Experimental BrainResearch

Exp Brain Res (1990) 79:530-538

9 Springer-Verlag1990

Trajectory control in targeted force impulses VII. Independent setting of amplitude and direction in response preparation M. Favilla*, J. Gordon, 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. We have previously shown that when aiming impulses of force to unpredictable flexion and extension targets, subjects prepare themselves to respond by preselecting a default amplitude and direction. In our preceding study, where flexion and extension target ranges were symmetrical, subjects prepared a single default amplitude near the center of each of the two ranges and selected a default direction arbitrarily (Favilla, Hening et al. 1989). Following target presentation, amplitude and direction were specified gradually and in parallel. By using an experimental paradigm in which target amplitudes had unequal probabilities in only one of the two directions, we no sought to determine to what extent the default amplitude prepared by subjects is dependent on the default direction that is selected for a given trial. Five normal subjects produced isometric elbow flexion and extension force impulses to match four targets requiring responses of two amplitudes in each direction. Flexion and extension targets were of the same size. In all biased probability conditions, targets were presented in unpredictable order and, in one direction, the probability of occurrence of the targets was biased (80% versus 20%). In the other direction, the two targets were equiprobable. To maximize the effect of the default settings on response parameters, we required subjects to initiate their response in synchrony with a predictable tone occuring less than a reaction time subsequent to the presentation of the visual target. Such a short interval was chosen to disclose the default parameters selected by limiting the influence of the target just presented on the amplitude and direction of 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)

the subjects' responses. If the default amplitude were prepared independently of the response direction selected, subjects should show a bias in the default amplitude in both directions. In contrast, if default amplitudes were associated with direction, the bias would occur only in the direction whose target amplitudes had unequal probability. For controls, we presented the same four targets both in predictable series (simple condition) and in unpredictable series without any bias (choice condition). When presented with targets whose amplitudes were of unequal probability in only a single direction, all subjects showed a strong bias in the amplitudes of their responses in both directions. This result was true of all subjects and occurred independently of the response direction (flexion or extension) or whether the response was in the same direction as the current target or the opposite direction (i.e. wrong direction). This bias was strongest when responses were made in the same direction as the targets with unequal probability. We conclude that subjects prepare default amplitudes that are, in large measure, independent of the direction of responses made to unpredictable targets. The present results expand our demonstration that subjects may perform different component processes of sensorimotor processing, such as specification of response amplitude and direction, by parallel processing channels.

Key words: Sensorimotor processing - Response preparation - Isometric - Trajectory specification - Reaction time - Human subjects

Introduction In previous reports in this series, we examined the processes by which human subjects extract information from visual targets in order to accurately

531

specify the peak forces of isometric elbow impulses (Hening, Favilla, and Ghez 1988; Favilla etal. 1989). Our studies utilized the Timed Response paradigm, in which an auditory cue that controls the timing of response onset is presented independently from a visual target shift that indicates the desired response amplitude. By varying the interval between the target shift and the auditory cue determining response onset, we were able to ascertain the time course over which information concerning the amplitude and direction of an unpredictable step change in a target level was incorporated into subjects' responses. When only the amplitudes of target shifts were unpredictable, subjects required an interval of about 300 ms between target shift and response initiation (stimulus-response or S-R interval) to accurately specify response amplitude. At S-R intervals of less than 100 ms, response amplitudes were distributed around the middle of the range of targets previously presented (Hening, Favilla, and Ghez 1988). Over an interval of about 200 ms, this default amplitude was gradually converted into one which appropriately matched the target. When both the direction (i.e. flexion vs extension) and the amplitude of the target step were unpredictable, the directions of responses initiated at short latencies ( < 100 ms) were only randomly related to those of the targets (Favilla et al. 1989). The amplitudes of these responses were, however, distributed around the center of the two ranges of targets. Specification of amplitude was again gradual, but was delayed in comparison to the unidirectional case. Moreover, amplitude and direction were apparently specified concurrently, as shown by the improving specification of the amplitude of responses made in the wrong direction at increasing S-R intervals (Favilla et al. 1989). In contrast, when the direction and the amplitude of the target were predictable in advance, subjects produced accurate responses at all S-R intervals. These findings led us to conclude that, in matching response features to those of a target, subjects initially prepare a set of parameters reflecting their expectation. If the target is predictable, these parameters reflect the actual target. For unpredictable targets, the default amplitude is set close to the center of the expected range of responses while direction, a discrete variable, is assigned arbitrarily. Then, as additional information from the target becomes available, different response features, such as direction and amplitude, are gradually specified by processes operating in parallel. Because these results indicated that subjects can initiate responses even if their amplitudes and directions are not yet accurately specified in

relation to the current target, we concluded that specification and initiation can be controlled by parallel processes. The parallel specification of two response features following target presentation raises the question of whether the preparation of the default amplitude and direction, prior to target presentation, also occurs in parallel. Two possibilities can be envisioned. In one, the preparation of a response in a given direction would include the selection of a particular default amplitude associated with that direction. In the other, amplitude and direction would be prepared independently and a single default amplitude would be associated with both directions. In our preceding bidirectional study (Favilla et al. 1989), flexion and extension targets had the same amplitudes. Therefore, the subjects' default amplitudes would be equal for responses in the two directions both if subjects prepared a single default amplitude independent of response direction and also if they prepared distinct default amplitudes for flexion and extension responses. The present experiments were undertaken in order to determine whether amplitude is prepared independently of direction. In order to address this question we presented subjects with a symmetrical range of targets whose probability of occurrence was made unequal in only one direction (flexion or extension). Using the Timed Response paradigm, subjects were induced to initiate responses at short latencies (less than 150 ms) in order to allow us to identify the default parameters of the response before they were obscured by the specification to the current target. Based on a prior study using targets of unequal probability in a reaction time task (Hening, Vicario, and Ghez 1988), we expected that the default amplitude would be biased toward the most probable target. We reasoned that, if a separate default amplitude were prepared for each direction, this biasing would influence only responses produced in the direction of the bias. On the other hand, if the amplitude of the default were prepared independent of its direction, response amplitude would be biased in both directions. Some of the results have been previously summarized in an abstract (Favilla et al. 1986) and in a review (Ghez et al. 1990). Methods The present set of experiments was conducted using 5 normal adult subjects (3 men and 2 women, ages 30 to 42) who had participated in previous studies of this series (Hening, Favilla, and Ghez 1988; Favilla et al. 1989). The subjects, therefore, all had prior experience with the task of producing force ira-

532 pulses at the elbow and with the Timed Response Paradigm, in which they were required to synchronize their responses with a predictable auditory cue. Since the apparatus we utilized and the general task of producing force impulses, as well as most of our analytic procedures, have been previously described in detail (Ghez and Gordon 1987; Hening, Favilla, and Ghez 1988; Favilla et al. 1989), we now describe these aspects of our methods only briefly. Subjects sat with their right arm 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 flexor and extensor muscles. 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. A storage oscilloscope, placed below the monitor, allowed subjects to check the timing and the trajectories of their responses at the end of each trial.

reaction time when making elbow flexions in a comparable reaction time task (Favilla et al. 1987). This interval (approximately one standard deviation less than the meala reaction time) was long enough to allow some degree of specification. This preserved the subjects' motivation to attend to and make use of target information. The interval between the step change in the target level and the actual time of onset of the subject's response is termed the Stimulus-Response (S-R) interval. The S-R interval differs from the conventional reaction time in that the time of response onset does not depend upon a decision by the subject of when to initiate responses. The mean S-R interval for all subjects was 112 ms, an average of 20 ms less than the interval between the target shift and the synchronizing tone (see Table 1). Within each block of 60 trials, targets were presented in one of three types of block: simple, choice, and biased. In simple blocks, the targets were presented in a predictable cyclical sequence (1, 2, 3, 4, 1, 2 . . . . ), while in choice and biased blocks, the targets were presented in an unpredictable order. In choice blocks, all four targets had equal probability of occurring, whereas in biased blocks, target probabilities were unequal in one direction and equal in the other (Fig. 1). In the direction with unequal probabilities, the more probable target occurred four times as often as the less probable target. There were four bias conditions: Large Flexion (LF), Small Flexion (SF), Large Extension (LE), Small Extension (SE). For example, in the LF bias condition, a block consisted of 24 large flexion targets, 6 small flexion targets, 15 large extension targets, and 15 small extension targets. Thus, in any given bias condition, one target amplitude (either small or large) occurred in 65% of the trials, while the other target amplitude occurred in 35% of the trials. However, target directions (flexion or extension) were equiprobable. Typically, sessions consisted of one block of practice, one simple block, one choice block, and four biased blocks. Biased blocks were presented later in the session in order to minimize performance deficits due to lack of intra-session practice. Subjects were given two or more blocks with the same bias condition in sequence in order to maximize bias effects. Two different bias conditions were presented in each of two sessions. To minimize use of conscious strategies, subjects were not told of the presence of bias. W h e n qnestioned after the final experiment, some of the subjects reported that they had not noticed the bias. Informed that a bias had been present, some then incorrectly identified the bias in a particular block.

Experimental procedure: the Timed Response Paradigm We used an auditory cue to control the time of response initiation (Fig. 1). Subjects were trained to initiate force impulses in synchrony with the last in a series of four tones of increasing pitch, each lasting 20 ms and presented at intervals of 500 ms. The required response amplitudes were signalled by upward (flexion) and downward (extension) shifts in the target line on the oscilloscope screen. On each trial, subjects were to produce a single impulse of force whose peak was to match the new level of the target. They were then to relax, allowing the force to return passively to baseline. Subjects were instructed and trained not to correct their responses after initiation if they noted an error in amplitude or direction. Four target amplitudes were presented, 2 in flexion and 2 in extension. The amplitudes in Newtons were the same for both flexion and extension; the ratio of the small to the large target amplitude was 1:5 in each direction. The largest force required was chosen at a level which did not produce significant fatigue, approximately 40% of the subjects' maximal isometric force. In the course of a single trial, subjects would first align their force with the target line at zero force for 4 s. Then, as shown schematically in Fig. 1, the sequence of ascending tones began. Between the third and fourth tones, the target was stepped to a new level. Subjects were required to initiate the force impulse at the time of the fourth tone. In order to ensure that subjects' responses reflected primarily the default amplitude, we limited the time available to process information from the target preceding the response. This was done by setting the interval between the target step and the fourth tone (S-T interval) on all trials at 30 ms less than the subject's simple

.... Tone # 1

Y ....

T,,,

2

3

,

i i

4

Large Small Small Large

..,~%',--"

0.5 S

Target Level

Flexion

Flex

Flex

Ext

Ext

l

40

10

25

25

10

40

25

25

25

25

40

10

25

25

J0

40

!" 7"'"

J

Force

i

Experiments were controlled and data were acquired and analyzed using a 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

Bias Condition (% Trials in block)

B

A

Data acquisition and analysis

. . . . . .

...... o

......

S-R Interval

Extension

Fig. 1 1 , B. Timed response paradigm with biased probabilities. The subject hears a series of four tones (arrows above) and at a fixed interval before the last tone (S-T interval, typically 120 ms), the target line (broad line) is displaced to one of four target levels. Subject initiates force pulse (fine line) in synchrony with fourth tone. S-R interval is the time from visual stimulus to response onset. Table at right (B) indicates the percentage of each of four targets in four different bias blocks

533 and triceps electromyographic (EMG) activity recorded with surface electrodes (Boston Elbow Myoelectrodes, Liberty Mutual). Automatic computer programs, whose results were reviewed by inspecting individual trials, marked the onset and peaks of the force trajectories and reduced the trial data to arrays of variables which described trajectory parameters (response direction, amplitude and timing of the peaks of force and of its first two time derivatives, timing error, and S-R interval). 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. Such rejected responses made up fewer than 5% of all trials in each subject. The arrays of trajectory variables were then subjected to statistical analysis. Statistical analysis was carried out only on responses made in biased blocks. Data from the simple and choice blocks are presented for purposes of comparison. Responses in the biased blocks were categorized by bias condition (LF, SF, LE, SE), target direction (FLEXION, EXTENSION), target size (SMALL, LARGE), and response direction (FLEXION, EXTENSION). These categories were then used to classify trials according to the following conditions: bias direction (SAME as response direction, D I F F E R E N T from response direction), bias size (LARGE target more probable, SMALL target more probable), and response type (CORRECT-same direction as target direction, WRONG-opposite direction from target direction). In all analyses, the dependent measure was response amplitude (absolute value of peak force in Newtons; flexion force was otherwise considered to be positive, extension force negative). Statistical analyses were performed separately on the data from each subject and then across all 5 subjects. Because the number of trials performed in each condition varied within a subject's data, the within-subject analyses were carried out using multiple regression techniques to test whether selected conditions significantly influenced response amplitude. The across-subjects analysis was computed using means from the 5 subjects for a given condition. Therefore, the cell frequencies were balanced, and a standard repeated measures A N O V A was performed. The details of the statistical analyses are presented in the Results.

and choice blocks were not significantly different from these in any of the subjects. All but one of the subjects initiated their responses on average before the fourth tone, and all the mean S-R intervals are considerably shorter than a typical choice reaction time (Hening, Vicario, and Ghez 1988). Since the mean S-R intervals range from 73 to 132 ms, intervals in which subjects show little or no specification of response amplitude relative to the target (Favilla et al. 1989), response parameters must be based on the previous experience and expectations of the subject. Response amplitude would be little changed from a "default" value prepared prior to target presentation (Hening, Favilla, and Ghez 1988). The effect of different bias conditions on this default amplitude will first be described in detail for a single subject ($13), and then the results for all subjects will be reported. Table 1 also shows the percentage of wrong direction responses by each subject in the biased blocks. All but one subject responded incorrectly significantly less than 50% of the time. In simple blocks, the subjects did not make wrong direction responses; in choice blocks, the incidence of wrong direction responses was similar to that in biased blocks. A

Simple Equiprobable

125 N

..... ~

F,ex,on

!

C

Choice Large Flexion biased

.....

40% : : : : : : : : : : : : : = = : : : : : : :

\

25 N ... l.a'.~..."~...2~ .....

1

"

""~" 25N . . . .""~' . a~,,.~.~."~..

25% ~ . . . . ... . . ... . . . ~ "I~"'""'~ --

k f

Extension

Results

Table 1 shows the mean S-R interval for each subject in the biased blocks. S-R intervals in simple

'125 N . B

25% . . . . . . . . . . . . . . . . . . . . . . Choice

D

Choice

Equiprobable

Small Flexion biased

25% . . . . . . . . . . . . . . . . . . . . . .

10% . . . . . . . . . . . . . . . . . . . . . .

2

4

Table 1. Mean Stimulus-Response intervals (ms) and incidence of wrong direction responses for each subject in biased blocks SBJ

N of trials

$2 583 $5 530 S12 466 $13 533 S14 527 Mean across subjects S.D.

S-T interval a

S-R interval Mean

S.D.

Wrong direction (Per cent of all biased trials

120 150 120 150 120 132

73 123 103 132 127 112

30 23 33 37 23 29

52 43 40 40 28 41

16

24

6

9

" S-T Interval refers to stimulus-tone interval which, for a given subject, was constant across trials

5

~

25~/. . . . "~ . . . . . . . =,,,,,,r

25%

25% . . . . . . . . . . . . . . . . . . . . . .

25%

0

~

I

I

100 ms .

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

Fig. 2A-D. Average trajectories in Simple, Choice, and Flexion Bias Conditions. In each part of the figure there are four traces for averages of correct direction responses to large (solid line) and small (hatched line) targets. Target amplitudes are indicated in A and probability of each target, in choice and bias conditions (B-D), to left of target lines. All averages synchronized on force onset. N's for averages: simple, 1 ~ 1 4 ; unbiased choice, 17-40; large flexion bias, 8 32, small flexion bias, 7-41

534

A

Flexion Targets Biased

Extension Targets Biased

B

~::~:~:~:~::::~:~::~::~::~:~:~:::~:~:~:~:~:~:~:i~:~:~:~:~:~:~:~:~:~i:~i:~:i:~:~:~:i:!~:i:i:i~!~!:!:!:!:!:!:!:!:!:!:!:!:!:!:!~!:!|

li i i i

A

i

25 _ ~:~i~;i~~ .......................................................................................... .

14.

IJ

:::::::::::::: ~i~i:i~i~i!i!~ii!iii!i~i!i!i!i!i!i~i!i!i?i!i!i!i!i!i!~!i!i!:.ii!i!!i!!!~!~!:.!i!~!~!~!:.!:.!~!~!::?:.!~!~i~!~i~::i::i~{i~i{~i~i~[~i~i{~ ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: ........

~:~:::::~:~;::~:::::~:~:::~:~:::::::~:~::::~:~:~:~:::~:::::~:~:::::~:~::~::~:~:::::~::::~:~:::::~:~:~:::::~::::~:::::::::::~ ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: ::::::::::::: ::::::::::.: :::: :_:

o 0

!'iiii~iiiii~:~iiiiii~iiiiii~i~iii~ii~i~i~i~i~i~:~=~i~i~i~i~i~i~i~ii~iiiiii'

FIx

|:~::~:~:~:~:~:~::~:~:~:~:~:~:::~:~:~:~:~:~:::~:~:~:~:~::~:~:~:~:~:~:~:L:~:~:~:~:~:~:~:L:~:~:~:~:~:~:~:~:~:~:~:~:~:~::::::::::::::::::::::

fl" e-

~u

~i~iii~i~i~i~i~i~iii~iii~i~iiiiiiiii~iii~iiiiiiiiiiiii~iii~iiiiiiiiiii~iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii}iiiiiiiiiiiiiiiiiii!i!iiiiiiiiiii!i!i!]

r::~::~::~:::~:::::::::::~:~::::~:::~:::~:::::~:::~:::~:::::~:~:~:::~::~:::~:~:::::::~:~::~:~:~:~::::::~:~:~:~:~:~:::~:~:::::~:~

-25-

iiiiiiiiiiiiii:= Iiiii

t~

1

~iiiiiiiiii

iiiiiiii~t

}, ., ,. ". ,.. '. , .- } -"7Z'--'~4 [ii:ililiiiiiiii:i~i~i~iiiii!iiii'!iiiiiiiiiii}i{ii::iliiii!ii!iiiiii!:iii{i iiiiiiig:iil /,"

...... ....,........,.................. ,........ ':~==!i!ii!==!=/~!==!=:!==i!==i~!~!==?=r

//" ," EI~

//"

................... ::::::::::::::::::::::

!" Ext

~:~:?::::::::::::::::::::::::::::::::::::::...........

.................................... .~.=..=-".~:~...................:'::::::