latency with sevoflurane sedation is consistent with the results obtained in a previous study on isoflurane18 and other classes of sedative agents. After midazolam, saccadic latency increased and PSV decreased appropriately.19 Padoan and colleagues20 found that an increased latency persisted for longer after thiopental or diazepam than the decrease in saccadic velocity, but Paut and colleagues19 found that the changes in PSV were significantly different from baseline for longer than latency (180 min for PSV vs 120 min for latency). The limited number of trials (six runs of 30 trials) in each of their subjects, and the use of electroculography, meant that latency could not be determined with the precision that we could obtain in our study by having 2000 trials for each subject.

PSV may give a functional measure of levels of anaesthetic agent; 0.05 and 0.1 MAC concentrations of several agents, and benzodiazepines and propofol all depress PSV in a dose-dependent manner.1113 PSV is determined by the amplitude of the pulse of activity in burst neurones in the brainstem (in the pre-pontine reticular formation and mesencephalic reticular formation, regions described as the ‘pontine gaze centre’). Via a neuronal integrator, these units drive motor neurones which, in turn, activate the extraocular muscles. Thus, while PSV is a quantitative measure of the functional behaviour of these brainstem mechanisms, it is only an indirect way of estimating any impairment of higher cortical functions. As argued above, we propose that saccadic latency may be a more appropriate measure of sedation in certain situations. Saccadic latency can be measured non-invasively and essentially automatically in a matter of minutes, and could give a useful measure of residual impairment of higher sensorimotor function in recovery from general anaesthesia. Saccadic latency and SSRT could also be used as indexes of cortical recovery with a variety of other drugs, when we cannot easily establish the link between drug concentration in the brain and its effect.


    Appendix: a brief summary of the LATER model
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendix: a brief summary...
 References
 
Saccadic latency—the time between target presentation and the initiation of movement (Fig. 1)—is about 200 ms. This is much longer than that expected from nerve conduction times, or estimated from stimulation and recording in the shortest pathway linking visual input to oculomotor output via the superior colliculus.29 30 The reason for this ‘oculomotor procrastination’ is that cortical mechanisms, mediated by descending tonic inhibitory pathways acting on the superior colliculus, keep the lower structures in check while ‘deciding’ which of the many potential targets present in a typical visual field is of sufficient interest to merit a saccade. Saccadic latencies therefore reflect the time taken for these cortical decisions to be made, and the variation of saccadic latency, both randomly from trial to trial and in response to modifications of the environment, provides a way of investigating these decision-making processes in a quantitative manner. When large numbers of reaction times – whether saccadic or otherwise – are measured, they show considerable random variation from trial to trial. Nevertheless, the distribution of latency obeys a well-defined stochastic law (the recinormal distribution, with the reciprocal of latency being distributed in a Gaussian fashion). A convenient way to demonstrate this fact is to plot cumulative distributions using a probit ordinate scale with latency on a reciprocal abscissa (Fig. 5A); the resulting reciprobit plot will then be a straight line if the distribution is recinormal. Two parameters then characterize this line: the median latency {lambda}med, where it intersects the 50% ordinate, and the infinite-time intercept, I, the probability corresponding to t={infty}. LATER is a simple quantitative model of the decision process underlying the selection of responses to competing stimuli. It postulates a decision signal, initially at a level S0, that rises linearly at a rate, r, in response to a stimulus, until its arrival at a predetermined threshold level, ST, triggers a response. The reaction time, {lambda}, from stimulus to response is thus given by (STS0)/r. The rate r varies randomly from trial to trial, obeying a Gaussian distribution with mean µ and variance {sigma}2. Consequently, 1/{lambda}, the reciprocal of the latency, is distributed as a Gaussian with mean µ/(STS0) and variance {sigma}2/(STS0)2.



View larger version (12K):
[in this window]
[in a new window]
 
Fig 5 The LATER model. When reaction times are plotted as cumulative distributions on a probit scale as a function of the reciprocal of latency (a reciprobit plot), in general they are found to yield a straight line (A), implying that the reciprocal of latency is distributed as a Gaussian. The line intersects the P=50% ordinate at the median latency, and the t=infinity axis at a point I. This behaviour can be explained by a simple mechanism (the LATER model) in which a decision signal, S, initially at a level S0, rises linearly on presentation of a stimulus until it reaches a threshold, ST, at which point a response is initiated. On different trials, r varies randomly about a mean µ in a Gaussian manner, such that the distribution of latency is skewed (shaded area) and the reciprocal of latency is Gaussian, as observed.

 
If the LATER model were simply an empirical description, its value would be limited, like all statistical descriptions, to the fact that an entire set of latencies can be economically described with just two parameters. It is in fact rather more than that: it represents a mechanism for making decisions between competing hypotheses under conditions of uncertainty. Though our perceptions may generally seem to us clear and veridical, it is easy to show that they are subject to uncertainty and error, arising from two main sources. One is our own expectations, the other is noise in afferent sensory processes. Perception thus becomes a matter of judging the likelihood of the existence of objects around us, based on our past experience (prior probability) and actual information coming in through our senses. At any moment there will be many competing hypotheses, and it is the task of the brain to identify the one that is most likely. We update the probabilities on the basis of the incoming information, converting our prior probabilities into posterior probabilities. Mathematically this is described by Bayes’ theorem:

p'(Hi) = p(Hi).p(E|Hi)/p(E)

where p(Hi) and p'(Hi) are the prior and posterior probabilities, respectively, of the ith competing hypothesis, E is some experimental observation, and p(E|Hi) is the probability of making that observation, given Hi; p(E), the global probability of E, is common to all hypotheses. Taking logarithms:

log(p'(Hi)) = log(p(Hi)) + log(p(E|Hi)) – log(p(E))

If information is being presented continuously from a stimulus, then it is clear that log(p(Hi) will rise linearly. There is thus a one-to-one correspondence between this formulation of an ideal decision maker and the LATER model: S0 represents log(p), r represents the rate of arrival of information, and ST represents a criterion level at which the hypothesis is sufficiently supported for a response to be justified: it is equivalent to a ‘significance level’ in conventional statistics.

Experiments have tested whether this interpretation of the LATER model is correct, by seeing whether, when expectation or the supply of information or the criterion level are changed, the observed effects on the latency distributions are as predicted. So far these tests have proved satisfactory. Other experiments have studied situations where more than one LATER units operate in parallel, embodying the competition that results in one hypothesis being selected rather than another. These too appear to support the LATER model. Finally, neural correlates of such a mechanism have been identified in the frontal cortical eye fields of macaque monkeys through single-unit recording techniques.7 31 Recording from these ‘movement cells’, which increase their firing rate just before saccade generation, shows a linear rise in activity whose rate is correlated with latency, whereas the threshold of firing corresponding to saccadic activation remains essentially constant. Thus these neurones show similar behaviour to that predicted by the LATER model: the determination of saccadic latency appears to involve cortical centres that act indirectly on the more primitive brainstem circuitry. This converts the command ‘make the movement’ into an appropriate pattern of motor nerve activity that will move the eye rapidly and accurately on to its target (and which also, incidentally, determines PSV). Since manual responses as well as saccades appear to follow a recinormal distribution, and for other stimuli (auditory, tactile) as well as visual, there seems to be good grounds for supposing that something like the LATER model represents a fundamental neural process underlying decision making in the brain.


    References
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Appendix: a brief summary...
 References
 
1 Leigh RJ, Zee DS. The Neurology of Eye Movements, 3rd Edn. New York: Oxford University Press, 1999; 128

2 Carpenter RHS. Movements of the Eyes, 2nd Edn. London: Pion, 1988; 69

3 Laborit G, Angiboust R, Papin JP. A study of eye movement for assessing recovery from anaesthesia. Br J Anaesth 1977; 49: 805–10[Abstract]

4 Laborit G, Angiboust R, Papin JP. Monitoring the integrity of saccadic eye movements as a test of post-anaesthetic recovery. Rev Electroenceph Neurophysiol Clin 1977; 7: 70–5[ISI]

5 Khan OA, Taylor SRJ, Jones JG. Anaesthesia and saccadic eye movements. Anaesthesia 2000; 55: 877–82[CrossRef][ISI][Medline]

6 Carpenter RHS. SPIC: a PC-based system for rapid measurement of saccadic responses. J Physiol Proc 1994; 480: 4P

7 Schall JD, Thompson KG. Neural selection and control of visually guided eye movements. Ann Rev Neurosci 1999; 22: 241–59[CrossRef][ISI][Medline]

8 Carpenter RHS. A neural mechanism that randomises behaviour. J Consc Studies 1999; 6: 13–22

9 Logan GD, Cowan WB. On the ability to inhibit thought and action: a theory of an act of control. Psychol Rev 1984; 91: 295–327[CrossRef][ISI]

10 Schall J, Taylor DH. Neural control of behavior: countermanding eye movements. Psychol Res 2000; 63: 299–307[CrossRef][ISI][Medline]

11 Yoshizumi, J, Marshall RW, Sanders LD, Vickers MD. Effects of small concentrations of isoflurane on some psychometric measurements. Br J Anaesth 1993; 71: 839–44[Abstract]

12 Yoshizumi J, Marshall RW, Vickers MD. Effect of small concentrations of cyclopropane and halothane on peak velocity of saccadic eye movement. Br J Anaesth 1991; 67: 735–40[Abstract]

13 Gao F, Marshall RW, Sanders LD, Vickers MD. Effect of subanaesthetic infusions of propofol on peak velocity of saccadic movement. Eur J Anaesth 1991; 8: 267–76[ISI][Medline]

14 Kapoula Z. Evidence of a range effect in the saccadic system. Vision Res 1985; 25: 1155–7[CrossRef][ISI][Medline]

15 Wittenborn JR. The effects of benzodiazepines on psychomotor performance. Br J Clin Pharmacol 1979; 7: 61–7S

16 Griffiths A, Mercer A, Jones D, Richens A. The effects of zolpiden, a novel hypnotic, on saccadic eye movements, psychomotor performance and mood in healthy male volunteers. Acta Pharmacol Toxicol 1986; 59 [Suppl. 5]: 120

17 Hindmarch I, Bhatti JZ. Recovery of cognitive and psychomotor function following anaesthesia. A review. In: Hindmarsh I, Jones JG, Moss E, eds. Aspects of Recovery from Anaesthesia. New York: John Wiley & Sons, 1987; 113–70

18 Khan OA, Taylor SRJ, Jones JG, Swart M, Hanes DP, Carpenter RHS. The effects of low-dose isoflurane on saccadic eye movement generation. Anaesthesia 1999; 54: 142–5[CrossRef][ISI][Medline]

19 Paut O, Vercher J-L, Blin O, et al. Evaluation of saccadic eye movements as an objective test of recovery from anaesthesia. Acta Anaesthesiol Scand 1995; 39: 1117–24[ISI][Medline]

20 Padoan S, Kortilla K, Magnusson M, Pyykko I, Schalen L. Effect of intravenous diazepam and thiopental on voluntary saccades and pursuit eye movements. Acta Otolaryngol (Stockh.) 1992; 112: 579–88[ISI][Medline]

21 Carpenter RHS. Oculomotor procrastination. In: Fisher DF, Monty RA, Senders JW, eds. Eye Movements: Cognition and Visual Perception. Hillsdale: Lawrence Elbaum, 1981; 237–46

22 Hanes D, Patterson W, Schall J. Role of frontal eye fields in countermanding saccades: Visual, movement, and fixation activity. J Neurophysiol 1998; 79: 817–34[Abstract/Free Full Text]

23 Hanes D, Carpenter RHS. Countermanding saccades in humans. Vision Res 1999; 39: 2777–91[CrossRef][ISI][Medline]

24 Kendall MG, Stuart A. The Advanced Theory of Statistics. London: Griffin, 1968

25 Taylor SRJ, Khan OA, Swart ML, Lockwood GG, Jones JG. Effects of a low concentration of isoflurane on contrast sensitivity in volunteers. Br J Anaesth 1998; 81: 176–9[Abstract/Free Full Text]

26 Mapleson WW. Effect of age on MAC in humans: a meta analysis. Br J Anaesth 1996; 76: 179–85[Abstract/Free Full Text]

27 Kolmogorov A. Confidence limits for an unknown distribution function. Ann Math Stats 1941; 23: 525–40

28 Carpenter RHS, Williams MLL. Neural computation of log likelihood in the control of saccadic eye movements. Nature 1995; 377: 59–62[CrossRef][ISI][Medline]

29 Guitton D. Control of saccadic eye and gaze movements by the superior colliculus and basal ganglia. In: Carpenter RHS, ed. Eye Movements. London: MacMillan, 1992; 244–76

30 Sparks D. Translation of sensory signals into commands for control of saccadic eye movements: role of primate superior colliculus. Physiol Rev 1986; 66: 118–71[Abstract/Free Full Text]

31 Hanes DP, Schall JD. Neural control of voluntary movement initiation. Science 1996; 274: 427–30[Abstract/Free Full Text]





This Article
Abstract
Full Text (PDF)
E-Letters: Submit a response to the article
Alert me when this article is cited
Alert me when E-letters are posted
Alert me if a correction is posted
Services
Email this article to a friend
Similar articles in this journal
Similar articles in ISI Web of Science
Similar articles in PubMed
Alert me to new issues of the journal
Add to My Personal Archive
Download to citation manager
Search for citing articles in:
ISI Web of Science (1)
Disclaimer
Request Permissions
Google Scholar
Articles by Nouraei, S. A. R.
Articles by Carpenter, R. H. S.
PubMed
PubMed Citation
Articles by Nouraei, S. A. R.
Articles by Carpenter, R. H. S.