Signal-, set-, and movement-related activity in the human premotor cortex

Abstract

Single unit recording studies in non-human premotor cortex have revealed neurons with motor-related activity. Other neurons, however, seem to be involved in prior movement selection and preparation processes, and have activity related to visual instruction signals or movement preparation (set). We have used single pulse transcranial magnetic stimulation (TMS) to identify similar processes in human subjects. In Experiment 1 subjects performed a cued movement task while being stimulated with TMS over three sites: sensorimotor cortex, posterior premotor cortex and anterior premotor cortex. TMS slowed movements when applied at 140 ms after the visual cue over the anterior premotor site, at 180 ms after the visual cue over the posterior premotor site, and at 220 ms and later after the visual cue over the sensorimotor cortex. The results are consistent with a change from signal to movement-related processing when moving from premotor to motor cortex. In Experiment 2 there was a preparatory set period between the instruction signal that informed subjects which movement to make and the go signal that informed them when to actually make the movement. TMS was applied over the anterior premotor site and the sensorimotor site during the set period. At both sites TMS had similar effects on slowing subsequent movements. The results suggest set activity in both premotor and motor cortices in human subjects.

Introduction

In the macaque, anatomical labelling experiments have demonstrated that both the primary motor cortex, M1, and the more anterior dorsal premotor cortex, PMd, project to the spinal cord 10, 18. This has been interpreted [10]as evidence that the various premotor areas operate at the same hierarchical level as the primary motor cortex. By contrast, the results of single unit recording and lesion or inactivation based investigations, have emphasized a distinction in the hierarchical level at which PMd and M1 operate. There is a general consensus that the PMd is more concerned with some aspect of the planning, preparation, selection or visual guidance of responses while M1 is more concerned with the implementation of the response. Removal or temporary inactivation of the premotor cortex or just its more dorsal part, PMd, impairs selection between movements 17, 31, 40, 41, 42while an M1 lesion disrupts the implementation of appropriate muscle activation patterns during movement execution [20]. Single unit recordings have been made from monkey PMd and M1 in a variety of paradigms involving visually cued movements. A greater proportion of PMd cells have activity related to the visual cue (signal-related) while a greater proportion of the M1 cells have movement related responses 7, 8, 16, 23, 25, 26, 29, 30, 36, 46, 47, 64. There is a transition from signal- to movement-related activity as the recording electrode is moved in an anterior to posterior direction 22, 62, 64. We have previously used single pulse transcranial magnetic stimulation (TMS) to investigate the human PMd in the region of the superior precentral sulcus 51, 54. We have shown that disruption of PMd particularly interfered with a choice reaction time (RT) task which required selection between two possible finger movement responses. TMS pulses over M1 were most disruptive when they were delivered around the time of response execution. Pulses over PMd, however, are also disruptive when they are delivered shortly after the instruction cue. In the first experiment we attempted to characterize such signal- and response-related TMS effects by examining the effects of TMS delivered at a number of scalp sites and at a variety of different intervals after visual signal onset. The effects of TMS were assessed by comparing RT on trials with and without TMS. Single unit recording studies in the monkey have also characterized the difference between premotor and motor cortex in terms of preparatory set activity. In such experiments, monkeys are first presented with a instruction signal (IS) that tells them which movement to make, and then, only after a short delay, they are presented with a go signal (GS) that tells them that the instructed movement is now to be executed. Preparatory set activity can be recorded in the interval between the presentation of the IS and the GS. A greater proportion of PMd than M1 cells have set-related activity 23, 64. The second experiment attempted to use TMS to identify set-related activity. TMS pulses were applied during the intervals between the presentation of separate ISs and GSs. It was expected that the disruption of preparatory set activity would slow subsequent RTs. In this experiment it was not possible to compare the effects of TMS with non-stimulation trials. TMS may be considered to have two components. First it has a direct effect on cortical processing 3, 11, 21. Since the pulse interferes with normal cortical processing it is expected to slow RT 51, 54when it is delivered over a brain area involved in task performance. Second, TMS is also likely to have a general effect by virtue of its being a very salient tactile and auditory stimulus, which, conversely, will tend to speed simple RT 19, 35, 40, 41. The influence of the second, general facillitatory effect is minimized in the choice RT task used in Experiment 1. The subjects in Experiment 1 must attend and respond to the visual IS since this is the only stimulus that informs them about which movement to make. Any subject responding to the sound and feel of the TMS pulse would have just a 50% chance of making the correct response. In Experiment 2 however, it is likely that subjects will already know which response to make at the time the TMS pulse is delivered in the interval between the IS and the GS. If TMS over motor or premotor cortex has any slowing effect it is likely to be superimposed on a general facilitatory effect afforded by TMS even at non-motor sites. In Experiment 2 we therefore compared TMS over premotor and motor cortices with TMS over a control site outside the motor system.0452. Methods2.1. Experiment 1: signal- and movement-related activity2.1.1. SubjectsSeven subjects participated in the first experiment: five right handed males, one left handed male, one left handed female [37]. Ethics permission for this study was obtained from the Central Oxford Research Ethics Committee (COREC, 94.261), and consent was obtained from all subjects.

In order to localize M1, muscle potentials were recorded via two electrodes taped over the flexor digitorum profundus (FDP). These were connected to a Medelec MS20 EMG machine. An initial estimate of the scalp position over the hand representation in motor cortex was marked 4 cm lateral and 2 cm anterior to Cz (international 10/20 EEG system). To locate the position of the fingers representation in motor cortex, we searched to find a point of stimulation that resulted in the maximum Motor Evoked Potential (MEP) with the minimum stimulation strength. A figure-eight stimulation coil (each wing 70 mm in diameter), connected to a Magstim 200 with maximum output of 2.0 Tesla, was placed over the left motor cortex tangential to the skull, with the coil current flowing in an anterior to posterior direction parallel with the midline. The intensity of stimulation was increased from 30% of maximum output in 5% steps until an MEP was just visible. The coil was then moved in 0.5 cm steps medial, lateral, posterior and anterior until the point of the maximum MEP was ascertained. Stimulation intensity was then decreased to the lowest setting at which MEPs could still be induced with all pulses. The coil was then moved in lateral, medial, posterior and anterior steps of 0.5 cm to check that adjacent sites did not more reliably elicit MEPs on three trials. If no better site was found, then the provisional site was taken to be the optimal position of stimulation of the finger representation in motor cortex—the hot-spot. If a better site was found then the procedure was repeated iteratively until the hot-spot was identified. We then used a plastic grid to mark a series of reference positions on the scalp in relation to the hot-spot. Three positions were marked 1, 2 and 3 cm anterior to the hot-spot and at the same laterality; a further three positions were marked 1 cm medial to these. We have previously shown that single pulse TMS within a 0.5 cm radius of the point 2 cm anterior and 1 cm medial to the hot-spot disrupts movement selection in choice RT tasks 51, 55. We have also shown that this site is situated above the dorsal precentral sulcus which concurs with the location of PMd as established by positron emission tomography (PET) in single subject studies [13]. This will be referred to as the anterior (ANT) site. Stimulation was also used at two other sites. A sensorimotor site (SM) situated 1 cm posterior to the motor hot-spot was used. It is known from co-registration of TMS with and PET scans that the motor hot-spot is situated on the anterior lip of the central sulcus just anterior to the position of maximum rCBF change during hand movement [61]. The SM site chosen in this study should therefore be over the central sulcus. Stimulation over this site still induced MEPs. A third, intermediate (INT) site was also used. This was situated 1–1.5 cm behind the ANT site. The locations of ANT, INT and SM sites in all seven subjects have been plotted with respect to the motor hot-spot (marked by X) in Fig. 1. There was little variability in the position of the SM site (triangles in Fig. 1). The ANT site was usually about 2 cm anterior and 1 cm dorsomedial to the motor hot-spot (○). In two subjects, however, the site was more anterior. The position of the INT site showed the most variability (•). In general, however, the INT sites were situated between 0.5–1.5 cm anterior and 0–1 cm dorsomedial of the motor hot-spot. The subject with the most anterior ANT site also had the most anterior INT site. This subjects INT site was 2 cm anterior to the motor hot-spot. Fig. 2 shows a T1 weighted MRI scan (spin echo, 2 Tesla, Siemens Vision) of the brain of one representative subject on whose scalp the SM, INT and ANT sites had been marked by capsules containing garlic and soya oil. A line, representing the orientation of the stimulating coil, was drawn tangential to the surface of the skull on each section shown in Fig. 2. A line was then drawn at 90° to indicate the centre of the area where the induced field was at its maximum. The SM site is above the central sulcus region. The ANT site is over the superior precentral sulcus. The INT site is on the precentral gyrus adjacent to the posterior ramus of the superior precentral sulcus.

During all the experiments, the subjects sat facing a PC (486) computer with their chin on an adjustable chin rest approximately 50 cm from the screen and their right hands by a key pad. The subjects performed a choice RT task in which they were required to respond with either the middle or the index finger according to the shape that was presented at the centre of the screen. The shapes were presented for 100 ms and responses were timed from the moment of cue onset. If a small circle or a large rectangle were presented on the screen, the subjects responded with their index finger. If a large circle or small rectangle were presented, the subjects responded with their middle finger. Neither shape nor size alone determined the response required; this design was intended to prevent the process of response selection from becoming automated. The inter-trial interval was 4 s. This ensured enough time for the Magstim 200 capacitors to recharge between the delivery of pulses. During the experiment, the subjects were magnetically stimulated over the left hemisphere, contralateral to the right hand used to respond. The subjects performed the task in blocks of 14 trials. The subjects were stimulated for eight blocks at each site (ANT, INT, SM). The order of sites that were tested was varied across subjects. In the case of ANT and INT stimulation was first applied at the locations described above. If TMS pulses had no effect on RT then the coil was moved to an adjacent position within a radius of 0.5 cm of the original position. The subjects were stimulated on half of the trials. In each of the eight blocks of 14 trials, the subjects were stimulated once at seven different intervals. The onset of the TMS pulses were measured with respect to the onset of the instruction cues on the monitor screen. The intervals used were 60, 100, 140, 180, 220, 260 and 300 ms. The stimulation and non-stimulation trials were presented in a random order. The subjects were first adapted to stimulation by slowly increasing the output of the stimulator from 50% in 5% steps. These trials were not included in the analysis. During the experiment, if the subject made less than seven correct responses on stimulation trials for any TMS interval, a further two blocks were given. No more than 400 pulses were given to any subject on one day, this was required by the local ethics committee. All subjects completed the experiment within one session that lasted approximately 2–4 h. The task is summarized in Fig. 3a. All stimulation was at 60–80% of stimulator output. An effective level of stimulation could have been defined for each subject individually by reference to their FDP hot-spot threshold. However, we decided in this initial study of TMS over premotor cortex that the use of a high level of stimulation would avoid the possibility of false negative results. The subjects were initially given two sets of 28 practice trials without stimulation pulses. The subjects were first told which responses were correct, and during the first few practice trials they were given verbal feedback so as to ensure that they responded correctly. The subjects were encouraged throughout to be as quick and as accurate as possible. After two sets of 28 trials the response times and errors were briefly inspected, and the subjects were given a further 28 trials if their reaction times were still highly variable.

The RTs were recorded and stored on disk for later analysis. Subjects made only a few incorrect responses and these were not included in the analysis. The results were analysed using SPSS for Windows. The data were normalized: for each individual, we calculated the percentage change of the median response times with stimulation compared with the median response times with no stimulation. The percentage change is an indication of the effect of stimulation compared with the no stimulation baseline. ANOVAs were used to analyse the RT changes seen in the TMS trials. These were followed up with t-tests (Bonferroni corrected) on the response percentage changes with a null hypothesis that there was no percentage change.