Acta Psychologica 40 (1976), 49-56 0 North-Holland Publishing Company

EVIDENCE FOR AN OUTFLOW THEORY OF SKILL

Bill JONES and Martyn R. HULME

Carleton University* and University of Waterloo

Received May 1975

We argue that central monitoring of efferent signals (CME) is a necessary and perhaps sufficient condition for the most accurate reproduction of a linear movement. In a movement duplication task where the subject sets or is given a criterion movement which he then attempts to reproduce, we argue that voluntary criterion and reproduction movements may be based upon some ‘pre-set’ instruction whereas passive movement duplication must be based upon proprioceptive information. We show that subjects can perform a visual signal detection task during the voluntary criterion movement which does not affect accuracy of reproduction. The same task carried out druing the passive criterion movement does affect reproduction. These findings suggest that pick-up of proprioceptive information is not involved in voluntary movement.

Introduction

How are movements in a skill integrated in the correctly timed sequence? Many theorists (e.g., Bahrick 1957; Schmidt 1971) have more or less automatically assumed that proprioception (which for our purposes refers to information on the direction and extent of move- ment derived from receptors in the muscles, joints, and tendons) is a necessary condition of precise timing. An opposing notion that central monitoring of efferent signals (CME), in whatever manner, is a neces- sary and perhaps sufficient condition for voluntary motor control has had off and on support at least since Helmholtz. In fact, it is now fairly well accepted that eye movements are controlled by some central or ‘outflow’ system rather than a proprioceptive feedback or ‘inflow’ system (see e.g., Festinger and Canon 1965) and Jones (1974a) has

* Address for reprints: Bill Jones, Department of Psychology, Carleton University, Ottawa, Canada.

50 B. Jones, M. R. HulmejAn outjlow theory of skill

reviewed evidence to suggest that the outflow or CME hypothesis is a useful starting point for understanding movement control in general.

The present experiment is based upon the movement duplication paradigm in which the subject sets or is given some criterion movement which he then attempts to reproduce. Jones (1972, 1974b) has shown that subjects can more accurately duplicate a purely voluntary move- ment than a passive movement or one made to a stop set by the experimenter. Moreover, voluntary movements, though not movements in either of the other two classes, may be retained with no loss of accuracy over an unfilled retention interval (Jones 1974b). Retention of voluntary movements required ‘central capacity,’ in Posner’s ( 1967) operational sense, since reproduction accuracy dropped when the subject had to perform an attention demanding task during the reten- tion interval. Jones argues that when the subject makes a voluntary movement he is in a position to predict the extent of movement prior to moving. Both criteron and reproduction movements may be entirely ‘pre-set’ and the subject could use the set of ‘criterion instructions’ to perform more or less the same movement again without relying upon proprioceptive or other feedback. It follows that the subject may have spare capacity during the course of a voluntary criterion movement which is not available when he must depend upon proprioception for knowledge of the movement.

Experiment 1

To test the spare capacity notion we compared the effect on reproduction of having to

perform a signal detection task during criterion movements which could be either voluntary or

passive. Detection performance was assessed by means of the choice theory parameters (Lute

1963) of sensitivity (n) and response bias (b)-distribution-free indices computed directly from

the Yes-No matrix without manipulation of biases or confidence ratings. Accuracy of move-

ment reproduction was the mean absolute error for each S. Since it has been argued by many

(in most detail by Schutz and Roy 1973) that variable error (the within-S standard deviation

around the constant error) is more appropriate, our use of absolute error requires some

comment. Briefly, variable error is the best measure of precision when performance on any

given trial can be taken as dependent upon performance on previous trials; e.g., if the S aims at

the same target or reproduces the same movement over a series of trials. However, when the 5,

as in the present experiment, reproduces a different criterion on each trial, each trial is

essentially independent of previous trials. There exists therefore only one measure for each trial _ the extent to which the reproduction movement differs from the criterion (see Jones 1975,

for a fuller treatment of this argument).

B. Jones, M. R. Hulme/An outflow theory of skill 51

Method

Subjects

Ten right-handed undergraduates took part in the experiment as a general course require-

ment.

Apparatus A freely moving slide with handles which could be grasped by the S or by the E was

mounted on a straight track. In the Passive condition the slide could be moved by the E to a

stop mounted on the track. A pointer attached to the slide moved along a scale and allowed the

E to read off the extent of displacement. A microswitch wired to close when the slide moved

away from the rest position was fixed at one end of the track.

The slide was positioned to the right of the s’s body, parallel to a three-field tachistoscope

(Scientific Prototype Model G.B.). The S sat in front of the tachistoscope and looked into the

viewer which completely blocked sight of the limb moving along the slide. The microswitch on

the slide was connected to the start inputs of the tachistoscope. Timers in the tachistoscope

were wired in a sequence so that the slide moving away from the rest position closed the

microswitch, thus exposing the target field. The start inputs were then switched off for 12 sec.

This period permitted a visual detection task to be presented during the criterion movement but

not during reproduction. Luminance levels in all fields of the tachistoscope were equal. In front

of the tachistoscope, by the S’s left hand, was a two-way switch used in the detection task. The

S wore headphones which delivered a 70 dB white noise mask; thus preventing any auditory

feedback from the slide.

Procedure The S sat in a chair in front of the tachistoscope and looked into the viewer while grasping

the handle of the slide with his right hand. There were two criterion movement conditions. In

the Voluntary condition the S moved the slide along the track to whatever position he

preferred, returned the slide to the rest position and attempted to reproduce the initial

movement. He was instructed to adopt a range of criterion movements within approximately

5 cm to 60 cm. In the Passive condition the S grasped the handle of the slide which the E moved along the track to some pre-set position. The E returned the slide to the rest position

and the S attempted to voluntarily reproduce the passive criterion.

On half the trials under each movement condition the S was required during the per-

formance of the criterion movement to detect the presence (Signal) or absence (Noise) of a gap

in the centre of the top edge of a square presented in the tachistoscope between two fixation

dots in the target field. The square was presented for 10 msec, 150 msec after the beginning of

the criterion movement. The gap subtended 8’ of visual angle and the square subtended 1” 12’

of visual angle. In detection trials half the trials at random were Signal and half were Noise.

With his left hand the S responded ‘Signal’ by pushing the two-way switch to the right and

‘Noise’ by pushing the switch to the left. In the No Detection condition the S looked into the tachistoscope as before. The tachistoscope simply presented an empty target field on each trial.

Both criterion movement and detection conditions were within-S variables. Voluntary and

Passive conditions were counterbalanced in ABBA fashion. Within this sequence Detection or

No Detection trials were also counterbalanced ABBA. There were 80 trials in each movement

block and 20 trials in each Detection-No Detection block (320 trials in total for each subject).

That is, the order of blocks of movement conditions was Voluntary, Passive, Passive, Voluntary

52 B. Jones, M. R. Hulme/An outjrow theory of skill

and within each block No-Detection, and Detection ‘blocks were presented. To ensure approxi- mately equal ranges of criterion displacements in the voluntary and Passive conditions, the order of criterion movement extents in each Passive block followed the order of the corre- sponding cell of the Voluntary block. For example, the order of criterion movements in the first Detection block under the Passive movement condition followed exactly the order of criterion movements initially made by the S in the first Detection block under the Voluntary condition.

Knowledge of results was not given for either the movement or the detection task. The S was instructed to be as accurate as possible in both tasks and to perform the detection task as quickly as possible. Brief rest periods were given every 40 trials.

Results

Mean absolute errors for the different movement conditions are given in table 1. A 2 (criterion movement conditions) X 2 (Detection-No Detection conditions) analysis of variance with both factors as within-5 variables showed that both main effects and the interaction reached statistical significance: movement conditions, (F = 64.79, df = l/9, p<O.OOl), Deteo tion-No Detection (F = 10.79, df = l/9, p<O.Ol) and the interaction (F = 11.50, df = l/9, p<O.Ol). Newman-Keuls com_parisons showed that performing the detection task had a signifi- cant effect only on the reproduction of passive movements (p <O.Ol).

Table 2 shows mean values of n and b for the two movement conditions. There were no statistically significant differences between conditions in either sensitivity. (t = 1.95, df = 9) or bias (t = 0.04,df = 9).

Table 1 Mean absolute error and standard deviations (cm) for each condition (experiment 1).

Criterion movement

Voluntary Passive M SD M SD

Detection 1.28 0.53 2.25 0.70 No-Detection 1.07 0.39 1.59 0.40

Table 2 Mean sensitivity (n) and response bias (b) for the detection task in the two movement conditions (experiment 1).

9 b

Voluntary 0.32 1.29 Passive 0.43 1.29

B. Jones, M. R. Hulme/An outflow theory of skill 53

Experiment 2

procedures which conceivably allowed Ss to use both distance and location cues in reproduc- tion though Jones (1974a) has shown that eliminating location cues has no effect on the short-term retention of voluntary movements. In the second experiment, therefore, we compared voluntary and passive movement under the previous conditions except that the S had only distance cues available to him.

Method

Subjects The 6’s were 14 undergraduates who took part in the experiment as a general course

requirement.

Apparatus The apparatus was that used in the previous experiment except that a series of 5 micro-

switches were fitted along the track to allow the use of a number of different starting positions.

Procedure The procedure exactly followed the previous experiment except that the starting position of

each criterion movement was randomly chosen from the available five positions and the slide was returned before. The reproduction movement to one of the positions again chosen at random. This procedure ensured that the S was reproducing only distance information.

Results

Table 3 gives mean absolute errors and standard deviations for the different conditions. A 2 (criterion movement conditions) X 2 (Detection-No conditions) analysis of variance with both factors as within-S variables showed that both main effects and the interaction were significant: Movement conditions (F = 232.12, df = l/13, p<O.OOOl); Detection-No Detection (F = 58.80, df = l/13, p<O.OOl), and the interaction (F = 43.04, df = l/13, p<O.OOOl). As in the previous

Table 3 Mean absolute error and standard deviations for each condition (experiment 2).

Criterion movement

Voluntary Passive M SD M SD

Detection 1.19 0.21 2.53 0.20 No-Detection 1.15 0.17 1.17 0.23

54 B. Jones, M. R. HulmejAn outflow theory of skill

Table 4 Mean sensitivity (q) and response bias (b) for the detection task (experiment 2).

rl b

Voluntary Passive

0.35 1.11 0.41 1.03

experiment Newman-Keuls comparisons showed that the detection task significantly increased only the error of reproduction of passive movements @<O.Ol).

Table 4 shows mean values of q and b for the two movement conditions. Once again there were no statistically significant differences between conditions in either sensitivity (t = 0.71, df = 13) or bias (t = 0.07, df = 13).

Discussion

The results of both experiments corroborate our hypothesis that reproduction of voluntary movement is based upon some centrally monitored set of instructions for each movement which may be ‘pre- set’ in advance of movement and that reproduction may not require the pick-up of proprioceptive information. When proprioceptive informa- tion is a necessary condition, as in the passive task, the signal detection task interferes with movement performance to a significant extent. The same signal detection task did not effect reproduction of the voluntary criteria, suggesting that subjects may not have to pay attention to any sensory inflow during the voluntary movement. It cannot be argued that subjects making voluntary movements have chosen to trade-off accuracy in the movement task at the expense of detection accuracy since in neither experiment does detection performance differ between the two movement conditions. (In fact given a one-tailed test detection performance in experiment 1 during voluntary movement would be significantly more accurate than passive movement.)

Physiologically our outflow hypothesis assumes that efferent signals are monitored or stored prior to discharge. Ruth (1965) noted that the cerebellum is reciprocally linked with the motor cortex in such a way that the cerebellum has both facilitatory and inhibitory effects on movement initiated by the motor cortex. He argued that the central nervous system may pre-program movements if the motor cortex discharged efferent impulses into a circular loop involving the cerebel-

B. Jones, M. R. Hulme/An outflow theory of skill 55

lum. Thach (1970a, b) has recently shown that changes in the frequen- cy of discharge of some nuclear cerebellar cells occur well before movement. He argues that cerebellar activity could modify cortical output after its initiation but prior to movement.

Schmidt (1973) has suggested that the motor programming notion (e.g., Keele 1968) can explain the difference in the duplication of voluntary and passive movements without involving the CME idea. Output from the program would presumably be contingent upon proprioceptive feedback to the cortex or cerebellum. However, we would argue from our present results that since interference with proprioceptive perception does not effect reproduction of voluntary movement, the motor program can be ‘run off’ without requiring proprioceptive feedback. It is not surprising therefore that augmenting proprioceptive feedback does not increase the accuracy of voluntary movement reproduction (Jones 1974a).

To argue that proprioceptive inputs may not be required for motor control is not of course to deny the importance of exteroceptive (principally visual) feedback for motor control. Movements may be ‘sketched in’ in something like the way Ruth (1965) suggests and afferent sensory information would fine down control. If we consider CME in terms of some cortico-cerebellar loop it is clear that delayed efferent commands could constitute a motor program prior to any muscular contraction. In other words, the motor program would be based upon the subject’s intention to act rather than upon proprio- ceptive information about the movement.

References

Bahrick, H., 1957. An analysis of the stimulus variables influencing the proprioceptive control of movements. Psychological Review 64, 324-328.

Festinger, L., and L. K. Canon, 1965. Information about spatial location based on knowledge about efference. Psychological Review 72, 378-384.

Jones, B., 1972. Outflow and inflow in movement duplication. Perception and Psychophysics 12,95-99.

Jones, B., 1974a. Role of central monitoring of efference in short-term memory for move- ments. Journal of Experimental Psychology 102, 37-43.

Jones, B., 1974b. Is proprioception important for skilled performance? Journal of Motor Behavior 6, 33-45.

Jones, B., 1975. What is the best measure of movement accuracy? Perceptual and Motor Skills, in press.

56 B. Jones, M. R. Hulme/An outjlow theory of skill

Keele, S. W., 1968. Movement control in skilled motor performance. Psychological Bulletin 70, 387-403.

Lute, R. D., 1963. Detection and recognition. In: R. D. Lute, R. R. Bush, and E. Galanter (eds.), Handbook of mathematical psychology. New York: Wiley.

Posner, M. I., 1967. Characteristics of visual and kinesthetic memory codes. Journal of Experimental Psychology 75, 103-107.

Ruth, T. C., 1965. Basal ganglia and cerebellum. In: T. C. Ruth and H. D. Patton (eds.), Physiology and biophysics. Philadelphia: Saunders.

Schmidt, R. A., 1971. Proprioception and the timing of motor responses. Psychological Bulletin 76,383-393.

Schmidt, R. A., 1973. Proprioception versus motor outflow in timing: A reply to Jones. Psychological Bulletin 79, 389-390.

Schutz, R. W., and R. A. Roy, 1973. Absolute error: The devil in disguise. Journal of Motor Behavior 5, 141-154.

Thach, W. T., 1970a. Discharge of cerebellar neurons related to two maintained postures and two prompt movements. I. Nuclear cell output. Journal of Neurophysiology 33,527-537.

Thach, W. T., 1970b. Discharge of cerebellar neurons related to two maintained postures and two prompt movements. II. Purkinje cell output and input. Journal of Neurophysiology 33, 537-547.