spatial trajectories and reaction times of aimed …brain.umn.edu/pdfs/ag007.pdf · spatial...

JOURNALOF NEUROPHYSIOL~GY

Vol. 46, No. 4, October 198 I. Printed in U.S.A.

Spatial Trajectories and Reaction Times of

Aimed Movements: Effects of Practice,

Uncertainty, and Change in Target Location

APOSTOLOS P. GEORGOPOULOS, JOHN F. KALASKA, AND JOE T. MASSEY

Department of Physiology, The Johns Hopkins University Schoul of Medicine, Baltimore, Maryland 2120.5

SUMMARY AND CONCLUSIONS

1. The spatial and temporal characteristics of arm movements in two (X-Y) dimensions were studied in three rhesus monkeys a) during the acquisition of an aiming motor skill, b) under conditions of spatial and temporal uncertainty, and c) when the location of the target changed during the reaction or movement time, from 50 to 400 ms after the presentation of the target.

2. The acquisition of the aiming skill was marked by an exponential reduction in the spatial variability of the movement trajectories. The reaction time remained virtually unchanged, whereas the peak velocity increased by a small amount. The skill was transferred appreciably to the other hand within a short period of training.

3. Handpath variability increased under conditions of spatial uncertainty. Temporal uncertainty in target presentation had no effect. Reaction time did not change signif- icantly in either condition.

4. Change of target location during the reaction or movement time elicited a graded movement toward the first target, followed by reversal of direction and movement to the second target. The duration of the movement toward the first target was a linear function of the time that elapsed from the presentation of the first target to the change of targets (interstimulus interval, ISI): the later the change occurred, the longer the movement toward the first target. In contrast to the gradation of hand movement with change in target location, the eyes always fully made a saccade to the first or second target.

5. The trajectory of the initial hand movement deviated occasionally toward the second of the two targets when the interstimulus interval was short and the targets were adjacent.

6. No marked delays were observed, beyond the reaction time, when responding to the first or the second target. A large increase in peak velocity was attained after reversing the movement, i.e., on the way to the second target.

7. The results of this study indicate that the process that generates the aimed movement becomes less variable with practice and is influenced by the uncertainty of the subject about the location of the target but not by the time of its appearance.

8. The orderly modification of the movement produced by change in target location suggests that the aimed motor command is emitted in a continuous, ongoing fashion as a real-time process that can be interrupted at any time by the substitution of the original target by a new one. The effects of this change on the ensuing movement appear promptly, without delays beyond the usual reaction time. No appreciable “psychological refractory period” is observed under these conditions, and the second stimulus has continual and effective access to the process generating the aimed arm movement,

INTRODUCTliON

Arm movements aimed at visual targets in space are common in the motor repertoire of primates. They lead to the acquisition of objects of interest, an indispensable function

0022-3077/8 I /OOOO-OOOOSO 1.25 Copyright 0 198 1 The American Physiological Society 725

726 GEORGOPOULOS, KALASKA, AND MASSEY

in everyday life. It is unknown how the vi- suospatial information concerning target location is used by the brain for the generation of aimed movements, nor is it known how the trajectories of these movements are formed or what factors influence them. Ex- perimental studies in which the motion of the arm was restricted to a single joint (at the elbow) provided useful information on the possible mechanisms of the muscular actions by which targets could be reached (43): the hand attained the intended end position even after complete deafferentation and without sight of the arm. These results led Polit and Bizzi (43) to the hypothesis that the variable controlled by the motor command is “an equilibrium point resulting from the interaction of agonist and antag- onist muscles” (p. I83 of Ref. 43).

The temporal characteristics of aimed movements resemble those of arm movements in general; e.g., the target can be attained by a fast or slow movement, similar to those observed in self-paced, step-tracking tasks (3, 36). Rapid aimed movements pos- sess a single-peaked velocity curve, which is symmetric in naive subjects (4). The final acquisition of the target is usually not ac- complished by the rapid movement compo- nent but is the result of a second, slower, “homing-in” phase (1, 2,4, 22,42,60). Cor- rections of aimed movements based on visual feedback of information concerning the position of the moving limb require at least one visual reaction time (33) and, therefore, can be operative only for movements that last longer than this time.

Rapid aimed movements do not always termi- nate dead on target. The distance of the point of their termination from the target is a measure of the error in target acquisition. It has been re- peatedly observed since the pioneer study of Woodworth (61) that speed and accuracy are in- versely related, so that increase in speed results in loss of accuracy. This phenomenon has been studied extensively in human subjects and is known as the speed-accuracy trade-off (see Ref. 32 for review). Schmidt et al. (48) investigated recently the effects on accuracy of the two com- ponents of speed, i.e., distance traveled (movement amplitude) and movement time. They found that errors in target acquisition increased linearly with I) increasing speed, 2) increasing distance traveled in a constant time, and 3) decreasing movement time for a given movement amplitude.

Their findings provide experimental support for the hypothesis of these workers (48) concerning the source of inaccuracy in rapid aiming movements; namely, that it results from variability in the motor output, and particularly from variability in the initial force (“impulse”) given to the hand. They have shown (48) that this variability is proportional to the amount of force and, therefore, the more force applied to the hand the more variable that force will be. This force increases both when longer distances are traveled in a constant time, and when a given distance is traveled in a shorter time; therefore, under these conditions errors in target acquisition would be expected to increase.

In the approach described above the errors in target acquisition were allowed to vary, and the effe IS of the movement speed on accuracy examined. In other studies (18) a fixed accuracy was required; different movement amplitudes were then combined with various accuracy requirements. The resulting movement time and the specified movement amplitude and accuracy were then combined to generate a binary index of performance (1~):

1 WS Ip = - 7 log, 2~ bits/s

where t is the average movement time in seconds, Ws the effective target width (accuracy required), and A the amplitude of movement. Ip expresses the information transmitted per unit time and, if t is the minimum attainable (i.e., if the subject moves as fast as possible), Ip expresses the information capacity of the motor system under the specified conditions of amplitude and accuracy. This maximum rate of performance increased uniformly as movement amplitude was decreased and as tolerance limits were relaxed; it was relatively constant over a wide range of accuracy and amplitude requirements at approximately 10 bits/ s (18).

This analysis considers only some of the movement variables, and so is only a partial case of the general situation of the information capacity of the motor system in controlling all the parameters of movement. Fitts (18) mentioned that “the present rationale can be extended to the question of the rate of information involved in the selection of the direction, the force or the duration of movement” (p. 387 of Ref. 18) but to our knowledge no such formulations have been attempted.

In contrast to the temporal and accuracy characteristics of aimed movements, which have been thoroughly investigated, the spatial characteristics of their trajectories have attracted less attention. Our research interest lies in the latter and particularly in study-

AIMED ARM MOVEMENTS 127

ing the factors that influence these characteristics. Furthermore, we are interested in the elucidation of the neural mechanisms of the spatial attributes of aimed movements (24, 3 1). In this investigation we studied the trajectories of aimed arm movements made on a plane by monkeys under three conditions: I) during the acquisition of an aimed motor skill, 2) under conditions of uncertainty in the spatial (“where will the target appear?“) and temporal (“when will it appear?“) domains, and 3) when the target of the aimed movement changed location during the reaction and movement time. In addition, we measured the reaction times, since it has been observed that they a) decrease with practice (41), b) increase with uncertainty (10, 26, 28), and c) are lengthened appreciably when responding to the second of two stimuli presented in quick succession (see Ref. 6 for review). In contrast to these expectations, we found that the reaction times remained virtually unchanged in all conditions above. Instead, the spatial trajectories of the hand were greatly infl- :nced and became a) less variable with practice, b) more variable under conditions of spatial (but not temporal) uncertainty, and c) fragmented in an orderly way when their target changed location. These results underscore the importance of the movement trajectories as a variable in the control of aimed movements. Moreover, they indicate that several factors can influence these trajectories without increase of the reaction time.

Preliminary results of this study were presented (23, 30, 37).

METHODS

Subjects Three male rhesus monkeys (Macaca mulatta)

were used. They weighed between 2.5 and 4.5 kg.

Apparatus The animal sat comfortably in a primate chair.

He used one arm for the task while the other was loosely restrained. He sat in front of a plane surface situated at waist level and tilted 15O from the horizontal, toward the animal. The central 25 cm x 25 cm part of the plane served both as a display and as a working surface over which the animal moved a manipulandum. Nine light-emitting diodes (LEDs) were arranged on a circular pattern on the plane as follows: one at the center and eight on the circumference of a circle of 8 cm

radius (Fig. 1). The manipulandum was a handle that the animal grasped and moved over the working surface. It consisted of two tubular portions (“forearm” and “upper arm”) joined at the “elbow,” The upper arm could rotate relative to the plane (at the shoulder) and the forearm relative to the upper arm. Shaft angle encoders were placed in the two joints (shoulder and elbow) of the manipulandum and they measured the angle of each joint. The handle the animal grasped was located at the most distal end of the forearm of the manipulandum. A transparent Plexiglas circle was attached at that end so that it moved with the manipulandum. A large circle was used in the beginning of training; it was replaced by successively smaller circles as training progressed, and was fixed at 15 mm for the study described in this paper. The readings from the shaft angle encoders were fed to a microprocessor, which provided pairs of Cartesian coordinates of the X-Y position of the center of the circle with a resolution of 0.125 mm. These coordinates were then fed to a PDPl I/ 20 minicomputer through which the experiment was controlled (see below, Data collection).

Task

The basic requirement of the task was that the animal capture within a Plexiglas circle any LED that was illuminated. Only one LED was on at any time. The sequence of events was as follows: I) the center LED was turned on at the beginning of a trial-the animal had to capture that LED within a prescribed period of time; 2) after the animal held the LED captured for a period of time (foreperiod), one of the peripheral LEDs was turned on as the center LED went off; 3) the animal was then required to capture and hold the new target LED for a period of time in order to receive a liquid reward. No restrictions on the animal’s movements were imposed. Although upper limits of reaction and movement times were specified (2 s for each), these were much larger than the actual times the animal utilized, so that the latter were the results of the animal’s own behavior rather than the result of experimental control.

An initial period was devoted to familiarize the animals with the apparatus and the task. This period varied among animals; it was about 1 wk for two monkeys, but only 2-3 days for the third. At the end of that period the animals held the manipulandum at the center for a brief period of time and then moved it toward the target light; the areas of the “center” and the “target” within which the animal had to capture the light were relatively large (circles of 30 mm diameter) and the times for which they had to hold at those positions relatively brief (0.5 s). These circles were then decreased progressively in area and the hold-


0

A

\

FIG. I. Left: diagram of the apparatus. The monkey sits at A in front of the working surface, B. On that plane are nine light-emitting diodes (LEDs); one is located at the center and eight are arranged in a circle of 8-cm radius around it. The peripheral LEDs were used as targets. They are numbered counterclockwise and will be referred to as T-l, T-2, T-3, etc. The monkey holds the articulated manipulandum at the distal end, C, which is free to move in two (X-I’) dimensions across the working surface. The monkey moves the manipulandum toward the target that appears, so that it is captured within the clear plastic circle, D, at the end of the manipulandum. Right: a monkey performing the task (side view). Insert shows two trajectories.

ing times made longer until they were fixed, and not changed for the rest of the training. The center and target windows were 15 and 30 mm, respectively, and reaction and movement time limits were 2 s each. The animals performed approximately 800 successful trials per day.

The effects of uncertainty were studied in both the spatial (location of the target) and temporal (time of its appearance) domains. In the spatially certain condition (So) the same target was presented successively over 20-30 trials, so that the target of each successive movement was predict- able; in the spatially uncertain condition (S,) the location of the target varied, being selected for each trial from the population of eight possible peripheral targets in a random sequence. In the temporally certain condition (To) the foreperiod was fixed at 2 s; in the temporally uncertain condition (T,) it varied, being selected from normal distribution with mean = 2 s, SD = 0.4 s, and

range from 0.8 to 3.2 s (i.e., mean + 3 SD). The four combinations of these factors (&7’,, SIT,, SOT,, S, T,) were presented in a randomized block design.

A modified version of the task described above was used to study the effects of change in target location. At various times after the appearance of the target (interstimulus interval) a new target was turned on while the original target light went off. Trials were presented in a randomized sequence with event probabilities balanced (Table I), so that the animal could not predict whether the target would change position and, if so, the time of change. Only one pair of target LEDs was used in each run of the task. The targets were on opposite sides of the center LED in some runs, but adjacent in other runs. At the end of the interstimulus interval the first target was turned off while the other target (“second target”) was turned on. This second target stayed on until the

TABLE 1. Probabilities of various conditions in task with change in target location

Change at Interstimulus Interval, ms No

Target Change 50 100 150 200 250 300 400 Total

A 114 l/28 l/28 l/28 l/28 l/28 l/28 l/28 112 B 114 l/28 l/28 l/28 l/28 l/28 l/28 l/28 112

l/14 l/14 l/14 l/14 l/14 l/14 l/14

Total 112 112 1

VELOCITY

HRNDPf3Tl-l VELOCITY

. i . l

l ,’

. ’

:

.

. I r r

. .

.

I T

I-- R7 + MT-1

-i b

100 msec

I Tl I T2

FIG. 2. Top: data collected from a trial of the basic task. On the left is the movement trajectory from the center

to the target (in this case T-3). The sequence of points is the position of the center of the circle on the manipulandum, sampled at HI-ms intervals. On the right is the velocity of the movement every 10 ms, starting from the time the target appeared (T). The curve was calculated by differentiating the position data. A simple algorithm determined

the reaction time (RT) as the first significant change in the velocity curve and determined the peak velocity (V). In the trial illustrated the movement was completed within a visual RT (RT = 260 ms, movement time (MT) = 240 ms). In other cases, the movement was completed later. Bottom: data collected from a trial of the target- change task, in which a target (the “first target”; in this case at 12 o’clock) appeared first and stayed on for 200 ms, then went off and the second target came on. The small cross on the movement trajectory (left) indicates the point of reversal of the trajectory, at which point the movement toward T-3 stopped and the movement toward T-7 began. On the right is the velocity curve for this trial. Tl and T2: time at which the first and second targets appeared, respectively. RTl, reaction time to the appearance of the first target. MTl, duration of the movement

toward the first target, measured from the beginning of movement to the reversal of the movement. RT2: estimate of reaction time to the second target, measured from the time of its appearance to the reversal of movement. Vl, peak velocity of the initial movement toward the first target V2, peak velocity of movement toward the second target.

730 GEURGOPOULOS, KALASKA, AND MASSEY

monkey captured it with the manipulandum. The

given in Table 1. Note that the following condi-

animal received a liquid reward when he captured the second target but no restrictions were imposed

tions were equiprobable: target A versus target B,

on his movements. Seven time intervals for which the first target stayed on were chosen to fall at

change versus no change of target, and the inter-

several points within the reaction or movement time to the first target, The probability of occur-

stimulus intervals (ISI).

rence of the various conditions in this task are

X and Y means, respectively. This analysis pro- vides a measure of the two-dimensional spatial

the sum of these areas along the trajectory was

variability of the mean trajectory along its course. A single measure of the total spatial variability

used as an indicator of the total dispersion of tra-

of movement trajectories to a target was derived

jectories for a particular target. The avertige for

by a nonparametric method as follows. For each of the 19 points, the area was calculated within which 95% of the values from different trials lay;

all targets is used in Fig. 4 to denote the handpath variability on different days of training.

Data cdlectiun The experiment was controlled through a

PDPl l/20 minicomputer. The X-Y position of the manipulandum was sampled with a rate of lOO/ s. Eye movements were recorded using implanted electrooculogram (EOG) Ag-AgCl electrodes. The sampling frequency of the X-Y electrooculogram was 100/s. Both the manipulandum and EOG data were stored on-line in digital form.

Data analysis The reaction time was measured from the ve-

locity record of the movement, which was obtained by differentiating the handle position data (Fig- 2, top): it was the time that elapsed between the appearance of the peripheral target and the first reliable change of velocity, and was measured with a precision of 10 ms. The velocity curve also provided the value and the time of occurrence of the peak velocity of the movement toward the target or toward the first and the second target in the paradigm in which target location changed.

An exponential reduction in the spatial variability of the movement trajectories was observed during the acquisition of the aiming motor skill (Fig. 4). The coefficient of determination for the linear regression of the log of handpath variability versus time was r* = 0.83 1 for the curve shown in Fig. 4, and 0.863 and 0.964 for the training curves of the other two animals (data not shown). Figure 3 shows examples of performance from early, intermediate, and late stages of training. No large changes were observed in the reaction times or peak velocities, although the latter increased by a small amount.

The animals were trained in the task using one hand. Later they were tested using the

The reaction time to the second target in this task untrained hand. The performance with the was measured from the appearance of the second stimulus tO the reversal Of direction at the end Of

untrained hand was marked initially by high

the fragmented movement. handpathvariability. However,performance

The variability in movement trajectories was improved rapidly and in a few days of train-

analyzed as follows. Each trajectory was divided ing the animals attained levels of low hand-

into 20 equal portions by 19 equidistant points; path variability comparable to those reached

each point was defined spatially by its X-Y co- in the 5th wk of the original training period. ordinates. The mean trajectory toward a target The reaction times showed no significant over a number of trials was derived from the mean change during this process.

RESULTS

Acquisit ion of to other hand

aiming skill und transfer

X and Y positions at each of the 19 successive points, averaged over the contributing trials; the dispersion of the trajectories at each point was then derived by calculating 95% confidence intervals for these means. The results of this analysis

Effects of uncertainty

are illustrated in Fig. 3, bottom; the mean X-Y Spatial uncertainty concerning the loca-

at each point is the center of an ellipse, and the tion of the target caused a significant in- horizontal and vertical axes of the ellipse are pro- crease in handpath variability (Fig. 5); in portional to the 95% confidence intervals of the contrast, temporal uncertainty concerning

AIMED ARM MOVEMENTS

Day 24 Day 35 n IOMM

FIG. 3. Top: records of movement trajectories at early (16th day, left), intermediate (24th day, middle), and later (35th day, right) stages of training in the aiming motor task from one monkey. Movements were performed from central to peripheral targets. Note the reduction of dispersion (variability) of trajectories with practice. Bottom: data above, analyzed as described in the text. Lines are the mean trajectory of movement toward each target, and the ellipses represent the X-Y variability of the recorded movements about the mean trajectory.

the time of target appearance had no effect. Reaction times and peak velocities did not change in either condition.

The quantitative relation between spatial uncertainty and handpath variability was examined in one monkey. Four levels of uncertainty were obtained by varying the number of target lights in separate runs. Presen- tation of only one target in successive trials yielded log, 1 = 0 bits of uncertainty; presentation of two targets in a random sequence yielded log2 2 = 1 bit of uncertainty; and similarly runs with random sequences of four and eight targets yielded 2 and 3 bits of uncertainty, respectively. It was found that the handpath variability (Var,) increased with increasing spatial uncertainty (H,) according to the equation

Var, = 1M + H-I, (2)

where M is a constant and b the slope of the straight line. The coefficient of determination for equation 2 was r2 = 0.792.

Effects of change in target lucation

The movement of the hand was modified in an orderly fashion when the target location changed during the reaction or movement time. The hand moved initially toward the first target for a period of time, then changed direction and moved to the second target (Fig, 2, bottom; Fig. 6, bottom). No anticipatory responses were observed (e.g., movement before the target appearance) in any of the animals.

MOVEMENT TOWARD FIRST TARGET: The duration of the initial movement was a linear function of the time that elapsed from the appearance of the first target to its substitution by the second target: the later the change occurred, the longer the movement toward the first target (Fig. 7). The shape of the velocity curve for this initial movement was similar to the corresponding part of the curve for control trials, until the animal began to decelerate in order to change


Handpath Variability

1 15

I 20

1 I 25 30

Day of Training

I I 35 40

FIG. 4. Changes in movement variables with practice. Data from one monkey. Movement trajectory variability decreased exponentially. There was no consistent change in reaction times. Peak velocities tended to increase slightly. Each point on graphs is the daily average of values derived from hundreds of successful movements performed that day.

direction toward the second target (Fig. 8). The peak velocity before the change in direction increased gradually with longer interstimulus intervals, until it reached the control value. These results show that an aimed movement can be readily modified by change in its target. Moreover, the initial movement toward the first target emitted under these conditions seems to be a fragment of the control response that would have occurred if the target location had not changed. The reaction time to the first target did not change from the control value for any of the interstimulus intervals,

In addition to the duration, the direction of the initial movement was sometimes modified at short interstimulus intervals (e.g., up to 100 ms), when targets that did not involve opposite directions of movement were used. For example, when the first target was at 9

o’clock and the second at 12 o’clock, the trajectory of the initial movement sometimes deviated toward the second target at short interstimulus intervals (and similarly for other pairs; Fig. 9). This phenomenon was not observed when the targets were located at opposite sites (e.g., at 6 and 12 o’clock, 3 and 9 o’clock, etc.).

MOVEMENTTQWARDSECONDTARGET. The hand mov&d to the second target after reversing its course (Fig. 2, bottom; Fig. 6, bottom). The time taken from the appearance of the second target to the reversal of the direction of the movement was close to the control reaction time for that target. That time tended to increase somewhat with small interstimulus intervals, but no appreciable delays were observed (Fig. 10).

The peak velocity attained on the way to the second target was generally much higher


l-hdpth Vor iabil ity

(pmNo1) Reaction Time

(N.S.)

Peak Velocity

(N.S.)

FIGl 5. Effects of spatial uncertainty on movement trajectory variability, reaction time, and peak velocity for all three monkeys. Each point is the mean effect of spatial uncertainty, observed for all movements made to targets by each monkey over severa days. Note that spatial (target location) uncertainty caused a significant increase in movement variability, but had no significant effect on reaction time or peak velocity. Temporal uncertainty had no effect (data not shown).

(up to threefold) than that of the control. of movement (29). The cause of this phe- This was evident even at short interstimulus nomenon is unclear. intervals, when the distance from the reversal point to the second target was not sub- Eye movements

stantially different from control trials. When the animals moved the handle from Therefore, these high velocities cannot be the center to the peripheral target the eyes accounted for exclusively by a mechanism usually exhibited a saccade to that target that adjusts peak velocity to the amplitude before the hand movement and remained


HRND

FIRST LfWT STRYS ON CHSEC3 5Q

t’ .

. i;

.c

L .- :

. l .

’ . - .

.r: . - . - . . - . . . . .

. l i+

f i

. . :

/

* : . \

200 250

?-

3

/\ - . : ; - : . .

. - . - l :

. -

.

: i

.

.

.

.

.

:

‘.

k

\

300

.fZ

*

4 . . . . . . l

- :

- .

. l

.

* l

. -

.

. .

: i+

.

.

.

.

.

.

.

.

\

400

FIG. 6. Hand-movement trajectories (bottom) recorded during a single cycle of the target-change task. In the trials shown, T-3 (at 12 o’clock) was the first target. On the right, seven control trials in which only T-3 appeared, are superimposed. In the other seven trials, T-3 remained on for only the period of time shown, and then a second target (at 6 o’clock, T-7) appeared. Note the increased extent of the initial movement toward the first target with the increased duration of its presentation. In other trials of the task (not shown), T-7 appeared first, followed by T-3. The same effect was observed. Eye movements (upper) were recorded simultaneously with hand movements. The eyes made saccades all the way to the first target (T-3) before moving toward the second target (T-7). There was no evidence of a gradation of the initial saccade like that observed for the hand movements,

fixated there during the hand movement. the moving limb (e.g., the animal had plain When the location of the target changed the view of his arm) so that the experimental eye and hand movement differed. The eyes conditions resembled those of everyday life. typically made a saccade all the way to the The first allowed adequate variation of the first target followed by a second saccade to movement trajectories in space, so that they the second target (Fig. 6, top). There was could be studied as a dependent variable. no gradation in the amplitude of the initial The second and third served to minimize the saccadic response, which differed in that re- transformations of information during the spect from the limb movement. S-R processing (see below) and thus to ex-

clude factors irrelevant to the simple aiming DISCUSSION function under study.

Methodological considerations The three characteristics mentioned above dis-

The major characteristics of the task used tinguish this task from step tracking tasks used

in this study were: 1) the availability of 2 by other investigators. An apparatus that allows

deg of freedom for the movement, 2) the movements in two dimensions but obstructs vision

high spatial correspondence between stim- of the hand by the subject is currently in use by

ulus and response codes (i.e., high stimulus- E. Bizzi and collaborators (unpublished obser-

response (S-R) com+tibility, see below), vations) for the purpose of studying the relations b t e ween

and 3) the lack of restriction in the infor- hand trajectories and joint angles at the

elbow and shoulder in human subjects (38). Ear- mation available to the animal concerning lier studies in the same laboratory employed an


Time for which First Target Stays On (msec)

FIG.7. Relation between the duration of presentation of the first target and the duration of the movement toward the first target. Data from one monkey. Target at 12 o’clock was the first and at 6 o’clock the second target. The same straight-line relationship was observed for other pairs of targets and in all three monkeys. (Coefficient of determination in the linear regression for the data shown, # = 0.988.)

apparatus in which the vision of the arm was also obstructed while, in addition, the motion was restricted to a single joint (at the elbow) (43). A step tracking task was used by Brooks and collaborators (9, 36) for the study of discrete arm

50 MSEC 100 MSEC

250 MSEC 300 MSEC

movements. Again, the vision of the arm by the animal was obstructed and the motion restricted to the elbow. Moreover, that task was not a visual aimed task, for there were no targets to aim at. The animal was, instead, required to emit discrete

150 MSEC

w I t I I I I I I I

400 MSEC CONTRDL -1 I-

100 MSEC

FIG. 8. Average velocity curves from a run of the target-change task in which T-2 and T-5 were targets. In the figure, the control curve (dotted line) was produced from all trials in which only T-2 appeared. The solid lines are the average velocity curves for trials in which T-2 appeared first for the indicated period of time, followed by T- 5. Note the correspondence between the initial part of the experimental curves (solid lines) and the control velocity curves (dotted lines). The same correspondence was observed in the other trials of this task, in which T-5 appeared first, followed by T-2 (data not shown). To produce the average velocity curves, all trials of the same class were oriented to the start of movement, In control trials, the averaging was continued to the point 200 ms after reaching the target. In experimental trials, the averaged curve is shown up to the point 20 ms after the reversal of the movement from the first target toward the second, hence the truncated shape of the curve.

736 GEORGUPOULOS, KALASKA, AND MASSEY

+

+ +

FIG. 9. Examples of trials in which the initial trajectory of movement deviated away from the first target toward the second. The dotted lines are the trajectories from three trials, in which the first target (T-2, right) was on for 50, 100 or 300 msec, respectively, followed by the second target (T-4, left). The solid lines are trajectories of movements toward T-2 and T-4 in control trials. Note how the initial direction of movement was rotated away from T-2 toward T-4 when T-2 was presented for only 50 or 100 ms. This effect was not observed when T-2 was presented for longer periods of time, such as 300 ms.

movements of certain amplitudes, the correct step being indicated by a tone when the desired amplitude was achieved.

Of the numerous tasks used over the past 40 odd years for the study of various aspects of reaction-time processes, few resemble our task. Most employed nonspatial codes using keyboard paradigms or verbal responses. In a few studies simple spatial tasks were used, however, most no-

tably by Fitts and his colleagues (19, 20); in fact, the task used in the present study is similar to that used by Fitts and Deininger (19). In other studies aimed movements were emitted under more complicated conditions as, for example, when they were superimposed on continuous tracking (54, 55). It is worth mentioning that aimed tasks are frequently used as behavioral tools in neurophysiological experiments. For ex-

Duration of presentation of First Target (msec)

FIG. 10. Reaction time to the second target for one monkey, measured as the interval between its appearance and the reversal of movement toward it. Note that there is no pronounced delay in response to the second target compared to control trials, although there is a slight trend for the reaction tiines to be a little longer at very short intervals. Data were for trials in which T-3 was the‘first target and T-7 was the second target. Control reaction times were from trials in which only T-7 was presented. The same result was observed in the other trials of the task, in which T-7 was the first and T-3 the second target. Control reaction times shown were of the same magnitude as those obtained to T-7 in the basic task, without change of targets.


ample, they have been employed to control the direction and amplitude of movement (13) and to study the operations of the arm in extrapersonal space (39). They have also been used in motor behavioral studies to assess deficits in movement disorders (e.g., Parkinson’s disease; Refs. 15, 22), and to study eye-hand coordination (44, 45).

There has been, to our knowledge, no de- tailed study of the fact ors that influ .ence the trajectories of aimed movements. We fo- cused on the spatial charac teristics of these trajectories i .n the present study and observed their changes I) during acquisition of the aimed motor skill, 2 ) under conditions of increased uncertainty, and 3) when the target of the movement changed unpredict- ably. We discuss below the effects of these three factors separately.

Effects of practice The spatial variability of the aimed tra-

jectories decreased with practice in an exponential fashion, which is common for training cur ves (27). BY contra St, reaction times (RT) remained Vi rtually un changed and peak veloci ties increased only by a small amount. These results show that the spatial characteristics of the aimed movements were the major subst tice l Reduction

rate for the effects of prac- of spat i al variability with

practice has also been observed in pursuit tracking tasks (3, 9). I n general, the variability of learned responses per se does not always decrease with pract ice (27 ‘); therefore, the observed reduction in the variability of movement trajectories could not have been predicted on the basis of some general learning principle. On the other handy short- ening of the RT with practice has been described (41) but was not observed in this study. In fact, the spatial effects appeared without change in the RT, even when the handpath variability was decreasing rapidly (Fig. 4).

The acquisition of the aimed skill was rather specific for the hand that was trained in it; when the animals performed with the other hand the handpath variability was high, in spite of the fact that the cognitive aspects of the task (i. e., the meaning-of the target; “move to the light that comes on”) were well learned. H owever, t here were ob- vious savi ngs in th at learning process, since levels of handpath variability similar to those of the trained hand were achieved in only a few days of training.

We postulate that at least two factors con- tributed to the decrease in trajectory variability during practice: improved motor coordination and better estimation of the appropriate aimed trajectory. We believe that both factors have to be taken into ac- count, Motor coordination alone cannot suf- fice, for there can be well-coordinated but unsuccessful movements; that is, movements that ultimately do not succeed in capturing the target although elegantly performed. On the other hand, knowledge of the best aiming trajectory is useless if the latter cannot be carried out by a well-coordinated movement. Motor coordination involving two joints can be thought of as a reduction of the degrees of freedom available for independent angu- lar movements in the two joints, as proposed by Bernstein (5) for motor learning in general. This gradual loss of independence of single muscles and joints (in the statistical sense of reduction in their degrees of freedom) would result in reduction in the trajectory variability. This hypothesis predicts progressively higher correlations between the movements of the two joints during the course of practice, so that after sufficient training the aimed movement is performed as a “whole” by the interrelated (i.e., highly correlated) actions of muscles on these two joints. This prediction can be tested exper- imentally.

The second point concerns the trajectories per se. There is an infinite number of trajectories through which the arm could capture the target, yet only some were actually used. In fact, the number of trajectories used decreased with practice, and ultimately only a few were emitted. This gradual reduction in the variety of movement trajectories can be thought of as indicating increased information by the brain concerning the best (i.e., most efficient) aimed trajectory. This idea bears close relationship to Fisher’s (17) use of information in the theory of statistical estimation of the true value of a parameter. He proposed that, given a sample of values of an unknown parameter, the uncertainty about the true value of that parameter is proportional to the variability of the values in the sample; or that, conversely, the amount of information that is provided by that sample about the true value of the parameter is proportional to the inverse of the variability in the sample. An interpretation of our re-


sults in a similar way would suggest that the brain’s information about the best aimed trajectory increases as practice progresses; in fact, a measure of this gain of information with practice in terms of the variability of the movement trajectories might provide an assessment of a particular stage in the acquisition of the aiming motor skill. Mea- surement of uncertainty in terms of variability has also been employed by Klemmer (34) in the temporal domain.

These results indicate that the spatial characteristics of the movements are important variables in the aimed task. They also show that changes in these characteristics can occur independently of the length of the RT, This indicates that the neural processes underlying these changes do not become shorter but probably are altered in their pattern, becoming more and more similar with practice.

Effects of uncertainty

The spatial variability of the aimed movements increased when the uncertainty about the location of the target was increased. This effect was specific for the spatial factor, for temporal uncertainty had no effect. This result, together with those of the preceding section, strengthen the hypothesis that the spatial characteristics of the movements are of major importance in the aiming task.

The quantitative relation between spatial uncertainty (H,) and handpath variability (Var,) was examined in a preliminary experiment and was found to be of the form

Var, = 1M + bH, (2)

or, equivalently

Var, = iI4 + b log2 n (3)

where M is a constant, b the slope of the straight line, and rz the number of alternatives. It is noteworthy that these equations are identical to those describing the effect of event uncertainty on the reaction time in choice RT tasks (26, 28; see also Ref. 21)

RT = A + 6’ log, n (4)

where n is the number of alternatives and A is a constant.

It is remarkable that spatial uncertainty did not influence the reaction time in this experiment, yet the spatial characteristics of the aimed movements were affected in a wav

quantitatively identical to that governing the relations between RT and uncertainty in choice RT tasks. These results suggest that the effects of uncertainty, in general, may be qualitatively diverse, depending on the nature of the task, but quantitatively uni- form, the change in the dependent variable being a linear function of the amount of uncertainty.

In contrast to the spatial characteristics of the movement, the reaction times of the aimed movements were virtually constant under all levels of uncertainty. This was sur- prising, for an increase of the RT with increasing uncertainty is well documented (26, 28). We postulate that two factors were re- sponsible for the lack of increase in the RT in this experiment: namely, the long practice by the subjects and the high stimulus-response compatibility of the task. Both of these factors have been shown to decrease or abolish the slope in equation 4 above, so that RT can become independent of the level of uncertainty (7, 12, 20, 40).

The concept and the effects of the so- called stimulus-response compatibility require more discussion, for this is a key factor in explaining the results of the present study concerning both the lack of changes in the RTs (referred to above) and the orderly modification of the aimed movement without undue delays, discussed below.

Stimulus-response compatibility, reactiun processes, and psychological refractory period

An important and long-lasting contribution to the motor behavioral field was made by Fitts and Seeger (20) when they drew attention to the relations between the stimulus and response codes as determinants of performance in motor tasks. They defined the key concept of stimulus-response (S-R) compatibility as follows: “A task involves compatible S-R relations to the extent that the ensemble of stimulus and response combinations comprising the task results in a high rate of information transfer. * . . This interpretation makes use of the idea of a hypothetical process of information transformation or recoding in the course of perceptual-motor activity, and assumes that the degree of compatibility is at a maximum when recoding processes are at a minimum” (p. 199 of Ref. 20). Movements aimed at targets in space form one of the most compatible sets, as these workers showed (20). In other dimensions, some highly compatible combinations are, e.g., naming arabic numerals (from one to nine) and touching.


a switch with the same finger that receives the stimulus (see Ref. 47 for review), Fitts and Dein- inger (19) predicted that the rate at which the perceptual motor system can process information would be a function of the S-R compatibility; this prediction has been shown to be correct in a crucial area, namely in the influence of uncertainty on the RT: the increase of RT (i.e., the lowering of the rate of information processing) observed with increasing event uncertainty is lowered or abolished in cases of high S-R compatibility (7, 20). This has important consequences for theories concerning the processes taking place during the reaction time.

Various theoretical models have been proposed for the choice RT processes; some postulate a si- multaneous or successive search process (26, 46), and others a sequential statistical decision process (51, 52). The idea that the RT is composed of successive processes or stages (see Ref. 50 for review,) that are initiated by the stimulus and lead to the motor response dates back to Donders ( 14). A useful conceptualization of these stages is that of Welford (59), who postulated that three basic processes take place during the RT: identification of the stimulus, translation of it into motor response (i.e., response selection), and finally, formation and execution of the motor command. Welford (58) proposed that the increase in RT with log, n is due to lengthening of the translation stage, for it is during that stage that the appropriate response is sought among those possible. Sanders (47) drew the logical conclusion from this scheme, and argued that zero slope in the RT-log* n relationship indicates that the translation stage is bypassed. This leads to the prediction that any increase in RT thought to result from lengthening of the translation process would also be reduced or abolished in tasks with high S-R compatibility. One such case is the psychological refractory period in serial RT tasks.

When two stimuli appear in quick succession and a different response to each of them has to be made, the response to the second stimulus is delayed beyond the normal RT. This delay has been called the psychological refractory period (PRP) since the early study of Telford (53). Many studies in human subjects since the pioneer study of Vince (54) have documented the PRP in a variety of motor tasks (see Ref. 6 for review) and several hypotheses have been proposed to explain it (see Refs. 6,49, 57 for review). The most widely accepted is the single-channel theory, which pos- tulates that somewhere in the central mechanisms of S-R processing there is a gated channel of lim- ited capacity that cannot handle the S-R requirements of both stimuli simultaneously; therefore, the two S-R sequences are treated sequentially and without overlap. This successive processing of information results in the delay of the response

to the second stimulus by an amount of time approximately equal to RTI - ISI (where RTI is the RT to the first stimulus, and ISI is the interstimulus interval), for the case where the second stimulus is given during the RTI. Therefore, the predicted RT to the second stimulus (RT,) would be

RT2 = RT’ + (RT, - ISI) (5)

where RT’ is the normal RT to the second stimulus when presented alone (11, 56). Delays of the predicted magnitude have indeed been observed in many studies (see Ref. 6 for review). In other studies, even longer delays were observed, which led to the proposal that the single channel may continue to be occupied during the initial part of the motor response to the first stimulus for response-monitoring purposes (56). The PRP may not be a fixed, absolute delay but may reflect a decreased ability of the single channel to process new information after the first stimulus, an ability that is probably restored gradually in an exponential way (35).

The PRP has thus received wide experimental support and it has been taken for granted that it represents inherent limitations of the perceptual- motor apparatus in handling successive stimuli presented at short ISIS. That a PRP is present under certain conditions, there is no doubt. The real question is whether it can be avoided and, if so, under which circumstances. Beyond the prac- tical viewpoint, this question raises the important theoretical issue of whether there is an ultimate limitation of the perceptual-motor mechanism in responding to a second stimulus without undue delays beyond the RT. This issue is still unsettled. In fact, some of the task conditions may them- selves impose the apparent limitations on the perceptual-motor apparatus. For example, it has been pointed out (25) that the PRP has been most prominently observed in tasks that involve responses with different anatomical units (e.g., right and left hand) and stimuli of different modalities (e.g., visual-auditory). True, these were chosen to obtain accurate measurements of RTZ and also to avoid artificial lengthening of the RT2 that may occur when the same hand responds to two con- secutive stimuli (in which case the inertia of the first response has to be overcome). Yet, these sup- posedly “clean” conditions of the keyboard paradigms that require responses from both hands may introduce new factors and, as was pointed out by Gottsdanker (25), “cannot be assumed to represent the sequence of events in the single response-unit case” (p. 238 of Ref. 25). The early study of Vince (54) and that of Ellson and Hill (16) involved step stimuli in a continuous tracking paradigm. Delayed responses to the second stimulus were found in both studies, although of smaller magnitude in the Ellson and Hill (16) study. Continuous tracking, however, is itself de-


manding for the perceptual motor mechanisms, and this may superimpose additional limitations to the system. Discrete, step tracking tasks would be more appropriate but they have not been used frequently. It is interesting that in one study with such a task, no undue delays were observed when responding to the second stimulus with the same hand (25).

In summary, then, it is widely accepted and well documented that when a second stimulus arrives while a first is being processed (i.e., during RT,), the motor response to the new stimulus is delayed. Absence of these delays have been observed in isolated instances (25) but they have been generally treated as exceptions to the general rule of PRP. We wish to propose a different hypothesis; namely, that the PRP is a quantitative variable that can be influenced by various factors. One such factor is the S-R compatibility: the higher the S-R compatibility the less the PRP; that is, the higher the rate of information processing.

Modification of aimed movement

The major finding of this study is that the rapid aimed movement of the hand was interrupted in a continuous and orderly man- ner when its target changed during the reaction and movement time, so that the initial movement became a fragment of the complete response. The duration and amplitude of the interrupted movement were linear functions of the time for which the first target stayed on. This modification occurred without undue delay beyond the usual reaction time; that is, without any appreciable psychological refractory period.

Only a small increase in RT2 (by about 59 ms) was observed at the shortest ISI tested (50 ms) (Fig. lo), which is far below the value predicted by equation 5 above (see Refs. I 1, 56) (RT2 observed = 264 ms, versus RT2 predicted = RT normal + (RTI - ISI) = 205 + (220 - 50) = 375 ms). Moreover, the small increase of RTz observed is probably more apparent than real, for we took as the beginning of the second response the reversal of movement; obviously, however, the second response had already started by decelerating the initial movement before the time of reversal. Therefore, our measurements of RT2 overestimate the true RT to the second target.

These results are of interest with regard to the issue of modifiability of rapid movements and the occurrence of a psychological refractory period in serial RT tasks, discussed above. It would be argued on the basis of the ideas reviewed in the

preceding section that the first of two fast responses in serial RT tasks cannot be modified for two reasons. First, if the new target is presented during the movement time (MT,) and the movement is fast enough to be completed within a visual RT (i.e., MT1 5 RT& then the new response will appear after the first movement has been completed, since its effects will be manifested after at least one RT. The second argument could be as follows: If the new target is presented during the RTI, the postulated single-decision channel will be occupied at least until the end of the RTI. Therefore, processing of the second stimulus will not start until the end of RTI, and its effects will thus appear after the normal termination of the first movement (if MT, 5 RTJ, or in any case, late in MT1 (if MT1 > RTJ. The prediction, therefore, is that in the limiting case of MT1 5 RT1 (see, e.g., Fig. 2, upper) the rapid aimed movement cannot be modified in its course. In fact, it was this prediction that we tested in this study and which was not fulfilled. Indeed, the aimed movement was promptly interrupted with no undue delay. Isolated instances of modifica- tions late in the first movement were evident even in the pioneer study of Vince (54) and then in the study of Ellson and Hill (16); in both of these studies, however, a definite increase in the second RT (i.e., a PRP) was present. Gottsdanker’s (25) study was perhaps the only one in which definite reduction of the ampIitude of the initial movement was seen in the absence of a PRP. Yet, the movements were very small (0.5-I inch) and were made with the fingers, so that a generalization of those findings for the arm movements could not be made. Our results provide definite evidence that aimed arm movements are modifiable without any appreciable PRP and, furthermore, provide data for the quantitative description of this relation as a linear function of the ISI. It could be argued that monkeys behave differently from hu- mans in this respect. We believe that this is not true, for qualitatively similar results (although under different circumstances) have been obtained in human subjects as well (l&25), as mentioned above.

We hypothesize that the virtual lack of PRP in our study, in contrast to the results of other studies, can be explained by the stimulus and response codes used; that is, by the high S-R compatibility present in our task. This hypothesis is based on experimental and theoretical grounds. On experimental grounds, high S-R compatibility has been shown to increase the rate of information processing, e,g., by allowing more alternatives to be processed during a normal RT.


Therefore, it would be only a step further to suppose that it would do so under other circumstances, as well; e.g., by allowing two

times, duration) of their activation. These specifications, in the aimed task, have to cor- respond to the spatial location of the target,

stimuli to be processed in quick succ&sion for the outcome of this process will be a without undue delay, thus increasing, again, the rate of information processing. On theoretical grou nds, the PRP i .s post ulated to occur d urrng the tran slation stage of the S-

movement that will bring the hand to the target. Our results indicate that these specifications can be altered if the target changes early in the RT, but that this altera tion can

R processing in the RT (58). it has been proposed, however, that this stage is by-

only occur when similar muscle groups are involved in both movements. Indeed we did

passed in the presence of high S-R compat- not observe this modification when the tar- ibility (47); therefore, a diminished or absent gets were on opposite locations on the circle. PRP would be expected under these condi- In conclusion, the results of this section

of our study indicate that the process t hat generates the aimed movement possesses the

tions, and this is indeed what we o lbserved. An indica .tion that the t .ranslation stage is bypassed is thought to be provided by a-zero slope in the relation RT = A + b log2 rt

when compatible, familiar, and/or highly

(equation 4), that is, by the lack of effect of event uncertainty on the RT, which was

practiced responses are required. Indeed, it

indeed observed in our experiments. This indicates a bypassing of the translation stage, which can explain the lack of PRP in this study. The bypassing of the tra .ns lation stage could mean that the S-R bond is very stable

in a continuous, ongoing fashion as a real- time

following characteristics. First, it is emitted

process that can be interrupted at any time. Second, it is sensitive to target-location information throughout the reaction and movement time; this informati .on has continual and effective access to the central mechanisms controlling the

motor commands already in progress. Fi-

limb motor apparatus and can interrupt and

nally, it is a highly efficient process that al-

modify the content of

was proposed by Fitts and Switzer (21) that in these cases a direct association mechanism between stimulus and response may replace the time-consuming decision mechanism usually postulated in the response-selection process.

An additional modification but of a different kind was observed when the two targets were spatially adjacent on the circle: the trajectory of the fragmented movement was sometimes rotated away from the first and toward the second target. This is an interesting instance of interaction between the first and second motor command and reflects an apparent change in the composition of the first command. The composition of a motor command can be thought of as a process that specifies the muscles to be activated and the strength and time characteristics (onset

lows prompt responses to changing targets, a property of crucial advantage in everyday

l

life .

ACKNOWLEDGMENTS

We thank Dr. Allyn Kimball for valuable advice on the statistical analysis of handpath variability in the two (X-Y) dimensions,

This research was supported by Public Health Service Grants 1 -ROl-EY03 167 and NS-07226.

J. F. Kalaska is a postdoctoral fellow of the Medical Research Council of Canada.

J. T. Massey received partial support from a William S, Parsons Visiting Professorship.

Present address of J. T. Massey: Applied Physics Laboratory, The Johns Hopkins University, Laurel, Maryland 208 10.

Received 4 November 1980; accepted in final form 27 April 1981.

REFERENCES

1. ALDRICH, J. W., LYMAN, J., AND STASSEN, H. A formal model for arm motion during target approach. IV&A Spec. Pub/. 215: 581-608, 1969.

2. ANNETT, J., GOLBY, C. W., AND KAY, H. The measurement of elements in an assembly task. The information output of the human motor system. Q. J. Exp. Psychol. 10: l-11, 1958.

3. BAHRICK, H. P., FITTS, P. M., AND BRIGGS,G. E. Learning curves-facts or artifacts? PsychoI. Bull, 54: 256-268, 1957.

4. BEGGS, W. D.A. ANDHOWARTH,C.I. Themove- ment of the hand towards a target. Q. J. Exp. Psy- ch& 24: 448-453, 1972.

5, BERNSTEIN, N. A. The Coordination and Regu- lation of Movements. Oxford: Pergamon, 1967.


6.

7.

8.

BERTELSON, P. Central intermittency twenty years later. Q. J. Exp. Psychol. 18: 153-163, 1966. BRAINARD, R, W., IRBY, T. S., FITTS, P. M., AND

ALLUISI, E. A. Some variables influencing the rate of gain of information. J. Exp. Psychol. 63: 105- 110, 1962.

BROOKS, V. B., COOKE, J. D., AND THOMAS, J. S. The continuity of movements. In: Control of Pos- ture and Locomotion, edited by R. B, Stein, K. G. Pearson, R. S. Smith, and J. B. Redford. New York: Plenum, 1973, p. 257-272.

25.

26.

27.

28.

9. BROOKS, V. B., REED, D. J., AND EASTMAN, M. J. Learning of pursuit visuo-motor tracking by monkeys. Physiol, Behav. 21: 887-892, 1978.

BROWN, I. D. Many messages from few sources, Ergonomics 3: 159- 168, 1960.

DAVIS, R. The limits of the “psychological refractory period.” Q. J. Exp. Psycho). 8: 24-38, 1956.

DAVIS, R., MORAY, N., AND TREISMAN, A. Imi- tative responses and the rate of gain of information, Q. J. Exp. Psychol. 13: 78-89, 1961.

DELONG, M. R. AND GEORGOPOULOS, A. P. Motor functions of the basal ganglia as revealed by studies of single cell activity in the behaving primate. Adv. Neural. 24: 131-140, 1979.

DONDERS, F, C. Over de snelheid van psychische processen. Onder zoekingen gedaan in het Physiol- ogisch Laboratorium der Utrechtsche Hoogeschool, l&68- 1869, Tweedereeks, II, 92- 120. (Translated by W. G. Koster. Acta Psycho/. 30: 4 12-43 1, 1969.)

DRAPER, I. T. AND JOHNS, R. J. The disordered movement in parkinsonism and the effect of drug treatment. Bull. Johns Hopkins Hosp. 115: 465-

480, 1964,

ELLSON, G. D. AND HILL, H. The interaction of responses to step function stimuli. I. Opposed steps of constant amplitude. Washington, DC: Air Ma- teriel Command, Mem. Rep. MCREXD-694-2P, 1948.

29.

10.

11.

12.

30.

aiming arm movements. I. Aiming at different tar-

get locations. Sot. Neurosci. Abstr. 6: 156, 1980. GOTTSDANKER, R. The effect of superseding sig- nals, Q. J. Exp. PsychoI. 18: 236-249, 1966. HICK, W. E, On the rate of gain of information. Q. J. Exp. Psycho/. 4: 1 l-26, 1952. HOVLAND, C. I. Human learning and retention. In: Handbook of Experimental Psychology, edited by S. S. Stevens, New York: Wiley, 1951, p. 613-689. HYMAN, R, Stimulus information as a determinant of reaction time. J. Exp. Psychol, 45: 188-196, 1953. JOHNS, R. J. AND DRAPER, I. T, The control of movement in normal subjects, Bull. Johns Hopkins Hosp. 115: 447-464, 1964.

KALASKA, J. F., MASSEY, J. T., ANDGEORGOPOU- LOS, A. P. Visually guided discrete arm movements in two (X-Y) dimensions. II. The effects of spatial and temporal uncertainty. Federation Proc. 39: 601,

1980. 31,

13.

32.

14,

33.

15,

KALASKA, J. F., MASSEY, J. T., ANDGEORGOPOU- LOS, A. P. Cortical mechanisms of two-dimensional aiming arm movements. II. Change of target location. Sot. Neurosci, Abstr. 6; 156, 1980.

KEELE, S. W. Behavioral analysis of movement. In: Handbook of Physiology. The Nervous System. Bethesda, MD: Am. Physiol. Sot. 198 1, sect. 1, vol. II, part 2, chapt+ 31, p+ 1391-1414. KEELE, S, W, AND POSNER, M. I. Processing of visual feedback in rapid movements. J. Exp. Psy- cho/. 77: 155-l 58, 1968.

34. KLEMMER, E. T. Simple reaction time as a function of time uncertainty. J, Exp. Psychol. 54: 195-200, 1957.

16. 35.

36.

17.

18.

KOSTER, W. G. AND PEACOCK, J. B. The influence of intensity of visual stimuli on the psychological refractory phase. Acta Psychol. 30: 232-253, 1969.

KOZLOVSKAYA, I., UNO, M., ATKIN, A,, AND BROOKS, V. B. Performance of a step-tracking task by monkeys. Commun. Behav. Biol. 5: 153-l 55, 1970.

37.

19. 38.

39.

20.

21.

FISHER, R, A. The Design of Experiments. London: Oliver & Boyd, 1925*

FITTS, P. M. The information capacity of the human motor system in controlling the amplitude of movement. J. Exp. Psycho/* 47: 381-391, 1954.

FITTS, P. M. AND DEININGER, R. L. S-R compatibility: correspondence among paired elements within stimulus and response codes. J. Exp. Psychol. 48: 483-492, 1954.

FITTS, P. M. AND SEEGER, C. M. S-R compatibility: spatial characteristics of stimulus and response codes. J. Exp. Psychol, 46: 199-210, 1953.

FITTS, P. M. AND SWITZER, G, Cognitive aspects of information processing: I. The familiarity of S- R sets and subsets. J. Exp. Psycho!. 63: 321-329,

1962.

40.

41.

22.

23.

24.

FLOWERS, L,. A. Ballistic and corrective movements on an aiming task: intention tremor and parkinson- ian movement disorders compared. Neurology 25: 413-421, 1975.

GEORGOPOULOS, A. P., KALASKA, J. F., ANDMAS- SEY, J. T. Visually guided discrete arm movements in two (X-Y) dimensions. III. All-or-none or modifiable? Federation Proc. 39: 601, 1980,

GEORGOPOULOS, A. P., KALASKA, J. F., AND MAS- SEY, J. T. Cortical mechanisms of two-dimensional

42.

MASSEY, J. T., GEORGOPOULOS, A. P., AND KA- LASKA, J. F. Visually guided discrete arm movements in two (X-Y) dimensions. I. Changes during motor learning. Federation Proc. 39: 601, 1980.

MORASSO, P. Spatial control of arm movements. Exp. Brain Res. 42: 223-227, 1981.

MOUNTCASTLE, V. B., LYNCH, J. C., GEORGO- POULOS, A., SAKATA, H., AND ACWNA, C. Posterior parietal association cartex of the monkey: command functions for operations within extrapersonal space.

J. Neurophysiol. 38: 871-908, 1975.

MOWBRAY, G. H. Choice reaction time for skilled responses. Q. J. Exp. Psychol. 12: 193-202, 1960.

MOWBRAY, G. H, AND RHOADES, M. V, On the reduction of choice reaction times with practice. Q. J. Exp. Psychol. 11: 16-23, 1959.

NAVAS, F. AND STARK, L. Sampling or intermittency in hand control system dynamics. Biophys, J. 8: 252-302, 1968.

43.

44.

POLIT, A. AND BIZZI, E. Characteristics of the motor programs underlying arm movements in monkeys. J. Newophysiol. 42: 183-194, 1979.

PRABLANC, C., ECHALLIER, J. F., JEANNEROD, M., AND KOMILIS, E. Optimal response of eye and hand


45.

46.

47.

48.

49.

50.

51.

52.

motor systems in pointing at a visual target. II,

Static and dynamic visual cues in the control of hand movement. Biol. Cybern. 35: 183-187, 1979.

PRABLANC, C., ECHALLIER, J. F., KOMILIS, E., AND JEANNEROD, M. Optimal response of eye and hand motor systems in pointing at a visual target. 1, Spatio-temporal characteristics of eye and hand movements and their relationships when varying the amount of visual information. Biol. Cybern. 35: 113-124, 1979.

RAPPAPORT, A. A study of disjunctive reaction times. Behav. Sci. 4: 299-315, 1959.

SANDERS, A. F. Some aspects of reaction processes. Acta Psychol. 27; 115-l 30, 1967.

SCHMIDT, R. A., ZELAZNIK, I-I., HAWKINS, B., FRANK, J. S., AND QUINN, J. T., JR. Motor output variability: a theory for the accuracy of rapid motor acts. Psychol. Rev. 86: 415-451, 1979.

SMITH, M. C Theories of the psychological refractory period. Psychol. Bull. 67: 202-213, 1967.

STERNBERG, S. The discovery of processing stages: extension of Donder’s method. Acta Psychol. 30: 276-315, 1969,

STONE, N. Models of choice-reaction time. Psy- chometrika 25: 25 l-260, 1960,

TANNER, N. P. Physiological implications of psy-

53.

54.

55.

56.

57.

58.

59.

60.

61.

chophysical data. Ann. NYAcad. Sci. 89: 752-765, 1961.

TELFORD, C. W. The refractory phase of voluntary and associative responses. J. Exp. Psychol. 14: l-

36, 1931. VINCE, M. A. The intermittency of control movements and the psychophysical refractory period. Br. J. Psychol. 38: 149-157, 1948. VINCE, M. A. Corrective movements in a pursuit task. Q. J. Exp. Psychol. 1: 85-103, 1948. WELFORD, A. T. The “psychological refractory period” and the timing of high-speed performance. A review and a theory, Br. J. Psychol. 43: 2-19, 1952. WELFORD, A. T. Evidence for a single channel decision mechanism limiting performance in a serial reaction task, Q. J. Exp. Psychol. 11: 193-210, 1959. WELFORD, A. T. The measurement of sensory-motor performance: survey and reappraisal of twelve years’ progress. Ergonomics 3; 189-229, 1960. WELFORD, A. T. On the sequencing of action. Brain Res. 71: 381-392, 1974. WELFORD, A. T., NORRIS, A. H., AND SHOCK,

N. W. Speed and accuracy of movement and their changes with age. Acta Psychol. 30: 3- 15, 1969. WOODWORTH, R. S. The accuracy of voluntary movement. PsychoI. Rev. 3: 1-l 14, 1899,

spatial trajectories and reaction times of aimed …brain.umn.edu/pdfs/ag007.pdf · spatial...

Documents