Human hlo\cmrnt Scwnce I2 (IYY)i) ;5:-?6-I North Holland

The control of discrete and reciprocal target-aiming responses: Evidence for the exploitation of mechanics ”

Introduction

Recently. Chamberlm and Mng~ll (1989) reported that ‘I movement toward a single target has n shorter movement time I e IS ewcuted

t,l\tt‘r. th,w the Fame movement \\hen It IS tollwed t-q ;I wzond rno\ement twxd d wcond taget Chaniberlrn and M;lgrll (IYSY)

rnterpreted thl\ otwtot,qtv rrtirwltrzge phenomenon a e\wlrnce tor on-lme programm~np 111 \t hlch some ot the progr,m~mtng fog the

second movement occurs durmg ewcutron of the trrst movement Ftschman and Reese (1903) proposed an altern,ltn’r e~pl,lnatwn tar

the one-target ad\,mt,lg? The\. argued that mcnement time to a tttrt target (hIT1) Is lengthened because the addltmndl requirement ot ewcutlng ‘I wcond movement plxes cowtranth on the ewcutlon ot the tlrbt movement Sprc~trc;ill~. thti) rewxwd that M hen cublects ‘Ire tnlng to mo\tz through ‘1 multr-segment responw as raprdly AS pow- Me. the! must conwan the Irmb (or shluc) ‘IF rt completes the Inrtlnl wgment at the tlrbt target III order to moothly aid qu~chly ewcute the nelt mwement wgment (p 133) kxordtng to Fwhman and Rtwe f lYY7). thl\ ,iddrtlon,ll denw~d for control could be entrrel! ~~rt~progr animed

Flschmdn ad Ree\e I lYY2) tested the on-ltne progr,~mmtng h\,- puthesls by dtxgntng n procedure which mJwmlzed the opportunlo

tar complete programming pt Ior to response rnrtl,ttlon The) xcom- pltbhed this b> ;Illo\rrn_g subyxts dn unlmirted amount of respone pl,mnmg ttmt: betore response Inltldtlon and by srrnpllt\lng the second movement Thus. the\ ,lrhsd subwt5 to t&e rhelr time ,ind plan the entire respone before Inltl,ltrng It. the response D;IS to \trlhe the tlr\t target and then slrnpl), lrtt the shlus ott and rno\e rt c)\er the xcond target \rtthout rtrrhlny It E\en though this procedure mlnlmrzed the need tar on-line progr:~mmrng the rewltb shw’ed ‘1 robust one-tLllrget ahCmtagr Flschmnn aid Rer\e (lYY2I Interpreted this tlndlng ,I\

r~ltlence that the longer MT1 rn t\fo-element responws I\ prob,ibll not due to on-line progranmlng Insteal the) Lnored ,I wnwalnt conceptron M hlch clarms that subtectb tmng ‘1 hio-ekment ta-get

amIng re~p~~nrs adopt ;I crtateg> ot ~e~tr~unlng the limb <I\ It up- proaches the tlrst tltrget

We agree filth Flwhrmn md Rtmr that the experlmentnl e~ldencr

stkxm to ta\or &i constraint csplamtion mer dn on-lrntz progrmmnin~

r\plm,ltwn Hwe\er Flcchrnm ;Ind Reoe \ comtr,ilnt explancttwn 14

not cornpletelv mlsblng It IS e\prrrsed In cery gent’rdl r‘lther ~c~gut‘

term ,ind does not ~peclflc‘all~ dddresr the questIon, ot N~/z\ cmf Irm

wbJect3 ccmtrain their first mwerncnt rn a trio-element strrhlng

xtlon The gwl nt thrs study \ias to shed some Irght on thrs ISSW

Many studies (Adam 1992. Te;lsdale 1087. Teasdale ,md Schmidt

1991. Waters and Strlch. 1981. Zelaznlk et aI lc)86) hae pointed to

the Importance of the 7t77pocr trifle f/w rnrger a\ an essentlA contrlbu-

tor to movement deceleratwn That IS. at the tak ~rnpows some

Impact constramts. part ot the deceleration ot the mo\srnent rnq be

xcompllshed pnss~vsly or mechanlcall) t-q txget rmpxt thl\ IS slmph

a consequence ot prlnclplec of Ne\vtonl;ln mechanics (TeardAe and

Schmidt 1991) For Instance. Te,lsd;lle ( 1087. cited 111 Tensdale ,jnd

Schmidt 1991) reported that the pe,lk Impact force ot td5t clngle

aiming movements (mean movement duration 1hO ms) b;ls 205”~

larger than that ot slower movements (mean rno\‘ement duration 2-M

ms) This recult strongly suggests th,lt pawve Impxt tortes contribute

to movement decelerntlon

The Insight that the Impxt \ilth ;I target surtxe c;111 pro\lde

movement decelerdtron might help eMorate the constr,llnt s\pl,m,l-

tlon ot the one-target ad\ant,lge That I$. the trrst mwwnent rn a

wo-element response rn,l! rely to a lesser twent on paswe deceler,l-

tlon through txget Impact th,m d one-element movement Thl\ mrght

he $0 because hitting target 1 \tlth ‘1 large Impact may hinder quick

rel?asr trom this target and theretore delq Inltldtlon ,lnd cmooth

ewcutlon ot the cecond movement When t,lced \\lth J ti\o-t,lrgst

response. theretore. sulqsctc might choose to progr,im mixfement 1

such th,lt It I\ charactcrlzed tx ‘1 rel,ltlvel> Ixge portlon ot uct1lv’

deceleration In order to ensure that Impxt torte ;Ind Impxt time

mirth target 1 ~111 not Intertere Mlth qu~ch ,tnd wwoth Inltl,ltlon ot

mo\femsnt 3 For the same re<tson. wblects mai choose to progr,lm

smaller peal\ \elocltw tar the brst nwement In d hio-element response than tar a one-element response

In sum xcordlng to the Impxt con,tl,llnt e\pl,ln,ltlon. the one-

target ,idLnnt‘lge results tram ‘1 motor control str,~teg! In \ihlch

one-target movements xe characterized tg rel,~tnel> shorter decelera-

tion phase ;Ind larger peak velocltles th,tn the colre5pondlng moie-

ment in d trio-target response

In this stud\ ue tested the Impxt conw;llnt h!pothesls ot the

one-target ad\ant,ye t-q studying the hlnematlcs of one-element ,lnd

t\\o-element target nlmlng rerponws Hoiie\‘t:r. Inste,ld ot employing

target srnhg mwements \LLZ asked sublects to m;tke ~l~l~rzg mo\e-

ments tward targets orw r/w wrtc7ccP of 17 cirg7rr~tv Also we dwgned the hto-element response such that movement 1 M~S a right-lett

movement. and movement 2 a lett-right mwement. that 15. (I re\ers;ll

of the tlrst movement Thus. the one-element movement HIS a d~s-

crew. rrght-left moi’ement and the t\\o-element maement a recrpro-

cdl. &-left. Iett-right responw

The ratlon;lle tar employmg drscrete and teclprocA slldlng mo\w

mats \\a that It created the poss~b~l~~ to stud\ the control ot

one-element and hfo-element movements in the &wnce ,ind pro-

encr ot target Impact constrnlnts That IS. 4rdlng movementr MZ

rl~twzsel~~t~r call kx- ~ctll’e control ot the \i hole deceleration phase since

there IS no opportunity for pestle deceleration through target Impact

Hcwee\vr. by Introducmg n mechanIcal stop. I e A btooden barrier ,tt

txget 1. subjects could. at leait pxth. under thew circumsLlnces

nchws deceleration ot the mwement pawvely or mechanIcally lx

w-get Impact

The ~mpxt conwant hypothew ot the one-target ;id\ant,lge pre-

dicts that. for sliding ma’ements wrth a mech,lnlcA stop ,~t target I.

MT1 IS shorter for the one-element (I e discrete) movement thm for

the hto-element (I e reclpiocA) mw’ement Moreover. these shorter

movement times die xcomp~imed by larger pe,ih velocitie and shorter

decelerntlon phases For slldlng mwements \ilthout mechdnvxl stop

dt target 1. the Impact constrant hypotherls would predict no dotter-

ences III MTI. pedk ~eloc~tv. and duration ot the deceleration phase

toI discrete and reaproc;ll mo\emrnts since both type ot movements

rely completely on actne. deceleratne control

Since target SIX ha been sho\fn to Influence mwtmenn time, peak

velocrty and the rel,ltne duration ot the acceler;ltlon dnd deceleration

phases (Adam 1992 hlacKenzle et <II lY87. Mnrtemuk et aI 1987

hIlIner and IJ~Z lYYO1 \fr: JIW rnanlpuldted target SIZC m this study h!,

employing 5niCill and large targets

hlet hod

NIneteen students (1-I female and 5 malet) ot the Unr\el\In ot

Lunburg partlclpated In this study (mean ‘lge. 22 2 years. range 20 to

31) The) \iere nil rght-hnnded nnd volunteered to partlclpdte None

ot them had any e\perlence \ilth the eupermxntal task or the loge

behmd the study

A 61 x 91 cm X-Y dlgltlzmg tablet (Scrlptel Corporation), mounted

on a 80 cm high table. was used In conynctmn i\lth a MS-DOS AT computer to record time-X data paws Sampling rate ~a?s 135 Hz. and

spatial accuracy ot the dygtlzmg tablet was set at 0 1 mm

T\\o target sheets ivere constructed. each consisting of hvo equally

sized circular targets horlzontallly separated by 10 cm The diameter of the two targets was either 3 mm or 24 mm Target sheets uere placed on top of the dlgltlzer and coxred by a pvxe of clear glas Subjects were asked to execute one-element (I e discrete) and two-

element (reciprocal) target almmg responses The discrete amilng

response required subjects to moie the stylus from the right target to

the left target The reciprocal aiming response required subjects to

move trom the right target to the lett target and then back to the right

target SubJects had to produce the amimg movements either In the

absence or presence ot a mechanical stop This mechanlcal stop

conswed ot a \\ooden barrier idImensIons. 32 x 9 x 9 cm. Height 606

gr) covering the left side of the left target Subjects had to keep this

wooden barrier m place by pressrng It l\lth their left hand In total.

three Independent variables \\ere orthogonally combined - Type of

Response (dlscrete/reclprocal). Target Size (3/2-l mm). and Mechan-

real Stop (wlth/wthout) - resulting In 8 movement conditions

Subjects stood taclng a table on which the X-Y dlgltlzlng tablet was

mounted They L\ere Instructed to powon themsehes such that their

body midline wac biased toward the center ot the left target They

were ashed to hold a stylus m a pen-grip fashion and to slide the stylus

smoothly and as quickly as possible toward the target(c) It was

emphasized that the) had to stop (for discrete movements) or to

reverse (for reciprocal movements) the movement wthln the bound-

aries of the left target In order to ensure optlmal response prepara-

tion. response rnitratron \%as under subJect control Subjects per-

Movements here dncil>zed otf-line using n nomnterxti~e computei program The follw! ing dependent mesuw \\ere cnlcul,~ted tar both mwenient 1 and 2 nio\ement time. pr,A ielocrty. ncceletdtion time (time tram initi‘ition ot a mo\sment to that ot peak ~~IocIQ). deceler-

dtwn time (time tram pe,lk \eloat> to zero \eloclt> ‘It the end ot the

movement). normalized nccelerat~on (percentage ot mo~enient time spent In acceleration). awl normalized deceleration (percentage ot

mo~ernent time spent in deceleration) In dddition tar the reciproccil aiming recponw d~iell time was calculated the period ot tmw the

<peed of the stylus \{a zero during the re\erwl ot the rno\wiient on

the hrrt target These dependent \xl;lbles l\erts termed tram the

~~\er,tge ot the subyxt’s 10 tot trials

Results and Discussion

The means ot the hlTI5 tar the discrete and reclproc;il dlmlng

rwxements to\\;ird the sm,~ll tdrget(s) <IS a functwn ot the mechwlcnl

<top condltlon (\ilthout,‘\\lth) drr: depicted In tlg 1~ A 1 x 3. (me-

chwvxl stop Y opt: ot response) ivlthln-suhwct nn,llvsls ot \nrlance (,ANOF A) sho\\ed ‘1 5;lgnlflcnnt mdln eftect tar the mech,mlcal stop

condltron. Fc I IS) = 67 1 p < 0 001. Indwxtlng much Lister responws

for the condltlon \\lth mech,inlcdl stop (Jf = 323 rns) th,ln tar the

condltlon without mechCmlcnl 5top (,\I = 5-N ms) The signiticant interxtion bet\\een mech;lnml stop ,md t!pr ot response. F( 1. 18) = 4 52. !J < 0 05. Indicated that \ilth mechanic‘4 stop MT1 ot discrete

response5 (,\I = 305 msb MI?S shorter than th,lt ot rtwprocal reponses

( .\I = 31-I 111s). hut that Mlthout mechanIcA stop hIT1 ns about the

wne tar both type ot responses The wme pattern ot rewlts \\a eildent tar the \arlabla norm,ll-

lzed decelrratlon and peak \~~OCI~J (see tigs lb and lc. re\pectnel!,) Thnt IC. mwemrnts In the condrtlon wth mech,inlcA <top \iere

MOVEMENT TIME

MT (ms) 600,

550 5

500

450 I-

\\

RECIPROCAL

% DECELERATION

I RECIPROCU

400 - PEAK VELOCITY

DISCRETE ,ooP” ILnb.,

350 - s k w I , DISCRETE ’

300 - b 80

10. I ’

250i WITHOUT WITH

MECHANICAL STOP

chnracterlzed by smaller portlons of normJlze:d decelerCltwn. F( 1. 120

= 103. p < 0 001. ;Ind larger peak \!elocltles. F( 1. 18) = 136. /I < 0 001.

than movements m the condltlon iblthout mechanlcal stop (37 vs YE.

;Ind 81 is 56 cm/s. respectnely) In addition. slgnlflcant Interactions

lndlcated that nlthout mechdnlcal stop. the first movement of the

reclprocdl response \\a~ not dlfterent trom that of the discrete re-

sponse m term5 of normalized deceler;ltron and pe& \eloclty. how-

e\er. \\lth mechanIcal stop dwrete mowments had shorter normA-

lzed deceleration phases and Idrger peak \elocltles (F( 1. 18) = 6 31.

p < 0 025. and F( 1. 18) = 8 33. p < 0 01. for normalized deceler&lon

and peak \elocq. respectnely)

Thrs pattern of result IS consistent wth the predlctlons dernred

from the Impact ccmstralnt hypothesis of the one-target ;Id\,mtage

\\hrln the physlcal constraints ot the task allow paswe deceleratron of the movement through target Impact. MT1 of a discrete rno\ement IS

shorter than MT1 ot d reciprocal movement ,md. moraxer. has d

Ixger ped \eloc~t~ ,Ind a shorter normalized deceleration phase

HoweLet-. ashen the t,lsk constrants do not permit passive decekra-

twn through txget Irnpxt there IS no one-target aivantage T&en

together. thee results \trongll wggest that passne Impact force

contrlhute to mwernent deceler~tlon and that Impact with target IS dn

important control parameter of movement programming and organi-

zation

The means of the MTlr tar the discrete ,md reaprocA wnmg

mwements toward the large t,lrgt:t(~) as a tunctlon of the mechanlcxl

stop condltlon (wthout/wth) are shwn In fig 2a A 2 x 2 (mechanl-

cdl stop X t)pe of response) \tithin-subject an;ll!sis of idridnce

(ANOVA) shwed ‘1 slgnltrclmt maIn ettect tar the mrchanvxl stop

MOVEMENT TIME

MT (ms) 350 I-

250

200

l-8

+ F

RECIPROCAL

PEAK VELOCITY

J” Ic”3.l

condltlon. F( 1. 18) = 17 1. p < 0 001. lndlcatlng taster responses for

the condltlon wth mechanlcal stop (M = 218 rns) than tor the condl-

tlon wthout mechanical stop (A1 = 2-W ms) Howe\vr. the slgnlflcant

InteractIon between mechanical stop and ape of response. F( 1. 18) =

29 3. p < 0 001. quaIltIed this maln effect by lndlcatrng that the

ddiantage for the mechanical stop only materlallzed tor the discrete

response but not for the recrproc;il response That IS. MT1 ot the

reciprocal response was virtually the same for the condltlons ~lth ,md

\vlthout mechanical stop Apparently. MT1 ot reciprocal almrng re-

sponses toH;Irds large targets does not seem to be affected b) the

presence or Axence of a mechanlcal stop at txget 1

Importantly. the slgnltlcant mteractlon ;~lso Indicated that wth

mechanlcal 5top MT1 ot the discrete and the reciprocal response \iere

the same. and that wthout mechanlcJ stop MT1 of the reciprocal

response was fl7swl than that ot the discrete response (313 versus 285

ms. respectrvely) The tlrst flndrng reflects d Lulure to obtain the

one-target ad\;lntage. ithilt: the second flndlnp demonstrates a fnw

tnrger ~7~11~712~~7g:~~ This InteractIon WAS also eLIdent tor the \;lr~Ales

normalized deceleration. F( 1. 18) = 13 0. P < 0 001. and peak \eloclb.

F( 1. 18) = h 36. ~1 < 0 03 (see fig 2b and ?c. respectIveI!)

Clearly. the pattern of results tor discrete and reciprocal moie-

ments toward lc7rge targets IS quite different from the pattern ot

result\ tor movements toward snznll targets HON to e\plam these

unexpected fIndIngs” A promrslng dIrectIon tor possible elplanatlon

would be an account that postulates that a reciprocal response toi\xd

k7rg~ targets does not ha\e to stop but rather to rt’lvvse twl twtwt tlumwt~ on the first target ImportantI!. the control mechnnrsmg ot

stopping a movement and re\erslng a movement are different The

e’recutlon of a discrete movement typ~call\ require\ d three-burst

EMG pattern of agonlst-Clntagonlst-agonlst xtlvrty \\lth the last

agonist actl\lty dnmpmg possible osclllatlons In order to stop and

faate the limb (or s@lus) on the target (Enoha 198X. Hallett et al

1975 Hannatord and Stark 19X5) In a fast reversal movement. on the

other hand. the last component of this three-burst xtnlty pattern IS

usually absent (Enoka 1988) The fact that a discrete unldlrectlon;il

movement IS controlled bl J three-burst actlk~ty pattern Mhlle d

reciprocal. wersal movement IS controlled b\ a wo-burst EMG xtwlty pattern might account for the shorter MT1 ot the reclproc;ll

response touxd large target($)

For recqxwxl mwemrntc tward vmll rcrtgerr ( Z I~WZ L hwe\el.

the dlstlnctlon brt\ieeen jtoppmg and re\erslng a mwement might

not be tele\‘lnt This IS so. bsc;luse wiall target\ lequlre that both

dwrete and reclprocA moiementz ~rcy~ on the tlrst txget so that

subwts can evaluate their mwement 5 endpoint accurxy. that IS.

they ha\e to cornplus Intended dnd xhleved endpoint accurxv In

other words. reciprocal almrng mcnernentr toward small target< mai

be thought of 2s two dlscrrte mwemrntc sepx,lted by ‘1 dlstlnct

period ot stCmdstlll’ The ~n~lgses ot d\tell time on t,u-get I support5

this suggectmn dwell tlme on the small w-get \fds much longer th,in

diiell time on the large txget ( 130 iersub 15 m\. respectnel)) Accord-

rng to the llns ot re;lsonlng. reclprocJ ‘lnd dwrete movements

twxd small targets In the Jxence ot target Impact constrdlnts are

h.mctlon;lllg the wme. datd concerning hlT1. nornwllzed deceler~~tlon

and peak \eloclt) support this InterpretCltmn (see top I)

The dxence ot the one-txget ‘id\ant;lge tar Ixge targets In the

presence ot A mech~inlc~il stop mght ref-kct <I celling eftect’ Th,\t IS. ‘3

one-txpet ad\,tnt;igt: tar large w-get5 would require ‘1 dwxete re-

sponw that hit, the mechanlcA stop i\lth d Ltrger Impact than the

reclprocA responw \~ould do Hweer. since the mechanIcA stop was

CI rrlatl\ely small ;Ind Ight-iieelghted Mooden bxrler Mi 111 phc-e h\

r/w dwtr /rrt~~//ztvst~l~ (that IF. i\lth the I&t hand, SubJects might

h,l\e shied awa\ tram hrttlnp the b,irrlsr Mlth such ‘I Large Irnpxt th‘lt

iiould c‘\usz it to be dl\plxed

At this point. it should be noted that Mhlle the prewnt t,lsl\

constraints rsqurred subwx to reverse mwement dlrrctlon dfter

hrttlng txget 1. previous studies regarding the one-t,lrget ad\ant;tge

required wblects to continue the second mwement In the dlrectlon ot

the tlrst movement Possibly. the klnematlc protrle of <t trrst rtwement

tollwed b> A second mcnement In the same dIrectIon 15 dltterent tram

th,lt tollowed b! J mwement In the opposite dlrectwn This obsew,l-

tlon warrant5 cdutlon III gener;lllzlng the present tIndIng ,md Inter-

pret‘itmns to multwlement rmxements ewcuted In the sxntl dlrec-

tICIt

The lmplwxtmns ot the present result5 tar motor control theon drt:

twotold FII~;~. because the txget Impxt constr,ilnt~ dr;lmatlcall\

affected the reldtwe durdtwn ot the acceler~itlon ;Ind deceler‘itlon

ph,ws the notion ot simrnetq or bell-shaped lelocltv-time protlles IS

not suppo~ t?d This outcome IS not compatible wth the aswmptwn ot

rmnrlance ot relatwe tune. and theretore questwns the \1ab111ty ot

theories of motor control In l\hich a general motor program produces

sptw~ic movements by adJustlng d sckibie pa-ameter in the trme

dom,un (e g . Meyer et al 1982. Munh,lll et ‘11 19S5) Rather. the

results support a VKN of motor control that 15 fund;lment,llly prag-

matIc and spec~hc to the oblectlves ot the pertormer and the con-

straints ot the task (Adam 1W2. A-bib 1985. Cole and Abbs 1986.

Marteniuh et al 1087)

Second. the one-target ad\,mtdge belongs to <I gentzral class ot

phenomen;l alled cotItc’\f effrcfs movements embedded In A se-

quence are not Independent ot each other but mnq mutuAli Inf’luenct:

exh other Studies ot h‘md\irltlng. tar Instance. have shwn th,tt the

shape and tlmlng ot a letter depends on \i hdt letter precedec It and on

wh,it letter tollous It (Rosenbaum 1991) Conkit eftew A;0 hnw

been described In the stud! of bpwrltlng (Terzuolo ,md VnICmr

1080) The present research points to the Importwce of mechCmlcdl

constraints (I e Impact l\lth txget) In producing contekt ettects. a

point corrobor‘ited b! Gentner et al (1988l \vho shwed that contekt

etkcts In tqpewrltlng can be centr,illy based but also mech,mlcall)

Importantly the rele\~nce ot mechanIcal constralnt5 In the control ot

mo\‘ements has been emphasrzed b> Rosenbnum ( 1991) M ho drpued

that rel)Ing on mechanics’ cdn smipll~ the degrees-of-freedom prob-

lem (see also. BIZZI and Muss,l-I\xldl 1989)

In conclwnn The one-target ;Idvantnge m;lterlnllzes when t;lrk

constrdlntc rtqurre d mwernent 5top and Jlo\.\ tar target Impact. It 5

underlyng mechdnlsm IS the e\ploltatlon of paw\e. mech,mlcal decel-

eration through target Impact The t\\o-target advantage IS tound

ithen task constrarnts do not require a mcxement \top but ctllo~ a tart

mwemrnt re\erwl In the absence of target Imp;lct constraints. It s

underhung mechamsm IS prob;lblv d txo-burst pattern ot ;Igonlst-

ant;lgonlst actlilt\ Future rewxch should Imestlg;lte the one- and

tuo-target ‘idvantage more directly by monltorlng rmpxt tortes ,md

reysterlng EhlG xtnlty pattterns

References