UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The development of stochastic vestibular stimulation and its application to dynamic vestibular evoked… Dakin, Christopher James 2012

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2012_fall_dakin_christopher.pdf [ 3.93MB ]
Metadata
JSON: 24-1.0073200.json
JSON-LD: 24-1.0073200-ld.json
RDF/XML (Pretty): 24-1.0073200-rdf.xml
RDF/JSON: 24-1.0073200-rdf.json
Turtle: 24-1.0073200-turtle.txt
N-Triples: 24-1.0073200-rdf-ntriples.txt
Original Record: 24-1.0073200-source.json
Full Text
24-1.0073200-fulltext.txt
Citation
24-1.0073200.ris

Full Text

The Development of Stochastic Vestibular Stimulation and its Application to Dynamic Vestibular Evoked Responses.  by  Christopher James Dakin  B.H.K., The University of British Columbia, 2007   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF  THE REQUIREMENTS FOR THE DEGREE OF   DOCTOR OF PHILOSOPHY   in   The Faculty of Graduate Studies   (Kinesiology)   THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)  September 2012   © Christopher James Dakin, 2012   ii Abstract   The vestibular system provides sensory information regarding linear and angular motion of the head for tasks such as spatial navigation and postural stabilization. In these dynamic environments examination of vestibular signals is experimentally difficult given current techniques. Recently, continuous stochastic stimuli have shown promise in addressing some limitations in current vestibular probes and might provide a useful tool for investigating the dynamic behaviour of the vestibular system. The purpose of this thesis is a) to develop further the stochastic stimulus format by examining the customizability of the stimulus bandwidth and the stimulus’ effectiveness in extracting dynamic responses, and b) to use these advancements to explore dynamic vestibular function during locomotion and head rotation.  Exploration of the customizability of stimulus bandwidth revealed that a single broad bandwidth stimulus provides similar information to the sum of a series of sinusoidal stimuli or narrow bandwidth stimuli, but in much less time, and that stimulus bandwidth can be modified, by removing frequencies below 2 Hz, to attenuate the postural perturbation created by the stimulus. In a dynamic context the stochastic stimulus was also shown to be very effective in extracting the time varying modulation of vestibular-evoked responses during motion by identifying phase-dependent vestibular responses in the gastrocnemius during locomotion. The stochastic stimulus was then used to examine vestibular modulation and suppression during locomotion and vestibular spatial transformation during head turn. During locomotion, phase-dependent modulation of vestibular responses was observed in muscles of the leg and hip. In some muscles around the ankles these responses are attenuated with increasing cadence and walking speed. Lastly the transmission and spatial transformation of these vestibular-evoked responses are not hindered by motion and the spatial transformation   iii occurs in nearly real time during head rotation.  In general, the stochastic stimulus can be customized to reduce postural sway and is effective in extracting the dynamic modulation of vestibular influence on muscle activation. The identification of widespread phase-dependent vestibular coupling in the lower limbs and continuous spatial transformation of vestibular signals demonstrates that the stochastic waveform is an effective tool for the investigation of human vestibular physiology in dynamic contexts.   iv Preface A version of Chapter 2 has been published: Dakin CJ, Inglis JT & Blouin JS. (2011). Short and medium latency muscle responses evoked by electrical vestibular stimulation are a composite of all stimulus frequencies. Exp Brain Res 209, 345-354. I was responsible for 70% of the work. My role included collecting the data analyzing and interpreting the results and writing the published manuscript. Dr Blouin aided in planning the experiment interpreting the results and reviewed the manuscript. Dr Inglis aided in planning the experiment and reviewed the manuscript. This research was conducted under ethics provided under UBC’s Clinical Ethics Board, ethics number: H07-03119, H05- 70609. Check the first pages of this chapter to see footnotes with similar information.  A version of Chapter 3 has been published: Dakin CJ, Luu BL, van den Doel K, Inglis JT & Blouin JS. (2010). Frequency-specific modulation of vestibular-evoked sway responses in humans. J Neurophysiol 103, 1048-1056. I was responsible for 65% of the work. My role included collecting the data analyzing and interpreting the results and writing the published manuscript. Dr Luu collected some of the data and reviewed the final manuscript. Dr van den Doel developed the mathematical approach for the data analysis. Dr Inglis aided in planning the experiment and reviewed the final manuscript. Dr Blouin aided in planning the experiment interpreting the results and reviewed the manuscript. This research was conducted under ethics provided under UBC’s Clinical Ethics Board, ethics number: H07-03119. Check the first pages of this chapter to see footnotes with similar information.   v  A version of Chapter 4 has been published: Blouin JS, Dakin CJ, van den Doel K, Chua R, McFadyen BJ, & Inglis JT (2011) Extracting phase-dependent human vestibular reflexes during locomotion using both time and frequency correlation approaches. Journal of Applied Physiology, 111, 1484-1490. Dr Blouin collected the data, interpreted some of the results and partially wrote the manuscript. My role entailed analyzing and interpreting the results and writing the published manuscript. I was responsible for 45% of the work. Dr van den Doel developed the mathematical approach for the data analysis. Dr McFadyen aided in planning the experiment and reviewed the final manuscript. Dr Inglis aided in planning the experiment and reviewed the final manuscript. This research was conducted under ethics provided under UBC’s Clinical Ethics Board, ethics number: H07-03119. Check the first pages of this chapter to see footnotes with similar information.  Chapter 5 is based on work conducted in UBC’s Sensorimotor physiology lab and neurophysiology lab by Christopher James Dakin, Dr Romeo Chua and Dr Timothy Inglis and Dr Jean Sébastien Blouin. I was responsible for 65% of the work which consisted of data analysis and interpretation as well as production of the final manuscript. Dr Blouin collected the data, and reviewed the manuscript. Drs Chua and Inglis contributed to the initial planning of the experiment and reviewed the final manuscript. This research was conducted under ethics provided under UBC’s Clinical Ethics Board, ethics number: H07-03119. Check the first pages of this chapter to see footnotes with similar information.     vi Chapter 6 is based on work conducted in UBC’s Sensorimotor physiology lab and neurophysiology lab by Christopher James Dakin and Dr Jean Sébastien Blouin. I was responsible for 70% of this project which included planning, collecting, analyzing and interpreting the results and writing the published manuscript. Dr Blouin collected the data, and reviewed the manuscript. This research was conducted under ethics provided under UBC’s Clinical Ethics Board, ethics number: H12-00689-A001 Check the first pages of this chapter to see footnotes with similar information.    vii Table of Contents  Abstract ..................................................................................................................................... ii Preface .......................................................................................................................................iv Table of Contents ................................................................................................................... vii List of Tables .............................................................................................................................ix List of Figures ............................................................................................................................ x Acknowledgments .................................................................................................................. xii Dedication .............................................................................................................................. xiii  1 General Introduction ........................................................................................................ 1 1.1 A Brief Overview of Vestibular Anatomy and Physiology ..................................... 2 1.2 Integration of Vestibular Information for Muscle Control ...................................... 3 1.3 Vestibular Stimulation ........................................................................................... 12 1.4 Goals of this Thesis ............................................................................................... 18  VOLUME 1 Exploration of the Stochastic Vestibular Stimulus ........................................ 23  2 Frequency-Specific Modulation of Vestibular-Evoked Sway Responses in Humans 2.1  Introduction ........................................................................................................... 24 2.2 Methods ................................................................................................................ 26 2.3 Results ................................................................................................................... 35 2.4 Discussion ............................................................................................................. 44 2.5 Conclusion ............................................................................................................ 49 2.6 Summary: Study 1 ................................................................................................ 50  3 Short and Medium Latency Muscle Responses Evoked by Electrical Vestibular Stimulation are a Composite of all Stimulus Frequencies 3.1 Introduction ........................................................................................................... 51 3.2 Methods ................................................................................................................ 54 3.3 Results ................................................................................................................... 59 3.4  Discussion ............................................................................................................. 67 3.5 Conclusion ............................................................................................................ 71 3.6 Summary: Study 2 ................................................................................................ 72  4 Time-varying Vestibulo - Myogenic Coupling Extracted using Stochastic Vestibular Stimulation 4.1 Introduction ........................................................................................................... 73 4.2 Methods ................................................................................................................ 75 4.3 Results ................................................................................................................... 82 4.4 Discussion ............................................................................................................. 90 4.5 Conclusions ........................................................................................................... 95 4.6 Summary: Study 3 ................................................................................................ 96     viii   VOLUME 2 Stochastic Vestibular Stimulation and its Application to Dynamic Vestibular Physiology ........................................................................................................................ 97  5 Vestibular Ex-afference Modulation during Locomotion and its Suppression with Increased Locomotor Velocity and Cadence 5.1 Introduction ........................................................................................................... 98 5.2 Methods .............................................................................................................. 100 5.3 Results ................................................................................................................. 104 5.4 Discussion ........................................................................................................... 110 5.5 Conclusion .......................................................................................................... 116 5.6 Summary: Study 4 .............................................................................................. 117  6 Dynamic Transformation of Vestibular Ex-afferent Signals by Head Rotation 6.1 Introduction ......................................................................................................... 118 6.2 Methods .............................................................................................................. 121 6.3 Results ................................................................................................................. 130 6.4 Discussion ........................................................................................................... 136 6.5 Conclusion ...................................................................................................... 14040  7 General Discussion and Conclusions 7.1 General Summary and Discussion ...................................................................... 142 7.2 Methodology - Advantages and Limitations ....................................................... 149 7.3 Future Directions ................................................................................................ 157 7.4 Conclusion .......................................................................................................... 159  Bibliography .......................................................................................................................... 160    ix List of Tables  Table 2.1 Early / first, middle / second and late EMG latencies for each of the      stochastic stimuli ...................................................................................................... 38   x List of Figures   Figure 1.1 Muscle and force responses to a 0 – 25 Hz stochastic vestibular stimulus ............ 17  Figure 2.1 Vestibular stimuli and corresponding power spectra .............................................. 28  Figure 2.2 Raw data and vestibular-sway pathway .................................................................. 30  Figure 2.3 EMG, force and trunk position cumulant density estimates elicited by        the 0-1 Hz, 0-2 Hz, 0-25 Hz, 1-25 Hz and 2-25 Hz stimuli ................................... 39  Figure 2.4 Root-Mean-Square of antero-posterior sway and peak antero-posterior sway correlation .............................................................................................................. 42  Figure 2.5 Gain plots along six stages of the SVS-sway relationship ..................................... 43  Figure 3.1 Stimuli, stimulus power spectra and 0 - 10 Hz, 10 - 25 Hz and 0 - 25 Hz                   cross-covariance functions ...................................................................................... 60  Figure 3.2 Raw left medial gastrocnemius EMG recording and corresponding EMG                    power spectrum ...................................................................................................... 62  Figure 3.3 Description of stimulus EMG entrainment ............................................................. 63  Figure 3.4 Cross covariance functions for the sinusoidal stimuli ............................................ 64  Figure 3.5 Average phase frequency functions for the 0 -20 Hz sinusoidal stimuli versus        the 0 - 20 Hz broad bandwidth control stimulus .................................................... 66  Figure 4.1 Coherence and cross-correlations between SVS and medial        gastrocnemius muscles .......................................................................................... 84  Figure 4.2 Decrease in time-dependent coherence related error associated with        increasing the total steps in the average ................................................................. 85  Figure 4.3 Time-dependent SVS-EMG coherence and gain for the right and left        medial gastrocnemius muscles ............................................................................... 87  Figure 4.4 Time-dependent cross-correlation between SVS-EMG for the right        and left medial gastrocnemius muscles ................................................................. 89  Figure 5.1 Averaged muscle activity for each of the three trial conditions............................ 105  Figure 5.2 Coherence plotted for each muscle in each walking condition ............................ 107   xi  Figure 5.3 Coherence and cross-correlation for four muscles over the 0.4m/s        52 steps/min condition ......................................................................................... 108  Figure 5.4 Change in coherence across trial conditions ......................................................... 111  Figure 6.1 Experimental set up, motion trajectory and raw data ........................................... 125  Figure 6.2 Schematic displaying an overview of the procedures and expected changes   in muscle response polarity .................................................................................. 128  Figure 6.3 Time-dependent cross-correlations for active, passive and fixed head   motion conditions ................................................................................................ 131  Figure 6.4 Gain and average EMG for the active and passive motion conditions ................. 133  Figure 6.5 EMG-SVS coherence for the active and passive motion conditions .................... 134  Figure 6.6 Change in the amplitude of the medium latency response at five head        angles ................................................................................................................... 137     xii Acknowledgments   Firstly, I would like to thank my family for providing the opportunity and environment to finish this thesis. To my parents, Jim and Marcia, and brother, Shawn, thank-you for the love and support through all these years without you this thesis would likely have never been started! To my wife, Laura, thank you for making these past years a joy, without your unconditional love this thesis would certainly never have been finished. To my daughter, Allie, thank you for making me smile after even the toughest days at work. To Ken and Merlayna thank you for all the support you have provided our family during this process.     Secondly, I would like to thank my mentor and friend Jean-Sébastien Blouin for providing an exceptional experience over the past five years. It has been an honour to have you as a supervisor. I would also like to thank Tim Inglis and Romeo Chua, if it wasn’t for the opportunities you provided and your guidance and mentorship over the years I would not have taken this extraordinary journey. I would also like to thank all my lab-mates (Greg, Dan, Marty, Billy, Tammy, Harrison, Sam & Brian) for all the help and advice they have provided over the years. Without good co-workers this experience could never have been as good as it was.  To everyone involved I am truly thankful for sharing this experience with you!   xiii Dedication To my family   1 1 General Introduction   Human movement, although seemingly effortless, is derived from an extremely complex interaction between incoming sensory information and retained information pertaining to past experience and the current state of the body. The sensory systems, which represent our brains’ interface to both the external world as well as our internal environment, allow us to estimate the current state of the body in the context of the world around us. Of these sensory systems four are important for movement (some more important than others): the auditory system, the visual system, the somatosensory system and the vestibular system. Each of these systems is normally attributed to a particular perceptual sensation, the auditory system with sound, the visual system with sight, the somatosensory system with touch and the vestibular system with a sense of equilibrium or dizziness. In the vestibular system (the focus of this thesis) the corresponding percept is usually not noticeable until after a prolonged acceleration, such as a spin on a merry go round for instance. As subtle as it might be, the vestibular system plays important roles in the stabilization of gaze, the estimation of directional heading and the activation of skeletal muscle. In animals much has been learned about the function and contribution of the vestibular system to overall behaviour. In humans on the other hand, it is difficult to assess or probe the function of the vestibular system and particularly so during dynamic tasks. Because of this difficulty human vestibular function is usually assessed during relatively static postural tasks and as a consequence little is known about the contribution of the vestibular system to dynamic movement and postures.  In contrast, the inertial properties of the vestibular receptors dictate that vestibular afference increases as body and head motion increases, suggesting the prevalence of vestibular information also increases with motion. As such, the vestibular system plays an essential role   2 in eye stability during head movements as well as important roles in spatial navigation and self-motion estimation. Arguably however, the most important function of the vestibular system might be its role in measuring accelerations of the head, and therefore the body, for the purpose of postural compensation and movement correction. Unfortunately, the tools available to test this component of human vestibular function are limited and not well suited for examining vestibular function in a dynamic context. Therefore the overarching theme of this thesis is the development of a new approach to examining vestibulo-motor interactions and the subsequent application of this approach to the investigation of vestibulo-muscular control during both static and dynamic tasks.  1.1 A Brief Overview of Vestibular Anatomy and Physiology  The vestibular system is a bilateral set of inertial accelerometers encased in a bony and membranous labyrinth located anterior and deep to the mastoid processes.  Each labyrinth is composed of three orthogonal angular motion receptive organs (termed the anterior, posterior and horizontal semicircular canals) and two orthogonal linear motion receptive otolith organs (termed the saccule and utricle), which are pitched up approximately 30° relative to earth horizontal when the eyes are level and the head is facing forward (Fitzpatrick & Day, 2004). Bilaterally the vestibular apparati are arranged redundantly such that left vestibular apparatus is roughly the mirror image of the right. Because of this mirrored arrangement the semicircular canals are organized into antagonistic functional pairs in which an angular movement resulting in an increase in the firing rate of the vestibular neurons on the right side is also represented by a decrease in firing rate in neurons on the left side.   3 Vestibular signals are carried by first order neurons from the receptors through the eighth cranial nerve to the vestibular nuclei located in the medulla of the brainstem where these signals are initially processed and rerouted. Depending on the source and destination the vestibular signal might enter one or more of the four vestibular nuclei. The superior vestibular nuclei relays and processes information related to eye stabilization whereas the medial vestibular nuclei processes information related to both neck and eye control. The inferior vestibular nucleus combines information from both the vestibular organs and the cerebellum, and projects to the vestibulospinal and reticulospinal tracts as well as to higher brain centers while the lateral vestibular nucleus also integrates vestibular and cerebellar information projecting to the vestibulospinal and reticulospinal tracts for the control of axial and appendicular muscle control (Kelly, 2000). The vestibulospinal and reticulospinal tracts are the major conduits for transmitting vestibular signals to the axial and appendicular muscles for influence over posture and movement control (Peterson & Azbug, 1975; Wilson & Peterson, 1978; Highstein et al., 1987).  1.2 Integration of Vestibular Information for Muscle Control   The vestibular system has long been known to contribute to postural control (Sherrington, 1898). Indeed many early researchers were interested in the vestibular systems influence over muscle tone. These early researchers were specifically concerned with fluctuations in decerebrate animal limb extensor rigidity (extensor tone) which was observed when the animals were rotated in pitch about the transverse plane. This rotation causes a change in the position of the head relative to gravity while preserving the position of the head   4 relative to the body. In these studies, extensor muscle activation was minimized when the animals head was upright and facing forward while activation increased for all other head positions, reaching a maximum when the head was 180° inverted in pitch from the head upright and forward position (Magnus, 1924). Because only head orientation relative to gravity was modified and these responses are abolished with excision of the vestibular labyrinths, the change in extensor tone to changes in head position was therefore attributed to the otolith organs of the vestibular system (Sherrington, 1910; Magnus, 1924). Similar results were also observed in several human case studies in which pathological de-cortication or localized cortical lesions resulted in similar modification of extensor tone to changes in head in space orientation (Magnus, 1924). In practice however, the vestibular system rarely functions under such pathologically isolated conditions and is normally modified by both sensory information from the various sensory systems as well as motor outflow 1  related to a movement’s generation.    Cancellation of Vestibular Afference during Voluntary Movements   In felines, sensory information from somatosensory receptors in the neck cancel vestibular afference 2  related to slow voluntary movements. Afferent responses to movement which might be derived from connective tissue in the joints of the cervical spine (McCouch et al., 1951) or the cervical musculature (muscle spindles) (Chan et al., 1987; Kasper et al.,  1  For the purposed of this thesis motor outflow will be defined as a general term to describe the motor component of movement production encompassing the efference copy, corollary discharge and the output of a forward model.  2  Afference refers to sensory signals from the body’s many sensory receptors whereas efference refers to motor signals used for the generation of movement.   5 1988) are inverted from those of the vestibular system (Lindsay et al., 1976; Ezure & Wilson, 1984).  This antagonistic relationship allows coinciding afferent signals from the neck and vestibular system to cancel during movements of less than 1 Hz (Lindsay et al., 1976; Ezure & Wilson, 1983, 1984). When head on body movement increases above 1 Hz however, neck and vestibular afference becomes mismatched, reducing the effectiveness of this cancellation (Ezure & Wilson, 1983). In maintaining posture, cancellation of movement induced sensory afference (termed re-afference) is beneficial as it acts to isolate sensory information related to errors or external disturbances (termed ex-afference) (von Holst, 1973). Once sensory information related to the error or disturbance is isolated, it can be used to aid in compensation to maintain the posture of the organism. Alternatively, a movement-related motor command or efferent signal might also be used to cancel vestibular signals. In 1950, under a concept described as “the re-afference principle”, von Holst and Mitterstaedt theorized that a bifurcation of the motor efferent signal, described as an efference copy, could be used to effectively cancel sensory re-afference and isolate the residual ex-afference. Around the same time Sperry (1950) coined the less specific term corollary discharge, to describe any motor related signal that influences sensory processing in a means analogous to the efference copy. This distinction in semantics is important because the corollary discharge allows for sensory cancellation by higher level motor centers and is not limited to efference at the motor neuron level. Since 1950, empirical support for “the re-afference principle” and corollary discharge has been observed at a variety of evolutionary levels: the electric fish (Bell, 1981), the swellfish (Sperry, 1950), the crayfish (Krasne & Bryan, 1973; Edwards et al., 1999) and in the rhesus and squirrel monkey (McCrea et al., 1999; Roy & Cullen, 2001). For review and summary see Crapse and Sommer (2008).   6 Several studies have provided support for the cancellation of vestibular re-afference by a corollary discharge. In squirrel monkeys similar behaviour is observed in central vestibular neurons. These neurons’ sensitivity to head rotation is absent or greatly attenuated during active versus passive head movements, suggesting they primarily encode externally applied head movements (Boyle et al., 1996; McCrea et al., 1996; McCrea et al., 1999). Second order vestibular neurons of the rhesus monkey also display this behaviour exhibiting modulation to both passive rotation of the head over the body and coincidental rotation of the head and body (Roy & Cullen, 2001). In contrast, modulation in these neuron’s firing rate is reduced in movements which require activation of the neck musculature to drive the motion. Based on these observations Roy & Cullen (2001) argued that since muscle activation was required to reduce vestibular signal modulation the reduction was likely caused by an efferent copy or, more abstractly, a corollary discharge of the neck motor command. In 2004 this hypothesis was updated when Roy and Cullen (2004) observed inhibition of vestibular signals in rhesus monkeys only when neck movement re-afference corresponded to the expected afferent return from the movement. They proposed the motor command might be used to predict the sensory consequences of a movement through what was described as an internal forward model (Ito, 1970; Miall & Wolpert, 1996; Wolpert & Kawato, 1998). In the forward model the sensory consequences of a movement are predicted based from the motor command and are then compared to the re-afference generated by the movement and if it matches to a reasonable degree vestibular re-afference is suppressed. Parallels between animal models and human function are often implied. However, due to the difficulty in assessing human vestibular function it can only be assumed that comparable cancellation of vestibular re-afference during voluntary movements occurs in   7 humans. After cancellation of re-afference the vestibular ex-afferent signal remains. How these ex-afferent signals are transformed to aid postural control in a dynamic environment is still relatively unknown.  Temporal and Spatial Transformation of Vestibulospinal Input   Assuming human vestibular re-afference is attenuated during voluntary movement the remaining ex-afferent signal therefore should reflect disturbances to posture generated by the surrounding world. In order to effectively compensate for these disturbances, vestibular ex- afference would need to be temporally and spatially organized into a profuse motor efference or vestibulo-muscle interaction to result in a corrective postural action. The timing of this vestibulo-muscle interaction would be important in order to influence muscle action during periods when the muscle can affect meaningful change, contributing to postural stability. As well, depending on the orientation of the head relative to the feet similar disturbances will be represented by different patterns of vestibular afferent discharge. To account for this, the vestibulo-muscle interactions must also be spatially transformed, taking into consideration the head-over-feet orientation, in order to effectively compensate for disturbances. For example, with the head facing forward if someone is pushed from behind the appropriate compensatory response would be to either plantar flex the feet, to bring the body back to a centered position, or to take a step forward. In contrast, if the head is turned over the left shoulder then a push towards the left on the right shoulder will register in the vestibular organs as a similar motion to the previous scenario however compensation is now different requiring a push to the right by the left leg, to center the body, or a step to the left.   8  Temporal Modulation  During locomotion vestibular input to muscles must be appropriately timed to constructively influence motor output shaping both the timing [In felines: (Russell & Zajac, 1979; Udo et al., 1982)] and amplitude [In guinea pigs: (Marlinsky, 1989, 1992) ]of muscle activity during locomotion. This modulation likely aids the shaping of vestibular ex-afferent input into appropriate compensatory responses when a disturbance to locomotion occurs. The timing of vestibulo-muscle interactions appears to be at least in part determined by activity arising from or passing through the cerebellum, as its removal abolishes gait specific modulation (Orlovsky, 1972). Presumably, inhibitory cerebellar purkinje cells exert their influence over central vestibular neurons in the lateral vestibular nucleus resulting in the phase dependent behaviour of vestibulo-muscle interactions during gait (Walberg & Jansen, 1961; Ito & Yoshida, 1964). In felines, collections of central vestibular neurons are modulated in accordance with fore and hindlimb extensor muscle activity while others are thought to modulate general extensor tone (Orlovsky, 1972; Matsuyama & Drew, 2000a, b). These multiple distinct groups of vestibular neurons exhibit increases in activity at different points in the step cycle (Orlovsky, 1972; Matsuyama & Drew, 2000a) and are likely timed to influence muscle action during periods when specific muscle groups can most effectively evoke change.  In humans, vestibular influence over skeletal muscles during locomotion also appears to be phase dependent. However, in contrast to the direct measurement of central vestibular neurons, as in the animal models, evidence of this phase dependency is derived from the application of an artificial vestibular stimulus and is therefore better described as an examination of the influence of vestibular ex-afference on muscle activation. Bent et al.   9 (2004) used electric (Galvanic) vestibular stimulation (discussed below) timed to three events in the gait cycle (heel contact, mid stance and toe off) to provide a gross estimate of the phase dependency of vestibular interaction with skeletal motor neurons. They observed that vestibular stimuli timed to heel contact elicited larger compensatory postural responses than stimuli timed to either mid stance or toe off suggesting the greatest vestibular influence is during the early stance phase of the step cycle. Subsequently, the phase dependencies of these vestibulo-muscle interactions were examined more closely and observed to depend not only on the phase of the step cycle but also the muscle group examined (Iles et al., 2007; Roskell et al., 2007). In the leg vestibulo-muscle interactions have been observed at different phases of the step cycle depending on the muscle observed (Iles et al., 2007; Roskell et al., 2007). Unfortunately, however, the stimulus format and analytical methods used in these studies were limited in their capacity to resolve the presence of responses, demonstrating interactions in only a few muscles (Roskell et al., 2007), and their capacity to document the time profile of the induced responses (Iles et al., 2007).  Importance for this Thesis: To date most studies examining the role of ex-afferent signals during human locomotion have mainly looked at the net postural effect of vestibular stimulation on the trajectory of gait (Fitzpatrick et al., 1999; Bent et al., 2000; Fitzpatrick et al., 2006). Attempts to understand the vestibulo-muscle interactions underlying these postural effects have been restricted due to limitations inherent in both the available stimuli and their accompanying analysis techniques.  One of the aims of this thesis is therefore to develop a stimulus format and analysis approach capable of extracting vestibular ex-afferent signals influence on muscle activity during locomotion. This development is essential to   10 understanding the role that vestibular ex-afferent signals play in stabilizing human locomotion.  Spatial Transformation The perspective or ‘reference frame’ from which vestibular signals operate under is important to ensure vestibular signals provide a beneficial contribution to motor control throughout the body. Vestibular signals are initially produced under a head centered reference frame. This is because vestibular organs are fixed in the skull, reporting only accelerations applied to the head irrespective of the position of the head relative to the body. These head referenced vestibular signals are useful for controlling eye movements as in the case of the vestibulo-ocular reflex (Grossman et al., 1989; Anastasopoulos et al., 1996). However to contribute to postural control of the body, head centered vestibular signals must be transformed to provide appropriate whole body postural responses for a given head over feet orientation. Spatial transformation of ex-afferent vestibular signals by head on body orientation has been observed in several human studies using electrical vestibular stimulation (described below) (Nashner & Wolfson, 1974; Lund & Broberg, 1983; Iles & Pisini, 1992; Britton et al., 1993). In these studies postural and muscular responses to the vestibular stimulus were observed to change their direction (postural responses) and polarity (muscular responses) depending on the static head on body orientation. In animal models both static and dynamic head on body vestibular transformations have been observed and are suggested to occur at a variety of locations.    11  In the vestibular nuclei of the brainstem, vestibular afference is modified by head on body orientation (Boyle & Pompeiano, 1981; Anastasopoulos & Mergner, 1982; Gdowski & McCrea, 1999, 2000) to change the vestibulo-motor reference frame. This transformation was inferred by the observation of neurons in the vestibular nuclei which are correlated more closely to trunk velocity than head velocity during a full body rotation (Gdowski & McCrea, 1999). However, the vestibular nuclei do not appear to perform this transformation in isolation. The modulatory influence of neck afferents on the vestibular signal in the lateral vestibular nuclei (the nucleus which primarily contributes to transmission of vestibular signals down the spinal cord) is partially dependent upon the function of the cerebellum. When part of the cerebellum is removed the modulatory influence of the neck on the vestibular signal is attenuated (Boyle & Pompeiano, 1981). In addition to regulating the modulatory role of the vestibular nuclei, the cerebellum and its deep nuclei also appear to have a prominent role in the spatial transformation of vestibular signals. Manzoni et al. (1998) found that the spatial transformation of vestibular responses in felines could be frozen through functional inactivation of the cerebellum’s vermal cortex. They also later found purkinje cells of the cerebellum whose response vector to vestibular stimulation was modified by head turn (Manzoni et al., 1999). The authors proposed that spatial transformation of these vestibular signals was mediated by a convergence of proprioceptive neck afferents onto vestibular neurons within the anterior cerebellar vermis. More recently, Shaikh et al. (2004) also identified neurons in the fastigial nucleus (a deep cerebellar nucleus) of a macaque monkey which appeared to code an intermediate head-body reference frame providing further support for the role of the cerebellum in mediating the spatial transformation of vestibular signals. In humans, however   12 it remains unclear as to the role vestibular nuclei and cerebellum play in the spatial transformation of vestibular signals. Recent exploration of the spatial transformation of vestibular responses in patients with a degenerative cerebellar disease (cerebellar ataxia) has mixed results. One study has shown a deterioration in the spatial transformation of vestibular signals in cerebellar patients (Kammermeier et al., 2009) while two unpublished studies have found no changes in the direction of vestibular evoked responses with static head turn over those of healthy controls (unpublished data (Bunn & Day, 2008; Dakin et al., 2009)).  Importance for this Thesis: In animal models, the spatial transformation of vestibular signals has been widely explored and localized to the vestibular nuclei and cerebellum. We know comparatively little about these same transformations in humans and particularly so during movement. Currently it is unclear whether spatial transformation requires active movement to occur and even more generally whether vestibular ex-afferent signals exert the same influence during head motion. The development of a stimulus format and analysis approach capable of easily extracting the spatial transformation of vestibular ex-afferent signals during movement is therefore key to understanding how these signals are transformed to contribute to directionally specific compensatory behaviour during motion.  1.3 Vestibular Stimulation   In humans, examination of central vestibular neuronal behaviour is difficult. Current technologies lack the spatial or temporal resolution to resolve natural vestibular neuronal behaviour in vivo. To compensate, electrical vestibular stimuli are typically used to probe   13 human vestibular function. While these stimuli are not natural stimuli, they provide a means to examine the role vestibular ex-afference plays in human perceptual, muscular and postural control. Electric vestibular stimulation is usually performed by applying a small amplitude short duration electric current to the mastoid processes behind each ear in a procedure called galvanic vestibular stimulation (GVS). The applied electric current passes through the skin (later described as a percutaneous application) and underlying vestibular organs and nerve resulting in changes in the firing rate of the vestibular afferents leaving the vestibular organs (Goldberg et al., 1984; Aw et al., 2008). By changing the direction of the electric current flow the firing rate of the vestibular afferents can be either increased (cathode) or decreased (anode) depending on the pole of the electrode (anode or cathode) nearest to the vestibular organ and nerve. As such, two electrode pole configurations are normally used when implementing electric vestibular stimulation: monopolar or bipolar. With a monopolar electrode configuration a single electrode pole is placed over one (unilateral configuration) or both (bilateral configuration) mastoid processes while the opposing poles are placed somewhere else on the body, usually the posterior neck or forehead, resulting in similar modulation in both vestibular nerves (Severac Cauquil et al., 1998; Severac Cauquil et al., 2000; Day et al., 2010). With a binaural bipolar electrode configuration opposite poles are placed over each of the mastoid processes causing opposite responses in nerves on each side of the head. Each of these different electrode configurations results in spatially differing perceptual, postural and muscular effects. GVS is usually performed using a square or step shaped stimulus profile. The discrete nature of these square shaped stimuli can be both beneficial, such as when examining the   14 body’s response to a single stimulus or detrimental, such as when trying to examine dynamic changes in vestibular function or achieve an interpretable muscle response in a short period of time. The influence of GVS over the body is typically very small requiring anywhere from 6 stimuli, to examine the postural responses (Day & Guerraz, 2007), to 2000 stimuli (Dakin et al., 2010a), in order to observe a reproducible single motor unit response. When one also considers the time required between stimuli (inter-stimulus interval, ISI), which depends on the duration of the body’s response to the stimuli, testing multiple conditions with a square wave stimulus can be prohibitively long and uncomfortable for the participant. Electric vestibular stimulation elicits responses in most skeletal muscles throughout the body active in the maintenance of balance. In particular, a single vestibular stimulus is believed to elicit at least two responses. These responses, when examined at both the postural (the forces generated by the body on the ground) and muscular (EMG) levels, are of opposite polarity or direction and exhibit latencies in the muscle of between 50 - 70 ms for the short latency response and between 100 - 120 ms for the medium latency response (Nashner & Wolfson, 1974; Iles & Pisini, 1992; Britton et al., 1993; Fitzpatrick et al., 1994; Fitzpatrick & Day, 2004; Dakin et al., 2007; Lee Son et al., 2008; Dakin et al., 2010b) (similar to what is observed in Figure 1.1). To date there is still some uncertainty as to the origins of these responses. The medium latency response appears to originate from the semicircular canals (Cathers et al., 2005) because the direction of this response corresponds to the response direction predicted by modelling the effect of GVS on semicircular canal afferents (Fitzpatrick & Day, 2004; Mian & Day, 2009). In contrast, the short latency response was initially proposed to originate from otolith afferents (Fitzpatrick & Day, 2004; Cathers et al., 2005) however a recent study found the direction of the short latency response to be   15 incongruent with its predicted direction casting doubt onto a proposed otolithic source for this response (Mian et al., 2010). Alternatively, these responses have also been proposed to derive from separate pathways in the spinal cord. Due to the timing (Britton et al., 1993) and frequency composition (Dakin et al., 2007) of the responses the early latency response has been attributed to the reticulospinal pathway and the medium latency response to the vestibulospinal pathway.  Unfortunately, direct recording of these tracts during GVS has not yet taken place and therefore the assignment of either response to a particular pathway is currently speculative. At the postural level, responses to GVS are predictable and have been modelled by Fitzpatrick and Day (2004). In their model Fitzpatrick and Day proposed electric vestibular stimulation non-specifically activates vestibular afferents leaving the vestibular organelles. This non-specific activation results in a vestibular error signal which is shaped by symmetrical imbalances in the structural geometry of the organelles and bilateral imbalances in the organelles afferents’ firing rate induced by the electrode configuration. On a larger scale, when one vestibular apparatus is excited and the other depressed, as occurs with a binaural bipolar electrode configuration, imbalances in symmetry within the organelles and differences in excitation between the two apparatuses are believed to result in two postural responses. One which exhibits an inter-aural acceleration towards the anode electrode due to the otoliths, though as previously stated this otolith component is much debated, and another resulting in a rotation of the body in the roll plane about an axis 18 degrees up from Reids plane 3  also toward the anode electrode due to the semicircular canals. As stated previously, postural responses to vestibular stimulation are in a head centered reference frame and  3  Reids plane is a theoretical plane that intersects the lower orbital rims (the bottom of the eye sockets) and the external auditory meatus (the external opening for the ear).   16 therefore change their direction depending on the position of the head relative to the body. Similarly, the polarity of the GVS evoked muscle response is also dependent upon the orientation of the head relative to the feet (Nashner & Wolfson, 1974; Lund & Broberg, 1983; Iles & Pisini, 1992; Britton et al., 1993). When the head is facing forward and GVS is provided in a binaural bipolar electrode configuration the evoked muscle response in each leg is opposite in polarity (Britton et al., 1993). In contrast, when the head is turned to the side, facing over one of the shoulders, the polarity of the GVS response in one leg inverts so that the polarity of the responses in both legs correspond. If the head is turned to look over the left shoulder with a binaural bipolar electrode configuration, anode on the right, for example, the muscle responses in both lower legs (soleus or gastrocnemius) would exhibit a positive short latency peak and a negative medium latency peak with the net sway response directed forward. To date, electric vestibular stimulation has proven a useful tool for probing the body’s response to ex-afferent vestibular perturbations. However the time required generating visible force and muscle responses combined with instability caused by the stimulus itself limit the usefulness of the step stimulus in certain contexts. In contrast, the recent use of continuous random stimuli has shown promise in overcoming some of the limitations of discrete stimuli and might provide an effective substitute for the assessment of dynamic vestibular function. Continuous electric vestibular stimuli randomly varying in both amplitude and frequency (stochastic vestibular stimulation: SVS) have been widely used to simulate postural control (Fitzpatrick et al., 1996; Pavlik et al., 1999; Scinicariello et al., 2001; MacDougall et al., 2006; Moore et al., 2006). The effectiveness of these stimuli as tools to investigate vestibular function in a manner similar to GVS has remained largely unclear. Recently I   17 demonstrated the applicability of broad bandwidth (0 - 50 Hz) stochastic vestibular stimulation (SVS) in generating vestibular evoked muscle and force responses (Dakin et al., 2007). This preliminary study demonstrated that muscle responses similar to those evoked by GVS could be evoked in as little as three minutes using the stochastic stimulus. As well, the random nature of the stimulus provided the opportunity to extend analysis beyond the standard time varying measures, allowing spectral analysis of the evoked response. Consequently, the phase relationship between stimulus frequencies reaching the muscle offered additional support for the independence of the two components of the vestibular- evoked muscle response. The slope of the phase relationship between the 10 - 20 Hz frequencies suggested these frequencies contribute primarily to the short latency response whereas the slope of the phase relationship between the 0 - 10 Hz frequencies indicated this bandwidth primarily contributes to the medium latency response. Aside from this study the usefulness of this type of stimulus in assessing vestibular function is not yet known. However, two fundamental Figure 1.1 Muscle and force responses to a 0 - 25 Hz stochastic vestibular stimulus. A. The two peaks of the vestibular evoked muscle response from the left medial gastrocnemius. B. The analogous two peaks as represented as forces at the feet.   18 characteristics of SVS signify its potential usefulness in advancing understanding of vestibular function:  a)  SVS has the potential to serve as a frequency specific stimulus allowing investigation of vestibular attributes that are frequency dependent.  b) The continuous nature of the stimulus might allow the resolution of time varying dynamic processes with a comparatively short testing time.  1.4 Goals of this Thesis   The primary aim of this thesis is to tie together an innovation in the technique of electric vestibular stimulation through the development of the stochastic vestibular stimulus, and use this new technique to advance current understanding of dynamic human vestibular function. The first volume of this thesis has two primary methodological points of focus: the first is an investigation into the possibility to customize the stimulus bandwidth and the second is to develop an appropriate methodological framework for the extraction of dynamic vestibular behaviour. Once the appropriate methodological framework for extracting dynamic vestibular behaviour was developed, I used this approach to investigate vestibular influence on muscle activation during locomotion and head rotation. The first methodological focus is to investigate modifying the frequency content of the stimulus bandwidth to maximize or minimize a specific response (Chapters 2 and 3). Previously I observed that specific stimulus bandwidths seem to contribute independently to   19 the two components of the vestibular-evoked muscle response. This observation suggests that specific frequencies within the stimulus bandwidth might be more closely associated with a particular muscular or postural behaviour and that these behaviours could be isolated or attenuated by providing or omitting only these frequencies. An example of this is stimulus induced sway. One of the primary limitations of galvanic stimuli is the postural sway it induces. Since these perturbations are mainly associated with low frequency stimuli the removal of these frequencies from a broad bandwidth stimulus could possibly reduce the postural consequence but retain the higher frequency components of the muscle response. A reduction in the postural consequence of vestibular stimuli would be beneficial for contexts such as clinical work with patients in which stability is a concern or in studies in which excessive postural motion interferes with the experimental objectives.  Modifications in the stimulus bandwidth might also be used to isolate the two components of the biphasic muscle response in addition to reducing the postural sway associated with the stimulus. My previous work has shown that the 0 - 10 Hz bandwidth is closely associated with the medium latency response component and the 10 - 20 Hz bandwidth associated with the short latency responses component (Dakin et al., 2007). By providing a stimulus limited to these bandwidths either the short (10 - 20 Hz) or medium (0 - 10Hz) latency response might be preferentially elicited. This result is important because if the two responses components indeed originate from independent sources (otoliths or semicircular canals) or travel independent paths to the motor neuron (vestibulo or reticulo-spinal pathways) then these customized bandwidths would provide an ideal tool for clinicians to test these pathways or sensory organs, or for researchers to preferentially investigate them. The second methodological focus (Chapter 4) is on the applicability of the stimulus for   20 extracting time varying or dynamic vestibular influence on muscle activation. Recently the stochastic stimulus was shown to provide similar muscle responses to galvanic stimuli but in much less time (Dakin et al., 2007). This ability to quickly produce results coupled with the stimulus’s continuous format should allow a sample by sample correlation between the stimulus and a dependent variable of interest (EMG, forces or sway). By choosing a motion which is repeatable the stimulus-response correlations can then be averaged over many trials producing a much higher resolution indication of the time varying relationship between the stimulus and response than is available using a discrete galvanic probe. This combination of stimulus-analysis techniques should provide a valuable tool for identifying the dynamic coupling between vestibular ex-afferent like signals and their associated responses. Once this new approach to examining the dynamic vestibular-response coupling was validated it provided the foundation to investigate both the phasic modulation of vestibular coupling during locomotion and the spatial transformation of vestibular signals during head rotation. These two studies will comprise the second volume of this thesis which aims to investigate dynamic vestibular behaviour. As described previously vestibular signals are believed to contribute to locomotion in a phase dependent manner. In humans this phase dependency has proven difficult to unveil, given the time constraints of current technical approaches. Presumably, phasic vestibular coupling should be present in nearly all muscles active in stabilizing upright posture during locomotion. However current research has only observed phase dependency in muscles acting around the ankle (Iles et al., 2006) and in the general whole body postural response to the stimulus (Bent et al., 2004). In addition to phase dependency, vestibular influence during locomotion appears to be suppressed with higher walking speeds. Using the technical advances I derived in the final methodological chapter   21 (Chapter 4) I investigated the phase dependent and speed dependent modification of vestibular influence on muscle activation as this information provided key insight into how vestibular ex-afferent signals contribute to the maintenance of upright posture during locomotion (Chapter 5). The final study (Chapter 6) in this thesis also used the advances made in Chapter 4 to investigate dynamic vestibular-response coupling during head rotation. In humans, investigation of the relationship between head position relative to the body and the spatial orientation of vestibular induced responses is usually performed in a static environment. How this relationship behaves during motion is unclear. The combination of motion related signals with the ex-afferent like stimulus might negatively influence compensation to vestibular signals, attenuating vestibular-induced responses during head motion. In addition, it is also unclear whether vestibular signals are continuously transformed during motion. In this final study (Chapter 6), I investigated the dynamic transmission and spatial transformation of vestibular signals during head rotation. This study clarified how the body compensated for a vestibular error while the head is in motion.  Overall this thesis contains five studies with the following specific goals: In Volume One I a) determined if removal of low frequencies from the 0 - 25 Hz stochastic stimulus will attenuate the balance perturbation resulting from the stimulus (Chapter 2); b) determined if SVS can be tuned in its bandwidth to isolate either the short or medium latency component of vestibular induced responses (Chapter 3); and c) determined if SVS is amenable to extracting the time varying modulation of vestibular influence over postural and muscular activity through the application of a time-frequency analysis protocol (Chapter 4). The general goal of Volume Two was to apply the advancements made in the first three studies towards filling in   22 the gaps in knowledge presented in the general introduction with the specific goals of a) describing the phase dependent contribution of vestibular ex-afferent signals to the muscular control of locomotion (Chapter 5) and b) determining if head movement reduces our ability to compensate for vestibular ex-afferent signals (Chapter 6). These studies provided the first in depth examination of the role of vestibular ex-afferent signals in contributing to muscle activation during motion in humans.   23 VOLUME 1 Exploration of the Stochastic Vestibular Stimulus  Volume Synopsis  This volume aimed to advance the technique of stochastic vestibular stimulation (SVS) through three studies. The first study examined whether removal of the low frequency content of the stimulus reduced the postural sway associated with vestibular stimulation. Removal of the prominent sway response to vestibular stimulation could be beneficial for testing patients with balance deficits and limit the effect sway induced feedback has on the muscle responses. The second study examined whether narrow bandwidth SVS can be used to isolate the components of the vestibular evoked muscle response. Additionally, sinusoidal GVS was used as a comparison to identify a) if these two GVS response components are independent and b) if the muscle response to SVS is a linear sum of each of the contributing frequencies. The third study introduced the use of time frequency analysis to determine if time varying vestibular responses can be extracted using the stochastic stimulus. SVS was then used to quantify the time varying characteristics of vestibulo-muscle interactions in the medial gastrocnemius during gait. The study specific hypotheses are presented following each chapters specific introduction.   24 2 Frequency-Specific Modulation of Vestibular-Evoked Sway Responses in Humans 4   2.1  Introduction Galvanic vestibular stimulation (GVS) has long been used as a means to probe vestibular function (for review see Fitzpatrick and Day 2004). In humans maintaining standing balance, GVS provides an isolated vestibular error signal allowing the study of the resulting whole-body movements and myogenic responses in muscles involved in the control of balance (Nashner & Wolfson, 1974; Lund & Broberg, 1983; Iles & Pisini, 1992; Britton et al., 1993; Fitzpatrick et al., 1994; Lee Son et al., 2008). Recently, I have showed that stochastic vestibular stimulation (SVS) over a 0 - 50 Hz bandwidth elicits vestibular-evoked balance and muscle responses similar to those observed using GVS (Dakin et al., 2007). SVS- electromyographic (EMG) coupling was observed over the 0 - 20 Hz bandwidth coinciding with previous estimates of the dynamic range of vestibular function (Grossman et al., 1988; Armand & Minor, 2001; Huterer & Cullen, 2002). Whole-body responses to vestibular stimulation, however, appear to follow the vestibular stimulus when frequencies less than 5 Hz are provided (Lund & Broberg, 1983; Fitzpatrick et al., 1996; Pavlik et al., 1999; Latt et al., 2003; MacDougall et al., 2006; Moore et al., 2006), with the largest responses elicited when the frequency content of the stochastic vestibular stimulus is below 2 Hz (Fitzpatrick et al., 1996; Pavlik et al., 1999).  4  A version of Chapter 2 has been published: Dakin CJ, Luu BL, van den Doel K, Inglis JT & Blouin JS. (2010). Frequency-specific modulation of vestibular-evoked sway responses in humans. J Neurophysiol 103, 1048- 1056.     25 Ongoing control of upright balance is thought to occur through low-frequency lower- limb muscle activity (~2.5 Hz) which is mechanically filtered to produce an even lower frequency body sway (< 1 Hz) (Bawa & Stein, 1976; Fitzpatrick et al., 1996; Latt et al., 2003; Loram et al., 2005). As the activation frequency of lower-limb muscle activity increases, its amplitude must also increase to maintain impulse magnitude (and therefore sway amplitude) due to the inertial load of the body. However, muscle-sway gain decreases with increasing frequency, such that muscle activity at 5 Hz evokes body sway which is 100 times smaller than that evoked by similar muscle activity amplitude at 0.5 Hz (Fitzpatrick et al., 1996). Hence, higher frequency muscle activation is mechanically filtered out in its transfer to lower- limb moment production and resulting sway. SVS, in contrast, evokes muscle activity at frequencies from 0 - 20 Hz. By removing the lower frequency content of the SVS stimulus (< 2 Hz), I predict that the remaining higher frequency signal is mechanically filtered out; resulting in attenuation of the whole body sway response while maintaining the biphasic lower-limb muscle response.  The primary aim of this study was to determine if stochastic vestibular stimuli excluding low-frequency bandwidths could elicit vestibular-evoked lower-limb biphasic myogenic potentials with a reduction in associated sway responses. This could be particularly beneficial for exposing basic physiological phenomena that would otherwise be masked by the balance response, or for studying vestibular responses in patients with balance disorders who are already unstable. To unveil frequencies specific to the SVS evoked sway responses participants were exposed to five SVS stimuli: two meant to maximally elicit SVS related whole body sway (0 - 1 Hz and 0 - 2 Hz) and three meant to dissociate sway from SVS evoked muscle responses (0 - 25 Hz, 1 - 25 Hz and 2 - 25 Hz). Two control trials were also   26 performed to compare sway associated with SVS to sway observed with random 1 mA GVS pulses and free standing (no vestibular stimulation). Hypothesis: I hypothesized that vestibular stimuli with low frequencies removed (stimulus bandwidths between 1 - 25 Hz and 2 - 25 Hz) would evoke biphasic muscle responses with minimal whole-body sway due to mechanical filtering of the SVS-evoked higher frequency muscle responses.  2.2 Methods Subjects  Twelve healthy subjects (9 male, 3 female;  mass 70 ± 10 Kg and height 1.72  ± 0.10 m ( X ± SD)) between the ages of 21 and 33 years, with no known history of neurological disease or injury participated in this study. The experimental protocol was explained to each subject and their written, informed consent was obtained. All procedures used in this study conformed to the standards of the Declaration of Helsinki and were approved by the University of British Columbia’s clinical research ethics board.  Stimulus Stochastic vestibular stimulation (SVS) and galvanic vestibular stimulation (GVS) were delivered using a bipolar binaural electrode configuration with carbon rubber electrodes (9 cm 2 ), coated with Spectra 360 electrode gel (Parker Laboratories, Fairfield, USA), secured over the mastoid processes with an elastic headband. The stimuli were created on a PC computer using Labview software (National Instruments, Austin, USA), and delivered as an analog signal via a data acquisition board (PXI-6289, National Instruments, Austin, USA) to   27 an isolated constant-current unit (Model 2200 Analog Stimulus Isolator: AM Systems, Carlsborg, WA). The stochastic signals (Figure 2.1A) lasted 133 seconds and were designed to provide similar power amplitude to each frequency component within and across all stimuli (Figure 2.1B). This resulted in different Root Mean Square (RMS) amplitudes for the different bandwidth stimuli: 0.20 mA (0 - 1 Hz), 0.29 mA (0 - 2 Hz), 0.98 mA (0 - 25 Hz), 0.96 mA (1 - 25 Hz), 0.93 mA (2 - 25 Hz). On additional trials, 1 second 1 mA GVS pulses were provided, with each trial consisting of 5 anode right (cathode left) and 5 anode left (cathode right) pulses presented randomly for a total of 20 pulses. Galvanic pulses were delivered with a variable inter-stimulus interval of 10 - 15 seconds.  Test Procedures  Participants were required to stand on a force plate (Bertec 4060-80: Bertec Corp., Columbus, OH) with their feet 2-3 cm apart (as measured at the medial malleoli). The participants were instructed to stand relaxed with their eyes closed, arms by their sides and their head turned to the left with Reid’s plane tilted nose up 18° from parallel to the floor. This head position maximizes the postural response to vestibular stimulation in the antero-posterior direction (Cathers et al., 2005; Day & Fitzpatrick, 2005), aligning the postural response to the line of action of the soleus, gastrocnemius and tibialis anterior muscles. Head pitch and yaw and trunk sway, measured at the level of the sternal notch, were monitored online using a 3D motion tracking system (TrakStar: Ascension Technology Corp., Burlington, VT) to control for changes in head position (Cathers et al., 2005). Monitoring of head pitch limited RMS variability around the desired position to 0.33 degrees Participants were exposed to five stochastic stimuli 0 - 1 Hz, 0 - 2 Hz, 0 - 25 Hz, 1 -   28 25 Hz and 2 - 25 Hz to determine if prolonged sway responses to vestibular stimulation are primarily associated with vestibular stimulus frequencies below 2 Hz. Two trials were designed to induce sway (0 - 1 Hz and 0 - 2 Hz) and three trials examine the possible dissociation between muscle and sway responses (0 - 25 Hz, 1 - 25 Hz and 2 - 25 Hz). Each subject also performed a single free stance trial and two GVS trials (see stimulation parameters above) as controls to compare mean removed RMS trunk sway amplitude elicited by the SVS trials. Rest periods were provided at the request of the participant to avoid any sign of fatigue.   Figure 2.1 Vestibular stimuli and corresponding power spectra. A. Vestibular stimuli for each bandwidth of stimulation. B. Log-based power spectra for each of the bandwidths of stimulation. Signal power for each bandwidth of stimulation is localized in the bandwidth of interest.    29 Electromyography and Signal Analysis EMG was collected for the soleus, medial gastrocnemius and tibialis anterior of the right leg (Figure 2.2A). EMG was amplified (×2000; NeuroLog, Digitimer, Hertfordshire, England) and bandpass filtered (10-1000 Hz). EMG, vestibular stimuli and force plate data were sampled and digitized at a rate of 5000 Hz with a standard data acquisition board (PXI- 6289, National Instruments, Austin, USA) using a custom Labview software program. Horizontal forces acting on the subject and antero-posterior / medio-lateral trunk displacement were used to describe the balance and sway behaviour of the subjects (Figure 2.2A). Sway was recorded with the TrakStar at 240 Hz, low-pass filtered with zero phase shift at 20 Hz and interpolated up to 5000 Hz for correlation analysis with the vestibular stimuli. Cumulant density estimates were used to represent the time-domain relationship between vestibular stimulation and muscle activity (Halliday et al., 1995). The cumulant density estimate provides similar temporal and spatial characteristics to the muscle responses observed with trigger averaged GVS (Dakin et al., 2007). A consequence of this technique is that the cumulant density estimate between two measured signals provides a correlation-like measure (equivalent to the cross-covariance) and therefore must be interpreted as an associative rather than causal relationship. In this study, cumulant density estimates were determined for a random, controlled input signal (vestibular stimuli) and measured physiological signals (EMG, force and sway), therefore responses correlated to the vestibular stimuli must have been evoked by the vestibular stimuli. This is supported by the phase relationship between SVS and EMG responses exhibiting linear slopes consistent with GVS- evoked responses (Dakin et al., 2007). SVS-EMG, SVS-force and SVS-Sway cumulant density estimates are accordingly referred to as related responses that hold no physical values   30 (e.g. N or m). It should be noted however, that the physiological signals likely contain SVS- evoked contributions from both an open loop process and the associated recurrent feedback.   Figure 2.2 Raw data and vestibular-sway pathway. A. Ten seconds of raw data displaying muscle EMG, antero-posterior directed force, antero- posterior directed trunk sway and SVS stimulus. B. Schematic describing the vestibular-sway pathway. AP: antero-posterior; r-Tib: right tibialis anterior, r-mGas: right medial gastrocnemius, r-Sol: right soleus, SSF: somato-sensory feedback, EMG: electromyography.  Gain functions were used to provide an estimate of the SVS transfer function at six subsections in the SVS-sway pathway (Figure 2.2B) (Rosenberg et al., 1989; Halliday et al., 1995; Fitzpatrick et al., 1996). Neural control of sway is limited to muscle excitation therefore   31 changes in SVS-evoked movement patterns resulting after muscle activation should be related to mechanical factors. Gains were therefore calculated along the SVS–sway pathway for the 0 - 25 Hz stimulus bandwidth (SVS-EMG, SVS-AP moment, SVS-AP sway, EMG-AP moment, EMG-AP sway and AP moment-AP sway) to 1) determine if reductions in SVS to sway signal bandwidth are due to mechanical factors and to 2) localize sources of mechanical filtering of the signal (Figure 2.2B). Gains were calculated using two methods: the first method was to directly calculate the SVS-EMG, SVS-AP moment and SVS-AP Sway gains (Figure 5A). The direct approach has one central caveat in that it assumes an open loop pathway and neglects the effects of feedback. The resulting closed loop transfer functions therefore must be assumed to include the frequency characteristics of associated feedback. To separate the effects of feedback I used a second method to calculate the gains: the joint input-output approach [described by van der Kooij et al. (2005); Figure 5B]. By dividing two closed loop transfer functions, a mathematical cancellation of feedback occurs allowing estimation of the inferred open loop transfer function (Fitzpatrick et al., 1996; Kiemal et al., 2008) for example, to identify the inferred open loop EMG-sway transfer function the SVS-sway closed loop transfer function is divided by the SVS-EMG closed loop transfer function.  This approach was used to estimate the inferred open loop transfer functions between EMG-AP moment, EMG-AP sway and AP moment-AP sway (Figure 2.5B). Coherence, cumulant density and direct gain estimates were derived using a Matlab script based on the methods described by Rosenberg and colleagues (1989). Coherence estimates were calculated for each participant using segments of 2 12  data points and evaluated for significance using 95 percent confidence limits based on the number of disjoint segments. Gains were only calculated at frequencies exhibiting significant coherence since gain only has   32 meaning when a relationship exists between two comparison signals (Halliday et al., 1995). For the cumulant density and gain estimates, digitized EMG, force plate and sway data recorded during the stochastic SVS trials were time-locked to SVS onset and cut to provide 10 disjoint segments of 2 16  data points within each participant for each condition. Participant data were then averaged across all participants to provide grouped means. EMG data were full- wave rectified and cumulant density and gain functions between SVS-EMG, SVS-Horizontal Forces and SVS-Sway signals were estimated for each trial condition within and across all participants. SVS-EMG, SVS-Horizontal Forces, and SVS-Sway cumulant densities and gain estimates were analyzed with resolutions of 0.076 Hz (13.1 sec/segment) to identify the lower frequency components of both functions. Amplitudes of the cumulant density functions were normalized by the product of the vector norms of the input (SVS) and the output (EMG, force or sway) signals:  where, fxy is the cross spectrum, λj are the Fourier frequencies, T is the number of points in the Fourier transform, u is the lag, i is the square root of -1  and x and y are the input and output data series, respectively. This normalization procedure transforms the cumulant density values into standard coefficients of correlation (r values bounded between -1 and +1), providing meaningful units of magnitude. A consequence of the normalization process is that the amplitudes of the cumulant density estimates are scaled by the RMS of the contributing signals thereby changing the cumulant density estimates relative magnitudes. This normalization procedure primarily influenced the relative magnitudes of the 0 - 1 and 0 - 2 Hz cumulant density estimates, while having little influence on the 0 - 25, 1 - 25 and 2 - 25 Hz   33 cumulant density estimates and did not change whether or not a stimulus exhibited significant EMG, force or sway responses on a subject per subject basis. Confidence intervals were calculated for individual subject cumulant density functions based on the methods described by Rosenberg and colleagues (1989). Cumulant density functions were evaluated on a subject-by-subject basis to determine the presence of significant responses (i.e. when the values exceeded the computed 0.95 confidence intervals) and then averaged across subjects. The final averages are presented without confidence intervals. By convention, anode right currents are represented as a positive vestibular signal. Hence, a positive cumulant density function indicates that anode right currents induced muscle facilitation or anterior directed forces or sway.  Multi-phasic muscle and force responses are observed following GVS (Fitzpatrick et al., 1994) and broad bandwidth SVS (Dakin et al., 2007), however only the early and middle latency vestibular-evoked force or muscle correlations are described due to their potential physiological relevance (Britton et al., 1993; Cathers et al., 2005). In contrast, narrow bandwidth (0-1 and 0-2 Hz) muscle and force correlations are described as having first and second peaks, rather than early and middle latency responses, as they also exhibit multi-phasic patterns but with different spatial and temporal characteristics than the broad bandwidth stimuli. Local maxima and minima for the correlated trunk sway, force and muscle responses were extracted from the individual subjects’ cumulant density function only when they reached significance. Mean removed root-mean square (RMS) values for AP trunk sway were also calculated to provide estimates of AP trunk sway variability around the mean value during the vestibular stimulation and control trials.    34 Data Reduction Mean-removed RMS values for AP trunk sway measured for each SVS frequency bandwidth, square-wave GVS and control trials were compared using a one-way repeated measure ANOVA. The early and middle latency components of the EMG and force responses were determined using the peak correlation (or trough) observed in the time cumulant densities and compared between the 0 - 25 Hz, 1 - 25 Hz and 2 - 25 Hz trials using a one way repeated measures ANOVA; while first and second peaks of the EMG and force in the 0 - 1 Hz and 0 - 2 Hz trials were compared using paired t-tests. Direct comparisons of the 0 - 1 Hz and 0 - 2 Hz trials with the 0 - 25 Hz, 1 - 25 Hz and 2 - 25Hz trials were not performed because: a) the narrow bandwidth of the 0 - 1 Hz trial limited the random nature of the wave leading to correlations prior to zero time lag, similarly to what is observed when the input signal is a sine wave, and b) the 0 - 1 Hz and 0 - 2 Hz stimuli resulted in biphasic waveforms which did not resemble the spatial and temporal characteristics of the 0 - 25 Hz, 1 - 25 Hz and 2 - 25Hz trials. Magnitude comparisons were not performed in the tibialis anterior because significant correlations in SVS-muscle activity were not observed for all participants (see table 1). Decomposition of the main effects were performed using Fisher’s least significant difference (LSD) tests due to my a priori hypotheses, namely that sway would be reduced when vestibular stimuli excluded frequencies below 1 or 2 Hz. Statistical significance was set at p<0.05 Data are presented as means ± SD.       35 2.3 Results Electromyographic Responses    Lower-limb muscle activity was correlated with each of the 0 - 25 Hz, 1 - 25 Hz and 2 - 25 Hz vestibular stimuli (Figure 2.3). The plantar flexor responses exhibited early latency positive and middle latency negative going peaks while the right tibialis anterior exhibited corresponding peaks of opposite polarity (see Table 1 for latencies). The timing and polarity of the early and middle latency muscle responses were comparable across each of the 0 - 25 Hz, 1 - 25 Hz and 2 - 25 Hz bandwidths (Figure 2.3), while the amplitude of the early and middle latency peaks varied depending on the bandwidth of the stimulus provided. In the soleus muscle, the peak correlation of the early latency response for the 2 - 25 Hz stimulus (0.073 ± 0.05) was 26% and 35% larger than for the 1 - 25 Hz and 0 - 25 Hz stimuli, respectively (F(2,11) = 37.5 , p<0.05; Fisher LSD multiple p<0.05); whereas in the medial gastrocnemius, the peak correlation of the early latency response for the 2 - 25 Hz stimulus (0.105 ± 0.06) was 38 % and 64 % larger than for the 1 - 25 Hz and 0 - 25 Hz stimuli, respectively (F(2,11) = 63.4, p<0.05; Fisher LSD multiple p<0.05). There was no significant difference in the magnitude of early latency muscle responses for the 0 - 25 Hz and 1 - 25 Hz stimuli (multiple p>0.05; 0.054 ± 0.03 vs 0.058 ± 0.03, r-Sol; 0.064 ± 0.03 vs 0.076 ± 0.03, r- mGas). The medium latency correlation, however, was not consistent across muscles showing similar magnitude between stimuli in the soleus (F(2,11) = 36.9, p>0.05; -0.084 ± 0.06 (2 - 25 Hz), -0.090 ± 0.05 (1 - 25 Hz), -0.092 ± 0.05 (0 - 25 Hz)), but smaller magnitude for the 2 - 25 Hz trial than both the 0 - 25 Hz and 1 - 25 Hz trials in the medial gastrocnemius (F(2,11) = 46.2, p<0.05; -0.100 ± 0.07 (2 - 25 Hz), -0.121 ± 0.06 (1 - 25 Hz), -0.121 ± 0.05 (0 - 25 Hz)). The muscle responses correlated with SVS at bandwidths of 0 - 1 Hz and 0 - 2 Hz did   36 not resemble the typical motor responses triggered by GVS or 0 - 25 Hz SVS. The low frequency SVS signals were associated with a biphasic correlation in lower-limb muscles, exhibiting a first peak starting before or at the zero second time lag due to the narrow bandwidth of the 0 - 1 Hz and 0 - 2 Hz SVS (Figure 2.3). In the plantar flexors, the related muscle correlation exhibited a negative followed by a positive going waveform, while the opposite was observed in the right tibialis anterior (Figure 2.3). The correlation for the first muscle peak to the 0-2 Hz bandwidth was greater than the 0-1 Hz bandwidth in both the soleus (T(11)=4.47, p<0.05; -0.020 ± 0.01 vs -0.009 ± 0.08) and medial gastrocnemius (T(11)=2.87, p<0.05; -0.056 ± 0.026 vs -0.033 ± 0.018). Magnitudes for the second peak of the biphasic correlation were similar for the 0-1 Hz and 0-2 Hz bandwidths in the plantar flexors (multiple p>0.05; 0.031 ± 0.014 vs 0.023 ± 0.015, r-Sol; 0.068 ± 0.039 vs 0.066 ± 0.033, r- mGas).  Force Responses All of the SVS stimuli evoked significant force responses as illustrated by the correlations between SVS and Force (Figure 2.3). The force correlations to the 0 - 25 Hz, 1 - 25 Hz and 2 - 25 Hz stimuli bandwidths were characterized by an initial posterior (negative) peak followed by an anterior (positive) peak in the direction of the anode (see Table 1 for latencies). The magnitude of the early latency correlation for the horizontal anterior-posterior (AP) force was 77% and 54% larger for the 1 - 25 Hz and 2 - 25 Hz stimuli, respectively, when compared to the 0 - 25 Hz stimulus (-0.057 ± 0.03) (F(2,11) = 57.6, p<0.05; Fisher LSD multiple p<0.05), but was similar in magnitude between the 1 - 25 Hz and 2 - 25 Hz stimuli (- 0.101 ± 0.04 and -0.088 ± 0.05). In contrast, the magnitude of the middle latency correlation   37 for AP force grew larger as the low frequency content of the stimuli increased. The 2 - 25 Hz trial exhibited the smallest middle latency peak correlation response (0.055 ± 0.05), followed by a larger response for the 1 - 25 Hz trial (0.114 ± 0.05) with the largest response recorded in the 0 - 25 Hz trial (0.154 ± 0.06) (F(2,11) = 58.8 , p<0.05; Fisher LSD multiple p<0.05). Stochastic stimuli with bandwidths of 0 - 1 Hz and 0 - 2 Hz were associated with biphasic force correlations (see Table 1 for latencies). The first anterior directed peak correlation was similar (T(11) = 1.89 p>0.05) in magnitude for 0 - 1 Hz (0.310 ± 0.058) and 0 - 2 Hz stimuli (0.347 ± 0.079). The second posterior directed peak correlation was 35% larger (T(11) = 3.23 p<0.05) for the 0 - 1 Hz (0.335 ± 0.074) stimulus than the 0 - 2 Hz (0.249 ± 0.078) stimulus.     38    Table 2.1 Early / first, middle / second and late EMG latencies (ms) for each of the stochastic stimuli. Table 2.1 displays EMG peak correlation latencies and peak antero-posterior force correlation latencies, in msec, for each of the stochastic stimuli. The 0 - 1 Hz and 0 - 2 Hz trials exhibited biphasic patterns, which are labelled the first and second responses whereas the 0 - 25 Hz, 1 - 25 Hz and 2 - 25Hz trials exhibited tri-phasic patterns which are labelled as the early, middle and late latency responses. Responses in the right tibialis anterior were unreliable in some subjects. Values are presented as means with standard deviations and the number of subjects contributing to the mean is presented to the right (n). r-Tib: right tibialis anterior, r-mGas:   39  Figure 2.3 EMG, force and trunk position cumulant density estimates elicited by the 0 - 1 Hz, 0 - 2 Hz, 0 - 25 Hz, 1 - 25 Hz and 2 - 25 Hz stimuli. A. Pooled (n = 12) muscle responses for the right medial gastrocnemius and (n = 7) tibialis anterior. Biphasic muscle correlations were observed for the 0 - 1 Hz and 0 - 2 Hz stimuli, while tri-phasic muscle responses are observed for the 0 - 25 Hz, 1 - 25 Hz and 2 - 25 Hz stimuli.  B. Antero-posterior force correlations to the five stimulus bandwidths. C. Trunk position correlations showed SVS related sway correlation is attenuated as the low frequency content of the stimulus is removed. The 2 - 25 Hz exhibited no observable correlated sway in the cumulant density estimate. Cumulant density magnitude was measured as an r-value. AP: antero-posterior; r-Tib: right tibialis anterior, r-mGas: right medial gastrocnemius.        40 Trunk Sway The correlation between trunk sway and SVS decreased or was absent as frequencies below 2 Hz were removed from the stimuli (Figure 2.3). The SVS stimuli evoked a single trunk sway response in the direction of the anode peaking at 1198 ± 178 ms for 0 - 1 Hz, 1155 ± 266 ms for 0 - 2 Hz and 1008 ± 354 ms for 0 - 25 Hz. In contrast, the 1 - 25 Hz stimulus evoked very small biphasic sway responses that were only present in some subjects and were characterized by an early negative correlation at 284 ± 90 ms (n = 6) and a later positive correlation in the direction of the anode at 833 ± 381 ms (n = 4). The 2 - 25 Hz stimulus was not associated with any apparent sway (Figure 2.3). Trunk sway correlations were not formally compared across stimulus conditions due to the inconsistent sway patterns: the responses associated with the 0 - 1 Hz, 0 - 2 Hz and 0 - 25 Hz stimuli were monophasic and observed in all subjects, a measurable sway response to the 1 - 25 Hz stimulus was observed only in 1/3 of the subjects, while the 2 - 25 Hz stimulus produced no measurable trunk sway in any of the subjects tested. While both the 1 - 25 Hz and 2 - 25 Hz stimuli appeared to minimize or abolish the prolonged AP sway response observed in the cumulant density function, only the 2 - 25 Hz stimulus did not increase RMS AP trunk sway when compared to the no stimulation trial (F(6,11) = 89.6 , p<0.05; 12.3 ± 4.4 (1-25 Hz) vs 9.6 ± 4.0 (Control), Fisher p<0.05; 10.5 ± 3.6 (2-25 Hz) vs 9.6 ± 4.0 (Control), Fisher p>0.05) (Figure 4). All trials except the 2 - 25 Hz stimulus exhibited increased mean-removed RMS trunk sway when compared to the control trial (Figure 2.4).      41 SVS-Sway Gain Response  From the initial 0 - 25 Hz SVS stimulus, signal power was successively low pass filtered at the muscle, then at ankle moment production and finally at trunk sway (Figure 2.5A). The gain of the closed loop transfer function from SVS to EMG exhibits two peaks localized within the 0 - 10 Hz bandwidth: one low frequency peak (<1 Hz) and a second peak around 4.5 Hz. When the closed loop transfer function is extended a step further to the SVS to AP ankle moment, the signal gain exhibited a large low-pass filtering effect, decreasing in signal power by 50 times at 5.5 Hz. Signal power for SVS to AP sway exhibited a 50 times decrease in power at 1.5 Hz. Signal transfer from EMG to sway was determined through inferred open loop transfer functions for EMG to AP moment, EMG to AP sway and AP moment to AP sway (Figure 2.5B). The inferred open loop transfer functions were similar to the closed loop estimates. Signal power decreased by 50 times, at 6.5 Hz, from EMG to AP moment and by 40 times, at 1.9 Hz, from EMG to AP sway. These results suggest that SVS signal power is low pass filtered by the mechanics of the body as muscle activation is converted into body sway.    42   Figure 2.4 Root-Mean-Square of antero-posterior sway and peak antero-posterior sway correlation. A. Average Root-Mean-Square antero-posterior trunk sway for each trial condition. Square trials are 1 mA 1 second square wave pulses whereas participants received no stimulus during control trials. Stimulus bandwidths are labelled on the abscissa. Error bars are standard deviations. B. Average peak antero-posterior trunk sway correlation as measured on the cumulant density function. Stimulus bandwidths are labelled on the abscissa.      43  Figure 2.5 Gain plots along six stages of the SVS-sway relationship. A. Gain of the closed loop transfer functions from SVS input to three output measures: EMG muscle response (r-mGas), AP moment and AP sway. SVS-AP moment gain shows a strong filtering effect reducing prominent signal power transfer to frequencies below 5.5 Hz. SVS- AP sway gain shows a further reduction in signal power transfer to frequencies below 1.5 Hz B. Gain of the inferred open loop transfer functions between EMG (r-mGas) and AP moment, EMG (r-mGas) and AP sway, and AP moment and AP sway. EMG (r-mGas)-AP moment and EMG (r-mGas)-AP sway gain functions mirror the SVS-AP moment and SVS-AP sway functions with filtering of signal transfer frequency to below 6.5 Hz and below 1.9 Hz. AP moment-AP sway gain function shows signal transfer below 1.5 Hz. AP: antero-posterior; r- Tib Ant: right tibialis anterior, r-mGas: right medial gastrocnemius, r-Sol: right soleus, SSF:   44 somato-sensory feedback.  2.4 Discussion The primary aim of the current study was to develop a vestibular stimulus using a specific bandwidth to minimize SVS-related sway responses. The main findings in this study supported my hypotheses namely, a) a trunk sway response wasn’t present to high pass filtered stimuli while multi-phasic muscle responses were retained and b) decreases in trunk sway to the 1 - 25 Hz and 2 - 25 Hz stimuli appear to derive from mechanical low-pass filtering attenuating the higher frequency muscle responses as muscle activation is transferred to sway. Vestibular stimuli at low frequencies (< 2 Hz) were mainly associated with prolonged vestibular related trunk sway whereas typical muscle responses were evoked by vestibular stimulation excluding these low frequencies. Indeed, removal of the 0 - 1 Hz or 0 - 2 Hz content from the stochastic stimuli abolished prolonged trunk sway responses to the SVS stimulus, as represented in the time cumulant densities, resulting in a very small or absent residual sinusoidal sway pattern. This reduction of sway in response to bandwidth-limited SVS appears to be the result of a change in the relative magnitude of the early and middle latency components of the correlated AP force peaks (Figure 2.3). Although correlated responses between SVS and EMG, or force, have no physical values, they do possess similar spatial and temporal characteristics to the trigger-averaged EMG (Dakin et al., 2007) and force responses (Mian & Day, 2009) observed with GVS. Hence, the relative area within the correlation is discussed using a similar framework to the responses evoked by GVS. Removing the 0 - 1 Hz bandwidth increased the peak correlation of early latency force   45 response while attenuating that of the middle latency response, resulting in each component having nearly equivalent areas. The short time course of the SVS-related early force response and rapid onset of the opposing middle latency force response of similar magnitude provide little time for sway to accumulate, thereby reducing the magnitude of the resulting correlated sway. Similarly, the SVS-related force response to the 2 - 25 Hz bandwidth stimuli exhibits three small peaks occurring over a very short time period, preventing the production of any observable sway response. In contrast, the correlation between force and the 0 - 25 Hz stimulus exhibits a very small short latency component but a larger opposing medium latency component leading to a prominent sway response in the direction of the medium latency force response. This allows significant sway to be produced by the middle latency force response prior to correction. Regarding the standing balance behaviour, GVS pulses (1 mA 1 second duration) delivered randomly every 15 - 25 seconds, as well as most stochastic vestibular stimuli bandwidths (0 - 1 Hz, 0 - 2 Hz, 0 - 25 Hz and 1 - 25 Hz), increased the variability of trunk sway observed when compared to the control of normal standing balance. Only the 2 - 25 Hz vestibular stimuli were not associated with more variable trunk sway in the AP direction (Figure 2.4). This is a likely consequence of the minimal amount of horizontal force and the absent trunk sway that is associated with this stimulation bandwidth. Frequency related changes in sway could be explained as a purely mechanical process. By increasing the frequency of the vestibular stimulus, the frequency of the corresponding envelope of muscle activity also increases requiring higher gain to maintain the impulse necessary to overcome the inertia of the body and induce sway. In contrast, signal transfer gain from stimulus to sway experiences a strong low pass filtering effect when transferring from net muscle activity to AP trunk sway (Figure 2.5). The majority of the filtering effect   46 arises from the conversion of net muscle activity to AP moment reducing signal power to frequencies below 6.5 Hz while transfer from AP moment to AP trunk sway further limits signal power to frequencies below 2 Hz. Low pass filtering of muscle activation to force output is a mechanical consequence of intrinsic muscle properties and net motor activity across one or several muscles acting on a single common effector, in this case is the ankle joint, resulting in a smoothed force output with a bandwidth usually below 3 to 5 Hz (Bawa & Stein, 1976; Olney & Winter, 1985; Fitzpatrick et al., 1996). Signal filtering is also observable between AP moment and AP sway (Figure 2.5A & B) and is likely owing to the inertial load of the body resisting applied forces (Fitzpatrick et al., 1996; Latt et al., 2003). It is conceivable that the central nervous system specifically encodes for the mechanical filtering of the vestibular-sway pathway. Identifying the locus of potential neural structures representing the mechanical filtering described here cannot be performed in awake humans but experiments in animals might provide potential mechanisms. For example, animal models revealed that sinusoidal electrical vestibular stimulation affects vestibular afferents (albeit predominantly the irregular ones) and the second order vestibular neurons at stimulation frequencies up to 100Hz (Goldberg et al., 1984; Kleine & Grusser, 1996). Once in the fastigial nucleus, however, two distinct populations of vestibular related neurons have been described: one group that responds to low frequency (<1 Hz) sinusoidal vestibular stimulation and another group that is tuned towards frequencies up to 10 Hz (Schlosser et al., 2001). The fastigial nucleus contains vestibular related neurons that respond to vestibular signals with the appropriate coordinate transformation necessary to elaborate motor commands appropriate for whole-body responses (Manzoni et al., 1999; Kleine et al., 2004; Brooks & Cullen, 2009). These behaviours are similar to the vestibular-evoked muscle and   47 sway responses that show well-defined spatial transformations related to the position of the head relative to the feet (Lund & Broberg, 1983; Britton et al., 1993; Fitzpatrick et al., 1994; Mian & Day, 2009) and distinct low and higher frequency response characteristics to vestibular stimuli (as described here). Although it is possible that the nervous system and potentially cerebellar nuclei represent or independently code the mechanical dissociation between muscular and whole-body responses, additional experiments are required to test this hypothesis for balancing actions. Removal of the low frequency content of the stochastic stimuli enlarges the early latency component of the correlated AP force response that is applied to the body, suggesting that SVS void of frequencies below 1 or 2 Hz might be an effective way to investigate the early latency response. Traditionally, the early and middle latency vestibular-evoked force or muscle responses have been described as independent entities (Britton et al., 1993; Cathers et al., 2005), the early latency response has been suggested to originate from the otoliths and travel via the reticulo-spinal pathways whereas the middle latency response originates from the semicircular canals and travel via the vestibulo-spinal pathways. The early and middle latency peaks also appear to be contributed to by stimuli of different frequency content. The early latency peak is strongly contributed to by stimuli with higher frequency content (Rosengren & Colebatch, 2002; Dakin et al., 2007) while the middle latency response is largely influenced by stimuli with increased low frequency content (Britton et al., 1993; Rosengren & Colebatch, 2002; Dakin et al., 2007). Physiologically, there is some evidence to suggest that low frequency natural stimuli are transduced somewhat differently, mainly through regular firing vestibular afferents, than higher frequency natural input, which are transmitted through both regular and irregular firing vestibular afferents (Sadeghi et al.,   48 2007). Thus adjusting stimulus bandwidth to evoke a response component of interest (early or middle latency response) might be an important attribute of SVS, enabling further examination of the often difficult to isolate early latency response with less confounding interference from the middle latency response.  Functional Relevance SVS could provide an ideal tool for examination of lower-limb vestibular evoked responses in patient populations. Firstly, SVS requires relatively short stimulus durations to evoke prominent EMG and forces responses; 180 seconds total stimulus duration in Dakin et al. (2007) and 133 second trials in this study in comparison to 300 seconds for square wave GVS in Dakin et al. (2007) and between 260 - 300 seconds, for twenty pulses, in this study. Secondly, if induced sway resulting from vestibular stimulation confounds interpretation of results or is unwanted when assessing vestibular function in patients with a balance disorder (Pastor et al., 1993; Marsden et al., 2005; Liechti et al., 2008), the stimulus could be tuned to reduce unwanted sway responses by excluding frequencies below 2 Hz. Thirdly, stimulus bandwidths might also be tuned to study a particular physiological response such as the early (1 - 25 Hz or 2 - 25 Hz) or middle (0 - 25 Hz) components of the vestibular-evoked responses. This could prove particularly important to investigate the possible mechanisms proposed to contribute to the vestibular-evoked balance phenomenon: otoliths and semicircular canals (Cathers et al., 2005) or independent spinal pathways (Britton et al., 1993).      49 Limitations  A limitation regarding interpretation of this study is that the use of narrow bandwidth, low frequency stimuli (0 - 1 Hz and 0 - 2 Hz) results in correlations in the cumulant density occurring prior to the zero lag point. Prior to zero correlations occur with narrow bandwidths because the random nature of the stimulus is reduced creating a quasi sinusoidal stimulus which, much like a sine wave, will experience some correlation with the waveforms previous period along with the current one (Matthews, 1993). This prior to zero correlation confounds interpretation of the timings of the associated vestibular related response peaks but can be avoided by using vestibular stimuli with large bandwidths (e.g. 0 - 25, 1 - 25 or 2 - 25 Hz).  2.5 Conclusion  Removal of the low frequency content of the SVS stimulus resulted in attenuation of vestibular evoked sway responses and an increased amplitude of the early latency vestibular evoked force response. These results demonstrate that SVS with a bandwidth containing frequencies above 2 Hz might be used to provide a vestibular stimulus which does not cause a destabilizing sway response. Overall, SVS provides multiple benefits over GVS: it is more comfortable (Dakin et al., 2007), can elicit similar responses in a shorter stimulation period (Dakin et al., 2007) and its bandwidth can be modified to limit prolonged sway responses or assess specific components of the muscle or balance behaviour.   50 2.6 Summary: Study 1  The aim of study one was to examine whether or not the sway response to a vestibular stimulus could be attenuated by removing stimulus frequencies below 2 Hz. Indeed this is what was observed. The effect of the stimulus appears to be transferred to lower and lower frequencies in its transmission from muscle activity to torque production and finally to sway. Since stimulus induced sway is predominantly below 1.5 Hz the removal of these frequencies attenuated sway while retaining the biphasic muscle response. This finding provides the opportunity to investigate the role of vestibular ex-afferent type signals using electric vestibular stimulation in contexts in which the maintenance of stability is essential. The second study again examined the modification of stimulus bandwidth but this time for the purpose of a) determining whether the bandwidth of the stochastic stimulus can be modified to isolate either the short or medium latency component of the vestibular evoked response and b) identifying whether the provision of sinusoidal stimuli from 1 to 20 Hz provides additional information over the broad bandwidth stimulus regarding interaction between the short and medium latency responses. If the sinusoidal stimuli do not provide additional information over the stochastic stimuli it provides additional support for the use of broad bandwidth stochastic stimuli due to the greatly reduced testing time required for these stimuli when compared to multiple sinusoidal stimuli.   51 3 Short and Medium Latency Muscle Responses Evoked by Electrical Vestibular Stimulation are a Composite of all Stimulus Frequencies 5    3.1 Introduction Galvanic or stochastic vestibular stimulation (GVS and SVS) entail the trans- mastoidal percutaneous application of a small electric current, which is used to probe vestibular function (GVS:(Coats, 1973; Nashner & Wolfson, 1974)  for review see Fitzpatrick and Day 2004; SVS: (Fitzpatrick et al., 1996; Dakin et al., 2007)). The applied electric current modulates the firing rate of the underlying vestibular afferents producing responses in muscles active in the maintenance of balance. In lower-limb muscles these responses exhibit a biphasic pattern with two opposing peaks, termed short (50 - 70 ms) medium (100 - 120 ms) latency components (Nashner & Wolfson, 1974; Iles & Pisini, 1992; Britton et al., 1993; Fitzpatrick et al., 1994; Fitzpatrick & Day, 2004; Lee Son et al., 2008; Reynolds, 2010). While the exact physiological mechanisms underlying these two peaks remain unclear, results from previous studies indicate they possibly derive from independent sources (Britton et al., 1993; Cathers et al., 2005; Dakin et al., 2007). Two main hypotheses have been formulated to account for the biphasic vestibulo-motor response: the short and medium latency components might originate from two distinct regions in the brain (the vestibulo and reticulospinal systems; see Britton et al. 1993) or from the stimulation of afferents from the otoliths and  5  A version of Chapter 3 has been published: Dakin CJ, Inglis JT & Blouin JS. (2011).Short and medium latency muscle responses evoked by electrical vestibular stimulation are a composite of all stimulus frequencies. Exp Brain Res 209, 345-354.    52 semicircular canals (Cathers et al., 2005). Recently, I proposed these two components could be preferentially elicited by specific vestibular stimulus frequency bandwidths (Dakin et al., 2007), which could provide a simple and effective way to assess the function and physiological relevance of the short and medium latency components in humans. In the present experiment, I developed specific electrical stimuli to determine if altering the frequency content of an electrical vestibular stimulus could prove useful in dissociating the two vestibular response components.   The two components of the biphasic vestibulo-myogenic response might receive independent contributions from specific stimulus frequency bandwidths, as suggested by two regions (2 - 10 Hz and 11 - 20 Hz) of stronger coupling between SVS and muscle activity (Dakin et al., 2007). The time lag estimated by the slope of the SVS-muscle phase function within these two bandwidths indicates that the 2 - 10 Hz bandwidth contributes primarily to the medium latency component and the 11 - 20 Hz bandwidth to the short latency component. These findings predict that an SVS stimulus within either the 0 - 10 or 10 - 20 Hz bandwidth should modulate selectively either the short (10 - 20 Hz) or medium latency (0 - 10 Hz) peak of the vestibulo-myogenic response. Selective modulation of either response component would provide further support for the independence of the two components of the vestibular response. Therefore the first aim of the current study was to determine if distinct stimulus frequency bandwidths contribute to the individual peaks of the biphasic muscle response. I tested this hypothesis by designing stimuli within each of the two previously identified bandwidths in order to exclusively modulate the short (10 - 25 Hz stimulus) or medium latency (0 - 10 Hz stimulus) lower-limb vestibulo-myogenic responses.    53 Hypothesis: I hypothesized that the short latency component of the muscle response would be triggered by the 10 - 25 Hz vestibular signals and the medium latency component by the 0 - 10 Hz signals.  The second aim of this study was to test the independence of the two components of the vestibulo-myogenic response by determining whether activation of distinct brain regions with a fixed central delay, as proposed by Britton et al. (1993), could account for the presence of a biphasic vestibulo-myogenic response. Assuming linear transmission of reflexes, a consequence of examining a system exhibiting two distinct reflexive responses triggered by brain regions with a fixed delay is that the two reflex components will exhibit a temporal overlap (described as reflex component interaction) at specific frequencies of stimulation. Specifically, the interaction of these reflex components will depend on the time delay that exists between these components. Matthews (1993) used this rationale to examine the components of the human stretch reflex, identifying perturbation frequencies (around 20 Hz) for the wrist muscles at which the short and long latency components of the stretch reflex exhibit a temporal overlap. This interaction between short and long latency stretch reflex components was demonstrated in two ways: a) as abrupt changes in the phase frequency relationship between sinusoidal stimuli and associated electromyographic response; and b) as modulation of electromyographic activity at a frequency other than at the stimulation frequency. A similar interaction between the two vestibular response components might occur at specific frequencies of stimulation if the vestibulo-myogenic response originates from two independent central sources exhibiting a fixed delay between them (as proposed by Britton et al. 1993). For example, SVS-muscle coherence decreases at 10 Hz and exhibits a small inflection in phase at this frequency (Dakin et al., 2007), potentially reflecting an interaction   54 between the short and medium latency vestibular response. The approximate 50 ms delay between the short and medium latency response components in the leg muscles (Britton et al. 1993) could be produced by two inputs to the motor neuron pool that are out of phase, resulting in destructive interference of the response to a 10 Hz vestibular stimulus. To test this second hypothesis, subjects were exposed to a series of sinusoidal stimuli (1 to 20 Hz) from which I examined the stimulus-EMG correlations and phase frequency function on a frequency by frequency basis. Hypothesis: I hypothesized that, for vestibular stimuli around 10 Hz, an interaction between the short and medium latency response components would be observed, yielding an abrupt shift in phase of the stimulus-EMG phase frequency function as well as EMG modulation at frequencies other than the stimulus frequency.  3.2 Methods Subjects  Twelve healthy subjects (7 male, 5 female) between the ages of 20 and 34 years, with no known history of neurological disease or injury participated in this study. The experimental protocol was explained to each subject and their written, informed consent was obtained. All procedures used in this study conformed to the standards of the Declaration of Helsinki and were approved by the University of British Columbia’s clinical research ethics board.  Vestibular Stimuli  Vestibular stimulation was delivered using a binaural bipolar electrode configuration. Carbon rubber electrodes (9 cm 2 ), coated with Spectra 360 electrode gel (Parker Laboratories,   55 Fairfield, USA), were secured over participants’ mastoid processes with an elastic headband. Testing was performed in two separate experimental sessions (See below). Vestibular stimuli were generated on a PC computer using Labview software (National Instruments, Austin, USA) and were sent directly to a constant current isolation unit (Model 2200 Analog Stimulus Isolator: AM Systems, Carlsborg, WA) via a multifunction data acquisition board (PXI-6289, National Instruments, Austin, USA). The 0 - 10 Hz, 10 - 25 Hz and 0 - 25 Hz stochastic signals were delivered as two 105 seconds trials which were normalized to provide similar amplitude at each frequency component (Figure 3.1), resulting in different RMS amplitudes for each bandwidth stimuli: 0.65 mA (0 - 10 Hz), 0.73 mA (10 - 25 Hz), and 0.98 mA (0 - 25 Hz). Frequency normalization of the stimuli ensured equal contribution for each frequency within the stimulus. Sinusoidal stimuli, on the other hand, had amplitudes of ± 2 mA and were 90 seconds in length while the comparison 0 - 20 Hz stochastic stimuli were also 90 seconds in length but ± 4 mA in amplitude. The sinusoidal stimuli were provided with amplitudes of ± 2 mA to ensure adequate responses at all frequencies but also small enough to limit the uncomfortable effects of the vestibular stimulus. The 0 - 20 Hz stochastic stimuli control trial was provided with an amplitude of ± 4 mA to ensure an adequate response magnitude for comparison with the average of the sinusoidal stimuli (Dakin et al., 2007).  Testing Protocol  During each testing session, participants were required to stand with their feet 2 - 3 cm apart (as measured at the medial malleoli). For each trial, participants were asked to keep their arms at their sides, stand relaxed and keep their head turned to the right and eyes level, parallel to the floor. To maintain head position, participants were asked to focus on a target to   56 the right of them. By maintaining this head position the postural response to the vestibular stimulus was primarily aligned to the anterior – posterior directions, along the line of action of the left gastrocnemius (Lund & Broberg, 1983; Cathers et al., 2005; Day & Fitzpatrick, 2005). Electromyography (EMG) was collected for the left medial gastrocnemius as vestibular- evoked muscle responses are larger in the leg opposite to a head turn (Britton et al., 1993; Dakin et al., 2007). To test my first hypothesis ten subjects were exposed to three vestibular stimuli: a 0 - 10 Hz stimulus to modulate preferentially the medium latency response, a 10 - 25 Hz stimulus to modulate the short latency response and a 0 - 25 Hz stimulus as a control. The bandwidths of the 0 - 10 and 10 - 25 Hz stimuli were chosen based on the frequency range of the phase estimate described previously (Dakin et al., 2007). Participants were provided a total of six trials: two trials for each of the three stimulus bandwidths.  Rest periods were provided at the request of the participant to avoid any sign of fatigue. To test my second hypothesis five participants (including two new participants) were provided twenty sinusoidal stimuli spanning 1 to 20 Hz and a 0 - 20 Hz SVS comparison control trial. Participants were tested for a total of twenty one trials which were collected on a separate day from the first experimental session. To reduce the number of vestibular stimuli for participants, I limited the sinusoidal stimuli to the 1 - 20 Hz bandwidth since SVS-muscle coherence is largest in this region (Dakin et al., 2007) and it encompasses the frequencies at which an interaction between reflex components might occur (~10 Hz).      57 Data Collection and Analysis EMG was amplified (Gain ×5000 - 20000; Grass P511, Grass - Telefactor, West Warwick, USA) and band-pass filtered (30 - 1000Hz) prior to being digitized at 5000 Hz for the stochastic stimuli and 8192 Hz for the sinusoidal stimuli and associated 0-20 Hz control trial. EMG was sampled at 8192 Hz for the sinusoids to allow Fast Fourier Transform (FFT) windows of a power of two (8192 points). Using this sampling frequency and FFT window length, frequency values occur at each integer frequency, congruent with the sinusoidal stimuli which were delivered at integer frequencies. SVS trials within each testing session were time-locked to SVS onset, and concatenated within each condition for each participant. The sinusoidal stimuli were delivered as a single trial and therefore within participant concatenation was not necessary. EMG data were full-wave rectified, concatenated within subjects and concatenated across all subjects. Concatenated data were then used to estimate cross-covariance and phase functions between SVS (or sinusoidal stimuli) signal and EMG. For statistical analysis, dependent variables were extracted from the cross-covariance and phase functions estimated using concatenated data within participants whereas the functions estimated using concatenated data across all participants were used for illustrative purposes only. Within subject cross-covariance and phase estimates provided timing and amplitude information regarding the time-domain characteristics of the identified coherent frequencies (Rosenberg et al., 1989; Dakin et al., 2007).   Cross-covariance functions and phase estimates were calculated using a Matlab script based on the method described by Rosenberg and colleagues (Rosenberg et al., 1989; Halliday et al., 1995). SVS-EMG cross-covariance functions were analyzed with resolutions of 0.076 Hz (13.1sec/segment) to identify the low frequency contributions. Sinusoidal data   58 were analyzed with resolutions of 1 Hz (1 sec/segment) to correspond to the frequency content of the sinusoidal stimuli (1 Hz). By convention, anode right / cathode left currents are represented as a positive vestibular signal. Hence, a positive cross-covariance function indicates that anode right / cathode left currents induced muscle facilitation. The local maxima and minima in the muscle responses were estimated from the cross-covariance functions computed from the data concatenated for individual subjects. To address my first hypothesis the polarity and timing of the peaks (and troughs) obtained from the cross-covariance function for the 0 - 10 and 10 - 25 Hz bandwidths were compared to those obtained following the broad bandwidth (0 - 25 Hz) vestibular signals. This comparison was made to determine if a single component of the biphasic muscle response was triggered by the narrow bandwidth stimuli. Similarity between the 0 - 10 and 0 - 25 Hz covariance functions were assessed using paired t-tests (statistical significance was set at p = 0.05). To address my second hypothesis, phase estimates were calculated for both the sinusoidal stimuli and the 0 - 20 Hz stochastic control stimuli. The phase estimates for the sinusoidal stimuli were extracted at each stimulus frequency, concatenated across frequencies and realigned by values of 2π using the 1 Hz trial as a reference to obtain a continuous phase estimate. For illustrative purposes, the phase values obtained for the 0 - 20 Hz stochastic control trials are displayed as a continuous function of frequency (unwrapped) to provide an uninterrupted slope. As a secondary measure, I examined changes in EMG power to the sinusoidal stimuli. Similarly to Matthews (1993), if the two reflex components interact it should be apparent as an increase in EMG power at frequencies other than the stimulus frequencies.    59 3.3 Results Do 0 - 10 Hz and 10 - 25 Hz Vestibular Stimuli contribute Independently to the Short and Medium Latency Components of the Vestibulo-Myogenic Response?  The time-domain correlations between SVS and left medial gastrocnemius EMG revealed that the 0 - 10 and 10 - 25 Hz bandwidth stimuli contribute to the shape and timing of both components of the vestibulo-myogenic response. The EMG-SVS cross-covariance obtained for the 0 - 10 Hz bandwidth exhibited spatial properties similar to those obtained for the 0 - 25 Hz vestibular stimuli, with both stimuli eliciting biphasic responses (Figure 3.1). The timing of the two peaks, however, differed between stimuli. The short latency component of the 0 - 10 Hz cross-covariance occurred earlier than the corresponding peak for the 0 - 25 Hz control (50 ± 18 ms vs 65 ± 6 ms),  while the medium latency component of the 0 - 10 Hz response was later than the medium latency component of 0 - 25 Hz control (127 ± 18 ms vs 115 ± 14 ms). This resulted in a longer period between the short latency peak and medium latency peak for the 0 - 10 Hz stimuli compared to the 0 - 25 Hz stimuli (77 ± 6 ms vs 50 ± 13 ms; t(10) = 8.61, P < 0.05 ). On the other hand, the cross-covariance for the 10 - 25 Hz vestibular stimuli exhibited a triphasic response and was therefore not compared to the broad bandwidth responses (42 ± 7 ms, 73 ± 5 ms and 106 ± 6 ms). The spatial and temporal characteristics of the second and third peaks of the triphasic response triggered by 10 - 25 Hz stimuli appear to contribute mainly to the short and medium latency components of the muscle response observed following 0 - 25 Hz stimuli (Figure 3.1). Overall, neither the 0 - 10 or 10 - 25 Hz stimuli independently contributed to short or medium latency component of the biphasic muscle response.   60  Figure 3.1 Stimuli, stimulus power spectra and 0 - 10 Hz, 10 - 25 Hz and 0 - 25 Hz cross- covariance functions. A. Five seconds of the 0 - 10 Hz, 10 - 25 Hz and 0 - 25 Hz stimuli, B. Power spectra for the 0 - 10 Hz, 10 - 25 Hz and 0 - 25 Hz stimuli, C. SVS - EMG cross- covariance functions for the 0 - 10 Hz, 10 - 25 Hz and 0 - 25 Hz stimuli (n = 10). Neither the 0 - 10 Hz or 10 - 25 Hz stimuli evoked a single component of the biphasic response, although the 10 - 25 Hz stimulus appears to exhibit its greatest influence on the short latency component of the biphasic response. Hz: Hertz, s: seconds, ms: milliseconds  Are the Two Components of the Vestibulo-Myogenic Response Caused by a Fixed Central Delay?  To test my second hypothesis, sinusoidal stimuli with frequencies ranging from 1 to 20 Hz were provided. I examined the phase between the muscle responses and vestibular stimuli as well as EMG power for indications of interactions in the vestibular-evoked muscle responses. All stimulus frequencies yielded increases in EMG power isolated to the stimulus frequency and not multiple frequencies as would be suggested by an interaction (Figure 3.2B).   61   Figure 3.2 Raw left medial gastrocnemius EMG recording and corresponding EMG power spectrum for a single subject. A. Raw EMG showing entrainment to the stimulus over frequencies between 4 to 12 Hz. B. EMG power spectrum displaying increased power only at the stimulus frequency between 4 to 12 Hz.    62 However, stimulus frequencies between 4 to 12 Hz exhibited stronger entrainment of lower- limb EMG than stimulus frequencies outside of this bandwidth (Figure 3.2 & 3.3). For stimulus frequencies exhibiting obvious entrainment, the association between the vestibular stimulus and the muscle response was detectable in the raw data (Figure 3.3): an anode right / cathode left vestibular stimulus was associated with increased muscle activity about 90 - 100 ms later while anode left / cathode right currents yielded a decrease (or absence) in gastrocnemius muscle activity with a similar delay. Similar entrainment was not evident in the 0 - 20 Hz stochastic trials (Figure 3.3). All cross-covariance for the 1 to 20 Hz sinusoidal stimuli and associated EMG exhibited clear sinusoidal modulation at the stimulation frequency, correlating both prior to and following the zero time mark (Figure 3.4A). Cross- covariance functions within the 4 to15 Hz bandwidth displayed the largest modulation. None of the vestibular stimuli elicited muscle responses at frequencies other than the stimulation frequency, further indicating a lack of interaction between two central processes at a fixed delay (Figure 3.2B).  Examination of the vestibular stimuli-EMG phase relationship shows a comparable phase frequency function between the broad bandwidth stimulus and the individual sinusoidal stimuli over the 0 - 20 Hz frequency range (Figure 3.5). For the 0 - 20 Hz stochastic trial, the slope of the phase frequency function over 0 - 10 Hz corresponds to a phase lag of 108 ms compared to 116 ms for the sinusoidal stimuli phase frequency function over the same frequency range (i.e., 0 - 10 Hz). For the 10 - 20 Hz interval, the slope of the 0 - 20 Hz broad bandwidth SVS indicated a phase lag of 71 ms compared to 77 ms for the sinusoidal stimuli (Figure 3.5). Despite replicating the previously observed differences in the estimated time lags from the phase functions (Dakin et al., 2007), no prominent inflection in the phase   63 functions was present for either stimulus protocols.   Figure 3.3 Description of stimulus EMG entrainment for a single subject. Rectified EMG for a stimulus frequency (3 Hz) which shows minimal EMG entrainment compared to a stimulus frequency (8 Hz) displaying stronger EMG entrainment. EMG from the 0 - 20 Hz control trial does not exhibit any obvious signs of entrainment to the stochastic stimulus.  Are Both Peaks of the Biphasic Response rather a Composite of All Frequencies within the Stimulus? As the 0 - 10 and 10 - 25 Hz stimuli did not independently modulate the short or medium latency responses, I examined whether the biphasic response is rather the net result of a broad bandwidth of stimulus frequencies. If this is true the resulting averaged muscle response across all the individual sine waves should resemble the muscle response to broad bandwidth stimuli. To examine this, the cross-covariance functions for each of the sinusoidal   64  Figure 3.4 Cross-covariance functions for the sinusoidal stimuli. A. Cross-covariance functions for the sinusoidal stimuli and the control broad bandwidth stimuli. Sinusoidal stimuli produce sinusoidal covariance functions which when averaged produce a biphasic response similarly to the 0 - 20 Hz. In grey are each subjects muscle response to the stimulus frequency listed on the left with the pooled response for all subjects superimposed in black (n   65 = 5). Sinusoidal stimuli induce correlations in the cross-covariance function both following and preceding the zero time mark. Such pre-zero correlations result from a single period of the vestibular stimulus correlating not only with the response elicited by the stimulus but also with the response elicited by the stimulus period prior and the stimulus period post.  B. Superposition of the sinusoidal covariance functions displays the phase relationship between the different frequencies underlying the biphasic mean. Muscle responses to stimulus frequencies of 5Hz and below (grey segmented lines) are detrimental to the shape of the early latency response observed in the cross-frequency mean (black line). These plots are means (n = 5). Muscle responses to the 1 - 5 Hz stimuli are grey segmented lines, muscle responses to the 6 - 20 Hz stimuli are grey solid lines and the mean muscle response across all stimulus frequencies are black solid lines.  stimuli were determined and averaged across frequencies within each subject. The timing and polarity of the mean of all frequencies was then compared to the cross-covariance from the 0 - 20 Hz control trial. Similarity between the mean sinusoidal response and the 0 - 20 Hz control trial would suggest that vestibular reflexes stimulated at all frequencies between 0 - 20 Hz summate linearly to contribute to both the short and medium latency response components. Presumably, the relative phase for each frequency contributing to the average is important in determining the spatial and temporal characteristics of the biphasic profile of the average for the sinusoidal cross-covariance functions. Frequencies from 1 - 5 Hz, in this study, appeared to decrease the short latency component of the biphasic response as their cross-covariance function exhibits a single period for the duration of the biphasic response which is opposite in polarity to the short latency response (Figure 3.4A & B). In contrast,   66 frequencies from 6-13 Hz exhibit the correct period length, polarity and phase to contribute to both the short and medium latency component of the biphasic response (Figure 3.4A & B). Frequencies greater than 13 Hz exhibit multiple periods over the duration of the biphasic response suggesting these frequencies contribute to shaping the contours of the net waveform, especially sharpening the peaks of the short and medium latency responses (Figure 3.4A & B). When all the sinusoidal cross-covariance functions were superimposed their phase relationship becomes apparent highlighting how the biphasic waveform could result from the average of the sinusoidal cross-covariance functions. When the sinusoidal cross-covariance functions were averaged across frequencies the result was a biphasic response which approximated the spatial and temporal properties of the 0 - 20 Hz control cross-covariance function, exhibiting an early negative followed by later positive going peak (Figure 3.4A). The timing for the short latency component however, was significantly different from the control at 65 ± 6 ms (mean of sines) and 62 ± 6 ms (0 - 20 Hz control) [t(4) = 3.35, P < 0.05], while the timing of the medium latency component was not different from the control 104 ± 7 ms (mean) and 104 ± 10 ms (0 - 20 Hz control) [t(4) = 0.1, P > 0.05].  Figure 3.5 Average phase frequency functions for the 0 - 20 Hz sinusoidal stimuli versus the 0 - 20 Hz broad bandwidth control stimulus (n = 5).   67 3.4  Discussion The aim of the current study was to determine if parameters of an electrical vestibular stimulus could be tuned to assess the function and physiology of the short and medium latency vestibulo-myogenic response. Specifically, I tested the hypotheses that a) frequencies from the 0 - 10 and 10 - 25 Hz bandwidths independently influence the short or medium latency components of the vestibular-evoked biphasic lower-limb muscle response and b) the two components of the biphasic myogenic response, due to their fixed central delay, would interact at frequencies of stimulation near 10 Hz. The present results did not support either hypothesis. Instead the results suggest that the biphasic lower-limb muscle response is a linear composite of all frequencies within the stimulus bandwidth (0 - 20 Hz) and that the two components of the biphasic response cannot be explained by two independent sources generating muscle responses with a fixed delay between 25 to 500 ms. Previously I observed a dual sloped phase frequency relationship using a 0-50 Hz broad-bandwidth stimulus (Dakin et al., 2007) which suggested, along with previous research (Britton et al., 1993; Cathers et al., 2005; Day & Fitzpatrick, 2005), that the two peaks of the biphasic muscle response might be the product of separate physiological processes. The earlier of the two peaks appeared to be derived from frequencies in the 10 - 20 Hz bandwidth while the latter peak derived from frequencies in the 0 - 10 Hz bandwidth. These results led me to hypothesize that the frequency content of the stochastic stimuli could be adjusted to modulate independently the short or medium latency components of the biphasic response. This hypothesis, however, was proven incorrect as provision of low (0 - 10 Hz) or high (10 - 25 Hz) frequency stimuli did not independently modulate the short or medium latency components (Figure 3.1). Instead, it appears that the spatial and temporal characteristics of the   68 short and medium latency components of the lower-limb vestibulo-myogenic response result from activation of vestibular afferents by a stimulus comprised of frequencies from a broader bandwidth (0 - 20 Hz) (Figure 3.4) as indicated by linear summation of the individual sine waves. The lack of interaction between the two vestibular components to sinusoidal stimuli suggests that reducing the stimulus bandwidth to isolate components of the vestibular response does not provide any information that is not already provided by a broad bandwidth stimulus (0 - 20 or 0 - 25 Hz). Instead, it appears more appropriate to use a large bandwidth vestibular signal (0 - 20 or 0 - 25 Hz) for physiological or clinical testing due to increased subject comfort and reduced testing times. While no single peak of the biphasic response is contributed to exclusively by a specific frequency bandwidth, there do appear to be frequency bandwidths which contribute to specific attributes of the biphasic response. The biphasic shape of the myogenic response appears to be largely derived from frequencies between 6 - 13 Hz whilst the higher frequencies sharpen the peaks of the response. This is evident in the 10 - 25 Hz trial (Figure 1) where these frequencies appear to increase the amplitude and shorten the duration of the short latency component as well as shift the peak of the medium latency response earlier in time. On the other hand, frequencies below 6 Hz draw out the length of the medium latency response while attenuating the short latency response (Figure 3.4B). Accordingly, I have previously observed that as lower frequencies are omitted from a vestibular stimulus (2 - 20 Hz), the amplitude of the short latency response increased (Dakin et al., 2010b).  The paradoxical contribution of low frequencies to the short and medium latency response components might explain the slope difference in the phase frequency relationship for frequencies below 8 - 10 Hz when compared to frequencies above 10 Hz between the   69 stochastic vestibular stimuli and muscle response (c.f. Figure 5 and Figure 6 in Dakin et al. 2007). The two components of the vestibular evoked muscle response also do not appear to interact with each other over stimulation frequencies of 1 to 20 Hz. Evidence of such an interaction would have manifested as an increase in EMG power at a frequency other than that of the stimulus frequency and a shift in the stimulus-EMG phase frequency function similar to those observed by Matthews (1993). In contrast, EMG activity was modulated only at the frequency of stimulation for each of the sinusoidal stimuli and the phase frequency function did not exhibit a marked shift in continuity. Therefore my results do not support the premise that the two peaks of the biphasic myogenic response originate from two sources with a fixed time delay ranging between 25 - 500 ms. This interpretation would be appropriate for a linear system and, despite the non-linear input-output characteristics of individual motor neurons, the assumption of linearity is adequate for population of motor neurons (such as recorded with surface EMG here; Matthews 1993). This assumption is further supported by the observed similarity between the linear summation of the muscle responses evoked by sinusoidal stimuli between 0 - 20 Hz and those evoked by broadband stimuli over the same frequency range. Despite my failure to demonstrate an interaction between vestibular components, the two peaks could still originate from afferents of the semicircular canals and, although less likely, the otolith organs as proposed by Cathers et al. (2005). For example, it is possible that the delay between the two inputs could be less than 25 ms or that each of the individual components exhibit temporal features that are different than what can be inferred by their combined response in surface EMG. As well, the attenuated modulation of EMG to stimulus frequencies outside the 4 to 15 Hz bandwidth might mask signs of interaction in these   70 frequencies. Strong evidence attributes the semicircular canals to the medium latency peak as this peak can be predictably modulated based on the orientation of the head and the assumption of non-specific activation of semicircular canal afferents (Cathers et al., 2005; Day & Fitzpatrick, 2005; Mian & Day, 2009). In contrast, the origin of the short latency component is not as clear. The directionality of the short latency vestibular response does not appear to correspond to the net activation of otolith afferents as predicted by the model proposed by Fitzpatrick and Day (2004) (Mian et al., 2010).  Limitations Three methodological limitations must be considered when interpreting the results from this study. The first limitation pertains to the averaging of the muscle responses observed with 1 - 20 Hz sinusoids to replicate the results observed using the 0 - 20 Hz control stimulus. The 0 - 20 Hz stochastic stimulus contains power at non-integer frequencies. In order to most accurately replicate stochastic control condition, participants would have been required to be provided additional stimuli at intervals between each integer frequency making testing prohibitively long. The absence of non-integer stimulus frequencies likely led to the small difference (3 ms difference for the short latency component) in the average cross-covariance when compared to the 0 - 20 Hz stochastic control condition. The second methodological limitation pertains to the use of continuous repetitive stimuli. While continuous vestibular stimuli provide an advantage over discrete stimuli, by reducing testing time and the amplitude of stimuli that can be used, continuous repetitive stimuli likely induce correlated feedback into the muscle response. Sensory feedback correlated to the stimulus has been observed to influence vestibular responses with delays as   71 late as 400 ms (Day & Guerraz, 2007). Feedback occurring with shorter delays might be incorporated into the later aspects of the muscle response potentially interfering with interpretation of these later responses. The third methodological limitation pertains to the use of surface EMG as a measure to identify the underlying synaptic contributions to the biphasic responses. The biphasic muscle response as observed through surface EMG is composed of two underlying motor phenomenon: individual motor unit firing behaviour dictated by incoming synaptic currents and secondary ensemble phenomenon such as synchronization related oscillations due to the high autocorrelation function of the individual motor units (Moore et al., 1970). Because surface EMG provides an estimate of the sum of the total muscle activity it is difficult to isolate which component of it is caused by independent synaptic inputs versus synchronization-related behaviour.  3.5 Conclusion   This data demonstrates that selective bandwidth SVS cannot be used to selectively modulate the short or medium latency vestibular evoked muscle response. Instead the biphasic response as observed to a broad bandwidth stimulus (0 - 20 Hz) is the linear sum of all stimulus frequencies, with each frequency contributing to specific attributes of the short and medium latency vestibular evoked muscle response. The lack of non-linear interaction in the vestibular responses evoked at frequencies between 0 - 20 Hz supports the use of large bandwidth vestibular stimuli for clinical and physiological testing.   72 3.6 Summary: Study 2 The aim of this study was twofold: the first aim was to test whether the bandwidth of the stochastic stimulus could be modified to isolate either the short or medium latency component of the vestibular evoked response; the second aim was to identify whether the provision of sinusoidal stimuli from 1 to 20 Hz would allow for the detection of an interaction between the evoked response components at specific frequencies of stimulation (~ 10 Hz). The results of this study did not support the use of a 0 - 10 Hz or 11 - 25 Hz stimulus bandwidth to isolate the short or medium latency response components, nor did it identify an interaction between the two response component at frequencies around 10 Hz. It did however demonstrate that a broad bandwidth stochastic stimulus (0 - 25 Hz) provides the same information as the administration of a broad spectrum (1 to 20 Hz) of individual sinusoidal stimuli but in much less time (3 minutes versus 30 minutes). This finding validates the use of broad bandwidth stochastic stimuli (0 - 25) over their sinusoidal or narrow bandwidth counterparts. Study two concludes the investigation of the advantages of modifying the stimulus bandwidth. In these first two chapters I have shown that in general the administration of a broad bandwidth stimulus provides the same information as several narrow bandwidth stimuli, I have cast doubt on the timing of the two components of the biphasic muscle response and demonstrated that the removal of frequencies below 2 Hz attenuates the postural response to the vestibular stimulus due to mechanical filtering. The third study extended the usefulness of the stochastic stimulus to dynamic contexts. Specifically, I examined the application of stochastic vestibular stimulation to the extraction of phase dependent changes in vestibulo-muscle coupling during dynamic movements using locomotion as my example.   73 4 Time-varying Vestibulo - Myogenic Coupling Extracted using Stochastic Vestibular Stimulation 6   4.1 Introduction The evolution of locomotion from a multipedal to a bipedal task has resulted in a locomotor pattern with increased postural demand. Compared to quadrupeds, humans have a high centre of mass perched over a small base of support resulting in a generally unstable vertical posture. The inherent passive instability generated by this unstable vertical posture poses a challenge for human locomotor activity because the alternating flexor-extensor muscle activity, which propels the body forwards, requires the active maintenance of upright balance to be successful (Misiaszek, 2006). In humans, sensory systems, such as the vestibular system, aid in actively maintaining a vertical posture. During locomotion however, the contribution of these systems becomes less clear. The vestibular system, for instance, has been inferred to contribute to locomotion through the changes in gait trajectory observed both in patients with unilateral vestibular neurotomy (Borel et al., 2004) or acute unilateral vestibular failure (Brandt, 2000) and in participants receiving electric vestibular stimulation (Fitzpatrick et al., 1999; Bent et al., 2000; Fitzpatrick et al., 2006). In fact, most studies have been limited to describing the global effects of vestibular afference on locomotor control (Fitzpatrick et al., 1999; Bent et al., 2000; Fitzpatrick et al., 2006). In felines, in contrast, vestibular signals are modulated in phase with the locomotor step cycle and are believed to  6  A version of Chapter 4 has been published: Blouin JS, Dakin CJ, van den Doel K, Chua R, McFadyen BJ, & Inglis JT (2011) Extracting phase - dependent human vestibular reflexes during locomotion using both time and frequency correlation approaches. Journal of Applied Physiology, 111, 1484-1490.    74 modulate extensor muscle tone (Orlovsky, 1972; Matsuyama & Drew, 2000a, b). Similar modulation has been inferred in humans (Bent et al., 2004) but has only been directly documented once (Iles et al., 2007). Our limited knowledge of the vestibular system’s role in locomotion is partly due to the difficulty in probing and resolving vestibular reflexes during human locomotion. Typically galvanic vestibular stimulation is used to generate a vestibular ex-afferent like signal, which results in biphasic responses in lower-limb muscles actively involved in balance control (Nashner & Wolfson, 1974; Iles & Pisini, 1992; Britton et al., 1993; Fitzpatrick & Day, 2004). The evoked muscular responses are small, requiring the averaging of many responses to clearly identify the vestibular reflex. Thus, identifying these small biphasic vestibulo-motor responses from the on-going locomotor activity can prove challenging (Iles et al., 2007). Indeed, Iles et al. (2007) observed phase-dependent modulation of human vestibular reflexes but had difficulty in resolving both the time course of modulation and the amplitude of the short latency muscle response component. To overcome these limitations I propose to use stochastic vestibular stimulation (SVS) in combination with time-frequency and time- dependent cross-correlation analysis techniques to extract these time varying vestibular reflexes. SVS is an alternative method to galvanic vestibular stimulation (GVS), which has been shown to elicit similar vestibular responses to GVS delivered during human standing balance, but with reduced testing times and decreased postural disturbance (Dakin et al., 2007; Dakin et al., 2010b; Dakin et al., 2011). Compared to GVS, the continuous nature of the stochastic stimulus allows for the use of time-dependent decompositions that might help extract phasically modulated vestibular reflexes during cyclic activities such as human locomotion. Here, I propose to develop a framework to assess phasic modulation of vestibulo-   75 motor responses in humans using SVS. My primary aim is to characterize modulation of human vestibular reflexes across the locomotor step cycle using time-dependent correlations across both the time and frequency domains. In the time domain, time-dependent cross- correlations will be used to describe the shape and polarity of the evoked responses whereas in the frequency domain time-dependent correlations (coherence) will also be used to identify the time period and frequencies over which the SVS stimulus interacts with lower-limb muscle activity. Hypothesis: I hypothesized that vestibular reflexes will undergo phasic modulation during the gait cycle, exhibiting their largest responses during the stance phase, when ankle extensor tone should be at its maximum.  4.2 Methods Subjects Nine healthy subjects (4 females, height 170 ± 9 cm, mass 68 ± 13 kg) between the ages of 21 and 34 years, with no known history of neurological disease or injury participated in this study. Each subject gave their written, informed consent to the procedures approved by the University of British Columbia’s clinical research ethics board.  Stimulus Binaural bipolar vestibular stimulation was delivered using gel coated carbon rubber electrodes (9 cm 2 ) secured over the mastoid processes with an elastic headband. Bandwidth- limited stochastic vestibular signals (0 - 25 Hz; amplitude peak ± 4.5 mA, RMS 1.05 mA) lasting 305 seconds were created using Labview 8.5 software (National Instruments, USA)   76 and delivered using a constant current stimulus isolation unit (Model 2200, AM Systems, USA).  Protocol Participants walked on a treadmill at a speed of 0.4 m/s while maintaining a cadence of 52 steps per minute (guided by a metronome) with their eyes open. This treadmill speed was chosen to replicate the speed used by Iles et al. (2007) in their initial demonstration of phase dependent vestibulo-muscle responses during human locomotion whereas the cadence was matched to that used in previous locomotor studies using electrical vestibular stimulation (Fitzpatrick et al., 1999; Fitzpatrick et al., 2006). Subjects maintained their Frankfurt plane (the auriculo-orbital plane) 18° nose up relative to the floor by keeping a headgear-mounted laser on a target located on a wall 2m ahead of them (Fitzpatrick & Day, 2004; Cathers et al., 2005; Day & Fitzpatrick, 2005). These locomotor parameters and head positions were chosen to maximize the amplitude of the vestibular responses to medio-lateral perturbations (Fitzpatrick et al., 2006; Iles et al., 2007). Volunteers walked while receiving SVS for three trials of 5 minutes. Rest periods were provided between trials at the request of the participant to avoid any signs of fatigue. Across trials, a minimum of 380 strides were performed for each individual.  Electromyography and Signal Analysis Surface electromyography (EMG) was collected bilaterally from the medial head of gastrocnemius using self-adhesive Ag - AgCl surface electrodes (Ambu Blue Sensor M, Denmark). The medial gastrocnemius was chosen because it exhibits large responses to   77 medio-lateral vestibular perturbations during locomotion (Iles et al., 2007). Foot switches were placed bilaterally under the surface of the shoe, near the heel and first toe regions, to estimate bilateral heel strike and toe off during locomotion. EMG was amplified 5000:1and band-pass filtered from 30 Hz to 1000 Hz (Neurolog NL-844, Digitimer, UK), digitized along with the vestibular and foot switches signals at 2 kHz (PXI-6289, National Instruments, USA) and saved on a personal computer for subsequent off-line analyses. Time and frequency correlations between the SVS and EMG signals were performed to estimate the averaged SVS - evoked muscular responses during locomotion (Dakin et al., 2007; Dakin et al., 2010b). Briefly, data from the three individual walking trials were pooled for each subject yielding a total of 447 disjoint sections to perform the analysis (the length of each segment dictates the frequency resolution which was 0.48 Hz). Pooled EMG data were full-wave rectified and then the auto and cross-spectra for the SVS and muscle activity signals were computed for the bilateral gastrocnemius muscles. Frequency domain coherence between the SVS signal and EMG were estimated using a Matlab script for each trial condition within each participant to estimate the linear relationship between the two processes across various frequencies (Rosenberg et al., 1989; Halliday et al., 1995). Cross-correlations between the stochastic vestibular signal and the recorded muscle EMG were determined using the inverse Fourier transform of the cross-spectral estimate (Brillinger, 1974; Rosenberg et al., 1989) and were normalized by the norm of the input vectors to obtain unit less correlation values between -1 and +1 (Dakin et al., 2010b). Confidence intervals (with positive and negative 95% confidence limits) were computed to determine significant values on a subject-by-subject basis. Given the convention for positive vestibular signals corresponding to anode right (cathode left) currents, a positive correlation   78 indicates that anode right (cathode left) currents induced a facilitation of muscle activity. Polarity, peak amplitude and peak latency of the SVS-evoked responses were estimated using cross-correlations. Cross-correlations and coherence were averaged between subjects to represent the global response from all subjects. To identify phasic modulations in the correlation between the vestibular stimulus and muscle activity during the locomotor cycle, coherence, gain and cross-correlations between the SVS and EMG signals were computed as a function of time (using time-dependent coherence, gain and time-dependent cross-correlations respectively). Due to the non- stationarity of the EMG signals during walking, continuous wavelet decompositions of the signals were performed using the Morlet wavelet transform to estimate the time-dependent coherence (Zhan et al., 2006). The SVS and EMG signals were pre-processed prior to estimating the coherence and cross-correlation over time. First, the signals were cut into strides synchronized on the right heel strike (determined from right heel foot switch contact signal). To avoid distortion in the correlations over time at either end of the signals (e.g. around right heel strike), the strides were padded with data from the previous (50%) and subsequent (50%) strides. EMG signals were full-wave rectified and then both EMG and SVS signals were low-pass filtered (100 Hz cut-off, 4 th  order dual-pass Butterworth filter) and resampled at 200 Hz. The time-dependent coherence and gain were estimated using a modified procedure based on the method by Zhan et al. (2006). Time-dependent cross-correlations were estimated using a custom written algorithm which is available as a Matlab freeware at: http://www.cs.ubc.ca /~kvdoel/publications/tc.zip. To account for the variability of stride duration during locomotion I normalized stride duration in time, for the time-dependent cross-   79 correlation and coherence, by re-sampling with respect to the average stride duration across trials. This time normalization was performed on the wavelet spectra for both the SVS and EMG signals as well as their cross-spectrum prior to estimating time-dependent coherence. Time normalization to the mean stride duration was defined as  NTT i N i / 1 ∑ = =                                                   (1)  where T  is the average stride duration, i is the stride number, N  is the total number of strides (380) and iT  is the stride duration for the ith stride. For each stride the scaled stride time τ was defined as iTtT /=τ  which satisfies T≤≤ τ0 . Following stride normalization the time- dependent cross-correlation C (τ ,λ) (λ is the lag time), between x and y as a function of scaled time was determined through the use of the following equation:  2 1 2 1 1 )/(~)/(~ )/(~)/(~ ),( λττ λττ λτ − − = ∑∑ ∑ == = TTyTTx TTyTTx C ji N i ii N i iiii N i                               (2)  where ˜ xi (t)  and similarly, ˜ yi (t) were defined by  Ntxtxtx i N i ii /)()()( ~ 1 ∑ = −=                                         (3)    80 In this calculation the SVS signal for each stride (i=1, …, N; N = 380) is denoted by )(txi  and the EMG signal for each stride (i=1, …, N) is described by )(tyi . Here iy  can represent either the time domain EMG signal or the wavelet coefficient at a particular frequency. The factor TTi /  adjusts for the different stride durations and ensures I am correlating quantities at the same phase in the stride. When computing the correlation I interpolate signals ix  and iy  where necessary, as TTi /τ  is not aligned with the sample times at which they were measured. Note that in the case of the frequency domain analysis C is also a function of frequency. Gain estimates were defined as (van der Kooij et al., 2005):  )( ),( )( xPSD yxCSD H =λ                                                            (4) where ),( yxCSD  is the cross-spectral density of the SVS and EMG signals and )(xPSD  is the power spectral density of the SVS signal. Time-dependent gain estimates were calculated using a similar Morlet wavelet transform and stride normalization as the coherence estimates. For illustrative purposes, time-dependent coherence and time-dependent cross-correlations were averaged between subjects to represent the global response from all subjects and plotted in terms of the percent of stride cycle. To validate the technique two additional analyses were performed. Firstly I examined whether the amplitude of the time-dependent SVS-EMG coherence changed over the testing session.  To do this I determined the peak coherence for the first and last 100 steps of each subject and then compared these values using a paired t-test. Secondly I determined the minimum number of steps required to obtain results that approximate the total 380 steps. To do this coherence was estimated for increasing number of steps in 25 step increments (25, 50,   81 75,… 375). The difference in coherence between the 380 step total and each of the 25 step increments was determined by computing the normative error: ( )∑ −= nCohCoherror 380 2                                                    (5)   where Coh is the time-dependent coherence and n is the lesser step total. The normative error for all step counts between 25 and 375 steps were then compared using repeated measures ANOVA. Since the normative error for the 380 step count is zero it was not compared to the other step totals. The statistical analyses were therefore restricted to determining the highest step count which had statistically greater error than the 375 step count. The minimum steps required to provide results with error comparable to the 375 step count was determined as the step total which was 25 steps larger than the step total which was statistically different from the 375 step count. Statistical significance was set at α = 0.05.  Data Reduction & Statistical Analysis On a subject-by-subject basis cross-correlation and coherence were determined to be significantly different from zero when values exceeded a 95% confidence limit using the methods described by Halliday et al. (1995). Both the confidence limit for the coherence and confidence interval for the cross-correlation were based on the number of disjoint segments (n = 447) used in the spectral averaging (for further detail see (Dakin et al., 2007)). The confidence interval values for the cross-correlations were then normalized by the product of the vector norms of the SVS and EMG signals providing limits appropriately scaled for units of correlation (Dakin et al., 2010b). Similar statistical analyses were used for the time-dependent coherence and cross-   82 correlations (Zhan et al., 2006). Confidence limits for time-dependent correlations (in both the frequency and time domains) were derived from the number of strides (n = 380) performed by each subject. However because of the two-dimensional nature of these correlations, my simulations indicated that a 99% confidence limit represented better an alpha level of 0.05. The significance levels for the time-dependent coherence and cross-correlation were respectively 0.012 and 0.11 for individual subjects. When significant, I extracted the timing of the peak time-dependent coherence, gain and cross-correlation. To determine if the modulation of the vestibular-evoked muscle responses occurred in parallel to EMG modulation during the stride cycle, I compared the timing of the peak time-dependent coherence, gain and cross-correlations to the maximal EMG response during the locomotor cycle using paired t-tests. I also examined if modulation of the short and medium latency muscle responses occurred concurrently by comparing the peak muscle response occurring at a lag of 60 and 150 ms using a paired t-test. The level of significance was set at α = 0.05.  4.3 Results Subjects maintained a repeatable locomotor pattern and did not lose balance despite the presence of the vestibular stimulation. On average, stride duration was 2.24 (0.09 s, while left toe off, left heel strike and right toe off occurred respectively at 15 ± 3%, 50 ± 5% and 64 ± 6% of the stride cycle. All subjects exhibited well-defined correlations between the SVS and EMG signals in both the frequency and time domains (Figure 4.1). Coupling between SVS and the gastrocnemius muscle activity exhibited significant coherence in the 0 - 20 Hz bandwidth for all subjects, with the largest averaged peak occurring at 8.7 ± 4.7 Hz and 7.2 ±   83 3.2 Hz for the right and left medial gastrocnemius muscles. The cross-correlation between the SVS and EMG signals showed significant responses of opposite polarity between legs in all subjects. The early peak correlation between signals occurred at 59 ± 3 ms and 60 ± 3 ms and reached maximal amplitudes of 0.048 ± 0.021 and -0.047 ± 0.14 for the right and left medial gastrocnemius muscles respectively. The medium latency responses were of opposite polarity and occurred at a latency of 146 ± 8ms and 144 ± 7ms, reaching maximal amplitudes of - 0.038 ± 0.020 and 0.039 ± 0.19 respectively for the right and left medial gastrocnemius muscles. In most subjects, a small but significant intermediate response of the same polarity as the medium latency response occurred between the short and medium latency responses (observed in 7/9 and 8/9 subjects, averaged amplitude -0.030 ± 0.019 and 0.029 ± 0.015 for the right and left medial gastrocnemius respectively).      84  Figure 4.1 Coherence and cross-correlations between SVS and medial gastrocnemius muscles. The upper panels illustrate coherence between the SVS and EMG and the lower panels show the corresponding cross-correlations between these signals. The thick black lines represent the average response from all subjects and the thin grey lines illustrate the individual responses from each subject. The horizontal dashed grey lines show the confidence intervals computed for individual subject responses. For the coherence estimates the confidence interval was 0.0067 where as for the cross-correlations the confidence intervals were ± 0.0066. Significant coherence was observed in all subjects in the 0 - 20 Hz bandwidth and cross-correlations were significant around 50 and 150 ms lag, exhibiting opposite polarity between the left and right muscles. l -: left; r -: right, mGAS: medial gastrocnemius.     85 Time-dependent representations exhibited characteristic phasic modulation of SVS-EMG coherence for all subjects. The dynamic modulation of SVS-EMG coherence did not change over time as peak coherence was similar between the first 100 steps and the last 100 steps (paired t-tests: p > 0.05; 0.27 ± 0.07 vs 0.27 ± 0.13 and 0.24 ± 0.08 vs 0.26 ± 0.1 for the right and left medial gastrocnemius respectively). The minimum number of steps required to identify this modulation in single subjects was as few as 250 steps when compared to 375 steps. This was demonstrated by the difference in time-dependent coherence observed at 225 steps for the left (F(14,112) = 106, p < 0.05) and 250 steps for the right (F(14,112) = 176,  p < 0.05) medial gastrocnemius (Figure 4.2).  Figure 4.2 Decrease in time-dependent coherence related error associated with increasing the total steps in the average. These two graphs represent the error between the coherence for the full 380 steps and lesser step totals at 25 step increments (25, 50, 75, …, 375 steps) for both muscles plotted with their standard deviations. The error was defined by equation 5. Time- dependent coherence error estimated with step numbers less than 225 - 250 steps is significantly greater than the coherence error exhibited with 375 steps. Therefore at least 250 steps are required to obtain a coherence difference error that is not different from that of the 375 step total. The star denotes statistical significance. l -: left; r -: right, mGAS: medial gastrocnemius.   86 Maximal time-dependent coherence was observed at 5.5 ± 0.8 Hz and 5.2 ± 0.5 Hz, reaching amplitudes of 0.21 ± 0.11 and 0.22 ± 0.11 for the right and left medial gastrocnemius muscles (Figure 4.3). Coherence increased during the stance phase of the stride cycle when the gastrocnemius muscles were active, reaching maximal values during the portion of single leg support (21 ± 4% and 70 ( 5% of the stride cycle for the right and left medial gastrocnemius respectively) (Figure 4.3). Maximal coherence was observed before background EMG activity reached its peak (paired t-tests: p < 0.001; 21 ± 4% vs 38 ± 5% and 70 ± 5% vs 89 ± 6% of the stride cycle for the right and left medial gastrocnemius respectively). Maximal SVS-EMG gain also peaked before background EMG activity reached its peak amplitude (paired t-tests: p < 0.001; 27 ± 3% vs 38 ± 5% and 74 ± 3% vs 89 ± 6% of the stride cycle for the right and left medial gastrocnemius respectively). Minimal or no coherence was observed between SVS and EMG signals during the swing phase of the locomotor cycle.   87  Figure 4.3 Time-dependent SVS-EMG coherence and gain for the right and left medial gastrocnemius muscles. The upper panels illustrate the averaged time-dependent coherence between the SVS and EMG (n = 9). The middle panels illustrate the average time-dependent gain between the SVS and muscle signals (n = 9) and the lower panels show the modulation of the corresponding muscle during the stride cycle (n = 9). All figures are represented with time 0 showing right heel strike. The color bar represents the amplitude of the coherence, gain and the shaded area, the period of single limb support for the right and left leg. Significant   88 coherence and gain were observed during the period of stance phase of the stride cycle (i.e. while the corresponding ankle extensor was active) both of which reached maximal amplitude before background EMG reached its peak amplitude. Note that for illustrative purposes, the time-frequency representations were shifted by 50ms to account for the delay between SVS and muscle activation (see Figure 4.1 lower panels). l -: left; r -: right, mGAS: medial gastrocnemius.  The time-dependent cross-correlations between SVS and EMG signals exhibited clear phasic modulations of the biphasic muscle responses (Figure 4.4). The polarity of the responses, however, remained constant throughout the stride cycle. Hence, when a response was visible in the left medial gastrocnemius (i.e. during stance phase) it exhibited a negative short-latency response followed by a positive medium-latency response. Responses of opposite polarity were observed in the right medial gastrocnemius. The time-dependent cross- correlations between the SVS and muscle responses exhibited the largest correlation occurring during the rising phase of the background EMG signal. On average, the peak short-latency SVS-EMG cross-correlation occurred earlier than the peak background EMG signal (paired t- tests: p < 0.001; 23 ± 8% vs 38 ± 5% and 71 ± 5% vs 89 ± 6% of the stride cycle for the right and left medial gastrocnemius respectively). On the other hand, the maximal SVS-EMG cross-correlation occurred at similar times for the short and medium latency responses (paired t-tests: p > 0.5; 23 ± 8% vs 23 ± 10% and 71 ± 5% vs 70 ± 7% of the stride cycle for the right and left medial gastrocnemius respectively).    89  Figure 4.4 Time-dependent cross-correlations between SVS-EMG for the right and left medial gastrocnemius muscles. The upper panels illustrate the averaged time-dependent cross- correlation between the SVS and EMG (n = 9) and the lower panels show the modulation of the short and medium latency responses during the stride cycle (n = 9). For the upper panels, the black lines along the Time (ms) axis represents the averaged cross-correlation for the stride cycle (see Figure 4.1) while the black lines along the Stride Cycle (%) axis represents   90 the modulation of muscle activity during the stride cycle. Maximal cross-correlations were observed during the stance phase of the stride cycle (i.e. while the corresponding ankle extensor was active) but reached maximal correlation before background EMG reached its peak amplitude. The transparent rectangles represent the time lags for which the short (light grey) and medium (dark grey) latency response are represented for the lower panels. The lower panels illustrate the concurrent modulation of the short and medium latency responses during the stride cycle. All figures are represented with time 0 showing right heel strike. The color bar represents the amplitude of the cross-correlation. Note that for illustrative purposes, the time-dependent cross-correlations were shifted by 50 ms to account for the delay between SVS and muscle activation. l -: left; r -: right, mGAS: medial gastrocnemius, SL: short latency, ML: medium latency.  4.4 Discussion The aim of the current study was to determine if time and frequency domain correlation techniques could be used to investigate human vestibular reflexes during locomotion. I hypothesized that, similarly to felines, vestibular reflexes would be simultaneously modulated with extensor tone across the gait cycle, exhibiting their largest responses in correspondence with peak extensor muscle activity. Partly supporting my hypothesis, analysis of the SVS-EMG correlations as time-dependent processes permitted the identification of phase-dependent modulation of the vestibular reflexes during the stride cycle. As such, vestibular information from the stimulus exerts its influence over agonist gastrocnemius muscle activity early in stance phase, eliciting a multi-phasic muscle response. In contrast to my initial hypothesis however, the period of maximal correlation between the   91 vestibular stimulus and muscle activity did not correspond to the period of maximal EMG. These results demonstrate usefulness of this technique in extracting the modulation of vestibular reflexes over a cyclic task such as locomotion. Phase dependent modulation of vestibular signals over the gait cycle has been documented in several species (Orlovsky, 1972; Marlinsky, 1989, 1992; Matsuyama & Drew, 2000a, b; Iles et al., 2007). These studies have suggested vestibular signals act to shape both the timing (In felines: (Russell & Zajac, 1979; Udo et al., 1982)) and magnitude (In guinea pigs: (Marlinsky, 1989, 1992) of muscle activity during locomotion. In humans, phasic modulation of vestibular input to muscles has been directly observed only once, with limited resolution (Iles et al., 2007). My results correspondingly demonstrate that vestibular input is indeed modulated with the phase of the gait cycle and the period of vestibulo-muscular coupling corresponds to the period of extensor muscle activation. There are however, differences in the amplitude profile of the SVS-EMG coherence, gain and the gastrocnemius EMG over time. In the right gastrocnemius, for example, the increase in SVS-EMG coherence and gain initially parallels the rise in muscle activation peaking at 21 and 27 percent of the gait cycle but then falling back to baseline levels at the 50 percent point of gait cycle. EMG, in contrast, continues to increase reaching a maximum at the 38 percent point of the gait cycle. The discordance between SVS-EMG coupling and EMG amplitude indicates a) vestibular reflex amplitude is not purely dependent on the level of excitation of the motor neuron pool and b) vestibular ex-afference does not appear to provide a large contribution to peak gastrocnemius muscle activation, instead contributing to the early rise in extensor activation.  As well, the difference in timing between peak SVS-EMG coherence and peak EMG might be suggestive of the functional role of vestibular input to the gastrocnemius   92 muscles during locomotion. Since the maximal influence of vestibular input to the gastrocnemius motor neurones was prior to maximal activation of the muscle, a potential role of the gastrocnemius to a vestibular error could be to stabilize the ankle prior to push-off. In addition, Bent et al. (Bent et al., 2004) have proposed that vestibular information plays an important role in whole body stabilization during locomotion and might influence foot placement. My results support this proposal as the high SVS-EMG coherence early in the single support phase is appropriately timed to influence medio-lateral placement of the contra lateral foot. The improved resolution of the technique presented here also allows a closer examination of vestibular reflex polarity across time. As seen in figure 4.4, vestibular stimulation during locomotion induces multi-phasic muscle responses reminiscent of those induced during standing balance (Nashner & Wolfson, 1974; Iles & Pisini, 1992; Britton et al., 1993). During standing balance vestibular stimulation elicits a biphasic waveform which has been described as containing a short (50-70 ms) and medium (100-120 ms) latency component (Nashner & Wolfson, 1974; Iles & Pisini, 1992; Britton et al., 1993; Fitzpatrick et al., 1994; Dakin et al., 2007). The waveform induced by vestibular stimulation during walking also appears to contain a short latency component. The medium latency component, however, appears more complex exhibiting its own multi-phasic waveform. During standing balance the two components of the vestibular response have been described as being derived from independent spinal pathways (vestibulo and reticulo spinal) (Britton et al., 1993) or originating from separate vestibular organelles (the otoliths or semicircular canals) (Cathers et al., 2005). However co-variation in the amplitude of these two components observed during locomotion does not allow us to distinguish between these two processes.   93 The current approach provides several advantages over the more traditionally used GVS method when characterizing dynamically modulated responses in a cyclic task. Firstly, changes in both the spatial and temporal parameters of the response can be observed as they occur in time. This is evident in both the strength of vestibulo-muscular coupling over the gait cycle (Figure. 4.3) as well as the bilateral modulation of reflex polarity (Figure. 4.4). Secondly, the continuous nature of the stimulus eliminates the need to time lock multiple stimuli to specific events in the task cycle thus reducing testing time. Finally, the disruption of the locomotor pattern caused by the postural response to the stimulus can be reduced. In the present experiment, subjects maintained a stereotypical locomotor pattern and did not lose balance. In addition, it is possible to adjust the parameters of the SVS to minimize any balance disturbance while eliciting the proper vestibular responses in the muscles of interest (Dakin et al., 2010b). Since the largest balance responses to vestibular stimulation are associated with frequencies of stimulation below 2 Hz (Dakin et al., 2010b), and the bandwidth of significant coherence observed in the medial gastrocnemius begins at 2 Hz, these frequencies can be easily removed from the stimulus. There are however also some limitations to this technique. One of the main limitations of the current approach is that, although phase dependent modulation of human vestibular reflexes is revealed, the locus of such modulation cannot be identified. For the current results, it is not possible to determine whether the phasic modulation of vestibulo-motor responses occurs due to inhibition of vestibular signals in the brainstem, inhibition by the cerebellum or due to modulation of the vestibulo-motor action at the spinal level. Currently, the timing of vestibular influence over muscle activity is believed to be in part determined by activity arising from or passing through the cerebellum, as its   94 removal, in felines, attenuates gait cycle specific modulation (Orlovsky, 1972). Presumably, inhibitory cerebellar purkinje cells exert their influence over central vestibular neurons in the lateral vestibular nucleus resulting in the phase dependent behaviour of vestibulo-muscle interactions during gait (Walberg & Jansen, 1961; Ito & Yoshida, 1964). In addition, the description of SVS evoked vestibular activity itself as an ex-afferent signal can also present limitations. Most sensory disturbances which create an ex-afferent signal also exhibit congruent input from multiple sensory modalities. With SVS the vestibular response is incongruent with both visual and somatosensory inputs and therefore could be interpreted by the body in a manner different than had the vestibular input been in congruence with visual and somatosensory inputs. Controlling locomotor speed by using a treadmill might present some further limitations. By constraining locomotor speed participants were required to walk at cadences which differed in various amounts from their natural cadence. These parameters were chosen to match those used in previous experiments (Fitzpatrick et al., 1999; Fitzpatrick et al., 2006; Iles et al., 2007), however, the likely slower than preferred locomotor speed likely resulted in alterations in participant’s gait mechanics as well as alterations in the balance requirements of the task. My final limitation relates to the limited analyses presented here. For the sake of brevity, analyses have been limited to a single muscle (bilaterally) and only for roll balance perturbations. Outside of the lower leg, the distribution and influence of vestibular signals on lower-limb motor control is still unknown. Future work needs to focus these techniques on examining widespread vestibular contributions to dynamic muscle activity in humans.     95  4.5 Conclusions The results of the current study demonstrate that time and frequency domain correlation techniques can be used to investigate human vestibular reflexes during locomotion and provide a basic framework for extracting reflex modulation to continuous randomly varying stimuli. This approach was effective in extracting the phase dependent modulation of vestibulo-muscular coupling during a cyclic task, revealing that the period of strongest vestibulo-muscular coupling occurred during single leg support prior to the period of maximal electromyographic activity in the gastrocnemius. Together these findings suggest a functional role of the gastrocnemius muscle to a vestibular error prior to the push-off phase of locomotion.   96 4.6 Summary: Study 3  The aim of study three was to develop a methodological framework for the application of the stochastic stimulus to dynamic movements. By correlating the muscle responses to the stimulus in both the time and frequency domains phasic modulation of vestibulo-muscle coupling was very clearly revealed. Also identified was the response of the muscle to the stimulus at each point in the step cycle indicating that, similar to standing balance, vestibular stimuli evoke a biphasic response during locomotion. The delivery of a continuous stochastic stimulus coupled with appropriate time and frequency based analytical measures was shown to very successfully identify the influence of vestibular ex-afferent signals on muscle activity during motion. This third study concludes the methodologically focused first volume of this thesis and serves as a bridge to bring focus on vestibular physiology during dynamic motion: the aim of the second volume of this thesis. The success of the stochastic vestibular stimulus in extracting changes in vestibulo-muscle coupling over the locomotor cycle provides the opportunity to explore vestibular function in contexts which have traditionally proven difficult. One such context is locomotion. Our understanding of how vestibular ex-afferent signals are used to stabilize locomotion is limited partially due to the techniques available to probe and extract dynamic vestibular function. A second context is during head rotation. The spatial transformation of vestibular signals is poorly understood and it remains unclear whether vestibular signals are dynamically transformed during head movement. The aim of Volume Two of this thesis was to examine the physiology of dynamic vestibular function which has been made possible by the advances of the previous three studies.   97 VOLUME 2 Stochastic Vestibular Stimulation and its Application to Dynamic Vestibular Physiology  Volume Synopsis  Volume Two used the advances in SVS developed in Volume One to answer physiological questions regarding human vestibular function. The first study had two aims: to examine the timing, polarity and distribution of vestibulo-muscle interactions during locomotion and to examine the suppression of vestibular ex-afference with increasing cadence and locomotor speed. The second study examined a new dynamic context namely the spatial transformation of vestibular ex-afference during head rotation. In this study I identified whether or not vestibular ex-afferent signals were attenuated during head rotation and examined how the residual signals are continuously transformed during head rotation. Ultimately these two chapters in Volume Two provided the first clear view of the dynamic contribution of vestibular ex-afferent signals to stability in humans.   98 5 Vestibular Ex-afference Modulation during Locomotion and its Suppression with Increased Locomotor Velocity and Cadence  5.1 Introduction The vestibular system provides an essential sensory contribution to the stabilization of balance during human locomotion. When this system is impaired locomotion can become unstable and in severe cases lead to falls (Brandt, 2000). Currently, our understanding of how vestibular information is used to stabilize the body during locomotion remains generalized to full body compensatory responses resulting from electrically induced vestibular errors (Fitzpatrick et al., 1999; Bent et al., 2004; Fitzpatrick et al., 2006). The role individual muscles play in contributing to this compensatory response remains unclear. Research in animal models would suggest vestibular influence over muscle activity is modulated with the step cycle (Orlovsky, 1972; Matsuyama & Drew, 2000a), but only recently have phase dependent vestibular responses been identified in humans. Iles et al. (2007) observed vestibular induced motor responses during the stance phase of the step cycle but only in muscles acting around the ankle. In contrast, growing evidence suggests vestibular influence should be more widespread than just at the ankle. Cats have groups of vestibular neurons whose firing rates vary in phase with a variety of lower-limb muscles, not only those acting around the ankle (Orlovsky, 1972; Matsuyama & Drew, 2000a). In humans, compensatory responses to vestibular stimuli encompass the whole body in a manner similar to standing balance (Day et al., 1997; Bent et al., 2004; Fitzpatrick et al., 2006). This resemblance suggests that, similarly to during standing balance, vestibular input during locomotion should influence all muscles currently engaged in the act of maintaining stability   99 (Britton et al., 1993) and therefore, should be present in active muscles throughout the body and modulated in phase with muscle activity over the step cycle. Phase dependence is one mechanism by which the vestibular system could contribute to balance stabilization during locomotion; the attenuation of vestibular afference with increased walking speeds could be another. Patients with acute unilateral vestibular neuritis normally exhibit large directional errors when they walk with their eyes closed however when these patients run the directional errors decrease (Brandt et al., 1999; Brandt, 2000; Borel et al., 2004). This apparent immunity to vestibular-related directional error while running has led to the proposal that vestibular afference is increasingly suppressed with rising locomotor velocity. Similar behaviour has since been demonstrated in healthy subjects as compensatory sway responses to galvanic vestibular stimulation are also attenuated when running as compared to walking (Jahn et al., 2000). An assumption these studies have made is that the vestibular signals themselves are attenuated; however one factor that was not considered and might contribute to an attenuated vestibular influence is the duration the legs are in contact with the ground. As locomotor velocity increases both cadence and the proportion of time stance phase contributes to the step cycle decrease, reducing contact time with the ground (Mann & Hagy, 1980). This reduction in contact time with the ground reduces the time available for muscle activity to generate torque therefore requiring larger muscle response amplitude to generate similar changes in momentum.  Since vestibular influence over ongoing muscle activity (such as those from galvanic vestibular stimulation) is relatively weak a reduction in the time available to develop torque, without a concomitant increase in gain, will result in a reduced postural response to an increase in cadence. Here I investigate the contribution of vestibular signals to locomotor stability a) by   100 examining the prevalence and phase dependent modulation of vestibular coupling with muscles of the hips and lower limbs and b) by examining the potential suppression of vestibular ex-afferent signals during locomotion at increased velocities. I hypothesized that a) all muscles engaged in the control of stability during locomotion will exhibit periods of coupling with the vestibular stimulus at periods over the step cycle and b) the magnitude of vestibulo-myogenic coupling will decrease both to increases in cadence and increases in locomotor velocity.  5.2 Methods Subjects Nine healthy subjects (height 170 ± 9 cm, mass 68 ± 13 kg, 4 female subjects) between the ages of 21 and 34 yrs participated in this study. The experimental protocol was explained to each subject and their written, informed consent obtained. All procedures used in this study conformed to the standards of the Declaration of Helsinki and were approved by the University of British Columbia’s clinical research ethics board.  Stimulus Electric vestibular stimulation is the percutaneous application of a small electric current which modulates the firing rate of the underlying vestibular nerve. The effect of this type of stimulation is to provide an isolated ex-afferent like vestibular signal but absent congruent multi-modal feedback. When delivered in a bipolar binaural electrode configuration with the head facing forward the postural response to electric vestibular stimulation is a roll response in the frontal plane (Fitzpatrick & Day, 2004). In the current study the vestibular   101 stimulus was delivered using a binaural bipolar electrode configuration with carbon rubber electrodes (9 cm 2 ), secured over the mastoid processes using an elastic headband. Bandwidth limited stochastic stimuli (SVS; 0-25Hz, peak amplitude of ± 4.5 mA, root mean square 1.05mA) lasting 305 s were created using LabVIEW software and were delivered using an isolated constant-current stimulation unit (model 2200, AM Systems).  Test Procedures Participants walked on a treadmill, metronome guided, at two speeds (0.4 m/s and 0.8 m/s) and two cadences (52 steps/min and 78 steps/min) for a total of three trial conditions (0.4 m/s @  52 steps/min; 0.4 m/s @ 78 steps/min and 0.8 m/s @ 78 steps/min). The 0.4 m/s treadmill speed replicates the speed used by Iles et al (17) in their initial demonstration of phase-dependent vestibular responses during locomotion. The 0.8 m/s velocity was chosen to provide a 100% increase in locomotor speed. The cadence of 52 steps/min was also chosen to match that used in previous locomotor studies incorporating vestibular stimulation (Fitzpatrick et al., 1999; Fitzpatrick et al., 2006) and the 78 steps/min chosen to provide an increase in cadence but one which is reasonably comfortable for participants to perform given the relatively slower locomotor velocities. Participants maintained their Frankfurt plane 18º nose up from the floor by keeping a headgear mounted laser on a target located 2 meters in front of them (Fitzpatrick & Day, 2004; Cathers et al., 2005; Day & Fitzpatrick, 2005). The locomotor parameters and head position were chosen to maximize the amplitude of vestibulo- motor responses in the medio-lateral directions (Fitzpatrick et al., 2006; Iles et al., 2007). Periods in the step cycle which exhibit strong vestibular responses will be interpreted as times in which vestibular information is useful for the medio-lateral stabilization of locomotion by a   102 given muscle. Subjects walked while being provided the stochastic stimulus for three 5 minute trials at 52 steps/min and two five minute trials at 78 steps/min in order to collect a minimum of 350 strides per condition. Participants were provided rest periods between trials and at their request to avoid fatigue.  Electromyography and Signal Analysis  Surface electromyography (EMG) was collected from eight muscles: bilaterally from the medial gastrocnemius, and from the right lateral gastrocnemius, soleus, tibialis anterior, biceps femoris, rectus femoris and gluteus medius. Self adhesive Ag/AgCl surface electrodes (Ambu Blue Sensor M) were positioned over the muscles of interest after cleaning and abrading of the skin. These muscles were chosen to provide a summary of vestibulo-motor interactions throughout the leg and hip muscles. Footswitches were fastened to toe and heel of each foot and used to estimate heel strike and toe off for each stride. EMG was amplified (×5,000), band-pass filtered from 30 to 1000 Hz (Neurolog NL-844, Digitimer) and digitized along with the vestibular stimulus and foot switches at 2 kHz (PXI-6289, National Instruments). All data were saved on a personal computer for later offline analyses.  Coherence and cross-correlation between the vestibular stimulus and EMG were computed as a function of time to estimate the phasic modulation between the vestibular stimulus and EMG over the gait cycle as in Chapter 4 (Blouin et al. 2011). Time-dependent coherence was estimated using Morlet wavelet decomposition due to the non-stationarity of the EMG signal (Zhan et al., 2006). Prior to wavelet decomposition, the SVS and EMG signals were cut into strides synchronized to the right heel strike (identified by contact in the right heel switch). To avoid distortion in the correlations at heel strike each stride was padded   103 at the start and end with data from the previous (50%) and subsequent (50%) strides. EMG signals were also full wave rectified and both SVS and EMG signals low pass filtered (100 Hz fourth-order dual-pass Butterworth filter) and re-sampled at 200 Hz for data reduction.  Time-dependent coherence was estimated using a modified procedure based on the method described by (Zhan et al., 2006) and time-dependent cross-correlations estimated using a method described by Blouin et al. (2011). To account for the stride-to-stride variability I normalized the stride duration in time by re-sampling each stride to the average stride duration across trials. Stride duration normalization was performed on the auto-spectra of the SVS and EMG signals, as well as their cross-spectrum, prior to estimating coherence. Mathematical derivation of the normalization and time-dependent cross-correlation procedures are presented in Blouin et al. (2011). Briefly, data were first re-sampled to 200 Hz. Cross-correlations were then calculated at each sample in the step cycle providing a window of correlation from -50 ms prior to 300 ms following the stimulus at each point in the step cycle. The cross-correlations were then normalized to provide values of correlation between - 1 and 1 used to identify both the timing and time course of the SVS-EMG coupling in all recorded muscles. For illustrative purposes, time-dependent coherence for each muscle was averaged across all subjects to provide a representation of the global response.  Data Reduction and Statistical Analyses SVS-EMG coherence was used to identify phase dependent coupling between SVS and muscles of the lower limbs and hips. Significant coupling was defined in each subject as the period over the gait cycle in which SVS-EMG coherence exceeded a confidence limit set at P = 0.01 (corresponding to a coherence magnitude of 0.013). This value was chosen   104 because it better represents an α level of 0.05 due to the bi-dimensional nature of these correlations. The interval over which SVS-EMG coherence exceeded this confidence limit was determined on a subject by subject basis. To determine if vestibular drive to the lower leg muscles is suppressed with higher walking speeds and cadences the maximum coherence for each walking condition was determined. Coherence was chosen to estimate the magnitude of vestibular responses because it is normalized by the power in both the SVS and EMG signals, attenuating the effect of changes in EMG magnitude related to walking speed or cadence (Nilsson et al., 1985; Yang & Winter, 1985) on the magnitude of SVS-EMG coupling. Gain, on the other hand, is not normalized by the EMG signal amplitude and therefore will rise and fall with changes in EMG amplitude not related to increases or decreases in SVS-EMG coupling. Peak coherence was then compared between the two cadences and two locomotor speeds using paired t-tests. The effect of cadence was determined by comparing the 0.4 m/s @ 52 steps/min condition to the 0.4 m/s @ 78 steps/min condition whereas the effect of locomotor velocity was established by comparing the 0.4 m/s @ 78 steps/min and 0.8 m/s @ 78 steps/min conditions. I did not use a full factorial design because a cadence of 52 steps/min was too slow to perform with a walking speed of 0.8 m/s. Since I anticipated a reduction in SVS-EMG coupling with increasing cadence and velocity I used a one-tailed test p = 0.05.  5.3 Results General Observations  Phase dependent modulation of EMG was visually apparent across all participants in all conditions and in general, the magnitude of each muscle’s contribution varied both with cadence and walking speed (Figure 5.1).   105   Figure 5.1 Averaged muscle activity for each of the three trial conditions (n = 9). Phase dependent modulation of muscle activity is unique in each trial condition and peak EMG tends to be highest in the 0.8 m/s @ 78 steps/min condition. r-MG: Right medial gastrocnemius; r-LG: Right lateral gastrocnemius; r-Sol: Right Soleus; r-TA: Right tibialis anterior; r-RF: Right Rectus Femoris; r-BF: Right biceps femoris; r-GM: Right gluteus medius; l-MG: Left medial gastrocnemius; mV: milliVolts.     106 Significant SVS-EMG coupling was observed in all recorded muscles (Figure 5.2). Muscles in the lower leg exhibited coupling only just prior to and during the stance phase of the locomotor cycle while muscles of the thigh and hip exhibited periods of coupling throughout the gait cycle. Overall, the largest average peak SVS-EMG coherence was observed in the medial gastrocnemius (0.21 ± 0.11; 0.4 m/s @ 52 steps/min) while the weakest was observed in the rectus femoris (0.07 ± 0.03; 0.8 m/s @ 78 steps/min). Significant SVS-EMG coupling in the medial gastrocnemius began at heel strike peaking early in mid stance phase and ending late in mid stance. As expected, coupling in the left medial gastrocnemius was the mirrored response to the right medial gastrocnemius. The lateral gastrocnemius, in comparison, exhibited significant SVS-EMG coupling beginning in mid stance and ending at toe off. The right soleus exhibited a pattern distinct from the right gastrocnemius exhibiting SVS-EMG coupling for the duration of stance phase, e.g. from heel strike to toe off. On the anterior of the leg, the tibialis anterior exhibited coupling from just prior to heel contact until early mid-stance. Muscles of the thigh and hip generally exhibited different patterns of coupling from those observed in the lower leg. The rectus femoris exhibited very little SVS-EMG coupling excluding a brief period around heel strike prominent only in the 0.4 m/s @ 78 steps/min condition. In contrast, the biceps femoris and gluteus medius exhibited SVS-EMG coupling during heel contact and early stance as well as at toe off. The gluteus medius also exhibited coupling during late swing phase extending through heel contact, peaking during early stance and ending in mid stance then reappearing briefly just prior to toe off. Overall when SVS- EMG coupling is averaged across all the muscles recorded in the right leg and the lone medial gastrocnemius from the left leg it appears vestibular ex-afferent signals influence muscle   107 activation over the entire gait cycle (Figure 5.2, bottom). If all the muscles presented here had been recorded bilaterally this effect would be even more prominent across the entire gait cycle.   Figure 5.2 Coherence plotted for each muscle in each walking condition (n = 9). A. Coherence for each muscle in the 0.4m/s @ 52 steps/min condition. B. Coherence for each muscle in the 0.4m/s @ 78 steps/min walking condition and C. Coherence for each muscle in the 0.8 m/s @ 78 steps/min walking condition. For illustrative purposes only the bandwidth of max coherence (6.2 ± 0.6Hz, n = 9, all muscles) and data above a significance level of 0.01 are displayed. The bottom row in each plot shows the average coherence across all muscles recoded and illustrates that the stimulus influences muscle activity over the entire step cycle. Coherence amplitude is indicated by the color bar.    108 During periods of significant SVS-EMG coupling each recorded muscle produced a bi or tri-phasic response in the time-dependent cross-correlation. The average timing of the peaks of these responses across all recorded muscles was 59 ± 8 ms for the short latency response and 118 ± 27 ms for the medium latency response. At heel contact, all muscles in the right leg exhibited positive followed by negative going waveforms. In most muscles response polarity was consistent over the entire step cycle however in three muscles the polarity of these responses reversed at different points in the step cycle. The biceps femoris response polarity reversed just prior to toe off, the tibialis anterior polarity reversed in late stance also just prior to toe off and the rectus femoris response reversed at right heel contact (Figure 5.3).   Figure 5.3 Coherence and cross-correlation for four muscles (n = 9) over the 0.4m/s @ 52 steps/min condition. A. Time frequency coherence displaying the phase dependent coupling between the stimulus and muscle EMG. Frequency is plotted on the ordinate axis.  B. Time   109 dependent cross-correlations corresponding for these four muscles. In the soleus and gluteus medius response polarity is similar over the duration of the step cycle whereas in the tibialis anterior and biceps femoris response polarity inverts near toe off (40-60 percent of the step cycle). Time lag (s) relative to the stimulus is plotted on the ordinate axis. Color bars indicate the coherence amplitude and correlation amplitude and for illustrative purposes data below the p = 0.05 confidence limit has been excluded. r-Sol: right soleus, r-TA: right tibialis anterior, r-GM; right gluteus medius, r-BF; right biceps femoris.  SVS-EMG Coupling Attenuation with Increases in Locomotor Velocity and Cadence  To address my second hypothesis I examined whether increases in locomotor cadence or velocity decreased vestibular coupling with lower-limb muscles. I found that increasing both locomotor cadence (0.4m/s @ 52 steps/min to 0.4m/s @ 78 steps/min) and velocity (0.4 m/s @ 78 steps/min and the 0.8 m/s @ 78 steps/min) reduced SVS-EMG coupling in specific muscles of the lower limb (Figure 5.4). However a reduction of SVS-EMG coupling was not observed in all muscles. Muscles of the thigh, hip and lateral gastrocnemius generally did not show signs of reduced coupling at higher cadences or locomotor velocities. Muscles that did show reduced SVS-EMG coherence were located around the ankle and on average exhibited a decrease of between 15 to 31 percent in coherence. The right medial gastrocnemius exhibited a 25 percent decrease in SVS-EMG coherence due to cadence but was not significantly influenced by locomotor velocity [Cadence: t(8) = 2.22, P = 0.029; Velocity: t(8) = 0.41, P = 0.35]. In contrast SVS-EMG coherence in the left medial gastrocnemius decreased to both cadence (20 percent) and velocity (15 percent) [Cadence; t(8) = 2.80, P = 0.011; Velocity: t(8) =   110 2.41, P = 0.02]). And in similar fashion, the tibialis anterior also exhibited reduced SVS-EMG coherence to both cadence (31 percent reduction) and locomotor (25 percent reduction) velocity ([Cadence: t(8) = 3.30, P = 0.007; Velocity: t(8) = 2.39, P = 0.02]).  5.4 Discussion  The aim of the current study was to identify the periods in which vestibular ex-afferent signals influence lower-limb muscle activity during locomotion and to identify whether or not this influence wanes with increases in locomotor velocity or cadence. I found all muscles examined exhibited periods of modulation by the vestibular stimulus and that when this SVS- EMG coupling is averaged across all muscles recorded vestibular ex-afferent like signals exert influence over the entire stride cycle. Lastly both cadence and locomotor velocity appear to suppress vestibular-muscle coupling however this effect is muscle-specific, affecting only muscles acting around the ankle.  Phasic Modulation of Vestibular Responses is Ubiquitous in the Lower Limbs  Phase dependent modulation of vestibular input to limb muscles has been observed both in animal models and humans (Orlovsky, 1972; Matsuyama & Drew, 2000a; Iles et al., 2007). One of my objectives was to revisit this phasic modulation in humans and determine if it is more widespread than had been initially reported. I observed SVS-EMG coupling in all measured muscles suggesting widespread influence of vestibular ex-afferent signals over the locomotor cycle.   111  Figure 5.4 Change in coherence across trial conditions. Both cadence and locomotor speed reduced peak coherence but only in the tibialis anterior and medial gastrocnemius. The grey dots indicate each subject’s mean and the black dots the group mean. The error bars indicate one standard deviation.    112  My findings extend those of Iles et al. (2007), who observed phase-dependent responses only in muscles distal to the knee, with the observation of phase dependent vestibular coupling in rectus femoris, biceps femoris and gluteus medius. Presumably, the differences in findings between these two studies are due to the stimulus format and advantages related to the analysis approach (Blouin et al., 2011). Overall, the observation of widespread phasically modulated vestibulo-muscle interactions is congruent with the full body postural response typically observed to electric vestibular stimulation and it indicates compensatory actions to a vestibular perturbation are dependent on the phase of the locomotor cycle.  Vestibular Signals Modify Thrust, Medio-lateral Placement of the Foot and Double support to contribute to Locomotor Stability  SVS-EMG coupling in the gluteus medius in particular, suggests compensation for a vestibular induced roll in the frontal plane is conducted not only by thrust provided by the muscles acting around the ankle (Iles et al., 2007) but also through an abduction action of the leg just prior to and following heel contact. This observation corresponds well with the proposed role of both the abductors and adductors in frontal plane stabilization during locomotion (MacKinnon & Winter, 1993; Winter, 1995). In addition, the primary period of gluteus medius SVS-EMG coupling was immediately following heel strike during the double support stage. This period of heightened SVS-EMG coherence loosely corresponds in time with increased coupling in the tibialis anterior, soleus and medial gastrocnemius suggesting the double support phase is important for compensation of frontal plane vestibular   113 perturbations. Indeed Lyon and Day (2005) proposed that double support phase might be of equal or greater importance than medio-lateral placement of the foot for direction changes during locomotion. Taken together it is likely some combination of thrust control, lateral placement of the foot and adjustment of body motion during double support is exploited to maintain stability in the presence of a vestibular perturbation in the frontal plane. The form of compensation to a vestibular perturbation is a generalized whole body response (Day et al., 1997) and therefore it is not surprising that SVS-EMG coupling is widespread in muscles of the lower limbs. Despite the limited number of muscles examined here, it is clear that vestibular ex-afferent signals can influence gait over the entire stride cycle. However, this influence might not be uniform in magnitude across the entire gait cycle. My data suggest that the largest net response to a vestibular stimulus occurs early during the stance phase of the locomotor cycle. This observation mirrors those of Bent et al, (2004) who found postural deviation to vestibular stimulation during locomotion was largest when the stimulus was provided at heel strike. Presumably, the timing of these vestibular responses is tied to two important factors: the activity level of the muscles and the mechanical effect of muscles’ actuation. Vestibular stimulation induces relatively small muscle responses which are usually not large enough to induce activity in a quiescent muscle on its own (Fitzpatrick et al., 1994). These responses are therefore likely to be observed in muscles already active and absent in muscles not active, such as the soleus during swing phase. Secondly, whether the mechanical action of the active muscle aids in compensation to a virtual roll perturbation induced by SVS or not likely plays a role in determining the timing or presence of these responses. Presumably these pathways have evolved to provide effective compensation to vestibular perturbations. Therefore SVS-EMG coupling is likely strongest during periods   114 when these muscles can produce effective or efficient compensation to the vestibular perturbation, similar to what occurs in these muscles to a mechanical perturbation (Misiaszek, 2006).  Vestibular Influence on Muscle Activity can Reverse over the Step Cycle  The functional mapping between the vestibular signals and the corresponding response also appears to vary across the step cycle in some muscles. At right heel contact all recorded muscles in the right leg exhibit an inhibitory medium latency response. Near right toe off and during swing the biceps femoris, tibialis anterior and rectus femoris experience periods where the medium latency response becomes excitatory (Figure 5.3). This behaviour is similar to reflex reversals that have been observed previously in animals (Forssberg et al., 1977; DiCaprio & Clarac, 1981; Akay & Buschges, 2006) and to sural and tibial nerve stimulation in humans (Duysens et al., 1990; Yang & Stein, 1990; Duysens et al., 1992). In these studies, reflex reversals are thought to be caused by peripheral sensory feedback or central pattern generators modifying input to the muscle at different phases of the step cycle. Coincidentally the pattern of response reversal observed to vestibular stimuli in the tibialis anterior closely resembles responses in the same muscle to sural and tibial nerve stimulation which, in humans, has been suggested to result from the actions of a central pattern generator (Duysens et al., 1990; Duysens et al., 1992). This explanation would fit some of these data as reversals in both the tibialis anterior and biceps femoris occur during the stance phase of the locomotor cycle and not during an event producing an obvious change in peripheral sensory feedback, such as toe off or heel contact. In contrast, the rectus femoris response, in the few subjects in   115 which it was evident, reversed at heel strike suggesting it could be due to either a central pattern generator or a load related peripheral source such as Ib feedback from Golgi tendon organs (Iles & Pisini, 1992). Regardless of the potential source of these reversals, it is clear vestibular influence is modulated both in its timing of influence and, in some muscles, its polarity over the step cycle.  Vestibular Influence is Selectively Suppressed with Increases in both Cadence and Locomotor Speed  In some lower-limb muscles SVS-EMG coupling also decreases as cadence or locomotor velocity increases. This result parallels previous studies (Brandt et al., 1999; Brandt, 2000; Jahn et al., 2000) which observed a decrease in vestibular-related heading error in both patients with acute unilateral vestibulopathy or subjects administered galvanic vestibular stimulation while running compared to walking. In the present study however, only the medial gastrocnemius and tibialis anterior exhibited this behaviour while the soleus, lateral gastrocnemius, gluteus medius and biceps femoris exhibited no decrease in SVS-EMG coupling. This finding implies responses to vestibular ex-afferent signals might not experience a general ‘down-regulation’ at these lower walking speeds but rather local suppression or inhibition in the spinal cord resulting in muscle specific suppression of SVS-EMG coupling. This suppression is likely important with increasing locomotor velocity because responses to vestibular ex-afferent input reach lower-limb muscles with a fixed time delay (as short as 50- 70ms to reach the muscle, 59 ± 8 ms in this study, and 300-500ms for peak force production) (Nashner & Wolfson, 1974; Britton et al., 1993; Fitzpatrick et al., 1996; Day et al., 1997;   116 Fitzpatrick & Day, 2004; Dakin et al., 2007; Day & Guerraz, 2007; Dakin et al., 2010b)). Without the introduction of a phase advance or a reduction in gain of the vestibular input these responses would occur later and later in the step cycle inappropriately influencing foot plantar flexion and potentially leading to falls.  This is not the first evidence of regionally specific vestibular influence during locomotion. Bent et al. (2004) observed phase dependent vestibular induced postural responses during locomotion but only in the lower limbs. The differing results for the upper and lower bodies were said to result from separate functional roles: in the upper body vestibular information is used for stabilization of the head in space whereas in the lower limbs vestibular information is used for appropriate placement of the feet. However, at higher locomotor velocities vestibular information might be poorly timed to positively contribute to ankle position (leading to effects such as toe drag) therefore its suppression with increasing locomotor speed could act as a protective mechanism to ensure an appropriate placement of the foot in the event of a perturbation.  5.5 Conclusion I have shown vestibular ex-afferent input to lower-limb muscles is modulated phasically across the locomotor cycle. These signals modify muscle activation in many muscles of the lower limb over the entire stride cycle and are selectively suppressed with higher walking speeds and cadences. Together these findings demonstrate the widespread and phase dependent contribution of vestibular ex-afferent signals to muscle activation for the purpose of balance stabilization during locomotion in humans.   117 5.6 Summary: Study 4 The purpose of study four was to investigate the dynamic modulation vestibular influence on muscle activity during locomotion. Using the techniques developed in Chapter 4 I identified phase dependent modulation in muscles throughout the lower limbs and hips over the locomotor cycle. Modulation patterns were specific to each muscle but overlapped most during the stance phase of the step cycle. In addition I demonstrated that vestibulo-muscle coupling is attenuated with both increases in cadence and locomotor velocity, but only in muscles acting around the ankle. This result contrasts with previous observations of the global suppression of vestibular related postural responses suggesting suppression might be specific to particular regions of the body. Also I observed phase dependent polarity inversion of some muscles’ responses over the gait cycle. Previously, vestibular responses had only been seen to invert in response to changes in static head position relative to the body. This study demonstrates that sensory cues unrelated to head rotation can invert vestibular responses in a manner similar to what has been observed in proprioceptive reflexes during locomotion. Changes in the polarity of vestibular responses during head turn have been widely described but are rarely investigated on their own. Description of this transformation typically arises as a consequence of the static head positions used during delivery of electric vestibular stimulation and as a result their behaviour in a dynamic context remains unclear. In the following study I investigated the dynamic spatial transformation of vestibular responses during head rotation. Initially I identified whether the transmission of vestibular ex-afferent signals is attenuated during motion and then determined whether or not the residual ex- afferent signals are continuously transformed during head rotation.   118 6 Dynamic Transformation of Vestibular Ex-afferent Signals by Head Rotation  6.1 Introduction  Human vertical posture is generally unstable requiring sensory-motor feedback mechanisms to maintain balance in a dynamic environment. Sensory receptors located throughout our body provide a window into the state of our body and the world around us. This information is then used to make decisions, plan future motion, or alleviate threats to our current posture. The vestibular system, one of the body’s sensory systems, provides information regarding the linear and angular motion of the head which is used for a variety of functions including stabilization of the head and body in space (Fitzpatrick & Day, 2004; Angelaki & Cullen, 2008). Functionally the vestibular organs act as inertial accelerometers providing simple measures of change in motion of the head. Once in the central nervous system these signals are decomposed into re-afferent and ex-afferent sub-components (von Holst, 1973) and spatially transformed to provide information in the appropriate head or body centered reference frame (Roy & Cullen, 2001; Roy & Cullen, 2004; Angelaki & Cullen, 2008; Brooks & Cullen, 2009; Cullen et al., 2011). The ex-afferent component in particular provides our body with an indication of unexpected events or movement errors and is therefore important for generating corrective postural responses. During static postures re-afferent signals are attenuated and therefore vestibular signals related to postural disturbances should reliably reflect the disturbance. In contrast, during periods of motion, or situations in which multiple ex-afferent sources sum together the separation and interpretation of these signals could become more difficult. If these vestibular   119 signals do not combine linearly, information related to each of the individual ex-afferent sources might become inseparable. Currently, not much is known about ex-afferent signal fidelity during motion and, in particular, whether the process of separating the re-afferent information alters the ex-afferent information. Theoretically, cancellation of vestibular re- afference is meant to allow central vestibular neurons to code primarily for externally applied vestibular perturbations (Boyle et al., 1996; McCrea et al., 1999). In primates, this cancellation is reasonably efficient, with between 66 and 80 percent of re-afferent information removed from the ex-afferent signal (McCrea et al., 1999; Roy & Cullen, 2001; Roy & Cullen, 2004). However, since this cancellation is incomplete the superposition of the ex- afferent signal with the residual re-afferent signal could result in degradation of information related to the original ex-afferent source. Similar behaviour might also be present when two or more ex-afferent signals superimpose. The non-linear addition of two or more ex-afferent signals could result in attenuated compensation to the postural disturbances. In addition to uncertainty regarding ex-afferent signal fidelity, the spatial transformation of vestibular signals during motion remains similarly unexplored. Since vestibular receptors are fixed in the skull, the signals they transduce are initially received in a head centered reference frame. In order to contribute to postural responses throughout the body vestibular signals must be transformed based on the head’s position relative to the feet, to ensure that compensatory responses at the feet are appropriate given the current head on body position.  Presumably, if vestibular signals are to contribute to postural compensation during motion they must be transformed spatially, during the motion itself, to correspond to a body centered reference frame. In a static posture the state parameters defining how ex- afferent signals must be transformed are believed to arise from a central representation, or   120 internal model, of head and body orientation (Gurfinkel et al 1989). This internal model is thought to be updated and calibrated by incoming sensory information (Gurfinkel et al., 1989; Wolpert, Goodbody, & Husain, 1998) and exerts its influence on vestibular signals through actions in the vestibular nuclei (Gdowski & McCrea, 1999) or cerebellum (Manzoni et al., 1998; Manzoni et al., 1999), providing vestibular responses which are correct given the current head on body orientation. In humans, spatial transformation of vestibular ex-afferent signals has typically been investigated by providing an electric vestibular stimulus to participants standing or walking with a fixed head on body posture (Fitzpatrick & Day, 2004). When participants maintain this position (with eyes open) the spatial transformation of vestibular ex-afferent signals remains constant for the duration of time this posture is held. In contrast, during periods of head motion the transformation of these signals must change in near real time in order for vestibular signals to contribute most effectively to postural control. Here I examine the transmission and transformation of vestibular ex-afferent signals during dynamic movement with two main aims. The first aim is to determine whether ex- afferent signal fidelity is degraded during periods of active and passive head rotation as compared to static postures. The second aim is to characterize the dynamic transformation of these signals in both active and passive motion. Hypothesis: I hypothesize a) vestibular responses to an electric vestibular stimulus will be attenuated in passive and active movement conditions when compared to a static condition, and b) during head rotation the delivery of a vestibular stimulus will result in lower-limb vestibulo-muscular interactions which are continuously modified in their spatial orientation by head on body position.    121 6.2 Methods Subjects Ten healthy subjects (height 172 ± 14 cm, mass 74 ± 11 kg, 9 males) between the ages of 20 and 36 years, with no known history of neurological disease or injury participated in this study. Each subject gave their written, informed consent to the procedures approved by the University of British Columbia’s clinical research ethics board.  Stimulus Stochastic stimuli were provided using a bipolar binaural electrode configuration with carbon rubber electrodes (9 cm2), secured over the mastoid processes using an elastic headband. The stimuli were created on a PC computer using LabVIEW 10 software (National Instruments, USA) and delivered using a constant-current stimulation unit (DS5, Digitimer, England). Bandwidth limited stochastic signals (0 - 25 Hz bandwidth,  peak ± 4 mA, root mean square(RMS): 2.18  ± 0.0065 mA) were generated with three different trial lengths: 180 s for fixed head position trials, 579 s for passive head rotation trials and 585 s for active head rotation trials. Six seconds were added to the active trial lengths to ensure participants completed the 65 trails and were not delayed due to errors in self pacing. Trial lengths for the active and passive motion trials were determined based on the time required to perform 65 head rotations  Signal Collection Surface electromyography (EMG) was collected bilaterally from the medial head of gastrocnemius using self-adhesive Ag - AgCl surface electrodes (Ambu Blue Sensor M,   122 Denmark). The medial gastrocnemius was chosen because it exhibits large responses to vestibular stimulation (Dakin et al., 2007).  EMG was amplified (×2000 - 20000), band-pass filtered from 10 Hz to 1000 Hz (Neurolog NL-844, Digitimer, UK), digitized along with the vestibular signals at 1000 Hz (PXI-6289, National Instruments, USA) and saved on a personal computer for subsequent off-line analyses.  Motion Control Passive head rotation was performed by a custom built motion platform and controlled using a motion control interface (7344 motion controller, National Instruments, USA).  The motion platform was comprised of a stepper motor coupled to a helmet but interfaced through a clutch, to provide an upper limit to the torque applied to the head (Figure 6.1A). The lower shaft of the motor was hinged and could telescope to allow subjects to sway freely while controlling axial rotation of the head. In addition to the helmet, the trunk was also stabilized by a series of braces (Figure 6.1A). These braces allowed participants to freely sway but restricted them in axial rotation (Gurfinkel et al., 2006). During trials the motion platform rotated participants’ head 120º (from 60º to the left to 60º to the right) relative to their bodies with a peak velocity of 57 ± 2.5 º/s. While strapped to the motion platform, movement of the head was captured by quadrature encoder and binned to provide 1 degree resolution. The encoder was coupled to the helmet to ensure it captured head motion in the event of motor shaft slippage relative to the clutch. Participants wore the helmet during all trial conditions to monitor head position relative to the feet.     123 Test Procedures In all conditions, participant’s heads were rotated relative to the shoulders in the yaw plane (Figure 6.2). This axis of rotation was chosen due to the shift in reference frame occurring in the transverse plane. As the head turns, from looking over the right shoulder to looking over the left shoulder, the muscles involved in compensating for the perturbation, which occurs in the direction of the anode electrode, will shift between the plantar flexors and the dorsi-flexors, maximizing modulation of vestibular-evoked responses in the gastrocnemius. Participants performed three conditions to a) determine if vestibular ex- afferent signals are degraded during motion (both active and passive) and b) characterize the transformation of these vestibular signals during motion. The first condition was a control condition consisting of five static or fixed head positions (-60º, -30º, head turned to the left; 0º head forward and 30º, 60º head turned to the right). At the start of the static trials the center head position was determined by aligning the midline of the face with a line centered and parallel to the feet. Following determination of the center head position the head was rotated under the guidance of the motor encoder to ensure the correct amount of head on body rotation. Once the appropriate head on body position was established the head was maintained in this position by the holding torque of the motor and the interfacing clutch. Static trials were collected as a single three minute SVS trial.  The second trial condition was passive head rotation relative to the body. Passive head rotations were collected over four 9.5 minute trials providing a total of 260 oscillations. The first and last head rotation in each trial was removed to exclude trajectory errors associated with the onset and cessation of movement (8 oscillations). Timing errors in one subject resulted in the removal of seven additional head rotations. To ensure an equal contribution   124 from each subject a total of 245 head rotations were used.  During the passive condition the motor rotated the head through 120º (-60º to the left and 60º to the right) while SVS was provided. The trajectory of the passive head motion was constructed prior to data collection to resemble motion in the active trials. Although inter-subject variability did not permit an exact match, the trajectory profiles of both motion conditions were similar (Figure 6.1b). In the third trial condition the motor was turned off and the clutch released to provide little resistance to head motion. Participants then actively rotated their head through 120º paced by a metronome. The timing of the metronome matched the passive head motion trials to ensure correspondence in the motion trajectory of the two trial conditions. As for the passive condition, the active head rotations were paced to provide a total of 260 oscillations over four 9.5 minute trials, however to match both motion conditions only 245 head rotations were used.  Data Analysis EMG data for each condition were high pass filtered offline at 30 Hz, notch filtered at 60 Hz (due to noise induced by the motor) and cut to the length of the SVS stimulus. Since changes in EMG amplitude will influence the gain of the muscle response mean EMG amplitude was determined between conditions. To provide a single value for EMG magnitude rectified EMG was extracted for each head motion condition and averaged within each subject. The RMS of the average EMG was then calculated for each subject for each condition.     125  Figure 6.1 Experimental set up, motion trajectory and raw data. A. Photo displaying the brace controlling axial rotation of the trunk and the helmet through which head position was controlled. The subject stood on a force plate mounted to a turn table. This turn table was not used in this study   B. Average motion trajectories for the active head rotation trials (top) and for the motor controlled passive rotation trials (middle). C. Encoder position for one head rotation with its accompanying raw EMG from the right medial gastrocnemius. In each graph a position of 60º is straight ahead, 120 degrees is turned right and 0º is head turned left. s: seconds, mV: millivolts, EMG: electromyography, º: degrees   126  EMG data from all conditions were then cross-correlated with the SVS signal to estimate the magnitude of the peak medium latency correlation. In the static head position condition cross-correlations were calculating by estimating the time-cumulant density function (Rosenberg et al., 1989) then normalizing it by the vector norm of the SVS and EMG signals to provide units of correlation (Dakin et al., 2010). Once the cross-correlation was estimated the amplitude of the medium latency response was identified for each subject. This amplitude was determined for each of the static head positions was used to provide a reference to determine the relative gain and spatial transformation of the vestibular response at each head position for comparison with the dynamic trials. For both dynamic conditions (active and passive head motion), trials were cut into segments based on each transition point in the head rotation (starting with the head turned ~60º to the right and finishing with the head turned ~60º to the left) then time-dependent cross-correlation, time-dependent gain and coherence were calculated and averaged across the each of the segments, as described previously (Chapter 4) (Blouin et al., 2011). Because all participant had many trials in which they did not reach a full 120º of head rotation data for the dynamic trials is only displayed for up to 55º of head rotation in either direction (110º in total). To calculate the time-dependent cross-correlations data were first re-sampled to 200 Hz. Cross-correlations were calculated for each sample in the head rotation cycle providing a window of correlation from 250 ms prior to 500 ms following the data point at each sample in the head rotation cycle. The cross-correlations were then normalized to provide values of correlation between -1 and 1 and used to estimate the peak of the medium latency response in the EMG responses. Data for the dynamic trials were then averaged within subjects by   127 binning the data in one degree bins then averaging within each bin. This process transformed the data from change per unit time to change per degree. Each trial was also broken down into rotation of the head to the left and rotation of the head to the right, to separate potential effects due to rotation direction. To estimate the amplitude of the medium latency response in the dynamic trials for comparison to the static trials, the following steps were taken. First 512 point windows of raw data were extracted from the dynamic trials centered at each of the encoder positions matching the static trials (from head left to right: -55º, -30º, 0º, 30º, 55º). These windows were then grouped together and concatenated to allow estimation of the time-cumulant density function, similarly to the static trials (Rosenberg et al., 1989). The peak medium latency response was then identified from the cumulant density function for each head angle in each subject. Time-dependent coherence and gain was estimated based on the modified procedure presented by Zhan (2006) (Chapter 4). Briefly, head motion segments (see previous section) were transformed into the frequency domain by Morlet wavelet transform. Gain was calculated by taking the ratio of the cross-spectral density of the SVS and EMG signals divided by power-spectral density of the SVS signal for each segment (Blouin et al., 2011). The resulting gain signals were binned and averaged to provide an indication of change of gain per degree as opposed to change over time. Coherence was calculated similarly to Chapter 4 then also binned to provide an indication of change in vestibulo-muscle coupling per degree of head rotation. Coherence and gain were used together to identify changes in the coupling between the vestibular stimulus and lower-limb muscle activity.   128  Figure 6.2 Schematic displaying an overview of the procedures and expected results. In the figures below head right is indicated with positive head angles, head center is indicated with zero head angle and head left is indicated with a negative head angle. Head rotation will occur in the yaw plane as indicated by the black arrows. The vestibular stimulus will induce a roll along the plane of the inter-aural axis as indicated by the red arrow and in the direction of the anode electrode (the red wire and circle located by the ear). The predicted polarity of the induced muscle responses in the medial gastrocnemius of each leg for each head position are displayed in the bottom row. R: Right leg, L: left leg.  Data Reduction and Statistical Analysis To determine whether vestibular responses are present during head motion, time- dependent cross-correlations were calculated and evaluated for significance using a 99% confidence interval (Zhan et al., 2006). Confidence limits were derived from the number of data in each 1 degree bin for each subject. Similar to Chapter 4, the 99% confidence limit is being used due to the two dimensional nature of the data. Two additional groups of   129 comparisons were also performed to determine whether active head motion attenuates vestibular responses relative to passive motion. First both peak coherence and gain were compared using paired t-tests. Because there was not expected to be any difference in these measures bilaterally the peak magnitudes for the left and right muscles were averaged and compared between active and passive conditions. In addition, to control for changes in EMG activity which might affect gain magnitude relative to coherence RMS EMG amplitude (averaged  bilaterally) between passive and active motion conditions were compared also using paired t-tests. The level of significance was set at α = 0.05 To evaluate the spatial transformation of vestibular signals for dynamic and static head positions, the change in amplitude of the medium latency response was compared between dynamic motion conditions (active and passive) and static head conditions. This comparison served to determine if motion influences the transformation of these signals above the static condition. Since there were no differences between motion directions or rate of change in the amplitude of the medium latency response between bilateral muscles these conditions were averaged to provide a single value and simplify all subsequent analyses. Comparison between the change in medium latency amplitude and head position was subsequently performed using a two-way repeated measures ANOVA (MOTION: active, passive and fixed; POSITION: - 55(-60) -30 0 30 55(60)). Main effects were decomposed using Tukey’s post hoc with the level of significance set at α = 0.05.       130 6.3 Results During active and passive head motion significant correlations were observed between the vestibular stimulus and muscle activity (Figure 6.3). These results demonstrate that vestibular ex-afferent signals still exert a significant influence during head rotation and are not attenuated. Between motion conditions however, the gain of the vestibular responses was larger during active motion than during passive motion [t(9) = 4.57, p =0.0013; 25.8 ± 15.6 (Active), 3.9 ± 1.3 (Passive)] (Figure 6.4 A). This difference in gain was due to a significant rise in muscle activity [t(9) = 4.96, p =0.00078; 0.22 ± 0.12 mV (Active), 0.025 ± 0.017 mV (Passive)] (Figure 6.4 B) in the active condition which led to a significant increase in response gain in this condition compared to the passive condition. When muscle activity is controlled for, as in the time-dependent coherence, there is no difference in peak amplitude between responses in the active and passive motion conditions [t(9) = 0.46, p =0.65; 0.31 ± 0.09 (Active), 0.32 ± 0.078 (Passive)] (Figure 6.5A & B). During head rotation vestibular responses were continuously transformed with changes in head position. Spatial transformation was evident as a flip in the polarity of the muscle response as the head rotates from right to left and back. Bilaterally the timing of the muscle response polarity inversion was different (Figure 6.3 left side).   131  Figure 6.3 Significant time-dependent cross-correlations for active, passive and fixed head motion conditions. Top Row: Time-dependent cross-correlations for the active head rotation trials. On the left is the left medial gastrocnemius and on the right is the right medial gastrocnemius. With the head turned to the right the vestibular responses initially exhibit a negative positive polarity but as the head is turned to the left the response disappears and then reappears in the opposite polarity. The polarity inversion occurs earlier in the right leg than   132 the left causing both legs to have opposite polarities when the head is centered. Middle Row: Time-dependent cross-correlations for the passive head rotation trials. Significant correlations are present in both the passive and active head rotation conditions. Spatial modulation in both active and passive conditions is similar. Bottom Row: Cross-correlations for the fixed head position trials. Cross-correlations have been plotted to resemble the time-dependent cross- correlations of the motion trials. At each head posture from head right to head left the polarity of the muscle response is similar to the two dynamic trials. The color bars on the right indicate the magnitude of the correlation and the ordinate axis indicates the time lag relative to the stimulus. A correlation at 100 ms indicates the response occurs 100 ms following the stimulus. HR: Head right of center, HL: Head left of center, ms: milliseconds, l-mGas: left medial gastrocnemius, r-mGas: right medial gastrocnemius. Arrows indicate the periods over which the head is rotated relative to the head forward position.   133  Figure 6.4 Gain and average EMG for the active and passive motion conditions. A. Gain for the active (top) and passive (middle) motion condition. Active motion exhibits roughly a 10 fold increase in SVS-EMG gain over the passive condition. B. Mean EMG for both the active and passive motion conditions. EMG in the active conditions (Dark line) is approximately 10   134 times the amplitude of the passive condition (grey line) (left leg: 8.55 ± 2.03; right leg: 10.31 ± 3.62, X ± SD). This increase in gain during active motion is therefore due to the increased EMG activity associated with active head rotation. When EMG amplitude is taken into consideration, as in the coherence traces, peak SVS-EMG coupling between the two conditions is similar (Figure 6.5). Arrows indicate the periods over which the head rotated relative to the head forward position.   Figure 6.5 EMG-SVS coherence for the active and passive motion conditions. A. Coherence in the active motion condition. B. Coherence in the passive motion condition. Between the two conditions there is no difference in peak coherence.   135 In the left medial gastrocnemius when the head is turning from 55º to the right to the head center position the polarity of the vestibular responses is a negative short latency response (SL) and a positive medium latency response (ML) (Figure 6.2 left side). Once the head passes -12º to the left of center (-12 ± 14º Active; -12 ± 11º Passive) vestibular responses disappear, reappearing at -34º head left (-36 ± 8º Active; -32 ± 11º Passive) but in the opposite polarity: a positive (SL) negative (ML) waveform. When turning the head from left to right the pattern repeats itself with responses disappearing at -34º head left of center (-37 ± 6º Active; -31 ± 5º Passive) and reappearing at -14º head left of center (-17 ± 8º Active; -11 ± 13º Passive). In the right leg (Figure 6.2 right side) a different pattern is observed. When the head is facing over the right shoulder the polarity of the muscle response is also negative (SL) positive (ML) but as the head is turned to the left these responses disappear at 22º head to the right (24 ± 10º Active; 21 ± 11º Passive) then reappear in the opposite polarity at around head forward (4 ± 9º (Head right) Active; 6 ± 11º (Head left) Passive). The reverse happens when the head rotates from left to right. Initially responses are of a positive (SL) negative (ML) polarity but at 0º (5 ± 9º (Head right) Active; -4 ± 10º (Head left) Passive) responses disappear and reappear in the opposite polarity at 26º head turned right (30 ± 13º Active; 22 ± 14º Passive). Overall there seems to be a period of approximately 20º where these muscles do not couple with the stimulus. The spatial transformation of the vestibular responses during motion exhibited a roughly similar pattern to the changes in polarity observed during the static trials (Figure 6.3 Bottom) with all motion conditions exhibiting significant modulation with change in head position [F(4,2) = 163, p < 0.05; Tukey’s, multiple p < 0.05]. There was however, an interaction between position and motion [F(4,2) = 12, p < 0.05; Tukey’s, multiple p < 0.05] due to the   136 amplitude of the medium latency response being less during passive motion at -55º degrees head turned left than in either the active or fixed motion conditions and less than the fixed condition when the head is turned 55º degrees to the right [F(4,2) = 6.37, p < 0.05; Tukey’s, multiple p < 0.05]. These comparisons suggest that relative to the fixed or active motion conditions, spatial transformation might be hindered with increasing head angle during passive rotation (Figure 6.6).  6.4 Discussion The purpose of this study was to examine the transmission of vestibular ex-afferent signals to muscles of the lower limb during head rotation. I hypothesized that vestibular ex- afferent signals influence on lower-limb muscles would decrease during periods of movement due to either the contamination of re-afferent signals during active movement or the summation of multiple ex-afferent signals during passive motion. The results of this study indicate otherwise. Significant vestibulo-muscle coupling was observed during both active and passive head rotation indicating these signals continue to contribute to balance control during motion. However as head rotation increases responses generated during passive motion are attenuated relative to fixed or actively generated motion suggesting the spatial transformation of vestibular signals might be hindered during passive motion.    137  Figure 6.6 Change in the amplitude of the medium latency response at five head angles. Average amplitude of the medium latency response (n = 10) averaged across both muscles and movement direction for each of the movement conditions. During passive rotation the amplitude of the medium latency response was significantly reduced relative to the active and fixed motion conditions when the head was turned -55º to the left and significantly reduced relative to the fixed motion condition when the head was turned 55º to the right. Stars indicate statistical significance and the error bars indicate one standard deviation. Head left is -60º and head right is 60º.     138 Vestibular ex-afferent signals are faithfully transmitted to lower-limb muscles during head movement providing spatially appropriate compensatory responses. The superposition of the stochastic vestibular stimulus on both active and passive motion did not appear to hinder the ability of the ex-afferent like stimulus to influence muscle activity. This result is in contrast with the apparent suppression of vestibular ex-afferent signals during motion in other dynamic tasks, such as locomotion (Brandt et al., 1999; Brandt, 2000; Jahn et al., 2000), but is congruent with, and might provide support for current views on the processing of vestibular re and ex-afferent signals. During active movement the re-afferent component of the movement is attenuated 70 percent, on average, early in vestibular processing (Cullen et al., 2011). One reason proposed for this separation is to isolate the ex-afferent signal for its detection and compensation (Boyle et al., 1996).  While this could be part of the reason these signal are separated these data suggests the body is capable of effectively compensating for the ex-afferent signal even when it is superimposed upon the movement related signal, at least within the motion profiles and central angles (-30º to 30º) used here. Alternatively separation of ex- and re-afferent components likely serves as a way to avoid compensating for the re- afferent component of the signal (McCrea et al., 1999). If sensory re-afference were compensated for then the body would compensate twice for a single movement as self initiated or active motion is already accompanied by postural actions meant to stabilize the body during motion. Vestibular re-afference is therefore likely suppressed to limit compensation to only the ex-afferent component of the total vestibular signal. The spatial transformation of vestibular responses also appears to benefit from whether the movement is performed actively as opposed to passively. This active-passive difference might arise from a disparity in sensory feedback between the two motion   139 conditions. Traditionally the cause and source of the spatial transformation of vestibular responses has been attributed to sensory feedback (Popov et al., 1986; Gurfinkel et al., 1989). Shifts in the direction of vestibular responses can be induced by vibrating the gluteus maximus muscle without any accompanying active motion (Popov et al., 1986) and drift in the direction of these responses can be corrected by simply opening the eyes (Gurfinkel et al., 1989). During active movement muscle spindles can increase their sensitivity to muscle lengthening and shortening (Burke et al., 1978) potentially increasing the sensory return from active movements. Passive movements, in comparison, are believed to provide reduced sensory return which negatively impacts systems reliant on sensory feedback and could degrade the spatial transformation of vestibular signals. However, movements in the current study might not be truly passive because participants must continue to stabilize their head relative to their trunk as they are still free to move in the roll and pitch planes. Therefore even in the passive condition postural muscles in the neck are probably still active. The sensory return between these two movement conditions might be very similar and might not account for the differences observed between the active and passive motion conditions. Alternatively, active motion might afford a better positional sense than passive motion due to the predictive contribution of forward internal models or efferent corollaries. Sensory predictions provided by forward models have been suggested as a means to overcome limitations provided by conduction delays, cancel the expected sensory return from voluntarily generated motion or generally improve estimation of body position (Miall & Wolpert, 1996). Recently the contribution of central mechanisms to body position estimation has gained additional support from studies on motion perception. Gandevia et al. (2006) demonstrated that during ischemic block of the arm, when both the sensory and motor   140 pathways are paralyzed, participants still perceive arm motion when attempting to move their arm. Since sensory feedback from the limb has been abolished the most parsimonious source for a perceptual motion signal is that of a motor corollary or sensory prediction. In a complimentary experiment by Ansems et al. (2006) participants performed a series of passive and active bilateral forearm matching tasks which found that conditioning related position errors, which are attenuated during loaded passive arm movement, persist during active motion. The persistence of these errors, which should have been attenuated if they were due to muscle spindles, led the authors to propose that they are centrally generated with the movement command and that this information is used to help determine the position of the limb during movement. Similar behaviour might also contribute to the determination of head position. A combination of sensory feedback from neck muscles, skin and joints might be combined with a central prediction of head position to provide a more accurate representation of head on body position or motion. In addition a central prediction could also provide a phase advance at higher motion velocities which would be beneficial to overcome conduction delays associated with the sensory feedback. (Miall & Wolpert, 1996). Ultimately, the reduced spatial transformation of vestibular responses at increasing yaw angle might reflect position error attributable to the absence of an incorporatable centrally produced position signal.  6.5 Conclusion The transmission and spatial modulation of vestibular ex-afferent signals is not hindered by head motion signifying compensation to vestibular ex-afferent signals continues unimpeded during head rotation. In fact, the gain of the compensatory muscle response to the vestibular stimulus appears larger in the active motion condition at large yaw angles than with   141 passive motion suggesting motor prediction may contribute to the spatial transformation of vestibular information.   142 7 General Discussion and Conclusions  7.1 General Summary and Discussion Discrete stimuli such as the step pulses used during galvanic vestibular stimulation have exhibited a variety of limitations in certain contexts. The extended time required for their application reduces the information that can be derived in a single testing session and the accompanying postural disturbance can be counter-productive to the goals of the study, as well as cause discomfort for the experimental subjects. The overall goal of this thesis was to advance the method of stochastic vestibular stimulation for its application in dynamic contexts, then use these advances to examine dynamic vestibular function during locomotion and head rotation. In the first volume of this thesis I examined whether modification of the stochastic stimulus and its analysis procedures could overcome some of the known limitations associated with discrete galvanic stimuli.  The first study investigated whether the postural instability created by the stimulus could be attenuated by modifying the stimulus bandwidth. Previous studies had shown that the sway response to a galvanic stimulus is at a very low frequency (< 1.5Hz) (Fitzpatrick et al., 1996; Pavlik et al., 1999). I hypothesized that by removing these frequencies from the stimulus bandwidth sway related to the stimulus would be attenuated due to the higher frequencies being mechanically filtered out by the body.  This is indeed what was observed. As the stimulus transfers from muscle activation to force generation and finally resulting sway, stimulus-related frequencies are repeatedly low-pass filtered until the final sway responses is driven primarily by stimulus frequencies below 2 Hz. If these frequencies are removed from the stimulus sway correlated to the stimulus is reduced and total sway during   143 stimulation is not significantly different from free standing. What the main findings of this study show is that the stochastic stimulus can generate clear muscle responses in a manner which does not generate much corresponding sway, a beneficial result for contexts in which sway is counter-productive to the outcome of the experiment.  The second study also examined modification of the stimulus bandwidth but this time for two purposes: the first purpose was to determine if the stimulus could be used to preferentially elicit either the short or medium latency responses. Based on my previous work (Dakin et al., 2007) it appeared as though frequencies from 0-10 Hz contributed to the medium latency component and frequencies between 11-20 Hz contributed to the short latency component. I hypothesized that a stimulus composed of frequencies from 0-10 Hz would preferentially elicit the medium latency response and a stimulus composed of frequencies between 11-20 Hz would preferentially elicit the short latency response. The second purpose of study two was to determine if the short and medium latency components interact at certain frequencies of stimulation as both the coherence and phase functions from the stochastic stimulus exhibit an inflection at approximately 10 Hz, the transition between the frequency bandwidths contributing to the two responses. I tested this hypothesis using sinusoidal stimuli to determine if they could provide additional information regarding the potential interaction than is provided by the broad bandwidth stimuli. The results of this study indicated that modification of stimulus bandwidth could not be used to preferentially elicit the short or medium latency response components and that the sinusoidal stimuli provided no further information than that provided by the stochastic stimulus. In fact the 0 - 20 Hz stochastic stimulus provided similar information to the mean of all the sinusoidal stimuli promoting the use of the stochastic stimuli due to its drastically reduced testing time (1.5 vs   144 20 min).  Even though modification of the stimulus bandwidth could not isolate either component of the vestibular induced response it still appears to play a big role in determining the shape of the observed muscle response. This is true for both the stochastic stimulus and more traditional rectangular type waveforms. Using a traditional rectangular waveform, as the rise time of the stimulus is increased the short latency response becomes attenuated independently of the medium latency response (Rosengren & Colebatch, 2002). The effect of increasing the rise time of the rectangular wave appears similar to the inclusion of lower frequencies in the stochastic stimulus. In the first study (Chapter 2) removal of the low frequency content resulted in an increase in the amplitude of the short latency response. While in the second study (Chapter 3) the superposition of the responses to each of the sinusoidal stimuli indicated that lower frequency stimuli (1 - 5 Hz) produce a response which opposes the short latency response. When these low frequency stimuli are summed with their higher frequency counterparts the result is an attenuation of the short latency response. In the extreme case, such as the administration of a low frequency sine wave there is no sign of a biphasic response; muscle activity instead appears sinusoidal following undulation in the stimulus (Figure 2.3). Overall these results appear to support previous observations that high frequency stimuli (fast rise times) facilitate the amplitude of the short latency response component and low frequency stimuli can prolong the medium latency response (Britton et al., 1993). From a physiological perspective this finding is somewhat incongruent with current views on the source of these two responses as the medium latency response is thought to be derived from the semicircular canals (an organ that provides phasic input to the nervous system) and the short latency response, potentially, the otoliths (an organ which is believed to   145 provide both phasic and tonic input to the nervous system). One might presume prolonged (low frequency) modulation in muscle activity might be associated with mechanisms tied to the otoliths (thought to provide tonic postural input) and high frequency responses (more associated with change in rate) more indicative of mechanisms associated with the semicircular canals. Alas, both of these responses are likely much more complicated than containing simple otolith and canal components, a point I will revisit in the limitations sections. One of the major advantages of the stochastic stimulus is the reduced time it requires to resolve vestibular responses. This improvement over its discrete cousin provided the foundation for the third study in Volume One which examined the effectiveness of coupling a stochastic stimulus with time-frequency analysis to extract the dynamic modulation of vestibular influence during locomotion. The results of this study demonstrate that the combination of stochastic stimulus and time-frequency analysis is a potent grouping for extracting the time-dependent relationship between the stimulus’s influence over muscle activity and the step cycle. Phase-dependent stimulus-muscle coupling was observed bilaterally in the medial gastrocnemius during the stance phase of the step cycle. Vestibulo- muscle coupling in the frequency domain was observed over a similar bandwidth to standing balance but with a lower peak frequency and a peak coherence that did not coincide with period of greatest EMG. The shape of the vestibulo-muscle interaction in the time domain was that of a multi-phasic waveform also reminiscent of those observed during standing balance. In the end phasic modulation of vestibulo-muscle coupling was extracted at a much higher resolution then had been in the past but with only 15 minutes of required testing. Ultimately, the success of this technique opens the door for further investigation of the dynamic influence   146 of vestibular ex-afference during motion. The timely extraction of phase dependent responses during locomotion formed the technical foundation for the last two studies in this thesis. Dynamic modulation in reflex behaviour could now be extracted in reasonable time to test and compare a variety of dynamic conditions. The first of which, and might be most obvious, was to describe the phase dependent behaviour of vestibular responses during walking. As previously discussed, several studies had tried to examine phase dependence during locomotion however the techniques available at the time were limited in their suitability to the extract modulatory behaviour during a dynamic task (Bent et al., 2004; Iles et al., 2007; Roskell et al., 2007). The second volume of this thesis continued where the first left off but with a more physiological focus. The fourth study (Chapter 5) examined the phase-dependent influence of vestibular ex-afference over muscles of the leg and hip at different cadences and speeds. Up to this point vestibular ex-afference was known to be modulated by the phase of the step cycle; however it had only been observed in postural responses and in muscles acting around the ankle. In contrast, during free standing a vestibular perturbation causes muscle responses throughout the body, seemingly in any muscle engaged in the act of maintaining balance (Britton et al., 1993; Ali et al., 2003). The ubiquity of these responses during standing would imply that widespread vestibular influence should also be present during locomotion. Therefore the first aim of this study was to explore vestibular coupling in muscles throughout the lower limbs and hips. The second aim was to determine what happens to this coupling when cadence or walking speed is increased. Several studies have suggested vestibular ex-afferent signals are suppressed at higher locomotor speeds. It is unclear though as to whether this effect is due purely to locomotor speed or also to the rise in cadence that accompanies increasing   147 locomotor speed. The results of this study demonstrated that the vestibular stimulus influenced all muscles recorded, though some more than others. Each muscle had its own pattern of phase dependence which reversed in effect for a few muscles at particular points in the step cycle. Also, increasing both cadence and speed appeared to decrease the influence of vestibular ex-afferent signals but only in a few muscles acting around the ankle, suggesting vestibular ex-afference is locally attenuated rather than globally suppressed. The phasic modulation of vestibular signals presumably shapes the timing and amplitude of compensatory responses to the vestibular perturbation in order to ensure that an effective compensatory response can be produced at each point in the step cycle. Intriguingly, the inversion of some of these responses at points in the step cycle resembles the spatial transformation of vestibular signals that occurs with head turn. Many studies have documented the spatial transformation of electrically-induced vestibular responses to head turn in humans maintaining a static posture. In contrast, little is known about this same transformation during active motion. When a primate rotates its head vestibular re-afference is cancelled and the remaining ex-afferent signal is then transformed spatially to provide appropriate compensatory responses. The aim of study five was to examine the transformation of vestibular signals in a dynamic environment to determine a) if the process of cancelling the re-afferent signal degrades the ex-afferent signal and b) whether motion contributes or impairs the spatial transformation of vestibular signals. In this study I compared both active and passive head rotation with vestibular responses observed during standing balance. These comparisons indicated that ex-afferent signals are not degraded during head rotation. In fact, vestibular-muscle coupling appears uninfluenced by head motion. Also, the gain of these vestibular responses is larger during active head rotation   148 condition due to increased lower-limb muscle activity. Finally, I found that active head rotation exhibits greater spatial modulation than passive head rotation at yaw angles greater than 30º. These results suggest that volitional motion contributes to the spatial transformation of vestibular induced responses and further study is required to determine the sensory or motor source of this contribution (See Future Directions). The use of a stochastic stimulus has proven effective in the extraction of time-varying modulation during dynamic tasks. Both locomotion and head rotation demonstrated appreciable responses over a reasonable testing period. The usefulness of this perturbation protocol is not, however, limited to only vestibular stimulation. The application of a stochastic perturbation might be applied to any system in which a continuous signal can be creatively applied. Fitzpatrick et al. (1996) used both a random vestibular stimulus and a random postural perturbation to identify the loop gain of reflexes during standing. The postural perturbation was provided by a servomotor attached to a weak spring anchored to participants at the waist. The subsequent analysis protocol served as the foundation for Chapter 1 of this thesis. More recently this perturbation protocol has been adapted to examine the role of vision during locomotion. Logan et al. (2010) used a broad bandwidth stochastic visual display to examine stability during locomotion. They observed a large increase in horizontal displacement gain for the hip and shoulder during locomotion versus standing which was thought to reflect the increasing postural demands of locomotion over standing balance. As well, combinations of visual and postural stochastic stimuli were used to estimate the role of feedback during upright stance (Kiemal et al., 2011). These authors found feedback from sway consistent with an optimal feedback model suggesting muscle activation was for the purpose of body stabilization and not sway minimization. In addition to postural or visual   149 perturbations stochastic stimuli might be useful in other modalities such as vibration, to examine muscle spindle or cutaneous receptor behaviour, and electric nerve or skin stimulation. Overall, the stochastic stimulus might provide an effective alternative approach for investigating the human neuromuscular system across many sensory modalities. Like all techniques however, it has its advantages and limitations.  7.2 Methodology - Advantages and Limitations All of the studies presented in this thesis have focused on the development and application of the stochastic vestibular stimulus and many of them have advocated its use over more traditional waveforms. And just like many of these other stimuli there are several limitations that must be considered in conjunction with the advantages this stimulus format provides. Since many of the chapters of this thesis have already reviewed the advantages I will start this section with a brief discussion of these advantages followed by an in-depth look at its limitations.  Advantages The stochastic stimulus has largely proven effective because of the amount of current that can be provided in a short period of time. Vestibular stimulation has typically used a step or rectangular type waveform to deliver current. In many cases the step pulses are timed to ensure the response to the stimulus prior is finished before the next stimulus arrives. Under this type of stimulation protocol the duty cycle can be less than 20% meaning during most of the trial participants are receiving no stimulation prolonging the time required to collect sufficient data. When looking at the postural response this might not be a problem since   150 depending on the pulse amplitude and width as few as 6 pulses are sometimes needed. However examination of the muscle response requires many more stimuli to resolve which can drastically prolong testing time. The reduction in testing time provided by using the stochastic stimulus is one of the main reasons why it is effective at extracting time-dependent changes in its associated responses.  An additional advantage of the continuous random format of the stimulus is it provides the opportunity to easily correlate it with many dependent variables (EMG, forces, sway). This analysis can be conducted in both the time and frequency domains permitting a system identification approach under the assumption of linear transmission from input to output. The continuous nature of the waveform also allows the stimulus to be administered for the duration of a cyclic dynamic task. Once collection is finished the data can be broken into cycles and averaged to provide a near continuous representation of modulation in the vestibular evoked responses in a short period of time. This attribute of the stimulus serves as the basis for the analysis approach used in Chapters 4-6. To achieve the same results using a discrete stimulus, stimuli would need to be timed and averaged at each sampled point in the cycle of interest. This can be both technically difficult and take a great deal of time to administer. Lastly, postural responses to the stimulus can be attenuated and the discomfort of the stimulus appears reduced. The ability to provide a stimulus with reduced postural consequence might be of great benefit depending on the task and population being examined. Tasks in which participants are required to maintain a particular position, such as locomotion on a treadmill, or in which additional sway might be detrimental might benefit from a stimulus with very little accompanying sway. Depending on the waveform amplitude and   151 length vestibular stimulation can cause a considerable sway response. This might also be detrimental in patients with balance deficiencies. An additional but anecdotal benefit of the stimulus appears to be its greater tolerability. In general participants appear less negatively affected by the application of the stochastic stimulus when compared to discrete step pulses of similar amplitude. This benefit appears to increase participant adherence to the study protocol.  Limitations  While the stochastic stimulus might be more comfortable than a rectangular pulse it can still be uncomfortable. In general higher stimulus amplitudes and large high frequency swings in current are usually reported as uncomfortable for participants. Discomfort can normally be controlled by limiting the peak to peak amplitude of the stimulus as well as localizing the electrode gel to only the mastoid area, avoiding the auricle. As well, the more prominent the low frequency component of the stimulus is the more nauseating the stimulus seems. Since the low frequency components are easier for the body to follow, narrow bandwidth low frequency stimuli tend to have more accompanying sway and interpretable perceptual effects. In the extreme case of a single low frequency sinusoidal stimulus the accompanying sway and perceptual effects resemble the experience of being on a rocking boat (Bent et al., 2006; Grewel et al., 2009). In addition to these effects participants also report similar sensory experiences to GVS namely, a mild flushing or warmth (MacDougall et al., 2006), a metallic taste in the mouth (Bense et al., 2001), flashing in the visual field and in some cases nausea (Grewel et al., 2009) and dizziness.  To date, transfer of the stochastic stimulus to lower-limb muscles and the associated force and sway responses has been treated as a linear process. The system identification   152 approach used to calculate coherence, gain and phase identifies functional relationships between frequencies in the stimulus and those same frequencies in the dependent measure (i.e. 5 Hz in the stimulus with only 5 Hz in the dependent measure). Off axis relationships between stimulus frequencies and muscle, force or sway responses have only been superficially examined and have yet to be published. Within the bandwidth used in this thesis muscle responses to a stochastic stimulus appear linearly related to the input stimulus. In chapter 3 I presented data displaying the rectified EMG power spectra to individual sinusoidal stimuli from 1 to 20 Hz. Increases in EMG power were only observed at the stimulus frequency and not at frequencies unrelated to the stimulus. As well, in the first study the transfer function for the 0 - 25 Hz stimulus seemed to explain the most of the output response for each of the various stimuli suggesting no more than a linear addition of frequencies between stimulus conditions (0 - 1, 0 - 2, 0 - 25, 1 - 25 and 2 - 25 Hz) (unpublished results). Assuming a one-to- one transfer of stimulus frequencies through the central nervous system (1 Hz in the stimulus is 1 Hz entering the muscle), lower-limb muscles appear fully capable of responding within this bandwidth.  Sway, on the other hand, likely exhibits some component of off-axis coupling at different frequencies of stimulation. Typically, the human body is described as behaving like an inverted pendulum during sway (Winter, 1995; Gage et al., 2004). In this model sway occurs only around the ankle joints which might be valid at low frequencies of oscillation. However, depending on the cause and frequency of motion this comparison is invalid. The multi-segmented human body can have multiple concurrent modes of oscillation which might be excited depending on the frequency of the source driving sway (Jeka et al., 1998; Creath et al., 2005). In addition, the inertial properties of the body segments limit sway to lower frequencies causing sway to higher frequency stimuli to occur at less than a one-to-one ratio   153 with the stimulation frequency. This effect has been observed in sway driven above 0.4 Hz by a tactile stimulus (Jeka et al., 1998). Ultimately, it is probable that stimuli just above 2 Hz induce some form of sway which is likely not at stimulus frequencies. In contrast, stimuli at much higher frequencies likely do not cause much sway because the high frequency alternating in stimulus polarity limits the magnitude of the compensatory impulse created by the muscles thereby attenuating changes in whole body velocity. In fact, anecdotal data of mine suggests narrow bandwidth high frequency stimuli might actually decrease sway (unpublished results); however this result remains an observation and has yet to be fully explored. In summary, while stimuli with a bandwidth of 2-20 Hz might induce some sway, the amplitude of this sway is not greater than what is observed during normal stance, and is limited in its potentiation by the random nature of the stimulus.  In addition to technical limitations associated with the stochastic stimulus itself there are also two major uncertainties associated with electric vestibular stimulation in general which must be considered when interpreting responses to the stimulus. The first consideration relates to the different sensory systems that might be activated by the stimulus and the second consideration is what exactly the responses to the stimulus represent.  During electric vestibular stimulation current is passed between two electrode poles located either superficial to both mastoid processes or over one mastoid process and a second location (e.g. shoulder, neck or forehead). The presumption is that the electric current activates vestibular organs and afferents near the electrodes inducing vestibular-related responses (Aw et al., 2008). In animal models application of galvanic currents near the vestibular nerve causes non-specific activation of all vestibular afferents (irregular afferents exhibiting a lower threshold than regularly firing afferents) (Goldberg et al., 1982; Kim &   154 Curthoys, 2004). The percutaneous application of similar currents in guinea pigs seems to result in comparable non-specific vestibular afferent activation (Kim & Curthoys, 2004). Based on these results it has been assumed that in humans vestibular afferents are also activated in a non-specific fashion. However the current might also activate other nearby tissues due to the fixation of electrodes on the skin and the variance in resistance provided by the different tissues beneath and around the electrodes. Case in point, surface stimulation is also accompanied by paresthesia like symptoms in the skin beneath the electrodes, twitching in scalp muscles in phase with the stimulus and reports of both a metallic taste in the mouth and flashing in the visual field during stimulation. The potential for competing responses from activation of sensory receptors in the surrounding tissues as well as the possibility of a startle response at trial or stimulus onset is disconcerting. Some attempts have been made to examine or control for these potentially confounding effects. Likely the most important is determining whether stimulation of the skin beneath the electrodes results in its own reflexive response. Because of the skin’s location between the electrode and the vestibular nerve it is certain to be activated every time the stimulus is provided and probably presents the most probable confounding influence. A recent study tried to examine the influence of cutaneous afferents during vestibular stimulation finding facilitation of tendon jerk responses in the soleus while sitting. This response facilitation was partially attenuated when the skin beneath the electrodes was anesthetised leading the authors to suggest cutaneous and vestibular inputs could summate to elicit a startle response (Ghanim et al., 2009). However, the effect of skin anaesthetization has also been investigated in standing participants with a stochastic stimulus and in this case no difference was observed between when the skin is anesthetised and when it is not (unpublished results). Additionally, susceptibility to startle is known to attenuate with   155 repeated stimuli implying that startle might only be a concern at the onset of a trial (Brown et al., 1991). Electric vestibular stimulation likely activates many sensory-motor pathways however current research has yet to show that this supplemental activation exerts a significant influence on the induced lower-limb muscle responses.  In addition to uncertainty as to what is being stimulated there is much debate regarding what the observed responses to vestibular stimulation actually represent. Firstly, it is questionable as to whether the responses observed at the muscle are purely vestibular and secondly, where the two responses originate is still much debated. In as early as the first synapse vestibular signals are combined with a variety of other sources (Cullen et al., 2011) and once they enter the vestibular nuclei they could already be considered multi-modal signals. Neurons in the lateral vestibular nucleus, the nucleus primarily responsible for transmitting vestibular signals to the thoracic and lumbo-sacral spinal cord, are highly convergent, receiving input from a variety of sources (Wilson et al., 1967; Allen et al., 1972; Wilson & Peterson, 1978; Cullen et al., 2011). This convergence might be most apparent during locomotion where the firing rates of these neurons are strongly modulated with the step cycle shaped by cerebellar relay of either ascending sensory information from the peripheral nervous system (Wilson et al., 1967; Allen et al., 1972), or informed by the actions of a central pattern generator (Orlovsky, 1972; Matsuyama & Drew, 2000a). Once transmitted down the spinal cord the earliest appearance of vestibular responses in the lower leg is 50 - 70 ms (Nashner & Wolfson, 1974; Britton et al., 1993; Fitzpatrick et al., 1994); this is 15 - 30 ms longer than should occur through direct transmission from supra-spinal centers (Britton et al., 1993). The additional 15 - 30 ms is sufficient for vestibular integration with a variety of sources such as the aforementioned multimodal sensory convergence or central interpretation   156 and transformation of the volley prior to arrival at the motor neuron pool. Some of this delay might also result from additional synapses in the spinal cord. While direct projections from the vestibular nuclei to motor neurons exist (Wilson & Peterson, 1978) vestibular input might exert much of its effect via pre-motorneuronal inter-neurons. Vestibular signals are known to converge upon type II inter-neurons (Davies & Edgley, 1994), Ia inhibitory inter-neurons (Hultborn et al., 1976), presynaptic inhibitory inter-neurons (Pierrot-Deseilligny & Burke, 2005) and are suspected to contribute to non-reciprocal group I inhibitory inter-neurons (Iles & Pisini, 1992). Each of these inter-neurons is highly convergent providing yet another source of multimodal integration. Peripherally, feedback also likely shapes these responses. In the leg feedback from somatosensory receptors can influence muscle activity as early as 45 ms after the onset of the response (Gottlieb & Agarwal, 1979) roughly at the start of the medium latency component. Therefore changes in feedback resulting from the early components of the response might influence the latter components of the response potentially modifying its shape and amplitude. In addition to many sites of multimodal convergence the sources of these two responses are still very contentious. For a while the two responses to vestibular stimulation were thought to derive from separate end organs, the medium latency response from the semi circular canals and the short latency response from the otoliths (Fitzpatrick & Day, 2004; Cathers et al., 2005) however recent evidence has cast doubt on its otolithic nature suggesting it might be a vestigial response (Mian et al., 2010). Still other researchers have suggested the whole response observed to vestibular stimulation is due to an otolithic source because components due to the semi-circular canals are thought to be habituated, possibly by the cerebellum, or are only present at higher amplitudes of stimulation (Zink et al., 1998; Cohen   157 et al., 2012). As is apparent, much debate still exists around the exact source, transmission and meaning of these responses and until consensus has been reached, the muscle and postural responses, at least, seem best described simply as corrective responses to primarily an induced vestibular error.  7.3 Future Directions Over the course of this thesis several advancements have been made in understanding how the stochastic stimulus works and what it might be used for. During this process I have examined varying stimulus bandwidth to isolate components in the associated responses and attenuate sway, and I have used it to extract dynamic changes in coupling between muscle activity and the stimulus. There are however a few fundamental ways in which the stimulus might still be improved. The first is by cresting it. The stimulus used in each of these studies was based on white noise filtered to the bandwidth of interest. The variance of this signal in the time domain therefore exhibits a Gaussian profile with a few outliers reaching the peak-to- peak amplitude limit of 3 - 4 mA. Cresting is the process of eliminating the outliers and rescaling the signal to the desired peak-to-peak limit. This has the effect of increasing the root mean square (RMS) value of the signal and improving its signal to noise ratio. Some basic cresting has been performed on the stimulus in Chapter 6; however the improvements it provides are currently anecdotal and a formal comparison yet to be published.  An additional improvement over the current methodological approach is to use a multi-sine signal as opposed to a white noise signal. This approach has the advantage of localizing stimulus power only to frequencies of interest. It also brings the additional benefit of reducing spectral noise. Depending on the window length used to perform the Fourier   158 transform, frequencies whose periods are not multiples of this window length will produce some form of spectral leakage. This reduces the signal to noise ratio and adds noise to any subsequently estimated frequency response functions (coherence, gain, phase functions). By combining signal cresting with a multi-sine based stochastic signal the signal peak- to-peak amplitude can be reduced while retaining or even increasing signal RMS. In addition the subsequent reduction in signal noise brought by using a multi-sine based signal with an appropriate Fourier transform window should increase the accuracy of measures derived from this information. This combination should theoretically result in a more comfortable stimulus (lower peak-to-peak amplitude) requiring less testing time but providing more accurate results.  Currently, comparison of the crested multi-sine signal with the white noise based signals is being performed in collaboration with Delft University (Forbes et al., 2012)  Another study stemming from this thesis will arise as a continuation of Chapter 6. The final study of this thesis left off with some speculation regarding the source of one of the primary findings. Namely, the increased spatial modulation in the active motion trials might be due to two sources: a greater sensory return during active motion or the central prediction accompanying the active motion. The aim of my future work will be to examine a combination of active and passive head, trunk and head and trunk rotations to determine if the differences in spatial modulation are due to sensory feedback from the neck and spine or the predictive output of a forward internal model or motor corollary.       159  7.4 Conclusion   The aim of this thesis was to advance the methodological framework associated with stochastic vestibular stimulation and to use these advances to examine dynamic vestibular behaviour during locomotion and head rotation. The stochastic stimulus was shown to provide similar information to the provision of a series of sinusoidal stimuli from 1-20 Hz but in much shorter time. The stimulus could be limited in its bandwidth to produce muscle responses at the expense of an associated sway response and dynamic modulation of vestibulo-muscle coupling could be extracted by coupling the stimulus with time dependent correlations in both the time and frequency domain. Phase dependent modulation was explored during locomotion identifying step cycle phase specific vestibulo-muscle coupling in all muscles recorded in the lower limb. These responses were shown to be suppressed at higher cadences and locomotor speeds in muscles around the ankle suggesting vestibular influence is locally attenuated rather then globally suppressed at these speeds and cadences. Dynamic modulation was also examined in response to active and passive head rotation observing no difference in response gain and spatial transformation between these two conditions. However there were large increases in the change in response magnitude per degree of head rotation during motion trials compared to when the head was fixed suggesting compensation to a vestibular error during head motion is greater than when the head is fixed. Overall stochastic vestibular stimulation has proven to be an effective alternative to galvanic vestibular stimulation providing many benefits over its rectangular counterpart. With this stimulus I have shown the time-varying contribution of vestibular ex-afferent signals to the stabilization of balance during motion.    160 Bibliography Akay T & Buschges A. (2006). Load signals assist the generation of movement-dependent reflex reversal in the femur-tibia joint of stick insects  J Neurophysiol 96, 3532-3537.  Ali AS, Rowen KA & Iles JF. (2003). Vestibular actions on back and lower limb muscles during postural tasks in man. J Physiol 546, 615-624.  Allen GI, Sabah NH & Toyama K. (1972). Synaptic actions of peripheral nerve impulses on Deiters neurons via the climbing fibre afferents. J Physiol 226, 311-333.  Anastasopoulos D, Gianna CC, Bronstein AM & Gresty MA. (1996). Interaction of linear and angular vestibulo-ocular reflexes of human subjects in response to transient motion. Exp Brain Res 110, 465-472.  Anastasopoulos D & Mergner T. (1982). Canal-neck interaction in vestibular nuclear neurons of the cat. Exp Brain Res 46, 269-280.  Angelaki DE & Cullen KE. (2008). Vestibular system: the many facets of a multimodal sense. Annu Rev Neurosci 31, 125-150.  Ansems GE, Allen TJ & Proske U. (2006). Position sense at the human forearm in the horizontal plane during loading and vibration of the elbow muscles. J Physiol 576, 445-455.  Armand M & Minor LB. (2001). Relationship between time-and frequency-domain analyses of angular head movements in the squirrel monkey. J Comput Neurosci 11, 217-239.  Aw ST, Todd MJ, Aw GE, Weber KP & Halmagyi GM. (2008). Gentamicin vestibulotoxicity impairs human electrically evoked vestibulo-ocular reflex. Neurology 71, 1776-1782.  Bawa P & Stein RB. (1976). Frequency response of human soleus muscle. J Neurophysiol 39, 788-793.  Bell CC. (1981). An efference copy which is modified by reafferent input. Science 214, 450- 453.  Bense S, Stephan T, Yousry TA, Brandt T & Dieterich M. (2001). Multisensory cortical signal increases and decreases during vestibular galvanic stimulation (fMRI). J Neurophysiol 85, 886-889.  Bent LR, Bolton PS & Macefield V. (2006). Modulation of muscle sympathetic bursts by sinusoidal galvanic vestibular stimulation in human subjects. Exp Brain Res 174, 701- 711.  Bent LR, Inglis JT & McFadyen BJ. (2004). When is vestibular information important during   161 walking? J Neurophysiol 92, 1269-1275.  Bent LR, McFadyen BJ, Merkley VF, Kennedy PM & Inglis JT. (2000). Magnitude effects of galvanic vestibular stimulation on the trajectory of human gait. Neurosci Lett 279, 157-160.  Blouin J, Dakin CJ, van den Doel K, Chua R, McFadyen BJ & Inglis JT. (2011). Extracting phase-dependent human vestibular reflexes during locomotion using both time and frequency correlation approaches. J Appl Physiol 111, 1484-1490.  Borel L, Harley F, Lopez C, Magnan J, Chays A & Lacour M. (2004). Walking performance of vestibular-defective patients before and after unilateral vestibular neurotomy. Behav Brain Res 150, 191-200.  Boyle R, Belton T & McCrea RA. (1996). Responses of identified vestibulospinal neurons to voluntary eye and head movements in the squirrel monkey. Ann N Y Acad Sci 781.  Boyle R & Pompeiano O. (1981). Responses of vestibulospinal neurons to neck and macular vestibular inputs in the presence or absence of the paleocerebellum. Ann N Y Acad Sci 374, 373-394.  Brandt T. (2000). Vestibulopathic gait: you're better off running than walking. Curr Opin Neurol 13, 3-5.  Brandt T, Strupp M & Benson J. (1999). You are better off running than walking with acute vestibulopathy. Lancet 354, 746.  Brillinger DR. (1974). Cross-spectral analysis of processes with stationary increments including the stationary queue. Ann Probab 2, 815-827.  Britton TC, Day BL, Brown P, Rothwell JC, Thompson PD & Marsden CD. (1993). Postural electromyographic responses in the arm and leg following galvanic vestibular stimulation in man. Exp Brain Res 94, 143-151.  Brooks JX & Cullen KE. (2009). Multimodal integration in rostral fastigial nucleus provides an estimate of body movement. J Neurosci 29, 10499-10511.  Brown P, Rothwell JC, Thompson PD, Britton TC, Day BL & Marsden CD. (1991). New observations on the normal auditory startle reflex in man. Brain 114, 1891-1902.  Bunn LL & Day BL. (2008). Sensory mechanisms of balance control in cerebellar disease. University College London, London.  Burke D, Hagbarth K & Lofstedt L. (1978). Muscle spindle activiy in man during shortening and lengthening contractions. J Physiol 277, 131-142.    162 Cathers I, Day BL & Fitzpatrick RC. (2005). Otolith and canal reflexes in human standing. J Physiol 563, 229-234.  Chan YS, Kasper J & Wilson VJ. (1987). Dynamics and direction sensitivity of neck muscle spindle responses to head rotation. J Neurophysiol 57, 1716-1729.  Coats AC. (1973). The variability of the galvanic body-sway test. Ann Otol Rhinol Laryngol 82, 333-339.  Cohen B, Yakushin SB & Holstein GR. (2012). What does galvanic vestibular stimulation actually activate? Front Neuro-Otol 2, 1-2.  Crapse TB & Sommer MB. (2008). Corollary discharge across the animal kingdom. Nature Rev Neurosci 9.  Creath R, Kiemal T, Horak F, Peterka R & Jeka JJ. (2005). A unified view of quiet and unperturbed stance: simultaneous co-existing excitable modes. Neurosci Lett 377, 75- 80.  Cullen KE, Brookes JX, Jamali M, Carriot J & Massot C. (2011). Internal models of self motion: computations that suppress vestibular reafference in early vestibular processing. Exp Brain Res 210, 377-388.  Dakin CJ, Blouin JS & Day BL. (2009). Spatial transformations of vestibular reflexes in cerebellar ataxia patients. University of British Columbia & University College London.  Dakin CJ, Heroux M, Luu BL, Inglis JT & Blouin JS. (2010a). Lower limb motor unit response to galvanic vestibular stimulation during human standing balance. In Motoneuron meeting. Paris.  Dakin CJ, Inglis JT & Blouin JS. (2011). Short and medium latency muscle responses evoked by electrical vestibular stimulation are a composite of all stimulus frequencies. Exp Brain Res 209, 345-354.  Dakin CJ, Lee Son GM, Inglis JT & Blouin JS. (2007). Frequency response of human vestibular reflexes characterized by stochastic stimuli. J Physiol 583, 1117-1127.  Dakin CJ, Luu BL, van den Doel K, Inglis JT & Blouin JS. (2010b). Frequency-specific modulation of vestibular-evoked sway responses in humans. J Neurophysiol 103, 1048-1056.  Davies HE & Edgley SA. (1994). Inputs to group II-activated midlumbar interneurones from descending motor pathways in the cat. J Physiol 479, 463-473.  Day BL & Fitzpatrick RC. (2005). Virtual head rotation reveals a process of route   163 reconstruction from human vestibular signals. J Physiol 567, 591-597.  Day BL & Guerraz M. (2007). Feedforward vs feedback modulation of human vestibular- evoked balance responses by visual self-motion information. J Physiol 582, 153-161.  Day BL, Marsden JF, Ramsay EJ, Mian OS & Fitzpatrick RC. (2010). Non-linear vector summation of the left and right vestibular signals for human balance. J Physiol 588, 671-682.  Day BL, Severac Cauquil A, Bartolomei L, Pastor MA & Lyon IN. (1997). Human body- segment tilts induced by galvanic stimulation: a vestibularly driven balance protection mechanism. J Physiol 500 661-672.  DiCaprio RA & Clarac F. (1981). Reversal of a walking leg reflex elicited by a muscle receptor. J Exp Biol 90, 197-203.  Duysens J, Tax AAM, Trippel M & Dietz V. (1992). Phase-dependent reflex reversal of reflexively induced movements during human gait. Exp Brain Res 90, 404-414.  Duysens J, Trippel M, Horstmann GA & Dietz V. (1990). Gating and reversal of reflexes in ankle muscles during human walking. Exp Brain Res 82, 351-358.  Edwards JH, Heitler WJ & Krasne FB. (1999). Fifty years of a command neuron: the neurobiology of escape behavior in the crayfish. Trends Neurosci 22, 153-161.  Ezure K & Wilson VJ. (1983). Dynamics of neck-to-forelimb reflexes in the decerebrate cat. J Neurophysiol 50, 688-695.  Ezure K & Wilson VJ. (1984). Interaction of tonic neck and vestibular reflexes in the forelimb of the decerebrate cat. Exp Brain Res 54, 289-292.  Fitzpatrick R, Burke D & Gandevia SC. (1994). Task-dependent reflex responses and movement illusions evoked by galvanic vestibular stimulation in standing humans. J Physiol 478 363-372.  Fitzpatrick RC, Burke D & Gandevia SC. (1996). Loop gain of reflexes controlling human standing measured with the use of postural and vestibular disturbances. J Neurophysiol 76, 3994-4008.  Fitzpatrick RC, Butler JE & Day BL. (2006). Resolving head rotation for human bipedalism. Curr Biol 16, 1509-1514.  Fitzpatrick RC & Day BL. (2004). Probing the human vestibular system with galvanic stimulation. J Appl Physiol 96, 2301-2316.  Fitzpatrick RC, Wardman DL & Taylor JL. (1999). Effects of galvanic vestibular stimulation   164 during human walking. J Physiol 517, 931-939.  Forbes PA, Dakin CJ, Geers A, Happee R, Schouten AC & Blouin JS. (2012). Comparison of multisine versus white noise stochastic stimulus. Delft University & University of British Columbia  Forssberg H, Grillner S & Rossignol S. (1977). Phasic gain control of reflexes from the dorsum of the paw during spinal locomotion. Brain Res 132, 121-139.  Gage WH, Winter DA, Frank JS & Adkin AL. (2004). Kinematic and kinetic validity of the inverted pendulum model in quiet standing Gait Posture 19, 124-132.  Gandevia SC, Smith JL, Crawford M, Proske U & Taylor JL. (2006). Motor commands contribute to human position sense. J Physiol 571, 703-710.  Gdowski GT & McCrea RA. (1999). Integration of vestibular and head movement signals in the vestibular nuclei during whole-body rotation. J Neurophysiol 82, 436-449.  Gdowski GT & McCrea RA. (2000). Neck proprioceptive inputs to primate vestibular nucleus neurons. Exp Brain Res 135, 511-526.  Ghanim Z, Lamy JC, Lackmy A, Achache V, Roche N, Penicaud A, Meunier S & Katz R. (2009). Effects of galvanic mastoid stimulation in seated human subjects. J Appl Physiol 106, 893-903.  Goldberg JM, Fernandez C & Smith CE. (1982). Responses of vestibular-nerve afferents in the squirrel monkey to externally applied currents. Brain Res 252, 156-160.  Goldberg JM, Smith CE & Fernandez C. (1984). Relation between discharge regularity and responses to externally applied galvanic currents in vestibular nerve afferents of the squirrel monkey. J Neurophysiol 51, 1236-1256.  Gottlieb GL & Agarwal GC. (1979). Response to sudden torques about ankle in man: myotatic reflex. J Neurophysiol 42, 91-106.  Grewel T, James C & Macefield V. (2009). Frequency dependent modulation of muscle sympathetic nerve activity by sinusoidal galvanic vestibular stimulation in human subjects Exp Brain Res 197, 379-386.  Grossman GE, Leigh RJ, Abel LA, Lanska DJ & Thurston SE. (1988). Frequency and velocity of rotational head perturbations during locomotion. Exp Brain Res 70, 470- 476.  Grossman GE, Leigh RJ, Bruce EN, Huebner WP & Lanska DJ. (1989). Performance of the human vestibuloocular reflex during locomotion. J Neurophysiol 62, 264-272.    165 Gurfinkel VS, Cacciatore TW, Cordo P, Horak F, Nutt J & Skoss R. (2006). Postural tone in the body axis of healthy humans. J Neurophysiol 96, 2678-2687.  Gurfinkel VS, Popov KE, Smetanin BN & Shlykov VY. (1989). Changes in the direction of vestibulomotor response in the course of adaptation to protracted static head turning in man. Neurophysiol 21, 159-164.  Halliday DM, Rosenberg JR, Amjad AM, Breeze P, Conway BA & Farmer SF. (1995). A framework for the analysis of mixed time series/point process data-theory and application to the study of physiological tremor, single motor unit discharges and electromyograms. Prog Biophys molec Biol 64, 237-278.  Highstein SM, Goldberg JM, Moschovakis AK & Fernandez C. (1987). Inputs from regularly and irregularly discharging vestibular nerve afferents to secondary neurons in the vestibular nuclei of the squirrel monkey. II. Correlation with output pathways of secondary neurons. J Neurophysiol 58, 719-738.  Hultborn H, Illert M & Santini M. (1976). Convergence on interneurons mediating the reciprocal Ia inhibition of motorneurons III. effects from supraspinal centers. Acta Physiol Scand 96, 368-391.  Huterer M & Cullen KE. (2002). Vestibuloocular reflex dynamics during high-frequency and high-acceleration rotations of the head on body in rhesus monkey. J Neurophysiol 88, 13-28.  Iles JF, Baderin R, Tanner R & Simon A. (2007). Human standing and walking: comparison of the effects of stimulation of the vestibular system. Exp Brain Res 178, 151-166.  Iles JF & Pisini JV. (1992). Vestibular-evoked postural reactions in man and modulation of transmission in spinal reflex pathways. J Physiol 455, 407-424.  Ito M. (1970). Neurophysiological aspects of the cerebellar motor control system. Int J Neurol 7, 162-176.  Ito M & Yoshida M. (1964). The cerebellar evoked monosynaptic inhibition of deiters' neurons. Experientia 20, 515-516.  Jahn K, Strupp M, Schneider E, Dieterich M & Brandt T. (2000). Differential effects of vestibular stimulation on walking and running Neuroreport 11, 1745-1748.  Jeka JJ, Oie K, Schoner G, Dijkstra T & Henson E. (1998). Position and velocity coupling of postural sway to somatosensory drive. J Neurophysiol 79, 1661-1674.  Kammermeier S, Kleine J & Buttner U. (2009). Vestibular-neck interaction in cerebellar patients. Ann N Y Acad Sci 1164, 394-399.    166 Kasper J, Schor RH, Yates BJ & Wilson VJ. (1988). Three-dimensional sensitivity and caudal projection of neck spindle afferents. J Neurophysiol 59, 1497-1509.  Kelly JP. (2000). Vestibular System. In Principles of neural science, 4th edn, ed. Kandel ER, Schwartz JH & Jessel TM, pp. 1414. McGraw-Hill Medical.  Kiemal T, Elahi AJ & Jeka JJ. (2008). Identification of the plant for upright stance in humans: multiple movement patterns from a single neural strategy. J Neurophysiol 100, 3394- 3406.  Kiemal T, Zhang Y & Jeka JJ. (2011). Identification of neural feedback for upright stance in humans: stabilization rather than sway minimization. J Neurosci 31, 15144-15153.  Kim J & Curthoys IS. (2004). Responses of primary vestibular neurons to galvanic vestibular stimulation (GVS) in the anaesthetized guinea pig. Brain Res Bull 64, 265-271.  Kleine JF & Grusser OJ. (1996). Responses of rat primary afferent vestibular neuron to galvanic polarization of the labyrinth. Ann N Y Acad Sci 19, 639-641.  Kleine JF, Guan Y, Kipiani E, Glonti L, Hoshi M & Buttner U. (2004). Trunk position influences vestibular responses of fastigial nucleus neurons in the alert monkey. J Neurophysiol 91, 2090-2100.  Krasne FB & Bryan JS. (1973). Habituation: regulation through presynaptic inhibition. Science 182, 590-592.  Latt LD, Sparto PJ, Furman JM & Redfern MS. (2003). The steady-state postural response to continuous sinusoidal galvanic vestibular stimulation Gait Posture 18, 64-72.  Lee Son GM, Inglis JT & Blouin JS. (2008). Short-duration galvanic vestibular stimulation evokes prolonged balance responses. J Appl Physiol 105, 1210-1217.  Liechti M, Muller R, Lam T & Curt A. (2008). Vestibulospinal responses in motor incomplete spinal cord injury. Clin Neurophysiol 119, 2804-2812.  Lindsay KW, Roberts TDM & Rosenberg JR. (1976). Asymmetric tonic labyrinth reflexes and their interaction with neck reflexes in the decerebrate cat. J Physiol 261, 583-601.  Logan D, Kiemal T, Dominici N, Cappellini G, Ivanenko YP, Lacquaniti F & Jeka JJ. (2010). The many roles of vision. Exp Brain Res 206, 337-350.  Loram ID, Maganaris CN & Lakie M. (2005). Human postural sway results from frequent, ballistic bias impulses by soleus and gastrocnemius. J Physiol 564, 295-311.  Lund S & Broberg C. (1983). Effects of different head positions on postural sway in man induced by a reproducible vestibular error signal. Acta Physiol Scand 117, 307-309.   167  Lyon I & Day BL. (2005). Predictive control of body mass trajectory in a two-step sequence. Exp Brain Res 161, 193-200.  MacDougall HG, Moore ST, Curthoys IS & Black FO. (2006). Modeling postural instability with galvanic vestibular stimulation. Exp Brain Res 172, 208-220.  MacKinnon CD & Winter DA. (1993). Control of whole body balance in the frontal plane during human walking. J Biomech 26, 633-644.  Magnus R. (1924). Body Posture Amerind Publishing Co., New Delhi.  Mann RA & Hagy J. (1980). Biomechanics of walking, running and sprinting. Am J Sports Med 8, 345-350.  Manzoni D, Pompeiano O & Andre P. (1998). Neck influences on the spatial properties of vestibulospinal reflex in decerebrate cats: Role of the cerebellar anterior vermis. Journal of Vestibular Research 8, 283-297.  Manzoni D, Pompeiano O, Bruschini L & Andre P. (1999). Neck input modifies the reference frame for coding labyrinthine signals in the cerebellar vermis: a cellular analysis. Neuroscience 93, 1095-1107.  Marlinsky VV. (1989). The influence of adequate vestibular stimulation on evoked locomotor muscle activity in the decerebrated guinea-pig. Neuroscience 33, 643-650.  Marlinsky VV. (1992). Activity of lateral vestibular nucleus neurons during locomotion in the decerebrate Guinea pigs. Exp Brain Res 90, 583-588.  Marsden JF, Playford DE & Day BL. (2005). The vestibular control of balance after stroke. J Neurol Neurosurg Psychiatry 76, 670-678.  Matsuyama K & Drew T. (2000a). Vestibulospinal and reticulospinal neuronal activity during locomotion in the intact cat. I. Walking on a level surface. J Neurophysiol 84, 2237- 2256.  Matsuyama K & Drew T. (2000b). Vestibulospinal and reticulospinal neuronal activity during locomotion in the intact cat. II. Walking on an inclined plane. J Neurophysiol 84, 2257-2276.  Matthews PBC. (1993). Interaction between short- and long-latency components of the human stretch reflex during sinusoidal stretching. J Physiol 462, 503-527.  McCouch GP, Deering ID & Ling TH. (1951). Location of receptors for tonic neck reflexes. J Neurophysiol 14, 191-195.    168 McCrea RA, Chen-Huang C, Belton T & Gdowski GT. (1996). Behavior contingent processing of vestibular sensory signals in the vestibular nuclei. Ann N Y Acad Sci 781, 292-303.  McCrea RA, Gdowski GT, Boyle R & Belton T. (1999). Firing behavior of vestibular neurons during active and passive head movements: vestibulo-spinal and other non-eye movement related neurons. J Neurophysiol 82, 416-428.  Miall RC & Wolpert DM. (1996). Forward models for physiological motor control. Neural Netw 9, 1265-1279.  Mian OS, Dakin CJ, Blouin JS, Fitzpatrick RC & Day BL. (2010). Lack of otolith involvement in balance responses evoked by mastoidal stimulation. J Physiol 588, 4441-4451.  Mian OS & Day BL. (2009). Determining the direction of vestibular-evoked balance responses using stochastic vestibular stimulation. J Physiol 587, 2869-2873.  Misiaszek JE. (2006). Neural control of walking balance: if falling then react else continue. Exerc Sport Sci Rev 34, 128-134.  Moore GP, Segundo JP, Perkel DH & Levitan H. (1970). Statistical signs of synaptic interaction in neurons. Biophys J 10.  Moore ST, MacDougall HG, Peters BT, Bloomberg JJ, Curthoys IS & Cohen HS. (2006). Modeling locomotor dysfunction following spaceflight with galvanic vestibular stimulation. Exp Brain Res 174, 647-659.  Muise SB, Lam CK & Bent LR. (2012). Reduced input from the foot sole skin through cooling differentially modulates the short latency and medium latency vestibular reflex responses to galvanic vestibular stimulation Exp Brain Res 218, 63-71.  Nashner LM & Wolfson P. (1974). Influence of head position and proprioceptive cues on short latency postural reflexes evoked by galvanic stimulation of the human labyrinth. Brain Res 67, 255-268.  Nilsson J, Thorstensson A & Halbertsma J. (1985). Changes in leg movements and muscle activity with speed of locomotion and mode of progression in humans. Acta Physiol Scand 123, 457-475.  Olney SJ & Winter DA. (1985). Predictions of knee and ankle moments of force in walking from EMG and kinematic data. J Biomech 18, 9-20.  Orlovsky GN. (1972). Activity of vestibulospinal neurons during locomotion. Brain Res 46, 85-98.    169 Pastor MA, Day BL & Marsden CD. (1993). Vestibular induced postural responses in Parkinson's disease. Brain 116 (Pt 5), 1177-1190.  Pavlik AE, Inglis JT, Lauk M, Oddson L & Collins JJ. (1999). The effect of stochastic galvanic vestibular stimulation on human postural sway. Exp Brain Res 124, 273-280.  Peterson BW & Azbug C. (1975). Properties of projections from vestibular nuclei to medial reticular formation in the cat. J Neurophysiol 38, 1421-1435.  Pierrot-Deseilligny E & Burke DC. (2005). The circuitry of the human spinal cord: its role in motor control and movement disorders. Cambridge University Press, Cambridge, UK; New York.  Popov KE, Smetanin BN, Gurfinkel VS, Kudinova MP & Shlykov VY. (1986). Spatial perception and vestibulomotor responses in man. Neurophysiol 18, 548-554.  Reynolds RC. (2010). The effect of voluntary sway control on the early and late components of the vestibular-evoked postural response. Exp Brain Res 201, 133-139.  Rosenberg JR, Amjad AM, Breeze P, Brillinger DR & Halliday DM. (1989). The fourier approach to the identification of functional coupling between neuronal spike trains. Prog Biophys molec Biol 53, 1-31.  Rosengren SM & Colebatch JG. (2002). Differential effect of current rise time on short and medium latency vestibulospinal reflexes. Clin Neurophysiol 113, 1265-1272.  Roskell W, Dakin CJ, McFadyen BJ, Chua R, Inglis TJ & Blouin JS. (2007). Vestibular reflexes evoked by stochastic vestibular stimulation in walking humans. Soc Neurosci.  Roy JE & Cullen JE. (2001). Selective processing of vestibular reafference during self- generated head motion. J Neurosci 21, 2131-2142.  Roy JE & Cullen KE. (2004). Dissociating self-generated from passively applied head motion: neural mechanisms in the vestibular nuclei. J Neurosci 24, 2102-2111.  Russell DF & Zajac FE. (1979). Effects of stimulating Deiters' nucleus and medial longitudinal fasciculus on the timing of the fictive locomotor rhythm induced in cats by DOPA. Brain Res 177, 588-592.  Sadeghi SG, Chacron MJ, Taylor MC & Cullen KE. (2007). Neural variability, detection thresholds, and information transmission in the vestibular system J Neurosci 27, 771- 781.  Schlosser HG, Guldin WO & Grusser OJ. (2001). Tuning in fastigial units during natural and galvanic labyrinth stimulation. Neuroreport 12, 1443-1447.    170 Scinicariello AP, Eaton K, Inglis JT & Collins JJ. (2001). Enhancing human balance control with galvanic vestibular stimulation. Biol Cybern 84, 475-480.  Severac Cauquil A, Martinez P, Ouaknine M & Tardy-Gervet M. (2000). Orientation of the body response to galvanic stimulation as a function of the inter-vestibular imbalance. Exp Brain Res 133, 501-505.  Severac Cauquil A, Tardy-Gervet MF & Ouaknine M. (1998). Body response to binaural monopolar galvanic vestibular stimulation in humans. Neurosci Lett 245, 37-40.  Shaikh AG, Meng H & Angelaki DE. (2004). Multiple reference frames for motion in the primate cerebellum. J Neurosci 24, 4491-4497.  Sherrington CS. (1898). Decerebrate rigidity, and reflex coordination of movements. J Physiol, 13.  Sherrington CS. (1910). Flexion-reflex of the limb, crossed extension-reflex, and reflex stepping and standing. J Physiol 40, 28-121.  Sperry RW. (1950). Neural basis of the spontaneous optokinetic response produces by visual inversion. J Comp Physiol Psychol 43, 482-489.  Udo W, Kamei H, Matsukawa K & Tanaka K. (1982). Interlimb coordination in cat locomotion investigated with perturbation. Exp Brain Res 46, 438-447.  van der Kooij H, van Asseldonk E & van der Helm FCT. (2005). Comparison of different methods to identify and quantify balance control. J Neurosci Meth 145, 175-203.  von Holst E. (1973). The behavioural physiology of animals and man, vol. 1. Methuen & Co, Coral Gables.  Walberg F & Jansen I. (1961). Cerebellar corticovestibular fibers in the cat. Exp Neurol 3, 32- 52.  Wilson VJ, Kato M, Peterson BW & Wylie RM. (1967). A single unit analysis of the organization of Deiters nucleus. J Neurophysiol 30, 603-619.  Wilson VJ & Peterson BW. (1978). Peripheral and central substrates of vestibulospinal reflexes Physiol Rev 58, 80-105.  Winter DA. (1995). Human balance and posture control during standing and walking. Gait Posture 3, 193-214.  Wolpert DM, Goodbody SJ & Husain M. (1998). Maintaining internal representations: the role of the human superior parietal lobe. Nat Neurosci 1, 529-533.    171 Wolpert DM & Kawato M. (1998). Multiple paired forward and inverse models for motor control. Neural Netw 11, 1317-1329.  Yang JF & Stein RB. (1990). Phase-dependent reflex reversal in human leg muscles during walking. J Neurophysiol 63, 1109-1117.  Yang JF & Winter DA. (1985). Surface EMG profiles during different walking profiles. Electroencephalogr Clin Neurophysiol 60, 485-491.  Zhan Y, Halliday DM, Jiang P, Liu X & Feng J. (2006). Detecting time-dependent coherence between non-stationary electrophysiological signals- A combined statistical and time- frequency approach. J Neurosci Meth 156, 322-332.  Zink R, Bucher SF, Weiss A, Brandt T & Dieterich M. (1998). Effects of galvanic vestibular stimulation on otolithic and semicircular canal eye movements and perceived vertical. Electroencephalogr Clin Neurophysiol 107, 200-205.    

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0073200/manifest

Comment

Related Items