Journal of

J Neurol (1989) 236:406-410

Neurology

© Springer-Verlag 1989

Force transition control within a m o v e m e n t sequence in Parkinson's disease* G. E. Stelmach, A. Garcia-Colera, and Z. E. Martin Motor Behavior Laboratory, University of Wisconsin-Madison, Madison, WI 53706, USA

Summary. An experiment was performed which examined movement planning and force transition control in six patients with Parkinson's disease (PD) during a sequence of five finger taps at either a fast (200 ms) or slow (600 ms) temporal speed. The patients acted as their own controls and performed finger taps under three task conditions: (1) where all taps had to be of the same force intensity: no stress; (2) where it was known that one of the taps had to be executed with an augmented force: stress simple reaction time (SRT); and (3) where it was not known prior to initiation which one of the taps was to be more forcefully produced: stress choice reaction time (CRT). Reaction time data revealed a between-condition effect where stress SRT was faster than stress CRT and no stress was faster than both. Under both speed conditions, the interval after the stressed tap was profoundly lengthened. It was found that the lengthening was due to increases in both lift and dwell times for the slow tapping rate. In contrast, at the fast tapping rate, proportionately more of the interval increase was due to the increase in lift time. These findings suggest that patients with Parkinsons disease, when performing under a motor program mode, have difficulty in initiating a sequence and making a transition to lighter force levels after a stressed tap. Key words: Parkinson's disease - Movement control - Movement planning - Force transition control

Introduction After reviewing the relevant literature, Marsden [6] concluded that simple motor programs are, for the most part, preserved in PD patients. Normal movement patterns are assumed to be composed of sequences of simple motor programs set to occur at designated time periods. Accordingly, planning a motor act consists of assembling the sequential and/or simultaneous motor programs necessary to execute an intended action. In Marsden's view, it is the plannning process that is mediated by the basal ganglia and that disintegrates with the onset of PD, leading to an inability to execute motor actions rapidly and accurately. * This research was supported by the National Institute of Neurological Diseases and Stroke, grant no. NS17421 Offprint requests to." G. E. Stelmach

While Marsden's view is clear, the data on whether patients with Parkinson's disease can adequately plan movement via advance information have been recently debated in the literature. [1-4, 10]. When patients with PD are tested on tracking tasks, the results appear to depend on the specific nature of the task used and whether the subjects are well practiced. Flowers [4] found a deficit in performance on a tracking task involving a stimulus of a predictable sinusoidal track that required reversals of movement direction and that included advance information about the movement to be performed. Bloxham et al. [1] expanded the findings of Flowers by demonstrating that PD patients can perform a continuous tracking task. However, they argued that the sinusoidal nature of the stimulus used by Flowers [4] involved direction changes that could be thought of as an initiation of a new movement with each change of direction. This hypothesis was supported by the use of a discrete finger-lifting task which showed that, with respect to choice reaction time, patients with PD and controls did not differ significantly without prior information, but that the reaction time (RT) of the controls dropped substantially when advance information was provided. Thus, Bloxham et al. [1] posited that Parkinsonians can use prior information to plan their movements, but have difficulty in using such information when the task is composed of discrete movements requiring movement reversals. Patients with PD are able to use advance information in predictable tracking task as shown by Day et al. [2] who found only a small decrement in performance as compared with controls. It was suggested that the decrement in performance of patients with PD was due to a lack of accuracy in response execution rather than a deficit in using advance information. According to Day et al. [2], PD patients can adopt predictive motor strategies; however, owing to a lack of precise control in fast movements, they are not able to achieve the same degree of accuracy. Although there are some results that indicate that the planning of a discrete movement sequence is impaired in PD, most studies on planning dysfunctions have focused on discrete responses. Stelmach et al. [10] concluded that Parkinsonians, when performing a discrete movement precuing task, can shorten latencies if given partial or complete advance information. The data were unequivocal in showing that advance information can be used by patients with PD. In a subsequent analysis and comparison [13], it was found that when simple

407 reactions and choice reactions are compared over a period of practice, simple reactions show a gradual and consistent decrease of lateney with practice. These data further establish that patients with PD can benefit from using precue information for planning purposes. The main purpose of the present work was to investigate the motor planning and execution of finger taps, using an experimental paradigm that allows patients with PD to act as their own controls to investigate three primary questions: Can patients with Parkinson's disease perform a serial tapping sequence involving an embedded force element? Can they use advance information in a movement involving both timing and force? Does response execution speed impair movement planning and force control characteristics? PD patients performed sequences of five taps in three different task conditions: (1) no stress (NST), in which all taps were to have the same intensity; (2) stress simple reaction time (stress SRT), in which the patients knew the location of the stressed tap; and (3) stress choice reaction time (stress CRT), in which the position of the stressed tap was unknown until the imperative signal was given. By using this paradigm, it was possible to further investigate whether Parkinsonians use advance information in preparing a movement sequence. In addition, this paradigm provides evidence on how they control slow (i.e. on-line control) and fast (i.e., programmed) sequences. Moreover, the execution of the tap sequence was inspected by studying the interaction between force production and the timing of successive movements.

Method

Subjects and apparatus. Six adult volunteers diagnosed as having PD (59-65 years of age) participated in the experiment. All subjects showed bilateral impairments. Subjects performed sequences of five taps on a contact key (6.0 x 6.5 cm) with the index finger of the right hand. The key was mounted on a wooden board that was fastened to a stationary table. Subjects sat facing the key with the forearm resting upon the board. They were instructed to hold their right index finger just above the key in a comfortable position. The tapping movement required involvement of the wrist, finger flexor, and extensor muscles. A tap was produced by hitting the key with a downward flexion movement and then releasing it. A photocell system attached to the key detected contact of the finger with the key surface by emitting a light beam across the key surface which was broken every time the finger contacted or was lifted from the key. A tap activated a strain gauge attached to the key that measured both mean voltage and peak voltage of the tap. Thus, voltage was a relative index of force production. Procedure. Each trial began with the presentation of a star on the terminal screen that had a duration of 500 ms along with a simultaneous auditory "beep" of 1 kHz for 20 ms. These signals served as a double warning signal (WS); the onset of the signals was synchronized. The WS was followed 2000 ms later by the presentation of the "temporal model" for the movement sequence delivered via a radio transmitter by auditory beeps which indicated the speed to be reproduced by the subject. The temporal model speed consisted of five auditory clicks of equal intensity and frequency (1 kHz) with a pulse

duration of 2 ms separated by equal time intervals. There were two click intervals, fast and slow; the former interval was 200 ms and the latter interval was 600 ms. The last click initiated a variable preparatory period (PP) of three equally likely periods (900, 1500, or 2100 ms) presented in pseudorandom order. At the end of the PP, the reaction signal (RS) was presented on the terminal screen for 500 ms. Either an equal sign (=) or a digit ranging from 1-5 was used as the RS. The former signal was used when no voluntary accentuation was required, whereas the latter type of signal was used when one of the taps was to be of a stronger force than the other four taps. The digit indicated the serial position of the stressed tap. Subjects were instructed that there were three components to the task and that all were equally important. The first was to initiate the sequence as fast as possible after the RS presentation. The second was to reproduce the temporal model as accurately as possible by tapping on the response key. The final component was to stress the appropriate finger tap at the designated location such that a larger force was produced relative to the other taps. The following measurements were recorded for each response sequence: delay between RS onset and onset of the first tap, duration of each tap or period during which the finger was in contact with the key, the period between successive contacts of the finger with the response key, and the maximum force for each tap. Design. The experiment was completed over two days. Each test day was run at one of the two temporal model speeds, fast (200 ms) or slow (600 ms), which were balanced across days. Each experimental session consisted of no stress (NST), stress simple reaction time (stress SRT), and stress choice reaction time (stress CRT) trials. For the NST condition, a block consisted of 12 trials. For the stress SRT block, 12 successive trials were given at each of the five stress locations. Prior to beginning the trial series, each subject was explicitly instructed where the stress tap should occur in the sequence. The stress CRT block was similar to the stress SRT condition except that the serial position of the stress was presented in random order, making the subject unaware of the response required. Therefore, each subject received a block of each condition (NST, stress SRT, or stress CRT) at one temporal speed (fast or slow) in the initial session. The second was identical to the first except that the speed of the temporal model was switched. The presentation of conditions was balanced over subjects. The use of withinsubject comparisons provides the opportunity to investigate movement planning and execution characteristics in patients with PD. Reaction time data excluded 2 patients because of spurious data recordings. The 2 subjects experienced excessive tremor on the day of testing, preventing collection of accurate RT data. Therefore, reaction time data were analyzed using only 4 patients. In all other comparisons, the data from 6 patients were analyzed.

Results

Reaction time The mean reaction times of the patients in each conditions are presented in Fig. 1. Overall, the main effect was that NST,

408 600

\

680

560

k

6;0

,00e,

520 600

2

~/

7-,.

\/

&

\o

\ \/ / xo

_

~80

Stow

560 X

L+~+O

520

o

o& °~o~

°

400

\\°'~-o--o o. "'o--o

X "~'o--o

480 360

370

~)~~°~. o

E 4s+O

~

I

~e

280

cu _c &O0 Ox

260 600 Tapping rate (ms)

Fig.1. The mean reaction times for each condition (no. stress, @; stress SRT, i , stress CRT, I ) as a function of the tapping rate

\o._/

2

760

c

/

\ 320 Fast

280 o

stress SRT, and stress C R T differed significantly from one another, F(2,6) = 28.59, P < 0.001. This main effect shows that subjects increased in response latency overall between task conditions irrespective of temporal model speed. Posthoc analysis shows that in the NST condition a mean RT of 319 ms was significantly lower than stress SRT, 378 ms, in which case the subjects still had complete advance information but a stressed tap was embedded in the sequence, ( P < 0.01). Further, response latencies increased significantly (to 473 ms) in the stress C R T condition to 473 ms where the serial location of the stress tap was unknown as compared with stress SRT, (P < 0.01). A significant increase (P < 0.01) in reaction time was also evident between NST and stress CRT. A significant main effect of temporal speed is clear from Fig. 1, F(1,3) = 23.04, P < 0.02). The execution of a fast tapping rate resulted in longer latencies, 438 ms, than the slow tapping speed, 342 ms. Although all three task conditions were significantly different overall, at the slow tapping rate one comparison was not significant between NST and stress SRT.

Force and error characteristics Because there were no differences in force profiles for the different taping rates, the results at both speeds were combined. A significant stress by tap interaction in the stress SRT and stress C R T conditions, F(16,80) = 59.81, P < 0.001, demonstrates that the patients were executing the force generation correctly. There was no effect of force as a function of stress location.

Tap intervals Inter-tap intervals were classified according to their relationship within the tapping sequence with regard to the stressed tap's serial position. An interval occurring prior to the position of the stressed tap was classified as a before stress (B) interval. The interval that occurred after the position of the stress was labeled an after stress (A) interval. The final classification is remote (R), which includes all the intervals that

240

\

o \\

°