Investigative Ophthalmology & Visual Science

APRIL 1978 Vol. 17/4 Investigative Ophthalmology & Visual Science A Journal of Clinical and Basic Research

Articles

Training of voluntary torsion

Richard Balliet and Ken Nakayama

By means of a visual feedback technique, human subjects were trained to make large conjugate cyclorotary eye movements at will. The range of movement increased with training at a rate of approximately 0.8° per hour of practice, reaching 30° at the end of training. Photographs recorded the ability to make voluntary cyclofixations at any amplitude within the subject's range. Cyclotorsional pursuit was also trained, with ability increasing with greater amounts of visual feedback. In addition, torsional saccadic tracking was trained, showing a magnitude vs. peak velocity relationship similar to that seen for normal saccades. Control experiments indicate that all of these movements were voluntary, with no significant visual induction. With extended practice, large torsional movements could be made without any visual stimulation. The emergence of voluntary torsion through training demonstrates that the oculomotor system has more plasticity than has generally been assumed, reopening the issue as to whether other movements could also be trained to alleviate the symptoms of strabismus.

Key words: eye movements, torsion, saccades, slow pursuit, fixation, orthoptics, oculomotor plasticity c' yclorotations are defined as rotations the eye can undergo a cyclorotation as it about the visual axis of the eye. These rota- moves from one tertiary position of gaze to tions are considered to be reflexive, with no another, but the amount of this cyclorotation indication of voluntary control. For example, is fixed, being dictated by Listings law.11 In addition, involuntary cyclovergence has been 1 From the Smith-Kettlewell Institute of Visual Science, reported to occur during convergence, and Department of Visual Sciences, University of the reflexive cycloversions have been demon- Pacific, San Francisco, Calif. strated to occur during lateral head tilt.14> 18 This paper is based on a Ph.D. dissertation submitted by Cycloversional movements can also be in- the first author to the Department of Visual Sciences, University of the Pacific, San Francisco, Calif. 94115. duced visually by large field stimuli (25° to The work was supported in part by NEI Grant 50°). These movements include optokinetic EY01582 and the Smith-Kettlewell Eye Research cycloversion,7* 8* 13 fusional cyclovergence,9 Foundation. and tonic cycloversions.8 All of these previ- Submitted for publication July 6, 1977. ously reported cyclorotary eye movements Reprint requests: Dr. Richard Balliet, Smith-Kettlewell Institute, 2232 Webster St., San Francisco, Calif. occur either as a necessary consequence of 94115. other movements or as part of a sensorimotor

0146-0404/78/0417-0303$01.20/0 © Assoc. for Res. in Vis. and Ophthal., Inc. 303 Invest. Ophthalmol. Visual Sci. 304 Balliet and Nakayama April 1978

however, a larger than average torsional response variability in all positions of gaze (S.D. = 0.75°). We reasoned that although small, M. N.'s torsional variability was clearly perceptible and perhaps sufficient to extend further by exploratory training procedures. As our first and pilot subject we worked closely with him, up to 5 hr/week over a period of 13 months, testing many methods to develop an apparently optimal approach, Then, having found successful procedures for M. N. (see later), we trained two normal individuals who showed no oculomotor abnormalities. Subjects R. B, (one of the authors, age 30) and C. H. (age 26) were chosen because of their high motivation. These subjects had normal stereopsis and eye movements and normal or corrected normal acuity. Contact lenses were worn by C. H. Both subjects were trained and tested over a period of 2 months. Fig. 1. Photographs showing a tonic voluntary cy- Training and testing procedure. Many different clotorsion of 20°. Bars drawn in across the irises training procedures were used with M. N. We are similar to those used for photographic analysis. report only upon the final method used most suc- Vertical reference marker (present in all photo- cessfully with him and exclusively with C. H. and graphs) is seen at right. R, B. In general, the following procedures were conducted during both training and testing. During visual feedback training, the subject reflex. There are no instances where one sees was seated in a dark room with the left eye purposeful cyclofixations, or smooth and sac- always occluded. Torsional head movement was cadic torsional tracking, phenomena which fixed to within 6' of arc by a full-mouth impres- are common to voluntary horizontal or verti- sion bite plate (see Ditchbum10). Fixating a tar- cal eye rotations. get with the right eye, the subject activated a In this paper we will demonstrate that if an brief vertical electronic flash (11 x 14°), leading adequate sensory feedback loop is created, to the formation of a clear vertical afterimage. providing subjects with a very sensitive indi- By horizontally sliding the base of the bite cator of their own eye position, voluntary cy- plate mechanism, the subject then imaged the clotorsional eye movements can be trained. afterimage parallel to a vertical luminous real- Methodology and results will be presented line of equal visual angle which was viewed at which will substantiate that once this re- 100 om distance in primary position. The real- line was a partially occluded "white" 110 volt lationship is established, large voluntary cy- luminescent panel. It was bisected by a 15' clorotations can be learned and brought circular black fixation point. The subject con- under relatively fine control. trolled the luminance of the real-line over a range of 0.2 to 1.0 log foot-lamberts to prevent Tonic cycloversion visual suppression of the afterimage line by the Methods target line (or vice versa). He also controlled a Subjects. We report the successful training of This rotation angle could be observed by the sub- three subjects. Our first subject, M. N. (age 31) ject directly and/or by a digital voltmeter readout. had unilateral intermittent exotropia, his left eye The meter had an accuracy of ±4'. deviating 30°. Normally he allowed this exotropia The subject was instructed to keep his afterim- to be manifest. Otherwise, he had normal stere- age matched parallel only by cyclorotating his eye opsis, eye movements, and corrected acuity. Ini- to the real-line which he progressively rotated tially both eyes were tested for possible deviations more and more (left or right) from the vertical as from Listing's law. They showed no systematic training progressed. Usually 20' of arc increments deviation, with the mean torsion in each eye being were used. No instructions were given as to what 16 predictable by the direction of gaze We did see, type of eye movement was to be used to arrive at Volume 17 Number 4 Training of voluntary torsion 305

this position, e.g., slow pursuit, saccades, nystag- C.H. mus. Such matches which could be subjectively held for a minimum duration of 5 sec were defined as "cyclofixations." Subjects were encouraged to train equally in both incyclotorsional and excyclo- R.B. torsional directions, training usually at the rate of 1 hr/day. Photographic measurement of eye torsion. Ob- jective measures of cyclofixations were obtained from over 1,500 photographs taken with a 35 mm Nikon F-2 motor-driven camera with lenses and bellows attachments equivalent to a lens with a focal length of 400 mm. Two Vivitar 292 strobe 10 15 20 25 flash units provided illumination. High-resolution HOURS OF TRAINING color photographs on Kodachrome II film were Fig. 2. Training of voluntary tonic cycloversion for taken at 100 cm distance with a first-surface mirror subjects C. H. and R. B. Hours of training at 1 mounted at 30 cm distance in front of the subject hr/day vs. the total range (degrees) of psy- at a 45° angle, so positioned to be 5.0° off the chophysically measured torsion is plotted by the subject's primary visual axis. At this angle sys- small symbols. These are maximum tonic ranges tematic inaccuracies due to perspective distortion which can be held for a 5 sec duration. The three are minimal, not exceeding 3%. (See Appendix of large squares and the one large circle are simulta- Nakayama and Balliet16.) neous measures obtained photographically. The camera was activated by a button which could be pressed by the subject while he simulta- bination of optimal stimulus arrangements as neously held a subjectively determined cyclofixa- well as physical and mental effort. Very small tional match. Fixed to the bite bar was a vertical differences in the brightness of the afterim- photographic reference marker positioned near the outer can thus. Thus each photograph con- age and the real line led to the inhibition of tained all the information necessary to determine the dimmer by the brighter, necessitating eye torsion, eliminating errors due to possible fine control of the intensity of the real-line to camera or film misalignment. During photo- ensure the visibility of both. Attentive con- graphic analysis slides were magnifided 12 times centration was necessary to refrain from mov- in size by a Kodak Carousel projector. The image ing the afterimage with an attempted head was rear-projected onto a horizontal table screen movement. For example, during initial train- made of translucent "white" Mylar. Landmarks ing, two subjects broke bite plates, attempt- used in measurement were two identifiable ing unsuccessfully to rotate the afterimage by limbal-scleral blood vessel junctions which bor- making lateral head movements. dered a diameter of the iris and approximately bisected the pupil. The orientation of a line Slowly, however, the subjects succeeded formed between these two points relative to the in the task, learning to increase their range of stationary vertical marker was used in determining torsional eye movements with greater and objective cyclotorsion. Fig. 1 illustrates an exam- greater facility. Fig. 2 shows the remarkably ple of this technique, showing a 20° voluntary cy- steady increase of voluntary torsion as a clotorsional difference between the two photo- function of hours of practice. Small symbols graphs; the bar represents the line joining the two represent the maximum range of torsional landmarks. All photographic cyclotorsional mea- difference measured psychophysically. The surements were defined relative to average eye large symbols (square and circle) show the position when the observer was relaxed and not torsional ranges derived from photographic making voluntary eye torsions. Torsional mea- measurements; standard errors are small, surements from the same photograph were re- within the size of the symbols. At the end of peatable to within ±5' of arc. training, subjects had the ability to make cy- Results clofixations at any cyclotorsional amplitude Training of cyclofixation. The training of within the ranges specified. cyclorotary eye movements requires a com- Subject C. H. acquired a cyclofixation Invest. Ophthalmol. Visual Sci. 306 Balliet and Nakayama April 1978

N., for example, could return to previous positions after having been away for 2 months. Furthermore, after approximately 1 year without any intervening experience, both C. H. and R. B. still had the capacity to make torsional movements of about one-half their previous torsional range. Trained cyclofixation is not visually induced. As mentioned earlier, very small amounts of tonic eye torsion can be induced by the static inclination at a multiple-lined field.8 As such, these movements appear to differ from the currently described movements in two significant respects. (1) They are much smaller (less than 1.0°). (2) They are 12 10 8 6 4 2 0 2 4 6 8 10 12 involuntary. In order to determine how INCYCLO VERTICAL EXCYCLO much of the measured cyclofixation was vi- OBJECTIVE TORSION-FIXATING R.E.. DEC sually induced by the real stimulus line we Fig. 3. Conjugacy of trained torsional eye move- conducted a control experiment. After train- ments. Comparison of the torsion in each eye ing, subjects were instructed to relax and (measured photographically) as the subject is simply look for 30 sec at the real-line, which matching the afterimage tilt to designated angles was held at different fixed lateral angles in of the comparison line. Viewing is monocular 3.0° increments. This was done both with and through the right eye. Solid line represents per- without the usual vertical afterimage line. At fect conjugacy. the end of the 30 sec, photographs of the range of 26.5° (12.0° incyclo, 14.5° excyclo) in right (fixating) eye were taken. The time in- approximately 30 hr. Subject R. B. acquired terval of 30 sec was comparable to the longest a cyclofixation range of about 20° (9.0° in- time required to obtain a afterimage/real-line cyclo, 11.0° excyclo) in approximately 25 hr. match using voluntary torsion. Both subjects showed about an 0.8°/hr train- Nearly identical results were found wheth- ing rate. These torsional ranges should not be er or not the subject had an afterimage. They considered as the maximum possible; for indicate that without voluntary effort there is example, C. H. is clearly not at an asymptotic essentially no effect; eye torsion remains level at the end of 35 hr of training. Subject nearly constant for any angle of the line. Only M. N. had a final objectively measured cy- R. B. showed a slight trend, attaining a max- clofixation range of approximately 20° (10.0° imal torsional response of less than 1°. These incyclo, 10.0° excyclo). His training results results are similar to that found by Crone8 are not shown because of differing proce- and demonstrate that in our stimulus situa- dures used in training of 250 hr over a period tion optostatically induced eye torsion is very of 13 months (as described in the Subjects weak. section). Binocular aspects. The training of torsion It should be emphasized that the effort re- was an exclusively monocular task, with no quired to make these torsional movements training or measurement of the occluded left decreased dramatically over the training eye. It is of interest to know the rotational period. Near the end of training, for exam- status of this occluded eye as the trained eye ple, torsional fixations of up to 6° to 8° were undergoes large torsional changes. For ex- "virtually effortless" and could be made with ample, cyclotorsional eye movements have the same ease as with ordinary fixations. It been reported to accompany vergence,1 with also appears that the acquired capacity to extorsion increasing for increasing amounts of make these movements is long-lasting. M. convergence. This relationship raises the Volume 17 Number 4 Training of voluntary torsion 307

question as to whether our trained torsions are conjugate wheel-like cycloversions which are simply an expression of this relation or occur around the visual axes. whether they indeed represent a new and Subjective observations. During the train- separate type of eye movement. ing and testing of all voluntary cycloversions To deal with this possibility, we separately (including voluntary cycloversional slow pur- photographed the rotations of each eye for suit and saccadic tracking—see below), sub- different cyclofixations made with the right jects experienced sensations of lateral rolling. eye. We did this photographing of the right They also experienced during initial training eye as described in the Methods section occasional nausea, headache, and body fa- above and photographing the occluded left tigue. Most notable and most consistent was eye as it viewed darkness. During experi- the sensation of the body turning in the same mentation, M. N.'s left eye was exodeviated direction as the eye, with concomitant sensa- by 30°, necessitating the camera to be ap- tions of stationary objects appearing to rotate proximately positioned to photograph along in the opposite direction when environmen- the visual axis. Thus, for a cyclotorsional fixa- tal cues were available. A more quantitative tion made by the right eye as defined by a description of these and related illusions is particular angle of the afterimage match, the presented elsewhere.3 rotations of either eye could be separately measured and compared. The degree of tor- Dynamic cycloversions sional conjugacy between the right and left Voluntary conjugate eye movements can eyes was determined by plotting the torsion be divided into two distinct categories: slow of the viewing eye against the torsion of the pursuit and saccadic tracking. Because our occluded eye for different matching angles of trained cyclofixations were clearly voluntary, the real-line. Fig. 3 shows a near-perfect cor- it was of interest to see whether their respondence between the cyclorotations of dynamic characteristics were comparable to the two eyes in all three subjects. voluntary horizontal and vertical movements, Horizontal movements of the left eye were specifically in the degree to which they had determined by measuring the relationship the appropriate smooth and saccadic charac- between the corneal reflex and the pupil cen- teristics. Alternatively, it was also of interest ter.10 This position remained constant to to see whether the dynamics were compara- within ±0.5° (experimental error ± 0.2°) over ble to the nystagmic pattern normally seen the whole range of torsional movements of during reflexive cyclorotations.7' 8> l3 the right eye. This exceptional stability of Method. In order to characterize the ability to horizontal eye position of the occluded eye make voluntary dynamic cycloversions in detail, during eye torsion effectively rules out any motion picture records were taken in one session explanation of our results based on changes in each for two subjects, M. N. and C. H. This oc- vergence.* All our voluntary cyclotorsions curred after only 10 additional training sessions of 1 hr each. The subjects' ability to make large cy- *All subjects demonstrated an exceptional degree of clofixations gave them the ability to make dynamic horizontal and vertical stability of the occluded eye and a cyclorotary pursuit or tracking of a moving target near-perfect conjugacy of torsion. This was even true for within this short duration. Considered alone, the M. N.'s left eye, which was exodeviated by 30°. Consid- training of dynamic tracking was not difficult; ering the presumed primary actions of the oblique and however, it was made difficult by the bright il- 6 l9 vertical muscles in 30° abduction, ' it may be lumination required for high-resolution photogra- hypothesized that either the combination of neural in- phy. The over-all apparatus and procedure were puts to the six muscles to produce torsion is greatly al- also similar to the methods described earlier, with tered for M. N. or that the primary mechanical actions of the muscles remain invariant in an eye-based coordinate differences noted below. system. In the former hypothesis, torsion would be or- Motion picture technique. Measures of dynamic ganized at a high level in the oculomotor system, thus torsion were made from photographs taken with a requiring a "torsion command" system to rotate the eye 16 mm Bolex H16 reflex movie camera running at around the visual axis, independent of gaze direction. . either 24 frames/sec for smooth or 64 frames/sec Invest. Ophthalmol. Visual Sci. 308 Balliet and Nakayama April 1978

C.H.

TIME. SECONDS TIME. SECONDS

Fig. 4. Photographic analysis (24 frames/sec) of voluntary cycloversional slow pursuit with complete visual feedback (rotating line and afterimage visible). Solid line represents a tracing moving at the correct velocity of 1.67sec. Numbers indicate order of tracing. for saccadic testing. A 150 mm P. Angenieux zoom line against a bright white field and the afterimage lens and 30 mm of extension tubes were used. was viewed as negative afterimage against this Movies of the right eye were taken at 30 cm dis- field. Instead of a single flash, numerous consecu- tance through a first-surface mirror in front of the tive flashes were required, this being necessary to subject at a 45° angle, placed 5.0° off the subject's obtain sufficient contrast for the afterimage. Rota- visual axis. Two-Color Tran 650 watt flood lights at tions of the target line were produced automati- 100 cm distance provided the required constant cally by driving the servomotor by means of a illumination. During analysis each frame was in- waveform generator, either at a constant velocity dividually measured for cyclotorsion as well as or in steps. horizontal and vertical movements. Over 10,000 frames of film were analyzed. A 16 mm L.W. Results Single-Frame Motion Picture Analyzer rear-projected a 10 times magnified image onto a horizon- Voluntary pursuit. In order to examine the tal table screen. Torsional movements of the eye characteristics of cyclorotary slow pursuit eye were measured with the use of the limbal termi- movements, the real-line was made to rotate nation points of the two radial iris markings. The at a constant angular velocity. Initially, ob- two termination points were chosen so that a line servers were asked to make their afterimage joining them would pass very close to the pupil "follow" the rotating real-line, reporting on center. The orientation of this line relative to a their slow pursuit ability by noting the stationary vertical marker was used in determining smoothness of the rotation of the afterimage. objective cyclotorsion. Measurement error was Some smooth pursuit was reported at the ±10' of arc. outset with perceptible improvement occur- Stimulus arrangements. Because of the large ring over several practice sessions. Curi- amount of illumination required to obtain high- ously, there was a relatively narrow range of resolution motion pictures, some stimulus altera- tions were required to ensure the simultaneous velocities over which smooth pursuit was ap- visibility of the real and afterimage line. Instead of parent, between 1° to 2°/sec. Outside this a light target and afterimage being viewed against range the afterimage could be made to move a dark background, the contrast of the over-all only in steps, indicating the preponderance array was reversed. The target was a rotatable dark of saccades. It is of interest to note that the Volume 17 Number 4 Training of voluntary torsion 309

TIME. SECONDS TIME. SECONDS

Fig. 5. Photographic analysis (24 frames/sec) of voluntary cycloversional slow pursuit with partial visual feedback (afterimage visible, no rotating line). Numbers indicate order of tracing.

M.N C.H

TIME. SECONDS TIME, SECONDS Fig. 6. Photographic analysis (24 frames/sec) ot voluntary cycloversional slow pursuit with no rotary visual feedback (neither afterimage or real target line is visible). Numbers indicate order of tracing. optimal velocity for optokinetic following of a afterimage and the target line, (2) partial vi- rotating stimulus is also in this slow range.13 sual feedback with the afterimage only, and For a photographic analysis of pursuit, a (3) no rotary or tilted visual feedback—this constant 1.6°/sec angular velocity was used consisting of a single black fixation dot. Fig. 4 over a total rotation of 16°, rotating clockwise to 6 show time records of the torsional eye from -8° to +8° for M. N. and the reverse for movements, obtained photographically, for C. H. the three conditions for each subject. The Three different conditions were used: (1) sloping solid line in Fig. 4 corresponds to a complete visual feedback utilizing both the movement at the velocity of the target line. Invest. Ophthalmol. Visual Sci. 310 Balliet and Nakayama April 1978

—r~ .5 1.0 1.0 TIME. SECONDS TIME. SECONDS Fig. 7. Photographic analysis (64 frames/sec) of voluntary cycloversional saccadic tracking: 4° step stimulus. Arrow represents "correct" destination of the saccadic eye movement. Numbers indicate order of tracing.

Traces are numbered in the order taken. ducing any pursuit, producing only a series of With complete visual feedback both subjects saccades. As a further point of comparison demonstrated a moderate ability to make between trained torsional pursuit and ordi- voluntary cyclotorsional slow pursuit move- nary pursuit, we had our subjects attempt to ments. Particularly accurate are trace 2 for C. make cyclotorsional pursuit movements H. and trace 3 for M. N., where movements without any stimulus other than a single fixa- at the actual target velocity were common tion dot. Under these conditions, movements (see Fig. 4). Intermixed with these pursuit consisted mostly of a series of steplike sac- movements are small corrective saccades. cades, with M. N. having peculiar overshoots With partial visual feedback (Fig. 5), sub- (see below) but with little evidence of smooth jects had only a single afterimage line and a eye movements (Fig. 6). This inability to black fixation dot, and they were required to make torsional pursuit in the absence of per- "imagine" a line moving at the same velocity ceived torsional movement provides yet an- as before (1.6°/sec). Under this condition, other comparison to the known characteris- subjects had more difficulty, showing less tics of voluntary horizontal and vertical pur- time in performing smooth movements and suit movements. showing more saccades instead. It should be It should be emphasized that even though noted, however, that C. H. was able to make both subjects showed little evidence of some long, slow pursuit movements in the smooth pursuit without a moving target, they absence of a moving line using only the af- could rotate their eyes over a large range, up terimage. Trace 5 begins with a movement to 24° (Fig. 6, M. N.), without any visual having a velocity of 1.6°/sec. Her other traces stimulus. This underscores the fact that the 6 and 7 have episodes with a velocity of about torsional eye movements are not visually in- 3°/sec. M. N. also shows some slow pursuit of duced. Subjects need no stimulus at all to about 3°/sec in his trace 7. Such findings voluntarily rotate their eyes. might be considered analogous to experi- Voluntary saccades. During the training ments showing smooth horizontal tracking and testing of voluntary cyclorotary saccades, with afterimages alone.12 the target line was moved in steps of 4°, 8°, Ordinarily, horizontal and vertical pursuit and 16°. M. N. used a clockwise rotation from eye movements cannot be made in the ab- 0° to +4° or +8°, or -8° to +8°, whereas the sence of perceived movement. Except under movements were in the opposite direction for special circumstances,22 a stationary observer C. H. Subjects were instructed to track the viewing a static scene is not capable of pro- vertical line using their afterimage. Motion Volume 17 Number 4 Training of voluntary torsion 311

M.N,

C.H.

.5 1.0 1.0 1.5 TIME. SECONDS TIME. SECONDS Fig. 8. Photographic analysis (64 frames/sec) of voluntary cycloversional saccadic tracking: 16° step stimulus. Arrow represents "correct" destination of the saccadic eye movement. Numbers indicate order of tracing. picture speed was increased to 64 frames/sec, movement. The 8° and 16° saccadic tasks maximizing the temporal resolution that were judged to be far easier. These results could be obtained photographically. Figs. 7 suggest a possible inability to accurately con- and 8 show sample records of the 4° and 16° trol cyclotorsional saccadic trackings of less saccades. than 4°. It is apparent that voluntary saccadic cyclo- C.H. demonstrated normal glissadic over- torsional eye movements exceeding 12° are shoots when compared to normal horizontal possible (Fig. 8, C. H.). From the records it saccades,23 and the saccadic tracking of M. N. can be seen that the voluntary saccades can often showed large dynamic overshoots of up be single, sequential, or nearly overlapping. to 2.0°, following the normal velocity-ampli- Accuracy of saccades, however, appears less tude relationship for saccades2 (see Fig. 8). than for ordinary saccades, showing errors as Although M. N. is a unilateral intermittent great as 4°. The records from M. N. were exotrope, it should be remembered that more irregular, some showing frequent drifts these traces are for his nonstrabismic eye. (0.2° to 1.0°/sec) with a magnitude of 1.0° or Also, we found no unusual horizontal eye less. movements for either of his eyes when mea- During the 4° voluntary cycloversional sac- sured by an infrared photoelectric technique, cadic tests (Fig. 7), either the subjects greatly accurate to 20'. We have one possible expla- 4 overshot the correct stimulus position or, in nation. Bahill and Stark have found that the the case of M. N., the appropriate eye occurrence and magnitude of dynamic over- positions were achieved by making small shoots increases with fatigue. M.N. had refixation drifts and/or small saccades. Both ported feeling very tired for several days subjects reported that an appropriate 4° sac- prior to and through the day of these tests. cade was virtually impossible to make and In addition, both subjects reported large had to be coaxed to even attempt such a amounts of saccadic suppression. Afterimages Invest. Ophthalmol. Visual Sci. 312 Balliet and Nakayama April 1978

1000:

1.0 10 1.0 10 MAGNITUDE, DEC MAGNITUDE. DEG Fig. 9. Comparison of voluntary cycloversional and horizontal saccades in terms of magnitude vs. peak velocity. Dots represent this analysis for voluntary cyclotorsional saccades. The regression line represents a best fit for a comparable analysis of horizontal saccades obtained by Bahill et al.3 with bars indicating the range of their data. disappeared during and somewhat after vol- TRAINED DYNAMIC CYCLOVERSION IS NOT VISU- untary saccades greater than about 4° or 5°, ALLY INDUCED. Rotating stimuli can visually thus diminishing retinal feedback which re- induce cyclotorsion.7'8> 13 As a further control lated eye position to the subject. Unusual to clarify the voluntary nature of our cyclo- oculomotor control may have occurred be- torsional slow pursuit and saccadic tracking, cause of these two factors. we had subjects view the rotating real-line as Systematic analysis indicates that horizon- in all previous test situations, instructing the tal and vertical saccades are highly stereo- subjects to relax and simply observe the typed. This is best understood by considering rotating stimulus (either 1.6°/sec pursuit the well-documented magnitude vs. peak ve- ramp; 4° or 16° saccadic step). This was done locity function for normal saccades,3' 24 a re- for both observers, with and without the af- lationship which emphasizes the fact that terimage line. Under all conditions no trace horizontal and vertical saccades do not come of systematic nystagmus or any movements in all sizes and velocities but that for any were found. No movements exceeded 1°. given size there is a characteristic velocity and duration. The dots in Fig. 9 show peak Discussion velocities of torsional saccades as a function of Voluntary nature of the trained cyclover- amplitude. The velocities have been multi- sions. Reflexive cyclotorsions have been plied by a factor of 1.43, compensating for known to exist for over 100 years. What is the fact that our frame rate of 64 frames/ new* is the existence of a cyclotorsional eye sec is effectively limited in its high-frequen- movement which fails to fall into any class of cy response to 32 Hz. Such a limitation reflexive movements, showing instead char- can be expected to reduce saccadic peak ve- acteristics which are decidedly voluntary. locity by 30%,3 thereby requiring a multipli- Two sets of observations should further em- cative factor of 1.43 to obtain the best esti- phasize this point. First is the finding that mate of the true peak velocity. Comparative neither static tilt nor rotary motion is data from horizontal saccades are plotted sufficient to visually induce our reported cy- as a regression line with error bars, these data being derived from Bahill et al.,3 utilizing a high-frequency recording sys- * Although not reported as such, we note that Noji,17 in a tem (1,000 Hz cutofi). What stands out supposed study of induced cycloversions, used afterim- is the remarkable congruence of the two ages paired with real-lines, reporting increasing "induction" up to 15° only after 5 successive days of "exercise". sets of data, suggesting a common or analo- Rather than induction, we consider this result to be the gous origin for both types of saccades. probable training of voluntary torsion. Volume 17 Number 4 Training of voluntary torsion 313

clorotary movements. Second and equally Clinical applications. The success of our important is the fact that large cyclorotary training procedure raises the issue as to eye movements, up to 26.5°, can be made in whether other types of oculomotor organi- the absence of any tilted or rotary stimulus zations are also possible and, if this is so, whatever (Fig. 6). Vision alone cannot drive whether or not such reorganization would the large torsional movements; they require have value in the treatment of eye movement voluntary effort. disorders. The sensory feedback technique Plasticity and limits of plasticity in the ocu- which we used to acquire this control is in lomotor system. The emergence of voluntary principle similar to other methods which eye torsion through training indicates an un- have used sensory feedback to gain control suspected plasticity of the human oculomotor over responses which are thought to be in- system. For example, under ordinary cir- voluntary; these include heart rate, visceral cumstances voluntary fixationsar e thought to function, peripheral blood flow, and the be governed by Listing's law such that, in permanent return of muscle control in pa- terms of its resting position, the eye rotates tients with various manifestations of dis- with only two degrees of freedom, with the turbed neuromotor control. (For a review of third degree of freedom (torsion) determined recent developments in sensory feedback, 21 by the horizontal and vertical direction of see Schwartz and Beatty. ) One of the key gaze. The nonmechanical nature of this one features for successful training of these in- degree of rotational constraint implies the voluntary responses is the availability of a existence of a supranuclear neural network, high-gain sensory feedback signal indicating assuring a fixed and remarkably systematic minute changes in direction of the desired degree of eye torsion for each gaze di- motor response. With this information, the rection.15' 24 Clearly, the existence of volun- subject gradually gains access to previously tary torsion demonstrated in this paper indi- inaccessible visceral or motor control mech- cates that such a seemingly "hard-wired" anisms. network can be modified or overridden. Considering this recent work on the train- It should be noted that despite the large ing of involuntary responses and our own degree of oculomotor plasticity demonstrated work on the training of voluntary torsion, it is by our training procedures, there are some of interest to determine whether other aspects of the newly trained movements sterotyped oculomotor routines are also mal- which are obviously familiar and clearly not leable. For example, both Hering's law and new. For example, the conditions under Listing's law are well-known rules of the which one finds slow pursuit are similar, both oculomotor system, reflecting the fact that for torsional as well as for normal pursuit, the pattern of innervation to the muscles is with greater amounts of slow pursuit appar- very orderly and seemingly permanent. ent with greater amounts of movement in- Normally, this neural permanence is thought formation. For saccades, the magnitude vs. to be functional, enabling the muscles to di- peak velocity relation affords the most sys- rect the eyes in a precise and repeatable tematic comparison. This relation is a con- manner. In strabismus, and especially with tinuous one for normal saccades25 and is ap- incomitant strabismus, however, these laws, plicable to the fast phase of nystagmus as instead of being helpful, are harmful. What is well.20 What is of interest is the fact that our required in the case of a muscle weakness is a newly trained movements should also follow new innervation pattern; the old pattern this rule very closely. We suggest that this leads to symptoms. What is needed is a similarity is not coincidental but reflects the modification of neural patterns in the appro- fact that all saccades, including torsional sac- priate direction, i.e., a functional violation of cades, are mediated by a primitive and rela- Hering's law and Listing's law. The question tively "hard-wired" neural network, original- is whether such fixedpattern s of coordination ly designed for the quick phase of nystagmus. can be reorganized. Invest. Ophthalmol. Visual Sci. 314 Balliet and Nakayama April 1978

Our results suggest that the limits of volun- 11. Helmholtz, H.: Treatise on Physiological Optics, tary oculomotor control are as yet unknown. translated from Third German edition, Southall, J. P. C, editor, New York, 1910, Dover Publications. Since effortless voluntary torsion can be 12. Heywood, S., and Churcher, J.: Eye movements maintained over a significant range (6° to 8° and the afterimage. I. Tracking the afterimage, Vi- minimum), it is possible that other eye sion Res. 11:1163, 1971. movements can be trained as well. For 13. Kertesz, A. E., and Jones, R. W.: The effect of angu- example, voluntary cyclovergence might be lar velocity of stimulus on human torsional eye movements, Vision Res. 9:995, 1969. applied to patients with superior oblique 14. Miller, E. F.: Counterrolling of the human eyes palsies, with the limits of such therapy de- produced by head tilt with respect to gravity, U.S. termined by the specific feedback training Naval School of Aviation Medicine Research Report procedure, patient motivation, and time. No. 75, 1962. 15. Nakayama, K.: Coordination of extraocular muscles. We thank Drs. Lawrence Stark and Brian Brown for In Lennerstrand, G., and Bach-y-Rita, P., editors: helpful comments. We especially thank subjects Basic Mechanisms of Ocular Motility and Their Matthew Noble and Candus Hedberg. Clinical Implications, Oxford, England, 1975, Per- gamon Press, pp. 193-207. REFERENCES 16. Nakayama, K., and Balliet, R.: Listings Law, eye 1. Allen, M. J.: Dependence of cyclophoria on con- position sense, and the perception of the vertical, vergence, elevation and the system of axes, Am. J. Vision Res. 17:453, 1977. Optom. & Arch. Am. Acad. Optom. 31:297, 1954. 17. Noji, R.: Uber optisch Erzwungene parallele 2. Bahill, T. A., Clark, M. R., and Stark, L.: Dynamic Rollungen der Augen, Arch. Ophthalmol. 122:562, overshoot in saccadic eye movements is caused by 1929. neurological control signal reversals, Exp. Neurol. 18. Petrov, A. P., and Zenkin, G. M.: Torsional eye 48:107, 1975. movements and constancy of the visual field, Vision 3. Bahill, T. A., Clark, M. R., and Stark, L.: The main Res. 13:2465, 1973. sequence, a tool for studying eye movements, Math. 19. Robinson, D. A.: A quantitative analysis of extraocu- Biosci. 24:191, 1975. lar muscle coordination and squint, INVEST. OPH- 4. Bahill, T. A., and Stark, L.: Overlapping saccades THALMOL. 14:801, 1975. and glissades are produced by fatigue in the saccadic 20. Ron, S., Robinson, D. A., and Skavenski, A. A.: eye movement system, Exp. Neurol. 48:95, 1975. Saccades and the quick phase of nystagmus, Vision 5. Balliet, R., and Nakayama, K.: An illusion of self Res. 12:2015, 1972. rotation during voluntary torsion, Science (submit- 21. Schwartz, G. E., and Beatty, J.: Biofeedback— ted, 1978). Theory and Research, New York, 1977, Academic 6. Boeder, P.: The cooperative action of extraocular Press, Inc. muscles, Br. J. Ophthalmol. 46:397, 1962. 22. Steinbach, M. J.: Pursuing the perceptual rather 7. Brecher, G. A.: Die optokinetische Auslosung von than the retinal stimulus, Vision Res. 16:1371, 1976. Augenrollung und rotatoischem Nystagmus, Pflue- 23. Weber, R. B., and Daroff, R. B.: Corrective gers Arch. Ges. Physiol. 234:13, 1934. movements following refixation saccades: type and 8. Crone, R. A.: Optically induced eye torsion. II. Op- control system analysis, Vision Res. 12:467, 1972. tostatic and optokinetic cycloversion, Albrecht von 24. Westheimer, G.: Saccadic eye movements. In Zik- Graefes Arch. Klin. Exp. Ophthalmol. 196:1, 1975. mund, V., Editor: The Oculomotor System and 9. Crone, R. A., and Everhard-Halm, Y.: Optically in- Brain Function, Bratislava, Czechoslovakia, 1973. duced eye torsion. I. Fusional cyclovergence. Al- Publishing House of the Slovak Academy of Science. brecht von Graefes Arch. Klin. Exp. Ophthalmol. 25. Zuber, B. L., and Stark, L.: Micro saccades and the 195:231, 1975. velocity amplitude relationship for saccadic eye 10. Ditchburn, R. W.: Eye Movements and Visual Per- movement, Science 150:1459, 1965. ception, Oxford, England, 1973, Clarendon Press.