Experimental BrainResearch
Exp Brain Res (1989) 75:280-294
9 Springer-Verlag 1989
Trajectory control in targeted force impulses VI. Independent specification of response amplitude and direction M. Favilla*, W. Hening**, and C. Ghez Center for Neurobiology and Behavior, New York State Psychiatric Institute, College of Physicians and Surgeons, Columbia University, 722 W. 168th Street, New York, NY 10032, USA
Summary. The preceding study of this series (Hening, Favilla and Ghez 1988) examined the time course of the processes by which human subjects use information from a target to set the amplitude of an impulse of isometric elbow force. In that study, subjects were provided with separate cues to time response initiation and to inform them of the required amplitude of the response. When the time between target presentation and response initiation was too brief for them to incorporate information from the target, subjects produced default responses whose amplitudes reflected their prior experience. At longer latencies, subjects specified response amplitude with a gradual time course, starting earlier and ending later than an average reaction time. The present study now examines how two distinct response features, amplitude and direction, are specified following presentation of a target. We sought to answer three main questions. What are the features of responses that are produced before target information is available? Are direction and amplitude specified serially or in parallel? Does the specification of one response feature interfere with the specification of the other? Six normal subjects were studied. They were trained to initiate impulses of isometric elbow force in synchrony with the last of a predictable series of regular tones. The amplitudes and directions were to match those of visual targets requiring flexions or extensions with one of three amplitudes. The targets were presented at random times (0M00 ms) before the last tone. Target directions and amplitudes were either predictable (simple condition) or unpredictable (choice condition). In the Present addresses: * Istituto di Fisiologia Umana, Universitfi di Ferrara, V. Fossato di Mortara, 64B, 1-44100 Ferrara, Italy 9* Department of Neurology, Lyons Veterans Administration Medical Center, Lyons, NJ 07939, USA Offprint requests to: C. Ghez (address see above)
simple condition, response amplitudes and directions were independent of the interval between target presentation and response onset (S-R interval). In the choice condition, both amplitude and direction varied with the S-R interval. At short S-R intervals (< 100 ms), the direction of the subjects' responses was not related to that of the target. The amplitudes of the responses were near the centers of the two target ranges. With increasing S-R intervals, the proportion of correct direction responses gradually increased. Over the same range of S-R intervals, the amplitudes of both right and wrong direction responses to the different targets separated and converged on their respective target amplitudes. Specification of both direction and amplitude was complete at S-R intervals greater than 300 ms. The time course of amplitude specification in this bidirectional paradigm was prolonged over that in a paradigm where response direction was predictable. As in our previous reports, subjects varied response amplitude by adjusting the time derivatives of force rather than force rise time. Response trajectories were similar for flexor, extensor, right and wrong direction responses. We conclude first, that the amplitudes of impulsive responses to unpredictable targets are specified from an initial default value even when both amplitude and direction are unpredictable; second, the amplitude and direction of such response are specified gradually by separate processing channels operating in parallel; third, the processing of the two response features is not, however, fully independent. It is suggested that the two processes share a common neural resource. Key words: Human subjects - Trajectory formation Isometric - Accuracy - Reaction time - Targeted responses - Sensorimotor processing
281 Introduction
The present series of studies has been conducted to characterize the processes by which human subjects use information from a visual target to accurately specify the size and direction of a transient impulse of flexion or extension force produced at the elbow. In previous reports we focused on how the amplitude of this simple response is controlled. We demonstrated that accuracy depends upon two sets of mechanisms; one controls the execution of a desired trajectory (Gordon and Ghez 1987a, b), and the other specifies the features of the desired response in accord with information obtained from the target (Hening, Vicario and Ghez 1988; Hening, Favilla and Ghez 1988). Trajectory control is achieved through the use of a general motor program for varying response amplitude and through mechanisms for correcting errors in implementation. Under a wide variety of conditions, subjects control the amplitude of the force by varying the rate of rise of force rather than by systematically adjusting the force rise time. In addition to this pulse height control policy by which a common trajectory waveform is scaled to the target (Gordon and Ghez 1987a), subjects make small, compensatory adjustments to the force rise time in order to correct initial errors in the trajectory (Gordon and Ghez 1987b). We have used the term specification to refer to the processes that set response parameters according to information from a target. To analyze the time course of amplitude specification, we used a new paradigm which brings the time available for processing target information under experimental control (Hening, Favilla and Ghez 1988). In this "timed response" paradigm, subjects initiate their responses in synchrony with a tone occurring at a predictable time, while attempting to scale their response to a visual target presented at variable times prior to the tone. We demonstrated that specification of the amplitude of a force impulse to unpredictable targets is a gradual process. This process involved both the progressive specification of the parameters of the motor program, indicated by the scaling of the initial phase of the trajectory and trajectory updating, presumably by continued processing of target information. During the period following target presentation, the amplitudes of the responses were initially clustered around a single default value, located at the center of the range of targets. As more time was available for processing, this value was replaced by ones that were progressively more accurate (Hening, Vicario and Ghez 1988; Hening, Favilla and Ghez 1988). Our results indicated that the neural processes that trigger response initiation can be independent of those that
specify response amplitude. The two processes can act as parallel channels processing visual information from the target. We now examine the specification of response direction in order to determine whether direction and amplitude, distinct features of the response, can also be specified in parallel. This would contradict the widely accepted notion that these features are specified sequentially. One such view, derived from reaction time experiments, has held that when both amplitude and direction are uncertain, the direction of the response needs to be determined before commands for amplitude can be elaborated (Megaw 1972). Other experiments, notably ones in which particular features of the response may be precued (Rosenbaum 1980; Bonnet et al. 1982; Zelasnik 1981; Stelmach et al. 1987), have supported the serial specification model (cf. however, Goodman and Kelso 1980). However, those results do not support an obligatory order of processing for different response features (Stelmach et al. 1987). For the present experiments, we have used the timed response paradigm introduced in our earlier study (Hening, Favilla and Ghez 1988) to determine the time course over which both the direction and the amplitude of an impulse of force are specified. Subjects were presented with targets requiring responses of three possible amplitudes in both the flexion and extension directions. These targets were either predictable or unpredictable. We sought to address three questions. First, before information about direction and amplitude are available, do subjects guess among the alternative responses, cocontract antagonist muscles to produce responses of intermediate direction or produce responses with default parameters? Second, are direction and amplitude specified serially or in parallel? We reasoned that if the specification of one feature was predicated on the prior specification of another, the specification of this feature would be delayed until that of the other was complete. Third, does the specification of one response feature interfere with the specification of the other? To address this issue we compared the rates of amplitude specification when only amplitude was uncertain to that when both amplitude and direction were uncertain. A preliminary account of this work has been published (Favilla et al. 1985). Methods
The present series of experimentswas conductedusing the same six normaladultsubjects (fourmen and two women,ages31 to 42) as in our previousstudyof isometricelbowflexions(Heninget al. 1988). All subjects, therefore, had prior experiencewith the task of producing force impulses at the elbow and had learned to
282
4, I
4, . . . .
Tone # 1
i
2
4, . . . .
I
4, . . . .
I
3
. . . .
!
4
0.5 S
Target
Force
S- R Interval / ~
.
-/,~,
.~,,,9
~.~
~
e/,,~
e.,O'.,4,
~
fO",,,'I
Fig. 1. Bidirectional timed response paradigm. Four successive tones of increasing pitch (above) allow subject to synchronize response initiation with the last tone. The target level (hatched area) shifts to one of six possible levels9 These levels consist of the same three target amplitudes in the flexion and extension directions. Interval between target shift and last tone is randomly varied between 0 and 400 ms. The S-R interval is the actual time from target shift to response onset
synchronize their responses with an auditory cue. Before data was collected for detailed analysis in the present experiment, however, all subjects were given one additional session of training with the 9new bidirectional task (Bidirectional Timed Response paradigm, Fig. 1). Since the apparatus, the task of producing force impulses and our analytic procedures, have all been previously described in detail (Ghez and Gordon 1987; Hening, Favilla and Ghez 1988) we now describe these aspects of our methods more briefly. Subjects sat with their right shoulder abducted to 70~, elbow flexed to 90~, and the arm and wrist immobilized with a series of padded but rigid metal restraints. A system of strain gauges measured the forces applied to the wrist cuff by elbow flexors and extensors. Subjects viewed two oscilloscopes. On the upper oscilloscope, a target level and the force registered by the strain gauge were displayed at a fast sweep speed and appeared as two horizontal lines. The force line moved up and down with flexor and extensor forces respectively. As in the earlier studies of this series, the required response was a single uncorrected impulse of force with as brief a rise time as possible. Subjects were instructed to allow the force to return passively to baseline after it had reached its peak. A storage oscilloscope, placed below the monitor, allowed subjects to check the timing and the trajectories of their responses.
The bidirectional timed response paradigm We used distinct auditory and visual cues to control the time of response initiation and to provide amplitude and direction information (Fig. 1). Response timing was controlled by training subjects to initiate force impulses in synchrony with the last in a series of four predictable 20 ms tones presented through earphones. Tones were separated by constant intervals of 500 ms and their intensity was adjusted to a comfortable level, approximately 40 to 50 dB above hearing threshold 9 To facilitate the prediction of the last tone, successive tones in the series had a progressively higher pitch (increasing from 625 to 1300 Hz). The visual cue signalling the desired response amplitude and its direction was a
step change in the horizontal target level displayed on the monitor oscilloscope. When the target line moved upward, flexion responses were required and when it moved downward, extension responses. Subjects were instructed to just match this target level with the peak of a correctly timed force impulse. The amplitudes of the target steps were the same in both flexion and extension and required forces in a ratio of 1 : 3 : 5. The largest force required was chosen at a level which did not produce significant fatigue. This level was 30% to 50% of the subjects' maximal isometric force. The time interval between the visual target step and the fourth tone was varied unpredictably from zero to 400 ms. In the course of a single trial, subjects would first align their force with the target for four seconds. Then, as shown schematically in Fig. 1, the sequence of ascending tones was initiated. Between the third and fourth tones, the target was stepped to a new level. Subjects initiated the required force impulse at the time of the fourth tone. The interval between the step change in the target level and the time of onset of the subsequent response was defined as the Stimulus-Response or S-R interval. It represents the time which the subjects had available to incorporate amplitude information from the target into their responses. The S-R interval differs from the conventional reaction time in that response onset is controlled by the experimenter and, therefore, does not depend upon a decision by the subject concerning when to initiate a response. Testing sessions consisted of six to ten blocks of 60 trials. The six target amplitudes were presented in both predictable (simple) and unpredictable (choice) order. Since behavior in the simple blocks was found to be quite consistent and relatively independent of S-R interval (Fig. 2), the majority of blocks were in the choice condition. Typically, sessions consisted of one block for practice, two simple blocks, and four to eight choice blocks. Choice blocks were presented later to minimize performance deficits due to lack of intrasession practice. The simple blocks consisted of the six target steps presented in a fixed order (1, 2, 3, 4, 5, 6, 1, 2 ...) while choice blocks consisted of the targets presented in a pseudorandom order. We presented the simple targets in a repeating series rather than in blocks of uniform amplitude to limit the distinction between simple and choice to that of predictability alone.
Data acqu&ition and analys& Experiments were controlled and data were acquired and analyzed using a general purpose computer (PDP 11/23, Digital Equipment Corporation). The time of target presentation and its amplitude were collected along with the subjects' force response and biceps and triceps electromyographie (EMG) activity recorded with surface electrodes (Boston Elbow Myoelectrodes, Liberty Mutual). Automatic computer programs marked the onset and peaks of the force trajectories and reduced the trial data to arrays of variables which described trajectory parameters (direction, amplitude, and timing of the peaks of force and of its first two time derivatives, and S-R interval)9 Trials with responses showing more than one peak in the first time derivative of force (dF/dt) or those with evident changes in direction were assumed to show evidence of voluntary corrections and were rejected. The arrays of trajectory variables were then subject to statistical (including multiple regression) analysis9 Data from individual sessions were first analyzed separately. Then, in order to achieve the highest resolution in determining the time course over which responses were specified, data from up to four experimental sessions of each subject were grouped. In addition, since we found that the shape of flexion and extension trajectories did not differ (see Results), for some analyses we also grouped responses made to targets of each amplitude regardless of direction. Curves were fit to bivariate
A
a
I f Ld _-I
I I o I
9 o_0
o o .
50
o .
.
.
B
b o
9
0
o
0
o
~
o
:: /
o
x
o
b
a
I |
...~___*....++*_'__~Z_.*__". -~- ' ~ "
.
283
x
K
~
OX
il
x x
xr~
X
I ! X
/ J
OX
x~
x X
/\
':\
X
/
'-
i
\
30 ox~
Z v
i0
.
x I
i
x+
9
.'1
oS*
.... %~..~r.:.~. ..
o~ 0
!
i
i
100
200
300
0
!
I
400
500
B lOO
!
0
E
100
I
200
I
300
!
400
100
80
O r
/,,aa.. . . . . . ~aa
,~
80
60-
• , • =s t* 9# 9
//;:.,:>-
' ~ 40139 =o
/,..~--:-,,.-o ~176 9 ,"
60
,,,,,,~176 .... .,,,,
40
..-
~
.-~ 2 0 -
20
> ....
0 0
9
100
-,
200
9
300
,
0
!
500
400
i
I
100
200
i
300
I
400
S-R Interval (ms) C
100 cO
o 9"- 80 s .... "
,"
'
:
60
/
9149
,",/ ..
0
..... ~. . ....
.... 0---
4
-.-...m-.-.
..*" ...-'"
","
"A
,.,s
. . . . .
40
........ t
S J :.," ~,~,,'.~"~ t/_,~:"
0 ,,. , . . ~ ;~~J Z' ""
(.9
....... . . . . . m----
Jr/
i
i
i
I
i
100
200
300
400
500
S-R Interval (ms)
have been averaged within the three latency intervals indicated as a, b, c in part B. The trajectories of responses initiated at S-R intervals of less than lOOms are similar for all three amplitudes. Responses to the three targets are separated for S-R intervals of 100 to 200 ms. Responses initiated at S-R intervals > 200 ms are closely matched to their targets and, since very few responses in the wrong direction occurred (and those only to a single target see part Ac), no average is shown. These findings indicate that, in this paradigm requiring a choice of both direction and amplitude, the specification of amplitude is gradual and begins from two default values of similar amplitudes but opposite directions. The progressive specification of the amplitude of
$2 $3 $5 $12
$13 $14
Fig. 8A-E. Response specification in the choice condition: all subjects. Responses have been grouped by 80 ms S-R intervals and the different statistics plotted against the midpoint of the inerval. A-C Right Direction Responses. The percent of variance in (A) peak force and (B) peak dZF/dt2 accounted for by the target and (C) the percent of right direction responses are plotted against the midpoints of S-R intervals. D, E Wrong Direction Responses. The percent of variance in (D) peak force and (E) peak d2F/dt 2 accounted for by the target are plotted against the midpoint of S-R intervals for intervals containing sufficient responses to correlate target amplitude and trajectory parameters. Total N's by Subject, right direction/wrong direction: $2,932/435; $3,483/145; $5,634/187; S 12, 787/113; S13,795/352; S14,231/186. Range of N's within individual intervals: right direction, 17 to 197; wrong direction, 16 to 216
wrong direction responses shows that amplitude specification is not dependent on the correct specification of direction.
Time course of direction and amplitude specification In order to obtain a more quantitative representation of the time course of direction specification and to compare the rate of amplitude specification in right and wrong direction responses, we grouped together responses initiated within discrete latency intervals. We then performed a linear regression on the data in each interval, as in our earlier study (Hening, Favilla and Ghez 1988). We used the percent of variance in
289 response amplitude accounted for by variation in the target (the squared correlation coefficient x 100) to estimate the degree to which target amplitude is reflected in response amplitude. This is a measure of the degree of amplitude specification achieved. The percentage of directional errors was also computed for each interval. Figure 7A shows the percent of variance in peak force accounted for by target amplitude plotted against the midpoint of successive 40 ms intervals. This figure is based upon the choice responses shown in Fig. 6. Right and wrong direction responses are again plotted separately. Simple responses to predictable targets (same data as Fig. 2) are also shown as controls. In the simple condition the percent of variance accounted for by the target does not vary systematically with S-R interval. In the choice condition the variance accounted for is close to zero for short S-R intervals. As the S-R interval increases beyond 100 ms, the variance accounted for gradually increases and, for S-R intervals greater than 200 ms, it approximates that of the simple responses. The time course of specification is similar for right and wrong direction responses. The percentage of responses made in the correct direction for the same data is plotted in Fig. 7B. The percentage of responses in the correct direction is significantly different from chance levels (empty triangles) for the first time in the interval centered on 100 ms (P < 0.01). Then, it progressively increases and eventually reaches 100% in the interval centered around 260 ms. The time course of direction specification therefore overlaps that of amplitude specification. Figure 8 depicts the time course of specification in all the subjects. Parts A and B show the time course of amplitude specification of right direction responses for peak force and for its second time derivative, respectively. The values of these peaks in the early derivatives of the responses must be centrally programmed, because they occur too soon in the response (< 50 ms after onset) to reflect the activity of corrective processes (Gordon and Ghez 1987b). A slight delay in the specification of the peaks in d2F/dt2 relative to the peak force is present here, much as we found in our earlier report of unidirectional specification (Hening, Favilla and Ghez 1988). The progressive specification of the peaks in d2F/dt2 indicates that the gradual specification of amplitude is achieved in part by a progressive increase in the accuracy of the earliest Components of the trajectories. The time course of direction specification is shown in part C. Specification of the peak force and peak d2F/dt 2 of wrong direction responses is shown in parts D and E respectively. All subjects show significant specifica-
tion in the amplitude of the responses made in the wrong direction in at least one 80 ms time interval (P < 0.01). Specification of peak force is always greater then that of its second time derivative in both correct and incorrect direction responses. Therefore it appears that trajectories are corrected to better match target amplitude, irrespective of whether direction is correct or not. We found that the rate of response specification sometimes differed slightly for the two directions. In five subjects, the degree of specification of extensor responses was significantly less than that of flexor responses for some of the targets. This difference could reflect the greater experience that our subjects had in producing flexor force impulses. In addition, in every subject wrong direction responses were specified more slowly than right direction responses.
Interactions between amplitude and direction specification Although our observations indicate that the specification of response amplitude is not dependent upon the correct specification of direction, the data presented above does not exclude the possibility that these processes might interact in other ways. In order to assess this possibility we compared the time course of amplitude specification in the presence and absence of a concurrent requirement to specify response direction. 'This was done by comparing the results of the present study with those obtained in our previous unidirectional experiment involving the same subjects (Hening, Favilla and Ghez 1988). We found that while amplitude was gradually specified in both experiments, the time course of amplitude specification was somewhat prolonged in the bidirectional condition. Figure 9 shows the percent of variance in response amplitude accounted for by the target plotted against S-R interval under the two conditions in subjects $2 and S12. In three subjects, both the rate of specification and the degree of specification achieved after 250 ms were lower in the bidirectional than in the unidirectional condition. The data from $2, shown in Fig. 9A, typifies this group. Amplitude specification in the bidirectional condition lags that in the unidirectional condition and even at long S-R intervals a lesser degree of specification is achieved. (S13 and S14 showed the same behavior.) In the remaining subjects, a prolongation in the time needed for specification could be documented only from the peak d2F/dt 2. Figure 9B, taken from subject $12, is representative of this group of subjects. The filled squares replot the data from Fig. 7 while empty squares were obtained in the previous series of experiments, in which targets in
290 A
Discussion
1oo
80
60
Unidir Force
,~
"
o~ 4o
d2F/dt2
Bidir Force "
20
d2F/dt2
o~ o,
:
0
B
--
"
1O0
200
300
400
1oo
8O
//J
6o
