INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The concept of a motor program has been used as a model to reconcile the suggestion that a ballistic response (see Ghez 1991) can be prepared in advance and "run off" unaffected by outside influences or peripheral feedback (Keele 1968). A neural corollary of the motor program has been hypothesized to be composed of cortical cell assemblies, which are groups of cortical neurons with strengthened synaptic interconnections (Wickens et al. 1994). Recent evidence has shown that motor programs may be triggered by a startling stimulus. The presentation of an unexpected loud acoustic stimulus has been shown to result in a stereotyped startle response in humans, including activation of the sternocleidomastoid (SCM) and orbicularis oculi (OOc) muscles at short latencies (Brown et al. 1991; Davis 1984; Scott et al. 1999; Yeomans and Frankland 1996). However, when it is paired with a reaction time (RT) task, it has been shown that a startling acoustic stimulus will also elicit a prepared ballistic response at very short onset latencies (Valls-Solé et al. 1995, 1999). Two lines of evidence support the notion of a startle-elicited response. First, the response-related electromyographic (EMG) activation pattern (e.g., Wadman et al. 1979) triggered by the startling stimulus is similar in both burst duration and timing to that produced when participants perform the task in the absence of the startling stimulus (Valls-Solé et al. 1999). Second, task accuracy is maintained during the startle-elicited response (Carlsen et al. 2000). These observations indicate that the intended prepared response has been triggered and that it is not simply a later voluntary response superimposed on an early startle reflex. Since the response is produced at latencies presumably too short to involve the motor cortex, it has been suggested that the response is prepared and stored subcortically and its initiation is triggered by the startling stimulus (Valls-Solé et al. 1999). Since the scope of inquiry into the effects of the startle on prepared movements is limited mainly to RT tasks involving a prepared ballistic response from a static starting position (Carlsen et al. 2000; Siegmund et al. 2001; Valls-Solé et al. 1995 1999), it is not known how the presentation of a startling stimulus might affect an ongoing compound movement task.

Several types of compound movement sequences, such as typing (Terzuolo and Viviani 1979) and handwriting (e.g., Fischman 1984), have been previously examined with results indicating that the sequences are centrally controlled by a motor program. However, the use of proprioceptive information in triggering discrete movement sequences has recently been examined by Cordo (1990) and others (Bevan et al. 1994; Cordo et al. 1994). Participants performed an active elbow extension movement (22°/s) and were required to open their hand within a fixed target area without the use of visual feedback. Although the time taken to reach the target area changed in relation to movement distance and resistance imposed, participants were successful in completing the hand-opening task within the target area. This indicated that some form of kinesthetic feedback from the elbow extension movement was used to trigger the secondary movement, not simply time-to-target (Cordo 1990). Later studies showed that information regarding both the elbow joint angle position and angular velocity were used by the CNS in successfully triggering the secondary movement at the correct position (Bevan et al. 1994; Cordo et al. 1994 1995).

If the secondary movement (hand opening) in the targeting task described above is triggered based on kinesthetic information from the primary elbow extension movement (Bevan et al. 1994; Cordo 1990; Cordo et al. 1994), this suggests that the secondary movement must have been prepared at some point prior to its initiation. Preparation could have taken place either at the same time as the primary movement or at some point during the execution of the primary task. The purpose of the present experiment was to determine whether a secondary movement that is prepared and then initiated on the basis of proprioceptive information about joint angle from a separate ongoing movement could be triggered by a startling acoustic stimulus. If so, the point at which the movement is prepared and loaded for initiation can be probed based on whether the secondary response is evoked at a given probe position. We hypothesized that the secondary movement would be prepared at some point during the execution of the primary task and that this point could be probed successfully with the use of a startling stimulus.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Participants

Thirteen right-handed volunteers (5 male, 8 female; ages 26 ± 3 yr) with no obvious upper body abnormalities or sensory or motor dysfunctions volunteered to participate in the study. All participants gave informed consent, and the study was conducted in accordance with the ethical guidelines set by the University of British Columbia.

Participant set-up

Participants sat upright in a comfortable height-adjustable chair outfitted with an automobile racing harness (Racer Components) to constrain any movement to the forearm segment. The right arm was secured, in a semiprone position with the palm facing inward, to a custom-made aluminum manipulandum that moved in the transverse plane. The starting position of the arm was 80° of flexion (where 180° = full extension) at the elbow (indicated by a fixed physical stop) with the shoulder both flexed and abducted 90°. The medial epicondyle of the right arm was centered over the axis of rotation of the manipulandum.

Instrumentation

Surface EMG data were collected from the muscle bellies of the right extensor digitorum longus (EDL), right biceps brachii (BI), right triceps brachii (TRI), left OOc, and left SCM muscles using bipolar preamplified Ag/AgCl surface electrodes (Therapeutics Unlimited). The recording sites were prepared to remove excess debris, thereby decreasing electrical impedance. The electrodes were oriented parallel to the muscle fibers and then attached using double-sided adhesive strips. A grounding electrode was placed on the participant's left radial styloid process. EMG data were amplified onsite and the electrodes were connected via shielded cabling to an external amplifier system (model 544, Therapeutics Unlimited). The signals were fed from the amplifier and sampled using an analog/digital (AD) interface (Data Translation DT2821) controlled by a customized program written with LabVIEW software (National Instruments). Data collection was initiated by the computer program at the start of each trial.

Arm angular displacement data were collected using a potentiometer attached to the pivot point of a custom-made manipulandum. Hand opening was monitored using a simple finger-tip switch. The switch consisted of electrically conductive wires powered by a 9-V battery attached to thin pieces of aluminum, which were then attached to the thumb and middle finger of the participant's right hand. The switch was configured in such a way as to send a signal of 0 V when it was closed and 9 V when it was open. All signals were digitally sampled at 2,000 Hz and recorded for later analysis.

Stimuli

The warning tone consisted of three short beeps (100 ms, 1,000 Hz, 82 dB each, separated by 500 ms). Trial stimuli were generated using a custom-made tone burst generator, which generated a narrow-band noise pulse (1,000 Hz, 40 ms duration), and presented via a loudspeaker (<1 ms rise time) placed directly behind the head of the participant with intensities of either 84 dB (nonstartle stimulus) or 124 dB (startle stimulus). The stimulus intensities were measured using a sound-level meter (model CR:252B, Cirrus Research) at a distance of 30 cm from the loudspeaker (approximately the distance to the ears of the participant).

Training

Participants practiced extending their right elbow at a constant rate (22°/s) and opening their right hand at a constant target location (55° of extension from starting point) using a feedback-based training program presented on a computer monitor placed 1 m in front of the participant. The position of the manipulandum was represented by a short (1 cm) vertical line that moved in the horizontal plane, with the starting position indicated as a position near the left edge of the screen. The vertical marker line remained fixed at the position at which hand opening occurred providing target accuracy feedback. A 2° wide target area was represented by a 1-cm-wide blue area in the same horizontal plane on the monitor. A green arrow placed above the manipulandum marker moved from slightly left of the starting position, left to right horizontally across the screen at a fixed rate corresponding to an elbow extension velocity of 22°/s. The participants were instructed to mimic the movement of the green arrow with the vertical line and to open their right hand when the vertical line passed through the target area. Once comfortable, participants performed the movement without the use of this visual feedback and were instead only given knowledge of results pertaining to opening angle and movement velocity after each practice trial. Participants were deemed to be competent at performing the task when they could perform five consecutive trials without visual feedback in which both elbow velocity and hand opening were within 2° of their respective target values.

Experimental procedure

The experimental task was the same as the movement learned during training: an active elbow extension (22°/s) and an active hand opening at a target area located 55° into extension. During testing trials, no on-line feedback was given, although knowledge of results pertaining to average movement velocity and opening angle were given following each trial.

Control trials were simply trials in which the participant carried out the normal protocol of the experiment. A series of warning tones signified the start of the trial, at which time the participant performed the task as described above. Participants were advised that fast reaction times were not necessary, but movement soon after the warning tones was required.

Probe trials consisted of either startle (ST) trials or nonstartle (NST) trials. ST trials were trials in which the startle stimulus (124 dB) was given at some point during the trial, whereas NST trials were trials in which the nonstartle stimulus (82 dB) was given at some point during the trial. The probe was given in three distinct positions once the movement was underway, but in only one position on any given trial. The probe was given either early in the movement (5° of extension past the starting point), in the middle of the movement (25° of extension), or late in the movement (45° of extension). Participants were instructed to ignore any auditory stimulus that occurred.

Participants performed three blocks of 20 trials consisting of 13 control (no probe) trials, 3 ST trials (1 at each of 5°, 25°, and 45° of extension), and 4 NST trials (2 at each of 25° and 45° of extension). The order of the trials was randomized and the participants were not aware of the testing order. After the first and second blocks, participants were given two practice trials using the on-line feedback.

Data reduction

Surface EMG burst onsets were defined as the point at which the EMG first began a sustained rise above baseline levels. The location of this point was determined by first displaying the raw EMG pattern on a computer monitor with a superimposed line indicating the point at which activity increased to more than 2 SDs above baseline (mean of 200 ms of EMG activity preceding movement). Onset was then verified by visually locating and manually adjusting the onset mark to the point at which the activity first increased. This method allowed for correction of errors due to the strictness of the algorithm.

Peak EMG amplitudes were defined as the largest EMG amplitude, rectified and filtered with a 25-Hz low-pass elliptic filter, recorded within an interval of 100 ms following EMG burst onset. For each participant, mean peak EMG amplitudes for the five probe conditions were expressed as a percentage of the mean peak EMG amplitude for the control condition. A short time window of 400 ms (from 100 to 500 ms following the probe) was also examined for differences in angular velocity, since effects of the probe would take 100 ms to appear and could be corrected quickly.

Statistical analysis

All dependent variables were analyzed using a simple (6 condition) repeated measures ANOVA to determine whether differences existed between control and probe trials. Differences with a P value of <0.05 were considered significant. Dunnett's and Tukey Honestly Significant Difference post hoc tests were administered to determine the nature of these differences.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Evidence of a startle response

Table 1 shows the mean onset latencies for the OOc and SCM muscles for each condition following the presentation of the stimulus. The onset latencies of SCM were significantly shorter in all ST trials compared with NST trials (P < 0.001). Significant differences also existed in peak EMG amplitude for SCM between the conditions (F(5, 60) = 16.75, P = 0.001; Fig. 1). Post hoc analysis revealed that SCM peak amplitude in all three ST conditions, while not being significantly different from each other, was significantly larger than in Control and NST conditions (P < 0.05). Eyeblink (OOc) activity occurred in both ST and NST trials with no significant difference in onset latency.

View larger version (5K): [in this window] [in a new window]   Fig. 1. Mean peak sternocleidomastoid amplitude and SDs. Values for each condition are expressed as a percentage of the mean peak amplitude for the Control condition.

Task accuracy

Participants were successful in accurately performing the hand-opening task at the correct target angle and accurately extending the elbow at the correct angular velocity in the Control condition (see Table 2; see also Fig. 2 Control). Significant differences were found in the onset angle of the secondary movement depending on the condition (F(5, 60) = 7.377, P < 0.001). Post hoc analysis revealed that hand opening occurred significantly earlier in ST45 trials and significantly later in ST25 trials compared with the Control condition (P < 0.05), indicating that the startle differentially affected the secondary movement depending on the location of the probe.

View larger version (28K): [in this window] [in a new window]   Fig. 2. Individual participant sample data. Each box contains sample elbow angle displacement and triceps raw electromyographic (EMG) data over the entire course of a single 4-s trial for each condition. Dashed horizontal line represents the target angle (55°) while the vertical lines represent probe stimulus presentation (left line) and actual hand opening (right line). Graph for the Control condition shows only 1 vertical line (hand opening) since no probe occurred. In the ST45 condition the initial trajectory of the arm is continued predicting the unperturbed movement (dotted line). ST, startle trial; NST, nonstartle trial.

Analysis of mean angular velocity also revealed differences between the conditions. Mean angular velocities for the entire trial (movement onset to hand opening) were significantly larger in the ST5 and ST25 conditions than in the Control condition (F(5, 60) = 15.150, P < 0.001). However, these differences in velocity were relatively small in comparison to differences to angular velocity during a short time window (400 ms) following the probe (see Table 2). Much larger differences existed in the mean angular velocities during this short time period (F(5, 60) = 11.133, P < 0.001; Table 2). Post hoc analysis showed that it was the mean angular velocities for 400 ms following the probe in the ST25 and ST45 conditions that were significantly larger than comparable epochs during Control trials (P < 0.05), indicating that the startle affected the ongoing extension movement (see Fig. 2). Mean angular velocity during the 400-ms window in the ST5 condition, although not significantly different at the P = 0.05 level, was also larger than in the Control trials.

EMG data

Analysis of surface EMG data for EDL indicated that differences existed in both the timing and magnitude of the EDL response (related to hand opening) across conditions (see Table 3 and Fig. 3). The angle at which EDL onset occurred was found to be different between the conditions (F(5, 60) = 9.797, P < 0.001; Table 3). Post hoc tests revealed that EDL onset occurred at a significantly smaller amount of extension in ST45 trials and at a significantly larger amount in ST25 trials compared with Control trials (P < 0.05). Significant differences were also exhibited in the time from displacement onset to EDL muscle onset (F(5, 60) = 14.437, P < 0.001; Table 3). Time to EDL onset was significantly shorter in all ST trials (P < 0.05) compared with Control trials, while no differences existed between Control trials and NST trials.

View larger version (7K): [in this window] [in a new window]   Fig. 3. Mean peak EMG amplitudes for extensor digitorum longus (open squares) and triceps (solid squares). Peak EMG amplitudes in each condition expressed as a percentage of mean control amplitude for each muscle.

Although EDL peak EMG amplitude analysis revealed differences between the conditions (F(5, 60) = 3.265, P = 0.021), post hoc tests showed that EMG amplitude was only significantly larger than Control amplitude in the ST45 condition (P < 0.05; Fig. 3). Analysis of peak EMG amplitude for TRI also revealed differences between the conditions (F(5, 60) = 7.522, P = 0.008). Post hoc tests showed that TRI peak EMG amplitude was larger in both the ST25 and ST45 conditions than in Control trials (P < 0.05). Peak TRI amplitude in the ST5 condition was also larger than Control amplitude, although not sufficiently large to achieve statistical significance (Fig. 3).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Previous studies involving a "triggering effect" of a startle on a prepared movement have involved ballistic RT tasks (Carlsen et al. 2000; Siegmund et al. 2001; Valls-Solé et al. 1995, 1999). In the present study we examined the effect of a startling stimulus on the secondary component of a compound movement task. Our findings showed that a startle cannot only be used to probe response preparation in RT tasks, but also in an ongoing task (i.e., non-RT task). Furthermore, our findings revealed that the secondary component of the task (hand opening) was not prepared prior to task initiation but was prepared and "loaded" on-line, during execution of the primary component of the task (arm extension). Thus, although proprioceptive information was used successfully to execute the secondary movement under normal conditions, when the participant was startled late in the primary movement, the startle was effective in overriding the kinesthetic coding and inappropriately triggering the secondary movement.

This experiment built on the work of Cordo (1990), Cordo et al. (1994), and Bevan et al. (1994) by further describing the nature of compound movements that involve proprioceptive triggering. This approach was also novel since it was unknown whether the secondary movement was prepared along with the primary movement in advance of any movement or whether the secondary movement was prepared on-line during the execution of the primary movement. In addition, using the startle to trigger the secondary movement without the involvement of the CNS allowed the identification of the epoch in which the secondary movement was prepared. Similarly, this approach allowed us to test the hypothesis that the secondary movement, once prepared, was stored in lower brain structures.

The task

The experimental task was for participants to perform an active elbow extension movement at a constant rate of 22°/s coupled with an active hand opening at a prescribed target joint angle without the use of visual feedback pertaining to joint position. Consistent with data from Cordo et al. (1994), participants were successful in performing the secondary task (hand opening) at the target angle. Findings from the present study showed that, in Control trials, participants performed the hand opening task with a mean error of only 0.12°, while 67% of trials had an opening angle within 1.19° of the target (for the range of opening angle values see Table 1).

Evidence has previously been provided by Cordo et al. (1994) in support of the notion that participants in their studies used proprioceptive information regarding elbow joint position to trigger the execution of the secondary task, rather than a timing strategy involving time-to-target. Evidence for a timing strategy in the present experiment would be in the form of consistent times between conditions from movement onset to hand opening. Figure 2 and Tables 2 and 3 include data that would not support the notion of a timing strategy. Although hand opening occurred at similar joint angles across conditions, the time to hand opening and EDL onset were different. For example, although in both the Control condition and the ST5 condition hand opening accuracy was very high (Table 2), the time-to-target intervals in these conditions were very different (Table 3), indicating that timing was not used to trigger hand opening. Hand opening in ST trials occurred earlier than other trials, owing to increased movement velocity in these trials. For example, in the ST25 condition, if the participant had used a timing strategy based on the early part of the elbow trajectory, hand opening would have occurred much later than observed (see Fig. 2, ST25 dotted line).

Further evidence that participants used a timing strategy would also be expressed in the form of low variability in mean angular velocity across trials. This was not seen in Control trials, as the SD of angular velocity was 1.04°/s (Table 2). At a constant rate of 22°/s, a 55° elbow extension would take 2.5 s to complete. The angular velocity variability of 1.04°/s would therefore manifest itself in a targeting SD of 2.6° over a 2.5-s time interval. This is much higher than the 1.19° SD exhibited by participants in Control trials in the present experiment. Thus it was concluded that participants were not simply using a timing strategy of opening the hand 2.5 s into the movement and suggests that the assertion by Cordo et al. (1994) that participants used kinesthetic feedback information to trigger the hand opening was correct.

Effects of the startling stimulus

EMG data from the OOc and SCM were analyzed for evidence that a startle response was elicited due to the presence of a startling stimulus. These indicators have been used extensively in other studies involving a startle response (Brown et al. 1991; Carlsen et al. 2000; Siegmund et al. 2001). Analysis revealed that, although the latency of OOc activity (see Table 1) observed in the ST trials was similar to that previously observed by Brown et al. (1991) in response to a startle (36.7 ms), it was also observed at similar latencies in the NST trials (see Table 1). In contrast, SCM activity was present at shorter latencies and larger amplitudes in the ST trials compared with NST trials (see Table 1), suggesting that the responses to the two stimuli were different. Furthermore, since the latency of the SCM was similar to that described previously (58.3 ms) by Brown et al. (1991), we are satisfied that a startle response was elicited in the ST trials and not in the NST trials. The occurrence of a startle in the present study resulted in several effects. One of these effects was that the secondary movement was elicited early in the ST45 trials. On average, the hand was opened 3.2° before the target when participants were startled late in the primary movement. This resulted because, in the ST45 trials, EDL muscle activity occurred 5.99° prior to the target. In comparison, in Control trials, EDL onset occurred only 2.00° prior to the target. We assert that, in the ST45 condition, the startling stimulus triggered the secondary movement. It has been shown previously that startle-evoked EMG activity in the forearm extensors occurs with a median latency of approximately 73 ms (Brown et al. 1991). In the present study, EMG onset in the EDL muscle resulting in the secondary movement occurred at a similarly short latency (median latency of 82.3 ms) following the presentation of the startling stimulus. Thus it appears that, rather than being initiated normally by the CNS, the secondary movement was elicited early by the startle.

Under normal conditions in discrete movement sequences, proprioceptive information from Ia and II afferents, in the form of velocity and positional information, can be used by the CNS to determine when to trigger the secondary movement (Cordo et al. 1994). Results from experiments by Valls-Solé et al. (1999), Carlsen et al. (2000), and Siegmund et al. (2001), however, indicated that prepared movements could also be triggered by the same structures that are activated by the startle response. In order for this to take place, the motor commands must have been accessible by or "loaded into" lower structures in the brain stem and midbrain, such as the reticulospinal centers, and have been ready to be released. As such, the triggering of the secondary movement by the startle in the present experiment is consistent with previous startle literature. Because the early elicitation of the secondary movement (observed in the ST45 condition) was not evident in other conditions, this also gives insight into the temporal structuring of movement preparation for secondary movements based on a primary movement. Thus results from the present study indicate that the motor program responsible for the execution of the secondary movement was "loaded" near the end of the primary elbow extension movement, likely based on the proprioceptive information processed on-line by the CNS (see Cordo et al. 1994). The startling stimulus evoked the secondary movement when it was presented 10° prior to the target, but not when presented 30° or 50° prior to the target, suggesting that the secondary movement was prepared and loaded between 10 and 30° prior to reaching the target. Based on an average angular velocity of approximately 22°/s, this corresponds to a temporal window of between 400 and 1,500 ms prior to the target. It should be noted that the 400-ms value is a conservative estimate based on the task employed and by no means indicates a lower limit on loading, as it has been previously shown that movement sequences lasting just 210 ms can be coordinated based on proprioceptive information (Cordo et al. 1994). As an alternative hypothesis to late program loading, it may be that the secondary movement was loaded at the onset of the primary movement but was not accessible to triggering by a startle stimulus. That is, if the movement was prepared and loaded at the beginning of the movement but is only accessible during a certain time window prior to the target, the same results would be observed.

The hand-opening movement was not evoked by the startling stimulus when it was presented 30° prior to the target (ST25). In fact, in this condition, hand opening occurred significantly later than in the other conditions (a 2.85° target overshoot). On average, EDL onset in these trials did not occur until the target elbow angle had been passed by 0.9°. In the present study, an increased burst of EMG activity was observed in the TRI during the ongoing extension at a short latency (median latency of 82 ms) following the startling stimulus. It has been shown previously that startle-evoked EMG activity in the TRI occurs with a median latency of approximately 71 ms (Brown et al. 1991). Since the observed burst latency value is similar to that reported previously, it appears as though the EMG in the TRI of the ongoing elbow extension was facilitated by the startle, possibly by the startle volley summing with the voluntary EMG activity (see Siegmund et al. 2001). Cordo et al. (1994) showed that, when participants' elbows were passively extended, they consistently overshot the target when elbow joint angular velocity was unpredictable. They hypothesized that, when velocity information was unreliable, joint position information was used by the CNS to trigger the secondary movement. However, when triggering was based on positional information, it occurred only after the target angle had been reached (Cordo et al. 1994). We assert that EMG facilitation due to the startle resulted in a large acceleration about the elbow, which would have caused the angular velocity to become less predictable. An overshoot of the target occurred during ST25 trials, suggesting that the new velocity information could not be used in the short amount of time between the startle and arrival at the target; thus triggering of the secondary movement was delayed until after the target had been reached.

The finding of a target overshoot was unique to ST25 trials. In the ST5 condition, the participant was startled at the beginning of the elbow extension, allowing more time for a correction in the angular velocity. Thus, as the movement progressed, the velocity became more predictable, allowing the participant to execute the hand opening on target. In the ST45 condition, a large acceleration of the elbow also occurred, resulting in a change in the angular velocity. However, since the program responsible for the execution of the hand opening was already loaded into lower brain structures, it was triggered by the startle. As such, the hand opening occurred early (prior to proprioceptive triggering) and thus was independent of the movement of the arm following presentation of the probe.

EMG activity was increased in both the TRI and EDL as a result of the startling stimulus and is consistent with previous literature (Siegmund et al. 2001). Yeomans and Frankland (1996) suggested that increased amplitude of the startle response was the result of the number of giant neurons recruited in the reticularis pontis caudalis (RPC) in response to the startling stimulus. Larger stimulus intensities recruit more RPC neurons, which may then conduct to the various levels of the spinal cord, along the reticulospinal tract. If the reticulospinal tract was an important part of the motor pathway for the task in the present study, the startle would have acted to facilitate the ongoing movement at the time of the startle through the increased activation of the RPC giant neurons. This would have led to lower recruitment thresholds in the spinal motor neuron pool and thus to the facilitated movements and movements of higher amplitude exhibited in the present study.