VOL. 70, No. 6, PART 1 DECEMBER 1968 Psychological...

VOL. 70, No. 6, PART 1 DECEMBER 1968

Psychologica l Bul le t in

MOVEMENT CONTROL IN SKILLED MOTOR PERFORMANCE1

STEVEN W. KEELE

University of Oregon

The speed and accuracy of single movements depend on several factors, suchas direction of movement, distance to the target, and accompaniment by simul-taneous movements. The relation between speed, accuracy, and distance appearsto be determined by the time required to process feedback and to make cor-rective alterations in the movement. For a repetitive series of movements, thereis some evidence suggesting that control is shifted from feedback to a motorprogram. This view receives further support from demonstrations that thereproduction of single movements may be under programmed control. Howthe study of movements may be relevant to understanding perceptual andmemory skills, as well as motor skills, is briefly mentioned.

The study of movements has great im-portance for understanding skilled perform-ance. This is, of course, most obvious for motorskills such as walking, driving, typing, partsassembly, athletics and so on. Psychologistslong have believed that movements are an im-portant component of nonmotor skills as well.For example, Watson (1924) was a proponentof the view that thought involved minutemovements of the vocal apparatus. Such aview has been discounted for many years,but recently motor theories of "thought" havebeen regaining acceptance. Hintzman (196S,1967) recently has suggested that the auditoryconfusion errors in short-term memory arenot of acoustic origin but are of articulatoryorigin. Rehearsal, according to Hintzman, in-volves subvocal articulatory movements, andit is the kinesthetic feedback from such move-ments that accounts for short-term memory.Thus, the research on kinesthetic memory re-ported later in this paper is relevant to the-ories of auditory short-term memory.

xThe preparation of this article was supported byNational Science Foundation Grant GB-3939, by theAdvanced Research Projects Agency of the Depart-ment of Defense monitored by the Air Force Officeof Scientific Research under contract F44620-67-C-0099, and by a National Institute of Mental Healthpostdoctoral fellowship. Appreciation is expressed toM. I. Posner and R. W. Pew for their comments.

A motor theory of memory need not neces-sarily involve actual movements. Visual orauditory input might be converted to motorcommands that control muscle movements,and the motor commands may be rememberedwhether or not the movement is actuallyinitiated (Hintzman himself recognized thispossibility). Sperling (1967) has proposedjust such a mechanism. He suggests that visualinformation is rapidly "read in" to a bufferstore which consists of motor programs forvocalizing the information. One could arguefurther that "seeing" the visual items is theactivation of motor programs. Evidence forthis interesting theory has been given byFestinger, Ono, Burnham, and Bamber (1967),and Liberman, Cooper, Shankweiler, andStuddert-Kennedy (1967) have proposed asimilar theory for speech perception.

The concept of a motor program may beviewed as a set of muscle commands that arestructured before a movement sequence be-gins, and that allows the entire sequence tobe carried out uninfluenced by peripheralfeedback. Evidence for motor programs hascome from recent physiological studies. Wil-son (1961), for example, has shown that thepattern of rapid wing movements in locustswas unchanged when sensory feedback fromthe wing was eliminated. Kennedy (1967) re-

387

© 1969 by the American Psychological Association, Inc.

388 STEVEN W. KEELE

ported that stimulation of single neurons inthe crayfish resulted in complex but highlyrepeatable movement patterns; stimulation ofother neurons resulted in slightly differentpatterns. It appears quite clear that the en-tire motion depended solely on the initialstimulation and not on subsequent feedback.Behavioral evidence suggesting that humansare able to learn motor programs will be re-ported later in this article.

It should be evident from these illustrationsthat a motor program is not a movement initself, but acts to control movements. It ispossible that a program may be activated insome sense without a movement actually beinginitiated. Such an occurrence, rather than anactual movement, may be important in short-term memory and in perception. Therefore,research on movements may be relevant tounderstanding perceptual and memory phe-nomena as well as to understanding skilledmotor performance. As will become obviousin the remainder of the paper, however, theemphasis on movements will be primarily to-ward understanding motor skills. First, anumber of variables that influence the speedand accuracy of movements will be reviewed.Then studies on feedback processing time willbe presented, and the implications of process-ing time for understanding the relation be-tween speed, accuracy, and distance of move-ment will be discussed. These studies will befollowed by evidence that for predictableevents, movement control is internalized.Finally, theories and data on the memory andreproduction of movements will be reviewed.

THE EXECUTION OF MOVEMENTS

The first step in understanding movementcontrol is to discover what variables influencespeed and accuracy. There has been a largeamount of research on this question, motivatedin part by the need for time and motionanalysts to determine the time that varioustasks should take (see Barnes, 1963, for ex-tensive tables of standard times). Standardtimes for various classes of movement havebeen experimentally determined by Bailey andPresgrave (19S8). Class A motions are stoppedby impact with an object and are the fastestmovements. If a movement is stopped byantagonistic muscle action (Classes B and C),

the time is increased. Class B movements(e.g., stopping the upstroke of a hammer) re-quire little precision and, therefore, are fasterthan Class C motions which require someprecision (e.g., grasping or placing an object).Further increases in time are required if thetarget is not visible until near the end of themovement—BV and CV movements.

Movement time also depends on the dis-tance, degree of precision, force, and numberof movements. For example, a Class C move-ment of 24 in. should take about 666 msec.If the end of the movement requires precisionof $ in., an additional 546 msec, is added.A weight of 8 Ibs. would add another 3 X 78msec, for starting, carrying, and stopping theweight, and a simultaneous movement with theother arm would add another 180 msec, formovements terminating 6 in. apart. Thus, thetotal movement time would be 1.626 sec.

Although the standard times developed byBailey and Presgrave are useful for timeanalysis of various jobs, the actual tasks usedare not well specified, results from severaltasks are grouped, and subjects did not makemovements as fast as possible. Consequently,their data is of lessened value for theoreticalanalysis of movements. Nevertheless, Baileyand Presgrave's work serves as an introduc-tion to the many variables that influencemovements, some of which have not beenrigorously studied by others. Later we shallreturn to see how their data compare with thatobtained in more specific experimental situa-tions.

Direction of movement. An extensive seriesof experiments on direction of movement hasbeen performed by Brogden and co-workers.Subjects used a stylus to follow a spot oflight moving at a constant velocity along astraight track varying from 0° to 360°. Withthe track in the horizontal plane, 0° representsa motion normal to the frontal plane, and armmotion is away from the body. Ninety de-grees involves movement to the left and so on.Accuracy was measured by the number oftimes the stylus touched the metal sides of thetrack.

Corrigan and Brogden (1949) studied right-handed movement in a horizontal plane at avelocity of 3 cm/sec. Accuracy (y) as a func-tion of direction of movement (x) was ap-

MOVEMENT CONTROL IN SKILLED MOTOR PERFORMANCE 389

proximated quite well by the trigonometricrelationship:

y — a — b cos2# + c sin2x [I]

where a, b, and c are constants. Movementsseparated by 180° are approximately equal inaccuracy with 135°-315° movements beingmost accurate and 45°-22S° being least ac-curate. Left-handed movements obey nearlythe same relationship except that it is invertedwith respect to right-handed movements(Briggs & Brogden, 1953). As target velocityincreases from 2.5 to 4.5 cm/sec, there is nochange in the number of errors and the trig-onometric relationship remains the same(Thompson, Voss, & Brogden, 1956). Brogden(1953) showed that with practice there is agradual shift of about 15° in the functionuntil worst performance occurs on the 60°axis (only angles from 0° through 150° wereused). When the plane of movement is tiltedfrom the horizontal to a near vertical planefacing the subject, the phase angle shiftsabout 60° (Briggs, Thompson, & Brogden,1954). Similarly, changing the plane of move-ment from 30° slant to the right to 30° slantto the left results in a change of phase ofabout 60° (Thompson & Brogden, 1955).

The trigonometric function apparently doesnot depend on visual factors. Begbie (1959)had subjects attempt to draw straight linesfrom one point through another point in ahorizontal plane, and calculated the averagedeviation of the line from the point at whichit was aimed. Even when subjects closedtheir eyes during the movement, they showedthe same trigonometric function as long asthe movements were not too long.

One possibility is that the function resultsfrom movement about two joints: (a) move-ment of the forearm about the elbow and (b)movement about the shoulder joint. If thearm were moved in a near horizontal planestarting at a point in the median plane suchthat the initial angle between forearm andupper arm was about 90°, moving on the135°-315° axis would involve primarily move-ment of the forearm about the elbow whereasthe 45°-225° axis would involve mainly move-ment at the shoulder joint. Goldscheider(1899, reported in Howard & Templeton,1966) showed that the threshold in degrees

for perceiving passive movement of the armis larger at the elbow than at the shoulder. Ifit is assumed that arm steadiness is inverselyrelated to the threshold for movement, thenas a subject moved along the 135°-315° axis,the relatively large unsteadiness at the elbowwould result in fluctuations parallel to thedirection of travel. For movement along the45°-225° axis, unsteadiness at the elbowwould be perpendicular to the line of travel.Thus, as the direction of movement is changedfrom 135° to 225°, there would be a rise inerror. As direction progresses to 315°, theerror would drop and then rise again as the45° angle is approached.

Speed-accuracy trade-off. It is clear fromthe earlier discussion of Bailey and Pres-grave's work that movement time is a jointfunction of the extent and required accuracyof movement. Fitts (1954) reported three ex-periments in which an attempt was made toquantify the relationship. The tasks were totap alternately two plates separated by somedistance, transfer washers from one pin toanother, and transfer pins from one hole toanother. He found that movement time iswell predicted by:

MT-a Iog2 (2A/W) [2]

where A is the amplitude of the movement(the distance from the center of one target tothe center of the other), W is the width of thetarget in the tapping task or the tolerance inthe other two tasks, and a and b are con-stants. Perhaps the most interesting aspect ofthe relationship (hereafter referred to asFitts' Law) is the trade-off between distanceof the movement and width of the target.According to Fitts' Law, if the distance of amovement is doubled, the movement timedoes not change if the width of the target isalso doubled.

The information content of a movement isgiven by ID = Iog2 (2A/W). Informationtransmission may be viewed as the "ability toproduce consistently one class of movementfrom among several alternative movementclasses . . . [Fitts, 1954, p. 381]." The rateof information transmission was denned byFitts as C = ID/MT and was found to beapproximately 10 bits/sec.

In later experiments Fitts and Peterson

390 STEVEN W. KEELE

(1964) showed that a discrete tapping taskresulted in a faster rate of information trans-mission (between 14 and 22 bits/sec) thanthe continuous tapping task of Fitts (1954).The higher rate of transmission was partly dueto the fact that time on target was not in-cluded in total movement time. In addition,the constant b of Equation 2 was smaller thanin the earlier study. Fitts and Peterson alsoshowed an increase in movement speed overthree sessions of practice due to a decreasein the constant a of Equation 2 with b re-maining the same. However, with 31-dayspractice on a peg transfer task (Kay, 1962),there was also a decrease in b resulting in anincrease of transmission rate from about 10bits/sec to over 20 bits/sec.

In a study by Fitts and Radford (1966),subjects were given monetary payoffs thatemphasized either speed or accuracy or wereneutral. As would be expected, increasing thepayoff for speed resulted in increased speedbut more target misses. The critical questionis whether the increase in misses balances theincrease in speed such that the rate of in-formation transmission is constant under thethree conditions. To determine this, Fitts andRadford calculated what target width wouldbe necessary at each speed for 95% of themovements to hit the target (hits were as-sumed to be distributed normally about thetarget center). The calculated target width(W) was then substituted for the actualtarget width (W) and solved for the rate ofinformation transmission. Comparing the ac-curacy condition with the speeded condition,small but consistent increases in the rate ofinformation transmission were found in thelatter.

Preparation time apparently has little effecton transmission rate. Fitts and Radford(1966) found little difference whether sub-jects were directed to hit a target as soon aspossible after the appropriate target was in-dicated or whether they were to take as muchtime as desired before initiating the movement.These results were substantiated in an un-published experiment by Keele.

Further applications of Pitts' Law. In orderto compare Fitts' Law with the extensive datareported by Bailey and Presgrave (1958)and cited earlier, ID was calculated for each

distance and tolerance of Class C movements.The relation between movement time and IDwas reasonably linear, and a least squareslinear regression accounted for 95.3% of thevariance. The rate of information transmissionvaried from 6.3 to 9.8 bits/sec and was some-what less than the rates found by Fitts(1954) and Fitts and Peterson (1964). Thisdifference in rates is probably explained byBailey and Presgrave's instructions to makemovements at a standard rate rather than ata rapid rate.

Although Fitts' Law accounted for a largeproportion of Bailey and Presgrave's data,there were systematic deviations from linear-ity. At a given level of tolerance, movementtime showed a slight positive acceleration withthe logarithm of distance. Welford (1960)has proposed a modification of the index ofdifficulty: ID = Iog2(4 + $W)/W. This mod-ification resulted in a slightly better fit to theoriginal data of Fitts (1954) and the morerecent data of Fitts and Peterson (1964).Consequently, it was also applied to Baileyand Presgrave's data. A least squares linearregression accounted for 95.6% of the vari-ance, a value nearly the same as for Fitt'soriginal formulation.

Fitts' Law does provide a good fit to Baileyand Presgrave's data for linear movements, buta large discrepancy arises for rotary move-ments. They reported that the time for turn-ing the hand is linear with the number of de-grees turned. Since movement time dependson the level of precision required and Baileyand Presgrave do not report the precision re-quired for different degrees of turn, it is dif-ficult to evaluate their data. In contrast,Grossman and Goodeve (1963) and Knightand Dagnall (1967) controlled the level ofprecision and reported logarithmic relation-ships.

Brown and Slater-Hammel (1949) have re-ported an experiment in which rapid move-ments were made to a line rather than to atarget of finite width. The line was 2.5, 10, or40 cm. from the starting position. Movementtime to come within .5 mm. of the target lineincreased almost perfectly linearly with thelogarithm of the distance. Thus, Fitts' Lawalso provides a good description for movementtime to line targets.


Using a different approach, Woodworth(1899) kept the time of repetitive movementsconstant for distances of 5, 10, IS, and 20 cm.to a line and measured error. According toFitts' Law, an increase in distance requires adecrease of the same proportion in precision tomaintain equal movement time. Woodworthfound that, on the average, the amount oferror at long distances was slightly less thanpredicted. Since he used only three subjects,the discrepancy could be due to individualdifferences. Another possibility is that move-ment times may not have been constant fordifferent distances. If, as distance increased,relatively more time were spent in moving andless time spent in reversing direction, then thelong movements would be more accurate thanpredicted. These questions could be answeredby experiments with more subjects and withmeasurements of the actual time spent moving.

Repetitive movements. One might expectthat a minimum time is necessary between themuscle activation which starts a movementand the activation of the opposing muscles toend it. The distance traveled in this timeshould depend on the strength of the initialcontraction, and, consequently, the minimumtime for various length movements (assumingthe movement must be stopped by muscularaction) should be the same.

Travis (1929) has shown that voluntarymovements of the finger normally occur inphase with finger tremors. Such tremors occurat the rate of 8-12 per sec. for the finger,wrist, and forearm (Dresslar, 1892; Stetson &McDill, 1923). Thus, if a tremor is on theupswing, downward movement is delayed untilthe downward tremor, suggesting that whenmovement is started in one direction, there isa minimum of 80-120 msec, before opposingaction will occur. In accordance with this view,Dresslar (1892) found finger-tapping rate tobe about 8-10 taps/sec. Bryan (1892) founda slower rate of tapping (6-7 taps/sec), butthe rate was about the same for finger, wrist,elbow, and shoulder movements. Furthermore,the rate was independent of the length offinger excursion, which ranged from 1 to 40mm. Bryan reported that earlier work by VonKries also showed tapping rate to be inde-pendent of length of finger movement.

Simultaneous and successive movements. A

few investigators have determined whetherthere is any decrement in time per movementwhen two movements are made. Langfeld(1915) studied tapping rates for: (a) singlefingers, (b) two fingers simultaneously, (c)two fingers in alternation—that is, while onefinger was being raised the other was lowered,and (d) complete alternation in which onefinger was moved up and down before theother began. As a rule, for any two fingersthe simultaneous tapping rate was no slowerthan the rate for the slowest finger alone, andin some cases the simultaneous tapping ratewas faster than for either finger alone. Whenfingers were tapped in alternation, the rate perfinger averaged about f the rate for singlefingers. With complete alternation, the rateper finger was only | the rate for singlefingers. Thus, simultaneous tapping results inno loss of efficiency per finger but alternatetapping does.

In finger tapping, little precision and novisual attention is required. Bailey and Pres-grave (1958) have shown for simultaneousmovements that as the degree of precision re-quired increases and as the separation betweentwo targets increases, the time for movementincreases. With no separation between thetargets, there is no decrement over singlemovements. This can be explained simply bythe necessity for looking back and forth be-tween the two targets when they are separated.

Time and motion analysts have long in-sisted that when motions are made with twohands, they should be symmetrical (e.g.,Barnes, 1963). To test this hypothesis, Peter-son (1965) had subjects move their hands asrapidly as possible and touch targets withtheir forefingers. There were four major con-ditions: (a) one-hand movement; (b) twohands in symmetry—that is, both forward,both back, both out, or both in; (c) twohands in the same plane but not in symmetry—that is, both left, both right, or one forwardand one back; and (d) two hands movingperpendicular to each other, such as oneforward and the other to the left. Two-handsymmetric movements were more accurate andslightly faster than one-hand movements (sup-porting Langfeld's work). Two-hand motionsin the same plane but not symmetric hadfewer misses than one-hand motion but were

392 STEVEN W. KEELE

somewhat slower. Finally, perpendicular move-ments resulted in the most misses and were asslow as nonsymmetric movements in the sameplane.

In a study by Trumbo, Rogers, and Avant(1967), however, symmetry was not impor-tant. Subjects moved a pointer in a matrix byturning two cranks in the frontal plane. Onecrank resulted in left-right movements of thepointer, and the other resulted in up-downmovements. The time to move the pointer topredesignated positions for symmetric move-ments (one crank clockwise and the othercounterclockwise) and for nonsymmetric move-ments (both clockwise or both counterclock-wise) was determined; no difference wasfound. To determine whether motions in dif-ferent planes interfere with each other, Norrisand Spragg (1953) used a two-hand trackingtask in which the target follower was con-trolled by two cranks. Having two cranks indifferent planes did not result in performancedecrement beyond that which resulted whenboth cranks were in the worst of the twoplanes.

Consecutive movements, in contrast to si-multaneous ones, result in substantial slow-ing. Langfeld's (1915) study has been men-tioned in which the making of alternatingfinger movements resulted in considerable lossof speed. Subjects, in an experiment by Rubin,Von Treba, and Smith (1952), moved fromone knob to another and turned each knobthrough varying degrees. The time taken tomove between knobs increased with the degreeof turn required by the knob. Similarly, Simonand Simon (1959) have shown that the timeto move between knobs is increased if theknob turn before or after the movement re-quired relatively high precision.

To summarize, multiple movements result insubstantial decrement as opposed to singlemovements only when simultaneous motionsrequire visual guidance to separate targets orwhen the movements are consecutive. Am-biguities in the importance of symmetry maybe due to the different types of motion in-volved (linear vs. rotary) or to differences insensitivity of the experiments. In any casethe effects of asymmetry are not large. Greaterinterference may be found between motions ofdifferent classes (e.g., one linear and the other

rotary), but apparently there has been noexperimental work on this question.

FEEDBACK AND MOVEMENT CONTROL

There are three possible ways in whichmovements may be controlled and each ofthese ways will be discussed. First, if move-ments are made slowly enough, correctionscan be made on the basis of visual feedback.Data on the processing time for visual feed-back are presented and implications of afeedback processing analysis for understand-ing Pitts' Law and continuous tracking arediscussed. Second, evidence that kinestheticfeedback is important, even in the presenceof visual feedback, and data on kinestheticprocessing time are reviewed. And last, move-ments may be preprogrammed in the sensethat the amount and timing of innervation tothe muscles may be determined before themovement begins, and once it has begun,peripheral feedback would exert no furthercontrol.

Visual feedback. An important question inunderstanding movement control is how longit takes to process visual feedback. Wood-worth (1899) and Vince (1948) studied thisquestion by varying the stroke rate of backand forth movements made with the eyesopen or closed. In Woodworth's experiment,subjects were asked to reproduce the lengthof the previous movement; in Vince's, theywere asked to move to a fixed line and back.At a rate of 100 to 180 strokes/min, accuracywas no better with eyes open than with eyesclosed. This corresponds roughly to abouttwo strokes/sec or 500 msec, to process visualfeedback.

One problem with the method of Wood-worth and Vince is that some time is spent inreversing movement so that feedback process-ing time is overestimated. To avoid this prob-lem, Keele and Posner (1968) studied dis-crete movements from a home position to atarget, and only the time in motion wasmeasured. Visual feedback was eliminated onhalf the trials by turning off the lights as thehome position was left. On those trials themovement was completed in the dark. Visualprocessing time, as estimated by the shortestmovement time at which hitting the target


was facilitated by having the lights on, was190-260 msec.

Subjects in an experiment by Pew (1966a)attempted to maintain a target in the centerof an oscilloscope by sequentially pressing twokeys, one of which caused target accelerationto the right and the other to the left. Whenthe oscilloscope display was blanked out forperiods up to 410 msec, after a response,the modal time before the next correctiveresponse was 300-350 msec, after the end ofblanking. Since some of the corrective re-sponses required less than the modal time,Pew's data are not inconsistent with that ofKeele and Posner which estimate minimumprocessing time. The processing time in Pew'sexperiment could also be slightly larger dueto the nature of the task. The reaction timefor pressing a key in response to the directionof movement of a target is probably slowerthan that for simply moving the hand in thedirection which would center the target. Fitts(1964), for example, reported that reactiontime to the onset of a light is faster if theresponse is merely to point at the lightrather than to press a key. Supporting thisview, Pew, Duffendack, and Fensch (1967)reported that in sine-wave tracking with de-layed feedback, corrective responses weremade about 190-220 msec, after feedbackdelay. In that experiment a compatible iso-metric lever control was used rather thankeys.

Feedback interpretation of Pitts' Law. Al-though Fitts' Law was originated in terms ofinformation theory, it is possible to derivethe theory from feedback considerations. Thepresent derivation is similar to one proposedby Grossman and Goodeve (1963).

Assume that there is a minimum time t forprocessing feedback—that is, the time for aninitial movement and each corrective move-ment is constant. Thus, if there is an initialmovement and n — 1 corrective movements,the total movement time is MT — nt. Nextassume that the relative accuracy of a move-ment is constant, that is, Xt/X^i = K, whereXt is the mean absolute distance from thecenter of the target after the ?th correctivemovement, and K is a constant. X0 = A isthen the distance from the starting position tothe target center. After starting a movement,

corrections are made until it is within thetarget area. In other words, Xn = $W whereW is the width of the target. Therefore,Xn = KXn^ = K*Xn_2 = • • • = K*A - W-Solving for n: n = - logo 2A/W/\o& K.Thus:

= nt = b logs 2A/Wwhere b = - t/\og2 K. [3]

If it is assumed that the initial movementtakes less time by a constant a than the othercorrective movements (since the time to decidehow far to make the initial movement isnot included in the movement time), then

MT= (n- 1 )t+ (t- a) = nt-a = blo&2A/W-a. [4]

This equation is the same as Fitts' Law(Equation 2).

The assumption that the time for each cor-rective movement is constant and independentof distance may not be strictly true. Vince(1948, Exp, IV) and Searle and Taylor(1948) showed that the time from the startof movement to the first inflection in thekymograph record tended to increase as thedistance of movement increased. The secondassumption that error is proportional to dis-tance is also only approximately true. Formovements of so short a duration that visualfeedback does not facilitate accuracy, increasesin error are not quite proportional to increasesin distance (Vince, 1948, Exp. Ill, and Wood-worth, 1899). Woodworth found the pro-portion of error was less than 4% in mostcases, but Vince found the relative error to beabout T%. The small deviations from con-stant duration and from constant relative errortend to cancel out each other as far as Fitts'Law is concerned.

One test of the adequacy of the model isto determine the constant b in Equation 3from empirical estimates of t and K. As dis-cussed earlier, Keele and Posner (1968)have suggested that t is between 190 and 260msec. Setting K equal to 4% ( Woodworth 'sestimate) and t — 190 msec., b = 40 msec.With K equal to 1% (Vince's estimate) andt = 260 msec., b = 70 msec. The latter valueis quite close to the empirically determinedvalue of 70-75 msec, found by Fitts andPeterson (1964).

394 STEVEN W. KEELE

According to feedback theory, a dispropor-tionate amount of the movement time shouldbe spent near the target. Peters and Wen-borne (1936) had subjects move a stylusrapidly along a track and lift up from thetrack when the stylus reached a line. Theyfound that movements started somewhatslowly, picked up considerable speed, andslowed drastically near the end of the move-ment. The initial starting speed probably wasinflated slightly due to a small reaction-timecomponent, and the ending time might havebeen even slower if subjects had had to hitan actual target rather than to lift up thestylus. Annett, Golby, and Kay (1958) ana-lyzed motion pictures of the movements in-volved in transporting pins and placing themin holes of various tolerance. The initialmovement time covering 15/16 of the totaldistance was nearly constant (about 250-300msec.) for different tolerances, but the finaladjustment time increased as tolerance de-creased. Final positioning of the pin variedfrom about half as long in time as the initial15/16 of the distance to twice as long, de-pending on the degree of tolerance. Thus,there seems to be quite good qualitative aswell as quantitative agreement for the feed-back analysis of Pitts' Law.

Feedback and continuous tracking. An in-teresting implication of feedback theory is thatthe amount of error during continuous track-ing should be predictable from the accuracyof single movements. As was mentioned in theprevious section, a movement to a target maybe viewed as a series of submovements, eachone reducing the remaining error. In a con-tinuous tracking task, however, the targetposition has changed after each submovement.Assuming as before that the error of eachsubmovement is proportional to the distance tothe target, each submovement should transmitthe same amount of information as in the fixedtarget situation. Further, if the duration ofeach submovement is the same as for the fixedtarget situation, then the rate of informationtransmission in continuous tracking shouldapproach that found by Fitts (1954) forreciprocal tapping (10 bits/sec). Grossman(1960) found that as the rate of informationinput in continuous tracking with previewincreased, the rate of transmission increased to

about 7 bits/sec. Unfortunately, he did nothave a high enough rate of information inputto determine whether or not the rate of trans-mission leveled off at about 10 bits/sec.

Kinesthetic feedback. Howard and Temple-ton (1966) have suggested that:

Kinaesthesis is best understood as a behaviouralterm. It includes the discrimination of movement andamplitude of movement of body parts, both pas-sively and actively produced. Visual and auditoryinformation is assumed to be absent. As well asafferents from muscles and joints, touch, stretch,and pressure signals from the skin serve thesediscriminations. The pattern of motor innervationis almost certainly also available as a source ofinformation for kinaesthetic judgments [pp. 71-72].

The present paper departs slightly fromHoward and Templeton's definition by notincluding motor innervation as part ofkinesthesis. The reason for this is that all theother sources of kinesthesis are afferent andof peripheral origin, whereas motor innerva-tion (or a motor program) is of central originand may have quite different consequencesfor movement control. For example, when askill becomes automatized in the sense thatit requires little attention, it may be primarilyunder motor program control.

Some studies have attempted to show theimportance of kinesthesis in the control ofrapid movements by blocking kinesthetic cues.Lazlo (1966, 1967) found that the loss ofkinesthetic sense from ischemia is very detri-mental to tapping. It is possible, though, thatsome of the decrement was due to efferentdamage as well as afferent. This technique alsoincurs novel feelings and considerable dis-comfort and pain, which could be responsiblefor some performance loss. In support of thesearguments, Provins (1958) found almost nodecrement in tapping rate when the indexfinger was anesthetized with xylocaine. How-ever, in his studies, kinesthetic cues werestill available from muscles and tendons, whichwould account for the lack of decrement.

A different approach is to look for be-havioral evidence that kinesthetic cues arebeing used. Gibbs (1965) noted that whensubjects were required to make rapid move-ments in the opposite direction to step func-tion signals, they occasionally started themovement in the wrong direction. Early in


practice the mean time to correct the erroneousresponse was .24 sec., but later in practicethe correction time was only .11 sec. Gibbssuggests that the latter time is too fast forvisual correction and instead is due tokinesthetic correction. An alternative, however,is that at the appearance of a signal, a motorcommand is issued to move in the most prob-able direction. As the signal is further proc-essed, the correct direction is determined andcompared with the just issued command, andif there is a discrepancy, a motor commandto reverse direction is issued. Thus, the feed-back involved in the correction could be ofcentral origin rather than peripheral origin.Models of this type are discussed by Rabbitt(1967).

Some studies have attempted to manipulatethe quality of kinesthetic feedback. Burke andGibbs (196S), Gibbs (1954), and North andLomnicki (1961) compared pressure controlswith freemoving (amplitude) controls in bothpursuit and compensatory tracking. Theyfound that pressure control using either fingeror forearm movements resulted in more ac-curate tracking. Gibbs argued that the ad-vantage of pressure control is due to the betterquality and greater rapidity of kinestheticinformation in isometric muscle contractionsas opposed to isotonic contractions. These re-sults certainly support the hypothesis thatkinesthesis plays an important role in move-ment control, but they are not conclusive. Itcould be argued that preprogramming ofmovements is more accurate for force than fordistance control; or, it may be that amplitudecontrols result in worse performance merelybecause it takes longer to make large move-ments than to make the very short isometricmovements with pressure controls. Similararguments may be made for the data ofBriggs, Fitts, and Bahrick (1957). Theyfound that manual compensatory trackingwas best with a combination of relatively highforce and high amplitude movement. Again,it is not clear that this would necessarily implythat kinesthetic information is used.

More conclusive evidence for the role ofkinesthesis in tracking comes from a study byNotterman and Page (1962). They comparedtracking, with a movable control stick havingvarious degrees of elasticity, viscous damping,

and inertia, to performance with an isometriccontrol in which elasticity, damping and in-ertia were computer-controlled constants. Thus,the relation between an operator's force on thecontrol stick and the oscilloscope output wasthe same in both cases, but with the movablecontrol, the operator had kinesthetic feedbackarising from movement as well as from force.The movable control resulted in better per-formance.

An experiment by Fleishman and Rich(1963) also indicated the importance ofkinesthesis in tracking. They showed that aspractice on the Two-hand Coordination Taskincreased, subjects with high kinesthetic sensi-tivity, as shown by small difference limens forjudgments of lifted weights, became increas-ingly better than subjects with low sensitivity.

Some attempt has been made to determinehow long it takes to process kinesthetic feed-back. In an experiment by Chernikoff andTaylor (1952), a blindfolded subject's armwas held horizontally in a sling and thendropped. Reaction time from the onset of thedrop until arm direction was reversed wasabout .11-. 12 sec. Subjects in a study byVince (1948) pulled a pointer down in re-sponse to a sudden displacement of a line.On some trials the spring tension opposingthe movement was increased. It took subjectsabout .16 sec. to react to the increased springtension. This value is slightly larger thanChernikoff and Taylor's estimate, perhaps dueto the increased uncertainty of when a cor-rection is necessary. Alternatively, Chernikoffand Taylor could be measuring spinal reflextime rather than central processing time. Therapidity of kinesthetic processing may ex-plain why, apparently, it is important in track-ing even with vision being used.

Attention to feedback. One might expect onthe basis of studies of the psychological re-fractory period (see Bertelson, 1966; Smith,1967; and Welford, 1967, for reviews) thatwhile feedback is being processed, the process-ing of another signal would be delayed. Re-cently, Keele and Posner (1967) have re-ported that in some conditions rapid motionsdo require attention. The attention demandof a rotary movement was determined by thedelay in reaction time to an acoustic signaloccurring at various positions within the move-

396 STEVEN W. KEELE

ment. Attention was greater at the beginningof the movement, decreased to a relatively lowvalue in the middle, and showed a slight in-crease near the end. At all positions the de-mand was greater for narrow targets than forwide targets.

Although movements do require attention,Leonard (19S3) and Jeeves (1961) haveshown that some surplus processing capacityis available. Subjects made a series of re-sponses to lights. After each response, theyreturned to a home position before makingthe next response. If the light indicatingthe next response came on during the returnmovement, much less time was spent on thehome position than when the next light cameon only upon reaching the home position.

In the experiments just described, the sub-jects made movements to visually dennedtargets. There apparently has been no studyof the attention demand of kinestheticallycontrolled movements, although Welch (1898)has shown that maintenance of a strong hand-grip deteriorates in proportion to the difficultyof a simultaneous mental task.

It has been suggested that repetitive move-ments, rather than requiring attention, mightbe performed better if attention were at-tracted to a secondary task. Although Bliss(1892) reported that the regularity of rapidtapping increased when attention was divertedby tasks such as mental addition, he did notperform any systematic studies on this ques-tion. The data which he did report were sovariable that no conclusions can be drawnfrom them. Boder (1935) also studied fingertapping during simultaneous performance ofa secondary task. The secondary task ofwatching lights and later reporting the orderof their occurrence tended to reduce the rateof tapping a slight amount but had very littleeffect on variability. When the secondary taskinvolved moving a lever in one direction for ared light and in the other direction for a greenlight, the rate of tapping was considerablyslowed, particularly while the lever was beingmoved. There was also a large increase invariability of tapping during the lever move-ment.

The data of Boder offer no evidence thatperformance of a simple tapping task isimproved during secondary task performance.

It is possible, nonetheless, that simple repeti-tive tasks with emphasis on regularity ratherthan speed could be improved by divertingattention. Scripture (1899) reported that forrhythmic tapping, there is an optimal periodat which the variability of the intertap intervalis minimum. Faster or slower rates resulted inincreased variability. Perhaps deviations froma natural rhythm require attention even forwell-practiced tasks. Michon (1966, 1967)has looked at variability in tapping ratesranging to 2,400 msec, between taps. Hisresults indicate that if processing of a second-ary task is occurring at the time a tap shouldbe executed, then the timing shows increasedvariability. If, however, there is time be-tween taps (e.g., with the 2,400 msec, intertapinterval) to complete processing of the sec-ondary task, there is no interference. Certainlymore study needs to be done on the importantquestion of attention to repetitive movementsand to series of nonrepetitive but highly prac-ticed movements, such as in playing a familiarpiece of music on a piano.

Formation of motor programs. In a rathertrivial sense, all movements involve motorprograms. Visual or kinesthetic informationindicating the distance to a target must beconverted to a muscle command, and oncethe movement is initiated, it is only undercontrol of the program and cannot be changeduntil some minimum time for processing feed-back. An important question, however, is themodifiability of the motor program. In anexperiment by McLaughlin (1967), subjectsfixated on a spot of light. Upon signal from abuzzer, they made a rapid eye movement toa target 10° to the left. The initial movementimpulse prior to corrective motions was nor-mally accurate within about .3°. Following afew initial trials of that type, the target lightwas switched after the start of the movementfrom the 10° position to a position only 9°from the fixation point. At first the eye tendedto overshoot the 9° position, but after a fewtrials the 10° signal came to initiate a nowappropriate 9° movement. Upon return to theoriginal condition, undershooting occurred forseveral trials, indicating that the motor pro-gram had indeed been modified. Young, Green,Elkind, and Kelly (1964) similarly showedfor continuous tracking that subjects adapted


to changes in polarity and gain of controlin about 2 seconds.

The concept of a motor program is muchmore important when it is related to a seriesof predictable movements. In that situation,movement control may shift from visual-kinesthetic feedback to preprogrammed con-trol. Such a shift could have at least threeadvantages. First, the degree of attention re-quired may be reduced. As already mentioned,there has been little work on this importantquestion. Second, successive stimuli may beanticipated so that appropriate movementsmay coincide with the stimuli rather than lagbehind. And third, it may be possible formovements to be made at a much faster rate.

Evidence that subjects learn to anticipateregular stimulus occurrences so that they donot lag behind has been found by Poulton(1952), Noble, Trumbo, and co-workers (seeNoble & Trumbo, 1967, for a review of theirwork), and Stark and Young (1965). Poulton(1957) has shown, in addition, that if subjectsclosed their eyes for a 5-sec. period whiletracking a 60-cpm input, tracking was oftenmaintained as well as when their eyes re-mained open for a 5-sec. period. The mostcommon error when tracking with eyes closedwas a gradual shift in timing.

The timing of movements may not becompletely independent of attention, sinceTrumbo, Noble, and Swink (1967) haveshown that a simultaneous mental task (antic-ipating numbers) interferes with the timing.Nonetheless, with a large amount of practice,the motor task becomes more impervious tosecondary task interference (Noble, Trumbo,& Fowler, 1967), suggesting that timing maybecome somewhat automated.

Recently, Pew (1966b) presented evidencesuggestive of a motor program. Subjects at-tempted to keep an oscilloscope target centeredon crosshairs by sequentially pressing twokeys, one of which caused target accelerationto the right and the other to the left. After alarge amount of practice, the best subjectswere making approximately four to five re-sponses per sec. Although the rate of respond-ing is close to the limits for processing visualfeedback, the pattern of responding suggestedthat, rather than modifying each responseon the basis of feedback, the subjects were

modifying a motor program. Quoting fromPew:

5 maintained a more rapid rate of responding thanwould have been possible if he were monitoringevery response on a closed-loop basis. . . . Whenthe target drifted off center to the left, S maintainedthe high rate of responding, but at the same timegradually increased the length of time the right keywas active relative to that of the left key, so thatover a series of responses the target was made todrift back toward the center [p. 769].

Other subjects, after drifting off target, madea single long duration corrective movement,suggesting that rather than gradually modify-ing a motor program, corrections utilizingvisual feedback were occasionally interspersedamong movements under programmed control.

Related evidence that for predictable eventsseparate movements might be organized as aunit and triggered as a whole was reported byStark and Young (1965). Subjects respondedto short pulses (less than 50-msec. duration)in a visual target by a rapid stroke followedby a return stroke. When the pulse durationwas known in advance, there was no delay inthe return movement, but when the shortpulses were mixed in with long duration pulses,there was a delay of about 200 msec, betweenthe beginning stroke and the end stroke.

The preceding studies quite convincinglyshow that movement control may become in-ternalized and, at least for short periods oftime, free of visual control. There is no evi-dence, however, that performance is main-tained without kinesthetic feedback. In viewof the rapid processing of kinesthetic informa-tion, it is still possible that the feedback fromone movement signals the next. It might beargued that anticipatory timing must be pro-grammed, but Adams and Creamer (1962)have proposed that even timing is based onkinesthesis. They suggest that movement gen-erates a kinesthetic trace which decays overtime, and the trace characteristics after vari-ous delays can be used for timing the start ofthe next movement. In support of their argu-ment they have shown that spring loading ona control stick does improve the timing to aregular series of step inputs, and, as will beseen in the next section, there is some evi-dence that there is a spontaneous decay insome movement cues.

398 STEVEN W. KEELE

As mentioned in an earlier section, there isevidence indicating that kinesthesis is used inperforming a series of movements. Still it isnot clear whether the individual movementswithin the series are initiated by feedbackfrom the previous movement or whether kines-thesis is used only intermittently in correctinga motor program. Separating the role of thesetwo factors in a series of movements appearsto be a formidable problem and may be pos-sible only with physiological techniques suchas discussed in the introduction.

REPRODUCTION OF MOVEMENTS

Although it appears difficult to determinewhether individual movements in a sequenceare under programmed control, it may be pos-sible to determine whether an individual move-ment in isolation is programmed. In this sec-tion, therefore, the reproduction of previousmovements will be considered, and evidencewill be presented that reproduction dependson a motor program, as well as other cues.

Motor program theory. Loeb (reported byHollingworth, 1909, and Woodworth, 1899)showed that when two successive movementsof subjectively equal length were produced,the movement involving the greater amountof active muscular contraction was actuallyshorter than the other. He assumed that in re-producing a previous movement, the musclesreceive the same innervation regardless ofkinesthetic spatial information arising fromthe movement. His results are explained bythe further assumption that a given amount ofinnervation produces smaller movements themore contracted the muscle. This theory isnot in accordance with findings of Holling-worth (1909) and Woodworth (1899) thatwhen a series of subjectively equal movementswas made from left to right, with each move-ment starting where the previous ended, themiddle movements were longer in extent thanthe movements on either side. Results byother investigators, however, do support amotor program theory.

When no feedback from a movement isavailable, reproduction should depend solelyon remembering the motor program of theprevious movement. Lashley (1917) studiedan individual in which kinesthetic feedbackfrom leg movements had been lost due to

injury. The patient was unable to perceive orreproduce passive movements of the leg, yetwas able to reproduce quite accurately move-ments that he had previously produced him-self. An additional finding of interest was thaton some occurrences, the subject stated thathe made a movement longer than intended;and indeed, such movements were longer thanothers intended to be the same length. Thisseems to imply that there is a central feedbackloop in which the issued command is com-pared with the intended program. (A similarmechanism was suggested earlier in this paperfor some results of Gibbs, 1965.)

Even in normal individuals, there appearsto be important differences in passive and ac-tive movements. Lloyd and Caldwell (196S)compared reports of leg angle when the legwas passively moved to a particular positionand of the angle produced when subjects wereasked to position actively their leg. The con-stant errors in the two situations were quitesimilar for leg extensions but different at legflexions. One interpretation of this effect isthat with passive movements, only kinestheticcues are available. With active movements, onthe other hand, the motor command to themuscles, as well as kinesthetic cues, might in-fluence judgments, resulting in different con-stant errors.

Other evidence for motor program theorycomes from studies on the control of eyemovements. Extensive reviews by Festingerand Canon (196S) and Howard and Temple-ton (1966) have suggested that one of thecues for keeping track of the direction of gazeis reference to preceding commands for eyemovement. Of particular relevance is an ex-periment by Festinger and Canon. Subjects ina dark room followed a spot of light withtheir eyes. Head position was kept fixed by abite board. The light was either suddenlyswitched from the center position to a termi-nal position or it slowly moved to the terminalposition. After the light turned off, theypointed with their arm to the position wherethe light had disappeared.

According to a motor program theory, acommand to move the eye from the centerposition to the terminal position would beissued in the sudden switching condition. Inthe slow movement condition, however, no


motor command corresponding to the totaldistance would be issued because there wasno way of knowing how far the light wouldmove before stopping. Since other evidencereviewed by Festinger and Canon and byHoward and Templeton indicates that the eyehas little kinesthetic sensation, there wouldbe no distance cues available in the slow condi-tion for directing the arm to the terminal posi-tion of the light. As predicted, rapid eyemovements resulted in much more accuratearm positioning than slow eye movements.When the head was not held by a bite board,kinesthetic cues from the neck were availablefrom the slow movement condition, and sub-jects then were able to position the arm asaccurately as in the step movement condition.

The evidence appears fairly conclusive thatmotor programs may be used in reproducingactive movements, but other cues might alsobe important.

Movement duration. Kramer and Mos-kiewicz, as reported by Hollingworth (1909),proposed that Loeb's results were due to theproduction of equal movement durationsrather than equal motor commands. Ifmovements with greater muscle contractionwere also made at a slower speed, thenthey would be shorter in extent. To test thishypothesis, Hollingworth (1909) measuredthe duration as well as the extent of subjec-tively equal-length successive movements. Hefound little systematic difference in durationof successive movements, and durations wereless variable than extent even though thesubjects had been instructed to reproduce ex-tent. These results give some support toKramer and Moskiewicz's interpretation.

Leuba (1909) had people make a standardforearm movement of 8°. This was followedby a movement of 7°, 8.5°, or 10°, and thesubjects judged whether the second movementwas less, equal to, or greater in extent thanthe standard movement. When the secondmovement of a given length was longer induration than the standard, there was atendency to judge it longer in extent; whenless in duration, the tendency was to judge itless in length. Unfortunately, Leuba presentedvery little data to support these trends. Inother data the duration of movement wasconfounded with the actual extent.

Duration by itself is not a sufficient cuefor reproducing movement, since the speedsmight differ. It is possible that, rather thanpreprogramming the total amount of innerva-tion to the muscles, people program the rateand duration of innervation. If one adoptsthis view, there is actually little differencebetween the motor program theory of Loeband the duration theory of Kramer and Mos-kiewicz. Alternatively, duration informationmay be combined with kinesthetic speed orforce information. Leuba (1909) showed thatif a constant resistance were added to thesecond movement, there was an increase induration, but there was little change in theaccuracy of judged extent. He concluded thatsubjects were able to use kinesthetic informa-tion to judge the speed and made compensa-tory changes in duration. When the resistanceto the movement changed throughout themovement, judgment of extent deteriorated,implying that, in this situation, subjects wereless able to judge the speed of movement andcompensate with changes in duration.

Movement extent. The experiment by Fest-inger and Canon (1965), as discussed earlier,presented evidence that kinesthetic cues fromthe neck could be used for movement repro-duction. Kinesthetic senses could give in-formation on the speed of movement, and,as discussed previously, that information couldbe combined with duration to determine ex-tent. Alternatively, there could be a directkinesthetic sense of extent that does not de-pend on speed or duration. In an experimentby Woodworth (1899), subjects with theireyes closed drew a line and then attempted toplace a dot at each end of the line. They werenearly as accurate in reproducing the endpoints as when they attempted to reproducethe previous line, suggesting that people havea direct sense of the distance between twopoints. Such a cue would seem to depend onkinesthetic knowledge of locations in spacerather than distance per se.

Memory for movements. As noted earlier,the internalization of movement control hasimportant implications for understandingskilled performance. However, for such inter-nalization to be of much use, the timing andextent of movements must be retained for atleast short periods of time. Results by Poulton

400 STEVEN W. KEELS.

(1957) have already been mentioned whichindicate that the timing of successive move-ments may be very accurately retained forperiods as long as 5 sec. Work on the reten-tion of distance information will now be re-viewed.

Several studies have shown that the ac-curacy of reproduction decreases as the timebetween the original movement and the repro-duction increases. Hollingworth (1909) showeda loss of accuracy between 2 and 30 sec. Asimilar decrease in accuracy over a periodof 60 sec. was reported by Scripture, Cooke,and Warren (1897), but their data are sovariable they are inconclusive. Schnieder (re-ported by Hollingworth, 1909) found thatthe decrease in accuracy continued over aperiod of 15 min., and it has been more re-cently reported by Bilodeau and Levy (1964)that increased forgetting of the movementoccurs between 2 days and 6 weeks.

As with memory for other types of material,increasing the number of repetitions of amovement increases the accuracy of reproduc-tion (Adams & Dijkstra, 1966). This is im-portant because it indicates that the internal-ization of movement control may have longlasting as well as transitory effects. However,even after as many as IS repetitions prior toreproduction, there is still some decline in theaccuracy over a retention period of two min-utes. As Adams and Dijkstra point out, theforgetting occurs even with an unfilled intervalbetween an original movement and the repro-duction. Moreover, Posner and Konick (1967)have shown that the retention of movementsis not further degraded by an interpolatedmental task, even though it involves handmovements from writing. Short-term memoryfor verbal units, in contrast, shows little lossduring an unfilled interval, but a large losswith an interpolated mental task (Posner &Rossman, 196S). Boulter (1964) and Blickand Bilodeau (1963) have also found that aninterpolated verbal or movement task has noeffect on movement reproduction. Boswell andBilodeau (1964), on the other hand, foundthat the simple act of picking up a pencilfrom the floor between a movement and itsreproduction 28 sec. later resulted in greaterinaccuracy. Although special rules for posture

were given, it is possible that there were somepostural changes after picking up a pencilwhich would account for the greater inac-curacy in that situation.

A few studies have varied the type ofmovement in order to separate the variouscomponents of movement control and to de-termine whether there are differences in mem-ory. Hollingworth (1909) compared memoryfor extent with memory for duration andfound similar changes over time. Recently,Posner (1967) compared memory for move-ment extent starting from the same positionas the original movement with memory forextent in which the two movements started indifferent positions. Although the overall ac-curacy was greater when reproduction wasfrom the same starting position, there was nosignificant difference in the loss of accuracyover time. When reproduction was from thesame starting position, there was a slight, butnonsignificant, trend for memory loss to in-crease with an interpolated mental task. Onlyan experiment by Keele2 has found differentmemory functions for different types of move-ment. Memory for location in space, likememory for verbal material, underwent littlespontaneous forgetting but did show forget-ting with an interpolated mental task. Simi-lar results were also found for memory ofshort distances, but results more congruentwith those of Posner were found for longerdistances, suggesting that different cues areinvolved in the memory of long and shortmovements.

More research is clearly needed before itcan be determined what cues are different inshort and long movements, whether the for-getting of motor programs is different thanfor kinesthetic cues, and whether differentkinesthetic cues show differential forgetting.With regard to the last point, Howard andTempleton (1966) state that muscle spindleafferents do not project as far as the sensorycortex, whereas joint receptors do. If it weretrue that only cues having cortical projectionsare rehearsable, then movements dependingon spindle information might be expected toshow different forgetting functions than move-ments depending on joint information.

2 S. Keele, unpublished manuscript (1968).


REFERENCES

ADAMS, J. A., & CREAMER, L. R. Proprioceptive vari-ables as determiners of anticipatory timing be-havior. Human Factors, 1962, 4, 217-222.

ADAMS, J. A., & DIJKSTRA, S. Short-term memoryfor motor responses. Journal 0} Experimental Psy-chology, 1966, 71, 314-318.

ANNETT, S., GOLBY, C. W., & KAY, H. The measure-ment of elements in an assembly task—the in-formation output of the human motor system.Quarterly Journal of Experimental Psychology,1958, 10, 1-11.

BAILEY, G. B., & PRESGRAVE, R. Basic motion timestudy. New York: McGraw-Hill, 1958.

BARNES, R. M. Motion and time study: Design andmeasurement of work. New York: Wiley, 1963.

BEGBIE, G. H. Accuracy of aiming in linear handmovements. Quarterly Journal of ExperimentalPsychology, 1959, 11, 65-75.

BERTELSON, P. Central intermittency twenty yearslater. Quarterly Journal of Experimental Psy-chology, 1966, 18, 153-163.

BILODEAU, E. A., & LEVY, C. M. Long-term memoryas a function of retention time and other condi-tions of training and recall. Psychological Review,1964, 71,27-41.

BLICK, K. A., & BILODEAU, E. A. Interpolated ac-tivity and the learning of a simple skill. Journalof Experimental Psychology, 1963, 65, 515-519.

BLISS, C. B. Investigations in reaction time and at-tention. Studies from the Yale Psychological Lab-oratory, 1892-1893, 1, 1-55.

BODER, D. P. The influence of concomitant activityand fatigue upon certain forms of reciprocal handmovement and its fundamental components. Com-parative Psychology Monographs, 1935, 11, No. 4.

BOSWELL, J. J., & BILODEAU, E. A. Short-term re-tention of a simple motor task as a function ofinterpolated activity. Perceptual and Motor Skills,1964, 18, 227-230.

BOULTER, L. R. Evaluation of mechanisms in delayof knowledge of results. Canadian Journal of Psy-chology, 1964, 18, 281-291.

BRIGGS, G. E., & BROGDEN, W. J. Bilateral aspects ofthe trigonometric relationship of precision andangle of linear pursuit-movements. American Jour-nal of Psychology, 1953, 66, 472-478.

BRIGGS, G. E., FITTS, P. M., & BAHRICK, H. P. Ef-fects of force and amplitude cues on learning andperformance in a complex tracking task. Journalof Experimental Psychology, 1957, 54, 262-268.

BRIGGS, G. E., THOMPSON, R. F., & BROGDEN, W. J.The effect of angle of tilt upon the trigonometricrelationship of precision and angle of linear pur-suit-movements. American Journal of Psychology,1954, 67, 475-483.

BROGDEN, W. J. The trigonometric relationship ofprecision and angle of linear pursuit movements asa function of amount of practice. American Journalof Psychology, 1953, 66, 45-56.

BROWN, J. S., & SLATER-HAMMEL, A. T. Discretemovements in the horizontal plane as a function

of their length and direction. Journal of Experi-mental Psychology, 1949, 39, 84-95.

BRYAN, W. L. On the development of voluntarymotor ability. American Journal of Psychology,1892, 5, 125-204.

BURKE, D., & GIBBS, C. B. A comparison of free-moving and pressure levers in a positional andcontrol system. Ergonomics, 1965, 8, 23-29.

CHERNIKOFF, R., & TAYLOR, F. V. Reaction time tokinesthetic stimulation resulting from sudden armdisplacement. Journal of Experimental Psychology,1952, 43, 1-8.

CORRIGAN, R. E., & BROGDEN, W. J, The trigonometricrelationship of precision and angle of linear pursuitmovements. American Journal of Psychology, 1949,62, 90-98.

GROSSMAN, E. R. F. W. The information capacity ofthe human motor system in pursuit tracking.Quarterly Journal of Experimental Psychology,1960, 12, 1-16.

GROSSMAN, E. R. F. W., & GOODEVE, P. J. Feedbackcontrol of hand-movement and Fitts' Law. Pro-ceedings of the Experimental Society, Oxford,1963.

DRESSLAR, F. B. Some influences which affect therapidity of voluntary movements. American Jour-nal of Psychology, 1892, 4, 514-527.

FESTINGER, L., & CANON, L. K. Information aboutspatial location based on knowledge about ef-ference. Psychological Review, 1965, 72, 373-384.

FESTINGER, L., ONO, H. BURNHAM, C. A., & BAM-BER, D. Efference and the conscious experience ofperception. Journal of Experimental Psychology,1967, 74(4, Pt. 2).

FITTS, P. M. The information capacity of the humanmotor system in controlling the amplitude ofmovement. Journal of Experimental Psychology,1954, 47, 381-391.

FITTS, P. M. Perceptual-motor skill learning. In A.W. Melton (Ed.), Categories of human learning.New York: Academic Press, 1964.

FITTS, P. M., & PETERSON, J. R. Information capacityof discrete motor responses. Journal of Experi-mental Psychology, 1964, 67, 103-112.

FITTS, P. M., & RADPORD, B. K. Information capacityof discrete motor responses under different cogni-tive sets. Journal of Experiment Psychology, 1966,475-482.

FLEISHMAN, E. A., & RICH, S. Role of kinesthetic andspatial-visual abilities in perceptual-motor learning.Journal of Experimental Psychology, 1963, 66,6-11.

GIBBS, C. B. The continuous regulation of skilled re-sponse by kinaesthetic feed back. British Journalof Psychology, 1954, 45, 24-39.

GIBBS, C. B. Probability learning in step-input track-ing. British Journal of Psychology, 1965, 56, 233-242.

HINTZMAN, D. L. Classification and aural coding inshort-term memory. Psychonomic Science, 1965, 3,161-162.

402 STEVEN W. KEELE

HINTZMAN, D. L. Articulatory coding in short-termmemory. Journal of Verbal Learning and VerbalBehavior, 1967, 6, 312-316.

HOLLINGWORTH, H. L. The inaccuracy of movement.Archives of Psychology, 1909, 2, No. 13, 87P.

HOWARD, I. P., & TEMPLETON, W. B. Human spatialorientation. New York: Wiley, 1966.

JEEVES, M. A. Changes in performance at a serial-reaction task under conditions of advance and de-lay of information. Ergonomics, 1961, 4, 320-338.

KAY, H. Channel capacity and skilled performance.In F. A. Geldard, (Ed.), Defence psychology. NewYork: Pergamon, 1962.

KEELE, S. W., & POSNER, M. I. Processing of feed-back in rapid movements. Paper presented at themeeting of the Western Psychological Association,San Francisco, May 1967.

KEELE, S. W., & POSNER, M. I. Processing of visualfeedback in rapid movements. Journal of Experi-mental Psychology, 1968, 77, 155-158.

KENNEDY, D. Small systems of nerve cells. ScientificAmerican, 1967, 216, 44-52.

KNIGHT, A. A., & DAGNALL, P. R. Precision in move-ments. Ergonomics, 1967, 10, 321-330.

LANGFELD, H. S. Facilitation and inhibition of motorimpulses: A study in simultaneous and alternatingfinger movements. Psychological Review, 1915, 22,453-478.

LASHLEY, K. S. The accuracy of movement in theabsence of excitation from the moving organ.American Journal of Physiology, 1917, 43, 169-194.

LAZLO, J. I. The performance of a simple motortask with kinaesthetic sense loss. Quarterly Journalof Experimental Psychology, 1966, 18, 1-8.

LAZLO, J. I. Training of fast tapping with reductionof kinaesthetic, tactile, visual and auditory sensa-tions. Quarterly Journal of Experimental Psy-chology, 1967, 19, 344-349.

LEONARD, J. A. Advance information in sensori-motorskills. Quarterly Journal of Experimental Psy-chology, 1953, 5, 141-149.

LEUBA, J. H. The influence of the duration and ofthe rate of arm movements upon the judgment oftheir length. American Journal of Psychology, 1909,20, 374-385.

LIBERMAN, A. M., COOPER, F. S., SHANKWEILER, D.P., & STTIDDERT-KENNEDY, M. Perception of thespeech code. Psychological Review, 1967, 74, 431-461.

LLOYD, A. J., & CALDWELL, L. S. Accuracy of activeand passive positioning of the leg on the basis ofkinesthetic cues. Journal of Comparative andPhysiological Psychology, 1965, 60, 102-106.

MCLAUGHLIN, S. C. Parametric adjustment in sac-cadic eye movements. Perception and Psycho-physics, 1967, 2, 359-362.

MICHON, J. A. Tapping regularity as a measure ofperceptual motor load. Ergonomics, 1966, 9, 401-412.

MICHON, J. A. Timing in temporal tracking. Nether-lands: vanGorcum & Company, 1967.

NOBLE, M. & TRUMBO, D. The organization of skilledresponse. Organizational Behavior and Human Per-formance, 1967, 2, 1-2S.

NOBLE, M., TRUMBO, D., & FOWLER, F. Further evi-dence on secondary task interference in tracking.Journal of Experimental Psychology, 1967, 73,146-149.

NORRIS, E. B., & SPRAGG, S. D. S. Performance on afollowing tracking task (modified SAM two-handcoordination test) as a function of the planes ofoperation of the controls. Journal of Psychology,1953, 35, 107-117.

NORTH, J. D., & LOMNICKI, Z. A. Further experi-ments on human operators in compensatory track-ing tasks. Ergonomics, 1961, 4, 339-353.

NOTTERMAN, J. M., & PAGE, D. E. Evaluation ofmathematically equivalent tracking systems. Per-ceptual and Motor Skills, 1962, 15, 683-716.

PETERS, W., & WENBORNE, A. A. The time patternof voluntary movements. British Journal of Psy-chology, 1936, 26-27, 388-406, and 60-73.

PETERSON, J. R. Response-response compatibility ef-fects in a two-hand pointing task. Human Factors,1965, 7, 231-236.

PEW, R. W. Performance of human operators in athree-state relay control system with velocity-aug-mented displays. IEEE Transactions on HumanFactors in Electronics, 1966, 7, 77-83. (a)

PEW, R. W. Acquisition of hierarchical control overthe temporal organization of a skill. Journal ofExperimental Psychology, 1966, 71, 764-771. (b)

PEW, R. W., DUITENDACK, J. C., & FENSCH, L. K.A quantitative description of the effects of delayedvisual feedback on motor performance. Paperpresented at the meeting of the Psychonomic So-ciety, October 1967.

POSNER, M. I. Characteristics of visual and kines-thetic memory codes. Journal of ExperimentalPsychology, 1967, 75, 103-107.

POSNER, M. I., & KONICK, A. F. Short-term retentionof visual and kinesthetic information. Organiza-tional Behavior and Human Performance, 1966, 1,71-86.

POSNER, M. I., & ROSSMAN, E. Effect of size andlocation of informational transforms upon short-term retention. Journal of Experimental Psy-chology, 1965, 70, 496-505.

POULTON, E. C. Perceptual anticipation in trackingwith two-pointer and one-pointer displays. BritishJournal of Psychology, 1952, 43, 222-229.

POULTON, E. C. On the stimulus and response inpursuit tracking. Journal of Experimental Psy-chology, 1957, 53, 189-194.

PROVINS, K. A. The effect of peripheral nerve blockon the appreciation and execution of finger move-ments. Journal of Physiology, 1958, 143, 55-67.

RABBITT, P. M. A. Time to detect errors as a func-tion of factors affecting choice-response time.Acta Psychologica, 1967, 27, 131-142.

RUBIN, G., VON TREBA, P. J., & SMITH, K. U. Di-mensional analysis of motion: III. Complexity ofmovement pattern. Journal of Applied Psychology,1952, 36, 272-276.

SCRIPTURE, E. W. Observations on rhythmic action.Studies from the Yale Psychological Laboratory,1899, 7, 102-108.


SCRIPTURE, E. W., COOKE, W. C., & WARREN, C. M.Researches on memory for arm movement. Studiesfrom the Yak Psychological Laboratory, 1897, 5,90-92.

SEARLE, L. V., & TAYLOR, F. V. Studies of trackingbehavior. I. Rate and time characteristics of simplecorrective movements. Journal of ExperimentalPsychology, 1948, 38, 615-631.

SIMON, J. R., & SIMON, B. P. Duration of movementsin a dial setting task as a function of the precisionof manipulation. Journal of Applied Psychology,19S9, 43, 389-394.

SMITH, M. C. Theories of the psychological refractoryperiod. Psychological Bulletin, 1967, 67, 202-213.

SPERLING, G. Successive approximations to a modelfor short-term memory. Ada Psychologica, 1967,27, 285-292.

STARK, L. R., & YOUNG, L. Biological control sys-tems: A critical review and evaluation of develop-ments in manual control. Contract Report CR-190,1965, National Aeronautics and Space Administra-tion.

STETSON, R. H., & McDn-L, J. A. Mechanism of dif-ferent types of movement. Psychological Mono-graphs, 1923, 32, (3, Whole No. 145), 18-40.

THOMPSON, R. F., & BROGDEN, W. J. The effect ofangle of slant upon the trigonometric relationshipof precision and angle of linear pursuit-movements.American Journal of Psychology, 1955, 68, 615-623.

THOMPSON, R. F., Voss, J. F., & BROGDEN, W. J.The effect of target-velocity upon the trigonometricrelationship of precision and angle of linear pursuitmovements. American Journal of Psychology, 1956,69, 258-263.

TRAVIS, L. E. The relation of voluntary movement

to tremors. Journal of Experimental Psychology,1929, 12, 515-524.

TRUMBO, D., NOBLE, M., & SWINK, J. Secondarytask interference in the performance of trackingtasks. Journal of Experimental Psychology, 1967,73, 232-240.

TRUMBO, D., ROGERS, S., & AVANT, L. L. Com-patibility effects in a two-hand cranking task.Journal of Applied Psychology, 1967, 51, 35-38.

VINCE, M. A. Corrective movements in a pursuittask. Quarterly Journal of Experimental Psychol-ogy, 1948, 1, 85-103.

WATSON, J. B. Behaviorism. New York: Peoples In-stitute, 1924.

WELCH, J. C. On the measurement of mental ac-tivity through muscular activity and the determina-tion of a constant of attention. American Journalof Physiology, 1898, 1, 283-306.

WELFORD, A. T. The measurement of sensory-motorperformance: Survey and reappraisal of twelveyears progress. Ergonomics, 1960, 3, 189-230.

WELFORD, A. T. Single-channel operation in thebrain. Ada Psychologica, 1967, 27, 5-22.

WILSON, D. M. The central nervous control of flightin a locust. Journal of Experimental Biology, 1961,38, 471-490.

WOODWORTH, R. S. The accuracy of voluntary move-ment. Psychological Review, 1899, 3, (2, WholeNo. 13).

YOUNG, L. R., GREEN, D. M., ELKIND, J. I., &KELLY, J. A. Adaptive dynamic response char-acteristics of the human operator in simple manualcontrol. IEEE Transactions on Human Factors inElectronics, 1964, 5, 6-13.

(Received March 11, 1968)

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

VOL. 70, No. 6, PART 1 DECEMBER 1968 Psychological...

Documents

Transcript of VOL. 70, No. 6, PART 1 DECEMBER 1968 Psychological...