466

Ann. N.Y. Acad. Sci. 1039: 466–469 (2005). © 2005 New York Academy of Sciences.doi: 10.1196/annals.1325.046

Tests of Hering- and Helmholtz-Type Models for Saccade–Vergence Interactions by Comparing Visually Guided and Memory-Guided Movements

ARUN N. KUMAR, YANNING H. HAN, KE LIAO, AND R. JOHN LEIGH

Department of Biomedical Engineering & Neurology Service, Veterans Affairs Medical Center, Case Western Reserve University, Cleveland, Ohio 44106-1702, USA

ABSTRACT: We compared the dynamic properties of memory-guided and visu-ally-guided saccade-vergence movements. For memory-guided responses, con-vergence components were slowed proportionally more than correspondingsaccadic components, compared with visually-guided responses. This result isconsistent with independent saccadic and vergence systems, and supports aHering-type model for saccade-vergence interactions.

KEYWORDS: burst neurons; disjunctive movements

Most natural shifts of the point of fixation are between targets that lie in differentdirections and at varying distances from the eyes, requiring a combination of saccadeand vergence movements. The two principal theories concerning the development ofsaccade–vergence eye movement are attributed to Hering1 and to Helmholtz.2 Animplementation of the Hering-type model by Zee and colleagues3 describes separatesaccade and vergence burst neurons, coordinated by omnipause neurons. The Helm-holtz-type model, on the other hand, developed by King and Zhou,4 proposes a setof monocular burst generators that determine the dynamic properties of both saccadeand vergence movements. There is no current consensus about which model iscorrect.

Memory-guided saccades have dynamic properties that differ from visually guid-ed saccades. Specifically, memory-guided saccades have lower peak velocities andlonger duration than visually guided saccades.5 In comparison, little is known aboutmemory-guided saccade–vergence movements. The study of the dynamics of com-bined saccade–vergence movements made toward remembered targets provides uswith an opportunity to distinguish between the two models.

In the King and Zhou model,4 burst generators are monocular, motor neurons re-ceive binocular inputs, and monocular neural integrator circuits provide an input tonear response vergence cells. These monocular burst neurons determine the dynamic

Address for correspondence: R. John Leigh, M.D. Neurology Service (127W), VeteransAffairs Medical Center, 10701 East Boulevard, Cleveland, OH 44106-1702. Voice: 216-844-3190; fax: 216-231-3461.

rjl4@case.edu

467KUMAR et al.: TESTS OF HERING- AND HELMHOLTZ-TYPE MODELS

properties of both saccades (similar discharge of bursters) and vergence (differencein discharge between bursters). The changes in the dynamics of memory-guided eyemovements (slowing of saccades or vergence) could be simulated by a decrease inthe burst generator outputs. However, a reduction in the outputs of the monocularburst generators would result in both saccade and vergence components being com-mensurately slowed. Thus, it does not seem possible for the model to slow down thesaccades and the vergence components by different amounts, because of the absenceof a separate vergence pathway. In contrast, the saccade–vergence burst neuron mod-el by Zee et al. postulates separate pathways for saccadic and vergence move-ments—separate burst generators, separate neural integrators—and the motorcommands from these two pathways are summed at the ocular motor neurons.3 Thisscheme allows for a greater degree of independence between the saccadic and ver-gence systems.

We studied four healthy normal subjects (age range 20 to 57 years), two of whomwere naive as to the purpose of the experiments. All subjects gave informed, writtenconsent and the study was approved by our Institutional Review Board. Horizontaland vertical movements of each eye were measured using the magnetic search-coiltechnique, as previously described.6 Subjects viewed a red laser spot (the “primarytarget”) that was projected from above onto a nearly horizontal plank of wood underthe control of MiniSax servocontrollers (GSI Lumonics, Inc.), at a distance from thesubject of either 23.5 cm or 51 cm on the midline. A green laser spot (the “secondarytarget”) was similarly projected onto the horizontal plank, at 30 cm or 60 cm, on ei-ther side of the midline, requiring a saccade of 15 deg. The general instruction wasto look at the primary target until it was turned off and then to look at the secondarytarget (or its remembered location). We employed two test paradigms: (1) visuallyguided saccades during fixation (GAP/FIX), and (2) memory-guided saccades dur-ing fixation (MEM/FIX). Timing of the target jumps in these paradigms was as pre-viously described.7 Instructions were given for each of two test paradigms and somepractice was allowed before search coils were inserted and data collection begun.

Vergence and version eye position signals were computed, filtered (bandwidth, 0to 50 Hz) and then differentiated, as previously described,6 to yield vergence veloc-ity and version velocity signals, respectively. The onset (or end) of the saccadicmovement was defined as the time when version velocity exceeded (or dropped be-low) 10 deg/s. The start and end of the initial vergence movement was assumed tocoincide with the start and end of the initial saccade. Responses were analyzed in-teractively, noting the peak vergence velocity, the peak version velocity, the durationand size of the vergence movement, and the duration and size of the version move-ment. A three-dimensional surface plot was generated in SigmaPlot (SPSS Inc, Chi-cago, IL), using an inverse-square smoother that computed the weighted average ofthe values at neighboring points.

FIGURE 1 compares the ratio of peak saccadic component speed to peak vergencecomponent speed during memory-guided and visually guided paradigms in our foursubjects. The mean ratio was 6.8 (SD = 2.5, n = 21) for memory-guided saccadeswith convergence movements; this was significantly higher than the mean ratio of5.1 (SD = 2.5, n = 23) for similar-sized visually guided saccades with convergencemovements (Student’s t-test, P < .05). Median ratios for saccades combined with di-vergence (nonnormal distribution) were similar (~2.9, n1 = 45, n2 = 37) for both par-adigms. Thus, memory-guided movements showed a proportionately greater

468 ANNALS NEW YORK ACADEMY OF SCIENCES

FIGURE 1. Comparison of the peak velocity ratios of saccade and vergence compo-nents during visually-guided (top) and memory-guided (bottom) movements. For diver-gence movements, the ratios for visually-guided and memory-guided movements are equal(~2.9), indicating that both saccades and divergence components are commensuratelyslowed during memory-guided movements. However, memory-guided movements showeda greater slowing of the convergence components than of the corresponding saccadic com-ponents. The ratio was ~6 to 8 for memory-guided movements and ~4 to 6 for similar-sizedvisually guided movements.

469KUMAR et al.: TESTS OF HERING- AND HELMHOLTZ-TYPE MODELS

slowing of convergence (but not divergence) compared with the saccadic componentthan similar-sized visually guided movements.

Our findings agree less with the predictions of the monocular Helmholtz-type andmore with a Hering-type model. They are consistent with an independent vergencepathway, which has separate convergence and divergence components. Nonetheless,demonstration of monocular control of saccadic eye movements on the one hand,8

and midbrain vergence burst neurons on the other,9,10 suggests that saccade–vergence generation may best be represented by a hybrid of the two models.

ACKNOWLEDGMENTS

This work was supported by NIH grant EY06717, Veterans Affairs, and the Eve-nor Armington Fund.

REFERENCES

1. HERING, E. 1977. The Theory of Binocular Vision. Plenum Press. New York.2. HELMHOLTZ, J.A. 1910. Treatise on Physiological Optics. Dover. New York.3. ZEE, D.S., E.J. FITZGIBBON & L.M. OPTICAN. 1992. Saccade-vergence interactions in

humans. J. Neurophysiol. 68: 1624–1641.4. KING, W.M. & W. ZHOU. 2002. Neural basis of disjunctive eye movements. Ann. N.Y.

Acad. Sci. 956: 273–283.5. LEIGH, R.J. & D.S. ZEE. 1999. The Neurology of Eye Movements, 3rd edn. Oxford

University Press. New York.6. KUMAR, A.N., Y. HAN, S. GARBUTT & R.J. LEIGH. 2002. Properties of anticipatory ver-

gence responses. Invest. Ophthal. Vis. Sci. 43: 2626–2632. 7. KUMAR, A.N. et al. 2003. Characteristics of a range effect for vergence movements.

Ann. N.Y. Acad. Sci. 1004: 394–398.8. ZHOU, W. & W.M. KING. 1998. Premotor commands encode monocular eye move-

ments. Nature 393: 692–695.9. MAYS, L.E. 1984. Neural control of vergence eye movements: convergence and diver-

gence neurons in the midbrain. J. Neurophysiol. 51: 1091–1108.10. MAYS, L.E., J.D. PORTER, P.D. GAMLIN & C.A. TELLO. 1986 Neural control of vergence

eye movements: neurons encoding vergence velocity. J. Neurophysiol. 56: 1007–1021.