8/9/2019 FRANZ, RAMACHANDRAN (1998). Bimanual Coupling in Amputees With Phantom Limbs

1/2natureneuroscience volume 1 no 6 october 1998 443

Bimanual coupling inamputees with phantomlimbs

Elizabeth A. Franz1 and V. S. Ramachandran2

1Department of Psychology, Uni versit y of Otago, Dunedin, Box 56, New

Zealand

2Brain and Percepti on Laboratory, Uni versity of Cali fornia, San D iego,

California 92093, USA

Corr espondence should be addressed to E.A.F (lf ranz@psy.otago.ac.nz)

People who experience phantom limbs following amputationsometimes report vivid movements in the phantom, bothspontaneous and voluntary13. Do these illusory movementsreflect the operation of neural mechanisms that govern nor-mal motor performance? By studying interference and cou-pling effects of the kind that arise during bimanualmovements in normal subjects, we show that voli tional move-ments of a phantom arm impose behavioral constraints com-parable to those evident in real movement, even when thearm has been missing for over ten years. Thus, the neuralmechanisms that generate this coupling continue to operatedespite the prolonged absence of any proprioceptive or visu-al reinforcement. If the amputation was preceded by a periph-eral nerve palsy, however, the phantom arm is reported asimmobile ( learned paralysis4). We did not observe couplingeffects in patients with immobile phantoms.

Subject F.A.s right arm was amputated justbelow the elbow following a boating accident in1984. He has experienced vi vid sensati on of themissing limb since then. Very striking is his ability

to move his phantom limb in a volit ional manner.For example, when asked to reach for an object(real or imaginary) that is placed in front of him,he will describe precisely how his phantom fingerscontact and grasp the object. Curi ously, i f t heobject is then pulled away by the experimenter, F.A.often experiences pain because the object is beingwrenched away from the fingers.

We investigated whether phantom limbmovements reflect the operation of neural mech-anisms that govern normal motor performance.We assessed phantom limb movement indirectlyby applying a standard test used to examine cou-pling effects that ari se when healthy subjects pro-duce bimanual movements. In this test, subjects

are asked to cyclically produce trajectories withboth limbs. The direction of the movements isvaried to be either the same (parallel) or differ-ent (orthogonal). Under normal movement con-ditions, the trajectories drawn by each hand areminimally affected in the parallel conditions incompari son to when either hand performs alone.In contrast, when the task requires movements inorthogonal directions, both trajectories deviatefrom their target paths57.

We capitalized on these known movement con-straints to test whether spatial coupling wouldoccur when one of the movements was producedby a subjectively movable phantom limb. F.A. pro-

scient i fic correspondence

duced bimanual movements by drawing on a digitizing tabletalong a vert ical axis with his intact l imb while moving theindex finger of his phantom limb in either a tapping (paral-lel) or twirl ing (or thogonal) motion, at a comfortable rate(Fig. 1). Standard deviation from the mean direction of l ines

computed across repeated cycles provides a precise measureof interference that cannot be predicted nor manipulated inany obvious way by subjects.

The performance of subject F.A. was indistinguishable fromthat of six control subjects who made actual movements withtwo int act li mbs (Fig. 2). Signi fi cantl y more spati al erroroccurred in the orthogonal condit ion compared to parallel orone-armed conditi ons for the phantom subject F.A. (usingtrial as a random factor, F(2,14) = 11.31, p< .001) and thecontrol group (F(2,10) = 18.42, p< .001). These results indi-cate that spatial coupling persists in subject F.A., even with-out proprioceptive input f rom the missing arm. Notably, useof fi nger movement for the task of the other (phantom) limbensures that we are not capturing aspects of coupling due toresidual peripheral input from the remaining stump of theamputated limb, given that insertion of the metacarpo-pha-langeal joint of the index finger would reside below the level ofthe amputation. The observed spati al coupling effects, there-fore, are not due to biomechanical properties associated withmovement of the phantom finger.

Do these effects depend on the subjective experience ofmovement?To explore this, we took advantage of a phenome-non that has been termed learned paralysis. Some amputeeswho do not experience phantom movement have had a pre-

Fig. 1. Schematic diagram of the experimental setup depicting the two bimanualconditions: parallel (tapping) and orthogonal (twirling). The dominant limb alwaysperformed the drawing movement. x-y coordinates of the y-axis lines wererecorded (at 78 Hz) using a digitizer tablet interfaced with a laboratory personalcomputer. Repeated cycles of movement were performed for eight ten-second tri-als per condition. Vision of the limbs was blocked by a shield appended to the wallbehind. Raw data of one representative trial from each bimanual condition repre-sent the form of interference observed in spatial coupling (inset). Informed consentwas obtained from all subjects in accord with human ethics guidelines at the V.A.Hospital in Martinez and University of California and the University of Otago.

1998 Nature America Inc. http://neurosci.nature.com

1998N

atureAmericaInc.http://neurosci.natu

re.com

8/9/2019 FRANZ, RAMACHANDRAN (1998). Bimanual Coupling in Amputees With Phantom Limbs

2/2444 nature neuroscience volume 1 no 6 october 1998

existi ng peri pheral nerve injury causing paralysis for a fewmonths prior to amputation. Learned paralysis might resultfrom a continuously experienced mismatch, prior to surgery,between the expected consequences of issued motor commandsand the resulting feedback. One such patient was D.S., whoseleft arm was amputated just below the shoulder in 1983 fol-lowing a brachial plexus avulsion in 1982. He experiences vividsensation of the fingers of his phantom limb, but no associat-ed movement. Two additional amputees without the subjec-tive experience of phantom movement were also tested by the

same method. Subject N.G., an amputee of 24 years, lost hisright arm just below the shoulder as a result of an industrialaccident. He experiences a dull pain in the tip of his stump atall times but has no sensation of phantom limb movement.Subject P.S. lost his right arm as a result of a motorcycle acci-dent seven years prior to testing, and reports vivid sensationof the limb without the experience of movement.

Amput ees without experience of phantom movementshowed no evidence of spatial coupling of the type that occurswith actual movements (Fig. 2; F(2,8) = 2.71, p> .05). Fur-ther, the original group of control subjects with two intactlimbs were tested on an imagery task, i n which they producedactual movement with one limb while vividly imaging move-ment of the other limb. As in the amputees with immovablephantoms, there was no evidence of spatial coupling of the type

that occurs wit h actual movement (F(2,4) = 2.48, p> .05).These findings were not due to differences in timing variables(duration, variance in duration) across condition or group.

Our data demonstrate that simi lar spatial coupli ng occursin normal bimanual movement and when an active limbmovement is combined with a phantom limb movement. Con-trol subjects under imagery conditions do not produce spa-tial coupling effects. Coupling, therefore, depends on thesubjective experience of movement. Although there is signif-icant overlap in the processes involved in imagery and action8,associated processes such as sending an efference copy ofmotor command signals to other brain regions may be spe-cific to movement, even the subjectively experienced move-ment of phantom limbs. Learned paralysis may account for

the lack of subjective experience of phantom movement insome amputees. A mismatch between the expected sensoryconsequences suppl ied by efference copy and the resul ti ngfeedback before amputation may cause plasticity in the neur-al system that integrates these sources of information, therebyeliminating the subjective experience of movement, and cor-responding spatial coupling effects, in those amputees who donot report moving phantoms.

Together, these findings have at least three important impli-cations. First, spati al coupling cannot be att ri buted to bio-

mechanical factors, but rather reflects spinal and/orsupraspinal mechanisms. Second, the neural processes asso-ciated with the spatial properties of movement may remainintact for as long as ten years, even when their associatedperipheral effects are absent. Third, residual spatial couplingis only manifest with the subjective experience of movement.We propose that learning-induced changes result from a mis-match between different sources of movement-related senso-ry information. Such changes seem to affect both thesubjective experience of phantom movement and the couplingmechanisms between the two l imbs.

AcknowledgementsWe thank the late Ir vi n Rock, Bob Rafal, Rich Ivr y, Steve Palmer, Bil l

Pri nzmetal and Cl if f Abraham for insights. Thanks also to Reuben Fischer

and Wil li am van der V li et for help wit h the figures.

RECEIVED 11 AUGUST: ACCEPTED 1 SEPTEMBER1998

1. Melzack, R. Sci. Am. 266, 120128 (1992).2. Sunderland, S. Nerves and Nerve Injuries(Saunders, Philadelphia, 1959).3. Ramachandran, V. S. Proc. Natl. Acad. Sci. USA 90, 1041310420 (1993).4. Ramachandran, V. S. in Neuronal Group Selecti on(eds Sporns, O., Tononi,

G. & Edelman, G.) (Academic Press, San Diego, 1994).5. Franz, E. A., Zelaznik, H . N. & McCabe, G. Acta Psychol. 77, 137151

(1991).6. Franz, E. A.Q. J. Exp. Psychol. 50, 684704 (1997).7. Franz, E. A., Eliassen, J. C., Ivry, R. B. & Gazzaniga, M. S. Psychol. Sci. 7,

306310 (1996).8. Jeannerod, M. Behav. Brain Sci. 17, 187245 (1994).

scient i fi c correspondence

Fig. 2. Spatial interference for one-armand two-arm conditions for all subjects.Spatial accuracy (zero is most accurate)and standard errors for drawing move-ments of one (intact) limb are shown foreach of four groups: actual, six subjects

with two intact limbs under actual move-ment conditions; phantom, a subjectwith vivid phantom limb movement;imagery, five subjects with two intactlimbs under imagery control conditions;and amputee, three amputees withoutphantom movement. Directional headerwas computed as the straight-line vectorconnecting the first and last data pointsof each cycle of movement within a trial.Standard deviation from the mean valueof this variable across repeated cyclescaptures the measure of spatial accuracydepicted here.

Condition

Standardd

eviationo

fspatialaccuracy

Unim anual Parallel Orthog o nal

Actual

Phantom

Imagery

Amputee

1998 Nature America Inc. http://neurosci.nature.com

1998N

atureAmericaInc.http://neurosci.natu

re.com