, Volume 187, Issue 1, pp 36–46

Sensorimotor effects of pergolide, a dopamine agonist, in healthy subjects: a lateralized readiness potential study


    • Georg-Elias-Müller-Institut für PsychologieUniversität Göttingen
  • Jutta Stahl
    • Georg-Elias-Müller-Institut für PsychologieUniversität Göttingen
Original Investigation

DOI: 10.1007/s00213-006-0400-9

Cite this article as:
Rammsayer, T. & Stahl, J. Psychopharmacology (2006) 187: 36. doi:10.1007/s00213-006-0400-9



The major purpose of the present study was to further elucidate dopaminergic modulation of sensorimotor processing in healthy human subjects.

Materials and Methods

To more specifically analyze dopaminergic effects on premotor and motor stages of sensorimotor processing, lateralized readiness potentials (LRPs) were obtained. In a randomized double-blind crossover design, either 0.075 mg of the D1/D2 dopamine (DA) agonist pergolide or placebo were administered to 12 healthy male volunteers ranging from 19 to 25 years in age. The subjects performed a two-choice visual reaction time task. In addition to behavioral measures, such as response speed and error rate, stimulus-locked LRP (S-LRP) and response-locked LRP (LRP-R) latencies were determined. To better dissociate potential central and peripheral motor effects, measures of response dynamics and response-locked electromyogram (EMG-R) recordings were also obtained.


Pergolide reliably enhanced speed of stimulus-related information processing as indicated by shorter S-LRP latencies while LRP-R latencies, reaction time, and indicators of response dynamics were not influenced by DA agonistic treatment. Furthermore, lower EMG-R amplitudes and an increased number of wrong-hand responses were observed under pergolide compared to placebo.


The results indicate that dopaminergic neurotransmission effectively modulates early perceptual and cognitive stages of information processing as suggested by neural network models of the functional role of prefrontal DA. The lack of an effect on aspects of motor processing may be due to a higher capacity of the nigrostriatal compared to the mesocortical DA system to compensate pharmacologically induced changes in dopaminergic activity.


PergolideHealthy subjectsSensorimotor processingLRPReaction time


Several lines of evidence suggest that the neurotransmitter dopamine (DA) effectively modulates sensorimotor processing. Nevertheless, a contentious issue refers to the question of what specific processing stages are controlled by DA activity in the brain. Within the information-processing framework (cf., Anderson 1980; Miller 1988; Sanders 1977, 1983; Sternberg 1969, 2001), a stage represents a functional set of elementary operations. Stages involved in sensorimotor processing can be assigned to premotor and motor processes. Premotor processes include early perceptual and cognitive processes, which are involved in stimulus-associated processing such as stimulus encoding, stimulus identification, and stimulus evaluation. Motor processes, on the other hand, refer to central and peripheral response-associated operations such as response selection, motor programming, muscular activation, and response execution (van der Molen et al. 1991).

Based on the results of animal studies, Stricker and Zigmond (e.g., Heffner et al. 1977; Zigmond et al. 1980) introduced a model for action of the mesostriatal DA system. According to their model, dopaminergic neurons reduce the threshold for behavioral responses to sensory input. A slight increase in dopaminergic activity should cause behavioral facilitation as indicated by faster response times while, on the other hand, a more pronounced increase in dopaminergic activity should prevent an appropriate discriminative response due to dopaminergic overstimulation. Decreased dopaminergic activity should also similarly result in impaired sensorimotor performance because the amount of sensory input is no longer adequate to elicit an appropriate behavioral response (Stricker and Zigmond 1986; Zigmond et al. 1980). Although animal and human studies provided convincing evidence for the notion that substantial deviations from the physiological level of dopaminergic activity, resulting in either hypo- or hyperdopaminergic states, effectively interfere with information processing (Barch 2004; Castner et al. 2004; Robbins 2000), this model fails to specify whether dopaminergic effects on sensorimotor processing are primarily related to premotor or motor processing stages.

Oades (1985) introduced a model in which mesolimbic DA activity is hypothesized to be the key mediator of stimulus–response selection. According to his model, enhanced dopaminergic activity increases the probability of task-irrelevant stimuli to effectively interfere with the processing of task-relevant information. Thus, Oades’ (1985) model proceeds from the assumption that dopaminergic modulation of sensorimotor processing occurs at the premotor level.

According to another theoretical account of dopaminergic modulation of sensorimotor information processing, the neurotransmitter DA contributes to the stability of cortical representations of internal and external stimuli by controlling the signal-to-noise ratio of neural networks involved in information processing (Servan-Schreiber et al. 1990; Winterer and Weinberger 2004). Within this framework, a moderate increase in dopaminergic activity may enhance the signal-to-noise ratio, while reduced dopaminergic activity causes a decrease in signal-to-noise ratio that is predicted to result in impaired stimulus processing at premotor stages such as stimulus encoding, stimulus identification, or stimulus evaluation (cf., Arnott et al. 2001; Bloxham et al. 1987; Cooper et al. 1994).

Although theoretical and neural network approaches appear to be consistent with the notion that premotor stages of sensorimotor processing are controlled by DA activity, the available empirical data are less conclusive. The vast majority of evidence for the involvement of DA in sensorimotor processing comes from the study of Parkinson’s disease (PD). PD is a neurodegenerative and progressive disorder of the basal ganglia characterized by a selective loss of dopaminergic neurons, predominantly in the substantia nigra pars compacta (Mandir and Vaughan 2000). The decline of these neurons results in a depletion of DA in the dorsal striatum (Marsden 1992). As the basal ganglia are strongly involved in motor control and impaired motor activity is a defining feature of PD, it is no surprise that PD patients tend to have longer response latencies than age-matched healthy controls across a wide range of experimental paradigms (e.g., Heilman et al. 1976; Jahanshahi et al. 1992; Rafal et al. 1987; Wascher et al. 1997). Thus, processing stages related to motor function have been considered the most obvious stages in which to expect DA to effectively modulate sensorimotor processing. This is supported by the fact that PD patients tend to have prolonged reaction times even in tasks with minimal perceptual and decision requirements (e.g., Bloxham et al. 1987; Evarts et al. 1981).

Central nervous dopaminergic activity, however, does not only play a crucial role in motor behavior but also in premotor processes such as selective attention and stimulus encoding (Crofts et al. 2001; Kahkonen et al. 2001; Saint-Cyr 2003; Servan-Schreiber et al. 1998b; Volkow et al. 1998). This is consistent with recent psychophysiological analyses of premotor and motor processes in PD, revealing impaired response selection and increased premotor and motor processing for PD patients compared to healthy controls (Low et al. 2002; Seiss and Praamstra 2004).

Neurochemical studies on the dopaminergic modulation of sensorimotor processing in healthy subjects clearly indicate that a pharmacologically induced decrease in DA activity, for instance by means of traditional neuroleptics such as haloperidol, interferes with sensorimotor processing at both the premotor and motor level (e.g., Grübel-Mathyl 1986; Rammsayer 1989, 1997). A less unambiguous picture, however, emerges with regard to pharmacologically induced increases in the effective level of brain DA activity. Schück et al. (2002) reported improved alertness and faster sensorimotor processing speed, as indicated by a visual simple reaction time task, after acute administration of the D2/D3 receptor agonist piribedil. A single 200-mg dose of the DA precursor levodopa similarly sped up visual choice reaction time in healthy volunteers (Rihet et al. 2002). Applying Sternberg’s (1969) additive factor method, Rihet et al. (2002) concluded that dopaminergic activity primarily modulates premotor stimulus processing rather than response selection or subsequent motor stages.

Psychopharmacological studies on the effect of methylphenidate on information processing yielded shorter response latencies under methylphenidate compared to placebo. This DA agonistic effect, however, was brought about by drug effects on processes involved in response organization rather than in stimulus analysis (Brumaghim et al. 1987; Fitzpatrick et al. 1988; Naylor et al. 1985).

Therefore, the major aim of the present study was to further elucidate the dopaminergic modulation of sensorimotor processing by investigating the effect of a moderate dose of the D1/D2 receptor agonist pergolide on behavioral and psychophysiological indicators of response speed and response dynamics. In addition to behavioral measures, such as reaction time and error rate, lateralized readiness potentials (LRPs) were obtained. The very few previous electroencephalogram (EEG) studies on the effects of a pharmacologically induced increase in brain DA activity on sensorimotor processing in healthy subjects focused on spectral EEG analyses (e.g., Luthringer et al. 1999; Schück et al. 2002) or the P3 component of the event-related potential (e.g., Brumaghim et al. 1987; Fitzpatrick et al. 1988). Over the last two decades, however, LRP has been established as a novel promising tool in chronometric studies of sensorimotor information processing (e.g., Gratton et al. 1988; Kutas and Donchin 1980). The LRP is derived from the readiness potential that appears several hundred milliseconds before voluntary hand movement and is larger contralateral to the hand to be moved. In choice response tasks with either of the two hands, amplitudes have been found to be larger at scalp sites contralateral to the activated hand (Gratton et al. 1988). Hence, LRP reflects the asymmetrical cortical activation contra- minus ipsilateral to the responding hand. The onset of the LRP can be considered a time marker intervening between the onset of stimulus presentation and response execution. The interval between stimulus onset and LRP onset is accordingly defined as the stimulus-locked LRP (S-LRP) latency. This interval indicates processes before response selection of the proper hand. Furthermore, the interval between the onset of the LRP and completion of the motor response is referred to as the response-locked LRP (LRP-R) latency and, thus, reflects the time course of processes involved in central response organization and execution of the motor response (Osman and Moore 1993).

Most surprisingly, up to date, LRP analyses do not appear to have been applied to investigate dopaminergic modulation of the transmission of sensory input into motor output in healthy volunteers. If enhanced dopaminergic activity accelerates speed of stimulus analysis, this should become evident in the S-LRP latency which represents the time required from onset of stimulus presentation until initiation of central motor activation. On the other hand, if aspects of response organization, such as response activation or response initiation, require less time with enhanced DA activity, this should be reflected by shorter LRP-R latencies. Based on these considerations, we introduce the use of LRPs to more specifically analyze dopaminergic effects on premotor and motor stages of sensorimotor information processing.

To better dissociate central and peripheral effects, measures of response force (RF), time to peak (TTP), and response-locked electromyogram (EMG-R) recordings were obtained as additional indicators of peripheral motor processing. While RF and TTP represent important aspects of motor response dynamics (Ulrich et al. 1999), averaged EMG-R latencies served as a chronometrical indicator of the time required from the onset of muscle activation to the onset of the overt response (Mueller-Gethmann et al. 2000). If response dynamics were effectively modulated by dopaminergic mechanisms, this should become evident in DA-induced changes in RF and/or TTP. Furthermore, as LRP latencies comprise aspects of both central motor activation and execution of the overt motor processes, EMG-R recordings are required to dissociate central and peripheral components of motor processing. Hence, if EMG-R measures were affected by the D1/D2 agonist pergolide, this would be indicative of a dopaminergic influence on peripheral aspects of motor response execution.

Materials and methods


The participants were 12 healthy male volunteers ranging from 19 to 25 years (mean±SD=22.4±1.8 years) in age. After filling in a health questionnaire and screening for drugs of abuse, the participants were selected according to the following inclusion criteria: non-smoking, no chronic drug intake, no past or present psychotherapy or psychiatric treatment, no allergy, no chronic endocrine or cardiovascular disease, and no signs of present acute or chronic infections or gastrointestinal diseases. All subjects were informed about the study protocol and the possible treatments applied in this experiment as well as the possible side effects of the drug and gave their written informed consent. The study was approved by the ethical committee of the German Psychological Society.

Sensorimotor task

Apparatus and stimuli

As response signals, the letters ‘V’ and ‘W’ were presented for 1,000 ms in the center of a computer screen subtending a visual angle of about 0.60°. Force-sensitive keys, composed of leaf springs, were used to assess RF and response time (RT) (cf., Giray and Ulrich 1993). At one end of the force key, an adjustable clamp held the leaf spring, while the participant pressed the other free end with his index finger. A force of 1,000 cN bent the free end of the leaf spring by approximately 2 mm. Strain gauges were attached near the leaf spring’s fixed end. Each response key was mounted on a board providing full forearm support. Thus, when the participant pressed the key, the strain gauges produced an analogue electrical signal corresponding to the force applied to the leaf spring. This signal was digitized and recorded for 2,000 ms with a sampling rate of 500 Hz starting at the onset of the visual response signal. To maintain a constant posture, a chin rest was fixed at a 65-cm distance to the computer screen.


The participants were instructed to respond as fast as possible with one hand to the ‘V’ and with the other hand to the ‘W’, but without making too many errors. The stimulus–response assignment was constant within each participant but balanced across participants. The participant responded with a brief flexion of either the right or the left index finger. An entire experimental session consisted of five blocks and lasted approximately 20 min. The first block served as practice block to ensure that the participants were familiar with the task and the stimulus–response assignment. The practice block was followed by four experimental blocks. Each block consisted of 25 left-hand and 25 right-hand trials. The participant initiated the next block when feeling ready to continue the experiment. The time between blocks averaged about 2 min across all participants. Order of trials was randomized within each block.

During the practice block, visual correctness feedback was provided after each trial. No such trial feedback, however, was presented for the experimental blocks. At the beginning of each trial, a white fixation cross was presented for 500 ms as a ready signal. To minimize anticipatory responses, the duration between the ready signal and the onset of the visual response signal corresponded to the sum of a constant period of 500 ms and an exponentially distributed random variable with a mean of 2,000 ms. Between blocks, visual feedback on overall mean RT and overall mean percentage of errors was provided on the monitor screen.

Mean RT, intraindividual variability of RT, and mean peak RF were determined for each type of trial across the four experimental blocks as measures of performance. The first moment at which RF exceeded a criterion of 50 cN after onset of the visual response signal was defined as RT, whereas the maximum force value recorded during a given trial represents peak RF. Time to peak was determined as an additional indicator of response dynamics. TTP referred to the temporal interval from onset of the overt motor response to the point in time when peak RF was reached. Wrong-hand responses, misses, and RTs greater than 1,000 ms were considered response errors and, therefore, also excluded from data analysis.

Electrophysiological recordings

The EEG was recorded with Ag/AgCl electrodes at five scalp loci. Scalp sites Fz, Cz, and Pz were selected according to the standard international 10–20 system (Jasper 1958). Sites C3′ and C4′ were positioned 1 cm anterior to C3 and C4, respectively. These positions were chosen because of their correspondence to the hand areas of the precentral motor cortex (cf., Coles 1989). Horizontal and vertical movements in the electrooculogram were recorded from electrode positions supra- and infraorbitally to the right eye and 2 cm external to the outer canthus of each eye. The active reference electrode was fixed on the left mastoid, the passive electrode on the right mastoid. Furthermore, EMG activity of the ventral forearm was obtained by bipolar recordings at the point trisected by the wrist–elbow distance and at the site 5 cm ventral to this point.

The recording of the electrophysiological data was continuous (1,000 Hz sampling rate). A NeuroScan data acquisition unit was used. The time constant was infinite and a low pass filter was set to 70 Hz. For the stimulus-locked analyses, the EEG signal was epoched off-line with epochs lasting from 100 ms before onset of the visual response signal until 900 ms after signal offset. For response-locked averaging of EEG and EMG, an epoch ranged from 300 ms before until 300 ms after reaching the response criterion of 50 cN. EMG was rectified off-line. For both, stimulus- and response-locked analyses, the 100-ms interval preceding the onset of the response signal served as a reference for baseline correction. All data were screened for artifacts, and contaminated trials exceeding maximum/minimum amplitude of ±70 μV were rejected. Data were digitally re-referenced to active and passive mastoid (cf., Picton et al. 1995).

LRP waveforms were recorded from electrodes located above the left and right motor cortices (C4′ and C3′, respectively) and were calculated as suggested by Coles (1989). Difference waveforms were computed between both electrode sites as a function of the responding hand in a given trial. On each trial, the recording from the ipsilateral hemisphere was subtracted from contralateral recordings. Thus, on a trial requiring a left-hand response, the difference wave C4′ minus C3′ was computed, whereas it was C3′ minus C4′ for right-hand trials. This procedure eliminates all electrical brain activity distributed symmetrically across the scalp. The single-trial difference waveforms were then averaged separately for left- and right-hand trials. Finally, in a last step, the LRP was computed as the mean of the average difference waves. Deviations of the resulting LRP from the zero line toward increased negativity indicate the activation of the correct response hand at the level of the motor cortex.

A jackknife-based scoring method with a relative criterion of 50% of the peak LRP amplitude was chosen for measuring S-LRP and LRP-R onset latencies, respectively (cf., Miller et al. 1998). This scoring technique has been developed to increase the low signal-to-noise ratio commonly observed with LRP data before onset detection. The jackknife procedure estimates the variability between participants by temporarily omitting each participant’s LRP waveform once. The contribution of a given individual to the grand average of all participants can be determined by the effect of leaving out his or her data. More specifically, if participants show a markedly different effect, the results should fluctuate notably depending on which participant is left out. On the other hand, if all participants show the same effect, the resulting means should only be slightly affected by leaving out a single participant. Hence, the influence of background noise is reduced before onset detection and LRP onset detection becomes much more reliable. For statistical testing, n grand average LRP waveforms were computed before LRP onset detection. Each of these grand average LRP waveforms included n−1 LRP waveforms because of the above-mentioned omitting procedure. Each LRP waveform was consequently included n−1 times in the data analyses, which conspicuously decreased standard error of the mean (SEM). For t test, the corrected SEM of differences was used (for a review, see Miller et al. 1998, Eq. 2). After a related rationale, the signs of correlation coefficients computed between a jackknifed and a non-jackknifed variable had to be inverted as shown by Stahl and Gibbons (2004). This applies to the respective correlation coefficients reported in the “Results” section.

Single trials were averaged according to their electrode position. Response-locked EMG (EMG-R) onsets were determined when EMG activity deviated 10% of the EMG amplitude from baseline (Osman et al. 1995, 2003). EMG-R latency was defined as the temporal interval from the EMG-R onset to the key press. EMG-R amplitude was measured as the maximum positive peak within an interval starting 100 ms before and ending 100 ms after having reached the response criterion. Behavioral and electrophysiological data were based on artifact-free correct responses.

Control for side effects

To control side effects of the drug, self-ratings on feelings of alertness, drowsiness, relaxation, concentration, nervousness, irritability, and energy were to be marked on a rating scale ranging from 1 (i.e., “not at all”) to 10 (i.e., “extremely pronounced”).

Pharmacological treatment and time course of the experiment

There is some evidence that both D1 and D2 receptors are involved in sensorimotor information processing. While premotor information processing appears to be modulated by prefrontal dopaminergic mechanisms depending primarily on D1 receptor activity (e.g., Müller et al. 1998; Goldman-Rakic et al. 1997), motor processes seem to be effectively influenced by mesostriatal D2 receptor activity (e.g., Frank 2005; Saint-Cyr 2003). Therefore, to investigate dopaminergic influences on premotor and motor components of sensorimotor information processing, the activity of both D1 and D2 receptors should be experimentally varied. For this reason, the mixed D1/D2 receptor agonist pergolide was applied in the present study. In a double-blind crossover design, a combination of either 0.075 mg of pergolide and 30 mg of domperidone or placebo and 30 mg of domperidone was administered orally in balanced order to each participant. A dose of 0.075 mg of pergolide was chosen based on the findings of a previous study of the effects of pergolide on human working memory performance (Müller et al. 1998) and on the results of a pilot study. Such a relatively low dose was expected to avoid unwanted side effects unrelated to sensorimotor processing. Furthermore, as both relatively low and high levels of dopaminergic activity should be associated with impaired performance (Barch 2004; Castner et al. 2004; Robbins 2000; Seamans and Yang 2003; Stricker and Zigmond 1986), a rather low dose of pergolide was used to ensure the occurrence of a facilitating dopaminergic effect on sensorimotor processing.

Pergolide is a semisynthetic ergoline dopamine D1/D2 receptor agonist. Receptor binding studies indicate that pergolide is more selective for D2 than for D1 receptors (Jenner 1995; Markham and Benfield 1997; Piercey et al. 1996). The mean time to reach peak plasma concentrations is 2.5 to 3 h after oral administration (Markham and Benfield 1997). The most common side effects of pergolide observed in Parkinson patients are nausea, dyskinesia, hallucinations, drowsiness, and insomnia (Olanow et al. 1994). To reduce unwanted peripheral side effects of pergolide, especially nausea, pretreatment with 30 mg of domperidone, a peripherally acting D2 antagonist, was performed. This is a well-established pharmacological procedure in the treatment of patients suffering from PD (cf., Oertel and Quinn 1996). When administered concomitantly with dopamine agonistic anti-Parkinson drugs, domperidone suppresses peripheral dose-limiting side effects such as nausea and vomiting, while the beneficial central effects of anti-Parkinson drugs are not aggravated (Brogden et al. 1982; Costall et al. 1979). Peak plasma concentrations are attained within approximately 30 min after oral administration of domperidone (Brogden et al. 1982). Unlike other dopamine antagonists, domperidone, a derivative of benzimidazole, is particularly well tolerated and rarely caused any marked side effects (Brogden et al. 1982). Lactose in identical capsules was used as placebo.

The experimental sessions were preceded by a practice session to familiarize the participants with the experimental tasks. The participants were tested at 1-week intervals. Each test day was initiated with oral administration of 30 mg of domperidone at 08:00 hours. Pergolide or placebo was applied 30 min later. One hour after drug intake, at 09:30 hours, a light standard breakfast was served. At 11:30 hours, the treatment measurement was started with the self-rating scales, followed by the sensorimotor task. Testing lasted approximately 30 min.


Screening for side effects that might affect sensorimotor performance yielded no evidence for drug-induced changes in subjective feelings of alertness, drowsiness, relaxation, concentration, nervousness, irritability, and energy (see Table 1).
Table 1

Mean scores and standard errors of the mean for self-rated feelings on arousal-related and emotional states under placebo and pergolide










Self-ratings on



















































Also given are t values and levels of statistical significance

Behavioral data

Table 2 shows means and standard errors of the mean for all behavioral measures. A reliable effect of pergolide could be established for hand errors [t(11)=3.6, p<0.01]; the percentage of wrong-hand responses was significantly increased by 30 mg of pergolide (3.33%) compared to placebo (0.25%). Under pergolide, the participants also tended to miss a response more often than under placebo. This tendency, however, failed to reach the 5% level of statistical significance [t(11)=1.9, p=0.08]. Intraindividual RT variability was significantly smaller under pergolide compared with placebo [t(11)=2.95, p<0.05]; mean within-subject standard deviations were 79.4 and 110.8 ms, respectively. All other behavioral measures were virtually unaffected by dopamine agonistic pharmacological treatment (see Table 2).
Table 2

Means and standard errors of the mean for behavioral and electrophysiological measures obtained under placebo and pergolide










Slow responses (%)







Wrong-hand responses (%)







No response (%)







Reaction time (ms)







RT variability (ms)







Response force (cN)







Time to peak (ms)







S-LRP latency (ms)







LRP-R latency (ms)







EMG latency (ms)







EMG amplitude (μV)







Also given are t values and levels of statistical significance

aAdjusted values as described by Miller et al. (1998) due to jackknife-based onset detection

Electrophysiological data

The stimulus-locked LRP waveforms for the placebo and the pergolide conditions are depicted in Fig. 1. As the applied jackknife procedure strongly reduces SEM, obtained t values and SEM were corrected according to a procedure described by Miller et al. (1998). Pergolide markedly decreased S-LRP latencies compared to placebo [t(11)=6.5, p<0.001]; S-LRP latencies were 289 and 332 ms under pergolide and placebo, respectively. This finding indicates a reliable enhancing effect of pergolide on the speed of stimulus-information processing. There was no indication of a similar effect on LRP-R latencies [t(11)=0.85, p=0.50]. Analysis of LRP-R latencies rather revealed that the time required for organization and execution of the motor response was almost the same for the pergolide (95 ms) and the placebo (90 ms) condition.
Fig. 1

Stimulus-locked (top panel) and response-locked (bottom panel) lateralized readiness potentials under placebo and pergolide. S-LRP latencies were 289 and 332 ms under pergolide and placebo, respectively. The time required for organization and execution of the motor response, as indicated by LRP-R latencies, was almost the identical for the pergolide (95 ms) and the placebo (90 ms) condition (see also Table 2)

Grand average response-locked EMG waveforms are presented in Fig. 2. Mean EMG-R latencies after 30 mg of pergolide and placebo were almost identical [t(11)=0.71, p=0.46]; mean latencies were 54 and 52 ms for pergolide and placebo, respectively. However, a significant effect of pergolide on EMG-R amplitude [t(11)=4.95, p ≤0.01] indicated that mean EMG-R amplitude was reliably reduced by pergolide (25.5 μV) compared to placebo (35.4 μV).
Fig. 2

Response-locked electromyogram under placebo and pergolide

Additional correlational analyses of the relationship between RT, S-LRP latencies, LRP-R latencies, EMG latency, and EMG amplitude yielded a reliable positive correlation between S-LRP latency and RT (r=0.62, p<0.05) after treatment with pergolide (see Table 3). This finding indicates shorter S-LRP latencies with faster responses. No other significant correlations could be revealed.
Table 3

Correlations between reaction time and electrophysiological measures


Reaction time

S-LRP latency

LRP-R latency

EMG latency



 S-LRP latency



 LRP-R latency




 EMG latency





 EMG amplitude







 S-LRP latency



 LRP-R latency




 EMG latency





 EMG amplitude





Please note that signs of correlation coefficients computed between a jackknifed and a non-jackknifed variable are inverted as suggested by Stahl and Gibbons (2004)

*p<0.05 (two-tailed)


In the present study, the effects of the DA D1/D2 receptor agonist pergolide on sensorimotor processing were investigated. To prevent adverse side effects that could interfere with performance on the sensorimotor task, a rather low dose of 0.075 mg of pergolide that did not produce any significant changes in self-rated measures of arousal and emotional states was administered. Furthermore, considering the well-documented functional relationship between DA activity and performance that can be characterized by an inverted U-shaped function, with rather low and high levels of dopaminergic activity being associated with impaired performance (Barch 2004; Castner et al. 2004; Robbins 2000; Seamans and Yang, 2003; Stricker and Zigmond 1986), such a low dose should warrant potential facilitating effects on sensorimotor processing to become evident. It should be noted that the present study did not aim at proving a U-shaped effect of DA on sensorimotor performance. For this latter purpose, the effects of both a DA agonist and antagonist should have been directly compared.

Behavioral measures of response speed and response dynamics were not affected by pergolide. In contrast to these behavioral indicators of sensorimotor processing, analyzing LRP latencies turned out to be a more sensitive approach for investigating pharmacologically induced effects on sensorimotor processing. S-LRP latency was reliably shorter under pergolide compared to placebo. As S-LRP latency reflects the duration of processes that occur before central response activation, the present finding indicates that pergolide effectively reduced the time required for early perceptual and cognitive stimulus processing including aspects such as stimulus encoding, stimulus identification, and stimulus evaluation. At the same time, LRP-R latency, representing the time necessary for central response organization and execution of the overt motor response, was not influenced by pergolide. This differential effect on S-LRP and LRP-R latencies is consistent with the finding that the DA precursor levodopa enhanced speed of stimulus-related information processing but did not affect succeeding stages of response selection and motor adjustment (Rihet et al. 2002).

Our finding of a reliable positive correlational relationship between S-LRP latencies and RT under pergolide shows that faster sensory stimulus processing was accompanied by faster responses. Although this result is indicative of a functional relationship between RT and speed of sensory information processing as assessed by S-LRP latency, the sum of S-LRP and LRP-R latencies cannot be considered functionally equivalent to behaviorally assessed RT (cf., Osman et al. 1995, 2003). The reason for this may be twofold. Osman et al. (1995, 2003) provided first evidence for a temporal overlap of premotor and motor processing stages which might violate the assumption of pure additivity. Thus, pergolide-induced acceleration of sensory/cognitive stimulus processing, as indicated by shorter S-LRP latency compared to placebo, resulted in a reduced overlap while behavioral RT remained unaffected. A second putative explanation refers to the common principle of averaging of EEG waveforms. In the present study, each participant’s averaged LRP waveform was based on approximately 180 single-trial waveforms. As there always is latency variability across trials, the shape of the averaged waveform is more or less blurred as a function of the amount of latency jitter. Proceeding from the assumption that pergolide reduced intraindividual S-LRP onset variability due to its beneficial effect on stimulus processing, the observed difference between pergolide and placebo could be caused by a smaller amount of latency jitter after pergolide treatment. Thus, under pergolide, LRP onset could occur earlier than under the placebo condition. Unfortunately, it is not possible to directly test this hypothesis as intraindividual LRP onset variability cannot be identified by means of commonly used LRP paradigms. Nevertheless, our finding of a reliable decrease of intraindividual RT variability under pergolide is consistent with the general notion of more efficient and, thus, less variable information processing with increased dopaminergic neurotransmission (cf., Bloxham et al. 1987; Cooper et al. 1994).

A neural network model introduced by Servan-Schreiber et al. (1998a,b) specifies that effects of pharmacologically increased dopaminergic neurotransmission depend on distinct task characteristics. According to this model, enhanced DA activity accelerates speed of premotor processing in situations in which participants have to maintain preparedness for a stimulus and rapidly execute a response that depends on the identity of the stimulus. On the other hand, when a task does not involve response competition, dopaminergic effects should become most noticeable in terms of changes in speed of response execution. The central task demands of the sensorimotor task applied in the present study included identification of the imperative stimulus and a speeded response depending on stimulus identity. This may account for our finding of a reliable pergolide-induced reduction in premotor processing speed as indicated by shorter S-LRP latency compared to placebo, while speed of motor processing, as indicated by LRP-R latencies, remains unaffected.

Our sensorimotor task also certainly involved a marked attentional component. To respond as quickly as possible, the participants had to pay attention to the onset of the imperative stimulus. Furthermore, there is no doubt that improved attentional control can be expected to lead to better encoding and evaluation of stimulus information. Both of these aspects of attention may be modulated by DA activity. Although the basal ganglia may be involved in fundamental aspects of attentional control (Saint-Cyr 2003), a growing amount of evidence from experimental studies and computational modeling (for recent reviews, see Barch 2004; Castner et al. 2004; Seamans and Yang 2003) suggests that prefrontal D1 receptor activity is crucial for attention and executive functions associated with working memory tasks that involve transient neural representations of information to guide an immediate response. From this perspective, the beneficial effect of pergolide on speed of early sensorimotor processes, such as stimulus encoding, stimulus identification, and stimulus evaluation, may be mediated by stimulation of D1 receptors in the prefrontal cortex. Such an explanation is also compatible with the neuromodulatory role of DA on signal-to-noise ratio within the context of internal and external stimulus processing (Servan-Schreiber et al. 1990; Winterer and Weinberger 2004).

Numerous studies on sensorimotor performance in PD provided converging evidence for the notion that DA modulates motor function through a nigrostriatal–thalamocortical loop (e.g., Alexander et al. 1986; Boraud et al. 2002; Mink 1996; Seiss and Praamstra 2004). The absence of an effect on speed of motor organization and motor execution in the present study points to the conclusion that dopaminergic neurons of the nigrostriatal–thalamocortical loop were less susceptible to pharmacological manipulations of DA receptor activity than dopaminergic neurons modulating prefrontal networks subserving working memory functions. Some indirect corroboration for such a notion represents the finding of an extremely high capacity for dopaminergic homeostasis within the nigrostriatal–thalamocortical loop (cf., Rammsayer 1989). This is, for instance, indicated by the fact that classical motor signs and symptoms of PD will not be observed unless nigostriatal DA cell loss exceeds a level of 70 to 80% (Agid 1991; Harden and Grace 1995). The observed lack of an effect on motor processes may alternatively be due to pre–post synaptic binding aspects. Pergolide is active at postsynaptic D1 and D2 receptors but is also a potent agonist at presynaptic D2 receptors (Markham and Benfield 1997). From this perspective, stimulation of presynaptic D2 receptors may have produced a DA antagonistic effect that could have partly compensated pergolide’s DA agonistic effect exerted at postsynaptic D2 receptors.

The rather unexpected finding of smaller EMG-R amplitude for the pergolide compared to the placebo condition indicates that less muscle activation was produced under pergolide. This phenomenon could be explained within the framework of Näätänen’s (1971) model of motor readiness (for review, see also Niemi and Näätänen 1981). According to this model, motor readiness is considered a function of the balance between motor excitation and motor inhibition when performing a reaction time task. With increasing motor excitation relative to motor inhibition, the effective level of motor preparedness increases until the threshold for response execution is reached. Among other factors, more efficient stimulus analysis, as for instance due to an enhanced signal-to-noise ratio, may exert a beneficial effect on motor preparedness (cf., Niemi and Näätänen 1981). In states of high motor preparedness, the modulation of an overt motor response is much easier than in states of poor response preparedness as less additional motor excitation is necessary for response initiation. Hence, less muscle activity is required. From this perspective, the smaller EMG-R amplitude observed under pergolide compared to placebo may reflect a higher level of motor preparedness as a consequence of more efficient premotor stimulus processing.

The significantly increased number of wrong-hand responses observed under pergolide suggests a deteriorating DA agonistic effect on the mapping between sensory stimuli and motor responses. Unfortunately, our LRP analyses do not allow to decide whether this stage of response selection represents an aspect of premotor rather than motor processing. Recent research also provides ambiguous findings on this issue; stimulus–response mapping and response selection have been related to the prefrontal dopaminergic network (e.g., Deco and Rolls 2003) as well as to basal ganglia–thalamocortical mechanisms (e.g., Seiss and Praamstra 2004).

In conclusion, analysis of S-LRP latencies revealed enhanced speed of stimulus-related premotor information processing after a low oral dose of the D1/D2 receptor agonist pergolide. This finding provides clear evidence that dopaminergic neurotransmission effectively modulates early perceptual and cognitive stages of information processing as suggested by neural network models of the functional role of prefrontal DA. The concurrent lack of an effect on response speed suggests higher procedural sensitivity of S-LRP-based compared to behavioral measures for elucidating dopaminergic effects on stimulus processing. There was no evidence for central or peripheral effects on motor processing speed as assessed by LRP-R and EMG-R latencies as well as measures of response dynamics. The absence of dopaminergic effects on aspects of motor processing may be due to a higher capacity of the nigrostriatal compared to the mesocortical DA system to compensate pharmacologically induced changes in DA activity.


This study was supported by the German–American Academic Council Foundation (Grant TransCoop-Program 2000).

Copyright information

© Springer-Verlag 2006