INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A long-standing question in motor control is how motor sequences are built up from individual movement components (Lashley 1951; Rosenbaum 1991). New approaches to this problem now are being generated by virtue of evidence that regions of the frontal cortex may be specialized for coding motor sequences and that these cortical areas together with the basal ganglia may form part of the central control mechanism for motor sequencing (Brotchie et al. 1991; Cromwell and Berridge 1996; Graybiel 1995, 1998; Kermadi and Joseph 1995; Kimura et al. 1992; Marsden 1987; Miyachi et al. 1997; Mushiake and Strick 1996; Roland et al. 1980; Tanji and Shima 1994; Tanji et al. 1987). Neuronal activity in the primate premotor cortex (Mitz et al. 1991) and supplementary eye field (Chen and Wise 1995a,b, 1996) is modified during the acquisition of sensory-motor associations. Tanji and coworkers have shown that many neurons in the supplementary motor area (SMA) fire preferentially in relation to particular learned movement sequences. These authors have suggested further that neurons in the cortical area anteriorly adjoining the SMA (the pre-SMA) are differentially active when a motor sequence is to be switched (Matsuzaka and Tanji 1996; Shima et al. 1996). In accord with these findings, inactivation of the SMA and particularly of the pre-SMA in monkeys produces deficits in learning and performing sequential motor tasks (Miyashita et al. 1996).

For the basal ganglia, it has been reported that neurons in the striatum and globus pallidus of monkeys are activated specifically at the first movement of a motor sequence or at other specific phases of the sequence (Brotchie et al. 1991; Kermadi and Joseph 1995; Kimura 1990; Kimura et al. 1992; Mushiake and Strick 1996). Similar task-specific activation patterns have been reported in the rat (e.g., Aldridge and Berridge 1998; Gardiner and Kitai 1992; Jog et al. 1997; Kubota et al. 1998; West et al. 1990; Woodward et al. 1995). Lesions of the striatum in the rat have been shown to impair sequential grooming behavior (Cromwell and Berridge 1996), but with the exception of a recent report by Miyachi et al. (1997), little has been reported about the consequences of lesions of the primate striatum for the ability of monkeys to learn and to retrieve motor sequences.

In the study reported here, we used the strategy of inactivating the dopamine-containing innervation of the striatum to impair selectively striatal function in monkeys before or after training on sequential push-button tasks. In previous work, we have found that local inactivation of dopaminergic function in the striatum by infusion of the neurotoxin 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) greatly decreases conditioned responses of striatal neurons previously acquired in a sensory-motor learning task (Aosaki et al. 1994a). These observations suggest that the nigrostriatal dopamine system is necessary for retrieving acquired neuronal activity from long-term storage of activity patterns in the striatum (Aosaki et al. 1994a; Graybiel et al. 1994). We reasoned that if the nigrostriatal system were critical for learning movement sequences or for retrieving them, we should detect deficits in performance of sequential motor tasks in monkeys with intrastriatal MPTP-induced dopamine depletion. The experiments reported here were designed to test this notion.

Our goals were to assess abilities of unilaterally dopamine-depleted animals to acquire programs for new motor sequences and to retrieve or to relearn previously learned motor programs. Our findings, briefly reported elsewhere (Matsumoto et al. 1994), strongly implicate the striatum and its dopamine-containing innervation as being involved in the learning of sequential motor tasks and in the retrieval of previously learned motor sequences.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Behavioral paradigms

Three male monkeys (Macaca fuscata, 5.5, 7.0, and 7.2 kg) were studied. The acquisition and care of the animals were based on National Institutes of Health guidelines (publication No. 80-23). The monkeys were trained to sit in a primate chair facing a small panel placed 50 cm in front of the animal. For experiments on monkeys KE and GN, three push buttons were situated on the panel so as to form the apices of a triangle ca. 9° of visual angle (10 cm) apart (Fig. 1). These buttons could be illuminated under computer control. The monkeys learned to push the three buttons sequentially in an instructed order for a total of three pushes (monkey KE) or four pushes (monkey GN). For monkey TZ, a touch panel (NEC 9873L) was attached to the surface of the computer display. A 1.6 × 1.6 cm blue rectangular light could be illuminated at any one of nine possible locations on the display (see Fig. 5). The monkey was trained to touch three sequentially illuminated rectangles on the touch panel with his index and middle fingers, in an instructed order, for a total of three pushes. All three monkeys were trained to make the button pushes with each arm. In a given trial, the unused arm was loosely restrained by a yoke.

View larger version (21K): [in this window] [in a new window]   Fig. 1. Sequential button-push task training protocol. Monkeys were trained to push buttons in response to illumination of the buttons in a clockwise or counterclockwise sequence. Left: diagrams illustrating the timing of the illumination of buttons, button pushes, and reward delivery during the task. Right: diagrams showing the sequential changes of button illumination and corresponding correct sequential button pushes. Only a clockwise sequence is shown; clockwise and counterclockwise sequences were presented in random order. Monkey KE learned the 3-push tasks, monkey GN learned the 4-push task. In this and succeeding figures: , button at which monkeys received a drop of water reward

The first step of training was the one-push task. As shown for the button task in Fig. 1, button 1 was illuminated, and the monkey received a drop of water as reward when he pushed the lighted button. During this phase of training, button 1 was illuminated at variable intertrial intervals of 5-7 s. In this and all subsequent behavioral sessions, an audible click was given 280 ms before the water reward. The second step of training was the two-push task (Fig. 1). After the monkey depressed button 1 for 1 s, the button 1 light was turned off, and simultaneously the second button was lighted. This served as a GO signal for the monkey to release button 1 and push button 2 to obtain a reward. The third training step was the three-push task, in which the monkey pushed the three buttons in either clockwise or counterclockwise sequences (Fig. 1). Button 2 was lighted after the monkey pushed the illuminated hold button (button 1) for 1 s; the monkey then had to release the hold button and push the illuminated second button, which in turn illuminated the third button, which the monkey had to push to receive reward. Monkey TZ was similarly trained on the touch panel to perform two blocks of trials. The first block was 30 trials of a fixed, cued sequence. The second block consisted of a total of 60 trials in which three different cued sequences of screen touches were triggered in random order but with equal probabilities (Fig. 5). Monkey GN, but not the other monkeys, was trained on a four-push task after he had learned the three-push task. The fourth button was illuminated after the third push, and then the monkey pushed the fourth button to receive reward (Fig. 1).

The monkeys performed a total of 800-1,500 correct trials in a day. In an effort to have identical training conditions for right and left hands, we adjusted the number of trials so that the total trials performed with the right and left arms did not differ by >25-100 trials in a day. Thus when the monkeys moved from one step to the next step of learning with the right hand, they also moved to the same next task with the left hand on the same day. Late in the experiment, task switch trials were introduced for monkey KE at irregular 4- to 7-day intervals. In these, the monkey unexpectedly had to perform a previously performed push task, rather than the current push task (for example, a 2-push instead of a 3-push task). This meant that reward and the associated click were delivered unexpectedly, one push early. When the task was switched, the monkey performed a total of 80-200 switch trials and 700-1,300 regular trials in a day. We carried out six switch experiments.

Recordings

Before behavioral training, each monkey underwent surgery for implantation of a head restraint bolt and a recording chamber to permit recording of single unit activity in the caudate nucleus and putamen. The activity of neurons in the striatum of these animals will be described in a separate report. Surgery was performed under anesthesia induced with ketamine hydrochloride (6-12 mg/kg im) and maintained with pentobarbital sodium (Nembutal, 28 mg/kg ip). Supplemental Nembutal (10 mg · kg¹ · h¹ im) was given as needed. Neuronal activity was recorded extracellularly with glass-insulated elgiloy microelectrodes inserted through the recording chamber at the start of each recording session.

Electromyographic (EMG) activity was recorded from the triceps and biceps brachii muscles (prime mover muscles of the arm) as well as from other supporting muscles and from the digastric muscle of the tongue (activated for licking the reward water) through chronically implanted, multistranded teflon-coated stainless steel wire electrodes (Cooner Wire AS631) with leads that led subcutaneously to the head implant. The EMG signals were amplified, rectified, integrated, and monitored on-line on a computer display simultaneously with the recorded neuronal activity. Licking movements also were recorded by means of a strain gauge fixed to the spoon in front of the animal's mouth.

The behavior of the animals was monitored routinely by a video camera. Eye movements were monitored by measuring the corneal reflection of an infrared light beam through the video camera at a sampling rate of 250 Hz. The computer system (RMS R-21C-A) determined the two-dimensional (x and y) signal of the center of gravity of the reflected infrared light beam. The spatial resolution of this system was ca. ±0.15°. The muscle activity and eye position signal from the video system were recorded at a sampling rate of 100 Hz by a laboratory computer (NEC 9801RA). The behavioral tasks also were controlled by the computer system. For the button tasks, the times of onset and offset of the button light stimuli, the time of depression and release of the push buttons and the onset of reward delivery and associated click stimuli were recorded during task performance. For the touch screen, the horizontal and vertical locations at which the monkey touched and released were detected as analogue data by means of a hardware device at a sampling rate of 500/s, and the computer system recorded the data at a rate of 100 Hz.

Unilateral dopamine depletion

To destroy the nigrostriatal dopamine system within the striatum, we locally infused MPTP (Sigma), a dopaminergic neurotoxin, into the caudate-putamen complex of one hemisphere in monkey KE and GN (Aosaki et al. 1994a; Imai et al. 1988). MPTP (4 mg) was dissolved in 200 µl of 0.15 M NaCl and was infused over a 14-day period into the right (monkey KE) or the left (monkey GN) hemisphere by means of an Alzet osmotic minipump (average rate, 0.5 µl/h). The infusion needle (0.6 mm OD, 0.3 mm ID) was connected to a minipump by a polyethylene tube, and it was implanted through the recording chamber with the monkey under Nembutal anesthesia (28 mg/kg im) at a depth of 11-12 mm and atlas coordinates A15.5-A16, a site between the putamen and the caudate nucleus. The infusion needle was fixed to the inner wall of the recording chamber by means of dental cement, and the minipump was implanted subcutaneously at the nape of the neck (Aosaki et al. 1994a; Imai et al. 1988). The infusion needle and minipump were implanted for 3 wk and then were removed. The monkeys were observed carefully throughout this period and were given injections of apomorphine (0.1 mg/kg im) to test for contralateral turning indicative of unilateral dopamine depletion (Imai et al. 1988). The apomorphine test was done once in monkey KE and three times in monkey GN.

We used two different protocols for the unilateral nigrostriatal dopamine depletion. In monkey KE, MPTP was given before behavioral training. After MPTP infusion and a 2-wk recovery period, training began and the monkey was required to use both the arm ipsilateral to the side of dopamine depletion (as the control) and the arm contralateral to the side of dopamine depletion. In monkey GN, MPTP was infused after behavioral training on the one- through four-button-push tasks. After dopamine depletion and the 2-wk recovery period, the monkey was required to perform both the three- and four-push tasks.

At the end of the experiments, the MPTP-treated monkeys were deeply anesthetized with Nembutal (70-90 mg/kg im) and were perfused through the left ventricle with 4% paraformaldehyde in 0.1 M phosphate buffer. Coronal 50-µm-thick sections through the right and left striatum were stained with cresyl violet. Sections spaced at 600-µm intervals through the striatum and substantia nigra were stained for tyrosine hydroxylase (Chemicon anti-TH, 1:2000). For monkey KE, the right and left hemispheres were cut separately, but sections from each were taken at approximately the same rostrocaudal levels and were stained in the same wells to ensure similar staining conditions for the MPTP-treated and untreated hemispheres. For monkey GN, the hemispheres were not separated.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Assessment of MPTP-induced dopamine depletion

Each of the MPTP-treated monkeys began to show neurological signs of dopamine depletion ~1 wk after implantation of the MPTP minipump, when about half of the total volume of MPTP had been infused. They exhibited a reluctance to use the arm and hand contralateral to the side of MPTP infusion when reaching for food, and they appeared to pay less attention to objects in the contralateral hemifield than to those in the ipsilateral hemifield. Their eye movements tended to be confined to the visual field ipsilateral to the dopamine depletion (Miyashita et al. 1995). They also exhibited some spontaneous turning toward the side of MPTP infusion, at a rate of a few turns/min in their home cages. These signs gradually became evident during the 2-3 wk after the implantation of the MPTP infusion pump and then stabilized and remained throughout the experimental period (6 mo for monkey KE and 10 mo for monkey GN). The monkeys consumed similar amounts of food and water before and after application of MPTP and gained weight normally (~500-700 g body wt in the 6 mo after MPTP infusion).

In each of the monkeys, we tested for the effects of the MPTP infusion by administering apomorphine (0.1 mg/kg im), a nonselective dopamine receptor agonist, within 16 days to 12 wk after MPTP infusion. Figure 2 shows the results of an apomorphine test given to monkey KE 12 wk after removal of the MPTP infusion needle. Before the application of apomorphine, monkey KE made spontaneous turning movements toward the side ipsilateral to the MPTP infusion at a rate of 1-4 turns/min. After the apomorphine was injected, the ipsilateral turning was abolished, and turning movements to the contralateral side began after 8 min. The rate of turning rapidly increased up to 22 turns/min, and then gradually declined. It took 60-70 min after the application of apomorphine for the contraversive turning to cease. The observation of apomorphine-induced turning in each monkey was used as behavioral confirmation that the unilateral infusion of MPTP via the implanted minipump had successfully produced unilateral dopamine depletion in the striatum and subsequent dopamine receptor sensitization.

View larger version (49K): [in this window] [in a new window]   Fig. 2. Apomorphine-induced turning in monkey KE. Apomorphine (0.1 mg/kg, i.m.) injected 12 wk after the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) infusion reversed the low level of spontaneous ipsiversive turning (negative values) exhibited by monkey KE and induced pronounced contraversive turning (positive values) that persisted for ~1 h.

The depletion of dopamine in the striatum and in the substantia nigra pars compacta, which gives rise to the nigrostriatal dopamine innervation, was confirmed histologically by staining for tyrosine hydroxylase (TH), the synthetic enzyme expressed in the cell bodies and axons of the nigrostriatal tract. The extent of dopamine depletion in the striatum, as judged by TH immunoreactivity, was different in the two animals (Fig. 3). In monkey KE, in which MPTP was infused into the right hemisphere before training (Fig. 3A, ), depletion of TH immunoreactivity occurred over ~7 mm in the caudal and midanteroposterior levels of the striatum, and the rostral striatum was left intact. The putamen was affected more severely than the caudate nucleus (Fig. 3A). In monkey GN, in which MPTP was infused into the left hemisphere after training (Fig. 3B, ), the depletion of TH immunoreactivity was much more extensive. It covered a large part of both the caudate nucleus and the putamen. Only the most ventral part of the rostral striatum and the rostrodorsal part of the caudate nucleus were left with dense TH immunoreactivity (Fig. 3B). The loss of TH immunoreactivity in the midbrain of the two monkeys differed in parallel to the levels of striatal depletion. There was no obvious loss of TH staining in the ventral tegmental area. Photographs of histological sections through the striatum and substantia nigra are shown for monkey KE in Fig. 4.

View larger version (40K): [in this window] [in a new window]   Fig. 3. Schematic illustrations of the extent of depletion of tyrosine hydroxylase (TH) immunostaining in the striatum following unilateral MPTP infusion in monkey KE (A) and monkey GN (B). Intensity of TH immunostaining was measured by densitometry and is shown in 6 grades from apparently normal (black) to maximum depletion (lightest shade). ND, no loss detectable. Numbers to left of each section indicate approximate Horsley Clarke coordinates.  and , level at which the MPTP infusion needle of MPTP was implanted. Photographs in Fig. 4 illustrate striatal level A 15.8 (Fig. 4, A and B) and nigral level A 12.4 for monkey KE (Fig. 4C).

View larger version (129K): [in this window] [in a new window]   Fig. 4. Loss of TH immunoreactivity in monkey KE. Photographs illustrate coronal sections stained for TH immunoreactivity. A: striatum on intact side. B: striatum on side of MPTP infusion. C: TH staining at the level of substantia nigra. Note almost complete loss of TH-like immunoreactivity on the side of MPTP infusion (right). PUT, putamen; Cd.N, caudate nucleus; GPe, GPi, external and internal segments of globus pallidus; TH, thalamus; SNpc, substantia nigra pars compacta; IC, internal capsule.

Sequential push movements are preprogrammed after learning

The push movements of the monkeys became faster as they were trained to perform particular fixed sequences of pushes over the course of several weeks. To test for the effects of predictability of the triggers for the movements, monkey TZ was trained first to perform >300 trials of three fixed sequences of pushes per day for 3 mo (Fig. 5A, left). The monkey then was asked to perform three new sequences of screen touches (Fig. 5B, left), in which the first and second targets were the same as that in the original sequence, but in which the third target was different. These three sequences occurred at random but with equal probabilities. The screen touches in the fixed and in the randomized sequences were performed in different 30 trial blocks. In a given day, we usually asked the monkey to perform 10-15 blocks of fixed sequences and 2-5 blocks of randomized sequences.

View larger version (44K): [in this window] [in a new window]   Fig. 5. Fixed sequence and randomized sequence tasks in monkey TZ. Left: locations of 1st, 2nd, and 3rd targets are illustrated. Right: histograms of serial reaction time from 2nd to the 3rd targets.

Figure 5A, right, shows the serial reaction times for the pushes performed after 3 mo of training. These were defined as the time between the onset of the target 3 light and the release of the target 2 light. The average serial reaction time was 81.7 ± 52.0 (SD) ms. The serial reaction times for the third push in the randomized sequences (time between onset of the 3rd target and release of the 2nd) are shown in Fig. 5B, right. They were significantly longer than those in the fixed sequence (131.5 ± 51.1 ms; P < 0.005, Wilcoxon single rank test). During the first and sometimes the second trial after the task was switched from the fixed sequence to the random sequence, the monkey tried to touch the target at the top right corner, which had been the third button of the fixed sequence, even though one of the other targets was illuminated. Only after that did he then press the illuminated button.

These observations suggest that the push movements from the second to the third target in the well-learned sequences no longer actually were guided by the visual stimuli, but were performed in a predictive, preprogrammed manner. By contrast, the longer serial reaction times for the movements in the randomized sequences suggest that these movements were guided by the visual cue for the third push, which was not predictable. In the random sequence task, the serial reaction times for movements from the second to the third button at the top right corner (124.8 ± 50.9 ms) were not significantly different from those to the third button at the top left (118.8 ± 50.8 ms) and bottom left (137.5 ± 35.5 ms), but they were significantly longer than those in the fixed sequence task (81.7 ± 52.0 ms, P < 0.005, Wilcoxon single rank test). These results indicate that the longer serial reaction times in the random sequence task did not occur because the random sequence task was less well learned than the fixed sequence task, but because the position of the third button was not predictable in the randomized sequence task. In the experiments with monkeys KE and GN, we tested the effects of dopamine depletion on the acquisition and performance of fixed sequences of push movements.

Unilateral dopamine depletion and movement times during sequence learning

In monkey KE, training on the sequential push button task started after dopamine depletion. The MPTP infusion needle and minipump were removed 3 wk before training onset. After 5 wk of training, the monkey had mastered both the one- and two-push tasks with his right and left hands. He then learned the three-push task. Figures 6 and 7 illustrate the activity of prime mover muscles (triceps brachii) for the arms ipsilateral and contralateral to the side of MPTP infusion as the monkey learned this new three-push task. The EMG traces illustrated were selected as representative of the activities occurring during the first 10 days of learning. The traces shown are aligned to the third button push, which elicited reward. Figure 6 illustrates performance in the clockwise sequence, and Fig. 7 illustrates performance in the counterclockwise sequence.

View larger version (28K): [in this window] [in a new window]   Fig. 6. Sequence learning after unilateral dopamine depletion in monkey KE. Electromyographic (EMG) activity of a prime mover muscle for the button-push task (triceps brachii) is shown for the 1st 10 days of training on the 3-push task in a clockwise sequence begun after the monkey had learned the 2-push task. Representative EMG traces are shown for the arms ipsilateral (right) and contralateral (left) to the side of the MPTP-induced dopamine depletion (right hemisphere, indicated by hatching in cartoon of monkey). EMG traces are centered on the time when the monkey pushed the 3rd button, after which he received a drop of reward water. Vertical tick marks above the EMG traces indicate the time when the monkey pushed the second button with the contralateral (L2) or ipsilateral (R2) hand.

View larger version (28K): [in this window] [in a new window]   Fig. 7. Sequence learning after unilateral dopamine depletion in monkey KE. Same as Fig. 6, but for the counterclockwise sequence.

Over the course of the 10 days of training, there were marked changes in the kinetics of the button presses when the monkey used the arm ipsilateral to the side of the MPTP lesion (Figs. 6 and 7). On the first day of training, when the monkey pushed the second button, which previously had elicited reward but now did not, he depressed the second button for long times, suggesting that he detected that something was wrong. The second button light was turned off, and the third button was illuminated simultaneously when the monkey pushed the second button. The monkey soon learned to release the second button and to push the illuminated third button to receive reward. He also depressed the third button for a long time. The time interval between the second and third button pushes became shorter and shorter when the monkey used his ipsilateral arm (from 836 ± 184 ms on day 1 to 319 ± 122 ms on day 10 for the clockwise sequence, and from 639 ± 98 ms on day 1 to 424 ± 92 ms on day 10 for the counterclockwise sequence).

The pattern of activation of the triceps also changed through the early period of monkey KE's training on the three-push task when he used his ipsilateral arm (Figs. 6 and 7). On the first 2-3 days, the duration of activation of the triceps at the second push was long (508 ± 134 ms for clockwise sequence, 375 ± 78 ms for counterclockwise sequence), as though the monkey did not understand that he had to move to the third button. But by the fourth day of training, the monkey moved his hand smoothly from the first to the second and then to the third button. The duration of the second and third button pushes became short, as shown in the brief periods of triceps activation (~ 334 ± 48.8 and 348 ± 71.4 ms for clockwise, 292 ± 48.0 and 275 ± 46.5 ms for counterclockwise).

In contrast to the large changes in movement kinetics and muscle activation shown when the monkey used the arm ipsilateral to MPTP infusion, relatively little change occurred when the monkey, over the same trials, used the arm contralateral to the MPTP lesion. As shown in Figs. 6 and 7, the interval between the second and third button presses remained long (592 ± 150 ms on day 1, 445 ± 131 ms on day 10 for the clockwise sequence; 804 ± 100 ms on day 1, 652 ± 172 ms on day 10 for the counterclockwise sequence).

After 3-4 days of training, the duration of activation of the triceps brachii muscle became short when the monkey pushed the third button. This shortening occurred for both the ipsilateral and the contralateral arm (and for both clockwise and counterclockwise sequences), as if the monkey realized that the third button was the final target to push in order to get reward.

Quantitative changes in task performance during learning of the three-push task are summarized in Fig. 8 and Table 1. The plots show the movement times of monkey KE for the first 18 days of training on the three-push task. An initial movement time (Fig. 8, A and C) was defined as the time interval from the release of the first button to the push of the second one. A serial movement time (Fig. 8, B and D) was defined as the time from release of the second button to the push of the third button. The initial movement times for movements with the arm contralateral to the MPTP infusion were significantly longer than those of the ipsilateral arm except for the first 3 days of learning in the clockwise series (Fig. 8A, for the 18 day averages, P < 0.0001, 2-tailed t-test). The serial movement times for the contralateral arm were also significantly longer than those of the ipsilateral arm, except for the first 4 days of learning (Fig. 8B, for the 18 day averages, P < 0.0001, 2-tailed t-test). Particularly striking was the variability in serial movement times for the contralateral arm, indicated by the large standard deviations (Fig. 8, B and D). Significant slowness of push button movements made with the limb contralateral to the dopamine depletion and large standard deviations also were observed in monkey GN (Table 1). These results confirmed the "hemi-Parkinsonian" status of the monkeys.

View larger version (29K): [in this window] [in a new window]   Fig. 8. Initial and serial movement times during learning of the 3-button push task in monkey KE in which MPTP was infused before training. The monkey was learning the 3-push task after having mastered the 1- and 2-push tasks. Cartoons of task above each group indicate movement made; black buttons indicate buttons at which reward was delivered. Plots show mean values for 20-300 trials in a given day. Error bars indicate standard deviations (SD). Filled and open symbols indicate, respectively, movement times for the contralateral and the ipsilateral arms. Asterisks indicate differences between the values for the ipsilateral and contralateral arms at P < 0.0001 (2-tailed t-test). A and C: initial movement time from button 1 to button 2. B and D: serial movement time from button 2 to button 3.

Unilateral dopamine depletion and changes in reaction times during learning of sequential motor tasks

To determine how prior dopamine depletion would affect the learning of motor sequences, we measured initial and serial reaction times through the learning process in monkey KE. This allowed us to compare performance in a situation that in the normal case would be reactive (initial reaction time) or predictive (serial reaction time). The initial reaction time was defined as the time between the onset of the second button light and the release of the first button. The serial reaction time was defined as the interval between the push of button 2 and the release of the same button in response to illumination of button 3.

When monkey KE underwent initial training on the two-push task having mastered the one-push task, his reaction times decreased for both arms. After the long hold time for button 1 (1 s), the reaction times for release of button 1 after illumination of button 2 decreased for the ipsilateral arm by ca. 91 ms from days 1 and 2 to days 5 and 6 (457 ± 65 to 366 ± 52 ms, P  0.0001, 2-tailed t-test). For the arm contralateral to the MPTP infusion, the decrease was ca. 41 ms (332 ± 119 to 286 ± 65 ms). Thus the percent declines for the ipsilateral and contralateral arms were quite similar (20 and 14%, respectively).

The initial and serial reaction times of monkey KE during training on the three-push sequential task are shown in Fig. 9 for the first 18 days after he had mastered the one- and two-push tasks. The initial reaction times were quite similar for the arms ipsilateral and contralateral to the MPTP infusion. The average initial reaction times for the full 18 days of learning were not significantly different for the ipsilateral and contralateral arms in the clockwise direction (Fig. 9A). Those for the ipsilateral arm were actually longer than those for the contralateral arm for the counterclockwise direction on some days of testing (P < 0.0001, 2-tailed t-test, Fig. 9C). The initial reaction times were also similar for clockwise and counterclockwise sequences, demonstrating that the monkey detected the button illumination for both the left and the right buttons and initiated reaching movements to them normally. The initial reaction times did not undergo significant shortening for either arm throughout the 18 day learning period.

View larger version (27K): [in this window] [in a new window]   Fig. 9. Initial and serial reaction times for monkey KE during learning of the 3-push task. The period of training shown is the same as that for Fig. 8. Plots show mean values of 20-300 trials in a day. Error bars indicate SD. A and C: initial reaction times. B and D: serial reaction times for movements from the 2nd to the 3rd button. Best fitting logarithmic curves are superimposed on the plots (B, y = 378 logx + 555 for ; y = 114 logx + 319 for ; D, y = 150 logx + 384 for ). *, significant difference between the values for the ipsilateral and contralateral arms at P < 0.0001 (2-tailed t-test).

In contrast to the initial reaction times, the serial reaction times were significantly different for the arms ipsilateral and contralateral to the side of MPTP infusion. As illustrated in Fig. 9B, the serial reaction times for the ipsilateral arm showed a decrease of >200 ms during the first 4 days of training (from ~520 ms on the first 2 days to ~310 ms on the 3rd and 4th days of learning the clockwise sequence; P < 0.0001, 2-tailed t-test). The high values for the ipsilateral arm reflect the long time during which the monkey continued to press the second button, which had provided reward (cf. Figs. 6 and 7). During this time, he also kept licking the spoon without reward (cf. Figs. 13 and 14). By the third and fourth days, the serial reaction times had shortened significantly in parallel with the rapid decline in duration of the triceps muscle activity in the ipsilateral arm (Figs. 6 and 7). Average serial reaction times for the ipsilateral arm decreased further to 120 ms by day 12. Values for the counterclockwise sequence fell less precipitously but otherwise showed similar patterns.

The serial reaction times for the arm contralateral to the side of the MPTP infusion did not undergo a rapid early decline and showed a much smaller decline during the rest of training. For the clockwise sequences, they fell only ~120 ms from an average of 320 ms (day 1-2) to an average of 200 ms (day 10-18), and values for the contralateral arm were significantly longer than those for the ipsilateral arm after 12 days of training (P < 0.0001, 2-tailed t-test). For counterclockwise sequences, the serial reaction times for the contralateral arm were also longer than for the ipsilateral arm throughout the learning period (Fig. 9D). Thus the effects of unilateral MPTP infusion on the acquisition of the sequential-button-push task appeared selectively for the arm contralateral to the side of MPTP infusion in both clockwise and counterclockwise sequences.

In performing the three-push task with the arm contralateral to the side of MPTP infusion, monkey KE continued for >12 days of training to lick for reward after the second push, which had been correct behavior in the previously learned two-push task. In the ipsilateral arm task, monkey KE almost completely abandoned the previously correct behavior in 6 days of training. This suggests that the monkey realized that the third rather than the second button was the reward-associated button after a few days of training when he performed with his ipsilateral arm, but that it took over 12 days for the monkey to realize the change of reward-associated button when he performed with his contralateral arm. This perseveration appeared to account for the lack of early decline in the serial reaction time for the contralateral arm.

Unilateral dopamine depletion abolishes the ability to learn predictive, preprogrammed movements in sequential motor tasks performed with the limb contralateral to the side of depletion

To visualize the temporal characteristics of the sequential button pushes by monkey KE, we constructed time trajectories for the three button pushes (Fig. 10). These time trajectories illustrate the averaged envelope of the movements by showing the initiation of the presses (rise times), the hold times (flat phases), and the releases (falling phases). Figure 10A depicts schematically how the time trajectory was calculated. The average time trajectories for clockwise sequences are plotted in Fig. 10, B and D, and those for counterclockwise sequences in Fig. 10, C and E. These plots were based on the performance of the three-push task during the fourth week of training on the three-push task. The average value for the hold times for each button are shown in Table 2.

View larger version (27K): [in this window] [in a new window]   Fig. 10. Temporal patterns of 3 sequential button pushes ("time trajectories") made by monkey KE. A: diagram showing how the time trajectory curves were obtained. In each trial, 3 button pushes are represented as 3 rectangles in which the time axes of both the abscissa and ordinate are durations of button pushes (a in A), and, in the abscissa, the 3 rectangles are arranged based on the times of push and release relative to the onset of GO signal (time 0). Time from the push of button 1 to the onset of the GO signal was fixed and was 1 s. In A, 3 rectangles obtained in 3 trials calculated in this way are superimposed. A time trajectory curve was then obtained by averaging ordinate values of the rectangles. B and D: average time trajectories in clockwise sequence. Values of means and SD are plotted. C and E: counterclockwise sequence. Vertical tick marks below the traces indicate the times at which the monkey pushed the 3rd button. Ipsilateral hand performance is shown on the right (D and E), contralateral hand performance on the left (B and C). In the average time trajectories, the onset of a push was determined as a time when the average hold time became significantly greater than the baseline (0), 2-tailed t-test, P < 0.01. Button-release time was determined as the time when the average hold time became so short that it was not significantly greater than the baseline. Hold times determined by the average time trajectories for the 3-push movements relative to the GO signal are longer than the average of hold times of each button push (Table 2), because they include movement times and reaction times.

The time trajectory plots show that the temporal pattern of sequential push movements was different for the arms ipsilateral and contralateral to the side of dopamine depletion. With the ipsilateral arm (Fig. 10, D and E), after the required long hold time for the first button, the second button was depressed for a brief time (230 ms for clockwise and 350 ms for counterclockwise sequences, determined by the time trajectory plot), and this was followed by a relatively long hold time for the third button (610 ms for clockwise and 580 ms for counterclockwise sequences), after which the monkey received reward water. The respective peak times for the second and third pushes in the average time trajectories (i.e., most frequent times of occurrence of the pushes) were at 490 and 800 ms after the GO signal for clockwise sequences and 560 and 920 ms afterward for counterclockwise sequences. There was considerable overlap of the average first-, second-, and third-push time trajectories.

By contrast, for the pushes that monkey KE made with his contralateral arm (Fig. 10, B and C), the average time trajectories for the three button pushes did not overlap. The different push movements clearly were separated. The average durations of the button pushes determined by the average time trajectories for the second and third buttons were, respectively, 340 and 910 ms for clockwise sequences and 460 and 600 ms for counterclockwise sequence. The average durations of pushes for the second and third buttons were longer for the contralateral arm than for the ipsilateral arm (by, respectively, 110 and 300 ms for clockwise, and 110 and 20 ms for counterclockwise directions). The total times from the GO signal to the release of the third button were also longer, by 710 ms in the clockwise direction and by 370 ms in the counterclockwise direction.

Unexpected task switch exposes automatic preprogrammed movement sequence

Despite the quantitative differences between movements made with the ipsilateral and contralateral arms during learning, these differences were not obvious on qualitative inspection. However, when we tested the monkey KE in an experimental paradigm involving an unexpected task switch, we found a clear qualitative difference in his performance when using the arms ipsilateral and contralateral to dopamine depletion.

The switch-task trials were introduced after the monkey had mastered the three-push task and his serial reaction times had shortened to asymptotic levels after 4-6 wk of training on the task. Then, during infrequent, randomly spaced trials, the task was switched from the three-push task currently being performed to the previously learned two-push task. The task switch experiments were performed once every 4-7 day during otherwise normal training. In the switched two-push task, when the monkey pushed the illuminated second button, the button 2 light was turned off and the liquid reward and associated click were given. The third button was left unilluminated. Thus,the monkey was given both visual and auditory stimuli indicating that the reward now would come after the button 2 press.

Figure 11 shows schematically the timing of button presses before (a) and after (b and c) the task switch for pushes made with the contralateral (Fig. 11B) and ipsilateral (Fig. 11D) arms. The filled symbols indicate pushes for which the monkey received reward. Before the switch, the monkey, with both the contralateral and ipsilateral arm, pushed the three buttons sequentially with the typical long time interval between the first and second button pushes and short time interval between the second and third button pushes. In the switched two-push task, when the monkey used the arm ipsilateral to the side of dopamine depletion, he continued to complete a sequence of three-push button movements without interruption at the second button for several tens of trials after the switch task began (Fig. 11D, b and c). Figure 11E shows the average number of pushes in 30 trials before and 112 trials after the task switch for pushes made with the ipsilateral hand for four switch experiments. The uncued third push did not fully disappear until after ~90 trials. It was as though the monkey had developed implicit knowledge of the three-push movement sequence during his previous training so that during the switch trials, even though the third button was not illuminated and even though reward and the reward click came at the second button press, the monkey completed the three-push movement sequence by automatically pushing the third button.

View larger version (32K): [in this window] [in a new window]   Fig. 11. Results in task-switch trials in which monkey KE had to switch unexpectedly from the 3-push task to the 2-push task. Monkey KE had learned the tasks after infusion of MPTP into the right striatum. A: diagram illustrating the task switch. B and D: timing of the button pushes during the 3-push task and the switched 2-push task performed with the contralateral (B) and ipsilateral (D) arms. Monkey received a liquid reward at times indicated (). Average number of pushes in 5 consecutive trials was determined for each test (see numbers at right). B and D, a-c: schematic examples of records of button pushes before (a) and after (b and c) the task switch. C and E: average number of pushes made before (3 push) and after (2 push) the task switch for the contralateral (C) and ipsilateral (E) arms as a function of task trials. Clockwise and counterclockwise sequences appeared in a random order in blocks of 100-300 trials, and results were averaged.

In sharp contrast, when monkey KE used the arm contralateral to the side of MPTP infusion, the push movements stopped promptly at the second button push after the task switch, with no sign of automatic execution of third-button pushes (Fig. 11, B, b and c, and C). This suggests that the sequential button-push movements made with the contralateral arm were performed in reaction to visual instruction and click sounds as cues for reward delivery, in contrast to the predictive, preprogrammed movements made with the ipsilateral arm.

To learn more about the differences in tactics of task performance of the monkey when he used his ipsilateral and contralateral arms, we measured the temporal parameters of the sequential push-button movements, including initial and serial reaction times, movement times, and hold times for each button before and after a set of switch trials. We reasoned that if the monkey performed the multiple button pushes as a single learned sequence of movements, then the temporal pattern of the entire set of push button movements, including the automatic extra pushes after the task was switched, should be similar to the temporal pattern of the movements before the task switch. This expectation was confirmed.

There was a close similarity between the temporal patterns of sequential button pushes made by monkey KE with his ipsilateral arm before and after the unexpected switch to the two-push task (Fig. 12, D and E). The average durations of pushes for the first, second, and third buttons after the switch were 1,440, 410, and 550 ms, respectively, as compared to values of 1,370, 350, and 580 ms before the switch. The peak trajectory times of pushes for the second and third buttons after the GO signal were 610 and 930 ms, respectively, after the switch, and 560 and 920 ms before the switch. The average time trajectories for button 2 and 3 pushes overlapped each other considerably even more than they did before the switch. In the switched task, the reward-associated clicks (vertical ticks in Fig. 12E) appeared immediately after the second-button push (520 ± 72.7 ms after the GO signal). The average onset of the automatic third button pushes was 700 ms after the GO signal, indicating that the automatic pushes occurred as late as 180 ms after the onset of the reward-associated clicks. Thus the monkey should have had enough time to detect the clicks before the third button push. Nevertheless, the average onset time of the third button pushes after the task switch was similar to that before the task switch (700 vs. 710 ms).

View larger version (32K): [in this window] [in a new window]   Fig. 12. Temporal patterns of performance on sequential button pushes made by monkey KE before (B and D) and after (C and E) the unexpected task switch in which the monkey received reward after the 2d button push instead of after the 3rd button push. A: task switch. B and D: average time trajectories of the push movements before the task switch. C and E: those of the push movements after the task switch. Format of illustration is the same as that in Fig. 10. Average time trajectories of button pushes in counterclockwise sequence are illustrated.

For the arm contralateral to the side of the MPTP infusion, the third push was missing, but the two pushes that were made after the task switch had a temporal pattern similar to that of the first and second pushes in the three-push task before the task switch. Both contralateral arm pushes showed a greater separation between the first and second holds and longer second hold durations than the ipsilateral arm pushes (Fig. 12, B and C).

Unilateral dopamine depletion and unilateral deficits in learning action strategies

Unilateral dopamine depletion not only induced slow movement times and prolonged serial reaction times for movements made with the contralateral arm but also affected other aspects of the monkeys' behavior. As they performed the sequential motor tasks, the monkeys made orofacial movements to consume liquid reward in addition to movements of the arms and eyes. To perform the tasks efficiently, they needed to learn a coordinated action strategy, which required activation of different body parts in the correct temporal order. For instance, arm muscles such as the triceps and biceps brachii and wrist extensor and flexors were activated during the sequential push-button movements, and then orofacial muscles were activated as the monkey consumed the reward delivered after the last push button movement. With training, the temporal pattern of activation of different body parts during the task performance became stereotyped when the monkeys used the arm ipsilateral to the side of dopamine depletion. By contrast, there was a striking retardation of the acquisition of such coordinated sequential movement patterns when the monkey used the arm contralateral to the side of dopamine depletion.

Figure 13 illustrates for monkey KE, representative activities of arm (triceps brachii) and orofacial (digastric) muscles during the course of the initial 12 days of training on the three-push task after the monkey had learned the one- and two-push tasks. During the first 2-3 days of training with the right arm (ipsilateral to the MPTP infusion), the monkey licked the spoon in front of his mouth when he pushed the second button, which had been the appropriate temporal order in the previous two-push task in which he expected reward delivery at button 2. The licks at the second push occurred in both clockwise (Fig. 13, right) and counterclockwise sequences (Fig. 14, right). The reward and the reward-associated click were now delivered after the third push, and the monkey licked the spoon again, after the third push, and he consumed the delivered liquid reward. The licks after the second push suggest that the monkey did not recognize that the task had been switched from the well-learned two-push task to the new three-push task. After 5 days of training, the monkey appeared to realize that reward would not be delivered after the second push because he stopped licking after the second push in most of the trials. In parallel with this change, activation of prime mover muscles for the second push became shorter in duration and shifted to the next activation for the third push (cf. Figs. 6 and 7). During the following several days of training, the duration of prime mover muscle activation and the interval between the second and the third push became shorter and shorter.

View larger version (59K): [in this window] [in a new window]   Fig. 13. Activities of arm muscles (triceps brachii) and orofacial muscles (digastrics) during the course of the initial 12 days of training of monkey KE on the 3-push task after the monkey had learned the 1- and 2-push tasks. Numbers on the right of traces indicate the occurrence of licks between the 2nd and 3rd button pushes (cf. Fig. 15). Muscle activity during clockwise sequence is shown. Symbols are the same as in Figs. 6 and 7.

View larger version (60K): [in this window] [in a new window]   Fig. 14. Same as Fig. 13, but for counterclockwise sequence.

Remarkably, when monkey KE used the arm contralateral to MPTP infusion during the same training period, he kept trying to lick at the second push for 8-12 days, even though by day 5 he no longer licked at button 2 when he used his ipsilateral arm (compare Figs. 13 and 14, left and right). Figure 15 plots the probability of occurrence of licks at the second push as a function of days of training. When the monkey used the arm ipsilateral to the MPTP infusion, licks at the second push occurred in <30% of trials after 6 days of training and in <5% of trials after 12 days of training. The declines occurred for both clockwise (Fig. 15A) and counterclockwise (Fig. 15B) sequences. When the monkey used his contralateral arm, there was no steady decline in licks after the second push. They declined only briefly, mainly for the clockwise series, for the first 3 days after the switch. The duration of activation of prime mover muscles for the second push also was still long even on the 8-10th day of training (Figs. 13 and 14, left). These observations suggest that the monkey was still expecting reward at the second push when he used his contralateral arm even after 8-12 days of training, when he was no longer showing signs of expecting reward at the second button when he used the ipsilateral arm.

View larger version (17K): [in this window] [in a new window]   Fig. 15. Probability curves of occurrence of licks at the second push in the 3-push tasks as a function of days of learning. , curves for contralateral hand task; , ipsilateral hand task. Each plot represents percentage of trials in which licking occurred at the second push in several tens of trials. A: clockwise sequence. B: counterclockwise sequence.

Eye movements before and after unilateral dopamine depletion

In agreement with the findings of Miyashita et al. (1995), we observed that both monkeys showed a tendency to pay less attention to visual objects appearing in the visual field contralateral to the MPTP infusion and tended to make fewer saccades to that side. We therefore examined eye movements as well as hand movements during the sequential button push tasks. Saccadic eye movements to each of the buttons occurred in addition to the arm movements. After learning, the saccades led movements of the ipsilateral arm during performance of the push button tasks (Fig. 16), suggesting that a coordinated, predictive eye-hand strategy had been acquired. There was greatly increased variability in the relative timing of the saccades and hand releases when the monkeys used the arm contralateral to the side of dopamine depletion.

View larger version (27K): [in this window] [in a new window]   Fig. 16. Timing of onset of saccades made by monkey KE from the 1st to the 2nd buttons relative to the time of hand release from the 1st button during 3-push task performed either with ipsilateral arm (C and D) or with the contralateral arm (A and B). Left: superimposed traces of vertical saccades from the 1st (hold) button to the 2nd button. Eye movement traces are centered at the saccade onset and the times when the monkey released the 1st button after an illumination of the 2nd button are marked by the vertical ticks above the traces. Two representative blocks of eye movement traces are illustrated. Right: relative times of saccade onset (0 on abscissa) and the hand releases of the 1st button. Data illustrated were obtained for 9 days between 120th and 128th day after MPTP infusion.

Figure 16 illustrates the timing of the onset of saccades from the first button to the second one relative to the times of hand release from button 1 made by monkey KE. In this monkey, not only the putamen but also the caudate nucleus was severely depleted of TH-like immunoreactivity, including the region representing oculomotor function (Hikosaka et al. 1989; Parthasarathy et al. 1992). Figure 16, A-D, left, illustrates superimposed traces of vertical saccades from the first (hold) button to the second button. The eye movement traces are centered at the saccade onset, and the times at which the monkey released the first button after illumination of the second button are marked by the vertical ticks above the traces. Two representative blocks of eye movement traces are illustrated. Figure 16, A-D, right, plots times of release of the first button relative to saccade onset. The plots show that saccade onset led hand release by ~120 ms on average when the monkey used the arm ipsilateral to the MPTP infusion (Fig. 16, C and D). With the contralateral arm, however, the lead times were smaller (average, 87.9 ms in the clockwise sequence; 99.2 ms in the counterclockwise sequence), and hand release occurred first and thus without guidance of saccades in a considerable number of trials, especially for the clockwise sequence (Fig. 16A). The trial-by-trial variations in relative saccade onset-hand release timing were greater when the monkey used the arm contralateral to the MPTP infusion than when he used his ipsilateral arm (6,128 vs. 3,592 for the clockwise sequence, and 3,896 vs. 2,935 for the counterclockwise sequence).

In addition to these ipsilateral and contralateral arm asymmetries, variations in relative saccade-hand release timing were larger in the clockwise sequences in which the initial saccade and hand movement went toward the hemifield contralateral to the MPTP infusion (ipsilateral hand, variance 3,592; contralateral hand, variance 6,128). In the counterclockwise sequences, in which the initial saccades and arm movements went toward the ipsilateral hemifield, the variance was 2,935 for the ipsilateral arm and 3,896 for the contralateral arm. The lowest variance in relative timing of saccade onset and hand release occurred when the monkey used the ipsilateral arm and performed the counterclockwise sequence, pressing the button in the ipsilateral half of extrapersonal space.

Unilateral dopamine depletion and change in reaction times during sequence retrieval and relearning

To examine the effects of dopamine depletion on the retrieval of learned sequences and on the relearning of sequences, we measured initial and serial reaction times in monkey GN, which received MPTP after learning the one- to four-push tasks. Figure 17, A and B, illustrates the monkey's performance before MPTP was given, after he had mastered the one-, two-, and three-push tasks and was learning the four-push task. Figure 17, C and D, illustrates the changes in reaction times as the monkey was tested on the four-push task after MPTP had been infused into the left striatum. The serial reaction times shown are for movements from the second to the third button.

View larger version (26K): [in this window] [in a new window]   Fig. 17. Serial reaction times from the 2nd to the 3rd button for monkey GN before and after MPTP infusion. A and B: performance of the 4-push task after monkey had learned the 1-, 2-, and 3-push tasks but before MPTP infusion. C and D: relearning after MPTP infusion. *, significant difference between values for ipsilateral and contralateral arms at P < 0.0001 (2-tailed t-test).

During monkey GN's original training on the four-push task before MPTP infusion, his serial reaction times from the second to the third button showed relatively small but significant shortenings for both the right arm and left arm tasks for both the clockwise and counterclockwise sequences (Table 3). The monkey's performance showed an arm asymmetry, however, because in the clockwise sequence, the reaction times were shorter for the right than for the left arm (Fig. 17A), whereas in the counterclockwise sequence, the reaction times were shorter for the left than for the right arm (Fig. 17B). The initial reaction times were shorter for the left arm for both directions of movement during the initial 4 days of training, but the difference became insignificant at the later stages of training (Table 3). These directional differences may have reflected a behavioral habit acquired during training.

During retraining on the four-push task after unilateral infusion of MPTP, no significant difference was observed between the initial reaction times from the first to the second button in the ipsilateral hand task and those in the contralateral hand task (Table 3). The serial reaction times from the second to the third button for the ipsilateral hand were short and stable for both the clockwise and counterclockwise sequences (Fig. 17, C and D). The serial reaction times during the initial relearning in the clockwise sequence (Fig. 17C, day 1-day 4, average 133.2 ± 42.0 ms) were slightly longer than those during the late stage of original learning (Fig. 17A, day 10-15, average 113.6 ± 29.8 ms). Contralateral to the side of MPTP infusion, serial reaction times during the first 4 days of retraining were longer than those of ipsilateral arm task, with one exception (day 4 for the clockwise sequence, Fig. 17C, day 4 for the counterclockwise sequence, Fig. 17D). In the clockwise sequence, the serial reaction times for movements with the contralateral arm (day 1-4, average 210.5 ± 100.3 ms, Table 3) were much longer than the stable serial reaction times before MPTP (day 10-15, average 85.8 ± 16.0 ms, Table 3). Similarly, in the counterclockwise sequence, the serial reaction times for movements with the contralateral arm (day 1-4, average 140.7 ± 71.4 ms, Table 3) were longer than the stable serial reaction times before MPTP (day 10-15, average 92.2 ± 18.6 ms, Table 3). On the other hand, the long serial reaction times for the contralateral arm diminished significantly through ~2 weeks of retraining, and the difference in the reaction times between the ipsilateral and contralateral arm became much smaller not only in the clockwise sequence (Fig. 17C) but also in the counterclockwise sequence (Fig. 17D).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our findings strongly suggest that the ability of monkeys to learn and to retrieve sequential motor tasks depends on the striatum and an intact nigrostriatal dopamine system. Our evidence for a role of the basal ganglia in procedural learning is in good accord with evidence for procedural learning defects in patients with Parkinson's disease (Knowlton et al. 1996; Pascual-Leone et al. 1993; Saint-Cyr et al. 1988) and in rodents with lesions of the caudoputamen (McDonald and White 1993). Our evidence further explicitly implicates the striatum and its nigral afferents in the development of predict