Patterned ballistic movements triggered by a ... - Wiley Online Library

8836

Journal of Physiology (1999), 516.3, pp. 931—938

931

Patterned ballistic movements triggered by a startle in healthy humans Josep Valls-Sole, John C. Rothwell*, Fatima Goulart, Giovanni Cossu and Esteban Mu˜noz Unitat d’EMG, Servei de Neurologia, Departament de Medicina, Hospital Clinic, Universitat de Barcelona, Barcelona, Spain and *MRC Human Movement and Balance Unit, Institute of Neurology, Queen Square, London, UK (Received 9 October 1998; accepted after revision 28 January 1999)

1. The reaction time to a visual stimulus shortens significantly when an unexpected acoustic startle is delivered together with the ‘go’ signal in healthy human subjects. In this paper we have investigated the physiological mechanisms underlying this effect. If the commands for the startle and the voluntary reaction were superimposed at some level in the CNS, then we would expect to see alterations in the configuration of the voluntary response. Conversely, if the circuit activated by the startling stimulus is somehow involved in the execution of voluntary movements, then reaction time would be sped up but the configuration of the motor programme would be preserved. 2. Fourteen healthy male and female volunteers were instructed to react as fast as possible to a visual ‘go’ signal by flexing or extending their wrist, or rising onto tiptoe from a standing position. These movements generated consistent and characteristic patterns of EMG activation. In random trials, the ‘go’ signal was accompanied by a very loud acoustic stimulus. This stimulus was sufficient to produce a startle reflex when given unexpectedly on its own. 3. The startling stimulus almost halved the latency of the voluntary response but did not change the configuration of the EMG pattern in either the arm or the leg. In some subjects the reaction times were shorter than the calculated minimum time for processing of sensory information at the cerebral cortex. Most subjects reported that the very rapid responses were produced by something other than their own will. 4. We conclude that the very short reaction times were not produced by an early startle reflex adding on to a later voluntary response. This would have changed the form of the EMG pattern associated with the voluntary response. Instead, we suggest that such rapid reactions were triggered entirely by activity at subcortical levels, probably involving the startle circuit. 5. The implication is that instructions for voluntary movement can in some circumstances be stored and released from subcortical structures. It is generally accepted that when subjects move voluntarily in response to a reaction signal, the cerebral cortex plays a major role in identifying the sensory stimulus and releasing the instructions to move. Indeed, theoretical models have been put forward in psychophysiological studies to symbolize the steps of such cerebral processing (Gratton et al. 1988). In contrast, startle reactions occur via a subcortical reflex mechanism. Sensory inputs activate the reticular formation and the descending reticulo-spinal tract to the spinal cord (Davis et al. 1982). Because of the differences in the length of the circuits, as well as in the amount of sensory processing, the latencies of the startle reaction are much shorter than those of a voluntary reaction. In muscles of the

forearm, a startle reaction occurs at less than 80 ms. In contrast, the voluntary reaction time to a visual ‘go’ signal is of the order of 150 ms. The reaction time to auditory and somatosensory stimuli is shorter but even then rarely less than 100 ms (Brown et al. 1991a; Thompson et al. 1992; Pascual-Leone et al. 1992). Recently, Valls-Sole et al. (1995) have shown that reaction times can be considerably reduced if a very loud, startling, sound is given at the same time as the visual ‘go’ signal. The amount of shortening is much greater than that observed in conventional intersensory facilitation (Nickerson, 1973), and presumably represents a specific startle-related effect. The question is what neural mechanisms are responsible for this

932

J. Valls-Sole , J. C. Rothwell, F. Goulart, G. Cossu and E. Mun˜ oz

phenomenon. Since the reaction times are the same as the startle reaction itself, the simplest explanation is that the activity observed in these cases consists of two components: an early startle reflex, and a late voluntary response, the true onset of which would be masked by the startle. An alternative explanation is that the high intensity acoustic stimulus somehow releases the movement being prepared for voluntary execution at a speed far quicker than usual, and probably using the same pathways as the startle reaction itself. The present experiments were designed to test these hypotheses for startle-induced reaction time shortening. To do this, we examined the effects of the startling stimulus on two stereotyped EMG patterns of activity thought to be generated in the central nervous system. One is the triphasic agonist—antagonist—agonist burst pattern of a rapid, selfterminated wrist flexion or extension movement (Hallett et al. 1975; Hallett & Marsden, 1979; Palmer et al. 1994). The other is the structured EMG pattern that accompanies the action of rising on tiptoes from the standing position (Nardone & Schieppati, 1988; Crenna & Frigo, 1991). Both patterns are centrally programmed and can be generated without significant contribution from peripheral afferent input (Sanes & Jennings, 1984; Forget & Lamarre, 1987). We have hypothesized that simple addition or collision between the startle reaction and the voluntary movement would be evident as a disruption of the stereotyped EMG pattern for both movements. Conversely, if the startling stimulus somehow bypasses some of the normal reaction time circuitry, then the response would be sped up with no alteration of the EMG pattern. Subjects

METHODS

The experiments were carried out in 14 healthy subjects. Eleven were paid pre-graduate students, eight men and three women, aged 19 to 26 years. The remaining subjects were three of the authors, aged 30 to 48 years. All signed a consent form after being informed of the nature of the experiment, which was approved by our Institutional Research Panel.

Recording systems

We recorded the electromyographic activity of the muscles under study with an electromyograph, MYSTRO5Plus (Vickers Medical Ltd, London, UK). The same electromyograph was used to capture the signal from a movement transducer placed on the moving segment. The bandpass frequency filter was 50—1000 Hz for the EMG recordings, and 1—500 Hz for the movement transducer. All EMG and movement signal recordings were printed on electromyographic paper and analysed off-line.

Experimental setting for the reaction time task

The subjects were required to react as fast as possible to a visual ‘go’ signal. The signal was a white 5 cmÂ mark appearing on a blank computer screen 5 s after the presentation of a short forewarning sentence and a verbal warning. The delivery of the ‘go’ signal triggered the electromyograph oscilloscopic sweep and, eventually, the stimulators used for the study (see below).

J. Physiol. 516.3

Before starting the experiment, the subjects had sufficient training to perform the task accurately with only minimum variations between trials.

Stimuli

The subjects were warned that in some of the reaction time trials, there could be an external stimulus linked to the ‘go’ signal (the test stimulus), but they were also encouraged to react to the ‘go’ signal regardless of other interfering stimuli. A loud sound, capable of producing a startle reaction, was obtained by discharging the magnetic coil of a magnetic stimulator over a metallic platform. The sound produced had an intensity of 130 dB SPL, measured at a distance of 1 m from the source with a Bruel and Kjaer Impulse Precision Sound Level Meter type 2204. In each individual, we obtained a minimum of 10 trials containing only the ‘go’ signal (control), and five trials containing the ‘go’ signal and the startling stimulus (test). In five subjects we also delivered single unexpected acoustic stimuli at the end of a sequence of trials when the subjects were expecting no further stimuli.

Experiments involving wrist flexion—extension movements

The subjects were sitting on a comfortable chair beside a table, with their forearm and hand enclosed in a metallic device made of two parts. The part containing the subject’s forearm was fixed to the table surface, while the part holding the subject’s hand could freely move within a range of −90 to +90 deg with respect to the fixed part. The two parts were linked by means of a hinge containing a built-in movement transducer, over which the wrist was carefully positioned. The subject’s forearm and hand were fully supported halfway between pronation and supination with the help of two movable metallic walls and pieces of foam, in such a way that only movements of the wrist joint were allowed. Surface EMG electrodes were placed over the wrist extensors, wrist flexors, biceps brachii (BIC) and triceps brachii (TRIC) muscles and also over the orbicularis oculi (OOc) and sternocleidomastoid (SCM) muscles. Subjects were instructed to react by making a brisk self-terminated wrist movement (of flexion in the first set of trials and of extension later).

Experiments involving rising onto tiptoe

Subjects stood barefoot facing the computer screen and were asked to perform a fast rise on tiptoes after the ‘go’ signal. Recording electrodes were placed over the tibialis anterior (TA) and triceps surae (TRS) muscles and also over the OOc and SCM muscles. A bar-shaped piezoelectric accelerometer was placed over the dorsum of the foot to record the movement. Some subjects needed to lean forward slightly in order to increase activity in the TRS muscles so that pre-movement silence could be measured clearly.

Analysis of the results

In the wrist flexion—extension task we measured the onset latency and total duration of the first EMG burst in the agonist muscle, the time between onset of the first agonist and antagonist bursts, the duration of the antagonist burst, and the time between the first and second agonist bursts. We also measured the onset latency of the first burst of EMG activity and the time between the bursts in biceps or triceps muscles. In the rising onto tiptoe task, we measured the onset latency and duration of the premovement silent period (pmsp) in TRS, the onset latency of TRS EMG activity, and the onset latency and duration of the burst of EMG activity observed in TA. In both experiments, onset of movement (handMOV or footMOV) was taken from the movement transducer (arm) or accelerometer (leg) records. The hand movements were calibrated in degrees of wrist rotation; the accelerometer data from the foot movement was

J. Physiol. 516.3

933

Startle-induced release of motor programmes

left in arbitary units because the acceleration depended on the precise position of the transducer on the dorsum of the foot. However, since this was constant between trials, arbitrary units were sufficient to detect within-subject differences of amplitude. Additionally, we measured the onset latency of the activity recorded in the OOc and SCM muscles in trials containing test stimuli. For each subject, we calculated the mean and standard deviation of handMOV or footMOV for control and test trials. To assess the effects induced by the startling stimulus on reaction time, we performed a one-factor analysis of variance (ANOVA) for handMOV and footMOV between control and test trials. Such statistical comparison was first made for each individual and also, because the same tendency was observed in all subjects, for the grand mean obtained after grouping all individual values in each type of trial. To assess whether the presence of the stimulus induced changes in the configuration of the pattern of movement, we normalized the data by expressing the values for handMOV or footMOV as a percentage of the mean value in each subject. Statistical comparisons between control trials and trials containing test stimuli were made for each variable, except for the normalized handMOV or footMOV, on the grouped data from all individuals, again using ANOVA.

In subjects receiving unexpected acoustic startling stimuli, we measured the onset latency of the EMG bursts recorded in forearm muscles and in OOc and SCM, and compared them with the onset latency of the bursts recorded in the same muscles in test trials.

RESULTS

Wrist flexion—extension movements

Wrist movements were accompanied in all subjects by a triphasic pattern of EMG activity in the forearm muscles. This consisted of an initial burst of activity in the agonist muscle, followed by a burst in the antagonist muscle, and a second burst in the agonist. Flexion or extension movements gave rise to similar patterns and, because no apparent differences were observed in the pattern or the onset latency of handMOV, we grouped together both movements for statistical analysis (control trials). Even though the forearm was well supported in the armrest, wrist movements were also accompanied by bursts of postural EMG activity in BIC and TRIC. The pattern of this activity was very similar to

Figure 1. Raw data example and mean data from all subjects in the arm movement trials

A, EMG recordings from a single subject performing a ballistic wrist flexion movement. Upper panel: control

trials in which the subject responded only to the visual cue. Lower panel: test trials in which the visual cue was accompanied by a loud startling auditory stimulus. MOV, signal from the movement transducer; BIC, biceps brachii; TRIC, triceps brachii; WF, wrist flexors; WE, wrist extensors. B, schematic representation of the mean EMG pattern of all subjects in control and test trials. The leftward extent of the bars represents the mean onset latency, whilst the horizontal line shows +1 s.d. The length of the bars represents the duration of the EMG bursts, and the horizontal lines at the right side of the bar show +1 s.d. Duration was only measured for the first agonist and the antagonist bursts. HandMOV, displacement of the wrist joint; note how the pattern of the EMG bursts (i.e. the interburst interval and the burst durations) is the same in control and test trials even though the onset latency is substantially reduced in the test trials.

934


J. Physiol. 516.3

––––––––––––––––––––––––––––––––––––––––––––––

Table 1. Onset latency and amplitude of the movement of the hand, and latencies and durations of the EMG bursts in the wrist movement task

–––––––––––––––––––––––––––––––––––––––––––––– Measurement Control Test –––––––––––––––––––––––––––––––––––––––––––––– Onset latency of handMOV 203·6 ± 74·5 104·3 ± 17·6 * Amplitude of handMOV 42·5 ± 19·5 48·0 ± 15·0 Onset of Ag1 EMG burst 171·4 ± 50·9 77·3 ± 10·7 * Duration of Ag1 EMG burst 76·6 ± 15·6 72·9 ± 11·2 Interval between onsets of Ag1 and Ant 36·2 ± 11·9 35·6 ± 10·4 Duration of Ant EMG burst 121·5 ± 47·0 109·3 ± 21·3 Interval between onsets of Ag1 and Ag2 143·2 ± 34·6 138·4 ± 24·8 Onset of postural EMG in biceps and triceps 181·9 ± 47·7 81·8 ± 13·2 * Interval between onsets of biceps and triceps 32·1 ± 14·4 29·3 ± 10·1 postural EMGs –––––––––––––––––––––––––––––––––––––––––––––– Values are the mean ± s.d. of the measurements listed in the left column. HandMOV, hand movement; Ag1, first burst of activity in the agonist muscle; Ag2, second burst of agonist activity; Ant, burst of activity in the antagonist muscle. All data are given in milliseconds, except for amplitude of handMOV, which is given in degrees of wrist rotation. * P < 0·05 comparing test and control trials. ––––––––––––––––––––––––––––––––––––––––––––––

Table 2. Onset latency and amplitude of the movement of the foot, and latencies and durations of the EMG bursts in the leg movement task

–––––––––––––––––––––––––––––––––––––––––––––– Measurement Control Test –––––––––––––––––––––––––––––––––––––––––––––– Onset latency of footMOV 237·6 ± 98·0 122·6 ± 25·3 * Magnitude of footMOV 209·2 ± 92·8 244·1 ± 102·5 Onset of pmsp in TRS 130·2 ± 46·8 68·5 ± 8·4 * Duration of pmsp in TRS 114·6 ± 57·1 60·9 ± 17·0 * Interval from onset of pmsp to onset of the TA burst 54·2 ± 22·2 46·2 ± 11·1 Duration of TA burst 81·8 ± 32·7 77·9 ± 12·4 Interval between EMG onset in TA and EMG onset in TRS 62·0 ± 18·1 53·3 ± 16·0 –––––––––––––––––––––––––––––––––––––––––––––– Values are the mean ± s.d. of the measurements listed in the left column. FootMOV, foot movement; TRS, triceps surae; TA, tibialis anterior; pmsp, premovement silent period. All data are given in milliseconds, except for amplitude of foot movement, which is given in arbitrary units. * P < 0·05 comparing test and control trials. ––––––––––––––––––––––––––––––––––––––––––––––

the triphasic pattern in the prime mover muscles except for the fact that the second agonist burst was often missing. The main effect of a startling stimulus on the wrist movements is shown in an example from one subject in Fig. 1A. It should be noted that the form, amplitude and duration of the EMG bursts, and the amplitude of the movement signal, are very similar in control and test trials. There was no sign that the startle had disrupted the pattern of the voluntary response of the control trial. This is shown schematically for the grand mean data of all subjects in Fig. 1B. Table 1 shows the values for all measured parameters and their statistical comparisons. The only significant effect of the startle was on response latency, while all measurements made on the parameters related to intrinsic characteristics of the triphasic EMG pattern or those concerning the temporal relationship between the triphasic pattern and the postural EMG activity were unchanged. In individual subjects, the onset latency of handMOV in test trials was 36—62% of that of the control trials.

In test trials, all subjects had bursts of EMG activity in the OOc and SCM muscles that were not present in control trials (Fig. 2A and B). Mean onset latencies of these EMG bursts were 46·2 ± 8·5 ms for the OOc, and 58·7 ± 11·1 ms for the SCM. These latencies were not different from those obtained in the same muscles when an unexpected acoustic startling stimulus was delivered with no previous instruction to react (Fig. 2C). The startle reaction induced bursts of activity in all muscles recorded from, at latencies ranging from 42·7 to 63·1 ms for the OOc, 49·7 to 74·0 ms for the SCM, 56·3 to 102·4 ms for wrist flexors, and 64·7 to 98·2 ms in the wrist extensor muscles.

Rising onto tiptoe

All subjects produced a well-defined EMG pattern of activity in control trials. This consisted of an initial silence in the ongoing activity of the TRS (TRS pmsp) and a burst of EMG activity in the TA, which partly coincided with the TRS pmsp and preceded the main burst of agonist activity in the TRS.

J. Physiol. 516.3


Figure 3A shows an example of the effect of a startling stimulus on the response in a representative subject. As with wrist movements, the main effect was to shorten the reaction time with no effect on the form of the response itself. Again, there was no significant change in the amplitude of the movement, nor in the duration or amplitude of the EMG bursts, between control and test trials. Figure 3B is a schematic representation of the grand mean group data. Table 2 shows the onset latency of footMOV and all the EMG parameters measured. In individual subjects, the onset latency of footMOV in test trials was 29—59% of the same values in control trials, while all intrinsic characteristics of the EMG pattern were generally unchanged. There were bursts of EMG activity in OOc and SCM muscles in all test trials but not in control trials. Mean onset latencies were 47·3 ± 10·6 ms for OOc and 62·8 ± 12·7 ms for SCM. These values were not different from those obtained in the same muscles when an unexpected acoustic startling stimulus was delivered with no previous instruction to react.

935

time by an amount of 20—50 ms (Nickerson, 1973) whereas the shortening seen in our subjects was generally more than 70 ms. Therefore, we believe that although intersensory facilitation might play some role, other factors must also be involved to account for our results. A clue to the mechanisms operating in our experiments comes from calculating the minimum time taken for sensory input to reach the motor cortex and release a command to move.

Other observations

All subjects were questioned about their perception of the movement. A general impression was that the movements made in test trials were less under the subjects’ own voluntary control than those of the control trials. No subject said that the loud sound helped them to move or accelerated the intended movement, and no-one was aware that the movement was faster in test than in control trials. Eight subjects specifically stated that, in trials with test stimuli, something other than their own will was making them move, and 11 felt that their own control of the movement was disturbed or even halted.

DISCUSSION

The main finding of the present study was that a loud unexpected sound sped up the execution of a voluntary movement without changing its main characteristics. Using voluntary responses that had a recognizable EMG pattern, we could be quite sure that the shortening of the reaction time was due to early release of the movement pattern and not to addition of an early startle reflex onto a later voluntary response. Addition of automatic and reflex responses has been described previously in the combination of a stretch reflex and a reaction time task (Day et al. 1983). However, in the experiments reported here, the startling stimulus did not produce any extra EMG activity in the responding muscles, even though evidence that a startle reaction had occurred was apparent in muscles of the face and neck. How could voluntary responses be released so much earlier than the usual reaction time? As we have argued before, it is unlikely that the reaction time shortening found in our subjects could be entirely due to the phenomenon of intersensory facilitation between the visual ‘go’ signal and the acoustic stimulus. This has been found to shorten reaction

Figure 2. Comparison of responses in prime moving muscles (wrist extensors and flexors), face and neck muscles

A, rapid wrist flexion movement in response to a visual stimulus on its own (control trial). B, rapid wrist flexion movement during a startling trial (test trial). C, response in

the same muscles when the loud startling stimulus was unexpectedly given on its own. Note the similar latency of muscle activity in B and C. OOc, orbicularis oculi muscle; SCM, sternocleidomastoid muscle; WE, wrist extensors; WF, wrist flexors; MOV, signal from the movement transducer.

936


Acoustic stimuli induce activity in many structures of the central nervous system. At the cochlear nuclei, the acoustic signal is converted into inputs to multiple channels. On their way to the auditory cortex of the temporal lobe, auditory inputs gave rise to responses in several structures of the brainstem. Evoked potential measurements suggested that they arrived at the nuclei of the lateral lemniscus with a latency of 6—7 ms, but no consistent information was available on the time taken to activate higher centres. The middle latency auditory potentials, thought to reflect the arrival of the first volley to the auditory cortex, occurred at a latency of about 35 ms (Erwin & Buchwald, 1986). With an efferent conduction time of about 20 ms to the forearm muscles, the time left for cortico-cortical activation of the motor cortex and generation of the first agonist burst would be only some 10 ms for the individual with the shortest Ag1 latency in our study (65 ms). In the leg, the shortest value observed for the TRS pmsp was 60 ms which, together with the fact that the normal efferent conduction time to the TRS was about 30 ms, left almost no room for any cortico-cortical transfer of activity. In addition, due to our experimental

J. Physiol. 516.3

setting, sound had to travel a distance of about 2 m from the place where it was generated to the subjects’ ears. This would take about 6 ms, which should be subtracted when calculating the effect of sound on reaction time. We must conclude that, unless cortical activation occurred via an unknown and very rapid input from auditory afferents, it is unlikely that the motor cortex could have contributed to the onset of the fastest reactions that we observed. We should like to propose an alternative mechanism to explain the rapid execution of a voluntary movement when a loud noise is delivered together with the reaction signal. This relies on the fact that an unexpected loud noise elicits a startle reflex that has a characteristic EMG pattern and latency (Brown et al. 1991a). Evidence from animal experiments (Davis et al. 1982) and from human pathophysiology (Brown et al. 1991a) suggests very strongly that this reflex is mediated through a centre in the reticular formation, and that impulses are conducted to the spinal cord via reticulospinal systems. Our subjects exhibited the EMG pattern characteristic of the startle reaction when the loud

Figure 3. Raw data example and mean data from all subjects when performing the rise onto tiptoe task A, recordings from a single subject performing a sudden rise onto tiptoes in control (upper panel) and test

trials (lower panel). OOc, orbicularis oculi; SCM, sternocleidomastoid; SOL, soleus; TA, tibialis anterior; Acc, accelerometric recording from the dorsum of the foot. B, schematic representation of the mean pattern of EMG activity from all subjects observed in the leg movement task. The leftward extent of the bars represents the mean onset latency of the events, whilst the horizontal line shows +1 s.d. The length of the bars represents the duration of the events, and the horizontal lines at the right side of the bar show +1 s.d. pmsp, pre-movement silent period. Duration was only measured for the TRS pmsp and the TA burst. Note the preservation of the burst durations and interburst latencies despite the shortening of movement onset.

J. Physiol. 516.3


noise stimulus was applied on its own, unexpectedly. Also, a fraction of the same EMG pattern was present in test but not in control trials as bursts of activity in OOc and SCM muscles. The implication is that (part of) a startle had been elicited in the test trials. However, even though the activity in the forearm muscles occurred at the same latency as that seen in a startle reflex, the form of the response was the same as in the control trials. We suggest that the voluntary reaction has been driven at the speed of a startle reaction while maintaining the features of the motor programme. As outlined in the Introduction, we had expected that if voluntary (corticospinal) and startle (reticulospinal) responses acted through separate channels, then their outputs would superimpose at the level of the spinal motor apparatus. However, since this did not occur, we assume that there must have been interaction between the startle and voluntary systems at a supraspinal level, which resulted in a single sped-up version of the normal voluntary response. Our proposal is that such interaction occurs in the reticular formation, where the startle responses originate. The implication of this hypothesis is that, under certain conditions, the motor programme that is being prepared for voluntary execution can be triggered by activation of the same reticular structures that are responsible for the startle reflex. For this to happen, the commands defining the movement (i.e. the interval between the EMG bursts, their duration and amplitude) must be accessible from the startle circuit and ready to be released at a much shorter latency than required for activation of cerebral pathways. Evidence that this is possible comes from many sources. Peparatory, pre-movement, activity occurs in parallel at many levels of the nervous system from cortex to spinal cord. For example, the spinal monosynaptic reflex changes before movement (Schieppati et al. 1986) as well as in the preparatory period before the reaction signal is given (Brunia, 1993). Motorevoked potentials to transcranial magnetic stimulation also change both in the reaction and preparatory periods before movement (Starr et al. 1988). Indeed, even the startle pattern itself can be modulated by the stance of the subject before the stimulus is given (Brown et al. 1991b). Our suggestion is that sufficient detail of the movement characteristics may be stored in brainstem and spinal centres so that, on occasion, the whole motor programme can be triggered without the expected command from the cerebral cortex. In the current theoretical model of the physiological processes taking place in a reaction time paradigm, movement performance requires time for recognition of the stimulus, transfer of the activity to the response channel and execution of the task. The response channel is conceived to include all the necessary structures to execute the response as rapidly as possible (Gratton et al. 1988; Pascual-Leone et al. 1992). From the findings reported here, we suggest that the reticulospinal system can be an important part of the response channel for ballistic reaction time tasks. Under certain conditions, its activation can trigger the neural commands required for correct execution of the prepared motor programme by-

937

passing the motor cortex and, therefore, escaping the normal process of stimulus recognition. This is not to deny the cerebral cortex a crucial role in preparing the response parameters, nor to exclude it from contributing to later parts of the response. However, it does not need to play the lead role for initiating voluntarily prepared responses. This effect could have interesting repercussions not only on areas of scientific knowledge, but also for those concerned with human physical performance and athleticism. We noticed that the International Amateur Athletics Federation regulations governing athletic running events state that reactions faster than 100 ms must be regarded as anticipation of the starting signal, and therefore classed as a false start (International Amateur Athletic Federation, 1998). Several of our subjects started leg movement within 100 ms of the startling noise, suggesting that this limit may need revision.

Brown, P., Rothwell, J. C., Thompson, P. D., Britton, T. C., Day, B. L. & Marsden, C. D. (1991a). New observations on the normal

auditory startle reflex in man. Brain 114, 1891—1902.

Brown, P., Day, B. L., Rothwell, J. C., Thompson, P. D. & Marsden, C. D. (1991b). The effect of posture on the normal and

pathological auditory startle reflex. Journal of Neurology, Neurosurgery and Psychiatry 54, 892—897. Brunia, C. H. M. (1993). Waiting in readiness: gating in attention and motor preparation. Psychophysiology 30, 327—339. Crenna, P. & Frigo, C. (1991). A motor programme for the initiation of forward oriented movements in humans. Journal of Physiology 437, 635—653. Davis, M., Gendelman, D. S., Tischler, M. D. & Gendelman, P. M. (1982). A primary acoustic startle circuit: lesion and stimulation

studies. Journal of Neuroscience 2, 791—805.

Day, B. L., Rothwell, J. C. & Marsden, C. D. (1983).

Interaction between the long latency stretch reflex and voluntary electromyographic activity prior to a rapid voluntary motor reaction. Brain Research 270, 55—62. Erwin, R. J. & Buchwald, J. S. (1986). Midlatency auditory evoked responses: differential recovery cycle characteristics. Electroencephalography and Clinical Neurophysiology 64, 417—423. Forget, R. & Lamarre, Y. (1987). Rapid elbow flexion in the absence of proprioceptive and cutaneous feedback. Human Neurobiology 6, 27—37. Gratton, G., Coles, M. G. H., Sirevaag, E. J., Eriksen, C. W. & Donchin, E. (1988). Pre and poststimulus activation of response

channels: a psychophysiological analysis. Journal of Experimental Psychology: Human Perception and Performance 14, 331—344. Hallett, M., Shahani, B. T. & Young, R. R. (1975). EMG analysis of stereotyped voluntary movements in man. Journal of Neurology, Neurosurgery and Psychiatry 38, 1154—1162. Hallett, M. & Marsden, C. D. (1979). Ballistic flexion movements of the human thumb. Journal of Physiology 294, 33—50. International Amateur Athletic Federation (1998). Rule 162, Start and Finish. In Rules and Regulations, Division IV, Rules for Men’s and Women’s International Competitions, Section III, Running Events. http:ÏÏwww.iaaf.orgÏrulesÏ90.html

938


(1988). Postural adjustments associated with voluntary contraction of leg muscles in standing man. Experimental Brain Research 69, 469—480. Nickerson, R. S. (1973). Intersensory facilitation of reaction time: energy summation or preparation enhancement? Psychological Review 80, 489—509. Palmer, E., Cafarelli, E. & Ashby, P. (1994). The processing of human ballistic movements explored by stimulation over the cortex. Journal of Physiology 481, 509—520. Nardone, A. & Schieppati, M.

Pascual-Leone, A., Brasil-Neto, J. P., Valls-Sole , J., Cohen, L. G. & Hallett, M. (1992). Simple reaction time to focal transcranial

magnetic stimulation: comparison with reaction time to acoustic, visual, and somatosensory stimuli. Brain 115, 109—122. Sanes, J. A. & Jennings, V. A. (1984). Centrally programmed patterns of muscle activity in voluntary motor behavior in humans. Experimental Brain Research 54, 23—32. Schieppati, M., Nardone, A. & Musazzi, M. (1986). Modulation of the Hoffmann reflex by rapid muscle contraction or realease. Human Neurobiology 5, 59—66. Starr, A., Caramia, M., Zarola, F. & Rossini, P. (1988). Enhancement of motor cortex excitability in humans by noninvasive electrical stimulation appears prior to voluntary movement. Electroencephalography and Clinical Neurophysiology 70, 26—32. Thompson, P. D., Colebatch, J. G., Brown, P., Rothwell, J. C., Day, B. L., Obeso, J. A. & Marsden, C. D. (1992). Voluntary

stimulus-sensitive jerks and jumps mimicking myoclonus or pathological startle sundromes. Movement Disorders 7, 257—262.

Valls-Sole , J., Sole , A., Valldeoriola, F., Mun oz, E., Gonzalez, L. E. & Tolosa, E. S. (1995). Reaction time and acoustic startle in

normal human subjects. Neuroscience Letters 195, 97—100.

Acknowledgements

This work has been accomplished partly as the result of a grant from Fundacio La Caixa (97Ï072).

Corresponding author

J. Valls-Sole : Unitat d’EMG, Servei de Neurologia, Hospital Clinic, Villarroel 170, Barcelona 08036, Spain. Email: [email protected]

J. Physiol. 516.3