Humans encounter unexpected transient perturbations to posture while standing, walking, and sitting. While standing, we may be bumped by passersby or have the bus, subway, or tram lurch beneath our feet. While walking, we may slip, stumble, or trip on contaminated or uneven surfaces. And while sitting, we may be struck from behind in a low-speed automobile crash. All of these perturbations consist of externally applied forces that either move the body or undermine its base of support. To handle these manifold perturbations, humans have developed numerous and widely varying responses that are presumably aimed at restoring balance, a desired posture, or protecting the body from injury. Other types of stimuli, such as loud noises or skin stimulation (e.g., an air puff or contact from an insect), can also perturb balance and posture by evoking a rapid protective response that is commonly called a startle reflex. Unlike the external perturbations described earlier, the external force applied by a loud noise or skin stimulation is negligible. Instead, the initial destabilizing movement and subsequent recovery strategy are generated internally by muscles (although gravity also is present).
These perturbations are generally transient - meaning they occur or are applied over a short duration. Our responses to these transient perturbations can vary widely with the magnitude, familiarity, and expectation of the perturbation. Large, unexpected, or unfamiliar perturbations are generally more destabilizing and more likely to evoke exaggerated responses than small, expected, or familiar perturbations. Indeed, the first exposure to a transient perturbation often evokes an exaggerated response that then diminishes with repeated exposures. This exaggerated response to the first exposure of an unexpected perturbation suggests that something more than a postural response is being evoked, and it is this first exaggerated response that is the focus of this article. The objective of this review is to examine the first response to a transient imposed stimulus, and based on our research, to present the novel hypothesis that the first response consists of both a postural and a startle response.
ADAPTATION, ATTENUATION, AND HABITUATION
The first exposure to an unexpected or unfamiliar whole-body acceleration generally evokes an indiscriminate muscle response wherein both agonist and antagonist muscles are activated concurrently. This response has been called a "clamp down" strategy and likely serves to stiffen the joints to protect against excessive movement (12). After repeated exposures to the same stimulus, there is an attenuation of the muscle response, particularly in muscles that oppose restoration of the initial posture (3,17). The antagonist muscle response can, in some cases, vanish, and the agonist muscle response becomes tuned to restoring balance or posture without overshooting the initial posture. For instance, during a seated forward acceleration that simulates a low-speed rear-end collision, the first exposure evokes a muscle response in both the anterior (sternocleidomastoid [SCM]) and posterior (cervical paraspinal [PARA]) muscles and a rapid restoration of the upright head posture (Fig. 1A). By the third exposure to the same stimulus, the response of the posterior muscles is largely extinguished, the response of the anterior muscles is attenuated, and the neck remains extended for a protracted period (Fig. 1B).
The attenuation of the neuromuscular response between the exaggerated response of the first exposure to a transient acceleration and the stabilized response after multiple identical accelerations is known as habituation (10) (Sensitization rather than habituation to a stimulus can occur if the stimulus is noxious. This review focuses on perturbations that generate habituation). Habituation of the neuromuscular response often changes the kinematic responses as well. The joints spanned by the muscles undergoing attenuation become less stiff, and this diminishes the peak acceleration and potentially increases the peak displacement of body segments that are distant to the site of the perturbation (Fig. 1C).
Many studies of transient whole-body accelerations expose subjects to a number of "familiarization" trials before beginning the experiment. Familiarization produces a stabilized neuromuscular and kinematic response that can then be repeatedly evoked over multiple trials. From an experimental perspective, stabilized responses offer two main advantages: (a) variability is reduced by averaging multiple similar trials, and (b) within-subject comparisons can be made to assess the influence of an intervention. Both advantages can produce impressive data and improve the chance of finding significant effects. Of course, the problem with using stabilized responses is that they may not be relevant to the problem being studied. For instance, much of our research has been focused on whiplash injuries that most commonly occur from a single exposure (no opportunity for habituation) and often occur without warning (no expectation or preparation). For our purposes, the stabilized response bears little resemblance to the response of interest (compare Fig. 1A with Fig. 1B), and thus we are almost exclusively interested in the first response to a transient acceleration. Therefore, using stabilized or habituated responses can potentially compromise the external validity of an otherwise excellent experiment - particularly if the goal is to study unexpected or unfamiliar perturbations.
INTERACTION OF POSTURAL RESPONSES AND STARTLE RESPONSES
As with whole-body postural responses, startle responses also habituate rapidly to repeated presentation (2,8). Unlike postural responses, however, habituated startle responses are often entirely extinguished in both the agonist and antagonist muscle groups spanning a particular joint. The time course of startle habituation is also similar to that observed after repeated whole-body accelerations. On the basis of this similarity, it was proposed that the initially elevated prehabituated muscle response during the first whole-body acceleration is, in part, a startle response (1). This proposal has a certain appeal: in its simplest form, the exaggerated part of the first response to a whole-body acceleration is caused by startle, which then disappears with habituation, leaving only the underlying appropriate response evoked by the postural control system. Perhaps more realistically, there is some tuning of the postural response with repeated exposures, although it is not clear whether the magnitude of any tuning effect is smaller or larger than the magnitude of the startle effect. Despite the simplicity of this proposal, it has not been simple to demonstrate. In our attempt to do so, we first tried to restore the exaggerated first response by combining a habituated response with an independently evoked startle response. We then turned to more sophisticated analytical techniques to determine whether there was evidence of a distinct startle response within the electromyogram itself. Both approaches are described later.
Can the First Response Be Restored With an Alternate Startle?
Our first approach was to restore the initial response by superimposing an acoustic startle response onto an already habituated postural response. We elected to do this in seated subjects exposed to abrupt forward horizontal accelerations on a linear sled that simulates a low-speed rear-end impact. Subjects were fitted with surface electromyography (EMG) electrodes over their SCM, scalene, and cervical PARA muscles to record the neuromuscular responses and with accelerometers and motion-tracking markers to record the kinematics. Subjects were first exposed to 11 identical whole-body accelerations (peak acceleration of 15.2 m·s−2, reaching 50 cm·s−1 after 59 ms) to achieve a stabilized habituated response and then exposed to 5 additional accelerations on which was superimposed a loud acoustic stimulus known to generate a startle response (124 dB, 40 ms, 1 kHz) (16). Peak EMG amplitude and peak kinematic responses across the 16 trials were then compared.
The addition of the startling tone restored the EMG amplitudes of all three neck muscles and the peak amplitudes of four (out of five) habituation-affected kinematic parameters to levels not significantly different from those measured before habituation, that is, during the first acceleration-only trial (Fig. 2) (6). These initial observations suggested that the first exposure to a forward acceleration contains a combined startle and postural response. Of course, an obvious difference between the postural-startle response evoked by the first sled-only perturbation and the postural-startle response evoked by the sled-tone perturbation stimuli resides in the sensory afferents mediating the startle response. Any startle response to the first sled-only perturbation was presumably triggered by somatosensory or vestibular afferents and not acoustic afferents. Our results demonstrated that a startle response triggered by a loud acoustic stimulus can interact with the neck postural responses elicited by the sled acceleration. This acoustically triggered startle response increases neck muscle coactivation during the acceleration, leading to a stiffer head-neck system, as suggested by the larger forward head acceleration and smaller head extension angle during exposures to the combined stimuli.
These findings are consistent with the first response being a combined postural and startle response, but fall short of proving this proposal. For one, we made no effort to match the amplitude of the startling tone to that of the forward acceleration, and therefore it is not known whether the startle response to the tone should restore the habituated component of the acceleration response. Moreover, restoration of the EMG and kinematic amplitudes shows only that the initial muscle and kinematic responses can be restored but does not prove that the neural activity that evoked the exaggerated responses in the first place constituted a startle response. Thus, to potentially identify a startle response in the first trial response, we had to scrutinize the electromyogram obtained from the initial response.
Changes in the Electromyograms
Correlation analysis in the frequency domain has shown increased coherence at 10 to 20 Hz between bilaterally homologous upper limb muscles after an auditory startle but not after a sham startle or during a voluntary contraction (9). Grosse and Browne (9) also proposed that neuromuscular coherence in this bandwidth represented a surrogate marker of reticulospinal activity because reticular structures are known to lie along the startle reflex pathway (18).
These numerical techniques were applied to data acquired from three separate experiments: one involving a single exposure to a horizontal acceleration (15), another involving 11 exposures to examine habituation (17), and a third - already described previously - wherein an acoustic startle was superimposed onto the habituated response (6). Our goal was to quantify the amount of coherence in the 10- to 20-Hz bandwidth between homologous neck muscles during accelerations simulating rear-end impacts (4). The specific goals were to determine (a) whether a single perturbation would evoke synchronous activity in neck motoneurons, (b) whether habituation would extinguish this synchronous activity, and (c) whether the superposition of an acoustic startle over a habituated postural response would restore the correlated activity.
In all three data sets, coherence between the left and right SCM muscles exhibited a peak in the 10- to 20-Hz region for the first exposure to the forward acceleration. In the two experiments with multiple trials, this peak was absent in thehabituated responses. In the final experiment, where the acoustic startle was added to the habituated response, the peak at 10 to 20 Hz reappeared (Fig. 3A). Moreover, there were no significant differences in the coherence peak between the first exposure and in the habituated + startle exposures (Fig. 3B). These results extended our earlier work by demonstrating that the EMG restored by an acoustic startle was in the same bandwidth as the EMG that disappeared with habituation. Although these findings do not prove that the first response contains a startle response, it does suggest that the same neuronal associations were responsible for the 10- to 20-Hz peak in coherence.
Of the two approaches we used to examine whether startle explained the exaggerated response to the first exposure to a forward acceleration, neither was able to disprove our hypothesis. Rather, both were consistent with the possibility that the exaggerated response to the first impact was some combination of a postural and startle response - an idea that we explored further with a simple model of the potential interaction between these two stimuli.
Changes in Muscle Onset Times
When we superimposed the acoustic startle and sled acceleration described previously (6), we delayed the onset of the startle tone by 18 ms because prior work has shown that muscle onset to the acoustic startle occurs 18 ms before muscle onset to the sled acceleration (14). Despite this attempt to contrive coincident sensory volleys at the motoneuron, we found that muscle onset occurred 9 to 12 ms earlier when both stimuli were presented (with an 18-ms offset) than when the acceleration-only stimulus was presented. This finding led to another experiment wherein we varied the relative timing of the startling tone (ST) and the sled acceleration (A) in five discrete steps from 0 ms (A + ST0, startle tone onset is coincident with acceleration onset) to 28 ms (A + ST28, startle tone onset occurred 28 ms after acceleration onset). Again focusing on the SCM muscle, we observed increasingly shorter muscle onset times as the delay between stimuli shortened (Fig. 4). We also observed larger EMG amplitudes in all conditions of combined stimuli compared with both the acceleration-only and startle tone-only trials. The absence of a significant difference between the onset times for the A + ST0 condition and the startle tone-only condition suggests that muscle onset during the A + ST0 trials was driven primarily by the acoustic stimulus. At the longest delay (28 ms), muscle onset still occurred earlier in the combined stimulus condition (A + ST28) than in the acceleration-only condition. The absence of a neck muscle response that was time locked to the acoustic startle stimulus indicates that the acoustic startle stimulus by itself did not trigger the earlier execution of the muscle responses to the forward acceleration in all cases, and that there was some form of interaction between the sensory information.
A Simple Posture-Startle Interaction Model
A simple explanation for the interaction we observed between the postural and startle stimuli is a summation of the two responses. For example, summation of the response amplitudes to cutaneous and acoustic stimuli has been previously observed in startle responses in the rat at both the functional and cellular levels (11,13) - a finding that provides a physiological basis for considering summation inour case. These prior studies, however, focused on summation of the response amplitude, whereas we focused on the relative timing of the two stimuli.
To further explore a summation-type interaction, we created a simple model of the net excitatory drive of both stimuli converging onto a pool of motoneurons (Fig. 5A). For purposes of evaluating the suitability of this model, the amplitude of the habituated excitatory drive from the acceleration (resulting from some combination of somatosensory, vestibular, and visual afferents) was arbitrarily set to one. The amplitude of the habituated drive from the acoustic stimulus was set to a fraction of the habituated acceleration-evoked drive (50% shown in Fig. 5B). A drive threshold for motoneuron activation between the peak drive levels of the acceleration and acoustic stimuli was chosen to produce neck muscle activity during habituated acceleration-only trials but not during habituated acoustic-tone-only trials (6,7,16). The excitatory drive for the acceleration-only trial in Figure 5B was temporally set to coincide with the onset of SCM EMG activity at 85 ms (indicated by the arrow) to match our experimental observations. The excitatory drive for the acoustic-only stimulus begins 18 ms earlier to account for the 18-ms-shorter latency of the acoustic startle response than the acceleration-evoked response in the SCM muscle.
In Figure 5C, the relative timing of the drive for the other four acoustic stimuli we presented are also shown (delayed by 13, 18, 23, and 28 ms). When the excitatory drive from each acoustic stimulus was independently added to the excitatory drive from the acceleration-only stimulus (Fig. 5D), the summed excitatory drives reached threshold earlier and thus predicted earlier muscle onset times. The model predicted onset times during the A + ST0 condition that were 5 to 8ms earlier than onset during the A + ST18, A + ST23, and A + ST28 conditions. This is similar to the 6.5-ms advance observed in our experimental data. Similarly, onset times for A + ST18, A + ST23, and A + ST28 conditions were predicted to be 7 to 11 ms earlier than acceleration-only condition - again similar to the 3 to 13 ms we observed experimentally.
The model also predicted that peak excitatory drive would increase, presumably generating a larger muscle activity through the possible recruitment of more or larger motoneurons. This too is consistent with the larger EMG we observed during combined stimuli conditions compared with the acceleration-only condition. Additional work varying the relative amplitude of the two stimuli is needed to further develop the model, and a more sophisticated model of excitatory drive should also be implemented. Viewed broadly, however, the temporal and spatial changes in EMG can be explained, at least partially, by a relatively simple summation of the excitatory drive converging at or before the neck muscle motoneurons.
CURRENT ISSUES AND FUTURE DIRECTIONS
To date, our experimental work and preliminary modeling work support the idea that a combined startle and postural response account for the exaggerated neuromuscular response observed during the first exposure to a transient acceleration. However, more work - perhaps using lesion experiments in an animal model - is needed to conclusively show that the exaggerated response during the first trial is a combined postural and startle response. In addition, because of our interest in whiplash injuries, we have focused on neck muscles. The fact that startle reflex connections to the neck muscles are particularly strong (8) has been fortuitous for us. It remains unclear, however, whether similar results will be found for the exaggerated responses in lower limb muscle during unexpected perturbations of standing posture.
Another issue requiring further work is the evidence of bilateral correlations between the neck muscles during nonstartle conditions (5). Nonreflexive activation of the neck muscles is thought to occur through the reticular formation, and the startle reflex is known to pass through this same structure. If Grosse and Brown (9) are correct that coherence in the 10- to 20-Hz bandwidth indicates reticulospinal activity, then it is not clear why this synchrony disappears during habituated responses of the neck muscles. Further work is needed to resolve this issue.
We set out to examine whether the exaggerated neuromuscular responses observed during the first exposure to a novel transient acceleration was the combination of postural and startle responses. If postural and startle responses combined to generate this exaggerated response, then habituation may be explained as the extinction of the startle response, leaving only the postural response as the stabilized or habituated response. The experiments we conducted did not generate any evidence to contradict our hypothesis, although additional experiments are needed to examine this possibility. A simple summation model of excitatory drive explains the changes in onset latency we observed, but additional work varying stimulus amplitudes is needed to further develop the model. Nonetheless, the current data support our hypothesis that postural and startle responses combine to generate the exaggerated neuromuscular responses evoked during the initial exposure to an unexpected or unfamiliar transient acceleration simulating a low-speed rear-end collision.
1. Allum, J.H.J., F. Honegger, and E.A. Keshner. Head-trunk coordination in man: is trunk angular velocity elicited by a support surface movement the only factor influencing head stabilization. In: The Head-Neck Sensory Motor System
, A. Bethoz, W. Graf, and P.P. Vidal (Eds.). New York, NY: Oxford University Press, 1992, pp. 571-575.
2. Bisdorff, A.R., A.M. Bronstein, and M.A. Gresty. Responses in neck and facial muscles to sudden free fall and a startling auditory stimulus. Electroencephalogr. Clin. Neurophysiol
. 93:409-416, 1994.
3. Blouin, J.S., M. Descarreaux, A. Belanger-Gravel, M. Simoneau, and N. Teasdale. Attenuation of human neck muscle activity following repeated imposed trunk-forward linear acceleration. Exp. Brain Res
. 150:458-464, 2003.
4. Blouin, J.S., J.T. Inglis, and G.P. Siegmund. Startle responses elicited by whiplash
perturbations. J. Physiol
. 573(Pt 3):857-867, 2006.
5. Blouin, J.S., J.T. Inglis, M.G. Carpenter, and G.P. Siegmund. Neural control of superficial and deep neck muscles
in humans. J. Neurophysiol
. 98:920-928, 2007.
6. Blouin, J.S., G.P. Siegmund, and J.T. Inglis. Interaction of acoustic startle and habituated neck postural responses
in seated subjects. J. Appl. Physiol
. 102:1574-1586, 2007.
7. Brown, P., B.L. Day, J.C. Rothwell, P.D. Thompson, and C.D. Marsden. The effect of posture on the normal and pathological auditory startle reflex
. J. Neurol. Neurosurg. Psychiatry
. 54:892-897, 1991.
8. Brown, P., J.C. Rothwell, P.D. Thompson, T.C. Britton, B.L. Day, and C.D. Marsden. New observations on the normal auditory startle reflex
in man. Brain
. 114(Pt 4):1891-1902, 1991.
9. Grosse, P., and P. Brown. Acoustic startle evokes bilaterally synchronous oscillatory EMG activity in the healthy human. J. Neurophysiol
. 90:1654-1661, 2003.
10. Harris, J.D. Habituatory response decrement in the intact organism. Psychol. Bull
. 40:385-422, 1943.
11. Li, L., and J.S. Yeomans. Summation between acoustic and trigeminal stimuli evoking startle. Neuroscience
12. Nashner, L.M. Adapting reflexes controlling the human posture. Exp. Brain Res
. 26:59-72, 1976.
13. Schmid, S., N.S. Simons, and H.U. Schnitzler. Cellular mechanisms of the trigeminally evoked startle response. Eur. J. Neurosci
. 17:1438-1444, 2003.
14. Siegmund, G.P. The Reflex Response of Human Neck Muscles to Whiplash-Like Perturbations (Doctoral thesis)
. Vancouver, BC: University of British Columbia, 2001.
15. Siegmund, G.P., D.J. King, J.M. Lawrence, J.B. Wheeler, J.R. Brault, and T.A. Smith. Head/Neck Kinematic Response of Human Subjects in Low-Speed Rear-End Collisions (973341). Proc. 41st Stapp Car Crash Conference
. Warrendale, PA: Society of Automotive Engineers, 1997, pp. 357-385.
16. Siegmund, G.P., J.T. Inglis, and D.J. Sanderson. Startle response of human neck muscles
sculpted by readiness to perform ballistic head movements. J. Physiol
. 535(Pt 1):289-300, 2001.
17. Siegmund, G.P., D.J. Sanderson, B.S. Myers, J.T. Inglis. Rapid neck muscle adaptation alters the head kinematics of aware and unaware subjects undergoing multiple whiplash
-like perturbations. J. Biomech
. 36:473-482, 2003.
18. Yeomans, J.S., and P.W. Frankland. The acoustic startle reflex
: neurons and connections. Brain Res. Rev
. 21:301-314, 1995.