Predictive Control of Eye Movements in Parkinson Disease K. A. Flowers, PhD, and A. C . Downing, P h D
Four parkinsonian patients who had shown evidence of an impairment of predictive manual control and 4 agematched normal subjects were tested for the predictive control of eye movements. Subjects tracked a target with their eyes as it moved in either irregular “noise” or regular (predictable) linear ramp or sine waveforms. Eye movements were monitored by electrooculography, and the overall tracking time lag for each condition was determined by cross-correlation. Both normal and parkinsonian subjects showed prediction in eye tracking on the regular waveforms (zero time lag or anticipation of the target track), indicating that (1) the parkinsonian loss of predictive control in manual tasks is not due to defective control of eye movements, and (2) there may be separate predictor mechanisms in the brain for eyes and hands. Flowers KA, Downing AC: Predictive Control of Eye Movements in Parkinson Disease. Ann Neurol4:63-66, 1978 Patients suffering from Parkinson disease are well known to have an impairment of the voluntary control of movement (akinesia). This impairment concerns most notably the coordination of movements into combinations and sequences, implying a functional loss at the “highest level” [ l , 71. In a recent series of experiments using a pursuit tracking task, it was suggested that a major aspect of the parkinsonian impairment is loss of prediction in the control of movement [4, 51. Where normal subjects make use of the redundancy of events in the external world to anticipate them and to act independently of sensory data, parkinsonian subjects d o not, but tend always to lag behind events. They appear, in other words, to behave in the fashion of a closed-loop servosystem reacting from moment to moment to current sensory information (as normal persons d o when their actions are directed toward events which cannot be predicted in advance). Their disability is therefore partly a cognitive one and is most marked in those conditions which are easiest for normal persons. Considerable evidence suggests that a similar kind of prediction normally operates in the control of eye movements during smooth ocular pursuit of a regularly moving target [8], though most investigations have revealed this in the form of a diminution of lag rather than its complete elimination o r replacement by anticipation, as is typical for skilled manual performance. It seemed worth comparing the visual tracking of parkinsonian patients and normal subjects to see if they remained similar in this ability, o r
whether the patients were as unable to utilize the predictability of target movement to improve ocular pursuit, as they seem to be in the case of manual tracking. Moreover, since Fender’s argument [3] that the relationship of visual target movement and tracking eye movements is nonlinear and so cannot be expressed as a transfer function (with its implication of a fixed phase difference between input and output), there has been a general need for more data on the relative timing of these movements. “Without this datum, who is to say that the system ‘predicts’, unless the phase lags are actually reduced to zero o r even turned into a phase lead? The experimental evidence is that neither of these two conditions ever occurs” ([3], p 540). There has been too little information on whether and when this can be the case. Studies of eye tracking in parkinsonian patients and normal subjects should both contribute to the understanding of Parkinson disease and be of general interest in elucidating oculomotor control. Materials a n d Method Four parkinsonian patients were tested. They had all shown distinct evidence of abnormal manual tracking, with an overall positive reaction time even when tracking a target moving in a predictable sinusoidal or linear ramp waveform (see Table 2), but they were not the most impaired of the parkinsonian group. They were not specially selected on the basis of any prior evaluation of their oculomotor performance. There were 2 men and 2 women, all outpatients with two or more years’ history of the disease; their ages were 37, 52, 63, and 74 years. All were
~
From the Brain and Perception Laboratory, Department of Anatomy, The Medical School, Bristol, England.
Address reprint requests to Dr Flowers, Department of Psychology, University of Hull, Hull HU6 7RX, England.
Accepted for publication Jan 25, 1978.
0364-5134/78/0004-Oll1$01.25 @ 1778 by K. A. Flowers 63
being treated with L-dopa, which they claimed had helped them greatly, mostly in reducing the resting tremor and rigidity of their limbs. Two had had a unilateral thalamotomy some years previously. Details of their clinical assessments are given elsewhere (in [S], Table 1, Subjects 1, 2, 7 , and 8). Four paid volunteer control subjects, 2 men and 2 women, were also tested. They were matched for age with the patients, their ages being 34, 51, 61, and 72 years. Two had performed o n the manual tracking test and 2 had not. A large-screen oscilloscope (Lanscope Model 419A display scope with a 17-inch short-persistence TV-type tube) was used for the display. The target was a small spot that moved horizontally across the screen, maintaining its vertical position constant, and was driven from side to side in the various waveforms from an Advance VLF generator (Model SG88) capable of providing any repetitive function, including repeatable noise. Subjects sat squarely in front of a table with the display set at head height 56 cm away from the table’s outside edge. They were positioned with the head resting securely on a chin rest and the weight of their body taken comfortably by the arms resting on the table. Subjects kept their heads steady during practice and recording periods; this was easily checked since any head movement resulted in a shift being registered on the eye movement recording apparatus. T h e horizontal component of eye movement was recorded by electrooculography. The experiment did not require information about absolute direction of gaze but only about the time relationships between target movement and ocular pursuit. To diminish any possible stress on our elderly subjects from the recording procedures, we made no attempt at absolute calibration of the eye movement signal for each subject, and we adopted an electrooculographic technique similar to that described by Geddes et a1 [61, which minimizes the fuss involved in preparing the skin and attaching electrodes. Two silver disc electrodes, mounted on insulating plastic blocks, were held in contact with the skin of the temples, level with the eyes, by the springy headpiece from a pair of audio headphones. (We found Eagle SF-20 headphones particularly suitable since
they are widely adjustable and the earpieces are held in gimbal-type clips, from which they are easily removed to be replaced by the electrode holders. The gimbals help to ensure good skin contact.) A very high input impedance J-FET source-follower preamplifier was mounted in each electrode holder. The output voltages from the two preamplifiers were fed to an Ac-coupled differential amplifier with long time constant and variable gain and shift controls, to give the final eye movement signal.
Procedure Subjects were tested in a single session, lasting about an hour and a half, some six weeks or so after they had been tested o n the manual tracking task. They were told that the purpose of the experiment was to see how closely they could follow a moving target with their eyes, but were not otherwise informed on the rationale of the study. Subjects were first fitted with the electrode headset for recording eye movements. At least ten minutes was allowed for the apparatus to settle down and for drifts to be reduced. After the subject had been positioned in front of the display with the chin rest adjusted to a comfortable height, he was told to watch the target spot as i t moved from side to side, keeping his head as still as possible all the time. Next there was a short practice period with the subject following a series of step functions of various amplitudes, which allowed the subject to show he understood the task and could follow target movements and the experimenters to adjust the amplifiers to a convenient gain. Before each test run with the various waveforms, the subject was shown the target movement and allowed as much practice as he wished (this was usually short). A single test run was made for each condition with a short rest between each one and whenever the subject needed it. O n e parkinsonian and 1 control subject were not tested on the fastest sine condition for lack of time. In addition, one record was lost. Details o f the target tracks presented are listed in Table 1 in order of presentation. Save for the fastest sine-wave tracks (conditions ( f ) and (g) in the list), they are all identical to the displays used in the manual tracking test. All the subjects were tested in the same order. The amplitude of
Table 1 . Details of Target Movenient in the Predictable and Unpredictable Conditions ~
~
Length of Test Run Condition
Amplitude o n Screen (mm)
Frequency (Hz)
Slow sine Slow sine sweep Fast sine Fast sine sweep (e) V fast sine (0 VV fast sine 8 (g) W fast sine 4
125 125 125 125 125 80 40
1/6
(h) Slow noise (i) Fast noise (j) Triangle
(a) (b) (c) (d)
116- 112
Cycles 8 18
112
16
1/2- 1 1
1%
25 32 24 24
Approx 110 average Approx 110 average
Approx 116 average Approx 112 average
Approx 6 Approx 16
125
Ill0
64 Annals of Neurology Vol 4 No 1 July 1978
195
6
Time (sec)
48 60 32 35 32 16
16 50 35
60
Table 2. Overall Tracking Time Lag (msec) on Irregular (Noise) Waveforms
target movement was 125 mm, equivalent to a visual angle of 12"44' at the eye. Tracking was recorded o n two channels of a twelvechannel UV oscillograph recorder (SE Labs Model 3006/ DL). Target and subject positions were also recorded o n punch tape by means of a multiplexer sampling the data at intervals of 80 msec, an A-D converter, and a tape punch. The overall lag or lead of ocular pursuit movements with respect to the target movement was determined by digital cross-correlation analysis of the punched tape record. A set of correlation coefficients was computed between the samples of target position and corresponding samples of eye movement record, with these two sets of data displaced from each other in time by successive multiples of 80 msec. That delay of target movement record with respect to eye movement record which gave the highest product-moment correlation between the two was taken as the overall time lag of ocular pursuit for a given run. It is very unlikely that the relationship between target position and eye movement signal departs sufficiently from linearity to give a spurious result, since we are not concerned with the absolute values of the correlation coefficients or with their ratios (as in the derivation of a phase angle), but simply with their rank order. Thus, we are able to measure lags or leads in our subjects' eye tracking in terms of time rather than phase. Since gain o n the eye movement apparatus was different for each subject, the values of the mean squared error scores in this calcu1atio.n are not comparable between subjects. This does not, however, affect the cross-correlation calculation, which simply measures the best fit in time of the two series of data. Scrutiny of the UV records enabled a check to be made that the subject was following the shape of the waveform reasonably well. When any doubt existed, o r if the subject's tracking broke down completely in the course of the track, or if the eye movement record drifted too much during the test run, the test was repeated until an acceptable full-length record was obtained.
Subject No.
Sex
Age (yr)
Slow Noise
Fast Noise
Parkinsonian group Patient 1
F
39
Patient 2
M
52
80 (340) 80
Patient 3
M
63
(260) 0
Patient 4
F
74
160 (420) 80 (320) 240 (320) 80 (320)
Control group Subject 1 Subject 2 Subject 3 Subiect 4
F F M M
34 51
61 72
~
~
(160) 0 (200)
0 0 0 0
80 80 80 80
~
~~
Figures in parentheses for the parkinsonian group indicate the median manual tracking lag on the same condition.
Results Tables 2, 3, and 4 show the overall phase lag or lead of each subject on each condition as determined by the cross-correlation analysis. The figures in parentheses show the equivalent measure for the parkinsonian group on the manual tracking task. Minus scores indicate a mean lead on the test run as a whole. From these figures it is clear that the parkinsonian patients show a definite ability to anticipate the movement of the target on regular, predictable tracks (see Tables 3, 4). Their mean timing difference is either zero o r a lead in all cases except o n the two fastest sine waves, and o n these tracks the normal subjects also showed a tendency to lag slightly. Their
Table 3. Overall Tracking Time Lug or L a d (msec) on Regular (Sine) Waveforms Subject No.
Slow Sine
Fast Sine
V Fast Sine
Parkinsonian group Patient I
-240
0 (144) 0) (20) - 80 (60) - 80 (240)
80 (280) 80 (0)
(180)
Patient 2
0
(160) Patient 3 Patient 4
-80 (144) - 80
(80)
Control group Subject 1 Subject 2 Subjcct 3 Subjcct 4
-80 - 160
-80 -240
0 80 - 80 -80 -
VV Fast Sine ( 8 )
0
. .) ... (. . .)
80 . .) 80 (. . .) 0 (. . .) ... (. . .)
80 80 0
0 80 80
...
...
(.
. .)
(.
. .)
0
... (80) -80
(300) 0 0 0 0
VV Fast Sine ( 4 )
0 (.
(.
Negative values indicate phase lead. Figures in parentheses for the parkinsonian group indicate the median manual tracking lag on the same condition.
Flowers and Downing: Eye Movements in Parkinsonism
65
Tuhle 4.0 1 erull Tracking Time Lag or Leud (mseri on Reguluv ( S t t i e Sweep and Rmip) Wat eforms
Subject N o . Parkinsonian g r o u p Patient 1 Patient 2 Patient
3
Patient 4
Control group Subject 1 Subject 2
Subject 3 Subiect 4
Slow Sine Swccp
Fast Sine Sweep
-80 (120) 0
0 (240) 0
(96) 160 (144) -80 (220)
-80 (160) -80 (240)
-
(64)
Ramp
-240 (180) -240 (200) 0 (160) -80 (160)
- 80 - 80
- 160 - 80
-80 - 80
-320 -400
Negative values indicate phase lead. Figures in parentheses for the parkinsoniati group indicate median manual tracking lag on the same condition.
performance is thus strikingly predictive and the difference compared to normal subjects very small, certainly much smaller than in manual tracking. The performance of the parkinsonian patients on the two noise tracks is also a good deal better than their equivalent manual tracking performance (see Table 2), even though they are not quite as good at keeping up with the target as are the normal subjects. The difference in lag between the fast and slow noise trials in both groups of subjects, and between the groups in both conditions, shows also that the zero o r negative lag on the regular waveforms is not an artifact of the sampling or cross-correlation procedure. That is, it is unlikely that a lag of 40 msec or less for the regular tracks is hidden by the sampling frequency of 80 msec. In all cases, therefore, it appears that the ability of parkinsonian patients to control their eye movements is essentially normal in terms of both reaction time and the ability to control movement predictively. Discussion Three conclusions may be drawn from these results. First, the manual tracking deficit shown by patients suffering from Parkinson disease is not due to faulty e y e movement control. These 4 patients, who all showed marked phase lags in the manual pursuit task, were able to track the same target movements presented on the same display quite adequately with their eyes. Also, in contrast to their performance on the manual task, they showed no particular difficulty in tracking the higher frequency, faster moving target tracks with their eyes. It seems unlikely, therefore, that the sensory (visual) input to the sensorimotor control system is disturbed in this disease. 66 Annals of N e u r o l o g y Vol 4 No 1 July 1978
Second, most of the 8 subjects, patients and normal persons alike, showed a negative overall time lag for eye tracking. This confirms results such as those of Drischel[21, which have indicated that smooth eye movements can actually precede predictable target movements. This confirmation is independent of the now suspect derivation of a phase ande. The smooth e y e tracking control system does seem to include a predictor, notwithstanding the objections raised by Fender [3] to previous evidence of prediction. Finally, a discrepancy exists between this evidence of fairly normal predictive control in eye tracking by parkinsonian patients and their inability, found by Flowers [ 5 ] , to show normal predictive control in tracking the same target motion manually. This implies that either (1) normal brains have at least two separate predictors which produce signals modelling probable target motion, one for voluntary control of smooth limb movements and the other for oculomotor pursuit, while in Parkinson disease the former, but not the latter, is impaired; o r ( 2 ) the same predictor generates signals which, in normal brains, go to both the smooth eye movement control system and also the system controlling smooth voluntary limb movements, whereas in Parkinson disease the output of this predictor to that part of the limb control system is impaired. Whichever of these explanations is true, parkinsonian patients evidently retain some capability for generating and updating running predictions from visual information. Supported by the Medical Research Council (UK). We are indebted to Dr R. Iangton Hewer for access to patients under his care and to Dr M. Hilary Morgan for clinical assessments and advice. We thank Mr I. Low and Mr M. Tajfel for technical and computing help.
References 1. Angel RA, Alston W, Higgins JR: Control of movement in Parkinson’s disease. Brain 93: 1-14, 1970 2. Drischel H : The frequency response of horizontal pursuit movements of the human eye and the influence of alcohol, in Asratyan EA (ed): Progress in Brain Research (Brain Reflexes). Amsterdam, Elsevier, 1968, vol 22 3 . Fender D H : Time delays in the human eye-tracking system, in Bach-y-Rita P, Collins CC, Hyde JE (eds): The Control of Eye Movements. London and New York, Academic, 1971 4. Flowers KA: Visual “closed-loop” and “open-loop” characteristics of voluntary movement in patients with parkinsonism and intention tremor. Brain 99:269-3 10, 1976 5. Flowers KA: Some frequency-response characteristics of parkinsonism on pursuit tracking. Brain 101:19-34, 1978 6. Geddes LA, Steinburg R, Wise G: Dry electrodes and holder for electro-oculography. Med Biol Eng 11:69-72, 1973 7. Home DJ de L: Sensorimotor control in parkinsonism. J Neurol Neurosurg Psychiatry 36:742-746, 1973 8. Young LR: Pursuit eye tracking movements, in Bach-y-Rita P, Collins CC, Hyde JE (eds): The Control of Eye Movements. London and New York, Academic, 197 1
Four parkinsonian patients who had shown evidence of an impairment of predictive manual control and 4 agematched normal subjects were tested for the predictive control of eye movements. Subjects tracked a target with their eyes as it moved in either irregular “noise” or regular (predictable) linear ramp or sine waveforms. Eye movements were monitored by electrooculography, and the overall tracking time lag for each condition was determined by cross-correlation. Both normal and parkinsonian subjects showed prediction in eye tracking on the regular waveforms (zero time lag or anticipation of the target track), indicating that (1) the parkinsonian loss of predictive control in manual tasks is not due to defective control of eye movements, and (2) there may be separate predictor mechanisms in the brain for eyes and hands. Flowers KA, Downing AC: Predictive Control of Eye Movements in Parkinson Disease. Ann Neurol4:63-66, 1978 Patients suffering from Parkinson disease are well known to have an impairment of the voluntary control of movement (akinesia). This impairment concerns most notably the coordination of movements into combinations and sequences, implying a functional loss at the “highest level” [ l , 71. In a recent series of experiments using a pursuit tracking task, it was suggested that a major aspect of the parkinsonian impairment is loss of prediction in the control of movement [4, 51. Where normal subjects make use of the redundancy of events in the external world to anticipate them and to act independently of sensory data, parkinsonian subjects d o not, but tend always to lag behind events. They appear, in other words, to behave in the fashion of a closed-loop servosystem reacting from moment to moment to current sensory information (as normal persons d o when their actions are directed toward events which cannot be predicted in advance). Their disability is therefore partly a cognitive one and is most marked in those conditions which are easiest for normal persons. Considerable evidence suggests that a similar kind of prediction normally operates in the control of eye movements during smooth ocular pursuit of a regularly moving target [8], though most investigations have revealed this in the form of a diminution of lag rather than its complete elimination o r replacement by anticipation, as is typical for skilled manual performance. It seemed worth comparing the visual tracking of parkinsonian patients and normal subjects to see if they remained similar in this ability, o r
whether the patients were as unable to utilize the predictability of target movement to improve ocular pursuit, as they seem to be in the case of manual tracking. Moreover, since Fender’s argument [3] that the relationship of visual target movement and tracking eye movements is nonlinear and so cannot be expressed as a transfer function (with its implication of a fixed phase difference between input and output), there has been a general need for more data on the relative timing of these movements. “Without this datum, who is to say that the system ‘predicts’, unless the phase lags are actually reduced to zero o r even turned into a phase lead? The experimental evidence is that neither of these two conditions ever occurs” ([3], p 540). There has been too little information on whether and when this can be the case. Studies of eye tracking in parkinsonian patients and normal subjects should both contribute to the understanding of Parkinson disease and be of general interest in elucidating oculomotor control. Materials a n d Method Four parkinsonian patients were tested. They had all shown distinct evidence of abnormal manual tracking, with an overall positive reaction time even when tracking a target moving in a predictable sinusoidal or linear ramp waveform (see Table 2), but they were not the most impaired of the parkinsonian group. They were not specially selected on the basis of any prior evaluation of their oculomotor performance. There were 2 men and 2 women, all outpatients with two or more years’ history of the disease; their ages were 37, 52, 63, and 74 years. All were
~
From the Brain and Perception Laboratory, Department of Anatomy, The Medical School, Bristol, England.
Address reprint requests to Dr Flowers, Department of Psychology, University of Hull, Hull HU6 7RX, England.
Accepted for publication Jan 25, 1978.
0364-5134/78/0004-Oll1$01.25 @ 1778 by K. A. Flowers 63
being treated with L-dopa, which they claimed had helped them greatly, mostly in reducing the resting tremor and rigidity of their limbs. Two had had a unilateral thalamotomy some years previously. Details of their clinical assessments are given elsewhere (in [S], Table 1, Subjects 1, 2, 7 , and 8). Four paid volunteer control subjects, 2 men and 2 women, were also tested. They were matched for age with the patients, their ages being 34, 51, 61, and 72 years. Two had performed o n the manual tracking test and 2 had not. A large-screen oscilloscope (Lanscope Model 419A display scope with a 17-inch short-persistence TV-type tube) was used for the display. The target was a small spot that moved horizontally across the screen, maintaining its vertical position constant, and was driven from side to side in the various waveforms from an Advance VLF generator (Model SG88) capable of providing any repetitive function, including repeatable noise. Subjects sat squarely in front of a table with the display set at head height 56 cm away from the table’s outside edge. They were positioned with the head resting securely on a chin rest and the weight of their body taken comfortably by the arms resting on the table. Subjects kept their heads steady during practice and recording periods; this was easily checked since any head movement resulted in a shift being registered on the eye movement recording apparatus. T h e horizontal component of eye movement was recorded by electrooculography. The experiment did not require information about absolute direction of gaze but only about the time relationships between target movement and ocular pursuit. To diminish any possible stress on our elderly subjects from the recording procedures, we made no attempt at absolute calibration of the eye movement signal for each subject, and we adopted an electrooculographic technique similar to that described by Geddes et a1 [61, which minimizes the fuss involved in preparing the skin and attaching electrodes. Two silver disc electrodes, mounted on insulating plastic blocks, were held in contact with the skin of the temples, level with the eyes, by the springy headpiece from a pair of audio headphones. (We found Eagle SF-20 headphones particularly suitable since
they are widely adjustable and the earpieces are held in gimbal-type clips, from which they are easily removed to be replaced by the electrode holders. The gimbals help to ensure good skin contact.) A very high input impedance J-FET source-follower preamplifier was mounted in each electrode holder. The output voltages from the two preamplifiers were fed to an Ac-coupled differential amplifier with long time constant and variable gain and shift controls, to give the final eye movement signal.
Procedure Subjects were tested in a single session, lasting about an hour and a half, some six weeks or so after they had been tested o n the manual tracking task. They were told that the purpose of the experiment was to see how closely they could follow a moving target with their eyes, but were not otherwise informed on the rationale of the study. Subjects were first fitted with the electrode headset for recording eye movements. At least ten minutes was allowed for the apparatus to settle down and for drifts to be reduced. After the subject had been positioned in front of the display with the chin rest adjusted to a comfortable height, he was told to watch the target spot as i t moved from side to side, keeping his head as still as possible all the time. Next there was a short practice period with the subject following a series of step functions of various amplitudes, which allowed the subject to show he understood the task and could follow target movements and the experimenters to adjust the amplifiers to a convenient gain. Before each test run with the various waveforms, the subject was shown the target movement and allowed as much practice as he wished (this was usually short). A single test run was made for each condition with a short rest between each one and whenever the subject needed it. O n e parkinsonian and 1 control subject were not tested on the fastest sine condition for lack of time. In addition, one record was lost. Details o f the target tracks presented are listed in Table 1 in order of presentation. Save for the fastest sine-wave tracks (conditions ( f ) and (g) in the list), they are all identical to the displays used in the manual tracking test. All the subjects were tested in the same order. The amplitude of
Table 1 . Details of Target Movenient in the Predictable and Unpredictable Conditions ~
~
Length of Test Run Condition
Amplitude o n Screen (mm)
Frequency (Hz)
Slow sine Slow sine sweep Fast sine Fast sine sweep (e) V fast sine (0 VV fast sine 8 (g) W fast sine 4
125 125 125 125 125 80 40
1/6
(h) Slow noise (i) Fast noise (j) Triangle
(a) (b) (c) (d)
116- 112
Cycles 8 18
112
16
1/2- 1 1
1%
25 32 24 24
Approx 110 average Approx 110 average
Approx 116 average Approx 112 average
Approx 6 Approx 16
125
Ill0
64 Annals of Neurology Vol 4 No 1 July 1978
195
6
Time (sec)
48 60 32 35 32 16
16 50 35
60
Table 2. Overall Tracking Time Lag (msec) on Irregular (Noise) Waveforms
target movement was 125 mm, equivalent to a visual angle of 12"44' at the eye. Tracking was recorded o n two channels of a twelvechannel UV oscillograph recorder (SE Labs Model 3006/ DL). Target and subject positions were also recorded o n punch tape by means of a multiplexer sampling the data at intervals of 80 msec, an A-D converter, and a tape punch. The overall lag or lead of ocular pursuit movements with respect to the target movement was determined by digital cross-correlation analysis of the punched tape record. A set of correlation coefficients was computed between the samples of target position and corresponding samples of eye movement record, with these two sets of data displaced from each other in time by successive multiples of 80 msec. That delay of target movement record with respect to eye movement record which gave the highest product-moment correlation between the two was taken as the overall time lag of ocular pursuit for a given run. It is very unlikely that the relationship between target position and eye movement signal departs sufficiently from linearity to give a spurious result, since we are not concerned with the absolute values of the correlation coefficients or with their ratios (as in the derivation of a phase angle), but simply with their rank order. Thus, we are able to measure lags or leads in our subjects' eye tracking in terms of time rather than phase. Since gain o n the eye movement apparatus was different for each subject, the values of the mean squared error scores in this calcu1atio.n are not comparable between subjects. This does not, however, affect the cross-correlation calculation, which simply measures the best fit in time of the two series of data. Scrutiny of the UV records enabled a check to be made that the subject was following the shape of the waveform reasonably well. When any doubt existed, o r if the subject's tracking broke down completely in the course of the track, or if the eye movement record drifted too much during the test run, the test was repeated until an acceptable full-length record was obtained.
Subject No.
Sex
Age (yr)
Slow Noise
Fast Noise
Parkinsonian group Patient 1
F
39
Patient 2
M
52
80 (340) 80
Patient 3
M
63
(260) 0
Patient 4
F
74
160 (420) 80 (320) 240 (320) 80 (320)
Control group Subject 1 Subject 2 Subject 3 Subiect 4
F F M M
34 51
61 72
~
~
(160) 0 (200)
0 0 0 0
80 80 80 80
~
~~
Figures in parentheses for the parkinsonian group indicate the median manual tracking lag on the same condition.
Results Tables 2, 3, and 4 show the overall phase lag or lead of each subject on each condition as determined by the cross-correlation analysis. The figures in parentheses show the equivalent measure for the parkinsonian group on the manual tracking task. Minus scores indicate a mean lead on the test run as a whole. From these figures it is clear that the parkinsonian patients show a definite ability to anticipate the movement of the target on regular, predictable tracks (see Tables 3, 4). Their mean timing difference is either zero o r a lead in all cases except o n the two fastest sine waves, and o n these tracks the normal subjects also showed a tendency to lag slightly. Their
Table 3. Overall Tracking Time Lug or L a d (msec) on Regular (Sine) Waveforms Subject No.
Slow Sine
Fast Sine
V Fast Sine
Parkinsonian group Patient I
-240
0 (144) 0) (20) - 80 (60) - 80 (240)
80 (280) 80 (0)
(180)
Patient 2
0
(160) Patient 3 Patient 4
-80 (144) - 80
(80)
Control group Subject 1 Subject 2 Subjcct 3 Subjcct 4
-80 - 160
-80 -240
0 80 - 80 -80 -
VV Fast Sine ( 8 )
0
. .) ... (. . .)
80 . .) 80 (. . .) 0 (. . .) ... (. . .)
80 80 0
0 80 80
...
...
(.
. .)
(.
. .)
0
... (80) -80
(300) 0 0 0 0
VV Fast Sine ( 4 )
0 (.
(.
Negative values indicate phase lead. Figures in parentheses for the parkinsonian group indicate the median manual tracking lag on the same condition.
Flowers and Downing: Eye Movements in Parkinsonism
65
Tuhle 4.0 1 erull Tracking Time Lag or Leud (mseri on Reguluv ( S t t i e Sweep and Rmip) Wat eforms
Subject N o . Parkinsonian g r o u p Patient 1 Patient 2 Patient
3
Patient 4
Control group Subject 1 Subject 2
Subject 3 Subiect 4
Slow Sine Swccp
Fast Sine Sweep
-80 (120) 0
0 (240) 0
(96) 160 (144) -80 (220)
-80 (160) -80 (240)
-
(64)
Ramp
-240 (180) -240 (200) 0 (160) -80 (160)
- 80 - 80
- 160 - 80
-80 - 80
-320 -400
Negative values indicate phase lead. Figures in parentheses for the parkinsoniati group indicate median manual tracking lag on the same condition.
performance is thus strikingly predictive and the difference compared to normal subjects very small, certainly much smaller than in manual tracking. The performance of the parkinsonian patients on the two noise tracks is also a good deal better than their equivalent manual tracking performance (see Table 2), even though they are not quite as good at keeping up with the target as are the normal subjects. The difference in lag between the fast and slow noise trials in both groups of subjects, and between the groups in both conditions, shows also that the zero o r negative lag on the regular waveforms is not an artifact of the sampling or cross-correlation procedure. That is, it is unlikely that a lag of 40 msec or less for the regular tracks is hidden by the sampling frequency of 80 msec. In all cases, therefore, it appears that the ability of parkinsonian patients to control their eye movements is essentially normal in terms of both reaction time and the ability to control movement predictively. Discussion Three conclusions may be drawn from these results. First, the manual tracking deficit shown by patients suffering from Parkinson disease is not due to faulty e y e movement control. These 4 patients, who all showed marked phase lags in the manual pursuit task, were able to track the same target movements presented on the same display quite adequately with their eyes. Also, in contrast to their performance on the manual task, they showed no particular difficulty in tracking the higher frequency, faster moving target tracks with their eyes. It seems unlikely, therefore, that the sensory (visual) input to the sensorimotor control system is disturbed in this disease. 66 Annals of N e u r o l o g y Vol 4 No 1 July 1978
Second, most of the 8 subjects, patients and normal persons alike, showed a negative overall time lag for eye tracking. This confirms results such as those of Drischel[21, which have indicated that smooth eye movements can actually precede predictable target movements. This confirmation is independent of the now suspect derivation of a phase ande. The smooth e y e tracking control system does seem to include a predictor, notwithstanding the objections raised by Fender [3] to previous evidence of prediction. Finally, a discrepancy exists between this evidence of fairly normal predictive control in eye tracking by parkinsonian patients and their inability, found by Flowers [ 5 ] , to show normal predictive control in tracking the same target motion manually. This implies that either (1) normal brains have at least two separate predictors which produce signals modelling probable target motion, one for voluntary control of smooth limb movements and the other for oculomotor pursuit, while in Parkinson disease the former, but not the latter, is impaired; o r ( 2 ) the same predictor generates signals which, in normal brains, go to both the smooth eye movement control system and also the system controlling smooth voluntary limb movements, whereas in Parkinson disease the output of this predictor to that part of the limb control system is impaired. Whichever of these explanations is true, parkinsonian patients evidently retain some capability for generating and updating running predictions from visual information. Supported by the Medical Research Council (UK). We are indebted to Dr R. Iangton Hewer for access to patients under his care and to Dr M. Hilary Morgan for clinical assessments and advice. We thank Mr I. Low and Mr M. Tajfel for technical and computing help.
References 1. Angel RA, Alston W, Higgins JR: Control of movement in Parkinson’s disease. Brain 93: 1-14, 1970 2. Drischel H : The frequency response of horizontal pursuit movements of the human eye and the influence of alcohol, in Asratyan EA (ed): Progress in Brain Research (Brain Reflexes). Amsterdam, Elsevier, 1968, vol 22 3 . Fender D H : Time delays in the human eye-tracking system, in Bach-y-Rita P, Collins CC, Hyde JE (eds): The Control of Eye Movements. London and New York, Academic, 1971 4. Flowers KA: Visual “closed-loop” and “open-loop” characteristics of voluntary movement in patients with parkinsonism and intention tremor. Brain 99:269-3 10, 1976 5. Flowers KA: Some frequency-response characteristics of parkinsonism on pursuit tracking. Brain 101:19-34, 1978 6. Geddes LA, Steinburg R, Wise G: Dry electrodes and holder for electro-oculography. Med Biol Eng 11:69-72, 1973 7. Home DJ de L: Sensorimotor control in parkinsonism. J Neurol Neurosurg Psychiatry 36:742-746, 1973 8. Young LR: Pursuit eye tracking movements, in Bach-y-Rita P, Collins CC, Hyde JE (eds): The Control of Eye Movements. London and New York, Academic, 197 1
