UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The neurophysiology of sensorimotor integration in healthy aging and chronic stroke Brown, Katlyn Elizabeth 2017

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

Item Metadata

Download

Media
24-ubc_2017_november_brown_katlyn.pdf [ 1.44MB ]
Metadata
JSON: 24-1.0355869.json
JSON-LD: 24-1.0355869-ld.json
RDF/XML (Pretty): 24-1.0355869-rdf.xml
RDF/JSON: 24-1.0355869-rdf.json
Turtle: 24-1.0355869-turtle.txt
N-Triples: 24-1.0355869-rdf-ntriples.txt
Original Record: 24-1.0355869-source.json
Full Text
24-1.0355869-fulltext.txt
Citation
24-1.0355869.ris

Full Text

 THE NEUROPHYSIOLOGY OF SENSORIMOTOR INTEGRATION IN HEALTHY AGING AND CHRONIC STROKE by  Katlyn Elizabeth Brown  B.Sc., University of Waterloo, 2011 M.Sc., University of British Columbia, 2013  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Rehabilitation Science)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  September 2017  © Katlyn Elizabeth Brown, 2017   ii Abstract Background: General decline in motor performance is often cited in healthy aging, and motor deficits persist into the chronic phase of stroke recovery. Abnormalities in sensorimotor integration may underlie these motor deficits; however, the effect of aging and chronic stroke on sensorimotor integration neurophysiology is not well understood. Further, investigation into the plasticity of sensorimotor integration is important to establish in populations experiencing sensorimotor decline. Methods: The overall objective of this thesis was to comprehensively understand the neurophysiology of sensorimotor integration, including the influence of aging and chronic stroke on sensorimotor integration and the reliability of common neurophysiological measures. The first research chapter (Chapter 2) explores age and stroke-related differences in measures of indirect sensorimotor integration. Chapters 3 and 4 investigate baseline differences in measures of direct sensorimotor integration induced by aging and chronic stroke, respectively. Further, they use an intervention to index plasticity of sensorimotor integration in these populations. The final chapter (Chapter 5) addresses the reliability of a variety of neurophysiological methods used to examine somatosensory and motor cortical excitability, as well as sensorimotor integration. Summary of Findings: In Chapter 2, older individuals and individuals with chronic stroke had reduced short-latency afferent inhibition, compared to younger individuals suggesting the difference is largely driven by age-related factors. Greater disinhibition post-stroke related to worse motor function and impairment. Chapter 3 showed that measures of direct sensorimotor integration are less susceptible to age-related changes   iiithan measures of indirect sensorimotor integration. Sensory training altered direct but not indirect sensorimotor integration, suggesting independent modulation of separate anatomical pathways of sensorimotor integration. Chapter 4 highlighted differences in direct sensorimotor integration between individuals with chronic stroke and older individuals such that vibration had less of an impact on baseline motor cortex excitability in individuals post-stroke and the intracortical response to sensory training was reduced. Chapter 5 showed high reliability in transcranial magnetic stimulation thresholds, the amplitudes of evoked potentials elicited at high stimulation intensities, and latency-based measures.  Conclusions: This dissertation contributes new knowledge to the impact of aging and chronic stroke on sensorimotor integration and the reliability of the measures used to quantify sensorimotor integration.   ivLay summary There is an important reciprocal relationship between sensation and movement. Sensation guides movement as it informs where our limbs are in space and in reference to where we want to move. Conversely, movement produces sensation that allows for online corrections; for example, if movement is perturbed, this sensory information must be incorporated in order to successfully adjust a movement to meet the original goal. There may be a disruption in this relationship that occurs with healthy aging and following a stroke that underpins movement-based deficits. The current thesis explores age and stroke-related changes to neurobiological systems that underlie this relationship. Additionally, the thesis examines whether these systems can be altered with an intervention in order to potentially combat age and stroke related decline. Finally, it explores reliability in the measurements used to test sensorimotor neurophysiology to ensure differences are being correctly identified.   vPreface The work in this dissertation was conceived, designed, conducted, analysed, and written by Katlyn Brown. The University of British Columbia’s Clinical Research Ethics Board approved all research included in this dissertation: Certificate #H12-03453, #H07-01759.  Katlyn Brown wrote chapters 1 and 6. Drs. Lara Boyd and Richard Staines provided assistance in editing these chapters.  Chapter 2 is based on work conducted by Katlyn Brown, Dr. Jason Neva, Samantha Feldman, and Drs. Richard Staines and Lara Boyd. Katlyn Brown conceived, designed, collected, analysed, and interpreted the data, as well as writing and revising the manuscript. Dr. Neva and Ms. Feldman assisted with data collection, interpretation, and manuscript editing. Drs. Staines and Boyd aided in the development of the study design and manuscript preparation.  Chapter 3 is based on work conducted by Katlyn Brown, Dr. Jason Neva, Samantha Feldman, and Drs. Richard Staines and Lara Boyd. Katlyn Brown conceived, designed, collected, analysed, and interpreted the data, as well as writing and revising the chapter. Ms. Feldman assisted with data collection. Dr. Neva aided in data collection, interpretation, and editing of the chapter. Drs. Staines and Boyd contributed to development of the study design and editing the chapter.    viChapter 4 is based on work conducted by Katlyn Brown, Dr. Jason Neva, Samantha Feldman, and Drs. Richard Staines and Lara Boyd. Katlyn Brown conceived, designed, collected, analysed, and interpreted the data, as well as writing and revising the chapter. Ms. Feldman assisted with data collection. Dr. Neva aided in data collection, interpretation, and editing of the chapter. Drs. Staines and Boyd contributed to development of the study design and editing the chapter.  Chapter 5 is based on work conducted as part of an international collaboration with sites in London, Leiden, Paris, and Vancouver. The list of collaborators is as follows: Katlyn Brown, Dr. Keith Lohse, Isabella Mayer, Dr. Gionato Strigaro, Dr. Mahalekshmi Desikan, Dr. Elias Casula, Dr. Sabine Meunier, Dr. Trian Popa, Dr. Jean-Charles Lamy, Omar Odish, Dr. Blair Leavitt, Dr. Alexandra Burr, Dr. Raymond Roos, Dr. Sarah Tabrizi, Dr. John Rothwell, Dr. Lara Boyd, and Dr. Michael Orth. Katlyn Brown, conceived, designed, collected, analysed, and interpreted the data, as well as writing and revising the manuscript. Ms. Mayer, Dr. Strigaro, Dr. Lamy, and Mr. Odish aided in data collection at the international sites. Ms. Mayer also helped in data analysis. Dr. Lohse aided in statistical analysis, as well as interpretation. Drs. Boyd, Rothwell, and Orth were important in conceptualisation of the study, interpretation, and manuscript preparation. All other contributors were involved in the larger collaboration and aided in manuscript preparation and editing.    vii A version of Chapter 2 has been submitted for publication: Brown KE, Neva JL, Feldman SJ, Staines WR, Boyd LA. Exploring the influence of age and chronic stroke on the neurophysiology of sensorimotor integration. A version of Chapter 5 is under review for publication: Brown KE, Lohse KR, Mayer IMS, Strigaro G, Desikan M, Casula EP, Meunier S, Popa T, Lamy J-C, Odish O, Leavitt BR, Durr A, Roos RAC, Tabrizi SJ, Rothwell JC, Boyd LA, Orth M. The reliability of commonly used electrophysiology measures.  Dissertation chapters may include additional details than those submitted for publication in order to improve cohesion across the full document.    viiiTable of contents  Abstract .............................................................................................................................. ii Lay summary .....................................................................................................................iv Preface ................................................................................................................................. v Table of contents ............................................................................................................ viii List of tables.....................................................................................................................xiv List of figures .................................................................................................................... xv List of abbreviations ..................................................................................................... xvii Acknowledgements .........................................................................................................xix Chapter 1: General introduction ...................................................................................... 1 1.1  Preamble ................................................................................................................. 1 1.2  Sensorimotor integration ......................................................................................... 2 1.2.1  Neuroanatomy of sensorimotor integration ..................................................... 3 1.2.2  Testing sensorimotor integration ..................................................................... 5 1.2.2.1  Transcranial magnetic stimulation ............................................................ 7 1.2.2.2  Testing indirect sensorimotor integration ................................................. 9 1.2.2.3  Testing direct sensorimotor integration .................................................. 11 1.2.3  Aging and sensorimotor integration ............................................................... 13 1.2.3.1  Aging: Somatosensory and motor function ............................................ 13 1.2.3.2  Aging: Somatosensory and motor neurophysiology ............................... 14 1.2.3.3  Aging: Sensorimotor integration neurophysiology ................................. 15 1.2.4  Chronic stroke and sensorimotor integration ................................................. 17   ix1.2.4.1  Chronic stroke: Somatosensory and motor function ............................... 17 1.2.4.2  Chronic stroke: Somatosensory and motor neurophysiology ................. 17 1.2.4.3  Chronic stroke: Sensorimotor integration neurophysiology ................... 20 1.3  Neuroplasticity ...................................................................................................... 22 1.3.1  Mechanisms of neuroplasticity ...................................................................... 22 1.3.2  Motor and somatosensory cortical plasticity ................................................. 24 1.3.3  Attention and neuroplasticity ......................................................................... 25 1.3.4  Plasticity of sensorimotor integration ............................................................ 27 1.3.5  Plasticity and aging ........................................................................................ 28 1.3.6  Plasticity and chronic stroke .......................................................................... 30 1.4  Reliability of neurophysiological measures .......................................................... 32 1.5  Thesis overview .................................................................................................... 33 1.5.1  Thesis impact ................................................................................................. 34 1.5.2  Research objectives ........................................................................................ 35 Chapter 2: Exploring the influence of age and chronic stroke on the neurophysiology of indirect sensorimotor integration ................................................. 37 2.1  Introduction ........................................................................................................... 37 2.2  Methods................................................................................................................. 40 2.2.1  Participants ..................................................................................................... 40 2.2.2  Experimental design....................................................................................... 40 2.2.3  Behavioural assessment ................................................................................. 41 2.2.4  Neurophysiological assessment ..................................................................... 41 2.2.4.1  Somatosensory evoked potentials ........................................................... 42   x2.2.4.2  Transcranial magnetic stimulation .......................................................... 42 2.2.4.3  Sensorimotor integration ......................................................................... 43 2.2.4.4  Statistics .................................................................................................. 44 2.3  Results ................................................................................................................... 44 2.3.1  Stroke and age-related changes in sensorimotor integration ......................... 46 2.3.2  Association between neurophysiology, impairment, and function ................ 49 2.3.3  Somatosensory and motor cortical excitability .............................................. 50 2.4  Discussion ............................................................................................................. 51 2.4.1  Stroke and age-related changes in sensorimotor integration ......................... 51 2.4.2  Relationship between neurophysiology, impairment, and function............... 54 2.4.3  Somatosensory and motor cortical excitability .............................................. 56 2.4.4  Limitations ..................................................................................................... 57 2.5  Conclusions ........................................................................................................... 57 Chapter 3: Sensorimotor integration and aging: baseline differences and response to sensory training ................................................................................................................ 58 3.1  Introduction ........................................................................................................... 58 3.2  Methods................................................................................................................. 61 3.2.1  Participants ..................................................................................................... 61 3.2.2  Experimental design....................................................................................... 62 3.2.3  Somatosensory functional tests ...................................................................... 62 3.2.4  Neurophysiological assessment ..................................................................... 62 3.2.4.1  Somatosensory evoked potentials ........................................................... 63 3.2.4.2  Transcranial magnetic stimulation .......................................................... 63   xi3.2.4.3  Direct sensorimotor integration .............................................................. 64 3.2.4.4  Indirect sensorimotor integration ............................................................ 65 3.2.5  Sensory training ............................................................................................. 65 3.2.6  Statistical tests ................................................................................................ 66 3.3  Results ................................................................................................................... 67 3.3.1  Baseline direct sensorimotor integration ....................................................... 67 3.3.2  Influence of sensory training on sensorimotor integration ............................ 69 3.3.3  Secondary measures – indirect sensorimotor integration .............................. 71 3.3.4  Sensory training performance ........................................................................ 72 3.4  Discussion ............................................................................................................. 72 3.4.1  Baseline sensorimotor integration .................................................................. 72 3.4.2  Neurophysiological response to sensory training .......................................... 74 3.4.3  Limitations ..................................................................................................... 77 3.5  Conclusions ........................................................................................................... 78 Chapter 4: Sensorimotor integration in chronic stroke: baseline differences and response to sensory training ............................................................................................ 79 4.1  Introduction ........................................................................................................... 79 4.2  Methods................................................................................................................. 83 4.2.1  Participants ..................................................................................................... 83 4.2.2  Experimental design....................................................................................... 84 4.2.3  Behavioural tests ............................................................................................ 84 4.2.4  Neurophysiological assessment ..................................................................... 85 4.2.4.1  Somatosensory evoked potentials ........................................................... 85   xii 4.2.4.2  Transcranial magnetic stimulation .......................................................... 85 4.2.4.3  Sensorimotor integration ......................................................................... 86 4.2.4.4  Sensory training ...................................................................................... 88 4.2.4.5  Statistical tests ......................................................................................... 88 4.3  Results ................................................................................................................... 89 4.3.1  Baseline direct sensorimotor integration ....................................................... 89 4.3.2  Functional relationships ................................................................................. 91 4.3.3  Influence of sensory training on direct sensorimotor integration .................. 92 4.3.4  Influence of sensory training on secondary measures ................................... 94 4.3.5  Performance ................................................................................................... 94 4.4  Discussion ............................................................................................................. 95 4.4.1  Baseline sensorimotor integration .................................................................. 96 4.4.2  Influence of sensory training on sensorimotor integration ............................ 99 4.4.3  Limitations ................................................................................................... 102 4.5  Conclusions ......................................................................................................... 103 Chapter 5: The reliability of commonly used electrophysiology measures .............. 104 5.1  Introduction ......................................................................................................... 104 5.2  Methods............................................................................................................... 105 5.2.1  Participants ................................................................................................... 105 5.2.2  Electrophysiology ........................................................................................ 106 5.2.2.1  Electroencephalography ........................................................................ 106 5.2.2.2  Long-latency reflexes ............................................................................ 107 5.2.2.3  Transcranial magnetic stimulation ........................................................ 107   xiii5.2.2.4  Statistical analysis ................................................................................. 108 5.3  Results ................................................................................................................. 111 5.3.1  Cohort .......................................................................................................... 111 5.3.2  Reliability and power simulations ............................................................... 113 5.3.3  Between site differences .............................................................................. 120 5.4  Discussion ........................................................................................................... 121 Chapter 6: General discussion ...................................................................................... 126 6.1  Aging and chronic stroke differentially impact sensorimotor integration .......... 126 6.2  Response to sensory training is altered post-stroke, not in healthy aging .......... 129 6.3  Measures of sensorimotor integration have varying levels of reliability ............ 131 6.4  Limitations .......................................................................................................... 133 6.5  Implications and future directions ...................................................................... 136 6.6  Conclusions ......................................................................................................... 139 References ....................................................................................................................... 140 Appendices ...................................................................................................................... 153 Appendix A .................................................................................................................. 153 A.1  Statistical rationale ......................................................................................... 153 A.2  Additional results ........................................................................................... 155    xivList of tables  Table 1.1: Neurophysiological measures of somatosensory excitability, motor cortical excitability, and sensorimotor integration. .......................................................................... 6 Table 2.1: Demographic information. ............................................................................... 45 Table 2.2: Group average data…………………………………………………………...47 Table 5.1: Attrition rates. ................................................................................................ 113 Table 5.2: ICC values for electrophysiological measures with high and moderate levels of reliability. ........................................................................................................................ 115 Table 5.3: Simulated independent t-test results. ............................................................. 117 Table 5.4: Simulated paired t-test results. ....................................................................... 118 Table 5.5: Simulated interaction results. ......................................................................... 119 Table 5.6: Between site differences. ............................................................................... 121 Table A.1: Low reliability measures. .............................................................................. 155 Table A.2: Between site effects for low reliability measures. ........................................ 156 Table A.3: Non-significant study site effects. ................................................................ 157 Table A.4: Between site effects of stimulation intensities. ............................................. 163    xvList of figures  Figure 1.1: Schematic depicting neuroanatomical connections underlying sensorimotor integration. .......................................................................................................................... 5 Figure 1.2: Overview of TMS techniques........................................................................... 8 Figure 2.1: Experimental procedures. ............................................................................... 41 Figure 2.2: Representative traces for measures of indirect sensorimotor integration in each group. ........................................................................................................................ 46 Figure 2.3: Group means for SAI (A), AF (B), and LAI (C). ........................................... 48 Figure 2.4: Relationship between SAI and motor impairment (A) and motor function (B)............................................................................................................................................ 50 Figure 3.1: Experimental procedures. ............................................................................... 62 Figure 3.2: Representative traces for measures of direct sensorimotor integration. ......... 68 Figure 3.3: Baseline direct sensorimotor integration. ....................................................... 69 Figure 3.4: Influence of sensory training on measures of direct sensorimotor integration............................................................................................................................................ 70 Figure 3.5: Influence of sensory training on indirect sensorimotor integration. .............. 71 Figure 4.1: Experimental overview................................................................................... 84 Figure 4.2: Representative trace depicting the influence of vibration on single-pulse MEP amplitudes. ........................................................................................................................ 90 Figure 4.3: Influence of vibration on measures of direct sensorimotor integration. ......... 91 Figure 4.4: Influence of sensory training on measures of direct sensorimotor integration............................................................................................................................................ 93   xviFigure 4.5: Effect of sensory training on measures of indirect sensorimotor integration. 94 Figure 5.1 Conceptual outline for including measures in study design. ......................... 111 Figure 5.2: Reliability values for each electrophysiological measure. ........................... 114    xvii List of abbreviations  ACh: Acetylcholine AF: Afferent facilitation AMPA: α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid AMT: Active motor threshold  ANOVA: Analysis of Variance APB: Abductor pollicis brevis  CNS: Central nervous system CS: Conditioning stimulus  CSP: Cortical silent period D-waves: Direct-waves EEG: Electroencephalography EMG: Electromyography FM: Fugl-Meyer fMRI: Functional magnetic resonance imaging GABA: gamma-aminobutyric acid I-waves: Indirect-waves ICC: Intraclass correlation ICF: Intracortical facilitation  ISI: Interstimulus interval LAI: Long-latency afferent inhibition LLR: Long-latency reflexes LTP: Long-term potentiation    xviiiMEP: Motor-evoked potential MVC: Maximal voluntary contraction M1: Primary motor cortex NHPT: Nine-hole peg test NMDA: N-methyl-D-aspartate PAS: Paired associative stimulation PNS: Peripheral nervous system RA: Rapidly adapting RMT: Resting motor threshold SA: Slowly adapting AF: Short-latency afferent facilitation  SAI: Short-latency afferent inhibition SEP: Somatosensory-evoked potential SICI: Short-interval intracortical inhibition SPD: Silent period duration S1: Primary somatosensory cortex TMS: Transcranial magnetic stimulation TS: Test stimulus  WMFT: Wolf Motor Function Test    xixAcknowledgements I am incredibly thankful to have had guidance from two excellent supervisors throughout this process. Lara, I have learned so much from you in the last four years and I could not have done this without your unwavering support. Your route to success through hard work, passion, and dedication, while still maintaining a strong work-life balance is inspiring. Rich, it is hard to believe that I walked into your office eight years ago; there is no doubt that I wouldn’t be here without your belief in me, your encouragement, and the endless hours of mentorship that you have provided me with. Your knowledge and insight, both professionally and personally, have been invaluable. The members of the Brain Behaviour Lab, both past and present, have all been essential to the completion of this thesis. I can’t emphasise enough how much every mental health run in the woods, every science-based conversation, and every laugh has meant. Personally, I am indebted to various friends and family members for supporting me every step of the way. Specific thanks to my parents, brother, and cousins who have patiently listened to every failure and every success, and lived the highs and lows alongside me. If I’ve learned anything throughout this degree, it’s that a PhD is not a singular effort, but depends on an extensive support system. Gratitude to everyone involved.  1Chapter 1: General introduction  1.1 Preamble Activities of daily living are highly dependent on extracting relevant somatosensory information from the periphery and incorporating it into movement through a process called sensorimotor integration. From the time we wake each morning, our brain is incorporating somatosensory information across various domains such as where our body is in space, cutaneous cues from contact with objects, temperature, and pain, and using this feedback to inform our movements to successfully interact with the environment. A reduced ability to perform activities of daily living, as well as changes to both somatosensory and motor systems, arise as a result of healthy aging and various pathologies such as stroke. Behavioural evidence is important given its functional relevance; however, it does not allow for disentanglement of the different components required to accomplish these motor tasks. For example, motor deficits may arise from a pure reduction in motor cortical excitability, less incoming somatosensory information, or from a disconnect between these two systems that limits the ability to properly use feedback to influence behaviour. Neurophysiological measures enable us to examine these processes and investigate which component(s) relate most directly to behavioural change.  The ability of the brain to change with experience, termed neuroplasticity, remains intact in older healthy individuals, and individuals with chronic stroke, though the degree of change may be reduced [1,2]. To date, neuroplastic change has primarily been evaluated using indices of the somatosensory or motor systems in isolation, with little work examining sensorimotor integration plasticity. Once the effects of aging and   2stroke on sensorimotor integration have been established, investigation into the plasticity of sensorimotor integration in these individuals is ripe for exploration. This will be especially relevant in populations with functional impairments that may be underpinned by abnormal sensorimotor integration.  The overall objective of this thesis is to comprehensively examine the neurophysiology of sensorimotor integration. The influence of both age and chronic stroke on various pathways of sensorimotor integration will be explored, followed by an investigation into the neuroplasticity of sensorimotor integration in these individuals.  The thesis concludes with an examination of the techniques used to probe the neurophysiology of sensorimotor integration. The general introduction will introduce key concepts before providing an overview of each research chapter included in this thesis. 1.2 Sensorimotor integration Sensorimotor integration broadly refers to the process through which sensory information from the periphery is incorporated into the central nervous system (CNS) to inform motor output [3]. Sensorimotor integration can take place across modalities (visual, auditory, etc), and occurs at various levels and in various structures, such as the cerebellum, association areas, and motor planning regions [3]. This thesis specifically considers sensorimotor integration as the neurophysiological process through which somatosensory information is incorporated into the primary motor cortex (M1). Somatosensory feedback is an essential component of voluntary movement; behaviourally this is demonstrated by data showing that deafferentation results in motor deficits [4-6]. Deficits in sensorimotor integration, therefore, contribute to motor impairment. This section will begin by exploring the neuroanatomical pathways   3underlying sensorimotor integration, and will then introduce experimental methodologies used to assess sensorimotor integration. The influence of both aging and chronic stroke on sensorimotor integration will then be discussed. Given the reliance of sensorimotor integration on both somatosensory and motor systems, the neurophysiology of these systems will also be addressed.  1.2.1 Neuroanatomy of sensorimotor integration  Evaluating neuroanatomical connections highlights the importance of sensorimotor integration; somatosensory information relevant to both touch and proprioception are transmitted to the cortex through large diameter, heavily myelinated afferent fibres to ensure somatosensory information can be rapidly integrated into the CNS and influence motor control [7]. Activation of peripheral somatosensory receptors determines the CNS pathways through which somatosensory information ascends to the cortex. Touch somatosensation results from the activation of cutaneous and subcutaneous mechanoreceptors, whereas proprioceptive information is derived from activation of muscle spindles, as well as Golgi Tendon Organs, and joint capsule receptors [7].  Following receptor activation, information passes along primary afferent fibres to the dorsal root ganglion, which connects the peripheral nervous system (PNS) and CNS. Both forms of sensation will primarily activate Aα and Aβ fibre groups, though the specific fibre type will differ; muscle spindles will transmit information through type 1a and type 2 afferents, whereas cutaneous mechanoreceptors will primarily travel in rapidly adapting (RA) or slowly adapting (SA) fibre types, depending on the particular receptor activated. From here, information ascends ipsilaterally to dorsal column nuclei, before   4travelling in the medial lemniscal pathway where decussation occurs at the level of the medulla [7].  Contralateral ascension to the level of the thalamus is the final relay before thalamocortical projections transmit information to the cerebral cortex. Cutaneous information synapses primarily in the lateral and medial ventral posterior nuclei of the thalamus and proprioceptive information is preferentially transferred to the superior ventral posterior nucleus [7]. From the thalamus, proprioceptive information is transmitted more directly to M1 than cutaneous information. There is evidence in humans that this information reaches Brodmann area 3a of the primary somatosensory cortex (S1), which has direct reciprocal connections to area 4 (M1), as well as animal work showing some connection directly to upper layers of M1 [8]. In contrast, cutaneous information travels through thalamocortical projections to area 3b in S1, which does not have direct connections to M1, but connects with areas 1 and 2 in S1, which can then connect with M1. More specifically, S1 monosynaptically connects to M1 through long-range pyramidal cells [9] into upper layers of M1, likely activating interneuronal circuits [9,10]. Therefore, somatosensory information may reach M1 through direct thalamocortical projections or corticocortical projections from 3a or, more indirectly through multiple Brodmann areas within S1; however, within M1, the sensorimotor integration from both routes likely takes place in interneuronal circuitry contained within the upper cortical layers of M1.   5 Figure 1.1: Schematic depicting neuroanatomical connections underlying sensorimotor integration. Somatosensory receptors carrying proprioceptive information (blue) and cutaneous information (red) ascend to the cortex in similar ways. Initially, they follow the dorsal-column medial lemniscus pathway to the level of the thalamus. From here, proprioceptive information is transmitted to Brodmann area 3a in S1, which has reciprocal connections with M1. In contrast, cutaneous information is primarily transmitted to Brodmann area 3b within S1 before being sent to M1 via areas 1 and 2.  1.2.2 Testing sensorimotor integration In order to quantify sensorimotor integration in-vivo, peripheral stimulation is paired with non-invasive brain stimulation. Specifically in this thesis, vibration or median nerve stimulation will be paired with transcranial magnetic stimulation (TMS). An overview of these methods can be found in Table 1.1 with further detail provided in the text below.   6 Somatosensory Motor Sensorimotor Integration Measure SEPs  (N20-P26) MEP SICI ICF SAI AF LAI MEPvib SICIvib ICFvib Proposed Mechanism Arrival of afferent information in S1, generated from posterior wall of central sulcus Transsynaptic activation of pyramidal neurons GABAA-receptor mediated Potential glutamatergic contributions ACh, GABAA receptor subunit Unknown, indirect facilitation of M1  Generally unknown, indirect inhibition of M1 through multiple areas (S1, S2, PPC) Influence of increased afferent feedback on transsynaptic activation of pyramidal neurons  Influence of increased afferent feedback on M1 GABAA-receptor mediated interneuronal circuits  Influence of increased afferent feedback on M1 excitatory interneuronal circuits  Table 1.1: Neurophysiological measures of somatosensory excitability, motor cortical excitability, and sensorimotor integration.SEPs: somatosensory evoked potentials, MEP: motor evoked potentials, SICI: short-interval intracortical inhibition, ICF: intracortical facilitation, SAI: short-latency afferent inhibition, AF: afferent facilitation, LAI: long-latency afferent inhibition, Vib: vibration, ACh: acetylcholine, S1: primary somatosensory cortex, GABA: gamma-aminobutyric acid, M1: primary motor cortex, S2: secondary somatosensory cortex, PPC: posterior parietal cortex. References for each measure can be found in the text. Adapted from Brown et al 2014.    71.2.2.1 Transcranial magnetic stimulation  TMS, when applied over the motor cortex, can activate neurons responsible for motor output [11]. Procedurally, a coil is placed over the scalp, and works through electromagnetic induction to generate activity in the underlying neurons. TMS can directly activate pyramidal neurons (D-waves) to stimulate the corticospinal tract; however, indirect activation through local interneurons, which synapse onto pyramidal neurons are more predominantly activated (I-waves) [12]. The contralateral muscle response to TMS can be quantified using electromyography (EMG), and is termed the motor-evoked potential (MEP; Figure 1.2). MEPs are considered to be indicative of corticospinal excitability. Using MEP amplitudes, experimenters can determine the cortical “hot spot” or the location on the scalp at which the largest amplitude MEP can be evoked in the muscle of interest [11]. This location can then be marked with real-time stereotaxic neuronavigation in order to ensure consistent coil placement within and between experimental sessions.   8 Figure 1.2: Overview of TMS techniques. A shows a figure-of-eight coil placed on the scalp producing an electrical field in the cortex. B depicts how, through electromagnetic induction, the TMS pulse activates intracortical interneurons, which in turn activate pyramidal neurons and produce a peripheral response that can be captured using surface electromyography (EMG) on the muscle of interest (C). D provides a graphical representation of a motor evoked potential that may be elicited by a single pulse of TMS. Adapted from Brown et al. 2014  In the current thesis, single-pulse MEP amplitudes are generated at a wide range of intensities based on an individual’s motor threshold to form recruitment curves. Resting motor threshold (RMT) is commonly defined as the lowest stimulation intensity required to evoke a MEP with a peak-to-peak amplitude of 50 μV in five out of ten trials [13]. Response to stimulation at threshold levels is thought to be indicative of cortical and lower motor neuron   9membrane excitability [14]. By increasing stimulation intensity, as during collection of recruitment curves, TMS recruits neurons with higher excitability thresholds than those collected at lower intensities, as well as those more distant from the coil centre [15]. As such, collecting a recruitment curve provides a more extensive examination of corticospinal excitability as compared to single-pulse MEP amplitudes at a single intensity.   In addition to providing information on corticospinal excitability, TMS pulses can also be paired in close succession to provide information on inhibitory and excitatory interneuronal circuitry. The current thesis uses short-interval intracortical inhibition (SICI) and intracortical facilitation (ICF) to probe interneuronal circuits. SICI occurs when two TMS pulses (a subthreshold conditioning stimulus (CS), followed by a suprathreshold test stimulus (TS)) are administered over M1 with an interstimulus interval (ISI) between 1-6 ms and result in a decreased MEP amplitude compared to that elicited from the TS alone [16]. The mechanism underlying SICI is thought to involve gamma-aminobutyric acid-A (GABAA) receptors, as demonstrated by an increase in SICI produced following pharmacological administration of a positive allosteric activator of GABAA receptors [17,18]. Further, SICI likely occurs in cortical layers 2 and 3 and selectively inhibits the circuits responsible for late I-wave generation [19]. ICF uses the same approach, with an ISI of 10-15 ms, in order to assess facilitatory circuits within M1. The neural mechanisms underlying ICF are less well understood than those underpinning SICI, though it is thought to have a glutamatergic component [20,21]. Nevertheless, SICI and ICF provide important and unique information on motor cortical circuitry in addition to corticospinal excitability derived from single-pulse measures. 1.2.2.2 Testing indirect sensorimotor integration  Sensorimotor integration has classically been tested by pairing single-pulse TMS over   10M1 with peripheral nerve stimulation in order to assess the influence of the incoming afferent feedback from the nerve stimulation on corticospinal excitability. Similar to paired-pulse TMS, by pairing the stimuli at specific ISIs, short and long-latency inhibitory networks can be assessed, as well as facilitatory networks [22]. Electroencephalography (EEG) can be used to individualise these measures such that they are based on the N20 latency derived from a somatosensory-evoked potential (SEP) [23]. The N20 generator has been localised as the posterior wall of the central sulcus, specific to Brodmann areas 3b and 1 within S1, and thus the latency is thought to reflect the time taken for afferent information to ascend from the periphery to the cortex [24]. Short-latency afferent inhibition (SAI) relies on an ISI that is 2 ms longer than an individual’s N20, allocating for transmission time to S1 as well as an additional 2 ms to relay from S1 to M1 where the TMS pulse is synchronously delivered. Afferent facilitation (AF) employs an ISI of 12 ms longer than the N20 latency; this facilitatory period extends to approximately 50 ms before another inhibitory period begins. Long-latency afferent inhibition (LAI), as used in the current thesis, is evoked at an ISI of 200 ms [25-27]. Mechanistically, SAI is the best understood of these measures of sensorimotor integration. SAI is thought to exert cortical influence through both a cholinergic [28] and GABAAergic component, specific to the α5-receptor subtype [25]. The longer ISIs employed in both AF and LAI suggest that there is potential for activation within non-primary cortices to contribute and they are likely metabotropic in nature [22,29]. While direct thalamocortical projections to M1 or a relay from Brodmann area 3a cannot be fully ruled out [30], evidence points to a relay in S1 being important for the effects seen with peripheral nerve stimulation [31]. Considering the neuroanatomy outlined above, direct stimulation of a mixed nerve activates a variety of large diameter fibres; while some of these will   11be 1a afferents that can be transmitted directly to areas 3a and 4, afferent feedback will preferentially be relayed to thalamic nuclei that project to S1 [8,32,33]. Through monosynaptic corticocortical projections from 3b to M1, information arising from nerve stimulation can reach M1 as the TMS pulse is being delivered. The corticocortical projections are thought to synapse in cortical layers 2 and 3, which are the primary layers through which interneurons are activated via TMS before synapsing on pyramidal neurons [9,34-36]. Specific evidence in support of this corticocortical effect arises from studies that have determined a linear relationship between the somatosensory cortical response and the amount of inhibition quantified with SAI pointing to the importance of the somatosensory relay, rather than a direct M1 response [31]. Additionally, studies reducing both somatosensory and motor cortical excitability have documented a change in SAI only when somatosensory excitability is lessened; this effect is not seen when motor excitability is lowered [37,38]. Therefore, it is likely that SAI is driven, at least in part, by these connections. The underlying mechanisms of AF and LAI are not understood; the initial cortical response to median nerve stimulation will be the same, however given the long ISIs, there is more time for activation of association areas such as the secondary somatosensory cortex that may also contribute to the effects [22,29]. Additionally, for all of these measures when collected using a mixed nerve, there may be contribution arising from the muscle twitch and activation of efferent fibres. However, based on the somatosensory relay in Brodmann area 3b thought to be important in these metrics, SAI, AF, and LAI will be referred to as quantifying indirect sensorimotor integration for the remainder of this thesis.	1.2.2.3 Testing direct sensorimotor integration  By altering the form of peripheral stimulation paired with TMS, different somatosensory receptors will be activated, and thus may allow for exploration of an additional pathway to M1.   12Peripheral vibration will activate cutaneous mechanoreceptors in the form of Pacinian corpuscles, but it will also activate muscle spindles, which preferentially transmit information through 1a afferent fibres and can have direct thalamocortical projections to M1, as seen in animal models, or to Brodmann area 3a of S1, which is reciprocally connected with M1 [8,32,33,39]. As was true with median nerve stimulation, somatosensory information will not be exclusively transmitted into M1 or area 3a without any information ascending to other regions of S1 (3b); however, projections will likely be more directly transmitted to M1 than is shown with median nerve stimulation. Thus, response of TMS measures to vibration will be considered direct sensorimotor integration for the duration of this thesis.   Concurrent vibration can be used in conjunction with both single and paired-pulse TMS techniques to assess the influence of incoming afferent information on multiple circuits within M1. Vibration-induced facilitation in single-pulse MEP amplitudes is driven by both spinal and cortical mechanisms; when transcranial electrical stimulation is used in place of TMS, facilitation still occurs, though the response is reduced [40]. When vibration interacts with paired-pulse TMS measures, the response is thought to occur at the level of the cortex [41,42]. Vibration has been shown to reduce SICI in numerous studies. In contrast, one study has suggested that there may be a less robust influence of vibration on increasing ICF. To our knowledge all past work exploring the impact of vibration on paired-pulse TMS measures has been done in young healthy individuals. To comprehensively understand sensorimotor integration, and how it relates to aging and neuropathologies, it is important to consider both direct and indirect pathways, as well as how both may contribute to somatosensory and motor function.      131.2.3 Aging and sensorimotor integration This section will begin by presenting evidence of motor behavioural changes that accompany healthy aging. This will be followed by an outline of age-related neurophysiological changes in somatosensory and motor systems, and will culminate in a description of what is known to date on the impact of healthy aging on sensorimotor integration, specifically.  1.2.3.1 Aging: Somatosensory and motor function Evidence for the impact of healthy aging on somatosensory function is mixed; this is largely due to methodological differences across testing modalities and location being assessed. Generally, it appears that there may be slight somatosensory decline with age in that individuals are less sensitive to touch; however, this has yet to be fully established [43]. A recent review by Goble and colleagues has described a similar decline in proprioceptive ability that accompanies healthy aging [44]. Furthermore, they highlight the impact of impaired proprioception on motor behaviour due to the important connections between the somatosensory and motor systems [45-47]. While it is believed that changes to both the PNS and CNS contribute to proprioceptive decline, Goble et al emphasise the lack of investigation into neural mechanisms that may underlie the deterioration of communication between the somatosensory and motor systems [44].  In contrast to the somatosensory system, there is a plethora of work indicating a decline in motor performance with healthy aging. These effects are widespread, affecting multiple components of the motor system such as strength, power, fatigue, speed, and coordination [48,49]. Further, they apply to both unimanual and bimanual tasks [50]. These changes in motor behaviour are relevant for activities of daily living, and also can be probed in a laboratory setting in an effort to understand potential neural underpinnings. Imaging work has shown alterations in   14movement-related networks in older individuals, as compared to young individuals, such that there may be a shift towards bilateral control of movement, and increased prefrontal activation, suggesting an increased cognitive demand [51]. Expanding on these broad, network wide changes the following section will explore neurophysiological changes in specific somatosensory and motor systems that are especially pertinent to this thesis. 1.2.3.2 Aging: Somatosensory and motor neurophysiology  The majority of work exploring age-related changes in sensorimotor cortical neurophysiology focuses on somatosensory and motor systems separately, rather than the impact of afferent feedback on motor cortical excitability.   Early components of SEPs in older individuals show an increase in amplitude (N20), indicative of the arrival of afferent feedback in S1. The hypothesised explanation for this is a reduction in cortical inhibition that is widespread across cortical regions in aging populations [52]. For example, there is a tonic inhibition exerted from the prefrontal cortex to S1 via the thalamus; alterations in prefrontal excitability, structure, or functioning, could then influence this inhibition and lead to increases in SEP amplitude [53,54]. Therefore, somatosensory changes that accompany healthy aging have been documented; however, they are likely the result of many different contributors given the neural networks between cortical regions.    The influence of healthy aging on measures of motor cortical circuitry has been comprehensively studied; age-related changes appear to influence specific neural circuits within M1 while others remain intact. By exploring the wide-range of measures that TMS provides, we can begin to elucidate which networks may be of primary interest. First, an increase in RMT occurs with age; this hypoexcitability may be caused by both spinal and cortical mechanisms. Likely it is driven, in part, by both factors, as there is evidence for a reduction in spinal   15excitability accompanying aging, in addition to evidence for a reduction in the synchronous recruitment of I-waves that contribute to TMS thresholds [1]. Response to single-pulse TMS delivered at higher intensities (120 and 140% of RMT) is reduced in older healthy individuals providing further evidence for this argument [1].   Studies of the effects of aging on SICI are highly variable. Overall, a meta-analysis showed no influence of aging on SICI; however, variability across methods and results is highlighted and it is suggested that this may mask a reduction in SICI indicative of age-related changes in GABAAergic transmission within the motor cortex [1]. Results for a recent large-scale study not included in the meta-analysis did show a significant reduction in SICI in older healthy individuals, compared to younger individuals [55]. Together, this indicates that when protocols are standardised and comparisons can be made across large groups, there may be aging-related changes to GABAA receptor mediated interneuronal circuits within M1. In contrast, no robust changes to interneuronal circuits responsible for ICF have been consistently documented [1]. Therefore, while motor cortical neurophysiology does appear susceptible to age-related decline, these changes are specific to independent circuits, rather than observed globally across all measures of M1 excitability.  1.2.3.3 Aging: Sensorimotor integration neurophysiology  Imaging work explored the sensorimotor cortical response to somatosensory input, as well as exploring functional connectivity within sensorimotor regions in elderly populations. More specifically, response to vibration within the sensorimotor cortex, quantified with functional magnetic resonance imaging (fMRI) techniques, is similar in older and younger healthy adults. Functional connectivity investigations, explored with different methodologies such as resting state fMRI, alone or in combination with diffusion tensor imaging, and functional   16density maps, are discrepant and fail to arrive at a solid conclusion on the impact of aging on sensorimotor integration [56,57]. Structurally, a reduction in gray matter volume within the sensorimotor cortices of older adults has been documented [58]. Together, while these results are inconclusive, in addition to the high degree of functional relevance, they provide enough evidence to warrant further exploration into changes in sensorimotor integration with aging.   Using the techniques outlined above, there has been preliminary investigation into potential influences of aging on these measures, with primary focus on indirect sensorimotor integration. SAI is the most well studied technique in the aging literature, and a recent meta-analysis concludes that there is a disinhibition associated with aging [1]. Importantly, this effect has been shown to have functional relevance such that disinhibition predicted performance on motor and cognitive tasks. This is in line with the proposed underlying cholinergic and GABAergic contributions to SAI, as they are both relevant in cognitive and motor domains [59]. Contradictory findings as to the effect of healthy aging on AF persist to date, with some work showing no age-related differences [29,59], and a single study showing an age-related reduction in AF [60]. In total, there are only three studies that have examined this measure of indirect sensorimotor integration in a population of older healthy individuals, and there are methodological differences that may account for some discrepancy. A single investigation that included LAI in their assessment of aging and indirect sensorimotor integration did not find any difference between age groups [59]. Together, these findings may indicate that higher order oligosynaptic networks are less susceptible to age-related change; however, as investigation into these measures is preliminary, future investigation with consistent methods is warranted.   To our knowledge, there has yet to be an investigation into the influence of vibration on single or paired-pulse techniques to probe direct sensorimotor integration in a population   17of older healthy adults. To date, work exploring vibration and paired pulse TMS has focused on individuals under the age of 50. Given the results from a recent study [55] suggesting that neurophysiological differences are not apparent until large age groups centered around the age of 50 (ie. < 50 compared to > 50) are explored, there is a gap in our understanding of the age-related changes that occur in this distinct pathway of direct sensorimotor integration.  1.2.4 Chronic stroke and sensorimotor integration Behavioural and neurophysiological changes persist into the chronic phase of stroke recovery, with the majority of individuals presenting with some degree of somatosensory and/or motor impairment, as well as changes in cortical excitability quantified with TMS. Additionally, both motor and somatosensory evoked potentials have been shown to have prognostic value [61,62]; the presence of either evoked potential is linked with greater functional recovery post-stroke.  1.2.4.1 Chronic stroke: Somatosensory and motor function  Motor impairments persist into the chronic phase of stroke recovery in 55-75% of individuals [63]. Further, somatosensory impairments are thought to occur in as many as 89% of individuals [64]. From such behavioural evidence, it is difficult to disentangle the influence of one system on the other, or parse out the contribution of sensorimotor integration to functional deficits. Understanding underlying neural mechanisms contributing to these deficits, beyond neuronal death within the infarct and peri-infarct regions, is essential to developing potential techniques to improve such behaviour.  1.2.4.2 Chronic stroke: Somatosensory and motor neurophysiology  The unique lesion characteristics of each stroke and the individualised nature of stroke pathology should be considered when interpreting group-based studies. There are advantages and   18disadvantages to both heterogeneous and homogeneous samples, as heterogeneity may provide increased generalisability and homogeneity providing insight into the influence of specific lesion location on neurophysiological parameters. The current discussion will focus on group averages found with heterogeneous samples, unless otherwise noted.   Work investigating the influence of chronic stroke on SEPs showed a general reduction in the early components [65,66]. Early SEP components (N20, P26) reflect the initial arrival of afferent feedback arising from peripheral nerve stimulation in S1; as such, reduction in these components suggests a potential reduction in the amount of somatosensory feedback reaching the cortical level. This reduction has been linked to both somatosensory and motor function; smaller SEP amplitudes relate to worse somatosensation and motor impairment [65]. These results demonstrate the strong links between somatosensory and motor systems. Hypothetically, reduced afferent feedback in S1 would translate to a decrease in sensorimotor integration that may then be reflected in motor deficits.   Cortical damage arising from strokes involving the prefrontal cortex leads to unique patterns of somatosensory processing given the strong anatomical connections between the prefrontal cortex, thalamus, and S1. Through a prefrontal-thalamo-cortical network, somatosensory gating can take place to allow for appropriate weighting of incoming somatosensory information based on external factors such as task-relevancy [54,67]. Additionally, prefrontal lesions lead to an increased SEP response, indicating a potential release of tonic inhibition exerted from the prefrontal to somatosensory cortex. Although this augmentation of SEP amplitude is not shown for the very early components (N20, P26), it is evident starting at approximately 28 ms [53]. As a result, when considering somatosensory   19processing and in turn sensorimotor integration, the ability of prefrontal cortices and attention to modulate SEP amplitudes must be considered.  Post-stroke assessment of primary motor cortical circuits has been extensively documented using TMS-based techniques. Comparisons are often made between the ipsilesional and contralesional hemispheres, as well as between the ipsilesional hemisphere and healthy controls, as stroke may also lead to contralesional changes that could influence hemispheric differences. A recent meta-analysis provides the largest exploration of post-stroke motor neurophysiology to date, and informs both aforementioned comparisons, as well as differentiating between stages of recovery, defining the early phase of recovery as within three months of the infarct, and the late phase being greater than 6 months post-infarct [68].  Overall, the general conclusions arising from this analysis were that, while there are no differences between the contralesional hemisphere and healthy controls, there is a reduction in ipsilesional M1 excitability quantified with RMT and MEP amplitude, as compared to both the contralesional hemisphere and healthy controls. This difference was pronounced for individuals in both early and late stages of recovery. In contrast, SICI was initially reduced post-stroke; however, these values were age-normative in the chronic stage [68]. This pattern follows the theory that initially after infarct, there is a reduction in inhibition to promote a neuroplastic environment and optimise recovery. This environment appears time-sensitive and reduces as time proceeds [69].  Specifically considering the chronic phase of stroke recovery reveals increases in both RMT and AMT, decreased MEP amplitudes, increased MEP latency, decreased map area, and a shorter silent period duration [68]. Cumulatively, these results suggest that there are widespread changes to various interneuronal circuits and populations that contribute to motor deficits   20following stroke. Further, such work validates the use of TMS as a methodological tool to assess neural circuitry post-stroke in that it is sensitive to detect change within these networks. Importantly, many TMS measures have been related to chronic motor functional deficits or impairment levels; increased motor thresholds, reduced MEP amplitudes, reduced motor map sizes, and alterations to the balance between SICI and ICF have all been reported as relating to poorer motor outcomes [68].  As presented above, both somatosensory and motor cortical neurophysiological measures have been shown to change in individuals with chronic stroke. Further, these changes have behavioural relevance in that they relate to or can be used to predict motor impairment. Despite this, there has been very little investigation into cortical sensorimotor integration in individuals with chronic stroke. Given the above results in combination with the known importance of sensorimotor integration for successful motor behaviour, neurophysiological investigation into sensorimotor integration in individuals in the chronic phase of stroke recovery is required. 1.2.4.3 Chronic stroke: Sensorimotor integration neurophysiology  To our knowledge, there has been a single study investigating indirect sensorimotor integration in acute stroke [70] as well as one study investigating indirect sensorimotor integration in chronic stroke [71]. Additionally, these investigations have been limited to SAI, without including longer-latency measures of AF and LAI, which are known to provide information that is unique to that probed with SAI. Acutely, in the ipsilesional cortex of individuals with acute stroke, there is a reduction in inhibition, quantified with SAI. This disinhibition is clinically relevant; the amount of disinhibition in ipsilesional SAI correlates with better scores on the Modified Rankin Scale [70]. Initial work in chronic stroke suggests that SAI   21may remain reduced post-stroke. To our knowledge, the only investigation that explored indirect sensorimotor integration in chronic stroke showed a reduction in ipsilesional SAI compared to the contralesional hemisphere. The sample used in this investigation was focused on individuals with subcortical lesions; there has yet to be exploration into the effects of cortical lesions on SAI in the chronic phase of recovery [71]. Additionally, whether the relationship between neurophysiology and motor function persists is unknown. Future work, therefore, should expand to include different lesion locations, different measures of indirect sensorimotor integration (AF, LAI), and the relationship between these measures of motor impairment and function.   In addition, it is important to understand how chronic stroke may affect other methods by which sensorimotor integration may take place. Using vibration and TMS to probe direct sensorimotor integration, outlined above, will provide unique neurophysiological information on sensorimotor integration post-stroke. To date, there has not been any investigation into how direct sensorimotor integration might be influenced by chronic stroke. Acutely, preliminary work indicates that vibration increases single-pulse MEP amplitudes beyond the level of facilitation seen in healthy controls, though this is reduced to age-normative values in the sub-acute stage [72]. To determine whether this trajectory continues towards a reduced impact of afferent feedback on the motor cortex in chronic stroke, and to understand how vibratory afferent feedback directly influences inhibitory and facilitatory interneuronal circuits within the motor cortex, further investigation is required.   The lack of investigation into both direct and indirect sensorimotor integration in individuals with chronic stroke led to the development of this dissertation. As sensorimotor integration has been shown to be important in motor function in earlier stages of recovery, as well as in healthy older individuals, it is important to understand the neurophysiological changes   22that persist into chronic stages of recovery and how this may influence motor behaviour. Identification of potential deficits in sensorimotor integration is essential to further our understanding of neurobiological change associated with chronic stroke, and is a critical first step to determining specific circuits that may be targeted with an intervention. Given that motor deficits are prevalent in this population and treatments have largely been unsuccessful, we must expand our understanding of the factors that contribute to post-stroke disability and use this knowledge to develop intervention strategies. 1.3 Neuroplasticity  Neuroplasticity is often cited as the ability of the brain to change in response to experience. As such, it is thought to underlie the learning of new skills and play a large part in motor recovery post-stroke [73,74]. Plasticity in sensorimotor networks has been shown in response to both single and multi-session interventions, with interventions ranging from behavioural practice to peripheral and cortical stimulation paradigms. Often, an intervention designed to optimise the cortical environment for neuroplasticity will be used as a primer to be followed by behavioural practice. This section will begin by outlining neurophysiological principles of neuroplasticity. This will be followed with an explanation of the influence of aging and chronic stroke on neuroplasticity, with a specific focus on somatosensory and motor cortices. Finally, the section will conclude with an exploration into intervention-induced changes in sensorimotor integration. As this thesis will focus on changes in measures of neurophysiology resulting from a single session, that will be of primary interest in this introduction. 1.3.1 Mechanisms of neuroplasticity Neuroplastic change may be driven by a combination of underlying neural mechanisms, with contributions from the unmasking of latent connections, activation of silent synapses, or   23strengthening of existing connections [75]. An unmasking of latent connections is related to the underlying neuroanatomy in the sensorimotor cortex. Specifically, in M1, a multitude of inhibitory connections exist between effector locations, allowing for efficient functioning. When a specific representation initiates movement via excitatory projections to the spinal cord, M1 neurons also activate GABAergic inhibitory interneurons, which inhibit surrounding representations from exciting pyramidal cells to neighbouring effectors [76]. Similar interconnections between locations within S1 exist, as has been illustrated with a reduction in surround inhibition following amputation or nerve transection [77]. Therefore, following interventions such as repetitive stimulation or behavioural activation of specific cortical motor neurons, with a release of GABAergic inhibition, the inhibitory/excitatory balance may be altered, leading to changes in motor map representations and underlying behavioural improvement [78].  Experience-dependent plasticity, specifically, is linked to the repetitive synchronous or simultaneous depolarization of specific synaptic structures over time, which strengthens connections and leads to post-synaptic excitability changes [79]. This can be induced experimentally, or result from repetitive behavioural practice. Long-term potentiation (LTP) is an important cellular mechanism underlying experience-dependent plasticity. LTP induces an increase in the strength of connection between two neurons that are repeatedly activated together through the development of temporally similar depolarization patterns based on the most salient input received during a given behaviour [79,80]. Neuronal depolarization triggers a cascade of cellular events leading to the enhancement of presynaptic neurotransmitter release and postsynaptic neurotransmitter uptake. Once an excitatory neural connection is activated, pre-synaptic glutamate release occurs. Post-synaptically, glutamate binds with α-amino-3-hydroxy-5-  24methy-4-isoxazolepropionic acid (AMPA) and N-methyl-D-aspartate (NMDA) receptors triggering sodium to enter the cell via AMPA receptors. Depolarization then causes NMDA receptor ion channels to open and calcium to rush into the cell; over time, repeated synchronous activation will in turn induce both an increase in the density of post-synaptic AMPA receptors and an enhanced neurotransmitter release from the pre-synaptic cell, both of which strengthen the specific synaptic connection. This process has also been linked to cortical GABAergic inhibition such that a reduction in experience-dependent plasticity is seen with the administration of a GABA agonist [81]. The changes in synaptic efficiency outlast the specific period of stimulation [7]. The combination of interventions inducing a release of GABAergic inhibition and repetitive stimulation or behaviour is therefore thought to be critical in optimising performance improvements [82,83]. 1.3.2 Motor and somatosensory cortical plasticity  Early work indicating that the somatosensory and motor cortices were able to rapidly change in response to experience arose from experiments on clinical populations such as amputees or individuals with nerve transections. In addition, animal work showed that both repetitive cortical stimulation and behavioural interventions could induce sensorimotor plasticity. Using techniques previously discussed, such as EEG and TMS, the influence of interventions on cortical circuitry can be examined. The sections below will outline the evidence for neuroplasticity within the healthy human somatosensory and motor cortices. Single and multiple sessions of motor practice on a piano sequence task induces changes in the hand representation of M1 in a group of neurologically intact individuals [84]. Moreover, TMS-derived motor maps shown an extended motor map representation for the extensor carpi radialis (ECR) representation in M1 following short-term single day training on a bimanual, in-  25phase motor task executed using the wrist flexors and extensors [85]. Similar work has shown that practice of a motor task induces changes in S1 as well, likely resulting from the repetitive somatosensory feedback arising from task practice [86]. The immediate map expansion following training may be indicative of an unmasking of latent connections, or the early phases of LTP; however, this is impossible to confirm without examining cellular changes.  Using a paradigm known as Paired Associative Stimulation (PAS) that repeatedly delivers stimulation over the median nerve at the wrist followed by a TMS pulse over M1 in quick succession, LTP-like plasticity can be more directly probed. Theoretically, PAS exploits LTP-like plasticity by repeatedly activating the same trans-synaptic structures to lead to long-lasting change beyond the period of stimulation. When quantifying motor cortical excitability following PAS, MEP amplitudes arising from single-pulse TMS are enhanced, suggesting an increase in corticospinal activity for a period of approximately 45 minutes [87]. Similarly, repetitive peripheral somatosensory stimulation or repetitive motor cortical stimulation can induce local excitability changes that outlast the period of stimulation [88]. Therefore, in healthy adult populations, experimental paradigms involving both motor practice and external stimulation can induce neuroplastic change in somatosensory and motor cortices. 1.3.3 Attention and neuroplasticity Attention has been shown to have a direct impact on sensorimotor plasticity. Using the PAS technique described above, an initial study showed that attention directed towards the median nerve stimulation was required to induce changes in motor cortical excitability [89]. This finding was expanded on in a follow-up study where the attentional demands were manipulated in three main conditions: (1) a condition in which attention was directed away from the hand and individuals performed arithmetic problems, (2) an attentional condition in which participants   26fixated on a cross on the computer screen but counted electrical stimulation delivered to the ipsilateral hand, and (3) a condition in which attention was again directed to the stimulation on the ipsilateral hand with visual attention also directed at the hand. The results provide support for an attentional continuum with increasing attention positively correlating with increased PAS effects on motor cortical excitability, thus allowing us to infer attentional implications to neuroplasticity induction in M1 [89]. Mechanistically, the authors postulate an underlying acetylcholine (ACh) based network contributing to this effect; ACh has been shown to be modulated by attention [90] and deficits in basal forebrain cholinergic function impair motor skill learning and associated neuroplasticity [91]. It is possible, therefore, that attention regulates the level of ACh, with high levels being associated with attentionally demanding conditions and lessened with less attentional demand, and this affects the ability to acquire a new skill or induce LTP-like plasticity, as was shown in the current study.  In addition, attention may also be an important component of inducing neuroplastic change through behavioural practice. Movement repetition is not sufficient to induce neuroplasticity in M1; however, similar movements for the purpose of learning a skill do induce such changes [84]. Attention may be a mediator of this effect; the learned skill would require more attentional demand than the simple repetitious task. In the sensory domain, passive stimulation of a digit does not prompt plastic change in S1 despite the same stimulation inducing cortical changes when attention had to be maintained in order to ensure contact between the digit and a rotating disk. In addition, if an auditory discrimination task demanded the participant’s attention during sensory stimulation, as compared to a sensory discrimination task, less plasticity was measured from S1 [92].   27Sensory gating describes a phenomenon in which decreased afferent information reaches S1 when it is not relevant to a task; sensory gating can occur at rest, in various modalities, or during movement [93]. Sensory gating is thought to involve a relationship between the prefrontal cortex and thalamic reticular nucleus in which they work in concert as a “gatekeeper” to ensure primarily relevant information is extracted from concurrent sensory feedback and reaches the cortex [53,54,67,94]. In the situations described above, it is possible that during purely repetitious movement or passive sensory stimulation, the attentional demand on the task is low, and the incoming afferents are not important for the task at hand, resulting in diminished cortical responses and therefore less plastic change. Conversely, task-relevant information may also be facilitated, with the same ultimate effect as inhibition of irrelevant information. This theory may underpin attentional effects that have been shown with the sensory training paradigm used in the current dissertation. Briefly, sensory training involves 15 minutes of cyclic vibration (2 s on/2 s off) delivered to a specific peripheral muscle. In order to control attention, individuals are instructed to detect frequency changes in the vibration, which appears essential for inducing changes in sensorimotor integration that have been reported. When vibration is passively applied to the same muscle in the same pattern, without attentional instruction, there is a reduction in the response [95,96]. Understanding the role of attention in modulating plasticity is of particular importance in this thesis as healthy aging influences prefrontal function and attention.  1.3.4 Plasticity of sensorimotor integration It has been well established that both somatosensory and motor cortical excitability can change with experience; however, less work has investigated intervention-induced change in measures of direct and indirect sensorimotor integration despite the functional relevance and thus potential as therapeutic intervention for pathologies such as stroke.    28In healthy individuals, one of the original studies to examine the plasticity of direct sensorimotor integration used the sensory training paradigm discussed above. Direct sensorimotor integration was indexed prior to and following using single and paired-pulse TMS with concurrent vibration. Following sensory training, vibration-induced release of SICI increased. Functionally, in young healthy individuals (mean age 35.6 years) these changes in direct sensorimotor integration correlated with motor performance on a thumb abduction task [97]. This complemented work using neuromuscular electrical stimulation to induce changes in indirect sensorimotor integration as indexed by SAI and AF in suggesting that neurophysiological measures of sensorimotor integration are modifiable with single-session intervention [98]. Imaging work found changes in resting state EEG, indicative of altered communication between sensory and motor cortices, following a single session of repetitive sensory stimulation (vibrotactile or electrical) delivered over the first dorsal interosseous for 30 minutes [99]. Taken together, these studies suggest that in healthy individuals, sensorimotor integration is modifiable, and that this plasticity has relevant functional implications. As evidence is gathered to support the idea of the plasticity of sensorimotor integration in healthy individuals, similar concepts should be examined in populations with neuropathologies. Both baseline levels of sensorimotor integration and plasticity of this system have yet to be thoroughly explored in individuals with chronic stroke; however, these individuals would greatly benefit from improved sensorimotor integration.  1.3.5 Plasticity and aging  Healthy aging has been associated with a reduction in somatosensory and motor cortical plasticity induced by experimental techniques such as PAS, as well as in response to behavioural practice. The induction of plasticity has largely been quantified as change in evoked-potential   29amplitude, likely due to experimental ease and quantifiable output. Nevertheless, these results provide important information on the ability of the aging cortex to change in response to single session interventions.   A recent study had a group of older and younger healthy individuals practice a ballistic thumb abduction task 300 times and examined the changes in motor thresholds, MEP amplitudes, and SICI induced by motor practice. Younger healthy individuals showed a robust increase in MEP amplitude following practice that was not seen in older healthy individuals. Further, older individuals did not improve their performance as much as younger individuals [100]. Thus, although the response to motor practice based interventions may be lessened with age, induction of neurophysiological and behavioural change is still possible.    Across multiple studies, a recent meta-analysis showed a trend towards a reduction in LTP-like plasticity in older healthy adults indexed with PAS, though it did not reach statistical significance [1]. This is likely due to inherent variability in response seen across age groups; further work is needed to fully evaluate this. However, this fits with animal work showing a reduction in LTP induction that could potentially be driven by a decrease in the quantity of functional synapses, reduced NMDA receptors, or decreases in calcium regulation [101-103], all of which are critical to the neurobiological process of LTP. Importantly, similar to that mentioned above with behavioural intervention, although there may be a reduction in degree of neuroplastic change occurring in the aging sensorimotor cortices, neuroplastic change does still take place. While animal work points to a reduction in axonal sprouting and changes in synaptic efficiency that may underlie these changes in capacity for experience dependent plasticity, the majority of literature emphasises that although the process may be slower, neuroplastic change does take place and underlie behavioural improvements in healthy older individuals.   30 To our knowledge, there has yet to be investigation into the ability of a targeted intervention to change sensorimotor integration in healthy older adults. Given preliminary evidence indicating potential age-related decline in these systems, such an investigation is warranted. 1.3.6 Plasticity and chronic stroke Despite the damage resulting from stroke, experience-dependent plasticity remains a mechanism underlying functional rehabilitation and motor learning. Similar to work in healthy populations, ipsilesional M1 excitability has been shown to change in response to behavioural interventions and stimulation-based interventions. A single intensive physiotherapy session taking place between 4 and 8 weeks post-stroke resulted in immediate expansion of the ipsilesional abductor pollicis brevis (APB) muscle representation. The changes in representational size correlated with function, as measured by the nine-hole peg test (NHPT) [104]. Due to the short time frame in which these representational changes occurred, they likely represent functional, rather than structural synaptic mechanisms, yet the ability to change cortical excitability with single session behavioural interventions is a promising baseline for functional reorganization and relearning post-stroke. When using PAS as a probe to index ability of the motor cortex to undergo LTP-like plasticity, individuals with chronic stroke show a reduction in the degree of neurophysiological change, as compared to age-matched controls, or the contralesional hemisphere; however, the intervention does still induce excitability changes [105]. Animal work supports these results in that it has shown that the remaining neural tissue can, in response to experience, increase synaptic density and efficiency, both of which are essential to LTP [106]. The amount of neuroplastic change occurring in response to single-session interventions then appears to be on a continuum influenced by both healthy aging and pathology.   31The reduction in plasticity occurring from interventions in individuals with chronic stroke has led to investigations into the possibility of combining techniques to create an optimal environment for neuroplastic change. Cortical and peripheral stimulation are able to induce short-term changes in ipsilesional motor and somatosensory cortical circuitry by releasing GABAergic inhibition or increasing excitability [107-109]. By pairing this type of intervention with behavioural practice these changes can be exploited, and thus potentially increase the amount of plasticity occurring from repetitious movement patterns, as compared to practice or stimulation alone. This hypothesis has proven true in individuals with chronic stroke when stimulation is delivered over M1 or S1 and paired with a motor sequence task [108,110]. Preliminary work suggests that in addition to the somatosensory and motor cortices themselves, indirect sensorimotor integration in chronic stroke may also be modifiable. An intervention designed to increase ipsilesional motor cortex excitability caused the inhibition typically associated with SAI protocols to shift to be closer to values seen in healthy controls. These data suggest that the motor cortex, with increased excitability, was more receptive of the incoming afferent information, and thus the afferent information produced more typical responses in the motor cortex [71]. This preliminary work suggests that sensorimotor integration may be modifiable in individuals with chronic stroke.  Despite many unique approaches to neurorehabilitation, approximately 60% of individuals with chronic stroke continue to live with some degree of motor impairment [68]. As a result, investigation into new techniques to optimise recovery persists. The intervention employed in the current thesis will specifically target sensorimotor integration. In young, healthy individuals, it has been shown to induce a release of afferent-induced GABAergic inhibition [97]. Considering the importance of a GABAergic release in the induction of neuroplasticity, if a   32similar change can be invoked in individuals with chronic stroke, pairing such an intervention with motor practice may prove beneficial in improving motor performance, in the short term, and motor function in the long term.   1.4 Reliability of neurophysiological measures  As outlined above, neurophysiological measures such as TMS and EEG can further our understanding of brain function in pathological populations and assess the efficacy of specific interventions. Therefore, understanding inherent reliability within these measures is essential to strengthen confidence that alterations observed are reflective of true differences before and after an intervention. Two main components can be considered when examining reliability: within-individual variability and between-individual variability. Within-individual variability, similar to test-retest reliability, refers to the stability of a measure that can be expected when tested repeatedly in the same individual [111-114]. When testing individuals at different time-points to assess the influence of an intervention, quantification of within-individual variation informs the degree of change to be expected regardless of an intervention, and thus an idea of the amount of change that must be overcome in order to be confident that is intervention-specific. Between-individual variation is of high importance when examining the influence of a specific pathology on neurophysiology; if there is known variability between healthy individuals within a given measure, this amount must be overcome in order to confirm that effects are actually related to the pathology of interest [111-114].  Studies of reliability centre on the fact that any measured score is reflective of a combination of the true score and measurement error. Reducing measurement error is paramount to increasing reliability and improving statistical power. There are many statistical approaches to assessing reliability; the work in this thesis employs intraclass correlations (ICCs), which   33represent the proportion of the variance in the true scores that can be captured when measured repeatedly by taking into account both within and between-individual variation. Broadly speaking, the ICC is a ratio of between-individual variation and the total amount of variation within a dataset, such that an ICC of 1 would be derived from measures that had no within-individual variation [112]. Past work has not reached a consensus as to the variability in neurophysiological TMS measures, though such discrepancies may arise from limitations such as small sample sizes and differences in methodologies that constrain the interpretation [115]. One study has previously looked at the reliability of a subset of TMS measures in older healthy individuals and individuals with stroke using various methodologies. They conclude that the included TMS measures can reliably detect between-individual change, though are less well-suited for detecting change within an individual, though they suggest that this limitation can be overcome with moderate sample sizes. Further, they indicate that further work is needed to fully understand reliability of TMS measures [116]. Specifically, expansion of sample size, age ranges, and measures included is required. For example, measures of sensorimotor integration, which are critical to the first three research chapters in this thesis, have yet to be assessed in a reliability study.  1.5 Thesis overview The overall objective of this thesis is to comprehensively examine the neurophysiology of sensorimotor integration. The research chapters will address the influence of healthy aging and chronic stroke on sensorimotor integration, the modifiability of sensorimotor circuits with a specific intervention, and the reliability of commonly used metrics to probe sensorimotor integration. Chapter 2, the first original research chapter of the dissertation, considers age and stroke-related differences in indirect sensorimotor integration and the relationship of these   34differences to somatosensory and motor function. Chapter 3 examines the impact of aging on direct sensorimotor integration at baseline, and the modifiability of these measures with a sensory training task. Chapter 4 provides a similar evaluation as was completed in Chapter 3, but will focus on stroke-related changes to direct sensorimotor integration and its modifiability. The original research chapters will conclude in Chapter 5 with an investigation into the reliability of the metrics that are commonly used to probe somatosensory and motor cortical excitability, as well as indirect sensorimotor integration. Chapter 6 presents a general discussion of the main findings of the thesis, as well as limitations and future directions. Final conclusions will also be presented in Chapter 6. 1.5.1 Thesis impact Motor behaviour is an integral component of quality of life, allowing for independent living and successful community engagement. Motor impairment is well documented in pathological populations such as stroke; however, even healthy aging is associated with motor decline. Sensorimotor integration is an essential physiological process that underlies all movement. Despite this behavioural relevance, there has yet to be a thorough investigation into the neurophysiological change that occurs within sensorimotor integration in older healthy individuals and individuals with chronic stroke, and how this change relates to somatosensory and motor function. Resultantly, this thesis was designed to provide basic neurophysiological knowledge on the impact of aging and stroke on sensorimotor integration, and to explore whether sensorimotor integration will respond to intervention, and thus provide a potential therapeutic target.   This work will provide important insight into neurobiological processes that accompany aging and stroke. This is highly important given the growing aging population that will increase   35the number of community dwelling seniors, as well as increasing the incidence of stroke. As motor deficits persist into the chronic phase of stroke recovery in a large percentage of individuals, understanding underlying neurophysiological processes that may be contributing to impairment is essential. In addition, targeting sensorimotor integration may provide a useful method to induce neuroplastic change in neurophysiological circuits underlying behavioural deficits. The work encompassed in this thesis is an important initial step to furthering this understanding that may, in the future, provide useful in designing rehabilitative interventions. 1.5.2 Research objectives Chapter 2 Aim: To establish how the neurophysiological profile of indirect sensorimotor integration changes in both a sample of healthy older adults and individuals post-stroke.  Hypotheses: Indirect sensorimotor integration will be modified such that afferent feedback will have less of an impact on the motor cortex in healthy older individuals, as compared to healthy younger individuals, and this will be further reduced in individuals with chronic stroke. Chapter 3 Aim: To understand the influence of age on direct sensorimotor integration at baseline and in response to a sensory training task.  Hypotheses: Direct sensorimotor integration will be reduced in healthy older individuals such that the motor cortex will be less affected by incoming afferent information, and there will be a reduced response to sensory training, as compared to younger healthy individuals. Chapter 4 Aim: To determine baseline patterns of direct sensorimotor integration in chronic stroke and examine the effect of sensory training on direct and indirect sensorimotor integration.  Hypotheses: Direct sensorimotor integration will be reduced in individuals with chronic stroke compared to older healthy controls; sensory training will induce less change in direct sensorimotor integration than seen in older healthy individuals.    36Chapter 5 Aim: To examine the feasibility, reliability, and susceptibility to group differences of commonly used neurophysiological measures quantifying sensorimotor excitability. Hypotheses: All measures will be well tolerated, leading to low attrition rates; measures of threshold and latency will be the most reliable and least susceptible to methodological differences between study sites.    37Chapter 2: Exploring the influence of age and chronic stroke on the neurophysiology of indirect sensorimotor integration  2.1  Introduction The ability to accurately incorporate somatosensory feedback from the periphery into the motor cortex to inform movement is essential for successful interaction with the environment. Sensorimotor integration is the neurophysiological process within the central nervous system by which somatosensory information informs motor output. Therefore, while the connection between the somatosensory and motor system is the core component of sensorimotor integration, it is also dependent on the integrity of both the somatosensory and motor systems. Biological processes such as healthy aging, or pathological processes such as stroke, are accompanied by changes to these somatosensory [117-119] and motor systems [1,62], which likely extend to sensorimotor integration [70].   Sensorimotor integration can be tested neurophysiologically by using a combination of techniques; as mentioned in the introduction of this thesis, to probe indirect sensorimotor integration, median nerve stimulation and transcranial magnetic stimulation (TMS) can be paired at various interstimulus intervals (ISIs) [25,29,120]. Median nerve stimulation in concert with electroencephalography (EEG) is used to index cortical excitability of the primary somatosensory cortex (S1) and motor evoked potentials (MEPs) elicited with TMS probe the excitability of the motor system. These methods, in addition to the indirect sensorimotor integration measures termed short-latency afferent inhibition (SAI), afferent facilitation (AF), and long-latency afferent inhibition (LAI), can be explored to disentangle the relationship between somatosensation, motor output, and sensorimotor integration. Collectively, each   38measure is thought to probe a different neuronal circuit, thus providing unique information on different components of sensorimotor integration. Due to the short ISI, SAI is thought to reflect direct projections from S1 to M1, and the inhibitory effect has been linked to cholinergic and GABAergic systems [25]. Though the mechanisms underlying AF and LAI are less well understood, they provide a measure of the facilitatory impact of afferent stimulation, and longer-latency inhibitory networks involving higher order cortical regions, respectively [22,29]. These measures have been comprehensively studied in young, healthy individuals and the patterns of inhibition and facilitation are well documented [25,29,120].  Various populations, such as older healthy individuals and individuals with chronic stroke often have behavioural symptoms that may be indicative of impairments in sensorimotor integration; however, it is challenging to disentangle the unique contributions of the somatosensory system, motor system, and sensorimotor integration to these impairments from behavioural evidence alone. Although there are mixed initial findings, a recent meta-analysis has suggested that neurophysiologically, healthy aging is accompanied with disinhibition quantified via SAI [1]. Limited and contradictory evidence exists for the patterns of AF and LAI that are associated with aging with preliminary work showing no age-related differences [59], or decreases in both AF and LAI [60]. Importantly, age-related changes that have been documented in SAI have related to performance on both motor and cognitive tasks highlighting the relevance of sensorimotor integration to motor behaviour [59]. The relationship between sensorimotor integration and somatosensory function, however, has yet to be investigated. Furthering our understanding of these relationships will provide additional information to a growing profile of neurophysiological changes associated with healthy aging and pathologies such as stroke.   39Despite the behavioural relevance of sensorimotor integration and its importance for motor recovery post-stroke, neurophysiological changes of sensorimotor integration in individuals with stroke has only been documented in a single study examining changes in the acute phase of recovery. Specifically, the authors examined stroke-induced changes in ipsilesional SAI and related this to functional outcome at 6 months. In this investigation, individuals in the acute phase of stroke recovery presented with disinhibition indexed with SAI in the ipsilesional hemisphere, which correlated with functional outcome as measured by the Modified Rankin Scale. The authors concluded that acute disinhibition was associated with better motor recovery [70]. This work suggests that sensorimotor integration is not only impaired post-stroke, but that it is relevant to functional outcome. To date, the neurophysiology of sensorimotor integration in the chronic phase of stroke recovery has not been evaluated. Understanding these measures of sensorimotor integration may provide insight into the trajectory of neurophysiological changes across stages of stroke recovery.  The current study aims to: 1) provide a comprehensive evaluation of sensorimotor integration, indexed by SAI, AF, and LAI in individuals with chronic stroke, as well as healthy older controls, 2) determine if neurophysiological measures of sensorimotor integration relate to somatosensory or motor function, and 3) evaluate whether there is a relationship between somatosensory evoked potentials (SEPs) and MEPs with measures of sensorimotor integration (SAI, AF, and LAI). We hypothesise that: 1) both aging and chronic stroke (to a greater extent) will lead to abnormalities in sensorimotor integration such that the impact of incoming afferent information will have less influence on motor output in older as compared to younger individuals, and in individuals with chronic stroke as compared to older healthy individuals (i.e. less inhibition or facilitation), 2) measures of sensorimotor integration will be associated with   40motor function and impairment post-stroke, and 3) the relationship between the somatosensory and motor systems with sensorimotor integration will change with age and after stroke.  2.2 Methods 2.2.1 Participants Thirty-seven individuals participated in the study, including twelve younger healthy individuals between the ages of 20 and 35 (27.7 ± 4.7 years, 3 M/9 F), twelve older healthy individuals between the ages of 50 and 80 (69.2 ± 10.0 years, 7 M/5 F), and thirteen individuals in the chronic phase of stroke recovery (71.5 ± 8.9, 7 M/6 F). Individuals in the chronic phase of stroke recovery (> 6 months) who experienced stroke-related sensorimotor deficits in the paretic limb were recruited for the study. Inclusion criteria was focused on sensorimotor impairment, rather than lesion location; however, individuals with prefrontal lesions were excluded as such lesions have been shown to uniquely influence SEP amplitudes. Informed consent was obtained from all participants prior to completion of the experimental protocol. Individuals were screened for contraindications to TMS using standard screening forms. The Clinical Research Ethics Board and the University of British Columbia approved all experimental procedures. 2.2.2 Experimental design Each individual participated in a single experimental session comprised of both behavioural and neurophysiological assessment. An overview of the experimental session can be found in Figure 2.1.      41  Figure 2.1: Experimental procedures. Behavioural assessment consisted of measures of motor function and impairment in individuals with chronic stroke, and somatosensory function in all individuals. Neurophysiological assessment consisted of SEPs to quantify somatosensory excitability, recruitment curves to quantify motor excitability, and measures of indirect sensorimotor integration (SAI, AF, LAI). Number of stimuli, interstimulus intervals, and intensities are noted in parentheses. FM: Fugl-Meyer, WMFT: Wolf Motor Function Test, SEPs: somatosensory evoked potentials, SAI: short-latency afferent inhibition, AF: afferent facilitation, LAI: long-latency afferent inhibition.  2.2.3 Behavioural assessment  Baseline Tactile Semmes-Weinstein Monofilaments were used to assess tactile perceptual thresholds (Quick Medical, Issaquah, WA, USA). Arm-position matching was used to quantify between limb proprioception using the KINARM (BKIN Technologies Ltd, Kingston, ON, Canada). Specifically, the measures of variability and absolute error were extracted and scores were presented as z-scores based on normative data. All measures were quantified bilaterally.   Behavioural assessment of the motor system was done with Fugl-Meyer (FM) Upper Extremity Scale [121] to quantify motor impairment and the Wolf Motor Function Test (WMFT) to index motor function [122]. These measures were collected only on individuals with chronic stroke. 2.2.4 Neurophysiological assessment For all neurophysiological assessments, individuals were seated in an upright, comfortable position and instructed to relax as much as possible. Neurophysiological measures of somatosensory cortical excitability, motor cortical excitability, and sensorimotor integration   42were collected. All measures were collected from the non-dominant hemisphere and non-dominant hand as self-reported in healthy individuals, and the ipsilesional hemisphere and paretic hand in individuals with chronic stroke. 2.2.4.1 Somatosensory evoked potentials  SEPs were recorded following median nerve stimulation (pulse width 200 s, square wave pulse, cathode distal, anode proximal) with surface electrodes corresponding to CP4 or CP3 positioning in accordance with the International 10-20 System (contralateral somatosensory cortex) and referenced to AFz (2000 Hz sampling rate) (NeuroPrax; Neuroconn, Ilmenau, Germany). Channel impedances were < 5 kΩ. Briefly, stimulation at 2 Hz (Digitimer DS7AH, Welwyn Garden City, Hertfordshire, UK) was delivered at motor threshold, defined as the minimum intensity required to evoke a visible twitch in the target muscle. Recordings from 300 stimuli were collected. Surface EMG was recorded from the abductor pollicis brevis (APB) muscle using silver/silver-chloride disc surface electrodes (1 cm diameter) in a belly tendon montage in order to monitor the amplitude of the M-wave. The EMG signal was amplified and analogue filtered (30 Hz to 1 kHz) with a Powerlab 4/30 EMG System (AD Instruments, Colorado Springs, CO). In order to analyse SEP data an average trace was produced to extract the component amplitudes (N20, P26, N30, P50). 2.2.4.2 Transcranial magnetic stimulation TMS was performed as previously described using established techniques [123]. Single pulse TMS was delivered using a monophasic figure-of-eight shaped coil (Magstim 70 mm P/N 9790, Magstim Co., UK) connected to a Magstim 2002 stimulator (Magstim Co., UK). Stimuli were delivered with a random ISI of 4-5 seconds. The coil was held to induce a posterior-anterior current direction in the underlying cortex with the coil handle positioned at an angle of 45   43pointing backwards. The APB ‘hot-spot’ was located using neuronavigation in combination with a template MRI to guide the search and to ensure consistent coil positioning throughout the experiment (Brainsight™, Rogue Research Inc., Montreal, QC, Canada). Resting motor threshold (RMT) was determined by finding the lowest stimulation intensity required to evoke MEPs of at least 50 µV in 5 out of 10 consecutive trials [124].   Recruitment curves were collected to probe corticospinal tract excitability. A total of 100 single pulse stimulations at ten intensities, ranging in 10% increments from 80-170% RMT, were delivered. The order of intensities was randomised. MEP amplitude at each intensity was averaged across the ten trials. 2.2.4.3 Sensorimotor integration SAI, AF, and LAI use TMS, in conjunction with peripheral nerve stimulation, to examine the influence of somatosensory information on the motor cortical output; specifically, an electrical stimulation was delivered over the contralateral median nerve prior to a TMS pulse delivered over the motor cortex while the participant was at rest. The median nerve stimulation was set at an intensity just above motor threshold where a twitch was visible and an M-wave was consistently produced. TMS pulse intensity was set such that an MEP of approximately 1000 µV was consistently produced (test stimulus (TS)) [25]. Interstimulus intervals (ISIs) were individualised such that the ISI for SAI was 2 ms longer than the N20 latency derived from the SEP trace, and the ISI for AF was 12 ms longer than the N20 latency [23]. LAI utilised an ISI of 200 ms [22,25-27,125,126]. Ten pulses of each conditioned technique, as well as ten pulses of unconditioned stimuli (TS) were collected.    442.2.4.4 Statistics  To determine if there were group (healthy younger, healthy older, stroke) differences in sensorimotor integration (Aim 1) a one-way ANOVA was performed for each neurophysiological measure (SAI, AF, LAI). Effect sizes derived from standard guidelines were also calculated to inform the strength of the effects [127]. Pre-planned contrasts were employed to directly test our aging and stroke related hypotheses. Post-hoc tests using Tukey’s HSD were conducted for all additional comparisons. To explore the association between behavioural and neurophysiological measures (Aim 2) a stepwise linear regression analysis (predictors: age, RC slope, SAI, AF) was performed. Finally, the relationships between somatosensory excitability, motor excitability, and sensorimotor integration (Aim 3) were assessed with bivariate correlations.  2.3 Results  Data were checked for normality, as indexed by significance of p < 0.001 in the Shapiro-Wilks test [128]. All variables were found to be normal. As such, for correlational analyses, Pearson’s (rp) bivariate correlations were used. Additionally, there was acceptable collinearity between predictors indexed by variance inflation factors below 10 and tolerance levels above 0.1 [129]. Data were normal and homoscedastic, with independent residuals as determined by the Durbin Watson test [129]. Demographic information can be found in Table 2.1. Group averages for all dependent variables can be seen in Table 2.2.     45 Subject Group Age Sex FM WMFTR Ipsi Hemi S01 Stroke 86 M 57 33.9 L S02 Stroke 68 M 51 36.8 R S03 Stroke 58 F 31 17.7 L S04 Stroke 70 F 64 54.5 R S05 Stroke 71 M 63 59.3 R S06 Stroke 77 F 54 30.2 L S07 Stroke 55 F 64 48.5 L S08 Stroke 75 M 58 57.5 L S09 Stroke 69 F 65 62.9 R S10 Stroke 78 M 58 52.0 L S11 Stroke 64 M 47 24.4 L S12 Stroke 80 F 63 53.8 R S13 Stroke 78 M 56 40 L Average  71.5  56.2 44.0  C01 Older 76 M   R C02 Older 73 M   R C03 Older 73 F   R C04 Older 64 F   R C05 Older 82 M   R C06 Older 50 M   R C07 Older 80 M   R C08 Older 58 F   R C09 Older 75 M   R C10 Older 74 F   R C11 Older 56 F   R C12 Older 69 M   R Average  69.2     Y01 Younger 32 M   R Y02 Younger 27 F   R Y03 Younger 20 F   R Y04 Younger 25 M   R Y05 Younger 35 F   R Y06 Younger 24 F   R Y07 Younger 25 F   R Y08 Younger 35 F   R Y09 Younger 30 M   R Y10 Younger 23 F   R Y11 Younger 29 F   R Y12 Younger 27 F   R   27.7     Table 2.1: Demographic information. FM: Fugl-Meyer, WMFTR: Wolf Motor Function Test Rate.   462.3.1 Stroke and age-related changes in sensorimotor integration  A one-way ANOVA (GROUP) was conducted for three separate dependent variables (SAI, AF, LAI). Results from these suggest that there are group differences in SAI (F(2,34)=3.854, p=0.031, η2partial=0.185) and AF (F(2,34)=4.314, p=0.021, η2partial=0.202), but not in LAI (F(2,34)=2.076, p=0.141). Sample traces for each measure can be seen in Figure 2.2.  Figure 2.2: Representative traces for measures of indirect sensorimotor integration in each group. A displays young healthy data, B shows older healthy data, and C presents stroke data. From left to right, examples of responses to test stimuli, SAI, AF, and LAI are shown. SAI: short-latency afferent inhibition, AF: afferent facilitation, LAI: long-latency afferent inhibition.  A pre-planned contrast to assess the impact of aging on SAI revealed a difference between older and younger healthy individuals (p=0.049) such that older individuals presented with age-related disinhibition. Examining the impact of chronic stroke on SAI through a pre-planned contrast revealed no differences between individuals post-stroke and age-matched controls (p=0.57). Post-hoc analysis for SAI demonstrated that individuals with chronic stroke have less ipsilesional   47SAI than young, healthy individuals (p=0.031). Similarly, post-hoc analysis for AF indicated a difference between individuals with stroke and younger healthy individuals (p=0.016) such that individuals with chronic stroke have greater AF. Pre-planned contrasts revealed no significant differences between older and younger healthy individuals, or older individuals and individuals post-stroke (ps=0.10, 0.24, respectively). These results can be seen in Figure 2.3.   SEP Amplitude (µV) RC Slope SAI (% TS) AF (% TS) LAI (% TS) Stroke 7.53 (3.75) 0.025 (0.019) 87.51 (46.30) 141.33 (64.14) 73.61 (32.33) Older 6.78 (8.42) 0.032 (0.018) 77.51 (52.17) 114.78 (55.66) 80.41 (44.56) Younger 4.34 (2.73) 0.032 (0.023) 41.07 (28.72) 76.49 (43.27) 51.77 (26.62) Table 2.2: Group average data. Mean (SD) values for each dependent variable shown separately for individuals with chronic stroke, older healthy individuals, and younger healthy individuals. SEP: somatosensory evoked potential, RC: recruitment curve, SAI: short-latency afferent inhibition, AF: afferent facilitation, LAI: long-latency afferent inhibition.   48 Figure 2.3: Group means for SAI (A), AF (B), and LAI (C). Individuals with chronic stroke and healthy older individuals have less SAI than young, healthy individuals (A). Individuals with chronic stroke have an increase in AF compared to young, healthy individuals (B). There are no group differences in LAI (C). Asterisks denote statistical significance. Error bars represent standard error of the mean. SAI: short-latency afferent inhibition, AF: afferent facilitation, LAI: long-latency afferent inhibition.    492.3.2 Association between neurophysiology, impairment, and function Stepwise linear regression analyses revealed an association between motor behaviour and ipsilesional SAI in individuals with chronic stroke. The significant model for both FM (R2=0.385, F(1,11)=6.900, p=0.024, β=-0.621) and WMFT rate, defined as repetitions per minute, (R2=0.380, F(1,11)=6.755, p=0.025, β=-0.617) allotted SAI as the only variable to significantly predict variance in each behavioural measure. This association was such that increased SAI corresponded with worse levels of motor impairment and function (Figure 2.4). For clarity, this has been depicted with raw data. There was a single individual in the current study who met our inclusion criteria despite having a severe FM score. Importantly, the aforementioned behavioural relationships remain even when that sample is removed from the data. In individuals with chronic stroke, there were no associations between somatosensory function and measures of sensorimotor integration.    50 Figure 2.4: Relationship between SAI and motor impairment, indexed with the Fugl-Meyer Assessment (A) and motor function, indexed with the Wolf Motor Function Test Rate (B). Reduced levels of SAI relate to poorer motor function and higher degrees of motor impairment. SAI: short-latency afferent inhibition, FM: Fugl-Meyer, WMFT: Wolf Motor Function Test.  2.3.3 Somatosensory and motor cortical excitability  There were no significant relationships between SEP components, recruitment curve slope, and neurophysiological measures of sensorimotor integration. This was true in all groups.   512.4 Discussion    Neurophysiological differences in sensorimotor integration persist into the chronic phase of stroke recovery in well-recovered individuals, yet these may be primarily driven by age-related changes. Compared to younger healthy controls, individuals post-stroke and older healthy individuals present with disinhibition in SAI. The level of disinhibition quantified with SAI in individuals with chronic stroke related to motor function and impairment such that worse motor performance was related to increased levels of disinhibition. AF was increased in individuals with chronic stroke compared to younger healthy individuals, with no difference between older and younger healthy controls. Finally, there was no relationship between somatosensory and motor cortical excitability with measures of sensorimotor integration in individuals with chronic stroke, older healthy controls, or younger healthy controls. 2.4.1 Stroke and age-related changes in sensorimotor integration This study investigated, for the first time, the impact of chronic stroke on sensorimotor integration quantified by SAI, LAI, and AF. These measures were compared between individuals with chronic stroke, healthy older controls, and young healthy individuals in order to distinguish the influence of aging from potential stroke-induced changes in sensorimotor integration. There was a difference in SAI and AF in individuals in the chronic phase of stroke recovery in comparison to younger individuals, with no difference between these individuals and older healthy individuals, suggesting that differences at this point in recovery may be driven by aging, rather than impact of stroke related changes.  Age-related disinhibition quantified by SAI, as was shown in the current work, is well documented in the literature [1]; however, the finding that a group of well-recovered individuals in the chronic phase of stroke recovery have no additional alterations in SAI is novel. Prior work   52investigating this neurophysiology post-stroke solely focused on the acute phase of recovery and showed further disinhibition in ipsilesional SAI, relative to age-matched controls [70]. Taken together with the current result, it appears that ipsilesional SAI levels may change throughout the trajectory of recovery post-stroke. Acutely after cortical infarct, it is theorised that cortical activity may shift towards disinhibition within neural networks in an attempt to promote an increased response to experience-dependent plasticity important for motor recovery. Once these new, potentially compensatory, networks are established, there is less of a dependence on reduced inhibition, and rather the focus is thought to shift towards a strengthening within these new pathways [69].  Neurophysiological changes matching this recovery have been shown with another TMS measure, short-interval intracortical inhibition (SICI), showing reductions up to 3-months post-stroke, but returning to more age-normal values by approximately 6-months after the infarct [69]. Despite differences in the exact systems they are testing, a common neurophysiological component links SICI and SAI. SICI has been shown to relate to GABAA receptors; similarly, SAI relates to GABAA α5-subunit [25] as well as to the cholinergic system [28]. When examining this measure across the trajectory of stroke recovery, it follows a similar path as SAI in that there is a disinhibition in the acute phase, with a return to more age-normal values as well-recovered individuals progress to the chronic phase [69]. Thus, it is possible then that GABAAergic measures such as SAI and SICI are reduced acutely post-stroke to optimise the neurophysiological environment for recovery, but as function is regained, this shifts back towards more normalised values.  The pattern of AF shown in the current work suggests that, similar to SAI, there was no difference between older healthy individuals and individuals with chronic stroke. Yet there was a   53difference between the chronic stroke group and the group of younger healthy controls. As this is the first investigation, to our knowledge, of the influence of stroke on AF, at any stage of recovery, we cannot compare with other studies presenting relevant information for different stages of recovery. Animal work suggests that in the chronic phase of stroke recovery, an increased sensitivity to glutamatergic signaling may be relevant for recovery [130]. Though the underlying mechanism of AF is unknown, because of its facilitatory nature, there is likely a glutamatergic component; therefore, the increased AF seen in the current work may be reflective of a glutamate increase. A potential discrepancy in this logic is that there was no difference between individuals with chronic stroke and healthy older individuals, suggesting that the results may be driven by age-related factors. Although this cannot be ruled out, the results in the current study also show no difference between younger and older age groups suggesting that the older healthy individuals may have a level of AF somewhere in between the two groups. Future work should probe this further to parse out potential differences in AF between healthy older individuals and individuals with chronic stroke that could be addressed with larger sample sizes, or multiple ISIs that provide a more comprehensive and individualised picture of AF.   Our AF finding is in line with age-related investigations into other cortical measures of facilitation. Intracortical facilitation, a measure of motor cortical facilitation, also has no age-induced changes when compared across a wide range of age groups [1]. Thus, it is possible that neurobiological processes associated with healthy aging have less influence on facilitatory pathways within the motor cortex, which are potentially glutamatergic in nature. Future work to understand the underpinnings of AF will help to inform potential reasoning for age-related differences. For example, AF has previously been cited as a disinhibition from SAI, rather than true facilitation [98]. Considering individual differences in AF relative to SAI across a variety of   54age groups and neuropathologies may provide additional insight into neurobiological processes that accompany these populations. The similarity between AF in younger and older healthy populations is in accordance with previous literature [59], but in contrast to a separate study showing a reduction in AF in a group of older healthy individuals [60]. There are methodological differences between the current work and that done previously in that we individualised the ISI such that the TMS pulse was delivered 12 ms after each participant’s N20 rather than set at a standard ISI of ~50 ms. The additional time between conditioning and test stimuli in the previous work may exploit a slightly different underlying neural network and thus results may not be directly comparable. Future studies should work to establish more definitive age-related changes in AF.   The current work found no group differences in LAI. This is in line with past work showing that age-induced changes relate more to short-latency variables [59]. To date, LAI has not been examined in a population of individuals with chronic stroke. As both AF and LAI have been cited as oligosynaptic networks involving higher order cortical areas, it is possible that with the longer ISIs and involvement of an increased number of cortical pathways, age or pathology related compensation can occur to maintain normal afferent-induced patterns of inhibition and facilitation [22,29]. Elucidating the underlying mechanism of LAI in the future may help to explain such discrepancies seen with methodological differences.  2.4.2 Relationship between neurophysiology, impairment, and function The current study aimed not only to quantify chronic stroke and age-related changes in sensorimotor integration neurophysiology, but also to investigate the behavioural relevance of these measures. In individuals with chronic stroke, we found that the amount of disinhibition indexed by SAI was associated with both motor impairment and motor function. Greater   55amounts of disinhibition were associated with higher levels of motor impairment on the FM and lower levels of motor function on the WMFT. The current results indicate that SAI may follow a similar recovery trend to other cortical measures of inhibition where there is commonly disinhibition acutely post-stroke, but this is reduced at 6-months. Therefore, it is possible that an absence of restoration of SAI to “normal” age values is indicative of a poorer-recovery. There was no group difference between older adults and individuals with chronic stroke in the current work, which is in line with aging literature that suggests disinhibition in SAI relates to poorer performance on behavioural tasks in both cognitive and motor domains, such as Go-No-Go tasks, Manual Dexterity tasks, Response Time tasks, and Choice Reaction Time tasks [59]. The current work only conducted motor assessments in the stroke group, as they are designed to determine motor function and impairment in clinical populations, so we cannot comment on the influence of this disinhibition on this motor performance in the current healthy populations. The stroke work in the acute phase of recovery has documented an opposite relationship than that presented currently such that a reduction in acute inhibition (within 10 days post-stroke) is associated with greater motor function when tested at 6-month follow-up. In line with the hypothesis presented above and the idea of a spontaneous recovery period, it may be that disinhibition quantified partially by SAI is beneficial in the acute phase while long-term potentiation is essential for developing new, compensatory, neuronal pathways but this importance is abolished by 6-months post-stroke. Therefore, we believe the present data build on previous work done acutely to provide new information on the progression of sensorimotor integration throughout various stages of stroke recovery. Sensorimotor integration, quantified by SAI, may provide a neurophysiological indicator of potential motor functional status in individuals with chronic stroke. Of course, this does not definitively indicate that SAI   56abnormalities are causally related to motor function and impairment, but rather that an association exists between the two. Further study is necessary to fully understand the nature of this relationship. 2.4.3 Somatosensory and motor cortical excitability   In the populations studied here, there were no relationships between N20-P25 SEP component amplitudes and sensorimotor integration neurophysiology. Similarly, there were no relationships between recruitment curve slopes and SAI, AF, or LAI. Previous work has shown the potential for both somatosensory and motor cortical excitability to influence sensorimotor integration depending on the population. For example, recent work suggests that the size of the N20-P25 SEP component relates to SAI and LAI in young healthy people; specifically, larger amplitude N20-P25, indicative of increased somatosensory information ascending to the cortex, correlates with greater inhibition [31,131]. Despite not reaching significance, the current work showed a similar, though only trending, relationship with SAI (p=0.08) in young healthy individuals, while no such trend was seen in older individuals or individuals post-stroke.  The relationship between motor cortical excitability and sensorimotor integration has commonly been assessed by extrapolating results from studies using non-invasive brain stimulation to reduce motor cortical activity and then quantifying the change in SAI, AF, and LAI. Collectively, these investigations largely show no effect of M1 excitability on neurophysiological measures of sensorimotor integration [23]. Our results are in line with this work, though tested in a different manner. Correlational assessment of recruitment curve slope, indicative of motor cortical and corticospinal tract excitability, and measures of sensorimotor integration showed no relationship across both young and older healthy individuals, as well as individuals with chronic stroke. Taken together with these results, neurophysiological   57investigation of sensorimotor integration seems to provide unique information in addition to that available from somatosensory and motor cortical excitability studies.  2.4.4 Limitations  The current study employed a heterogeneous sample of individuals in the chronic phase of stroke recovery. Our inclusion criteria focused on the presence or absence of motor deficits, rather than controlling lesion location. Further, all of the individuals with stroke who participated were well-recovered and presented with an ipsilesional MEP. This is largely driven by our methodological approach in that TMS requires an EMG output to quantify sensorimotor integration. Thus, the clinical population in the current work may not be representative of the broader population of individuals post-stroke. Future work could examine a similar question while considering the specificity of lesion location and the potential influence on neurophysiological or behavioural differences as well as the impact of severity on sensorimotor integration.  2.5 Conclusions Well-recovered individuals in the chronic phase of stroke recovery do not show neurophysiological differences in metrics of sensorimotor integration, SAI and AF, beyond those shown with healthy aging. Behaviourally, ipsilesional SAI relates to both motor impairment and motor function in individuals with chronic stroke such that greater disinhibition corresponds to worse motor function and impairment. Therefore, while the neurophysiology of sensorimotor integration examined here show similar patterns in healthy older individuals and well-recovered individuals with chronic stroke, quantification of these measures may provide unique insight into chronic levels of motor impairment.     58Chapter 3: Sensorimotor integration and aging: baseline differences and response to sensory training 3.1 Introduction   Somatosensory information is essential to inform motor output and allow for successful interaction with the environment. The neurophysiological process through which somatosensory information from the periphery ascends to the primary motor cortex (M1) and influences motor output is termed sensorimotor integration. Sensorimotor integration can occur through two related, yet distinct neuroanatomical pathways based on the somatosensory receptors activated. Through the activation of proprioceptive related sensory receptors such as muscle spindles, which in turn recruit 1a afferent fibres, somatosensory information can ascend to Brodmann area 3a of the primary somatosensory cortex (S1), which is directly connected to M1 (direct sensorimotor integration) [32,33,39]. Activation of other somatosensory fibres in the periphery, such as those arising from various cutaneous receptors, travel primarily through a variety of large diameter fibres to Brodmann area 3b of S1 before relaying to M1 (indirect sensorimotor integration) [32,33,39]. Animal work confirms these pathways, showing that somatosensory information can ascend directly to M1 after relaying through higher-order sensory thalamic nuclei such as the medial posterior nucleus [132] and the ventrolateral thalamic nuclei [133,134]. Alternately, information from S1 can excite pyramidal neurons in layers 2 and 3 of M1 through monosynaptic corticocortical connections [9,34-36]. While somatosensory information will not travel exclusively in a single pathway, there is preferential initial cortical activation in M1 or S1 depending on the peripheral afferent receptors being stimulated.    59Experimentally, using peripheral stimulation in concert with transcranial magnetic stimulation (TMS) provides a metric to assess sensorimotor integration. Using these tools, we cannot definitively address the specific neuroanatomical pathways associated with sensorimotor integration. However, evidence suggests that S1 excitability largely contributes to responses in M1 following peripheral nerve stimulation [31]. In contrast, muscle belly vibration is thought to more directly impact the motor cortex [42]. Therefore, while each type of peripheral stimulation may activate both pathways, responses to nerve stimulation may predominantly be driven by corticocortical connections from S1, while responses to vibration are more reflective of thalamocortical connections.  Neurophysiological and behavioural age-related changes in sensorimotor systems have been demonstrated, yet the majority of the literature focuses on the somatosensory and motor systems separately, with little work quantifying the influence of age on sensorimotor integration. Preliminary work in this area has exclusively focused on indirect sensorimotor integration [29,59]. To assess indirect sensorimotor integration, peripheral nerve stimulation is paired with TMS at various interstimulus intervals (ISIs) to probe afferent induced inhibition or facilitation in M1. Generally, an age-related reduction in the influence of afferent information on motor output has been shown [1,59]. Short-latency afferent inhibition (SAI) and afferent facilitation (AF) have previously been shown to be more susceptible to age-related changes than long-latency afferent inhibition (LAI). Importantly, age-related decline in indirect sensorimotor integration relates to poorer performance on behavioural outcomes in both cognitive and motor domains. Due to the unique functional relevance of indirect sensorimotor integration, investigation into potential age-related modulation in direct sensorimotor integration is warranted.    60Direct sensorimotor integration can be probed with muscle belly vibration in order to activate muscle spindles and test their effect on M1. Work in young healthy individuals assessed how vibration-induced activation of proprioceptive afferent input impacts multiple synaptic networks within M1, including corticospinal excitability, GABAergic inhibitory interneurons, and glutamatergic interneuronal circuits. More specifically, these results showed that activation of the direct pathway of sensorimotor integration increases corticospinal excitability and reduces short-interval intracortical inhibition (SICI) [41,42]. Though less conclusive, other work suggested that the direct pathway can also influence interneuronal facilitatory networks within M1 as quantified with intracortical facilitation (ICF) [42]. Yet to date the bulk of work into the direct pathway of sensorimotor integration has focused on young adults. Given other examples of age related changes to both the somatosensory [43,44] and motor systems [48-50] it is important to consider how afferent information is impacting the motor system in healthy older adults.  Plasticity resulting from single-session interventions has been shown to induce changes in somatosensory [135] and motor [87] neurophysiology in healthy individuals; while the degree of change may be reduced in older adults, the ability of the brain to undergo plasticity in response to single-session interventions remains [1]. In young healthy individuals, a vibration-based sensory training paradigm can induce change in measures of direct sensorimotor integration, and these neurophysiological changes translated to improved motor performance [97]. Specifically, fifteen minutes of cyclic vibration applied over the abductor pollicis brevis (APB), while individuals were instructed to attend to vibration and detect changes in frequency, changed the direct integration of afferent information into M1. Importantly these changes occurred without altering measures representing transmission through indirect pathways of sensorimotor integration or measures of motor cortical excitability alone [97,136]. Direct   61sensorimotor integration, quantified with peripheral vibration and paired-pulse TMS induced a disinhibition in SICI as compared to the influence shown prior to sensory training, with the degree of change correlating with improvements in single-session performance of a thumb abduction task. Importantly, if vibration was applied in the same manner without attention directed towards the stimulation, these patterns were not seen [95]. These results suggest that a sensory training task can be used to enhance sensorimotor integration in a young healthy population, though this has yet to be replicated in a population of older adults.   The current work has two main aims: 1) to understand age-related differences in direct sensorimotor integration, and 2) to determine if sensorimotor integration in older adults changes in response to a sensory training task. We hypothesised that: 1) similar to previous research examining indirect sensorimotor integration, direct sensorimotor integration would be affected such that afferent information would have less of an influence on motor cortical circuits in older individuals compared to younger healthy controls; and 2) sensory training would induce change in direct sensorimotor integration to a lesser extent in older individuals compared to their younger counterparts due to a reduced ascension of afferent information to the motor cortex. No change in measures quantifying indirect sensorimotor integration was expected in either group. 3.2 Methods 3.2.1 Participants  The same twenty-four individuals from Chapter 2 were recruited to participate in this study: twelve older healthy individuals (69.2 ± 10.0 years 7M/5F) and twelve younger healthy individuals (27.6 ± 4.7 years, 3M/9F). Informed consent was obtained from all participants and they were screened for contraindications to TMS using a standard screening form. The Clinical   62Research Ethics Board at the University of British Columbia approved all experimental procedures.  3.2.2 Experimental design Each individual participated in two sessions. In the first session, somatosensory functional tests were conducted, followed by baseline neurophysiological evaluation. In the second session, sensory training was delivered, followed by the same neurophysiological investigation from the first session. An overview of the experimental procedures can be found in Figure 3.1.  Figure 3.1: Experimental procedures. Behavioural assessment of somatosensory function was conducted on the first day, followed by neurophysiological assessment. The second day included sensory training, followed by the same neurophysiological assessment conducted on day 1. Mono: monofilaments, SAI: short-latency afferent inhibition, AF: afferent facilitation, LAI: long-latency afferent inhibition, SICI: short-interval intracortical inhibition, ICF: intracortical facilitation.   3.2.3 Somatosensory functional tests Baseline Tactile Semmes-Weinstein Monofilaments assessed tactile perceptual thresholds (Quick Medical, Issaquah, WA, USA). Arm-position matching quantified between-limb proprioception using the KINARM (BKIN Technologies Ltd, Kingston, ON, Canada). All measures were quantified bilaterally.  3.2.4 Neurophysiological assessment Individuals were seated in an upright, comfortable position and instructed to relax. Neurophysiological measures of somatosensory cortical excitability, motor cortical excitability,   63and sensorimotor integration were collected. All measures were collected from the non-dominant hemisphere and non-dominant hand.  3.2.4.1 Somatosensory evoked potentials SEPs were recorded following median nerve stimulation (pulse width 200 s, square wave pulse, cathode distal, anode proximal) with surface electrodes corresponding to CP4 or CP3 positioning in accordance with the International 10-20 System (contralateral somatosensory cortex) and referenced to AFz (2000 Hz sampling rate) (NeuroPrax; Neuroconn, Ilmenau, Germany). Channel impedances were < 5 kΩ. Briefly, stimulation at 2 Hz (Digitimer DS7AH, Welwyn Garden City, Hertfordshire, UK) was delivered at motor threshold, defined as the minimum intensity required to evoke a visible twitch in the target muscle. Recordings from 300 stimuli were collected. Surface EMG were recorded from the right abductor pollicis brevis (APB) muscle using silver/silver-chloride disc surface electrodes (1 cm diameter) in a belly tendon montage in order to monitor the amplitude of the m-wave. The EMG signal was amplified and analogue filtered (30 Hz to 1 kHz) with a Powerlab 4/30 EMG System (AD Instruments, Colorado Springs, CO). In order to analyse SEP data, an average trace was produced to extract the component amplitudes. 3.2.4.2 Transcranial magnetic stimulation TMS was delivered following established techniques [124]. Single pulse TMS was delivered using a monophasic figure-of-eight shaped coil (Magstim 70 mm P/N 9790, Magstim Co., UK) connected to a Magstim 2002 stimulator (Magstim Co., UK). Stimuli were given with a random ISI of 4-5 seconds. The coil was held in such a way to induce a posterior-anterior current in the brain with the coil handle positioned at an angle of 45 to the mid-sagittal plane pointing backwards. The APB ‘hot-spot’ was located using neuronavigation to guide the search and to   64ensure consistent coil positioning throughout the experiment using a template MRI (Brainsight™, Rogue Research Inc., Montreal, QC, Canada). Resting motor threshold (RMT) was determined by finding the lowest stimulation intensity required to evoke MEPs of at least 50 µV in 5 out of 10 consecutive trials [124].   Recruitment curves were collected to probe corticospinal excitability. A total of 100 single pulse stimulations at ten intensities, ranging in 10% increments from 80-170% RMT, were delivered. The order of intensities was randomised. MEP amplitude was averaged across the ten trials of each intensity. 3.2.4.3 Direct sensorimotor integration  Direct sensorimotor integration was tested by pairing APB muscle belly vibration with single and paired-pulse TMS as previously described by Rosenkranz where TMS measures were collected with and without vibration to enable a direct comparison motor circuitry alone and somatosensory influences on motor circuitry [42]. Briefly, vibration was applied over the muscle belly of APB at a frequency of 80 Hz (0.2-0.5 mm amplitude) using a 0.7 cm diameter probe. The amplitude of vibration was adjusted, if needed, to be below threshold for inducing or perceiving movement, and EMG was monitored for confirmation. Vibration was applied for 1.5 s with a 3.5 s ISI. Single or paired-pulses of TMS were delivered 1 s into the vibration train over the contralateral APB representation in M1. In order to quantify the influence of somatosensory information on motor cortical excitability, as well as the inhibitory circuits within M1, multiple TMS measures were collected (MEPs, short-interval intracortical inhibition (SICI), intracortical facilitation (ICF)) [42]. As previously described [16], SICI was conducted using two TMS pulses (a subthreshold conditioning stimulus (CS) followed by a suprathreshold test stimulus (TS)), administered over M1 with an interstimulus interval (ISI) of 2 ms, while ICF employs the same   65procedure with an ISI of 12 ms. In the present experiment, the CS was set at an intensity of 80% RMT with the TS intensity set to produce consistent unconditioned MEPs of ~1 mV amplitude. Inhibition and facilitation were then presented as a ratio of the paired pulse and unconditioned single pulse MEP amplitudes. Ten pulses of all measures (TS alone, CS+TS for SICI and ICF) were collected with and without vibration.  3.2.4.4 Indirect sensorimotor integration  Indirect sensorimotor integration was tested using common techniques (SAI, AF, and LAI) that pair TMS with median nerve stimulation. Specifically, an electrical stimulation was delivered over the contralateral median nerve prior to a TMS pulse delivered over the motor cortex while the participant was at rest [22,25,29]. Median nerve stimulation intensity was set just above motor threshold where a twitch was visible and an m-wave was consistently produced. TMS pulse intensity was set to produce a MEP of ~1 mV (TS) [25]. ISIs were individualised such that the ISI for SAI was 2 ms longer than the N20 latency derived from the SEP trace, and the ISI for AF was 12 ms longer than the N20 latency [23]. LAI utilised an ISI of 200 ms [25-27,125,126,137]. Ten pulses of each technique, as well as ten pulses of unconditioned TS were collected. M-wave amplitudes were analysed following the first session in order to ensure consistency and guide stimulation intensity in the second session. 3.2.5 Sensory training  Sensory training consisted of 15 minutes of 80 Hz vibration (0.2-0.5 mm amplitude) applied over the non-dominant APB muscle belly in 2 second trains with an inter-train interval of 2 seconds [97,136]. The frequency of vibration was changed with 300 ms left in 70% of trains. If a frequency change was detected, individuals were asked to respond by pressing the space bar on a keyboard in front of them. Vibration frequency was changed from 80 Hz to 65 Hz, 67.5 Hz, 70   66Hz, 72.5 Hz, 75 Hz, or 77.5 Hz in the group of young healthy individuals. In the older healthy population, to equate difficulty, account for age-related somatosensory decline, and attempt to equalise attention demand, deviations in frequency occurred in 10 Hz increments ranging from 70 Hz to 20 Hz. This adjustment was based on pilot work. 3.2.6 Statistical tests  To understand baseline differences in direct sensorimotor integration, two-way mixed-model ANOVAs were performed, including within-subjects factor VIBRATION (without vibration, with vibration) and between-subjects factor GROUP (younger, older individuals) for each neurophysiological measure of interest (single pulse TMS, SICI, and ICF). Post-hoc analysis was applied with Tukey’s HSD where appropriate. Additionally, pre-planned contrasts were employed to directly test the hypothesis that afferent information arising from vibration would have less of a cortical influence in older healthy individuals, as compared to younger healthy individuals.  Following the analysis of baseline sensorimotor integration, dependent variables were expressed as a percentage (i.e. (SICI with vibration/SICI without vibration)*100). Next, two-way mixed-model ANOVAs were performed, including within-subjects factor TIME (pre and post sensory training) and between-subjects factor GROUP (young, older individuals) for each dependent variable (MEP, SICI, ICF) in order to assess the influence of sensory training on sensorimotor integration.  As a secondary analysis, a mixed-model two-way ANOVA (Group x Time) was performed to determine if there was an effect of sensory training on measures of indirect sensorimotor integration (SAI, AF, and LAI), and corticospinal excitability, quantified with recruitment curve slope.    67 Finally, bivariate correlational analyses were performed to understand the relationship between somatosensory function and measures of neurophysiology. Non-normal data were assessed with Spearman’s correlations (rs), and normal data used Pearson’s correlations (rp).   For all statistical tests, a significance level of p≤0.05 was used. 3.3 Results  Data were checked for normality and log transformed when abnormal, as indexed by significance of p < 0.001 in the Shapiro-Wilks test [128]. In older healthy individuals, ICF with vibration was non-normal in the first session. All other variables were normally distributed in both groups. Data from one individual in the young healthy group was not included due to excess noise in all measures of direct sensorimotor integration; this also occurred in our measure of SICI with vibration in one older healthy individual.  3.3.1 Baseline direct sensorimotor integration  Sample figures for each measure of direct sensorimotor integration in both age groups can be seen in Figure 3.2.   68 Figure 3.2: Representative traces for measures of direct sensorimotor integration for younger healthy individuals (left) and older healthy individuals (right). The top panel depicts single pulse response to vibration. The middle panel shows the response of SICI to vibration. The bottom panel displays the influence of vibration on ICF. MEP: motor evoked potential, TS: test stimulus, Vib: vibration, SICI: short interval intracortical inhibition, ICF: intracortical facilitation.  Two-way ANOVA results revealed a main effect of Vibration (F(1,21)=16.24, p=0.001, η2partial=0.436; Figure 3.3) for single-pulse MEP amplitude. Post-hoc analysis determined that vibration increased MEP amplitude in both groups (p=.0007). Pre-planned contrasts revealed no group difference in the influence of afferent feedback at baseline (p=0.83). Two-way mixed-model ANOVA indicated that there was no influence of vibration on SICI amplitude, and no group difference (p=0.55; Figure 3.3). This was confirmed with the pre-planned contrast directly comparing influence of vibration between groups (p=0.52).   69Two-way mixed-model ANOVA disclosed a main effect of Vibration (F(1,21)=5.550, p=0.028, η2partial=0.209; Figure 3.3). Post-hoc analysis revealed that when concurrent vibration was applied ICF was greater than ICF alone in both groups prior to sensory training (p=0.030). Pre-planned contrasts revealed no group difference in the influence of afferent feedback at baseline (p=0.56).   Figure 3.3: Baseline direct sensorimotor integration. A shows the response of single-pulse TMS to vibration. B presents data showing the influence of vibration on SICI. C displays the impact of vibration on ICF. Vibration increases the MEP amplitude following single-pulse TMS, and increases ICF in both younger and older healthy individuals. MEP: motor evoked potential, SICI: short interval intracortical inhibition, ICF: intracortical facilitation. Asterisks denote statistical significance. Error bars represent standard error of the mean.  3.3.2 Influence of sensory training on sensorimotor integration Two-way mixed-model ANOVA displayed no significant effects of sensory training on single-pulse MEP amplitude in either group. There was a non-significant trend for an effect of Time (p=0.07), towards a reduction in vibration-induced facilitation following sensory training   70(Figure 3.4). Further, there were no significant differences in the influence of vibration on SICI following sensory training in either group (Figure 3.4).  Two-way mixed model ANOVA revealed a main effect of Time when examining the influence of vibration on ICF (F(1,21)=11.34, p=0.0029, η2partial=0.351). Post-hoc analysis revealed a decreased vibration-induced facilitation of ICF following sensory training in younger and older healthy individuals (p=0.0032; Figure 3.4).   Figure 3.4: Influence of sensory training on measures of direct sensorimotor integration. The response of single-pulse MEPs with vibration to sensory training is shown in A. B displays the impact of sensory training on SICIvib. C shows the response of ICFvib to sensory training. Following sensory training, there was a reduction on the influence of vibration on ICF in both younger and older healthy individuals. MEP: motor evoked potential, SICI: short interval intracortical inhibition, ICF: intracortical facilitation. Asterisks denote statistical significance. Error bars represent standard error of the mean.  Pre-planned contrasts directly testing the hypothesis that sensory training would have a smaller effect on older healthy individuals as compared to younger healthy individuals revealed no group difference for any measure.    713.3.3 Secondary measures – indirect sensorimotor integration  There were no significant differences in measures of indirect sensorimotor integration induced by sensory training (Figure 3.5). A main effect of Group for SAI was revealed by a two-way mixed model ANOVA (F(1,22)=4.844, p=.039, η2partial=0.18), as has previously been reported (Chapter 2). Post-hoc analysis revealed a reduction in SAI in older healthy individuals, compared to younger healthy controls (p=0.039). There were no changes in recruitment curve slope, indicative of no influence of sensory training on corticospinal tract excitability. Further, there were no significant correlations between measures of sensorimotor integration and somatosensory function.  Figure 3.5: Influence of sensory training on indirect sensorimotor integration. A shows SAI amplitudes prior to and following sensory training. B depicts the influence of sensory training on AF. C displays values for LAI prior to and following sensory training. There is a reduction in the amount of SAI shown in older, as compared to younger, healthy individuals. SAI: short-latency afferent inhibition, AF: afferent facilitation, LAI: long-latency afferent inhibition. Asterisks denote statistical significance. Error bars represent standard error of the mean.   723.3.4 Sensory training performance Group means for older and younger individuals were 58.19% and 63.79% accuracy, respectively. There was no significant group difference in sensory training performance (t(20)=-0.958, p=.348). There was no relationship between task performance and neurophysiological changes induced by sensory training.  3.4 Discussion   The current work showed, for the first time, that healthy aging does not influence measures of direct sensorimotor integration; single-pulse MEPs and ICF were increased with vibration in both younger and older healthy individuals. Response to sensory training was similar in younger and older healthy individuals, with changes seen in measures that quantified direct, but not indirect sensorimotor integration. Direct sensorimotor integration was reduced following sensory training, with decreased vibration-induced changes in ICF. Sensory training did not alter indirect sensorimotor integration, quantified with SAI, AF, or LAI. Further, sensory training did not induce any changes in measures of corticospinal or motor cortical excitability. This finding suggests that sensory training specifically impacts direct sensorimotor integration, rather than altering indirect sensorimotor integration or purely motor-based measures irrespective of afferent feedback. 3.4.1 Baseline sensorimotor integration Results from the current work show that healthy aging does not affect neurophysiological measures of direct sensorimotor integration. This is in contrast to past work showing reduced indirect sensorimotor integration quantified with SAI. Together, these results suggest differential age-related modulation of separate anatomical pathways responsible for sensorimotor integration.    73Activation of the pathway responsible for direct sensorimotor integration in the current work induced change in both single-pulse and facilitatory paired-pulse measures. Given that vibration influenced both single and paired-pulse TMS measures, afferent feedback through direct thalamocortical projections to M1 or corticocortical connections from Brodmann area 3a synapsing onto interneurons in layers 2 and 3 can modulate multiple circuits. Vibration-induced facilitation of single-pulse MEP responses is thought to result from both spinal and cortical components [41]. While our current methods do not differentiate between these two components, given that there was no age-related change in this measure, there is not evidence to suggest differences in the cortical response to afferent feedback that is of interest here. In contrast to single-pulse responses, vibration-induced changes in ICF are likely cortical in nature, as ICF is thought to relate to glutamatergic interneuronal circuitry distinct from interneurons activated during single-pulse MEPs [125]. Past imaging work showed that there is no relationship between age and cortical response to vibration; hand vibration produced a similar response in primary sensorimotor cortical blood-flow across individuals ranging from age 20 to 72 years [138]. Here, we show that in addition to that, the direct integration of this somatosensory information into different motor circuitry does not change with age.  Though not the primary aim of the study, contextualisation of our results with past work examining the influence of peripheral vibration on TMS measures is important. In data from our group of younger healthy individuals, we replicated previous work showing a vibration-induced increase in MEP amplitude, though we did not show the previously documented vibration-induced release of SICI [97]. The response of ICF to vibration has been less consistent, with some work suggesting no influence of vibration [95,97,136]. However, our results add support to a separate study showing a vibration-induced increase in the amount of facilitation [42].   74Vibration has previously been shown to reduce SICI, but this was not replicated in the current work. On average, there was less SICI when tested with concurrent vibration (68% in older individuals and 76% in younger individuals); however, the variability in this response was too great to reach statistical significance (standard errors of 22% and 56%, respectively). Beyond this, we cannot speculate what may underlie the difference between our results and those previously published.  Neurophysiological measures using peripheral nerve stimulation to index indirect sensorimotor integration have shown age-related changes. Though the techniques utilised here cannot determine the exact anatomical pathways through which afferent feedback is being transmitted, our results suggest a differential effect of aging on thalamocortical and corticocortical connections subserving direct and indirect sensorimotor integration, respectively. Given the reliance of movement and behaviour on processes of sensorimotor integration, being able to utilise the direct pathways to preserve sensorimotor processes may be a way in which to reduce age-related decline in motor behaviour. 3.4.2 Neurophysiological response to sensory training  Neurophysiological changes in response to a 15-minute block of sensory training were similar in younger and older healthy adults.  This finding suggests that this intervention can manipulate the neurophysiology of sensorimotor integration. Following sensory training, both age groups showed a decrease in vibration-induced enhancement of ICF. These results were independent of changes in ICF collected without concurrent vibration suggesting that sensory training targets neurophysiological networks specifically related to the incorporation of afferent feedback into motor cortical circuitry. Additionally, sensory-training only induced change in measures of the direct pathway of sensorimotor integration, without changing measures of the   75indirect pathway of sensorimotor integration (i.e., SAI, AF, LAI), or measures of motor cortical excitability without paired afferent stimulation. This indicates the potential to target specific sensorimotor integration circuits that are differentially influenced by healthy aging or pathology.  The findings from Aim 1 in the current work showed that aging does not influence the impact of vibration on M1. Further, this finding helps us interpret the results from Aim 2 as it follows that the response to repetitive vibration would likely be similar. If vibration did not influence specific cortical circuits at baseline, such as was seen with SICI, it is logical that repeated delivery of vibratory stimulation would not induce change. Finally, given that the vibration-induced facilitation of single-pulse MEPs is largely driven by spinal mechanisms, and the effect of sensory training is thought to be cortical in nature [136], this may provide an explanation for the lack of change seen in single-pulse MEP measures with vibration.  Our finding of a reduction in vibration-induced changes in ICF is in contrast with previous work, which has shown a further release of SICI with concurrent vibration, but no change to ICF. The neurophysiological response to sensory training is heavily dependent on attention, which may underpin the current results [89,95]. If, rather than instructing individuals to attend to the vibration in order to detect subtle changes in frequency, vibration is applied in the same pattern, without any attentional direction for 15-minutes, the physiological response to vibration is absent [95]. Despite our attempt to control attention by requiring individuals to respond when a frequency change was detected, our results follow this pattern. Ex-vivo results show diminished responses in neurons involved in processing frequently unattended stimulation, which slowly attenuates the response to repetitive stimuli [139]. Given that the sensory training paradigm used in this work applied 80 Hz vibration over the APB 225 times in 15 minutes, it is possible that a similar situation is occurring throughout the training. Specifically, if neurons in   76M1 responsible for the increase ICF when concurrent vibration is applied are less responsive after being repeatedly stimulated, sensory training may then negate the effect of vibration. If this is the case, our attentional manipulation was not effective enough to facilitate the response that has been previously shown.  Similarly, somatosensory processing can be heavily modified by attention-related prefrontal networks, which may provide an underlying mechanism for attention-dependent changes in responses to intervention [54]. If an individual is not attentive to the incoming somatosensory information, or it is deemed irrelevant, less information will ascend to the cortex, and the influence of this information may be lessened. Thus, if not attentionally engaged throughout the sensory training, the cortical response to vibration may be reduced to the point at which there is no longer a response to the specific stimuli. In the current work, this appears to be the case with ICF; following sensory training, vibration did not induce facilitation beyond that seen without vibration. If a sensory training paradigm is to be used in the future to alter sensorimotor integration, the attentional resources required in different populations must be considered in an attempt to drive neurophysiological change. Of secondary interest, we examined whether sensory training induced changes in indirect sensorimotor integration. Previous work has shown that effects of attention are modality specific such that attention to vibration changes vibration-based measures, where attention to vibration does not alter response to cutaneous stimulation, and vice versa. Additionally, work investigating passive vibration presented in 2s cycles for 15 minutes showed no effect on nerve-based measures [96]. However, it was noted that the sensory intervention did have easily identifiable responders and non-responders; this may just be indicative of variability. Future work could investigate this as a potential avenue for explaining differences in findings on the impact of   77vibration-based training. As has been previously highlighted, attention seems to be essential for inducing neurophysiological changes. Therefore, we aimed to control attention by having individuals detect frequency changes; however, this may not be adequate for driving attentional resources. Taken together, past results and the current findings support the notion that sensory training does not induce changes in SAI, AF, and LAI. Importantly, this is true in both younger and older healthy populations. 3.4.3 Limitations  One limitation to this study is that the pre and post measures were collected on separate days. The study was designed as such in an attempt to optimise attention to the sensory training task, given the importance previously noted in the literature of attention in inducing neuroplastic changes. Further, it enabled us to include additional measures of neurophysiology in our design, which provided unique information on the changes induced by sensory training. Including these measures added time to each collection session, which we felt could influence participant’s attention levels. Finally, given that our indices were ratios of measures both collected on the same day, we felt that they would be relatively stable over time. In support of this, there were no differences between MEP amplitudes in response to test stimuli collected in the first and second sessions, and it is unlikely that this factor was responsible for the effects seen.  Another limitation is that we only examined the influence of vibration on a single hand muscle. Previous work looking at sensorimotor organization using TMS and vibration examined how M1 excitability, inhibition, and facilitation are altered in multiple muscles, and changes in the pattern of these multi-effector relationships. Future work should examine whether similar responses are seen in other hand muscles or if these more complex relationships are altered in healthy older individuals.    783.5 Conclusions Direct sensorimotor integration remains intact with healthy aging, in contrast to indirect sensorimotor integration. In addition, response to sensory training is similar across ages, thus making this a viable path for investigating or potentially altering sensorimotor integration. Future work should investigate the behavioural relevance of these neurophysiological circuits and sensory training-induced changes.    79Chapter 4: Sensorimotor integration in chronic stroke: baseline differences and response to sensory training  4.1 Introduction   Motor impairment commonly persists into the chronic phase of stroke recovery, with only 5 to 20% of stroke survivors attaining ‘completely useful function’ of the upper-limb [140,141]. Utilising behavioural evidence of motor impairment alone, it is hard to understand the underlying neurophysiological mechanisms contributing to such deficits. For example, in addition to stroke-induced alterations to the somatosensory and/or motor systems, changes in the link between the two systems, termed sensorimotor integration, may also contribute to deficits in motor impairment and function. Thus, understanding the neurophysiology of sensorimotor integration in the chronic phase of stroke recovery is paramount to furthering knowledge of underlying factors leading to persisting motor deficit. Additionally, interventions may then be designed to target specific underlying neurophysiological change induced by stroke, with the ultimate goal of reducing motor symptomology.  Neuroanatomically there are two proposed pathways that contribute to sensorimotor integration. While somatosensory information will not be exclusively transmitted through a single pathway, separate pathways are preferentially activated based on peripheral receptor activation. Proprioceptive information arising from the activation of muscle spindles and predominantly travelling through 1a afferent fibres can be transmitted directly to the primary motor cortex (M1) or through a direct relay from Brodmann area 3a of the primary somatosensory cortex (S1) (direct sensorimotor integration) [32,33,39]. Somatosensory information arising from activation of other types of somatosensory receptors, such as cutaneous   80receptors, travels through a variety of large diameter fibres and is thought to predominantly ascend to Brodmann area 3b of S1 before relaying to M1 through corticocortical connections in areas 1 and 2 (indirect sensorimotor integration) [32,33,39]. Practically, there is likely contribution from both pathways underlying sensorimotor integration, as transmission will not occur exclusively in one anatomical pathway. Animal work shows direct anatomical connections to M1 from the medial posterior and ventrolateral thalamic nuclei, which likely contribute to direct sensorimotor integration [132-134], whereas indirect sensorimotor integration likely occurs through connections from area 3b of S1 synapsing into upper layers of the motor cortex and in turn exciting pyramidal neurons [9,34-36]. To date, there is little work examining sensorimotor integration in either pathway in individuals with chronic stroke.  Indirect sensorimotor integration is typically probed by pairing median nerve stimulation with transcranial magnetic stimulation (TMS) to assess the influence of somatosensory information being relayed from S1 to M1. Median nerve stimulation will activate a variety of large diameter afferent fibres that lead to preferential cortical activation in S1 before being transmitted to M1 [142]. Evidence for this arises from studies showing a relationship between S1 response to nerve stimulation and the response in M1 [142]. Further, altering levels of S1 cortical excitability, but not M1 excitability changes lead to changes in short-latency measures of indirect sensorimotor integration [37,38]. By altering the interstimulus interval (ISI) between the nerve stimulation and cortical stimulation, distinct neuronal circuits can be assessed, which provide unique information on different contributors to indirect sensorimotor integration. Afferent stimulation induces inhibition in M1 at both short and long latencies. Short-latency afferent inhibition (SAI) is thought to relate to cholinergic and GABAergic systems, and reflect direct projections from S1 to M1 [25]. Long-latency afferent inhibition (LAI) likely involves higher   81order cortical regions, though the underlying neural contributors are not fully understood [22,29]. In between these two periods of inhibition the facilitatory impact of afferent feedback can be assessed using a metric termed afferent facilitation (AF). Little work has characterised sensorimotor integration in individual who are in the acute stage post stroke (< 3 months post-stroke [143]) [70] and also those who are in the chronic stage post-stroke (Research Chapter 2). Results indicate that SAI may be decreased in the acute stage post stroke; however, this inhibition reaches age-normative values in well-recovered individuals with chronic stroke. Importantly, the amount of inhibition quantified with SAI relates to motor function and impairment in both stages of recovery, indicating the behavioural relevance of sensorimotor integration post-stroke (Research Chapter 2). Given behavioural evidence suggesting impairments in sensorimotor integration in individuals with chronic stroke [144], it’s possible that despite age-normative patterns in indirect sensorimotor integration, there are stroke-related changes in direct sensorimotor integration that contribute to these behavioural deficits. Direct sensorimotor integration can be probed using muscle belly vibration in combination with TMS as it preferentially recruits 1a afferents. By pairing vibration with both single and paired-pulse TMS methods, the influence of afferent feedback on various motor cortical networks can be assessed [41,42]. In young healthy individuals, using these techniques, vibration increases corticospinal excitability, reduces GABAergic receptor mediated inhibition within M1 interneuronal circuits, and may increase facilitatory interneuronal networks within M1 [41,42]. Preliminary work indicates that following stroke, vibration still increases single-pulse motor-evoked potential (MEP) amplitudes, though this response varies between the acute and sub-acute stages of recovery [145]. No work to date has examined paired-pulse measures of TMS that reflect motor cortical circuitry and are less dependent on spinal mechanisms than   82single-pulse TMS. To our knowledge, there has yet to be investigation into the influence of focal hand vibration on single and paired-pulse TMS measures of motor cortical excitability in individuals with chronic stroke. Such an assessment will extend knowledge of sensorimotor neurophysiological changes after stroke, and in conjunction with behavioural tests, determine if there is a relationship between direct sensorimotor integration, and somatosensory and motor function post-stroke.  Once a profile of sensorimotor integration in individuals with chronic stroke has been established, it is important to understand whether these measures are plastic and thus can be influenced by both single and multi-session interventions. Arguably, if baseline deficits exist following stroke, these individuals may be more likely to benefit from an intervention targeting sensorimotor integration. Post-stroke plasticity has been shown in various systems in both the acute and chronic phases of recovery [146,147]; however, investigation into the plasticity of sensorimotor integration is just beginning and has yet to be thoroughly explored in individuals with chronic stroke. The current study will use a sensory training paradigm that has been effective in healthy individuals to determine if similar intervention-induced changes in sensorimotor integration can be seen post-stroke. Specifically, fifteen minutes of attended cyclic vibration delivered over the muscle belly of an upper-limb or hand muscle can induce muscle specific neurophysiological changes in how afferent feedback is incorporated into the motor cortex, but not affect motor cortical excitability [97]. An abundance of clinical studies [148-150] use various vibration-based interventions, either alone or paired with motor practice, to induce behavioural improvement, yet the underlying physiology of these methods is not understood. The current investigation will provide insight into potential underlying neurophysiological change   83that may occur with vibration-based interventions and that could be responsible for the corresponding behavioural improvements.  Collectively, the current study addresses two main aims: 1) to establish baseline patterns of direct sensorimotor integration in individuals with chronic stroke by indexing the impact of peripheral vibration paired with single and paired-pulse TMS; and 2) to examine the effect of sensory training on neurophysiological measures of direct and indirect sensorimotor integration in chronic stroke. We hypothesise that: 1) direct sensorimotor integration will be affected such that vibration-activated afferent feedback will have less of an impact on motor cortical measures than has been previously shown in healthy populations; for example, measures in which vibration typically induces facilitation will be less facilitated in a group of individuals with chronic stroke as compared to healthy controls; and 2) sensory training will induce changes in measures of direct sensorimotor integration in individuals in the chronic phase of stroke recovery, potentially shifting these measures to more age-normative patterns. This response will be specific to measures of direct sensorimotor integration, with indirect sensorimotor integration not being influenced by the sensory training. 4.2 Methods 4.2.1 Participants  Fifteen individuals with chronic stroke (71.8 ± 8.2 years, 9M/6F) and twelve older healthy individuals (69.2 ± 10.0 years 7M/5F) from previous chapters were recruited to participate. Data from older healthy individuals had previously been used in our lab to examine sensorimotor integration changes accompanying healthy aging (Research Chapter 2). Informed consent was obtained from all participants and they were screened for contraindications to TMS   84using a standard screening form. All experimental procedures were approved by the Clinical Research Ethics Board at the University of British Columbia.  4.2.2 Experimental design The experimental design is very similar to that outlined in Chapter 3. Each individual participated in two sessions. In the first session, somatosensory and motor functional tests were conducted, followed by baseline neurophysiological evaluation. In the second session, sensory training was followed by the same neurophysiological investigation as was conducted in the first session. This is outlined in Figure 4.1.  Figure 4.1: Experimental overview. Behavioural assessment of motor and somatosensory function was conducted on the first day, followed by neurophysiological assessment. Individuals with chronic stroke had motor impairment and motor function assessed with the Fugl-Meyer Assessment and Wolf Motor Function Test, respectively. The second day included sensory training, followed by the same neurophysiological assessment conducted on day 1. Mono: monofilaments, SAI: short-latency afferent inhibition, AF: afferent facilitation, LAI: long-latency afferent inhibition, SICI: short-interval intracortical inhibition, ICF: intracortical facilitation.   4.2.3 Behavioural tests  Baseline Tactile Semmes-Weinstein Monofilaments were used to assess tactile perceptual thresholds (Quick Medical, Issaquah, WA, USA). Arm-position matching was used to quantify between limb proprioception using the KINARM (BKIN Technologies Ltd, Kingston, ON, Canada). All measures were quantified bilaterally.  Assessment of motor system impairment was characterised with Fugl-Meyer (FM) Upper Extremity Scale [121]. Arm motor function was assessed with the Wolf Motor Function Test (WMFT) [122]. Motor testing was only performed in individuals with chronic stroke.   854.2.4 Neurophysiological assessment  Individuals were seated in an upright, comfortable position and instructed to relax as much as possible. Neurophysiological measures of somatosensory cortical excitability, motor cortical excitability, and sensorimotor integration were collected. All measures were collected from the ipsilesional or nondominant hemisphere (paretic hand). 4.2.4.1 Somatosensory evoked potentials SEPs were recorded following median nerve stimulation (pulse width 200 s, square wave pulse, cathode distal, anode proximal) with surface electrodes corresponding to CP4 or CP3 positioning in accordance with the International 10-20 System (ipsilesional/nondominant somatosensory cortex) and referenced to AFz (2000 Hz sampling rate) (NeuroPrax; Neuroconn, Ilmenau, Germany). Channel impedances were < 5 kΩ. Briefly, stimulation at 2 Hz (Digitimer DS7AH, Welwyn Garden City, Hertfordshire, UK) was delivered at motor threshold, defined as the minimum intensity required to evoke a visible twitch in the target muscle. Recordings from 300 stimuli were collected. Surface EMG were recorded from the right abductor pollicis brevis (APB) muscle using silver/silver-chloride disc surface electrodes (1 cm diameter) in a belly tendon montage in order to monitor the amplitude of the m-wave. The EMG signal was amplified and analogue filtered (30 Hz to 1 kHz) with a Powerlab 4/30 EMG System (AD Instruments, Colorado Springs, CO). In order to analyse SEP data, an average trace was produced to extract the component amplitudes. 4.2.4.2 Transcranial magnetic stimulation  TMS was performed as previously described using established techniques [124]. Single pulse TMS was delivered using a monophasic figure-of-eight shaped coil (Magstim 70 mm P/N 9790, Magstim Co., UK) connected to a Magstim 2002 stimulator (Magstim Co., UK). Stimuli   86were given with a random ISI of 4-5 seconds. The coil was held in such a way to induce a posterior-anterior flow with the coil handle positioned at an angle of 45 pointing backwards. The APB ‘hot-spot’ was located using neuronavigation in concert with a standardised MRI to guide the search and to ensure consistent coil positioning throughout the experiment (Brainsight™, Rogue Research Inc., Montreal, QC, Canada). Resting motor threshold (RMT) was determined by finding the lowest stimulation intensity required to evoke MEPs of at least 50 µV in 5 out of 10 consecutive trials [124].   Recruitment curves were collected to probe corticospinal tract excitability. A total of 100 single pulse stimulations at ten intensities, ranging in 10% increments from 80-170% RMT, were delivered. The order of intensities was randomised. MEP amplitude was averaged across the ten trials at each intensity. 4.2.4.3 Sensorimotor integration  Pairing APB muscle belly vibration with TMS tested direct sensorimotor integration. Vibration-based measures were collected as previously described by Rosenkranz et al. (2003) where TMS measures were collected with and without vibration to enable a direct comparison between motor circuitry alone and somatosensory influences on motor circuitry [42]. Briefly, vibration was applied over the muscle belly of the paretic/nondominant APB at a frequency of 80 Hz using a 0.7 cm diameter probe (0.2-0.5 mm amplitude). The amplitude was below the threshold for perceiving movement, and EMG was monitored for confirmation. Vibration was applied for a duration of 1.5 s with a 3.5 s ISI. TMS was delivered 1 s into the vibration train over the contralateral APB representation in M1. In order to quantify the influence of somatosensory information on corticospinal excitability, as well as the inhibitory and facilitatory circuits within M1, multiple TMS measures were collected (MEPs, short-interval intracortical   87inhibition (SICI), intracortical facilitation (ICF)) [42]. As previously described, SICI was conducted using two TMS pulses (a subthreshold conditioning stimulus (CS) followed by a suprathreshold test stimulus (TS)), administered over M1 with an interstimulus interval (ISI) of 2 ms, while ICF employs the same procedure with an ISI of 12 ms [16]. In the present experiment, the conditioning stimulus was set at an intensity of 80% RMT with the test stimulus intensity set to produce consistent unconditioned MEPs of 1 mV amplitude. Inhibition and facilitation were then presented as a ratio of the paired pulse and unconditioned single pulse MEP amplitudes. Ten pulses of all measures (MEPs, SICI, ICF) were collected with and without vibration.  Indirect sensorimotor integration was tested with common techniques (SAI, AF, and LAI) using TMS, in conjunction with peripheral nerve stimulation, to examine the integration of somatosensory information from S1 into the motor cortex; specifically, an electrical stimulation was delivered over the contralateral (paretic/nondominant) median nerve prior to a TMS pulse delivered over the ipsilesional/nondominant motor cortex while the participant was at rest. Median nerve stimulation was set at motor threshold where a twitch was visible and an m-wave was consistently produced. TMS pulse intensity was set such that an MEP of 500-1000 µV was consistently produced (TS) [25]. ISIs were individualised such that the ISI for SAI was 2 ms longer than the N20 latency derived from the SEP trace, and the ISI for AF was 12 ms longer than the N20 latency [23]. LAI utilised an ISI of 200 ms [25-27,125,126,137]. Ten pulses of each technique, as well as ten pulses of unconditioned TS were collected. When collected on the second day, the m-wave amplitude was matched to that shown on the first day to ensure that potential differences in SAI, LAI, and AF were not due to changes in afferent fibre recruitment.   884.2.4.4 Sensory training  As previously published [136], sensory training consisted of 15 minutes of 80 Hz vibration (0.2-0.5 mm amplitude) applied over the paretic/nondominant APB muscle belly in 2 second trains with an inter-train interval of 2 seconds. The frequency of vibration was changed with 300 ms left in 70% of trains. If a frequency change was detected, individuals were asked to respond by pressing the space bar on a keyboard in front of them. In previous work with young, healthy individuals, vibration frequency was changed from 80 Hz to 65 Hz, 67.5 Hz, 70 Hz, 72.5 Hz, 75 Hz, or 77.5 Hz. In the current work in order to equate difficulty, account for age and/or stroke-related somatosensory decline, and attempt to equalise attentional demand, deviations in frequency occurred in 10 Hz increments ranging from 70 Hz to 20 Hz.  4.2.4.5 Statistical tests In order to assess baseline direct sensorimotor integration, two-way mixed-model ANOVAs were performed, including within-subjects factor VIBRATION (without vibration, with vibration) and between-subjects factor GROUP (healthy older, stroke). This was done for single-pulse MEP amplitude, SICI, and ICF separately, as each is known to represent different circuits within M1. Post-hoc analyses were performed using Fisher’s LSD where appropriate. Previous work examined the impact of chronic stroke on baseline differences in nerve-based measures of indirect sensorimotor integration, and therefore these are not reported here.   To determine whether there was a relationship between measures of sensorimotor integration and somatosensory and motor function, a stepwise linear regression was employed (predictors: age, RC slope, SAI, single-pulse vibration MEP amplitude).  Following the establishment of baseline patterns of sensorimotor integration, vibration-based dependent variables were expressed as a percentage (i.e. (SICI with vibration/SICI without   89vibration)*100). In order to determine the impact of sensory training on sensorimotor integration, two-way mixed-model ANOVAs were performed, including within-subjects factor TIME (pre and post sensory training) and between-subjects factor GROUP (healthy older, stroke) for each dependent variable (MEP, SICI, ICF).   As a secondary analysis, two-way mixed-model ANOVAs, including within-subjects factor TIME (pre and post sensory training) and between-subjects factor GROUP (healthy older, stroke) examined the effect of sensory training on measures quantifying indirect sensorimotor integration (SAI, AF, and LAI) and corticospinal excitability (recruitment curve slope).   For all statistical tests, a significance level of p≤0.05 was used. 4.3 Results  All data were checked for normality and log transformed when abnormal, as indicated by significance in the Shapiro-Wilks test (p<0.001) [128]. At baseline, ICF with vibration was non-normal, while all other variables were normally distributed (ps>0.004). Predictor variables had acceptable collinearity with variance inflation factors below 10 and tolerance levels above 0.1 [129]. Further, data were homoscedastic and had independent residuals as determined by the Durbin Watson test [129]. Due to technical issues, one individual with chronic stroke does not have measures of direct sensorimotor integration on the second day. 4.3.1 Baseline direct sensorimotor integration   Two-way ANOVA results revealed a Group x Vibration interaction for single-pulse MEP amplitude Figure 4.2 (F(1,25)=4.036, p=0.05; η2partial=0.14; Figure 4.3). Post-hoc analysis revealed a reduction in the influence of vibration on single-pulse MEP amplitudes in individuals with chronic stroke compared to healthy older individuals (p=0.017), with no difference in MEP amplitude without vibration between groups (p=0.872).    90 Figure 4.2: Representative trace depicting the influence of vibration on single-pulse MEP amplitudes. Black traces show MEPs without concurrent vibration, and grey traces show MEPs with vibration. Vibration significantly facilitates the response to TMS in older healthy individuals, but not in individuals with chronic stroke.  There was no effect of group or vibration on SICI, nor was there an interaction between the two (F(1,25)=0.016, p=0.901). Though there was a significant interaction effect when examining ICF data (F(1,25)=4.439, p=0.045), post-hoc comparisons did not reach statistical significance. These results can be seen in Figure 4.3.   91 Figure 4.3: Influence of vibration on measures of direct sensorimotor integration. Vibration increases response to single-pulse TMS in older healthy individuals, but not in individuals with chronic stroke (A). There is no impact of vibration on SICI in either group (B). Vibration differentially modulates ICF in individuals with chronic stroke, as compared to older healthy individuals (C). MEP: motor evoked potential, SICI: short-interval intracortical inhibition, ICF: intracortical facilitation, vib: vibration. Asterisks denote statistical significance. Error bars represent standard error of the mean.  4.3.2 Functional relationships Stepwise linear regression analyses revealed an association between sensorimotor integration and motor impairment and motor function in individuals with chronic stroke. This analysis had a reduced sample size as two individuals with measures of direct sensorimotor integration did not have measures of indirect sensorimotor integration. Indirect sensorimotor integration, quantified with SAI, was the only predictor in the significant model for FM (R2=0.363, F(1,11)=5.703, p=0.038, β=-0.603). This relationship was oriented such that decreased SAI related to greater motor impairment, as has been previously reported (Chapter 2). The model that explained the most variance in WMFT rate contained both direct and indirect measures of   92sensorimotor integration. SAI and single-pulse MEP amplitude with vibration were the predictor variables that remained in the model to explain the most variance in WMFT rate (R2=0.677, F(1,11)=9.421, p=0.006, β=-0.629, -0.562). Decreases in SAI related to slower functional rates, as did an increased response to vibration. There were no significant relationships between neurophysiological measures and measures of somatosensory function. 4.3.3 Influence of sensory training on direct sensorimotor integration  Two-way mixed-model ANOVA revealed a main effect of Time (F(1,24)=4.966, p=0.035), with post-hoc analysis showing that sensory training reduced vibration-induced changes in MEP amplitude in both individuals with chronic stroke and older healthy individuals (p=0.039). There was no effect of sensory training on the influence of vibration on SICI (F(1,23)=0.159, p=0.694). One less individual was included in this analysis due to excess noise in the data. Two-way mixed-model ANOVA examining vibration-induced changes in ICF showed a Time x Group interaction effect (F(1,24)=4.112, p=0.05, η2partial=0.15). Post-hoc analysis revealed that there was a difference in the impact of vibration at baseline between the two groups (p=0.045), but not following sensory training (p=0.524). Therefore, in older adults, sensory training reduced the impact of vibration on ICF, whereas this remained constant in individuals with chronic stroke. These results are displayed in Figure 4.4.   93 Figure 4.4: Influence of sensory training on measures of direct sensorimotor integration. Sensory training reduces the impact of vibration n single-pulse MEP amplitude in both groups (A). There is no influence of vibration or training on SICI in individuals with chronic stroke or healthy older individuals (B). Vibration increases ICF prior to sensory training in healthy older individuals, but not following sensory training (C). MEP: motor evoked potential, SICI: short-interval intracortical inhibition, ICF: intracortical facilitation.  Asterisks denote statistical significance. Error bars represent standard error of the mean.   944.3.4 Influence of sensory training on secondary measures Sensory training did not influence any measures of indirect sensorimotor integration in older individuals or individuals with chronic stroke Figure 4.5. Additionally, corticospinal excitability, quantified with recruitment curve slope, did not change following sensory training.  Figure 4.5: Effect of sensory training on measures of indirect sensorimotor integration. There was no influence of vibration on SAI (A), AF (B), or LAI (C) in individuals with chronic stroke or older healthy individuals. SAI: short-latency afferent inhibition, AF: afferent inhibition, LAI: long-latency afferent inhibition, MEP: motor evoked potential. Error bars denote standard error of the mean.  4.3.5 Performance Performance during sensory training was assessed by evaluating the percentage of correct responses to frequency changes in vibration. Individuals with chronic stroke performed at chance (52%) with older healthy individuals performing slightly better (58%). There was not a significant difference in performance levels between the two groups. Anecdotally, though they could feel the vibration, many individuals from both groups could not detect any frequency   95changes, regardless of the amplitude differences. As a result, individuals were instructed to attend to the vibration as best as they could for the duration of the sensory training. 4.4 Discussion   This is the first work to examine neurophysiological measures of direct sensorimotor integration in the chronic phase of stroke recovery, using peripheral vibration to assess the impact of afferent feedback on both single and paired-pulse TMS measures. In contrast with past work showing no stroke-induced change in measures of indirect sensorimotor integration for well-recovered individuals in the chronic phase of recovery (Chapter 3), measures of direct sensorimotor integration are impacted by chronic stroke. More specifically, the influence of vibration on single-pulse MEP amplitude is reduced in individuals with chronic stroke, as compared to healthy older controls. There is also evidence to suggest that there may be an altered response to vibration within interneuronal circuitry responsible for ICF when comparing individuals with chronic stroke and older healthy individuals. The influence of vibration on single-pulse MEP amplitude, in concert with SAI, related to motor function (WMFT rate), such that increased amplitude corresponded with a slower functional rate. Additionally, we examined whether a single-bout of sensory training could alter neurophysiological measures of direct and indirect sensorimotor integration. Following sensory training, the facilitatory effect of vibration on single-pulse MEP amplitudes was reduced in both individuals with chronic stroke and our sample of older healthy controls; however, the reduction in ICF with vibration shown in healthy individuals was not shown in individuals with chronic stroke. Sensory training did not impact measures of indirect sensorimotor integration in either group, suggesting it has a differential influence on separate neurological pathways.    964.4.1 Baseline sensorimotor integration  Direct sensorimotor integration is altered in individuals with chronic stroke, compared to older healthy controls. The influence of vibration on single-pulse MEP amplitude was reduced in individuals with chronic stroke as compared to healthy controls, and there was a similar result suggesting less of a vibratory influence on ICF. Inhibitory and facilitatory measures of indirect sensorimotor integration did not show differences between individuals with stroke and older healthy controls.   As previously established, peripheral nerve stimulation and peripheral vibration likely ascend to the motor cortex in slightly different ways, thus providing unique information about sensorimotor integration [32,33,39]. Muscle belly vibration activates 1a afferent fibres, which can be incorporated directly into M1 (Brodmann area 4), as well as S1 (Brodmann area 3a) [32,33]. In contrast, median nerve stimulation may be predominantly incorporated into the primary somatosensory cortex areas 3b and 1, but not directly into M1 [151]. Therefore, it is interesting to consider that despite nerve-based measures showing no stroke-related changes, those tested here relying on peripheral vibration to probe the somatosensory system, do appear altered post-stroke. In our sample of individuals with chronic stroke, there was a reduction in the influence of vibration on MEP amplitude derived from single-pulse TMS, and an interaction suggesting an altered response of ICF to vibration in individuals with chronic stroke. This differential impact of stroke on various measures of sensorimotor integration may provide further evidence for the separate underlying mechanisms associated with these types of measures.   The effect of vibration on single-pulse TMS measures is partially driven by spinal mechanisms; yet a cortical component is also likely as evidenced by a differential effect of vibration on electrical and magnetic evoked potentials [40]. While we do not have spinal   97measures to assess this claim, previous work showed that there is typically an increase in spinal reflex amplitude in the paretic limb of individuals with chronic upper limb impairment post-stroke [152]. In the current work, we saw a reduction in the response to vibration, which is the opposite of what one would expect if solely driven by changes in spinal pathways. Therefore, it is possible that the increased amplitude one would expect from the spinal component is outweighed by a reduction in cortical receptiveness to incoming afferent information that negates the spinal change.  To our knowledge, there is one past study examining the influence of vibration on MEP amplitude in individuals with stroke. Despite slight methodological differences in vibration frequency, and stage of stroke recovery, this past work provides relevant insight into motor cortical response to vibration post-stroke. Initially, in the acute stage of stroke recovery, individuals show an increase in vibration-induced facilitation as compared to age-matched controls, but by the sub-acute stage of stroke recovery vibration-induced facilitation has returned to more age-normative values [145]. When contextualising this finding with the current results, it seems that the trend towards a reduction in the influence of vibration on corticospinal activity continues into the chronic phase of stroke recovery where individuals present with less vibration-induced facilitation than older healthy individuals.  As has been previously reported, the current work shows that there is a relationship between the indirect measure of sensorimotor integration, SAI, and motor impairment quantified with the FM (Research Chapter 1). Interestingly, while measures of direct sensorimotor integration did not help to explain variance in FM scores, single-pulse response to vibration explained unique variance in motor function, beyond that addressed with SAI. Potentially, the tasks used in the WMFT are more complex and demanding than those in the FM assessment, and   98thus a reliance on both pathways of sensorimotor integration is required to complete the tasks. In contrast to what was shown with SAI, increased response to vibration was related to worse motor function. Given the known spinal contributions to vibration-induced facilitation [153,154], this is in line with past research. Individuals in the chronic phase of stroke recovery show increased spinal excitability, indicated by H-reflex amplitudes, in the paretic compared to non-paretic limb [152]. Additionally, individuals who present with worse motor deficits also have increased spinal excitability [155]. Therefore, this increase in spinal excitability may contribute to the larger response to vibration and explain the relationship with motor function. Future work parsing out the specific contributions of spinal and cortical mechanisms would provide important insight into the stroke-related changes in sensorimotor integration documented in the current work. Although there were no statistically significant post-hoc comparisons in our follow-up analysis, there was an interaction indicating that vibration may have a differential effect on cortical paired-pulse measures of sensorimotor integration following stroke. As paired-pulse measures are known to probe cortical circuitry, this may indicate that interneuronal circuits in M1 responsible for ICF are less responsive to incoming afferent information in the chronic phase of stroke recovery. Though we did not quantify event-related potentials arising from the peripheral stimulation, it is possible that these would be smaller than those seen in controls, and the reduction in afferent feedback reaching the cortical level is responsible for these differences. Alternatively, it is possible that our results are more driven by the receptiveness of motor cortical circuits to this information. These hypotheses cannot be tested with the current experimental design, but warrant further investigation. It should also be noted that the lack of significant post-hoc comparisons, despite a significant interaction, when examining the impact of vibration on ICF likely relates to high   99variability within response to vibration in our group of older healthy individuals. When these values were translated into percent differences between ICF with and without vibration, as was completed in order to investigate the impact of sensory training on direct sensorimotor integration, baseline differences in ICF were apparent. This may result from a reduced variability when data are considered in this form, or from encapsulating the difference between ICF with and without vibration more directly. Past work has shown that amongst healthy individuals, there is a high variability in response to vibration, leading to some work to stratify data based on responders and non-responders [96]; as such, this finding may be in line with previous work.  4.4.2 Influence of sensory training on sensorimotor integration   Sensory training altered measures of direct sensorimotor integration, but had no influence on measures probing the more indirect route of sensorimotor integration. A 15-minute block of sensory training reduced vibration-induced facilitation of MEP amplitudes arising from single-pulse TMS, similar to patterns seen in older healthy controls. However, individuals with chronic stroke did not show a decrease in ICF with vibration, as was shown in healthy controls. Cumulatively, these results suggest that sensory training may alter specific measures of sensorimotor integration in individuals with chronic stroke, but the impact of training is different between these individuals and healthy older adults.  Previously, the effect of vibration-based interventions on neurophysiology and behaviour has been tested in individuals with various movement disorders with differing degrees of success. In the current work, we utilised an intervention that, in the past, was shown to alter sensorimotor integration, indexed by SICI paired with vibration in a population of young healthy individuals [97,136]. Examining baseline patterns of direct sensorimotor integration may provide insight into the response to sensory training. Baseline methods examined the cortical influence of   100a single burst of vibration; logically, given that sensory training comprised of 225 applications of this vibration, it would result in repeated activation of circuits indexed at baseline. Without a baseline influence of vibration, sensory training would likely not have an effect. The influence of vibration on single-pulse MEPs showed group differences at baseline, with afferent feedback having less of an impact in the stroke group; however, on average, individuals with chronic stroke still demonstrated an increase in MEP amplitude when concurrent vibration was applied (127% greater than without vibration). Therefore, it seems that the influence of vibration, though reduced, is still present in in individuals with chronic stroke. Repeated activation of this circuit with the sensory training paradigm then should induce a similar response. This is in line with the present results; both individuals with chronic stroke and older healthy individuals show a reduction in the vibration-induced change in MEP amplitude following sensory training.   In contrast, the response of ICF was not altered following sensory training in individuals with chronic stroke, differing from the group of healthy older controls. Again, looking at baseline trends contextualises this result. Prior to sensory training, there was a non-significant trend towards a reduced influence of vibration on ICF in individuals with chronic stroke. Looking at raw values reveals that individuals with chronic stroke did not show any increase in ICF with concurrent vibration (89% of ICF without vibration), whereas older healthy controls showed an increased facilitation when vibration was applied simultaneously (201%). These results indicate that vibration seems to be differentially impacting specific ICF circuitry within the motor cortex in individuals with chronic stroke. Given that there is not an influence on these circuits with a single application of vibration, it follows that sensory training would not induce an effect in individuals with chronic stroke, but it would in healthy individuals who have a baseline response.    101The decrease in response to vibration for both groups for single pulse, and older healthy controls for ICF measures, following sensory training is in contrast to past work in younger healthy individuals. Previous work showed an increase in vibration-induced disinhibition of SICI, with no changes to the impact of vibration on single-pulse MEP amplitudes or ICF [97]. As discussed in Chapter 3, the importance of attention has been determined to be essential to the induction of neurophysiological change [89,95]; if individuals receive the same vibration pattern for 15 minutes, but are not instructed to attend to that stimulation and respond when frequency changes are detected, vibration no longer influences the response to single or paired-pulse TMS measures [95]. Further, neural recordings show diminishing responses in the processing of repetitive, unattended stimuli [139]. Despite our instructions for individuals to attend to the frequency of vibration throughout sensory training, our results follow this pattern that may arise simply from repetition. Anecdotally, individuals reported finding the task difficult and this is reflected in performance values. We cannot rule out the fact then that attention waned throughout this period, and without attention-based facilitation of this sensory information, repetition of vibration alone is not enough to induce change on a neurophysiological level.   In addition to attention, the modality of somatosensory stimulation has been shown to be important for inducing neurophysiological change. Therefore, our results showing changes in direct but not indirect sensorimotor integration are in line with previous research. If the influence of sensory training results from repetitive stimulation of the same synaptic pathways, nerve-based measures would not be expected to change. As previously mentioned, afferent feedback arising from peripheral vibration and nerve stimulation ascend to the cortex through different neuroanatomical pathways. Strengthening synaptic connections or repetitive activation resulting   102from vibration, then would not lead to increased excitability or strengthened connections that would be probed with measures of indirect sensorimotor integration.  Finally, although this sensory training paradigm has not been used previously in individuals with chronic stroke, and the specific sensorimotor integration measures have not been tested, other vibration-based interventions have been shown to lead to behavioural improvements with little understanding of the underlying neurophysiology. For example, a single-session of 5 minutes of 60 Hz vibration led to improvements on the Box to Blocks test that lasted for 15 minutes after vibration in individuals with post-stroke chronic upper limb impairment [72]. The effectiveness of vibration is evident after multiple sessions of pairing repetitive muscle vibration with physiotherapy or some form of motor practice [156]. Further, in the leg, chronic Achilles tendon vibration has led to strength improvements, without changing measures of neurophysiology, yet the baseline measures of neurophysiology were indicative of the degree of response to vibration, suggesting a link between physiology and function [157]. Future work should expand the neurophysiology tested to broaden our understanding of the physiological underpinnings of functional improvement with vibration-based interventions. Additionally, capturing potential neurophysiological change occurring across multiple sessions in an attempt to quantify potential cumulative effects will be important. 4.4.3 Limitations  Sensory training has often been shown to change not only neurophysiological measures in the vibrated muscle, but the organization of sensorimotor integration in the surrounding muscles. The current work only explored the response in a single muscle, and therefore cannot exclude the possibility that sensorimotor patterns for cortical representations of other muscles are not being altered with vibration. Additionally, when examining paired-pulse measures, we did   103not adjust the TS amplitude to be 1 mV when vibration was concurrently applied. Past work that has adjusted for this, in addition to using the approach of the current study has shown no difference between the two measures and collapsed across those conditions [41,97,136]. However, we cannot rule out the possibility that this may influence the results. Future work could address these limitations and further our understanding of the influence of chronic stroke and aging on sensorimotor integration by studying multiple hand muscles and various TS amplitudes. 4.5 Conclusions  In conclusion, vibration applied over the muscle belly of a focal hand muscle appears to have a reduced impact on measures of corticospinal excitability and interneuronal circuitry within the primary motor cortex in individuals with chronic stroke. Response to vibration within corticospinal circuitry is behaviourally relevant and related to the degree of motor function in chronic stroke. Additionally, a 15-minute bout of sensory training reduced vibration-induced facilitation of corticospinal excitability in both older healthy individuals and individuals with chronic stroke, though it did not alter more cortically-based circuits probed with paired pulse TMS in individuals with chronic stroke. Therefore, we suggest that vibration-based pathways of sensorimotor integration are impacted by chronic stroke, and that an intervention designed to alter sensorimotor integration differentially influences cortical neurophysiology, as compared to healthy older adults.     104Chapter 5: The reliability of commonly used electrophysiology measures 5.1 Introduction Electrophysiological measures can improve understanding of brain function both in healthy individuals and in the context of a disease, as has been discussed in the previous chapters of this thesis. Numerous electrophysiological experimental paradigms probe the function of the cortical gray and white matter connections [158]; some paradigms examine brain function at the time of stimulation, while others aim to modulate brain function so that the effects outlast the time of stimulation thus reflecting plasticity [159]. Commonly used techniques examining brain function include somatosensory-evoked potentials (SEP) to examine primary somatosensory cortical (S1) excitability and sensory afferent connections following peripheral nerve stimulation, and transcranial magnetic stimulation (TMS) in combination with electromyography (EMG) to explore excitability in the corticospinal tract and local circuitry within the primary motor cortex (M1). Additionally, the combination of peripheral nerve stimulation and TMS can be used to test cortical circuitry involved in sensorimotor integration [126].  Given the amount of information that can be extracted from these measures and their frequent use, it is essential to know more about their inherent reliability. In studies assessing baseline group differences, experimenters must be confident that quantified differences are not simply the result of inherent variation between individuals [111,112]. In clinical trials or intervention-based studies that are more dependent on within-individual changes, as opposed to group differences, it is essential that measurement error is not responsible for differences in observed scores [111,113,114]. Both of these issues can be addressed by understanding the reliability of measures. Finally, when designing multi-centre studies the susceptibility to differences between sites in the interpretation of the protocol for electrophysiological measures   105can introduce further variability in the data. It is therefore critical to identify these sources of added variability as well.  To date, there is no consensus on the reliability of electrophysiological measures; however, most investigations suffer from small sample sizes, include few measures, and do not consistently use the same methodological approach [115]. Recent work outlined these shortcomings, and provided a more comprehensive evaluation of the reliability of TMS measures in healthy older individuals as well as individuals with stroke [116]. This work should be expanded upon to include measures of somatosensory excitability, additional TMS measures, and measures of sensorimotor integration.  The current study utilised a large sample of electrophysiological data in healthy individuals across 24-months to address three main aims. First, we examined feasibility of each electrophysiological measure by calculating attrition rates across time. Second, we determined the reliability of different electrophysiological measures and quantified how reliability influences statistical power when using electrophysiological measures. Thirdly, we determined which electrophysiological measures may be susceptible to methodological differences by assessing between-site differences. Broadly, we hypothesised that measures of threshold and latency would be the most reliable and least susceptible to methodological differences between study sites.  5.2 Methods 5.2.1 Participants This study used electrophysiology data from 112 control participants (67 females) collected in individuals enrolled in the Track-On HD study [160,161].  Participants were assessed at baseline, as well as at 12 and 24-month follow-ups, at four study sites: London (29 participants at baseline; 26% of total 112), Paris (29 at baseline; 26%),   106Leiden (28 at baseline; 25%), and Vancouver (26 at baseline; 23%). Local ethics committees approved the study, and written informed consent was obtained from each participant. The Leiden site only collected electroencephalography (EEG) measures; therefore, the long-latency reflexes (LLRs) and TMS-based measures have a total of 84 control participants’ data. 5.2.2 Electrophysiology For all data collection, participants were seated in a comfortable chair and asked to relax as much as possible, unless instructed otherwise. All measures were collected from the dominant hemisphere and hand, assessed with the Edinburgh Handedness Questionnaire [162].  5.2.2.1 Electroencephalography SEPs were recorded following median nerve stimulation (pulse width 200 s, square wave pulse, cathode distal, anode proximal) with surface electrodes using routine techniques [23]. SEPs were recorded with a silver/silver-chloride disk electrode over the somatosensory cortex (2 cm posterior of C3 in the international EEG 10-20 system) referenced against Fz. Briefly, stimulation at 3 Hz was delivered at three intensities: sensory threshold, defined as the minimum intensity at which participants could perceive the stimulation at the wrist; motor threshold, defined as the minimum intensity required to evoke a visible twitch in the target muscle; and at 150% of the motor threshold. Recordings from 300 stimuli were collected. At three sites (London, Paris, Vancouver), surface EMG were recorded from the right abductor pollicis brevis (APB) muscle using silver/silver-chloride disc surface electrodes (1 cm diameter) in a belly tendon montage. The EMG signal was amplified and analogue filtered (30 Hz to 1 kHz) with a Digitimer D360 amplifier (Digitimer Ltd., Welwyn Garden City, UK) in London and Paris, or Powerlab 4/30 EMG System (AD Instruments, Colorado Springs, CO) in Vancouver. Data were digitised (sampling rate 4 kHz) for offline analysis using Signal software   107(Cambridge Electronic Devices, Cambridge, UK) in London and Paris, or LabChart (AD Instruments, Colorado Springs, CO) in Vancouver. In order to analyse SEP data, an average trace for each stimulation intensity was produced to extract the N20 latency and N20/P25 amplitude. The N20 component was identified as the first negative peak in a time window of 15-25 ms post-stimulus; N20 latency was defined as the time from stimulation to this peak. Latency was determined from the 150% of motor threshold trace; if no peak could be detected in this trace, MT was used. N20/P25 amplitude was calculated as the peak-to-peak amplitude between the N20 and the following P25.  5.2.2.2 Long-latency reflexes Long-latency reflexes were collected using standard procedures [163]. Three hundred stimuli were delivered over the median nerve at the wrist as individuals maintained an APB contraction of 20-30% of their maximal voluntary contraction (MVC). To activate the APB, individuals were instructed to abduct their thumbs against a force transducer while monitoring visual feedback to ensure consistency. EMG was collected as described above. Average traces were used to determine LLR2 amplitudes, as well as the latencies of both LLR1 and LLR2. LLR1 was defined as the first visible deflection from baseline between 35-45 ms post-stimulus, while LLR2 was identified in a time window of 45-55 ms. LLR2 amplitude was defined as the peak-to-peak amplitude between LLR2 and the following peak. Cortical relay time (CRT) was calculated by subtracting SEP and MEP latencies from the LLR2 latency to provide an estimate of transmission time within the cortex. 5.2.2.3 Transcranial magnetic stimulation  TMS was performed as previously described using established techniques [23,164]. At all sites single pulse TMS was delivered using a monophasic figure-of-eight shaped coil (Magstim   10870 mm P/N 9790, Magstim Co., UK) connected to a Magstim 2002 stimulator (Magstim Co., UK). Stimuli were given with a random inter-stimulus interval (ISI) of 4-5 seconds. The coil was held in such a way to induce a posterior-anterior flow with the coil handle positioned at an angle of 45 pointing backwards. The APB ‘hot-spot’ was located and both resting and active motor thresholds (RMT and AMT) were determined as described [123]. The optimal spot for APB activation was marked with a felt pen (London) or coordinates recorded using neuronavigation software (Vancouver; Brainsight™, Rogue Research Inc., Montreal, QC, Canada; Paris; N eXimia 2.2.0, Nextim Ltd, Helsinki, Finland). Following threshold determination, TMS was used to collect input/output curves at rest (110%, 130%, 150% RMT) or with pre-activation (125%, 150%, 175% AMT) as described including cortical silent period determination [12,137].  MEP latency was defined as the time between the stimulus and MEP onset, while the CSP was defined as the time from the beginning of the MEP to the return of voluntary EMG activity [165]. This was determined from the MEP evoked from the highest intensity of stimulation for each individual. To quantify the size of the MEP, both peak-to-peak amplitude and curve area were calculated. Curve area was determined from the unrectified MEP using each waveform’s absolute amplitude multiplied by the time between samples on the channel. To investigate sensorimotor integration, median nerve stimulation was paired with TMS at various ISIs using short-latency afferent inhibition (SAI) and afferent facilitation (AF) as described [27]. These measures of sensorimotor integration (SAI, AF) were not collected at the third visit as a result of an updated protocol for the Track-On HD study.  5.2.2.4 Statistical analysis In order to measure the reliability of different neurophysiological measures, we first calculated the average measures two-way random-effects intra-class correlation coefficient   109(ICC), hereafter written as ICC(2,k) (for more details about the statistical rationale see the appendix). In the random-effects ICC, both people and observations are treated as random effects (i.e., we assume that both people and time points are samples from a larger population [166]). The three different time points were treated as a sample of possible time points at which people could have been measured, and data were filtered prior to analysis to remove participants with missing data. The ICC(2,k) can be interpreted as the ratio of true variance to total variance for k measures (i.e., across all three time points, [167]). In our analyses, we selected ICC(2,k) ≥0.80 as cut-point indicating a relatively stable and reliable measure with relatively little variation within a person over time compared to the individual differences between people [168].   We also calculated a single measures two-way random-effects ICC, referred to as ICC(2,1). The ICC(2,1) reflects the average ratio of true variance to total variance captured by a single measure (i.e., the ICC(2,1) is equivalent to the average of the off-diagonal of a correlation between the time points). Presenting ICC(2,1) as a compliment to ICC(2,k) is important because ICC(2,k) is sensitive to the number of measurements, whereas ICC(2,1) is not. That is, if the ratio of r2tx=var(T)/var(X)=0.25 in the population, ICC(2,1) will approximate 0.25 (especially in large sample sizes) regardless of the number of observations taken, whereas ICC(2,k) will increase to very high levels as k increases. We next used the ICC(2,1) observed in the current data to explore the statistical power available using different electrophysiological measures. To this end we constructed statistical simulations for three different study designs: 1) an independent samples t-test, 2) a paired samples t-test, and 3) the within-between interaction in a mixed-factorial ANOVA (i.e., a Group by Test interaction in a classic randomised controlled trial). The details of the statistical simulations and the code for running them are provided online   110(https://github.com/keithlohse/power_reliability) and these results were corroborated with power-calculation software (G*Power 3.1.9.2; [169]). These simulations were run at nine different sample sizes n/group = [10, 20, 30, 40, 50, 60, 100, 200, 300] and two different effect-sizes d = [0.5, 0.8], which correspond to traditionally moderate and large effects [127]. The code for these simulations is adaptable, however, so researchers could always adapt the code for their own power analyses.   Finally, we were interested in electrophysiological measures that were not statistically different across study sites. We conducted a series of MANOVAs in which the three different time points were treated as dependent measures and study site was treated as a between-subjects factor. For TMS latency measures, we also included arm length as a covariate. The Pillai-Bartlett Trace (denoted V) was used in the determination of statistical significance. In the event of a statistically significant effect of study site in the MANOVA, we conducted follow-up univariate ANOVAs at each time point, followed by pairwise comparisons of the study-sites (in the event the univariate ANOVA was statistically significant).   All analyses were conducted using SPSS v23.0 (IBM, Armonk, New York). All descriptive statistics are reported as mean (SD) unless otherwise indicated. A graphical depiction of our approach can be found in Figure 5.1.    111 Figure 5.1 Conceptual outline for including measures in study design. The reliability of each potential electrophysiological measure should be determined unless it is known. A power-analysis can then be used to assess the sample size required to accurately detect change based on the individualised reliability of the measure of interest. For multi-site studies, a further step should be undertaken to assess whether the measure is susceptible to site-differences before including it in the study design.  5.3 Results  5.3.1 Cohort At visits one to three, individuals had an average age of 48.1 years (SD: 10.7), 49.4 years (10.5), and 50.6 years (10.4), respectively; 67 participants were female (60%). 103 individuals were right-handed, 6 were left-landed, and 3 were ambidextrous.    112There was a significant difference in arm length across study sites, and thus arm length was controlled for in our MANOVAs for latency-based measures. Arm length in Leiden (75.98cm (5.08)), was significantly larger than in London (72.10cm (5.7)), and in Paris (71.35cm (8.05)), but not in Vancouver (74.22cm (5.23)) (F(3,100)=3.16, p=0.03).   All electrophysiological measures were well tolerated and attrition rates were low.  Attrition values for each study site as well as total participant numbers can be found in Table 5.1. As the Leiden site exclusively collected EEG measures, there are different values for those outcomes as compared to all the other electrophysiological measures. At visit two, retention of participants was 94% for EEG measures, and 92% for all other measures. At the third visit, EEG measures were assessed on 90% of the original sample, and TMS measures were assessed on 89% of the original sample. Between the second and third visits, attrition was low at 4% and 3% for EEG and TMS measures, respectively.      113 Site Time 1 Time 2 Time 3 London 29 28 (97) 25 (86, 89) Paris 29 29 (100) 29 (100, 100) Vancouver 26 20 (77) 21 (81, 105) Leiden Total 28 112 28 (100) 105 (94) 26 (93, 93) 101 (90, 96) Table 5.1: Attrition rates. The number of participants at each site and totaled across sites for each time point for the electrophysiological measures. The number in brackets is the percentage of participants returning from previous time points, such that the first number in brackets is in comparison to Time 1 and the second corresponds to Time 2. Note that Leiden only collected somatosensory evoked potentials.  5.3.2 Reliability and power simulations We first examined reliability using data from the same participants across several visits. Several measures met the ICC(2,k) cut-off  ≥ 0.80 indicating high reliability (Figure 5.2; Table 5.2). These were SEPs at motor threshold (0.91) and 150% of motor threshold (0.91), N20 latency (0.90), the latency of both LLR1 and LLR2 (0.97, 0.98), MEP amplitude at 150% of RMT (0.81), RMT and AMT (0.89 and 0.81, respectively), and both resting and active MEP latency (0.92 and 0.89 respectively). Other measures met the ICC(2,k) cut-off  ≥0.50 of moderate reliability while a number of measures had  low reliability (ICC(2,k) cut-off <0.50) (Table 5.2, Table A.1 for low reliability measures).    114 Figure 5.2: Reliability values for each electrophysiological measure. High reliability indicates ICC values greater than 0.8, moderate reliability indicates ICC values greater than 0.5, but less than 0.8, and low reliability indicates ICC values less than 0.5.  115 Measure Time1 Mean (SD) Time2 Mean (SD) Time3 Mean (SD) Exclusions ICC (2,k) ICC (2,1) High Reliability       RMT (% MSO) 45.17 (8.74) 45.96 (10.30) 46.25 (9.22) 31 (37) 0.89    0.73 AMT (% MSO) 35.73 (9.27) 35.35 (7.17) 36.13 (6.82) 30 (36) 0.81    0.59 MEP Lat Rest (ms) 22.40 (1.73) 21.77 (1.75) 22.18 (1.75) 33 (39) 0.92 0.80 MEP Lat Active (ms) 20.51 (1.85) 20.98 (2.05) 20.98 (1.73) 37 (44) 0.89 0.73 Amp 150% RMT (mV) 2.02 (1.84) 2.38 (2.30) 2.51 (2.13) 41 (49) 0.81 0.59 SEP Amp MT (µV) 1.81 (1.37) 2.01 (1.53) 2.12 (1.74) 48 (43) 0.91 0.78 SEP Amp150 (µV) 2.27 (1.56) 2.59 (1.81) 2.72 (2.21) 54 (48) 0.91 0.76 N20 Latency (ms) 19.78 (1.20) 19.6 (1.27) 20.06 (1.39) 44 (39) 0.90 0.76 LLR1 Latency (ms) 38.51 (3.31) 38.56 (3.03) 38.21 (2.95) 54 (64) 0.97 0.91 LLR2 Latency (ms) 49.61 (3.44) 49.29 (3.24) 49.52 (3.00) 54 (64) 0.98 0.93 Mod. Reliability        SEP Amp ST (µV) 0.86 (0.92) 0.69 (0.60) 0.81 (0.81) 53 (47) 0.58 0.32 LLR2 Amp (mV) 0.10 (0.07) 0.10 (0.07) 0.09 (0.08) 54 (64) 0.62 0.35 CRT (ms) 7.99 (2.87) 8.11 (2.51) 8.07 (2.62) 61 (73) 0.61 0.34 Amp 130%RMT (mV) 1.23 (1.23) 1.46 (1.47) 1.44 (1.33) 37 (44) 0.70 0.44 Amp 125%AMT (mV) 1.78 (1.36) 1.61 (1.12) 1.62 (1.39) 38 (45) 0.54 0.28 Amp 150%AMT (mV) 3.36 (1.95) 3.54 (2.27) 3.74 (2.34) 41 (49) 0.68 0.41 Amp 175%AMT (mV) 3.96 (2.03) 4.51 (2.31) 4.92 (3.42) 46 (55) 0.64 0.38 Area 130%RMT (mV ms) 8.29 (8.53) 5.97 (5.42) 5.09 (4.33) 41 (49) 0.61 0.34 Area 150%RMT  (mV ms) 13.39 (12.46) 9.85 (9.09) 9.33 (7.40) 43 (51) 0.53 0.27 SPD 175%AMT (ms) 166.56 (45.26) 159.31 (36.9) 151.75 (37.4) 48 (57) 0.53 0.27 SAI N22 Amp (mV) 0.78 (0.65) 0.81 (0.72)  46 (55) 0.65 0.48 SAI N24 Amp (mV) 0.85 (0.82) 0.84 (0.69)  45 (54) 0.67 0.50 AF N32 Amp (mV) 1.39 (1.26) 1.28 (1.17) 45 (54)  0.62 0.45 AF N32 Area  (mV ms) 130.83 (70.50) 135.07 (69.48)  40 (48) 0.55 0.38 Table 5.2: ICC values for electrophysiological measures with high and moderate levels of reliability. Mean (standard deviation) values for each dependent measure at each time-point, exclusions, ICC(2,k), and ICC(2,1). ‘Exclusions’ refers to participants who had to be excluded due to missing data (shown as count (% of total n)). High reliability denotes an ICC(2,k) ≥ 0.8. Moderate reliability denotes an ICC(2,k) between 0.5 and 0.8.  Values for dependent measures with low reliability can be found in the appendix.   116We next explored the hypothetical statistical power available using different electrophysiological measures. For a paired-samples t-test, for instance, any non-zero change is always going to be statistically significant when there is no measurement error. As measurement error is added, r2TX decreases, and statistical power decreases (Table 5.3, Table 5.4, Table 5.5). Consider then a study design using a paired-samples t-test to detect a medium-sized effect where the change from pre-test to post-test is one-half of the pooled standard deviation, d = 0.5. If the primary outcome were RMT then the ICC(2,1)=0.73. As ICC(2,1) is an estimate of the true r2TX, we can make a conservative estimate that r2TX=0.64 and having 40 total participants would yield ~81% power (Table 5.4). Conversely, if the primary outcome were SEP at sensory threshold, the ICC(2,1)=0.32.  Looking at Table 5.4, the closest r2TX is 0.36 and now having 40 participants would only yield ~36% power. Indeed, we would need to increase our sample size to 100 to achieve even 74% statistical power due to the lower reliability of the SEP at sensory threshold.  Consider now the same measures when examining an interaction effect in a 2x2 mixed-factorial ANOVA, with a medium-sized effect (d=0.5). For RMT, with an estimated r2TX of 0.64, having a sample size of 60 would yield 69% power, while increasing the sample size to 100 would yield 89% power (Table 5.5). In contrast, SEP at sensory threshold with an estimated r2TX of 0.36 would require a sample size of 200 in order to achieve 78% power, with a sample size of 40 only providing 23% power (Table 5.5).          117Simulated independent t-test results, Cohen’s d = 0.5.  Measurement error in dependent variable n/group T  (no error) X r2TX=0.81 X r2TX=0.64 X r2TX=0.49 X r2TX=0.36 X r2TX=0.25 10 0.18 0.16 0.13 0.11 0.09 0.08 20 0.32 0.27 0.22 0.19 0.15 0.12 30 0.47 0.39 0.31 0.27 0.20 0.16 40 0.58 0.49 0.39 0.34 0.25 0.19 50 0.69 0.58 0.48 0.41 0.31 0.23 60 0.77 0.66 0.56 0.47 0.37 0.27 100 0.94 0.87 0.76 0.69 0.54 0.41 200 0.99 0.99 0.97 0.94 0.83 0.71 300 1.00 0.99 0.99 0.99 0.95 0.87        Simulated independent t-test results, Cohen’s d = 0.8.  Measurement error in dependent variable n/group T  (no error) X r2TX=0.81 X r2TX=0.64 X r2TX=0.49 X r2TX=0.36 X r2TX=0.25 10 0.40 0.33 0.27 0.22 0.19 0.13 20 0.68 0.61 0.50 0.42 0.34 0.23 30 0.86 0.79 0.68 0.60 0.47 0.33 40 0.94 0.89 0.80 0.71 0.59 0.42 50 0.98 0.94 0.88 0.81 0.69 0.52 60 0.99 0.98 0.93 0.87 0.77 0.58 100 1.00 1.00 0.99 0.98 0.94 0.79 200 1.00 1.00 1.00 1.00 1.00 0.98 300 1.00 1.00 1.00 1.00 1.00 1.00 Table 5.3: Simulated independent t-test results. Statistical power obtained as a function of sample size and reliability for independent-samples t-tests. Note that cells contain the statistical power (power of ≥0.8 is highlighted) observed across 10,000 simulated t-tests with a 1:1 group allocation ratio and α = 0.05 (i.e., % of significant results out of 10,000). Columns reflect decreasing reliability, X at different levels of r2TX’, from the original “true” outcome, T. These data were simulated based on a classical test theory model of ࢄ࢏࢐ ൌ ࢀ࢏ ൅ ࢿ࢏࢐ where T is a standard normal variable representing the true score and ε is a random normal variable representing measurement error. The variance of ε is adjusted to produce the desired correlation between T and X in the population.     118 Simulated paired t-test results, Cohen’s d = 0.5.  Measurement error in dependent variable Total N = Tpre/post  (no error) Xpre/post r2TX=0.81 Xpre/post r2TX=0.64 Xpre/post r2TX=0.49 Xpre/post r2TX=0.36 Xpre/post r2TX=0.25 10 1.00 0.53 0.26 0.16 0.11 0.09 20 1.00 0.86 0.49 0.32 0.20 0.14 30 1.00 0.97 0.68 0.46 0.28 0.19 40 1.00 0.99 0.81 0.58 0.36 0.24 50 1.00 1.00 0.88 0.69 0.44 0.29 60 1.00 1.00 0.94 0.77 0.51 0.33 100 1.00 1.00 1.00 0.93 0.74 0.51 200 1.00 1.00 1.00 1.00 0.96 0.82 300 1.00 1.00 1.00 1.00 0.99 0.94        Simulated paired t-test results, Cohen’s d = 0.8.  Measurement error in dependent variable Total N Tpre/post  (no error) Xpre/post r2TX=0.81 Xpre/post r2TX=0.64 Xpre/post r2TX=0.49 Xpre/post r2TX=0.36 Xpre/post r2TX=0.25 10 1.00 0.90 0.55 0.35 0.22 0.15 20 1.00 0.99 0.89 0.66 0.43 0.28 30 1.00 0.99 0.98 0.84 0.60 0.41 40 1.00 1.00 1.00 0.93 0.74 0.51 50 1.00 1.00 1.00 0.97 0.83 0.62 60 1.00 1.00 1.00 0.99 0.90 0.69 100 1.00 1.00 1.00 1.00 0.98 0.90 200 1.00 1.00 1.00 1.00 1.00 0.99 300 1.00 1.00 1.00 1.00 1.00 1.00 Table 5.4: Simulated paired t-test results. Statistical power obtained as a function of sample size and reliability for paired-samples t-tests. Note that cells contain the statistical power (power of ≥0.8 is highlighted) observed across 10,000 simulated paired-samples t-tests with α = 0.05 (i.e., % of significant results out of 10,000). Columns reflect decreasing reliability, X at different levels of r2TX’, from the original “true” outcome, T. These data were simulated based on a classical test theory model of ࢄ࢏࢐ ൌ ࢀ࢏ ൅ ࢿ࢏࢐ where T is a standard normal variable representing the true score and ε is a random normal variable representing measurement error. The variance of ε is adjusted to produce the desired correlation between T and X in the population. Post-test T was a linear transformation of pre-test, Tpost = Tpre+d, and independent ε were added to the pre- and post-test to create the Xpre/post variables.     119Simulated interaction results, post-test Cohen’s d = 0.5 (no pre-test difference).  Measurement error in dependent variable n/group = Tpre/post  (no error) Xpre/post r2TX=0.81 Xpre/post r2TX=0.64 Xpre/post r2TX=0.49 Xpre/post r2TX=0.36 Xpre/post r2TX=0.25 10 1.00 0.34 0.16 0.12 0.09 0.08 20 1.00 0.62 0.29 0.19 0.13 0.10 30 1.00 0.79 0.40 0.28 0.17 0.12 40 1.00 0.90 0.51 0.34 0.23 0.15 50 1.00 0.95 0.61 0.42 0.26 0.18 60 1.00 0.98 0.69 0.48 0.32 0.21 100 1.00 1.00 0.89 0.71 0.48 0.32 200 1.00 1.00 0.99 0.94 0.78 0.55 300 1.00 1.00 1.00 0.99 0.91 0.73        Simulated interaction results, post-test Cohen’s d = 0.8 (no pre-test difference).  Measurement error in dependent variable n/group = Tpre/post  (no error) Xpre/post r2TX=0.81 Xpre/post r2TX=0.64 Xpre/post r2TX=0.49 Xpre/post r2TX=0.36 Xpre/post r2TX=0.25 10 1.00 0.70 0.35 0.22 0.14 0.11 20 1.00 0.95 0.62 0.40 0.27 0.18 30 1.00 0.99 0.80 0.57 0.37 0.24 40 1.00 0.99 0.90 0.69 0.47 0.31 50 1.00 1.00 0.95 0.79 0.57 0.38 60 1.00 1.00 0.98 0.86 0.64 0.44 100 1.00 1.00 0.99 0.98 0.85 0.65 200 1.00 1.00 1.00 0.99 0.99 0.92 300 1.00 1.00 1.00 1.00 1.00 0.98 Table 5.5: Simulated interaction results. Statistical power obtained as a function of sample size and reliability for the interaction term in a 2 (Group) by 2 (Time) mixed-factorial ANOVA. Note that cells contain the statistical power (power of ≥0.8 is highlighted) observed across 10,000 simulated interactions with α = 0.05 (i.e., % of significant results out of 10,000). Columns reflect decreasing reliability, X at different levels of r2TX, from the original “true” outcome, T. These data were simulated based on a classical test theory model of ࢄ࢏࢐ ൌ ࢀ࢏ ൅ ࢿ࢏࢐ where T is a standard normal variable representing the true score and ε is a random normal variable representing measurement error. The variance of ε is adjusted to produce the desired correlation between T and X in the population. Post-test T was a linear transformation of pre-test, Tpost = Tpre+d for each group, and independent ε were added to the pre- and post-test to create separate Xpre/post variables in each group.     1205.3.3 Between site differences We examined whether results of our measures differed statistically between study sites. All measures were analysed with a MANOVA to determine whether there were between-site differences. Full details of this analysis and associated post-hoc tests can be found in Table 5.6 for highly and moderately reliable measures, with low reliability measures in the appendix (Table A.2, A.3). Within the measures that met the ICC(2,k) criterion of 0.8, SEPs at both motor threshold and 150% of motor threshold differed between sites (ps<0.001). Absolute stimulation intensities across visits were similar in Leiden and London while at the Paris site intensities at Visit 2 were substantially larger than at Visits 1 or 3 (Table 5.6, Table A.4). Resting MEP latency also had a main effect of study site that remained despite controlling for arm length (p=0.003). Post-hoc analysis revealed that there was a significant difference between Paris and Vancouver (p=0.02) at the first time-point, and between London and Paris at the second and third time-points (p=0.005, p=0.001). In general, three main trends emerged. First, lower stimulation intensities were more likely to have between-site differences. Second, active measures were more likely to have between-site differences than their resting counterparts. Third, measures analysed by area under the curve had more site differences than when analysed by quantifying peak-to-peak amplitude.      121 Measure V df p London Paris Vancouver Leiden High Reliability       Rest Lat (ms) 0.35 6, 100 0.003 21.43 (1.55) 22.66 (1.75) 22.16 (1.41) N/A SEP Amp MT (V) 0.32 9, 180 <0.001 1.63 (0.87) 0.97 (1.14) 1.87 (1.18) 2.79 (1.60) SEP Amp 150 (V) 0.42 9, 162 <0.001 2.16 (0.99) 1.22 (1.14) 2.74 (1.81) 3.60 (1.96) Moderate Reliability       125 AMT Amp (mV) 0.44 6, 96 <0.001 2.10 (0.81) 1.41 (0.90) 1.26 (0.91) N/A 150 AMT Amp (mV)  0.31 6, 92 0.02 3.99 (1.40) 3.40 (2.03) 3.09 (1.87) N/A 130 RMT Area (mV) 0.29 6, 92 0.02 5.51 (3.76) 7.28 (6.17) 4.63 (2.84) N/A 175 AMT SPD (mV) 0.51 6, 80 0.001 159.08 (32.09) 165.72 (19.96) 138.95 (31.29) N/A Table 5.6: Between site differences. The effect of study site in a MANOVA. Results are shown for significant results for measures of high and moderate reliability. Similar results that did not reach statistical significance or for measures with low reliability can be found in the appendix. For latency variables, the MANOVA controlled for the arm length of each participant. Mean (SD) are shown for each study site averaging across time points. Only participants with data from all three time-points were included in this analysis. Leiden data was only available for EEG measures. Abbreviations: SEP: somatosensory evoked potential; MT: motor threshold; RMT: resting motor threshold; AMT: active motor threshold; SPD: silent period duration.  5.4 Discussion  Electrophysiological methods can help assess brain function. It is important to know how reliable data generated with different electrophysiological methods are in order to sufficiently power a study using such measures. Here, using healthy control data from a longitudinal study conducted at four study sites, we show that the study had low attrition rates indicating it was well tolerated. Generally, measures quantifying latencies, thresholds, and evoked responses at high stimulator intensities had the highest reliability, and required the smallest sample sizes to adequately power a study. Very few between-site differences were detected. Our data can assist in adequately powering research studies or clinical trials using electrophysiological measures.  The electrophysiological data was collected from healthy controls at four Track-On study sites in three study visits a year apart. Attrition rates were low suggesting that the electrophysiological measurements were well tolerated. Amongst the individuals who did drop   122out, the main reasons related to other aspects of the Track-On study including the duration of the study day, which included many other assessments. The electrophysiological data were collected at the end of the study day, so that time constraints or participant burden often affected the electrophysiological part of the protocol.  We then employed two complimentary approaches to understanding the reliability of each measure. Measures with high ICC(2,k) values have a favourable signal to noise ratio so that the chance is high to detect even small differences for example introduced by an intervention. In accordance with previous work, the most reliable measures included TMS thresholds, measures of latency, and sensory and motor evoked-potential amplitudes at higher stimulation intensities [14,170-177]. Most amplitude measures for recruitment curves at rest or when active, and sensory short-afferent inhibition or facilitation were of moderate reliability. Conversely, most area measures, or ratios (SPD/MEP size or conditioned/unconditioned MEP size) had low reliability. Measures with high reliability may relate most closely to brain structure; for example, latencies may reflect the integrity of a particular white matter tract. Since brain structure probably did not significantly change in our healthy controls over the 2 years of the Track-On study, the within-participant variability of these measures was also low. Such reasoning may also apply to evoked-response amplitudes, in particular SEPs that were equally reliable at motor threshold and 150% motor threshold intensities; however, it is also possible that in case of the MEPs they were most reliable when they were likely near a physiological maximum response and thus high reliability may also reflect a ceiling effect. Measures with lower reliability may record brain function that is less tightly linked to brain structure. This does not mean that the methods used for these measurements are inherently less reliable, as it is also possible that the greater within-participant variability results from physiological differences in brain function. For   123instance, the activity and excitability of the neuronal populations being stimulated may fluctuate so that the response to stimulation depends on the state of that population at the time of stimulation [178].  Given the a priori effect-size of interest, the necessary sample-size can be adapted to the reliability of the outcome measure in question. We next used the reliability analysis to hypothetically calculate statistical power assuming three simplified statistical analyses plans for simple or more complex data sets including a simple clinical trial design. As expected, studies using measures with high reliability require the smallest sample size. However, our results also indicate that measures with less than ideal reliability may still be usable if the sample size provides sufficient statistical power. We should note, however, that there are other methods for increasing power beyond increasing sample size. For instance, study design can improve statistical power through the proper use of covariates or the use of repeated measures to improve reliability (e.g., multiple baseline and post-tests; [179]. Furthermore, variance/standard deviation measures from previous real research will capture both variations in the true scores and measurement error, whereas simply assuming an a priori effect-size (e.g., adequate power to detect a hypothetical d=0.5) does not take measurement error into account. That said, effect-sizes can vary drastically from sample to sample (especially in under-powered studies). Therefore, we would recommend that research be conservative about effect-size estimates in the design of future experiments and use the reliability data from the present study to inform effect size calculations, helping to avoid underpowered study designs.     Finally, we assessed between-site differences. This is important when determining which metrics to include in large, multi-site studies. Altogether, there were few study site effects. Generally, measures that had higher stimulation intensities, and thus were probably close to a   124physiological ceiling had fewer between-site differences. Further, measures obtained at rest were often less influenced by study-site than measures where participants had to pre-activate the muscle that was recorded from. In keeping with previous work participants per protocol had to maintain a contraction level of about 20-30% of the maximum contraction strength measured either with a force transducer or an oscilloscope. The TrackOn electrophysiology protocol involved training of study site personnel and data monitoring. However, the data suggest that these measures may not have been sufficient and thus may not have prevented variability in participants’ levels of pre-activation. Finally, peak-to-peak amplitude measures were more consistent across sites than evoked-responses quantified as area under the MEP curve. We had expected that area measures would have been more robust since area measurements account for the possibility that the recruitment of additional motor units in response to higher stimulation intensities may not necessarily be synchronous. Temporal dispersion of responses would then be captured in an increase in the area under the MEP. However, our data indicate that this may not necessarily be the case so that amplitude measures may be better than area measurements. Another explanation, however, may be that at least with higher intensities amplitudes may be close to a physiological ceiling and hence the data are more stable than when measuring area. When comparing across sites and individuals, it is essential to control for anthropometric factors that may influence the data. For example, in the current work, we controlled for arm length to eliminate any impact that the length of neural connections in the periphery could have on our latency-based variables. Controlling for arm length, there were no consistent differences between the sites for any latency measure, except rest MEP latencies. The identification of thresholds for peripheral nerve stimulation could be another source of between-site variation for which to control. Our protocol specified that median nerve motor threshold was based on a   125barely visible twitch in the target muscle, though M-wave amplitudes were monitored throughout. However, this may not account for differences between participants in the anatomical make-up of the mixed sensory and motor median nerve so that in some individuals motor fibres may run closer to the skin and therefore stimulation electrode and in others deeper within the forearm. There were still some notable differences between visits in the same participants at the same site. In a study running 24 months it cannot always be guaranteed that the same investigator performs all the experiments at a given site. This may introduce variability for instance in the amount of pressure applied to the peripheral nerve stimulation electrode which then can translate into differences in threshold measures. There may be other factors that we are not aware of that also introduce variability. Our data suggest that collectively controlling for as many of these variables as possible is essential to reduce site influences in multi-centre studies.  In conclusion, we have systematically examined the variability of electrophysiology measures that are commonly used in clinical research studies. Across a large population of healthy individuals, attrition rates were low, suggesting that electrophysiological measures were well-tolerated. Levels of reliability varied substantially. This indicates that the choice of a dependent measure should be informed not only by its theoretical interest, but also its reliability, as poor reliability has profound, negative effects on statistical power. Similarly, methods susceptible to between-site differences should be used cautiously to minimise site related differences in large multi-centre studies. Taken together, these steps can help to ensure that any differences between measurements truly reflect underlying biological differences rather than inherent methodological variability.    126Chapter 6: General discussion  This thesis was designed to comprehensively examine the neurophysiology of sensorimotor integration, providing insight into the influence of aging and chronic stroke on different neurophysiological pathways of sensorimotor integration, the plasticity of sensorimotor integration, and the reliability of measures used to test sensorimotor integration. The major findings from this investigation are summarised below. Limitations to interpretation of the current work, implications of the findings, and future directions for the examination of sensorimotor integration are also discussed. 6.1 Aging and chronic stroke differentially impact sensorimotor integration Cumulatively, results from Chapters 2-4 suggest that neurophysiological changes in sensorimotor integration occur with healthy aging and also during chronic stroke recovery. Importantly, our work also suggests that these changes in sensorimotor integration were independent of shifts in somatosensory cortical excitability or corticospinal excitability alone. The techniques used to assess sensorimotor integration in this thesis enabled the comparison of different anatomical pathways underlying sensorimotor integration. Though both median nerve stimulation and muscle belly vibration recruit 1a afferents, the proportional recruitment of these fibres differs between techniques. Indirect sensorimotor integration can be probed with median nerve stimulation as it recruits 1a afferents amongst a variety of large diameter fibres, resulting in afferent feedback primarily travelling to Brodmann area 3b in S1 before being relayed to M1 through areas 1 and 2 [8,9,32-36]. In contrast, direct sensorimotor integration can be probed with muscle-belly vibration, which predominantly recruits 1a afferents that can directly ascend to M1 through thalamocortical projections or through reciprocal connections with Brodmann area 3a of   127S1 [32,33,132-134,180]. Therefore, examination of both of these pathways provides unique information on pathological alterations to sensorimotor integration.   Results from Chapter 2 indicate that indirect sensorimotor integration is altered in individuals with chronic stroke and healthy older individuals, as compared to younger healthy individuals; specifically, these groups present with disinhibition in SAI, thought to be reflective of shifts in cortical cholinergic and GABAergic inhibitory circuits. The similarity shown between individuals with stroke and older healthy individuals suggest that this neurophysiological change is driven by age-related alterations in this pathway, rather than resulting from stroke-related changes. Despite the similarities shown between individuals with chronic stroke and older healthy controls, the level of disinhibition in stroke was functionally relevant and related to both poorer motor function and greater levels of impairment; individuals who showed patterns of sensorimotor integration more similar to normative patterns in young healthy individuals presented with fewer motor deficits. Given the short latency between the median nerve stimulation and the TMS pulse, SAI is thought to reflect a direct relay from Brodmann area 3b in S1 to M1. Thus, the immediate transfer of incoming somatosensory information to the primary motor cortex appears to be important in motor behaviour.   In contrast to change in indirect sensorimotor integration being primarily altered by age, direct sensorimotor integration was similar across age groups, but differed in individuals with chronic stroke. Taken together, these results suggest that age-related changes may more specifically relate to the incorporation of afferent feedback through corticocortical pathways relaying indirectly from Brodmann area 3b of S1 to M1, whereas more direct pathways to M1 are affected in chronic stroke. The specific measure of direct sensorimotor integration that displayed differences in individuals with chronic stroke was the vibration-induced response to   128single-pulse TMS, which is influenced by both spinal and cortical mechanisms [40]. The differences documented in this measure between individuals with chronic stroke and older healthy individuals could be driven by either of these components. Past work indicated that spinal excitability, quantified with H-reflex amplitudes, is higher in individuals post-stroke as compared to age-matched controls [152]. This result would lead us to expect that there would be an increased response to vibration in these individuals if vibration-induced facilitation is driven primarily by spinal mechanisms, yet a reduced response to vibration was seen. Therefore, the cortical component of sensorimotor integration may be driving this stroke-related change. Although paired-pulse measures of TMS can probe cortical circuitry, the underlying neural mechanisms for SICI and ICF differ from single-pulse MEP amplitude, and as a result, one cannot inform the other.   As noted above, SAI was an important predictor in explaining variance in both motor impairment and motor function post-stroke. Single-pulse response to vibration adds to the relationship between motor function and circuits of sensorimotor integration. Interestingly, the importance of direct sensorimotor integration is only shown for motor function, not motor impairment. Arguably, this could result from the nature of the behavioural tests. The majority of tasks in the FM assessment are less complex, and less dependent on fine motor control than those used in the WMFT meaning that the impairment tests may not be as demanding as those used to assess motor function. As motor task demands increase, both direct and indirect pathways of sensorimotor integration may be required to assist in performance. Nevertheless, neurophysiological measures of sensorimotor integration contribute to behavioural outcomes post-stroke, and thus a comprehensive understanding of these measures is important.   1296.2 Response to sensory training is altered post-stroke, not in healthy aging   Following a 15-minute block of sensory training, involving cyclic vibration in which individuals were instructed to detect and respond to frequency changes, older and younger healthy individuals showed a reduced response to vibration-based measures of sensorimotor integration. This pattern was shown both in the influence of vibration on single-pulse MEP amplitudes as well as intracortical facilitation, but not in our measures of intracortical inhibition. There were no changes in measures of indirect sensorimotor integration or corticospinal excitability, suggesting that sensory training targets sensorimotor integration, rather than altering overall excitability levels, and more specifically the changes are isolated to circuits responsible for processing a specific type of somatosensory information. In addition, the differential modulation of each specific measure by our vibration intervention provides supplementary support to the argument that different underlying neural pathways may drive these systems.  Prior to sensory training, vibration induced an increase in MEP amplitude, and an increase in ICF in both age groups, with no change in SICI. Therefore, it follows that these circuits would be the ones to change following sensory training given that the application of vibratory sensory training would repetitively activate these specific circuits and plasticity is known to result from repeated activation within specific synaptic structures. Contrary to our hypothesis, following sensory training, the influence of vibration on each of these measures was reduced. While past work has emphasised the importance of attention for inducing change in these networks [136], the sensory training paradigm had previously been designed to control for this by instructing individuals to respond to frequency changes. Our sensory training used the same levels of frequency changes in younger healthy individuals as has previously been shown to be effective [97].    130 Attention may influence somatosensory processing, and thus sensorimotor integration through somatosensory gating and relevancy-based facilitation. Through prefrontal-thalamic networks, irrelevant somatosensory information is suppressed while relevant information facilitated [67,181]. In the context of the sensory training paradigm, we hypothesised that attending to frequency changes would make the information relevant, thus increasing somatosensory information relayed through the thalamus and reaching the cortex. We cannot rule out the possibility that this did occur, and lead to the reduction in measures of direct sensorimotor integration; however, past work has shown that without attentional instruction, the sensory training paradigm reduces the response to vibration, in line with the results shown here [136]. Therefore, we suggest that during sensory training, attention did not facilitate the somatosensory information arising from vibration. Without attentional facilitation, repetitive afferent stimulation may lead to diminished responses within the activated networks. Support for this hypothesis comes from cellular work; when neurons are stimulated in an ex-vivo environment, repetition leads to a reduction in the neuronal evoked responses [139]. Overall, these results suggest that measures of sensorimotor integration can be changed within a single session and that this response remains intact with healthy aging.   In the sample of individuals with chronic stroke studied in this thesis, we showed a similar response to sensory training in the response to vibration on single-pulse MEPs as was seen in older healthy individuals. In contrast, individuals with chronic stroke did not demonstrate a reduction in the influence of vibration on ICF as older healthy controls did. This could stem from baseline differences between groups. Prior to sensory training, individuals with chronic stroke, though reduced, showed a vibration-induced facilitation of single-pulse MEPs whereas vibration did not influence ICF. If repetitive activation of specific neuronal circuitry underlies   131the response to vibration following sensory training, such a reduction will be isolated to the specific circuitry being stimulated.   Interestingly, larger responses to vibration with single-pulse TMS related to poorer motor function. We posit that this may result from the spinal contributors to the response to vibration. Increased spinal excitability, beyond that seen in the non-paretic limb or healthy age-matched controls has been associated with worse function post-stroke [152]. Therefore, if spinal contributors are driving the increased response to vibration, this may underlie the relationship with function. Sensory training in individuals with chronic stroke, as discussed above, lead to a reduction in response to vibration. Potentially, this could be beneficial if it reduces activity within the hyperexcited spinal circuits, which may improve function. 6.3 Measures of sensorimotor integration have varying levels of reliability   Each study in this investigation relied on electrophysiological measures including SEPs quantified with EEG, single-pulse TMS measures, and standard measures of sensorimotor integration pairing peripheral nerve stimulation with TMS. These measures are commonly used to assess the effectiveness of an intervention or to establish group differences in neurophysiology. At present, standard practice typically uses sample sizes ranging from 10-15 individuals, as was done in much of the current work [22,25,41,97,136]. For the final component of this thesis, we determined the reliability of these metrics using a large, multi-site sample. We examined data from 112 healthy individuals across a variety of age-ranges, collected at four international sites. As these data we generated as a part of a larger, international trial we were limited to the most commonly used measures and therefore could not include all the measures used in the first three research chapters of this thesis. However, we investigated overall reliability of TMS, SEPs, and SAI and AF.   132 Overall, our analysis indicated that the most reliable measures were thresholds, latency-based measures, and evoked-potential amplitudes at high stimulus intensities. The measures that were used in Chapters 2-4; SEP amplitudes at motor threshold, higher intensity MEP amplitudes, SAI, and AF, all fell within either the moderate or high reliability ranges [168]. As such, standard reliability analysis would suggest that these are acceptable measures to probe the neurophysiology of sensorimotor integration.   The unique contribution of our reliability analysis in Chapter 5 is the inclusion of statistical simulations to determine the sample sizes that would be required to adequately power studies based on specific ICCs for a given measure. When we examined our measures of sensorimotor integration (SAI, AF) in this way, we revealed that depending on the effect size and exact statistic being employed, sample sizes should be a minimum of 50 individuals. This indicates a large discrepancy between common practice in the field and statistical ideals. In addition to our statistical suggestions, practical considerations such as participant/experimenter burden, time constraints, funding, and patient availability must also be taken into account. Many of these logistical limitations may be overcome with collaboration or data pooling between sites. Our data suggest that with standardisation of protocols and guidelines, between-site differences can be reduced, thus making it easier to compare and pool data. When contextualising our results with past investigations into sensorimotor integration and aging, for example, we show a similar reduction in SAI as has been previously reported [1].  In addition, confidence in the current neurophysiological results can be garnered by examining effect sizes and contextualising our findings with previous literature. All of the neurophysiological effects shown in this dissertation meet the criteria for a moderate to high effect size and were based on power calculations. In this way, we suggest that while the results   133presented here would benefit from an increased sample size, they provide a starting point from which further study can be initiated to contribute to our understanding of the neurophysiology of sensorimotor integration in healthy aging and chronic stroke. 6.4 Limitations There are a number of limitations to consider when interpreting the research contained within this thesis. The most broad is a technical limitation derived from the nature of TMS; TMS relies on cortical stimulation evoking a response that is quantified peripherally through electromyography. As such, the technique relies on both spinal and cortical mechanisms. Past work has noted the cortical nature of many paired-pulse techniques, including the ones used in this work; however, given the peripheral measure we cannot fully rule out the contribution of changes in spinal excitability. Despite this, the fact that there were not pathology related changes in all measures, and similarly that not all TMS measures changed in response to sensory training strengthens confidence that they are not being driven only by spinal changes, which would logically change all TMS measures.  Another technological limitation arises from our choice of peripheral stimuli used to probe direct and indirect sensorimotor integration. Muscle belly vibration was intended to primarily activate muscle spindles and 1a afferents but, as the vibration device is placed on the skin, it will also activate cutaneous receptors and associated pathways. Further, the median nerve is a mixed nerve that, when stimulated, often produces a muscle response; as a result, median nerve stimulation will recruit 1a afferents amongst a variety of other, more cutaneous-based, fibres that will also generate somatosensory information that is relayed to the cortex. As such, while muscle belly vibration may preferentially recruit muscle spindles that are received cortically in Brodmann area 3a before being relayed directly to M1, and response to median   134nerve stimulation is based on information being preferentially received in Brodmann area 3b and transferred to M1, there is overlap between the two modes of sensorimotor integration. To more specifically probe these underlying pathways, future work could stimulate a purely cutaneous nerve to inform indirect sensorimotor integration and a combination of muscle belly vibration of the first dorsal interosseous with blocked somatosensation via the superficial radial nerve to inform direct sensorimotor integration.   When evaluating our stroke data, it is important to consider that due to the nature of the techniques used in the current design, our sample consists of well-recovered individuals. In order to quantify neurophysiological measures of sensorimotor integration via TMS, it was imperative that a response could be evoked from the ipsilesional hemisphere. As has been previously shown, the presence of an ipsilesional MEP is an indicator of degree of recovery to be expected [62,182], and thus individuals with a MEP as were included in our study have less severe impairment than those without a MEP. Additionally, we were quantifying responses evoked in the APB muscle in the hand. These may be harder to evoke than more proximal muscles due to the arrangement of cortical representations; proximal muscles are more diffuse, whereas distal muscles are more concentrated and thus may be more affected by stroke. Post-stroke recovery generally follows a similar pattern with restoration of movement in proximal muscles responsible for gross motor control preceding distal muscles required for fine motor control [183]. Therefore, while the results are reflective for the sample in the current thesis, and may extend to other well-recovered individuals in the chronic phase of stroke recovery, they do not encompass a wide range of severities.    Another consideration in the stroke group is that our inclusion criteria were based on behavioural parameters rather than imaging metrics. Individuals were included if they had   135sensorimotor impairments that had persisted into the chronic phase of recovery. Each individual had a stroke within the middle cerebral artery distribution, but beyond that, lesion location was not controlled for. As a result, our sample is heterogeneous and includes individuals with a variety of lesion locations. Heterogeneity within the sample may, in one aspect, lead to more generalizable results, given the individualised nature of all strokes, with various sizes and locations. In contrast, response seen in neurophysiological measures could be highly dependent on exact lesion location or volume, and this must be contemplated when examining the current results.   The sensory training paradigm used in the current work was selected from past research showing both an induction of neurophysiological and behavioural change in young healthy individuals [97]. In our younger healthy group, we used the identical paradigm, and with our older healthy adults and individuals with chronic stroke, we adjusted the changes in frequency that were required to detect to be larger to account for sensory decline and in an effort to equate difficulties between groups. Despite these attempts to control for attention and maintain a constant difficulty level, our results did not show the same effects that have been previously published [97]. Alternatively, our results were in line with past work employing passive vibration with no attentional demands that showed a reduction in response to vibration after applied for 15 minutes [95]. Without an objective method by which to quantify attention, we cannot understand if this factor is driving our effects. While our results do indicate that it is possible to induce change in neural circuitry responsible for sensorimotor integration in individuals with chronic stroke, as well as healthy individuals of various ages, it is possible that sensory training is not an optimal paradigm to do this. Understanding the baseline neurophysiology and how those measures are affected by healthy aging and stroke do provide   136important information on what directional change may be required to “normalise” these measures. One past study does suggest that there are responders and non-responders to vibration [96]; a larger sample size may enable the parcellation of this hypothesis, and guide interventions. However, given its success in previous research, if attention could be more strictly controlled and this effect widely replicated, potential remains.   Finally, given the results of our final study on the reliability of electrophysiological measures a limitation of the first three research chapters in this thesis is their small sample sizes. Although this work is in line with convention and were driven by practical limitations, increased sample sizes would only strengthen the result. A multitude of other studies investigating neurophysiology post-stroke, as well as in healthy aging, have used similar sample sizes to document population-related changes [1: for review]. Similar sample sizes are common when investigating responses to interventions. Nevertheless, it is impossible to discount the value in increasing the sample size. In order to coalesce practical and statistical realities, it may be optimal to employ standardized procedures across many sites to enable data pooling.  6.5 Implications and future directions  For the first time in 2016, Canada had more seniors than children [184]; as baby boomers age, the population of older individuals is quickly growing. Aging increases the prevalence of neuropathology including increased risk of stroke. As such, it is imperative that we understand the neurophysiological changes that accompany both healthy aging and stroke. Furthering our knowledge of what biological changes may accompany behavioural changes provides important targets for reducing functional decline. Understanding whether sensorimotor integration is a plastic phenomenon that can be changed with single or multi-session interventions will provide insight into the validity of these circuits as potential rehabilitative targets to improve function   137and impairment in clinical populations. The results enclosed in the current thesis provide important insight into these questions, but also serve as a starting point for future research.   The primary outcome measures in each of the research studies included here are forms of transcranial magnetic stimulation. To provide support for the current work, and extend our understanding of sensorimotor integration, imaging methodology could be employed. For example, we propose that there are two pathways of sensorimotor integration that are differentially impacted by aging and chronic stroke. By capitalising on tractography-based metrics, the impact of aging and stroke on the integrity of thalamocortical tracts to both M1 and S1 could be investigated. Of particular importance post-stroke, functional magnetic resonance imaging could be used to establish compensatory motor patterns that have been developed and investigate the role that these may have in sensorimotor integration. A shift towards contralesional or bilateral control of movement is often shown in individuals recovering from a stroke. A disconnect may occur in sensorimotor integration if the incoming afferent information ascends to the contralateral somatosensory and motor cortices, but motor control has shifted elsewhere.  In this case afferent information may not be appropriately translated to inform movement. Future investigation into the neurophysiology of sensorimotor integration should also include spinal measures such as H-reflexes in an attempt to parse out the cortical and spinal contributors to group differences.  In addition to expanding our understanding of neurophysiological changes, we must begin to explore the relationship between these changes and behaviour. As behavioural outcomes have great impact on daily life, it is important to understand how biological changes translate to behaviour, and if changing neurophysiology in a desired way translates to behavioural   138improvements. If this is not the case, investigation could occur into whether these neurophysiological changes may be early indicators of future changes in behaviour.   In individuals with stroke, methods such as transcranial magnetic stimulation that require a peripheral, or specific cortical, response commonly exclude a sub-group of individuals with severe impairment. The exclusion of these individuals is unfortunate, as it is of the utmost importance to develop interventions that can improve motor function and impairment in these individuals to bolster their quality of life. Therefore, we suggest that the efficacy of new techniques that are not dependent on motor output be investigated. TMS-EEG provides a potential exciting new technique through which this could be accomplished. This would allow for more inclusive study of neurophysiological investigation into sensorimotor integration and beyond.  Investigation into the reliability of neurophysiological measures is fairly new and not yet comprehensive. Future work in this area should expand to include less conventional measures that show promise and applicability in clinical populations. As previously mentioned, the development of guidelines of standardised procedures and optimal outcome measures would be helpful in increasing the cohesion of studies across sites and experiments, allowing for more comparable research and an increased ease in performing important meta-analyses to guide future experimental design.  Finally, given the recent trend towards individualised medicine, we believe that future work should examine individuals’ baseline neurophysiology to understand person-specific changes that are taking place. This would allow for more targeted intervention that could be adjusted based on the individual. For example, an individual could have “abnormal” sensorimotor integration in that they have less, or more, inhibition than typically seen. These two   139presentations should not be treated in the same way. Interventions typically show responders and non-responders, and having baseline biomarker data may help to inform how to best treat each individual. Personalised approaches are particularly important for heterogeneous conditions such as stroke. 6.6 Conclusions  This thesis furthers our understanding of the neurophysiology of sensorimotor integration across age-ranges and in chronic stroke by examining pathways activated by both vibration and peripheral nerve stimulation with unique methods of integration into the primary motor cortex. We document age-related alterations to indirect sensorimotor integration without changes in direct sensorimotor integration. Conversely, in individuals with chronic stroke, direct sensorimotor integration shows differences beyond those seen with healthy aging, but this is not true for indirect sensorimotor integration. Further, the current research shows that sensorimotor integration networks can be altered with a single-session of an intervention suggesting that sensorimotor integration remains plastic in healthy aging and chronic stroke. Finally, we stress the importance of understanding reliability of neurophysiological measures used in this thesis and commonly in investigations into the efficacy of interventions or pathological differences. Cumulatively, the research highlights many aspects of neurophysiological investigation into sensorimotor integration and motor cortical excitability in various populations. This thesis provides a wealth of data to foster future investigation into unique sensorimotor circuitry, its importance in clinical populations, and the relationship between this physiology and behavioural outcome measures.     140References [1] Bhandari A, Radhu N, Farzan F, Mulsant BH, Rajji TK, Daskalakis ZJ et al. A meta-analysis of the effects of aging on motor cortex neurophysiology assessed by transcranial magnetic stimulation. Clin Neurophysiol 2016;127:2834-45. [2] Castel-Lacanal E, Marque P, Tardy J, de Boissezon X, Guiraud V, Chollet F et al. Induction of cortical plastic changes in wrist muscles by paired associative stimulation in the recovery phase of stroke patients. Neurorehabil Neural Repair 2009;23:366-72. [3] Machado S, Cunha M, Velasques B, Minc D, Teixeira S, Domingues CA et al. Sensorimotor integration: basic concepts, abnormalities related to movement disorders and sensorimotor training-induced cortical reorganization. Rev Neurol 2010;51:427-36. [4] Nowak DA, Glasauer S, Hermsdorfer J. How predictive is grip force control in the complete absence of somatosensory feedback? Brain 2004;127:182-192. [5] Messier J, Adamovich S, Berkinblit M, Tunik E, Poizner H. Influence of movement speed on accuracy and coordination of reaching movements to memorized targets in three-dimensional space in a deafferented subject. Exp Brain Res 2003;150:399-416. [6] Sarlegna FR, Gauthier GM, Bourdin C, Vercher JL, Blouin J. Internally driven control of reaching movements: a study on a proprioceptively deafferented subject. Brain Res Bull 2006;69:404-15. [7] Kandel E, Schwartz J, Jessell T, Sigelbaum SA, Hudspeth AJ. Principles of neural science. 2012. [8] Jones EG, Coulter JD, Hendry SH. Intracortical connectivity of architectonic fields in the somatic sensory, motor and parietal cortex of monkeys. J Comp Neurol 1978;181:291-347. [9] Petrof I, Viaene AN, Sherman SM. Properties of the primary somatosensory cortex projection to the primary motor cortex in the mouse. J Neurophysiol 2015;113:2400-7. [10] Sherman S, Guillery R. The role of the thalamus in the flow of information to the cortex. Philosophical Transactions of the Royal Society B: Biological Sciences 2002;357:1695-1708. [11] Burke D, Hicks R, Gandevia SC, Stephen J, Woodforth I, Crawford M. Direct comparison of corticospinal volleys in human subjects to transcranial magnetic and electrical stimulation. J Physiol 1993;470:383-93. [12] Terao Y, Ugawa Y. Basic mechanisms of TMS. J Clin Neurophysiol 2002;19:322-43. [13] Rossini PM, Berardelli A, Deuschl G, Hallett M, Maertens de Noordhout AM, Paulus W et al. Applications of magnetic cortical stimulation. The International Federation of Clinical Neurophysiology. Electroencephalogr Clin Neurophysiol Suppl 1999;52:171-85.   141[14] Kamen G. Reliability of motor-evoked potentials during resting and active contraction conditions. Med Sci Sports Exerc 2004;36:1574-9. [15] Siebner HR, Rothwell J. Transcranial magnetic stimulation: new insights into representational cortical plasticity. Exp Brain Res 2003;148:1-16. [16] Kujirai T, Caramia MD, Rothwell JC, Day BJ, Thompson PD, Ferbert A et al. Corticocortical inhibition in the human motor cortex. J Physiol 1993;471:501-19. [17] Ziemann U, Lonnecker S, Steinhoff BJ, Paulus W. The effect of lorazepam on the motor cortical excitability in man. Exp Brain Res 1996;109:127-35. [18] Ziemann U, Lonnecker S, Steinhoff BJ, Paulus W. Effects of antiepileptic drugs on motor cortex excitability in humans: a transcranial magnetic stimulation study. Ann Neurol 1996;40:367-78. [19] Hanajima R, Ugawa Y, Terao Y, Sakai K, Furubayashi T, Machii K et al. Paired-pulse magnetic stimulation of the human motor cortex: differences among I waves. J Physiol 1998;509 ( Pt 2):607-18. [20] Ziemann U, Lonnecker S, Steinhoff BJ, Paulus W. Motor excitability changes under antiepileptic drugs. Adv Neurol 1999;81:291-8. [21] Ziemann U. Intracortical inhibition and facilitation in the conventional paired TMS paradigm. Electroencephalogr Clin Neurophysiol Suppl 1999;51:127-36. [22] Chen R, Corwell B, Hallett M. Modulation of motor cortex excitability by median nerve and digit stimulation. Exp Brain Res 1999;129(1):77,78-86. [23] Fischer M, Orth M. Short-latency sensory afferent inhibition: conditioning stimulus intensity, recording site, and effects of 1 Hz repetitive TMS. Brain Stimul 2011;4:202-9. [24] Allison T, McCarthy G, Wood CC, Williamson PD, Spencer DD. Human cortical potentials evoked by stimulation of the median nerve. II. Cytoarchitectonic areas generating long-latency activity. J Neurophysiol 1989;62:711-22. [25] Sailer A, Molnar GF, Cunic DI, Chen R. Effects of peripheral sensory input on cortical inhibition in humans. J Physiol 2002;544:617-29. [26] Sailer A, Molnar GF, Paradiso G, Gunraj CA, Lang AE, Chen R. Short and long latency afferent inhibition in Parkinson's disease. Brain 2003;126:1883-94. [27] Tokimura H, Ridding MC, Tokimura Y, Amassian VE, Rothwell JC. Short latency facilitation between pairs of threshold magnetic stimuli applied to human motor cortex. Electroencephalogr Clin Neurophysiol 1996;101:263-72. [28] Di Lazzaro V, Oliviero A, Profice P, Pennisi MA, Di Giovanni S, Zito G et al. Muscarinic receptor blockade has differential effects on the excitability of intracortical circuits in the human motor cortex. Exp Brain Res 2000;135:455-61.   142[29] Devanne H, Degardin A, Tyvaert L, Bocquillon P, Houdayer E, Manceaux A et al. Afferent-induced facilitation of primary motor cortex excitability in the region controlling hand muscles in humans. Eur J Neurosci 2009;30:439-48. [30] Di Lazzaro V, Profice P, Ranieri F, Capone F, Dileone M, Oliviero A et al. I-wave origin and modulation. Brain Stimul 2012;5:512-25. [31] Bailey AZ, Asmussen MJ, Nelson AJ. Short-latency afferent inhibition determined by the sensory afferent volley. J Neurophysiol 2016;116:637-44. [32] Heath CJ, Hore J, Phillips CG. Inputs from low threshold muscle and cutaneous afferents of hand and forearm to areas 3a and 3b of baboon's cerebral cortex. J Physiol 1976;257:199-227. [33] Hore J, Preston JB, Cheney PD. Responses of cortical neurons (areas 3a and 4) to ramp stretch of hindlimb muscles in the baboon. J Neurophysiol 1976;39:484-500. [34] Kaneko T, Caria MA, Asanuma H. Information processing within the motor cortex. II. Intracortical connections between neurons receiving somatosensory cortical input and motor output neurons of the cortex. J Comp Neurol 1994;345:172-84. [35] Kaneko T, Cho R, Li Y, Nomura S, Mizuno N. Predominant information transfer from layer III pyramidal neurons to corticospinal neurons. J Comp Neurol 2000;423:52-65. [36] Mao T, Kusefoglu D, Hooks BM, Huber D, Petreanu L, Svoboda K. Long-range neuronal circuits underlying the interaction between sensory and motor cortex. Neuron 2011;72:111-23. [37] Kojima S, Onishi H, Miyaguchi S, Kotan S, Sugawara K, Kirimoto H et al. Effects of cathodal transcranial direct current stimulation to primary somatosensory cortex on short-latency afferent inhibition. Neuroreport 2015;26:634-7. [38] Tsang P, Jacobs MF, Lee KG, Asmussen MJ, Zapallow CM, Nelson AJ. Continuous theta-burst stimulation over primary somatosensory cortex modulates short-latency afferent inhibition. Clin Neurophysiol 2014;125:2253-9. [39] Jones EG, Porter R. What is area 3a? Brain Res 1980;203:1-43. [40] Kossev A, Siggelkow S, Schubert M, Wohlfarth K, Dengler R. Muscle vibration: different effects on transcranial magnetic and electrical stimulation. Muscle Nerve 1999;22:946-8. [41] Rosenkranz K, Rothwell JC. Differential effect of muscle vibration on intracortical inhibitory circuits in humans. J Physiol 2003;551:649-60. [42] Rosenkranz K, Pesenti A, Paulus W, Tergau F. Focal reduction of intracortical inhibition in the motor cortex by selective proprioceptive stimulation. Exp Brain Res 2003;149:9-16. [43] Heft MW, Robinson ME. Somatosensory function in old age. J Oral Rehabil 2017;44:327-32.   143[44] Goble DJ, Coxon JP, Wenderoth N, Van Impe A, Swinnen SP. Proprioceptive sensibility in the elderly: degeneration, functional consequences and plastic-adaptive processes. Neurosci Biobehav Rev 2009;33:271-8. [45] Capaday C, Cooke JD. The effects of muscle vibration on the attainment of intended final position during voluntary human arm movements. Exp Brain Res 1981;42:228-30. [46] Cody FW, Schwartz MP, Smit GP. Proprioceptive guidance of human voluntary wrist movements studied using muscle vibration. J Physiol 1990;427:455-70. [47] Cordo PJ. Kinesthetic control of a multijoint movement sequence. J Neurophysiol 1990;63:161-72. [48] Hunter SK, Pereira HM, Keenan KG. The aging neuromuscular system and motor performance. J Appl Physiol (1985) 2016;121:982-95. [49] Serrien DJ, Swinnen SP, Stelmach GE. Age-related deterioration of coordinated interlimb behavior. J Gerontol B Psychol Sci Soc Sci 2000;55:P295-303. [50] Maes C, Gooijers J, Orban de Xivry JJ, Swinnen SP, Boisgontier MP. Two hands, one brain, and aging. Neurosci Biobehav Rev 2017;75:234-56. [51] Bernard JA, Seidler RD. Moving forward: age effects on the cerebellum underlie cognitive and motor declines. Neurosci Biobehav Rev 2014;42:193-207. [52] Desmedt JE, Cheron G. Non-cephalic reference recording of early somatosensory potentials to finger stimulation in adult or aging normal man: differentiation of widespread N18 and contralateral N20 from the prerolandic P22 and N30 components. Electroencephalogr Clin Neurophysiol 1981;52:553-70. [53] Yamaguchi S, Knight RT. Gating of somatosensory input by human prefrontal cortex. Brain Res 1990;521:281-8. [54] Zikopoulos B, Barbas H. Prefrontal projections to the thalamic reticular nucleus form a unique circuit for attentional mechanisms. J Neurosci 2006;26:7348-7361. [55] Cueva AS, Galhardoni R, Cury RG, Parravano DC, Correa G, Araujo H et al. Normative data of cortical excitability measurements obtained by transcranial magnetic stimulation in healthy subjects. Neurophysiol Clin 2016;46:43-51. [56] Yang AC, Tsai SJ, Liu ME, Huang CC, Lin CP. The Association of Aging with White Matter Integrity and Functional Connectivity Hubs. Front Aging Neurosci 2016;8:143. [57] Tomasi D, Volkow ND. Aging and functional brain networks. Mol Psychiatry 2012;17:471, 549-58. [58] Hafkemeijer A, Altmann-Schneider I, de Craen AJ, Slagboom PE, van der Grond J, Rombouts SA. Associations between age and gray matter volume in anatomical brain networks in middle-aged to older adults. Aging Cell 2014;13:1068-74.   144[59] Young-Bernier M, Davidson PS, Tremblay F. Paired-pulse afferent modulation of TMS responses reveals a selective decrease in short latency afferent inhibition with age. Neurobiol Aging 2012;33:835.e1,835.11. [60] Degardin A, Devos D, Cassim F, Bourriez JL, Defebvre L, Derambure P et al. Deficit of sensorimotor integration in normal aging. Neurosci Lett 2011;498:208-12. [61] Lee SY, Lim JY, Kang EK, Han MK, Bae HJ, Paik NJ. Prediction of good functional recovery after stroke based on combined motor and somatosensory evoked potential findings. J Rehabil Med 2010;42:16-20. [62] Stinear CM, Barber PA, Smale PR, Coxon JP, Fleming MK, Byblow WD. Functional potential in chronic stroke patients depends on corticospinal tract integrity. Brain 2007;130:170-80. [63] Post-stroke rehabilitation: assessment, referral and patient management. Post-Stroke Rehabilitation Guideline Panel. Agency for Health Care Policy and Research. Am Fam Physician 1995;52:461-70. [64] Borstad AL, Nichols-Larsen DS. Assessing and treating higher level somatosensory impairments post stroke. Top Stroke Rehabil 2014;21:290-5. [65] Shibasaki H, Yamashita Y, Tsuji S. Somatosensory evoked potentials. Diagnostic criteria and abnormalities in cerebral lesions. J Neurol Sci 1977;34:427-39. [66] Tsumoto T, Hirose N, Nonaka S, Takahashi M. Cerebrovascular disease: changes in somatosensory evoked potentials associated with unilateral lesions. Electroencephalogr Clin Neurophysiol 1973;35:463-73. [67] Zikopoulos B, Barbas H. Circuits for multisensory integration and attentional modulation through the prefrontal cortex and the thalamic reticular nucleus in primates. Rev Neurosci 2007;18:417-438. [68] McDonnell MN, Stinear CM. TMS measures of motor cortex function after stroke: A meta-analysis. Brain Stimul 2017. [69] Swayne OB, Rothwell JC, Ward NS, Greenwood RJ. Stages of motor output reorganization after hemispheric stroke suggested by longitudinal studies of cortical physiology. Cereb Cortex 2008;18:1909-22. [70] Di Lazzaro V, Profice P, Pilato F, Capone F, Ranieri F, Florio L et al. The level of cortical afferent inhibition in acute stroke correlates with long-term functional recovery in humans. Stroke 2012;43:250-2. [71] Ackerley SJ, Stinear CM, Barber PA, Byblow WD. Combining theta burst stimulation with training after subcortical stroke. Stroke 2010;41:1568-72. [72] Liepert J, Binder C. Vibration-induced effects in stroke patients with spastic hemiparesis--a pilot study. Restor Neurol Neurosci 2010;28:729-35. [73] Asanuma H, Keller A. Neuronal mechanisms of motor learning in mammals. Neuroreport 1991;2:217-224.   145[74] Nudo RJ. Recovery after brain injury: mechanisms and principles. Front Hum Neurosci 2013;7:887. [75] Butler AJ, Wolf SL. Putting the brain on the map: use of TMS to assess and induce cortical plasticity of upper-extremity movement. Phys Ther 2007;87:719-36. [76] Nudo RJ. Neurophysiology of motor skill learning. In: Anonymous Learning and Memory: A Comprehensive Reference: Academic Press; 2008, p. 403-421. [77] Calford MB, Tweedale R7. Acute changes in cutaneous receptive fields in primary somatosensory cortex after digit denervation in adult flying fox. J Neurophysiol 1991;65:178-87. [78] Jacobs KM, Donoghue JP. Reshaping the cortical motor map by unmasking latent intracortical connections. Science 1991;251:944-47. [79] Hess G, Aizenman CD, Donghue JP. Conditions for the induction of long-term potentiation in layer II/III horizontal connections of the rat motor cortex. Brain Res 1996;413:360-4. [80] Bliss TV, Lomo T. Long lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of perforant path. J Physiol 1973;232:331-56. [81] Butefisch CM, Davis BC, Wise SP, Sawaki L, Kopylev L, Classen J et al. Mechanisms of use-dependent plasticity in the human motor cortex. Proc Natl Acad Sci U S A 2000;97:3661-5. [82] Huntley GW. Correlation between patterns of horizontal connectivity and the extend of short-term representational plasticity in rat motor cortex. Cereb Cortex 1997;7:143-56. [83] Huntley GW, Jones EG. Relationship of intrinsic connections to forelimb movement representations in monkey motor cortex: a correlative anatomic and physiological study. J Neurophysiol 1991;66:390-413. [84] Pascual-Leone A, Nguyet D, Cohen LG, Brasil-Nito JB, Commarota A, Hallet M. Modulation of muscle responses evoked by transcranial magnetic stimulation during acquisition of new fine motor skills. J Neurophysiol 1995;74:1037-45. [85] Neva JL, Legon W, Staines WR. Primary motor cortex excitability is modulated with bimanual training. Neurosci Lett 2012;514:147-51. [86] Andrew D, Haavik H, Dancey E, Yielder P, Murphy B. Somatosensory evoked potentials show plastic changes following a novel motor training task with the thumb. Clin Neurophysiol 2014. [87] Stefan K, Kunesch E, Cohen LG, Benecke R, Classen J. Induction of plasticity in the human motor cortex by paired associative stimulation. Brain 2000;123:572-84. [88] Huang YZ, Edwards MJ, Rounis E, Bhatia KP, Rothwell JC. Theta burst stimulation of the human motor cortex. Neuron 2005;45:201-6. [89] Stefan K, Wycislo M, Classen J. Modulation of associative human motor cortical plasticity by attention. J Neurophysiol 2004;92:66-72.   146[90] McGaughy J, Dalley,J.W.,Morrison,C.H., Everitt BJ, Robbins TW. Selective behavioural neurochemical effects of cholinergic lesions produced by intrabasalis infusions of 192 IgG-saparin on attentional performance in a five-choice serial reaction time task. J Neurosci 2002;22:1905-1913. [91] Connor JM, Culberson A, Packowski C, Chiba AA, Tuszynski MH. Lesions of the basal forebrain cholinergic system impair task acquisition and abolish cortical plasticity associated with motor skill learning. Neuron 2003;38:819-29. [92] Recanzone GH, Merzenich MM, Schreiner CE. Changes in the distributed temporal response properties of S1 cortical neurons reflect improvements in performance on a temporally based tactile discrimination task. J Neurophys 1992;67:1071-1091. [93] Chapman CE, Jiang W, Lamarre Y. Modulation of lemniscal input during conditioned arm movements in the monkey. Experimental Brain Research 1988;72:316-34. [94] Zikopoulos B, Barbas H. Pathways for emotions and attention converge on the thalamic reticular nucleus in primates. J Neurosci 2012;32:5338-5350. [95] Rosenkranz K, Rothwell JC. Spatial attention affects sensorimotor reorganisation in human motor cortex. Exp Brain Res 2006;170:97-108. [96] Lapole T, Tindel J. Acute effects of muscle vibration on sensorimotor integration. Neurosci Lett 2015;587:46-50. [97] Rosenkranz K, Rothwell JC. Modulation of proprioceptive integration in the motor cortex shapes human motor learning. J Neurosci 2012;32:9000-6. [98] Mang CS, Bergquist AJ, Roshko SM, Collins DF. Loss of short-latency afferent inhibition and emergence of afferent facilitation following neuromuscular electrical stimulation. Neurosci Lett 2012;529:80-5. [99] Freyer F, Reinacher M, Nolte G, Dinse HR, Ritter P. Repetitive tactile stimulation changes resting-state functional connectivity-implications for treatment of sensorimotor decline. Front Hum Neurosci 2012;6:144. [100] Rogasch NC, Dartnall TJ, Cirillo J, Nordstrom MA, Semmler JG. Corticomotor plasticity and learning of a ballistic thumb training task are diminished in older adults. J Appl Physiol (1985) 2009;107:1874-83. [101] Akopian G, Walsh JP. Pre- and postsynaptic contributions to age-related alterations in corticostriatal synaptic plasticity. Synapse 2006;60:223-38. [102] Burke SN, Barnes CA. Neural plasticity in the ageing brain. Nat Rev Neurosci 2006;7:30-40. [103] Foster TC. Calcium homeostasis and modulation of synaptic plasticity in the aged brain. Aging Cell 2007;6:319-25.   147[104] Liepert J, Miltner WH, Bauder H, Sommer M, Dettmers C, Taub E et al. Motor cortex plasticity during constraint-induced movement therapy in stroke patients. Neurosci Lett 1998;250:5-8. [105] Castel-Lacanal E, Marque P, Tardy J, de Boissezon X, Guiraud V, Chollet F et al. Induction of cortical plastic changes in wrist muscles by paired associative stimulation in the recovery phase of stroke patients. Neurorehabil Neural Repair 2009;23:366-72. [106] Adkins DL, Hsu JE, Jones TA. Motor cortical stimulation promotes synaptic plasticity and behavioral improvements following sensorimotor cortex lesions. Exp Neurol 2008;212:14-28. [107] Meehan SK, Randhawa B, Wessel B, Boyd LA. Implicit sequence specific motor learning after subcortical stroke is associated with increased prefrontal brain activations: an fMRI study. Hum Brain Mapp 2011;32:290-303. [108] Brodie SM, Meehan S, Borich MR, Boyd LA. 5 Hz repetitive transcranial magnetic stimulation over the ipsilesional sensory cortex enhances motor learning after stroke. Front Hum Neurosci 2014;8:143. [109] Etoh S, Noma T, Takiyoshi Y, Arima M, Ohama R, Yokoyama K et al. Effects of repetitive facilitative exercise with neuromuscular electrical stimulation, vibratory stimulation and repetitive transcranial magnetic stimulation of the hemiplegic hand in chronic stroke patients. Int J Neurosci 2016;126:1007-12. [110] Meehan SK, Dao E, Linsdell MA, Boyd LA. Continuous theta burst stimulation over the contralesional sensory and motor cortex enhances motor learning post-stroke. Neurosci Lett 2011;500:26-30. [111] de Vet HC, Terwee CB, Knol DL, Bouter LM. When to use agreement versus reliability measures. J Clin Epidemiol 2006;59:1033-9. [112] de Vet HC, Terwee CB, Mokkink LB, Knol DL. Measurement in Medicine: A Practical Guide. New York, NY: Cambridge University Press, 2011. [113] Guyatt G, Walter S, Norman G. Measuring change over time: assessing the usefulness of evaluative instruments. J Chronic Dis 1987;40:171-8. [114] Terwee CB, Bot SD, de Boer MR, van der Windt DA, Knol DL, Dekker J et al. Quality criteria were proposed for measurement properties of health status questionnaires. J Clin Epidemiol 2007;60:34-42. [115] Beaulieu LD, Flamand VH, Masse-Alarie H, Schneider C. Reliability and minimal detectable change of transcranial magnetic stimulation outcomes in healthy adults: A systematic review. Brain Stimul 2016. [116] Schambra HM, Ogden RT, Martinez-Hernandez IE, Lin X, Chang YB, Rahman A et al. The reliability of repeated TMS measures in older adults and in patients with subacute and chronic stroke. Front Cell Neurosci 2015;9:335.   148[117] Zumsteg D, Wieser HG. Effects of aging and sex on middle-latency somatosensory evoked potentials: normative data. Clin Neurophysiol 2002;113:681-5. [118] Park BK, Chae J, Lee YH, Yang G, Labatia I. Median nerve somatosensory evoked potentials and upper limb motor function in hemiparesis. Electromyogr Clin Neurophysiol 2003;43:169-79. [119] Fierro B, La Bua V, Oliveri M, Daniele O, Brighina F. Prognostic value of somatosensory evoked potentials in stroke. Electromyogr Clin Neurophysiol 1999;39:155-60. [120] Tokimura H, Di Lazzaro V, Tokimura Y, Oliviero A, Profice P, Insola A et al. Short latency inhibition of human hand motor cortex by somatosensory input from the hand. J Physiol 2000;523 Pt 2:503-13. [121] Fugl-Meyer AR, Jaasko L, Leyman I, Olsson S, Steglind S. The post-stroke hemiplegic patient. 1. a method for evaluation of physical performance. Scand J Rehabil Med 1975;7:13-31. [122] Hodics TM, Nakatsuka K, Upreti B, Alex A, Smith PS, Pezzullo JC. Wolf Motor Function Test for characterizing moderate to severe hemiparesis in stroke patients. Arch Phys Med Rehabil 2012;93:1963-7. [123] Rossini PM, Barker AT, Berardelli A, Caramia MD, Caruso G, Cracco RQ et al. Non-invasive electrical and magnetic stimulation of the brain, spinal cord and roots: basic principles and procedures for routine clinical application. Report of an IFCN committee. Electroencephalogr Clin Neurophysiol 1994;91:79-92. [124] Rossini PM, Barker AT, Berardelli A, Caramia MD, Caruso G, Cracco RQ et al. Non-invasive electrical and magnetic stimulation of the brain, spinal cord and roots: basic principles and procedures for routine clinical application. Report of an IFCN committee. Electroencephalogr Clin Neurophysiol 1994;91:79-92. [125] Chen R. Interactions between inhibitory and excitatory circuits in the human motor cortex. Exp Brain Res 2004;154:1-10. [126] Manganotti P, Zanette G, Bonato C, Tinazzi M, Polo A, Fiaschi A. Crossed and direct effects of digital nerves stimulation on motor evoked potential: a study with magnetic brain stimulation. Electroencephalogr Clin Neurophysiol 1997;105:280-9. [127] Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Hillsdale, New Jersey: Lawrence Erlbaum, 1988. [128] Gamst G, Meyers LS, Guarino AJ. Analysis of Variance Designs: A Conceptual and Computational Approach with SPSS and SAS. New York, New York, USA: Cambridge University Press, 2008. [129] Field A. Discovering statistics using SPSS: And sex, drugs, and rock n' roll. London: Sage, 2009. [130] Carmichael ST. Brain excitability in stroke: the yin and yang of stroke progression. Arch Neurol 2012;69:161-7.   149[131] Turco CV, El-Sayes J, Fassett HJ, Chen R, Nelson AJ. Modulation of long-latency afferent inhibition by the amplitude of sensory afferent volley. J Neurophysiol 2017:jn.00118.2017. [132] Ohno S, Kuramoto E, Furuta T, Hioki H, Tanaka YR, Fujiyama F et al. A morphological analysis of thalamocortical axon fibers of rat posterior thalamic nuclei: a single neuron tracing study with viral vectors. Cereb Cortex 2012;22:2840-57. [133] Strick PL, Sterling P. Synaptic termination of afferents from the ventrolateral nucleus of the thalamus in the cat motor cortex. A light and electron microscopy study. J Comp Neurol 1974;153:77-106. [134] Asanuma H, Larsen KD, Yumiya H. Receptive fields of thalamic neurons projecting to the motor cortex in the cat. Brain Res 1979;172:217-28. [135] Hoffken O, Veit M, Knossalla F, Lissek S, Bliem B, Ragert P et al. Sustained increase of somatosensory cortex excitability by tactile coactivation studied by paired median nerve stimulation in humans correlates with perceptual gain. J Physiol 2007;584:463-71. [136] Rosenkranz K, Rothwell JC. The effect of sensory input and attention on the sensorimotor organization of the hand area of the human motor cortex. J Physiol 2004;561:307-20. [137] Chen R, Lozano AM, Ashby P. Mechanism of the silent period following transranial magnetic stimulation. Evidence from epidural recordings. Exp Brain Res 1999;128:539-42. [138] Tempel LW, Perlmutter JS. Vibration-induced regional cerebral blood flow responses in normal aging. J Cereb Blood Flow Metab 1992;12:554-61. [139] Eytan D, Brenner N, Marom S. Selective adaptation in networks of cortical neurons. J Neurosci 2003;23:9349-56. [140] Gowland C. Recovery of motor function following stroke: Profile and prediction. Physiother Can 1982;34:77,78-84. [141] Nakayama H, Jorgensen HS, Raaschou HO, Olsen TS. Recovery of upper extremity function in stroke patients: the Copenhagen Stroke Study. Arch Phys Med Rehabil 1994;75:394-8. [142] Bailey AZ, Asmussen MJ, Nelson AJ. Short-latency afferent inhibition determined by the sensory afferent volley. J Neurophysiol 2016;116:637-44. [143] Bernhardt J, Hayward KS, Kwakkel G, Ward NS, Wolf SL, Borschmann K et al. Agreed definitions and a shared vision for new standards in stroke recovery research. The Stroke Recovery and Rehabilitation Roundtable taskforce. Int J Stroke 2017. [144] Bolognini N, Russo C, Edwards DJ. The sensory side of post-stroke motor rehabilitation. Restor Neurol Neurosci 2016;34:571-86. [145] Tarlaci S, Turman B, Uludag B, Ertekin C. Differential effects of peripheral vibration on motor-evoked potentials in acute stages of stroke. Neuromodulation 2010;13:232-7.   150[146] Carey LM, Seitz RJ. Functional neuroimaging in stroke recovery and neurorehabilitation: conceptual issues and perspectives. Int J Stroke 2007;2:245-64. [147] Hodics T, Cohen LG, Cramer SC. Functional imaging of intervention effects in stroke motor rehabilitation. Arch Phys Med Rehabil 2006;87:S36-42. [148] Costantino C, Galuppo L, Romiti D. Short-term effect of local muscle vibration treatment versus sham therapy on upper limb in chronic post-stroke patients: a randomized controlled trial. Eur J Phys Rehabil Med 2017;53:32-40. [149] Liepert J, Hamzei F, Weiller C. Motor cortex disinhibition of the unaffected hemisphere after acute stroke. Muscle Nerve 2000;23:1761-3. [150] Sim SM, Oh DW, Chon SC. Immediate effects of somatosensory stimulation on hand function in patients with poststroke hemiparesis: a randomized cross-over trial. Int J Rehabil Res 2015;38:306-12. [151] Kaas JH. Plasticity of sensory and motor maps in adult mammals. Ann Rev Neurosci 1991;14:137-167. [152] Phadke CP, Robertson CT, Condliffe EG, Patten C. Upper-extremity H-reflex measurement post-stroke: reliability and inter-limb differences. Clin Neurophysiol 2012;123:1606-15. [153] Claus D, Harding AE, Hess CW, Mills KR, Murray NM, Thomas PK. Central motor conduction in degenerative ataxic disorders: a magnetic stimulation study. J Neurol Neurosurg Psychiatry 1988;51:790-5. [154] Claus D, Mills KR, Murray NM. Interaction of transcranial magnetic stimulation and mechanical stimuli. EEG EMG Z Elektroenzephalogr Elektromyogr Verwandte Geb 1988;19:222-7. [155] Pizzi A, Carlucci G, Falsini C, Verdesca S, Grippo A. Evaluation of upper-limb spasticity after stroke: A clinical and neurophysiologic study. Arch Phys Med Rehabil 2005;86:410-5. [156] Marconi B, Filippi GM, Koch G, Giacobbe V, Pecchioli C, Versace V et al. Long-term effects on cortical excitability and motor recovery induced by repeated muscle vibration in chronic stroke patients. Neurorehabil Neural Repair 2011;25:48-60. [157] Beaulieu LD, Masse-Alarie H, Camire-Bernier S, Ribot-Ciscar E, Schneider C. After-effects of peripheral neurostimulation on brain plasticity and ankle function in chronic stroke: The role of afferents recruited. Neurophysiol Clin 2017. [158] Boniface S, Ziemann U. Plasticity in the Human Nervous System: Investigations with Transcranial Magnetic Stimulation. 2003. [159] Rossini PM, Burke D, Chen R, Cohen LG, Daskalakis Z, Di Iorio R et al. Non-invasive electrical and magnetic stimulation of the brain, spinal cord, roots and peripheral nerves: Basic principles and procedures for routine clinical and research application. An updated report from an I.F.C.N. Committee. Clin Neurophysiol 2015;126:1071-107.   151[160] Kloppel S, Gregory S, Scheller E, Minkova L, Razi A, Durr A et al. Compensation in Preclinical Huntington's Disease: Evidence From the Track-On HD Study. EBioMedicine 2015;2:1420-9. [161] Orth M, Gregory S, Scahill RI, Mayer IS, Minkova L, Kloppel S et al. Natural variation in sensory-motor white matter organization influences manifestations of Huntington's disease. Hum Brain Mapp 2016. [162] Oldfield RC. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 1971;9:97-113. [163] Deuschl G, Eisen A. Long-latency reflexes following electrical nerve stimulation. The Inter national Federation of Clinical Neurophysiology. Electroencephalogr Clin Neurophysiol Suppl 1999;52:263-8. [164] Orth M, Rothwell JC. The cortical silent period: intrinsic variability and relation to the waveform of the transcranial magnetic stimulation pulse. Clin Neurophysiol 2004;115:1076-82. [165] Schippling S, Schneider SA, Bhatia KP, Munchau A, Rothwell JC, Tabrizi SJ et al. Abnormal motor cortex excitability in preclinical and very early Huntington's disease. Biol Psychiatry 2009;65:959-65. [166] Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull 1979;86:420-8. [167] Trevethan R. Intraclass correlation coefficients: Clearing the air, extending some cautions, and making some requests. Health Serv Outcomes Res Methodol 2016:doi:10.1007/s10742-016-0156-6. [168] Kline P. The handbook of psychological testing. London: Routledge, 2000. [169] Faul F, Erdfelder E, Lang AG, Buchner A. G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 2007;39:175-91. [170] Koski L, Schrader LM, Wu AD, Stern JM. Normative data on changes in transcranial magnetic stimulation measures over a ten hour period. Clin Neurophysiol 2005;116:2099-109. [171] Ellaway PH, Davey NJ, Maskill DW, Rawlinson SR, Lewis HS, Anissimova NP. Variability in the amplitude of skeletal muscle responses to magnetic stimulation of the motor cortex in man. Electroencephalogr Clin Neurophysiol 1998;109:104-13. [172] Truccolo WA, Ding M, Knuth KH, Nakamura R, Bressler SL. Trial-to-trial variability of cortical evoked responses: implications for the analysis of functional connectivity. Clin Neurophysiol 2002;113:206-26. [173] Wassermann EM. Variation in the response to transcranial magnetic brain stimulation in the general population. Clin Neurophysiol 2002;113:1165-71. [174] Malcolm MP, Triggs WJ, Light KE, Shechtman O, Khandekar G, Gonzalez Rothi LJ. Reliability of motor cortex transcranial magnetic stimulation in four muscle representations. Clin Neurophysiol 2006;117:1037-46.   152[175] Ngomo S, Leonard G, Moffet H, Mercier C. Comparison of transcranial magnetic stimulation measures obtained at rest and under active conditions and their reliability. J Neurosci Methods 2012;205:65-71. [176] Hermsen AM, Haag A, Duddek C, Balkenhol K, Bugiel H, Bauer S et al. Test-retest reliability of single and paired pulse transcranial magnetic stimulation parameters in healthy subjects. J Neurol Sci 2016;362:209-16. [177] McDonnell MN, Ridding MC, Miles TS. Do alternate methods of analysing motor evoked potentials give comparable results? J Neurosci Methods 2004;136:63-7. [178] Siebner HR, Hartwigsen G, Kassuba T, Rothwell JC. How does transcranial magnetic stimulation modify neuronal activity in the brain? Implications for studies of cognition. Cortex 2009;45:1035-42. [179] McClelland GH. Increasing statistical power without increasing sample size. Am Psychol 2000;55:963-4. [180] Jones SJ, Halonen JP, Shawkat F. Centrifugal and centripetal mechanisms involved in the 'gating' of cortical SEPs during movement. Electroencephalogr Clin Neurophysiol 1989;74:36-45. [181] Brown KE, Ferris JK, Amanian MA, Staines WR, Boyd LA. Task-relevancy effects on movement-related gating are modulated by continuous theta-burst stimulation of the dorsolateral prefrontal cortex and primary somatosensory cortex. Exp Brain Res 2015;233:927-36. [182] Stinear CM, Barber PA, Petoe M, Anwar S, Byblow WD. The PREP algorithm predicts potential for upper limb recovery after stroke. Brain 2012;135:2527-35. [183] Hayward KS, Schmidt J, Lohse KR, Peters S, Bernhardt J, Lannin NA et al. Are we armed with the right data? Pooled individual data review of biomarkers in people with severe upper limb impairment after stroke. Neuroimage Clin 2016;13:310-9. [184] Grenier E. Canadian seniors now outnumber children for first time, 2016 census shows. CBC 2017.   153Appendices  Appendix A    A.1 Statistical rationale In various theories of measurement, any individual observed score, X, can be conceptualised as the result of some underlying true score, T, plus error, ε [112]. Assuming that errors are random, the true score can be thought of as the average of an infinite number of measurements. ௜ܺ௝ ൌ ௜ܶ ൅ ߝ௜௝ For individual participants over time, each participant’s (i’s) observed score is the result of that participant’s true score and measurement error (j’s) at each time point. While we can collectively model this deviance as “random” measurement error, in reality this variance could systematically be explained by differences between individuals or the environment. For clinical measurements, reducing measurement error (i.e., increasing reliability) is critical not only for diagnostic accuracy (i.e., better discrimination between individuals or within individuals over time) but also for experimental research, because poor reliability has a negative effect on statistical power.  Although there are various methods for assessing reliability, we chose to focus on the Intraclass Correlations (ICCs). Broadly speaking, the ICC represents the proportion of the variance in the true scores that is captured by a set of repeated measurements. That is, because the variance of sums is equal to the sum of the variances: ݒܽݎሺܺሻ ൌ ݒܽݎሺܶሻ ൅ ݒܽݎሺߝሻ It follows that:   154ܫܥܥ ൌ 	 ݒܽݎሺܶሻݒܽݎሺܺሻ ൌݒܽݎሺܶሻݒܽݎሺܶሻ ൅ ݒܽݎሺߝሻ	 Thus, the ICC is essentially the ratio of between-individual variation to the total amount of variation within the data, and measurements with no within-individual variation have an ICC = 1.0. That said, there are several different calculations of the ICC that make different assumptions about the data. The specific ICCs that we calculated are explained below, but this general definition of the ICC as a measure of reliability applies to all calculations.    155 A.2 Additional results  Measure Time1 Mean (SD) Time2 Mean (SD) Time3 Mean (SD) Exclusions* ICC (2,k) ICC (2,1) Low reliability       110% RMT Amp (mV) 0.44 (0.55) 0.36 (0.37) 0.34 (0.31) 32 (38) 0.37 0.17 110% RMT Area (mV ms) 2.55 (2.15) 1.99 (1.47) 1.61 (1.32) 35 (42) 0.32 0.13 125% AMT Area (mV ms) 13.76 (8.78) 10.06 (6.26) 7.69 (5.96) 38 (45) 0.27 0.11 150% AMT Area (mV ms) 25.89 (13.10) 19.51 (10.29) 17.26 (10.12) 44 (52) 0.39 0.17 175% AMT Area (mV ms) 36.15 (21.61) 27.02 (11.43) 22.64 (11.87) 48 (57) 0.21 0.08 125% AMT SPD (ms) 84.08 (35.59) 81.91 (28.29) 76.33 (31.86) 38 (45) 0.23 0.07 125% AMT SP ratio 0.16 (0.07) 0.12 (0.06) 0.10 (0.06) 38 (45) 0.40 0.18 150% AMT SPD (ms) 132.96 (45.96) 125.91 (36.19) 126.39 (39.95) 44 (52) 0.37 0.16 150% AMT SP ratio  0.19 (0.06) 0.15 (0.06) 0.14 (0.06) 44 (52) 0.41 0.19 175% AMT SP Ratio  0.21 (0.10) 0.17 (0.06) 0.15 (0.07) 48 (57) 0.16 0.06 AF N34 Amplitude (mV) 1.34 (1.21) 1.28 (1.24)  45 (54) 0.48 0.32 SAI N22 Area (mV ms) 71.60 (38.71) 77.23 (33.27)  40 (48) 0.30 0.17 SAI N24 Area (mV ms) 78.50 (35.45) 85.22 (46.2)  40 (48) 0.02 0.01 AF N34 Area (mV ms) 136.46 (78.22) 141.18 (73.27)  40 (48) 0.44 0.29 SAI N22 Ratio 0.79 (0.32) 0.89 (0.64)  46 (55) 0.49 0.32 AF N32 Ratio 1.38 (0.61) 1.22 (0.54)  45 (54) 0.26 0.15 AF N34 Ratio 1.37 (0.68) 1.20 (0.56)  45 (54) 0.15 0.08 Table A.1: Low reliability measures. Denotes mean (standard deviation) values for dependent measures with low reliability (ICC(2,k) ≤ 0.5). For each measure, all time-points, as well as exclusions, ICC(2,k), and ICC(2,1) are shown. ‘Exclusions’ refers to participants who had to be excluded due to missing data (shown as count (% of total n)). * note that the Leiden site only recorded SEPs hence the large number of excluded RMT: resting motor threshold, AMT: active motor threshold, SPD: silent period duration, SAI: short-latency afferent inhibition, SAF: short afferent facilitation. 22, 24, 32, 34 represent inter-stimulus intervals between the nerve stimulation and TMS.    156 Measure V df p London Paris Vancouver Leiden Low Reliability        110 RMT Area (mV ms) 0.43 6, 102 <0.001 1.90 (0.92) 1.50 (0.86) 2.33 (1.06) N/A 125 AMT Area (mV ms) 0.41 6, 96 <0.001 11.92 (4.30) 10.44 (5.48) 8.46 (3.46) N/A 150 AMT Area (mV ms) 0.52 6, 86 <0.001 22.63 (7.29) 19.55 (7.35) 17.61 (8.01) N/A 175 AMT Area (mV ms) 0.39 6, 80 0.01 30.67 (13.41) 27.15 (7.55) 25.89 (10.24) N/A 125 AMT SPD (ms) 0.39 6, 96 0.002 86.04 (17.29) 84.27 (18.09) 74.35 (17.26) N/A 150 AMT SPD (ms) 0.59 6, 86 <0.001 132.97 (28.33) 131.42 (23.13) 114.17 (29.92) N/A 125 AMT Ratio 0.36 6, 96 0.003 0.13 (0.04) 0.12 (0.05) 0.12 (0.04) N/A 150 AMT Ratio 0.53 6, 86 <0.001 0.17 (0.04) 0.15 (0.04) 0.15 (0.04) N/A SAI 24 Area (mV ms) 0.20 4, 96 0.04 0.96 (0.57) 0.87 (0.60) 0.66 (0.74) N/A AF 34 Ratio 0.23 4, 86 0.03 1.40 (0.43) 1.18 (0.33) 1.09 (0.47) N/A Table A.2: Between site effects for low reliability measures. This table shows significant effects study site in a MANOVA for electrophysiological measures with low reliability. Mean (SD) are shown for each study site averaging across time points. Only participants with data from all three time-points were included in this analysis. Leiden data was only available for EEG measures. RMT: resting motor threshold; AMT: active motor threshold; SPD: silent period duration; SAI: short-latency afferent inhibition; SAF: short-latency afferent facilitation. 22, 24, 32, 34 denote inter-stimulus intervals between the nerve stimulation and TMS.   157 Measure V df p London Paris Vancouver Leiden High Reliability       RMT (%MSO) 0.20 6, 108 0.07 45.73 (10.55) 42.60 (8.32) 49.31 (8.54) N/A AMT (%MSO) 0.12 6, 110 0.34 36.32 (7.42) 33.48 (6.02) 37.52 (6.87) N/A Active Latency (ms) 0.15 6, 94 0.26 20.07 (1.77) 21.00 (1.82) 21.38 (1.34) N/A 150 RMT Amp (mV) 0.09 6, 92 0.67 2.50 (1.83) 2.42 (1.89) 1.79 (1.63) N/A N20 Latency (ms)  0.17 9, 189 0.27 19.53 (1.26) 20.13 (1.27) 19.86 (0.79) 20.20 (1.19) LLR1 Latency 0.20 6, 68 0.29 37.05 (3.26) 39.12 (3.66) 38.15 (1.83) N/A LLR2 Latency 0.25 6, 68 0.15 47.86 (3.51) 50.41 (3.65) 50.13 (2.39) N/A Moderate Reliability       SEP Amp ST (V) 0.23 9, 165 0.15 0.74 (0.37) 0.42 (0.47) 0.76 (0.38) 0.93 (0.72) LLR2 Amp (mV) 0.15 6, 70 0.49 0.10 (0.05) 0.08 (0.03) 0.07 (0.07) N/A CRT (ms) 0.19 6, 54 0.47 7.81 (1.85) 8.10 (1.85) 9.56 (0.95) N/A 130 RMT Amp (mV) 0.10 6, 98 0.53 1.34 (1.04) 1.44 (1.11) 1.18 (0.99) N/A 175 AMT Amp (mV) 0.20 6, 82 0.19 4.44 (1.32) 4.02 (2.05) 4.96 (3.19) N/A 150 RMT Area (mV ms) 0.20 6, 88 0.14 10.87 (7.35) 12.36 (7.54) 6.68 (5.23) N/A  SAI22 Amp (mV) 0.02 4, 84 0.94 79.02 (28.71) 70.06 (24.94) 68.28 (24.53) N/A  SAI24 Amp (mV) 0.05 4, 86 0.71 93.79 (32.42) 74.77 (18.94) 70.64 (28.47) N/A  AF32 Amp (mV)  0.14 4, 86 0.18 163.78 (60.94) 124.04 (44.22) 114.20 (69.06) N/A  AF34 Amp (mV) 0.13 4, 86 0.22 160.58 (59.81) 136.70 (70.98) 112.11 (62.38) N/A Low Reliability      110 RMT Amp (mV) 0.11 6, 106 0.42 0.31 (0.24) 0.43 (0.33) 0.41 (0.29) N/A 175 AMT Ratio 0.29 6, 80 0.15 0.19 (0.07) 0.16 (0.04) 0.18 (0.04) N/A SAI22 Area (mV ms) 0.07 4, 96 0.43 0.88 (0.55) 0.75 (0.54) 0.77 (0.85) N/A AF32 Area (mV ms) 0.15 4, 96 0.12 1.46 (0.66) 1.49 (1.23) 1.00 (1.08) N/A AF34 Area (mV ms) 0.09 4, 96 0.33 1.49 (0.74) 1.38 (1.09) 0.88 (0.96) N/A SAI22 Ratio 0.07 4, 84 0.55 0.76 (0.35) 0.82 (0.50) 0.99 (0.47) N/A SAI24 Ratio 0.12 4, 86 0.25 0.90 (0.36) 0.83 (0.34) 0.83 (0.48) N/A SAI32 Ratio  0.14 4, 86 0.18 1.40 (0.45) 1.25 (0.40) 1.23 (0.40) N/A Table A.3: Non-significant study site effects.This table shows non-significant study site effects from MANOVA results. For latency variables, the MANOVA controlled for the arm length of each participant. Mean (SD) are shown for each study site averaging across time points. Only participants with data from all three time-points were included in this analysis. Leiden data was only available for EEG measures. SEP: somatosensory evoked potential; ST: sensory threshold; LLR: long-latency reflex; CRT: cortical relay time; RMT: resting motor threshold; AMT: active motor threshold;   158SAI: short-latency afferent inhibition; SAF: short-latency afferent facilitation. 22, 24, 32, 34 denote interstimulus intervals between the nerve stimulation and TMS.  159Post-hoc ANOVAs for significant between-site effects High reliability For SEP Amp MT, at Time 1, Leiden was significantly different from London and Paris (ps < 0.006), but not Vancouver (p = 0.24). Neither London, Paris, nor Vancouver were statistically different from each other. At Time 2, Leiden was again different from London and Paris (ps < 0.005), but not Vancouver (p = 0.15). Neither London, Paris, nor Vancouver were statistically different from each other. At Time 3, Leiden was significantly different from London, Paris, and Vancouver (ps < 0.04), but these three groups were not statistically different from each other. For SEP Amp 150, at Time 1, Leiden was significantly different from London and Paris (ps < 0.001), but not from Vancouver (p = 0.94). At Time 2, Leiden was significantly different from London and Paris (ps < 0.011), but none of the other sites were statistically different from each other. At Time 3, Leiden was again statistically different from London and Paris (ps < 0.04), London was statistically different from Paris (p 0.03), and Vancouver was not statistically different from any of the other study sites. For Rest Latency, at Time 1, there was a significant difference between Paris and Vancouver (p = 0.02), but there were no other significant differences between study sites (ps > 0.11). At Time 2, there was a significant difference between London and Paris (p = 0.005), but there were no other significant differences between study sites (ps > 0.19). For Time 3, there was a significant difference between London and Paris (p = 0.001), but there were no other significant differences between study sites (ps > 0.09).      160Moderate reliability  For 125 AMT Amp, at Time 1, London was significantly different from Paris and Vancouver (ps < 0.001), which were not different from each other (p = 0.59). At Time 2, there were no significant differences between sites (ps > 0.26). At Time 3, there were no significant differences between study sites (ps > 0.05). For 150 AMT Amp, at Time 1, London was significantly different from Paris and Vancouver (ps = 0.001), which were not significantly different from each other (p = 0.63). At Time 2, there were no significant differences between study sites (ps > 0.53). At Time 3, there were no significant differences between study sites (ps > 0.71). For 175 AMT SPD, at Time 1, Vancouver was significantly different from London (p < 0.001) and Paris (p < 0.001), which were not different from each other (p = 0.72). At Time 2, there were no significant differences between study sites (ps > 0.44). At Time 3, there was a significant difference between London and Paris (p = 0.04) but no other significant differences between study sites (ps > 0.16).  Low Reliability  For 110 RMT Area, at Time 1, there were no significant differences between study sites (ps >0.13). At Time 2, London was significantly different from Paris (p = 0.02) and Vancouver (p = 0.02), which were significantly different from each other (p < 0.001). At Time 3, Vancouver was different from London (p = 0.01) and Paris (p = 0.03), which were no different from each other (p = 0.73). For 130 RMT Area, at Time 1, Paris was significantly different from London (p = 0.4) and Vancouver (p = 0.003), which were not different from each other (p = 0.24). At Time 2, there   161were no significant differences between sites (ps > 0.43). At Time 3, there were no significant differences between groups (ps > 0.31). For AMT 125 Area, at Time 1, Vancouver was significantly different from London and Paris (ps < 0.001), but London and Paris were not different from each other (p = 0.86). At Time 2, there were no significant differences between study sites (ps > 0.06). At Time 3, there were also no significant differences between study sites (ps > 0.34). For AMT 150 Area, at Time 1, there was a significant difference between London and Vancouver (p < 0.001), but no significant difference between London and Paris (p = 0.39). At Time 2, there were no statistically significant differences between groups (ps > 0.20). At Time 3, there was a statistically significant difference between Vancouver and London (p = 0.04) and Paris (p = 0.03), but no statistically significant difference between London and Paris (p = 0.71).  For AMT 175 Area, at Time 1, there was a significant difference between Vancouver and London (p = 0.03), but not London and Paris (p = 0.68). At Time 2, there were no statistically significant differences between sites (ps > 0.09). At Time 3, Vancouver was significantly different from London (p = 0.008) and Paris (p = 0.009), but London and Paris were not different from each other (p = 0.94). For 125 AMT SPD, at Time 1, Vancouver was significantly different from London (p < 0.001) and Paris (p = 0.03), which were also different from each other (p = 0.01). At Time 2, there was no statistically significant differences between groups (ps > 0.63). At Time 3, London was significantly different from Vancouver (p = 0.04) and Paris (0.03), which were not significantly different from each other.  For 150 AMT SPD, at Time 1, Vancouver was significantly different from London (p < 0.001) and Paris (p = 0.002), which were different from each other (p = 0.004). At Time 2, there   162were no statistically significant differences between study sites (ps > 0.74). At Time 3, there we no statistically significant differences between study sites (ps > 0.05). For 125 AMT Ratio, at Time 1, Vancouver was significantly different from Paris (p = 0.001), but no other differences were significant (p > 0.05). At Time 2, London and Paris were significantly different (p = 0.02), but no other differences were significant (ps > 0.09). At Time 3, there were no statistically significant differences between groups (ps > 0.05).  For 150 AMT Ratio, at Time 1, Vancouver was significantly different from London (p = 0.03) and Paris (p = 0.005), which were not different from each other (p = 0.30). At Time 2, London and Paris were significantly different from each other (p = 0.02), but no other differences were significant (ps > 0.06). At Time 3, Vancouver was significantly different from Paris (p = 0.01), but no other differences were statistically significant (ps > 0.11).  For SAI 24 Area, at Time 1, there were no significant differences between groups (ps > 0.88). At Time 2, London was significantly different from Paris (p = 0.005) and Vancouver (p = 0.04), which were not different from each other (p = 0.66). For AF 34 Ratio, at Time 1, Vancouver was not significantly different from London (p = 0.41) or Paris (0.24), although those two sites were different from each other (p = 0.03). At Time 2, Vancouver was significantly different from both London (p = 0.04) and Paris (p = 0.04), but these sites were not different from each other (p = 0.91).     163 Effect of Site in MANOVA      Measure V df p Time London Paris Vancouver Leiden SEP-ST (mA) 0.76 (6,116) <0.001 V1 2.43 (1.04) 3.23 (1.99) N/A 4.19 (1.18)     V2 2.08 (0.39) 3.93 (1.59) N/A 3.72 (0.97)     V3 1.94 (0.37) 2.81 (1.00) N/A 4.20 (1.45) SEP-MT (mA) 1.03 (6,116) <0.001 V1 4.29 (1.48) 5.99 (3.00) N/A 7.87 (2.67)     V2 3.58 (0.85) 9.08 (2.78) N/A 8.50 (2.28)     V3 3.17 (0.64) 5.79 (2.49) N/A 8.63 (2.17) SEP-150 (mA) 0.91 (6,122) <0.001 V1 6.09 (1.97) 7.45 (3.75) N/A 10.76 (3.47)     V2 5.19 (1.23) 11.55 (4.01) N/A 11.90 (3.36)     V3 4.61 (0.87) 7.71 (3.67) N/A 12.06 (3.19) Table A.4: Between site effects of stimulation intensities. Stimulation intensities across study sites and time are shown as mean (SD). Vancouver is not included as stimulation intensities at that site were quantified in V and thus a direct comparison cannot be made.  Post-hoc ANOVAs for significant between-site effects For the SEP-ST, follow-up ANOVAs revealed significant effects of study site at V1, V2, and V3 (ps < 0.01). At V1, Leiden had significantly higher stimulation intensities than London (p < 0.001) and Paris (p = 0.03), which were not significantly different from each other (p = 0.09). At V2, London had significantly lower stimulation intensities than Paris (p < 0.001) and Leiden (p < 0.001), which were not significantly different from each other (p = 0.52). At V3, Leiden had significantly greater stimulation intensities than London (p < 0.001) and Paris (p < 0.001), and Paris had significantly greater stimulation intensities than London (p = 0.02). For the SEP-MT, follow-up ANOVAs revealed significant effects of study site at V1, V2, and V3 (ps < 0.001). At V1, Leiden had significantly higher stimulation intensities than London (p < 0.001) and Paris (p = 0.02), and Paris had marginally greater stimulation intensity than London (p = 0.04). At V2, London had significantly lower stimulation intensities than Paris (p < 0.001) and Leiden (p < 0.001), which were not significantly different from each other (p = 0.38). At V3, Leiden had significantly greater stimulation intensities than London (p < 0.001) and Paris (p < 0.001), and Paris had significantly greater stimulation intensities than London (p = 0.02).   164For the SEP-150, follow-up ANOVAs revealed significant effects of study site at V1, V2, and V3 (ps < 0.001). At V1, Leiden had significantly higher stimulation intensities than London (p < 0.001) and Paris (p = 0.001), which were not significantly different from each other (p = 0.18). At V2, London had significantly lower stimulation intensities than Paris (p < 0.001) and Leiden (p < 0.001), which were not significantly different from each other (p = 0.71). At V3, Leiden had significantly greater stimulation intensities than London (p < 0.001) and Paris (p < 0.001), and Paris had significantly greater stimulation intensities than London (p = 0.001).    

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:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0355869/manifest

Comment

Related Items