manson gerome a 201906 phd thesis - university of toronto t … · manzone, animesh kumawat, john...
TRANSCRIPT
!Examining!the!Sensorimotor!Integration!Processes!Prior!to!and!During!Movements!to!Somatosensory!Targets!
! by! !
Gerome!Aleandro!Manson!
A!thesis!submitted!in!conformity!with!the!requirements!for!the!degree!of!Doctor!of!Philosophy!
Graduate!Department!of!Exercise!Science!University!of!Toronto!
Joint!with!École!Doctorale!des!Sciences!de!la!Vie!et!de!la!Santé!
AixGMarseille!Université!©!Copyright!by!Gerome!Aleandro!Manson!2019!
ii
Examining!the!Sensorimotor!Integration!Processes!Prior!to!and!
During!Movements!to!Somatosensory!Targets!
Gerome Aleandro Manson
Doctor of Philosophy
University!of!Toronto!AixGMarseille!Université!!
2019
Abstract!
Previous research on multisensory integration for movement planning and control has focused on
movements to targets external to the body. In this dissertation, three experiments were conducted
to examine the sensorimotor transformation processes underlying goal-directed actions to targets
defined by body positions (i.e., somatosensory targets). The goal of the first experiment was to
investigate if the modality of the cue used to indicate the location of a somatosensory target affects
the body representation used to encode the target’s position during movement planning. The results
showed that auditory cues prompted the use of an exteroceptive body representation for the
encoding of movements to a somatosensory target in visual coordinates. The goal of the second
experiment was to examine the neural processes associated with the visual remapping of an
auditory-cued somatosensory target. It was found that the sensorimotor transformation processes
responsible for the conversion of a somatosensory target position into visual coordinates engages
visuomotor cortical networks to a greater extent than movements to external visual targets. The
goal of the third experiment was to examine the sensorimotor transformation processes employed
for the online control of movements to a somatosensory target. The results of this experiment
revealed that the remapping of a somatosensory target into visual coordinates may not occur prior
to online corrections. Altogether the findings of this thesis reveal that sensory cues can facilitate
the remapping of a somatosensory target prior to goal-directed actions. However, these remapping
processes may be too costly for use during online control when there is no vision of the reaching
limb.
iii
Acknowledgements!Reflecting on the many people who helped me through this time made me feel extremely grateful.
The list of people acknowledged here is not exhaustive but just a few of the individuals who came
to mind. There were many more. Je vais commencer en français.
Merci beaucoup à tous mes collègues et mes amis à Marseille. Particulièrement : Caroline, Julien,
Marie C, Eva (et Téo), Stefania F, Christina, Sarah, Myelene, Joanna, Svetlana, Kyle, Rory, Shu,
Kevin, Raphael, Thibault, Stefania, Sebastian, Luisa, et Ana. Nous avons passé des bons moments
ensemble et vous me manquez.
Merci aussi aux membres de l’équipe : Merci à Didier Louber (l’ami des caraïbes), Laurence
Mouchnino, Marie Fabre, et Alain Guillaume. C’était vraiment un plaisir de travailler avec vous.
Merci à Olivia Lhomond pour toute ton aide et pour « English Day » chaque semaine. Merci à
Anahid Saradjian pour tout ce que tu m’as donné et pour notre rituel de donner un manger aux
chats. Merci infiniment à mes deux frères : Romain Chaumillion (le #1 fan de Raptors) et Nicolas
Lebar (« le mec ») simplement pour tous. Merci énormément à Mme Aurélie Aufray pour tout
votre travail administratif avec la co-tutelle.
Merci aux membres de comité de thèse : Merci à Franck Vidal pour les discussions et l’aide avec
le développement du document. Un grand merci aussi à mon superviseur Jean Blouin pour tout
votre travail avec la thèse et les connaissances scientifique que vous m’avez apporté.
Now to switch back to English. To the external examiners who took time to evaluate my work:
Dr. Denise Henriques and Dr. Jennifer Campos. Thank you for all of your insightful questions,
critiques, and the many discussions we had over the years. I would like to also acknowledge the
mentorship and guidance I received from Dr. Timothy Welsh. Tim, thank you for helping me
develop both as a scientist and as a person. Thank you also for your expertise in statistics.
Thank you also to all of the students and staff members, both past and present, of the Perceptual
Motor Behaviour Lab, the Action and Attention Lab, and the Department of Exercise Sciences
who made this thesis possible. Thank you for creating a supportive environment through both the
fun times and the hard times. In particular, I would like to thank Valentic Crainic, Damian
iv
Manzone, Animesh Kumawat, John DeGrosbois, and Intishar Kazi for their help with numerous
aspects of each project.
To all the present and former members of Team G: Sads, Taff, Lok, and Dov, for taking a big risk
by working with me, and then for becoming some of my best friends. I learned so much from you
all. To my mentors both official and unofficial: Danielle, Tanya, Darian Cheng, Heather, and
Matt Ray. Thank you for the words of advice and guidance. Thank you also to D-Millz, Jonah,
Sam, Nat, Steph, and Mo for friendship, lively discussions, and research support over the years.
Thank you to team bolt: Sharaf, Rach, Sharifa and Jo (also Jess). In addition to being great friends,
you all are my inspirations. A very special thank you to Cindy, Debra, and Rachel “getting
buckets” Goodman for your love, kindness, and support. Thank you to the Generals of the
Heavens: Kwasi, Gabriel, and Danny. You all provided me with the support and motivation to
keep going. Also, very special thank thanks to Kwasi and Steph for being my late-night study
partners in the ring and for your love and support over the years to me and my family during some
of the hardest times.
Thank you to my new lab and family at Houston Methodist. Thank you Dr. Dimitry Sayenko,
Rachel Markley, and Jonathan Calvert for your support during the past few months. Thank you
also to Masha, Sasha, Liza, and Misha for being part of my Texas Family.
To Dr. Luc Tremblay, I cannot express in words the gratitude I have for everything you have done
for me. You have been a colleague, mentor, and friend of the highest caliber. You believed in me
and pushed me to take risks, fail, and comeback better. Thank you also for all the lessons both in
science and in life. I will continue to pay it forward.
To my dear family Dolores (Mom), Richard (Dad) and Niclas (Brother). From pilot testing my
experiments both in Canada and France to planning surprise parties for everyone around, you all
have been with me every step of the way. Thank you all so much, for everything.
Lastly, I would like to dedicate this thesis to my late grandmother (Violet) and my late aunty
(Joanne). Thank you both for all the love and support and for helping me keep things in
perspective. Life is both long and short, I will do my best to make these days count.
v
Table&of&Contents&ACKNOWLEDGEMENTS ................................................................................................................ III
TABLE OF CONTENTS...................................................................................................................... V
LIST OF ABBREVIATIONS ............................................................................................................. IX
LIST OF TABLES ................................................................................................................................ X
LIST OF FIGURES ............................................................................................................................ XI
PREAMBLE .......................................................................................................................................... 1
.......................................................................................................................................... 3
RÉSUMÉ ....................................................................................................................................... 3
GENERAL INTRODUCTION AND LITERATURE REVIEW................................... 4
GENERAL INTRODUCTION ..................................................................................................... 4
PHYSIOLOGICAL SYSTEMS AND MODELS OF GOAL-DIRECTED ACTION .................................... 5
THE VISUAL SYSTEM .............................................................................................................. 5
Visual System Neuroanatomy in Brief ................................................................................. 5
Construction of the Visual World ........................................................................................ 8
Two Visual Streams ............................................................................................................ 8
Vision for Goal-directed Action Planning and Control: Multiple-Processes Model............ 11
The Multiple-Processes Model .......................................................................................... 13
THE SOMATOSENSORY SYSTEM ............................................................................................ 15
Somatosensory Sensory Receptors .................................................................................... 15
Somatosensory Information Processing: From Spinal Cord to the Cortex.......................... 17
Somatosensory Representations During Action – Vector Integration to Endpoint Model.... 20
SENSORIMOTOR TRANSFORMATIONS FOR MOVEMENT PLANNING AND CONTROL.................. 23
Multisensory Combination and Integration for Movement: General Principles.................. 23
Multisensory Combination and Integration ....................................................................... 24
Multisensory Combination and Integration During Movement Planning and Control ........ 25
LITERATURE REVIEW CONCLUSIONS, GAPS, AND FURTHER QUESTIONS ................................ 46
........................................................................................................................................ 49
GENERAL METHODOLOGY .................................................................................................. 49
vi
PARTICIPANTS ....................................................................................................................... 49
GENERAL PROCEDURES ........................................................................................................ 49
BEHAVIOURAL VARIABLES RECORDING AND ANALYSIS ....................................................... 50
Temporal Measures .......................................................................................................... 50
Error Measures ................................................................................................................ 50
Latency of Online Corrections .......................................................................................... 51
ROBOTIC GUIDANCE APPARATUS, DEVELOPMENT, AND TESTING .......................................... 54
ELECTROENCEPHALOGRAPHY APPARATUS, RECORDINGS, AND ANALYSES ........................... 55
Apparatus ......................................................................................................................... 57
Visual Evoked Potentials Features, Calculations, and Analysis ......................................... 57
Source Localization Procedures and Analysis ................................................................... 60
ELECTROOCULOGRAPHY (EOG) ............................................................................................ 61
EOG for EEG ................................................................................................................... 61
EOG for Eye Tracking and Gaze Direction Measurements ................................................ 61
FLEXIBILITY IN THE ENCODING OF REACHING MOVEMENTS TO
SOMATOSENSORY TARGETS: BEHAVIOURAL AND ELECTROPHYSIOLOGICAL
EXPERIMENTS .................................................................................................................................. 62
STUDY A ..................................................................................................................................... 63
ABSTRACT ............................................................................................................................ 63
EXPERIMENT A1 ................................................................................................................... 64
Introduction ...................................................................................................................... 64
Methods ............................................................................................................................ 65
Results .............................................................................................................................. 72
Discussion ........................................................................................................................ 74
EXPERIMENT A2 ................................................................................................................... 76
Introduction ...................................................................................................................... 76
Methods ............................................................................................................................ 77
Results .............................................................................................................................. 86
DISCUSSION .......................................................................................................................... 90
CONCLUSION ........................................................................................................................ 92
RAPID ONLINE CORRECTIONS FOR UPPER-LIMB REACHES TO
PERTURBED SOMATOSENSORY TARGETS: EVIDENCE FOR NON-VISUAL
SENSORIMOTOR TRANSFORMATION PROCESSES ................................................................. 93
vii
STUDY B ..................................................................................................................................... 94
Abstract ............................................................................................................................ 94
Introduction ...................................................................................................................... 95
Methods ............................................................................................................................ 98
Results ............................................................................................................................ 110
Discussion ...................................................................................................................... 120
Conclusions .................................................................................................................... 125
GENERAL DISCUSSION .......................................................................................... 126
THESIS FINDINGS, AND FUTURE DIRECTIONS .............................................................. 126
SUMMARY OF THESIS FINDINGS .......................................................................................... 126
Gaze- (in)dependent Encoding of Somatosensory Targets are Influenced by the Modality of
the Imperative Stimulus ............................................................................................................... 127
Activation of Cortical Networks Associated with Visuomotor Transformations for
Movements to Auditory-Cued Somatosensory Targets .................................................................. 129
Fast Correction Latencies Suggest Control Based on Optimized Feedback use for
Movements to Perturbed Somatosensory Targets. ........................................................................ 131
FUTURE DIRECTIONS AND PERSPECTIVES ............................................................................ 133
CONCLUSIONS ......................................................................................................... 134
REFERENCES .................................................................................................................................. 135
APPENDIX 1: POWER ANALYSES FOR EXPERIMENTAL STUDIES ..................................... 165
POWER ANALYSES FOR EXPERIMENTAL STUDIES ..................................................... 165
STUDY A: ............................................................................................................................ 165
Experiment A1 ................................................................................................................ 165
Experiment A2 Power Analysis ....................................................................................... 167
Study B Power Analysis .................................................................................................. 167
APPENDIX 2: SUPPLEMENTARY ANALYSIS STUDY A ........................................................... 170
ROBOT APPARATUS DEVELOPMENT AND TESTING ................................................... 170
APPENDIX 3: SUPPLEMENTARY ANALYSIS STUDY B ........................................................... 175
MOTION DETECTION TIME VERSUS LATENCY OF ONLINE CORRECTIONS ......... 175
NORMALIZED TRAJECTORY DEVIATIONS............................................................................. 176
viii
APPENDIX 4: LAPLACIAN-TRANSFORMED VEPS FOR STUDY A ........................................ 179
APPENDIX 5: EXPERIMENTAL QUESTIONNAIRES ................................................................ 180
PRIOR TO PARTICIPATION QUESTIONAIRES ................................................................ 180
HAND DOMINANCE TEST .................................................................................................... 180
EYE DOMINANCE TEST ....................................................................................................... 181
BRIEF NEUROLOGICAL QUESTIONNAIRE.............................................................................. 182
ix
List!of!Abbreviations!!
AUD-SOMA Auditory-cued Somatosensory Target AUD-VIS Auditory-cued Visual Target CSD Current Source Density DV Difference Vector DVV Desired Velocity Vector EEG Electroencephalography EOG Electrooculography fMRI Functional Magnetic Resonance Imaging FP Fixation Position IPL Inferior Parietal Lobule IPS Intraparietal Sulcus LED Light Emitting Diode LGN Lateral Geniculate Nucleus LIPv Ventral Lateral Intraparietal Area mIPS Medial Intraparietal Sulcus MT Movement Time OPV Outflow Position Vector PAcc Peak Acceleration PDecc Peak Deceleration PPC Posterior Parietal Cortex PPV Perceived Position Vector PV Peak Velocity RT Reaction Time SC Superior Colliculus SCARA Selectively Compliant Assembly Robot Arm SPL Superior Parietal Lobule TACT-SOMA Tactile-cued Somatosensory Target TMS Transcranial Magnetic Stimulation TPV Target-Position Vector V1 Primary Visual Cortex VEP Visual Evoked Potential VITE Vector-Integration-To-Endpoint
x
List!of!Tables!Table 1.! Descriptions of additional cellular processes contributing to local field potentials .. 56!
Table 2.! Mean (and standard deviation) of latencies (in ms) for the peaks used in the CSD-
VEP calculation. ....................................................................................................................... 85!
Table 3.! Mean and standard deviation for the temporal and kinematic variables of movements
to somatosensory (somato) and visual targets. ......................................................................... 113!
Table 4.! Mean and standard deviation (in cm) for the accuracy variables of movements to
somatosensory (somato) and visual targets. ............................................................................. 114!
Table 5.! Effect sizes of relevant studies ............................................................................. 166!
Table 6.! Number of participants in relevant studies............................................................ 168!
xi
List!of!Figures! A depiction of the distribution of rods and cones along the surface of the retina. The
density of rods, located on the peripheral retinal eccentricities is much greater than the density of
cones, which are concentrated more centrally around the fovea. At approximately 20 degrees of
visual eccentricity is the point at which ganglion cell axons exit the eye (known as the optic disk).
The presence of the optic disk creates a blind spot in the visual field. This figure was adapted from
Osterberg (1935) . ....................................................................................................................... 7!
Visual information enters through the occipital cortex and then is projected either
through the parietal lobe (dorsal stream) or the temporal lobe (ventral stream). This figure was
adapted from Milner and Goodale (1995). ................................................................................. 10!
The multiple-processes model of limb control. The planning stage occurs before
movement start. The initial movement vector is planned based on sensory information and a priori
knowledge from previous trials and the predicted sensory consequences. After peak acceleration
(PAcc) impulse regulation processes begin. Corrections during impulse regulation are based on
the contrast differences between expected and obtained sensory consequences. Impulse regulation
processes continue until after peak deceleration (PDecc) and the end of the primary sub-movement.
Limb-target regulation processes start at PDecc and continue until movement end. Limb-target
correction processes are based on visual and proprioceptive feedback obtained from prior to peak
velocity (PV) to the end of the trajectory. Figure adapted from Elliot et al. (2010)..................... 14!
The dorsal-medial-lemniscal pathway for the transmission of proprioception and touch
information to the somatosensory cortex. Information from muscle spindles and cutaneous
receptors enters through the dorsal roots of the spinal cord then ascends ipsilaterally to the medulla.
These fibers synapse then cross (decussate) at the medulla and ascend toward to the ventral
posterolateral nucleus of the thalamus and then to the cortex (adapted from: Noback, Ruggiero,
Demarest, & Strominger, 2005). ................................................................................................ 19!
The Vector Integration to Endpoint Model (VITE) of proprioceptive online control.
Before movement starts, a difference vector (DV) is calculated from a comparison of the target
position vector (TPV) and the current position of the arm (PPV). Movement commences with a
xii
voluntary and scalable GO signal that scales a velocity-based command (DVV) that activates
dynamic gamma motor neurons (γd). The DVV produces an outflow position vector (OPV) which
activate static gamma (γs) and alpha motor neurons (α). The activity of the OPV, which is driven
by movement, and is altered by sensory information from type Ia and II afferents, changes in the
PPV as arm position changes. This feeds back into the desired velocity vector and alters the
movement of the arm. Movement ends when the PPV matches the DV. This Figure was adapted
from Bullock et al. (1996) ......................................................................................................... 22!
Figure adapted from Pouget et al. (2002). A) The experimental paradigm, participants
fixated on a peripheral fixation point (FP) prior to making a movement to a central target. Targets
were either visual (led), auditory (sound from a speaker) or proprioceptive (projected position of
the participant’s right foot). Note that, in all conditions, participants pointed towards the targets
without actually touching them B) Gaze-dependent reaching errors for each target modality.
Movement endpoints to auditory, visual, and proprioceptive targets were all significantly biased
by gaze direction. ...................................................................................................................... 30!
Experimental setup and results of visual target condition adapted from Reichenbach et
al. (2009). Participants made movements to visual targets that either remained stable or were
shifted early or late in the trajectory. Movements without vision of the limb had larger endpoint
errors and longer correction latencies than movements made without vision of the limb. ........... 33!
Differences between hand-centred and gaze-centred neural receptive fields (RF). Hand-
centred neural activations shift with the position of the hand in space, and gaze centred activations
shift with the position of the eyes. Both types of representations have been noted in the PPC.
Adapted from Batista and Newsome (2000). ............................................................................. 37!
The common eye-centred reference frame gain field hypothesis. A) Transformations in
the PPC. The distribution of reach related activities in the PPC is arranged from primarily eye-
centred in the inferior parietal lobule (IPL) to heterogeneously limb-centred and eye-centred in the
superior parietal lobule (SPL). For this reason, the PPC is viewed as the sensorimotor interface for
transformations related to the planning and control of goal-directed actions. Sensory information
about target and effector positions from pertinent sources modify the reach related activities of
eye-centred neurons. Response patterns of neurons in the SPL suggest that information is then
xiii
transformed directly into limb-centred coordinates. B) Modulations of reach related neurons in
the PPC. Neurons in the PPC fire with respect to changes in eye position. The gain of these
responses can be modulated by visual information. Figure representations adapted from data
described in Buneo & Andersen 2006, and Cohen & Andersen 2002. ........................................ 38!
A) Adapted from Bernier et al. (2007). Participants exhibited lower errors due to
prismatic adaptation when reaching to a proprioceptively defined target than a visual target. B)
Adapted from Sarlegna et al. (2007), movements to somatosensory targets were less curved than
movements to visual targets in the same position. Both of these results argue that different
sensorimotor transformation mechanisms underlie movements to visual and somatosensory
targets. ...................................................................................................................................... 43!
Outline of the experiments conducted in this thesis, Experiment A1 and A2 investigated
the sensorimotor transformations used for planning movements to somatosensory targets, whereas
Experiment B investigated the role of these transformations in the online control of goal-directed
actions....................................................................................................................................... 48!
Figure outlining the extrapolation method for determining correction latency as
described by Oostwoud Wijdenes et al. (2014). The authors tested the predictive ability of different
computations of correction latency to estimate a 100 ms latency to target perturbations of 1 to 4
cm in simulated movements of 300 – 500 ms. The methods included threshold, confidence interval
and the extrapolation method. Panel A) shows the main findings of Oostwoud Wijdenes et al.
(2014) for the 3 cm perturbation stimulations, the extrapolation method applied to averaged
acceleration data yielded a very accurate prediction of correction latency across movement times.
Panel B) shows the method applied to data in the Study B of this thesis. Averaged acceleration
profiles for the perturbation curve showed in panel B were derived from movements to
somatosensory targets perturbed away from the body. ............................................................... 53!
To calculate an ERP, continuous EEG data were preprocessed and artifacts due to
external noise and interference were removed. After, the data were segmented into defined time
windows centered around the event of interest. Finally, by averaging all trials for certain
conditions, an ERP waveform could be observed for each electrode of interest (see below). ...... 58!
xiv
Characteristic peaks of a visual evoked potential. A negative deflection N1 occurs at
around 80 ms after the visual stimulation. The P1 occurs at about 100 ms after that, followed by
the N2 at around 140 ms. Lastly there is a P2 that occurs about 200 ms after the initial stimulation.
For the EEG analyses in the present thesis, the peak to peak amplitude between the P100 and N150
was used for analysis (shown in red), as this was the most recognizable peaks across participants.
................................................................................................................................................. 59!
Activities of associated with visual –sensorimotor processing were analyzed using the
electrodes overlapping the occipital cortex and the occipito-parietal junction of the left hemisphere
(electrodes PO7, PO3, O1). These electrodes were selected based on findings in previous studies
(Lebar et al., 2015). ................................................................................................................... 59!
Panel A) A drawn representation of the apparatus used in Experiment 1. Participants
sat facing an immersive display comprised of a computer monitor, a semi-reflective glass surface,
and a custom aiming apparatus (not to scale). Participants made movements from an unseen home
position (microswitch) to either visual targets projected onto the surface of the aiming console or
the perceived position of their fingers. In the somatosensory-target conditions, participants
positioned their target fingers beneath a plastic case and performed movements to the perceived
position of their fingers as if projected onto the case’s surface. Note that the visual items projected
on the semi-reflective glass were perceived by the participants as appearing on the table surface.
Panel B) A representation (not to scale) of the aiming surface in the visual target condition.
Piezoelectric buzzers were positioned to the left of the aiming surface, and provided the imperative
signals in the auditory-cued aiming conditions .......................................................................... 67!
Panel A) Averaged normalized reach trajectories for all participants in each cue-target
condition (error patches represent the between-participant standard error of the mean). Panel B)
Average normalized directional error for each cue-target condition (error bars represent the
between participant standard deviation). Participant’s reaching errors were significantly more
influenced by gaze fixation position (i.e., Right Fix vs. Left Fix) in the auditory-cued target
conditions (-VIS and –SOMA) than in the vibrotactile-cued target conditions. Panel C) Mean
normalized reaction times for participants in each experimental condition (error bars represent the
between-subjects standard deviation). Reaction times of gaze-shifted movements were
xv
significantly longer when participants aimed to auditory-cued somatosensory targets compared to
both auditory-cued visual targets and vibrotactile-cued somatosensory targets. ......................... 73!
Panel A) A drawn representation of the experimental apparatus for Experiment 2 (not
to scale). Participants fixated on one of three fixation locations and began their movements from
the associated home position. In the auditory-cued somatosensory-target condition (represented
above), participants performed movements to one of the three middle fingers of their non-reaching
limb. Panel B) A representation of the aiming surface in the visual target condition. Participants
fixated on of three target positions and placed their fingers on the corresponding microswitch to
begin each trials. Piezoelectric buzzers to left again provided the imperative signals to initiate
movement in both the AUD-VIS and AUD-SOMA conditions. ................................................. 79!
Grand average VEPs for each electrode in the AUD-VIS and AUD-SOMA conditions
(error patches represent the between-participant standard deviation). The peak-to-peak amplitudes
between P100 and N150 as determined by a current source density analysis were used for statistical
contrasts. ................................................................................................................................... 84!
CSD normalized VEPs for the occipital (O1) and occipito-parietal (PO7) electrode
(error bars represent the standard error of the mean). For both electrodes, CSD- VEPs were
significantly larger in the AUD-SOMA compared to the AUD-VIS condition. .......................... 88!
Grand average source activity for each condition between the P100 and N150 latency
time. Source activity (color maps reveal activation levels) was localized in parietal and occipital
areas in both experimental and control conditions. Statistical contrasts (paired-samples t-tests,
alpha set to 0.05, t-value map indicates direction of effects) revealed significantly more activity in
the left parietal and parieto-occipital regions (as indicated by the shade of red) for the AUD-SOMA
experimental condition compared to the AUD-SOMA control condition. .................................. 89!
A drawn representation of the experimental setup (not to scale) is shown in panels (a)
and (b). Panel (c) is a representation of the aiming console and stimuli for each target modality.
Participants sat comfortably facing the aiming apparatus in a completely dark room. A robotic
device was used to deliver target perturbations in both somatosensory (left finger) and visual
xvi
(LED) targets. Participants performed leftward arm movements to reach the position of the target
as if it was projected onto the aiming surface. ......................................................................... 100!
Examples of velocity profiles of the reaching hand and the robotic effector (Target) for
each perturbation time. Panel (a) shows a perturbation occurring before movement onset; panel (b)
shows a perturbation occurring ~100 ms after movement onset, and panel (c) shows a perturbation
occurring ~200 ms after movement onset. ............................................................................... 104!
The extrapolation method for determining the latency of online corrections. For each
participant average acceleration profiles in the direction axis were computed for both perturbation
and no-perturbation trials. The acceleration difference between these profiles (Accel Difference)
was then plotted to calculate correction latency. Correction latencies were computed by drawing a
line (Extrapolation Line) between 75% and 25% of the maximum difference in the Accel
Difference profile (Extrapolation points) and extrapolating the line to the first zero crossing. The
time between the perturbation and the zero crossing was defined as the correction latency. Panel
(a) shows this method applied to averaged data for somatosensory target perturbations and panel
(b) shows the method applied to averaged data for visual target perturbations. ........................ 109!
Average reaching trajectories for each condition in the reaching protocol. Panels (a),
(b), depict perturbations occurring before movement onset. Panels (c) and (d) depict perturbations
100 ms after movement onset. Panels (e) and (f) depict perturbations occurring 200 ms after
movement onset. Trajectories were normalized with each point representing 2% of movement
duration. Error bars indicated the between-subject standard deviation of spatial position ......... 112!
(a) Direction constant error. Participants were more accurate when performing reaches
to somatosensory targets perturbed after movement onset compared to when performing reaches
to visual targets. When aiming to visual targets participants exhibited a larger under-correction
relative to when making movements to somatosensory targets. (b) Correction magnitude.
Participants exhibited larger corrections in response to somatosensory target perturbations than in
response to visual target perturbations at all perturbation times. Furthermore, for both modalities,
participants exhibited smaller corrections in response to perturbations at 200 ms than in response
to the before and 100 ms perturbation times. ........................................................................... 117!
xvii
Correction latencies in response to target perturbations. Overall, participants corrected
faster for somatosensory target perturbations than for visual target perturbations. Furthermore, for
visual targets, correction latencies were longer in response to targets perturbed toward the body.
............................................................................................................................................... 119!
Reaction of the robot perturbation movement when signaled by motion tracking
processed on the same computer. ............................................................................................ 171!
Number of lost samples in 500 1-second recordings. Sixty (60) percent of data were
lost samples, this severely decreased the amount and resolution of the useable recordings. Also
55% during the robot’s trajectory. ........................................................................................... 171!
Differences between single and dual setups for the recording of perturbation data. The
dual station setup was found to be the best solution for both maximizing resolution of trajectory
data and reducing variability in the speed of perturbations....................................................... 173!
Most reaction times for the dual station setup fell within 60- 75 milliseconds after the
go signal. This reduction perturbation variability provided a basis for the protocol. ................. 173!
Overall the number of lost samples was reduced from 60% to 10% with only 4% of
lost samples occurring during the robot’s movements. ............................................................. 174!
Contrasts between the target modality differences in correction time and detection time.
The difference between the two target modality conditions is significantly larger incorrection time
compared to detection time. .................................................................................................... 176!
Analysis of trajectory deviations. Each curve represents a temporally normalized
averaged trajectory of each condition used in the experiment. Error patches represent the between
subject standard deviation. Targets were either perturbed Before, 100 ms or 200 ms after
movement onset or were unperturbed (control). Participants adjusted their trajectory to
somatosensory targets perturbed 100 ms after movement onset but no differences between spatial
trajectories were found for visual targets. Participants were capable to fully adjusting their
movements to targets perturbed before movement start. .......................................................... 178!
xviii
Grand average of the Laplacian transformed VEPs for each electrode of interest in the
AUD-VIS and AUD-SOMA conditions. The peak-to-peak amplitudes between P1 and N1 were
used for the normalization and comparisons made in Experiment A1. ..................................... 179!
Adapted from Miles, W.R. (1930). Ocular dominance in human adults. The Journal of
General Psychology, 3, 412-430.............................................................................................. 181!
1
Preamble!The three experiments in this dissertation address critical gaps in the literature with regard to
movements to somatosensory targets. The overall goal of the thesis was to determine if the
transformation processes used for movements to somatosensory targets are similar or distinct from
those underlying movements to visual targets. The goal of the first study (Study A) was to
determine if the manner in which a somatosensory target is encoded affects sensorimotor
transformation processes prior to upper-limb reaches. The theoretical basis for this study was
derived from numerous sources, including: previous work on body representations; clinical work
on patients with autotopagnosia; and psychophysical experiments examining movements to body
parts. A review of the extant literature revealed that somatosensory targets could be encoded in a
visual, gaze-dependent, reference frame, or a non-visual, gaze-independent, reference frame.
Experiment A1 examined if the sensory signals used to identify the somatosensory target
location impacted the reference frame employed for movement planning. Participants made
reaching movements to somatosensory targets, which were either cued by an auditory or a
vibrotactile stimulus. Gaze positions were also systematically altered for each cue-target condition
to measure the degree of gaze-dependent encoding. Larger gaze-dependent biases were found for
movements to auditory-cued somatosensory targets compared to vibrotactile-cued somatosensory
targets. These results indicated that participants used an exteroceptive, visual body representation
to encode auditory-cued somatosensory targets, and an interoceptive, somatosensory body
representation to encode vibrotactile-cued somatosensory targets. Importantly, the results of
Experiment A1 also revealed that shifts in gaze positions had a larger effect on reaction times for
auditory-cued somatosensory targets than both auditory-cued visual and vibrotactile-cued
somatosensory targets. The larger gaze-dependent errors and increases in reaction times could
indicate that additional transformation processes were required for the visual remapping of
somatosensory targets prior to goal-directed actions.
Experiment A2 expanded on the findings of Experiment A1 by examining the neural
correlates of the sensorimotor transformations underlying movements to somatosensory targets
planned in a visual reference frame. It was found that there was an increase in cortical visual
information processing prior to reaches to auditory-cued somatosensory targets, compared to
2
auditory-cued visual targets. The sources of this facilitation were localized to occipital and
occipital-parietal areas. These findings suggest that remapping of somatosensory target positions
prior to reaching movements involves cortical networks associated with visual information
processing.
The goal of Study B was to examine the sensorimotor transformation processes during the
online control of movements to somatosensory targets. Previous studies examining online
adjustments have suggested that somatosensory information about the limb position must be
transformed into visual coordinates prior to engaging in online corrections. However, as Study A
revealed, these processes require additional time and may result in an increase in error. Thus, it
remains unclear if this visual transformation strategy is also employed for online control.
In Study B, participants made aiming movements, with no vision of their reaching limb, to
either an external visual target or a somatosensory target (the left index finger of their non-pointing
hand). On a proportion of trials, target positions were perturbed either before, or 100 ms or 200 ms
after movement start. It was found that participants made faster, and more accurate, corrections to
perturbed somatosensory targets than to perturbed visual targets. These findings suggest that non-
visual sensorimotor transformations were performed when making online corrections to perturbed
somatosensory target
3
!!
! Résumé!La littérature sur les processus d’intégration multisensoriels (i.e., transformations sensorimotrice)
avant et pendant des mouvements volontaires ont principalement utilisé des cibles visuelles.
Considérant que la modalité d’une cible influence le cadre de référence utilisé pour contrôler un
mouvement volontaire, l’objectif principal des trois expériences de la présente thèse était
d’examiner ces processus de transformations sensorimotrices, et ce, spécifiquement pour des
mouvements vers un cible somatosensorielle. Les deux premières expériences ont examiné si les
transformations sensorimotrices pendant la planification du mouvement sont influencées par le
cadre de référence utilisée pour coder les cibles somatosensorielles. L’objectif de la troisième
expérience était d’examiner les transformations sensorimotrices utilisée pour le contrôle du
mouvement vers une cible somatosensorielle. Les résultats des deux premières expériences
indiquent que même des signaux auditifs peuvent faciliter l’utilisation d’un cadre de référence
visuel du corps pour transformer la position de cibles somatosensorielles. Ces processus de
transformation pourraient nécessiter des processus supplémentaires et engagent clairement des
réseaux corticaux visuels et visuomoteurs. De plus, les résultats de la troisième expérience indique
que des corrections apportées aux mouvements vers des cibles somatosensorielles perturbées
prennent moins de temps pour être initiées en plus d’être plus efficaces et précises que des
corrections apportées aux mouvement vers des cibles visuelles. De plus, cette troisième expérience
indique que des cibles somatosensorielles ne sont pas nécessairement transformées dans un
référentiel visuel avant de corriger un mouvement volontaire en cours d’exécution. Globalement,
les études dans cette thèse ont montré que des signaux auditifs peuvent faciliter l’encodage de
cibles somatosensorielles dans un cadre de référence visuel pendant la planification du mouvement
volontaire mais que le temps et les ressources nécessaire pour encoder des cibles
somatosensorielles pourraient être trop importantes pour nécessairement continuer d’utiliser un
cadre de référence visuel pour corriger la trajectoire d’un mouvement volontaire.
4
!!General!Introduction!and!Literature!Review!
! General!Introduction!Philosopher George Berkeley’s new theory of vision, originally published in 1709, was one of the
first cohesive accounts of why and how we develop the ability to see and interact with the world.
In his New Theory of Vision, Berkeley argued that an intelligent being who possesses a perfect
sense of sight but who is devoid of the sense of touch “…could have no idea of a solid or quantity
in three dimensions which followeth from its [the intelligent being] not having any idea of
distance” (Berkeley, 1709, p.154). Berkeley made the assertion that humans develop the ability to
perceive 3-dimensional objects only when they gain the capability to interact with these objects in
meaningful ways through touch (see also Held & Hein, 1963). This assertion was later paralleled
by Ian Hacking (1983) in his work Representing and Intervening to resolve the ongoing debate
between realists and anti-realists with regard to scientific entities. To Hacking, a scientist learns
“…to see through a microscope by doing, not just by looking ” and just as a “…scuba diver learns
to see in the new medium of oceans by swimming around…” any “… new ways of seeing acquired
after infancy, involve learning by doing ” (Hacking, 1983, p. 136). This connection between action
and perception also appealed to contextualists, sought to disentangle notions of separation between
thought and language. To contextualists, action, language, thought, and percepts could not be fully
understood without consideration of the context wherein they occurred. One of the strongest
champions of the contextualism was John Dewey, who, in his seminal work published in 1925,
challenged the spectator theories of knowledge by arguing that we must not only understand
context to understand the mind but that the “mind is constantly adapting to the environment
through action” (Nerlich & Clarke, 1996, p. 125). Thus, to truly understand the depths of any
philosophy or object of human nature, we must understand the nuances of the time, and
importantly, the context wherein they came to fruition.
Overall the purpose of this thesis was to examine the neuroscientific homologue of the
“action in context argument” as outlined by the contextualists. The series of experiments detailed
in this thesis employed behavioural and neurophysiological paradigms with the goal of examining
5
the sensorimotor transformations during the preparation and control movements to somatosensory
targets.
! Physiological!Systems!and!Models!of!GoalGdirected!Action!!Many different physiological systems are involved in the production of goal-directed movements,
the following section briefly surveys two of the main sensory systems that are considered in this
thesis: the visual system and the somatosensory/proprioceptive system. Both systems are discussed
with regard to their functional relevance for the planning and control of actions.
! The!Visual!System!
! Visual!System!Neuroanatomy!in!Brief!
Visual sensation begins at the retina, a thin sheet of densely packed neurons located in the posterior
part of the eye. The outermost layer of the retina consists of light-sensitive cells known as
photoreceptors, whose main function is to transform the refracted patterns of light into electrical
signals. There are two main classes of photoreceptors: cones and rods. Cones are short tapered
neurons that fire in response to high levels of illumination. The greatest concentration of cones on
the retina is found at 0 degrees eccentricity with respect to the fovea (i.e., foveola, see Figure 1).
As a result of their location and their frequency dependent firing pattern, cones are functionally
implicated in both visual acuity and colour vision (Green, 1970; Meister & Tessier, 2013; Shlaer,
1937). In contrast, rods are distributed along the retinal periphery (see Figure 1), and peak rod
density occurs at 18 degrees with respect to the fovea (Purves et al., 2001). Rods are long
cylindrical shaped neurons that respond to low levels of illumination, and their firing frequencies
become saturated in bright-light conditions. Functionally, rods have been implicated in dim light
perception and in the detection of peripheral motion (Gilbert, 2013a; Meister & Tessier, 2013).
Signals from groups of rods and cones from different regions of the retina are transmitted
to the brain via two main types of ganglion cells: ON cells and OFF cells. These cells alter their
firing patterns depending on their exposure to light intensity: ON cells fire more rapidly when
illumination increases, whereas OFF cells fire more rapidly when illumination decreases. The
anatomical arrangement of these ganglia gives rise to retinal receptive fields. The retinal receptive
field of a ganglion cell is best defined as the map between the spatiotemporal patterns of light and
6
the pattern of ganglion cell activations (Wienbar & Schwartz, 2018). In simpler terms, the
receptive field represents a region in visual space wherein the presence or change in a stimulus or
stimulus features can alter the response of the ganglion cell. Receptive field activation patterns
are linked to visual processes such as contrast perception, motion detection, and the recognition of
temporal characteristics of visual stimuli (Gilbert, 2013b).
Information gathered from these ganglion cells travels to the brain via multiple subcortical
pathways including: the superior colliculus (SC), a region in the midbrain associated with eye
movements (saccade generation and control), orienting behaviours, and attention (Becker, 1991;
Georgopoulos, 1990); the pretectum, another midbrain region associated with eye movements and
the pupillary light reflex (Gilbert, 2013a); and the lateral geniculate nucleus (LGN), a region of
the thalamus that projects to the visual cortex (Gilbert, 2013b; Meister & Tessier, 2013). The
information going through the LGN pathway is most associated with conscious perception of the
visual world and is the dominant pathway for the information used in the planning and control of
action.
7
&!A!depiction!of!the!distribution!of!rods!and!cones!along!the!surface!of!the!retina.!The!density!of!rods,!located!on!the!peripheral!retinal!eccentricities!is!much!greater!than!the!density!of!cones,!which!are!concentrated!more!centrally!around!the!fovea.!At!approximately!20!degrees!of!visual!eccentricity!is!the!point!at!which!ganglion!cell!axons!exit!the!eye!(known!as!the!optic!disk).!The!presence!of!the!optic!disk!creates!a!blind!spot!in!the!visual!field.!This!figure!was!adapted!from!Osterberg!(1935)!.!!
!
8
! Construction!of!the!Visual!World!!
The visual system also relies on cortical visual areas to construct a representation of the visual
scene. Such processes are supported by the neural pathways that extend from the LGN to the visual
cortex.
Neurons in the LGN primarily project to the primary visual cortex (V1). Responses of
neurons in V1 have been associated with: object orientation (Hubel & Wiesel, 1968; Meister &
Tessier, 2013); depth perception (Barlow, Blakemore, & Pettigrew, 1967); the direction of
stimulus motion (Treue, 2001); and stimulus speed (Maunsell & Van Essen, 1983). Furthermore,
activation patterns in the receptive fields of V1 neurons, as a consequence of the alignment of their
projecting cells, reveals the existence of selective firing patterns for specific subtypes of the
aforesaid visual features (Gilbert, 2013b). For example, a population of neurons in the visual cortex
possessing an alignment of inhibitory and excitatory projections in a certain orientation will only
respond to stimuli that match their orientation (Hubel & Wiesel, 1962). This response selectivity
of visual receptive fields has also been noted for stimulus direction, and speed (Maunsell & Van
Essen, 1983). By analyzing these selective firing patterns, it is possible to determine activity
correlated to intermediate visual perceptual features such as object contours, object motion, and
colour contrast (Gilbert, 2013b; Green, 1970; Maunsell & Van Essen, 1983).
! Two!Visual!Streams!
From V1, visual information travels through two separate but interconnected pathways (see
Budisavljevic, Dell’Acqua, & Castiello, 2018): the ventral visual pathway or ventral stream, and
the dorsal visual pathway or dorsal stream (Mishkin, Ungerleider, & Macko, 1983). Each stream
discussed below receives the same intermediate level (orientation, speed, direction response tuned
firing) information from V1 neurons (see Figure 2). Both visual streams receive information about
object structure, object orientation, and object spatial location. What most differentiates the two
streams is how visual information processing within each stream relates to the goal of the observer
(Goodale & Milner, 1992; Mishkin et al., 1983).
The ventral stream is associated with object recognition and representation. Anatomically,
the ventral stream is composed of projections from V1 to the temporal lobe (see Figure 2). Overall,
the functional role of the ventral stream is to discern the characteristics of objects and spatial
9
relationships by building perceptual representations. The representations formed in the ventral
stream allow us to identify and semantically describe relationships between objects observed in
visual scene.
In contrast to the ventral stream, the dorsal stream uses visual information from V1 to
construct an action-centered representation of the world. To enter the dorsal stream, visual
information travels from V1 to the parietal cortex and then to the frontal lobe. The dorsal stream
is thus associated with the transformation of visual information that is relevant to movement goals
and the moving effector.
To construct a representation of the environment for use in goal-directed actions, both
visual streams work together. Simply put, the ventral stream is engaged to enable the identification
and recognition of possible goal objects in the visual environment and the dorsal stream uses visual
information about the size and shape of objects to guide skilled movements.
10
&!Visual!information!enters!through!the!occipital!cortex!and!then!is!projected!either!through!the!parietal!lobe!(dorsal!stream)!or!the!temporal!lobe!(ventral!stream).!This!figure!was!adapted!from!Milner!and!Goodale!(1995).!!
11
! Vision!for!GoalGdirected!Action!Planning!and!Control:!MultipleGProcesses!Model!
The discovery of the two visual streams was paramount to the understanding of the role of vision
in the planning and control of action. However, models describing visuomotor control have been
in development for well over a century. In this section, the development of the multiple processes
model of online control is briefly described with relevance to visual information use. In this thesis,
when movements were made toward visual targets as a baseline measure, only initial visual
information about the hand was provided (Study A). When movements to visual targets were used
to examine online sensorimotor transformations, no initial visual information about limb position
was provided (Study B). In all cases, no visual feedback about limb position was provided during
the trajectory. Thus, the discussion of the utility of visual information for each phase in the multiple
processes model focuses on visual target information.
Woodworth, in his seminal monograph (see Woodworth, 1899), forwarded a model which
proposed that goal-directed movements consisted of two phases: an initial impulse phase wherein
the limb is propelled toward the target; and a current control phase wherein visual feedback is used
to “home in” on the target. Woodworth also noted that fast movements (i.e., less than 400 ms in
duration), showed similar trajectory characteristics independent of visual feedback availability
(vision of the limb and target). This led to the idea that the central nervous system takes time to
perform visual-based trajectory corrections based on target information, although recent estimates
are much shorter (i.e., less than 100 ms depending on when vision is presented in the trajectory)
than the originally proposed 400 ms (Blouin, Teasdale, Bard, & Fleury, 1993; Howarth, Beggs, &
Bowden, 1971; Kennedy, Bhattacharjee, Hansen, Reid, & Tremblay, 2015; Tremblay, Hansen,
Kennedy, & Cheng, 2013; Zelaznik, Hawkins, & Kisselburgh, 1983).
Since the proposal of Woodworth’s (1899) model, experimental evidence has revealed
insights into the use of visual target information in each component. For example, with the
emergence of motor programming theory, came the idea that the initial impulse phase was based
on a motor program specified before movement onset based on initial hand position and visual
target location (Schmidt, Sherwood, Zelaznik, & Leikind, 1985). Online trajectory amendments
were thought to be the result of separate motor programs initiated during the movement. Thus, the
12
efficacy of visual-based corrections was hypothesized to be related to the time required to organize
online motor programs (Howarth et al., 1971; Keele & Posner, 1968).
Perhaps the most crucial update to the Woodworth’s original model was proposed by
Meyer and colleagues in 1988. Meyer’s optimized control model was based on the impulse
variability theory of motor control (Schmidt et al., 1985). The authors surmised that the trajectory
specified in the initial impulse phase is programmed optimally such that no correction is required
in order to successfully land on the target. However, because of neuromotor noise, and the
variability associated with the impulse, endpoints would not always land directly on the target, but
have a normal distribution around the target’s centre. Similar to the previous models of motor
control, the optimized control model purported that any necessary corrective movements would
occur at the end of the initial impulse to compensate for endpoints that may land outside of the
target boundary.
While the above-mentioned models increased the explanatory power and mechanistic
understanding of the two-component model, these models fell short of accounting for a variety of
experimental findings with regard to visually-guided action. One of the main findings that this
model failed to address was corrections to early perturbations prior to the end of the initial impulse
(Goodale, Pélisson, & Prablanc, 1986; Hansen, Tremblay, & Elliott, 2008; Saunders & Knill,
2003). Goodale et al. (1986) performed a seminal study wherein participants reached without
online vision of their reaching limb to targets that changed location during the orienting saccade,
and prior to the onset of the hand movement (i.e., during saccadic suppression, see Bridgeman,
Hendry, & Stark, 1975). The authors found that participants corrected for shifts in target position
even though they could not consciously detect that the target position had changed. Moreover, the
kinematic features of the corrected trajectories could not be distinguished from those of
uncorrected trajectories (see also Prablanc & Martin, 1992). Together, these results showed that
changes in target position could be corrected for early in the movement even without vision of the
limb. The authors concluded that early rapid online corrections were likely based on updated visual
target location and limb position information derived from somatosensory and efferent information
(see also Prablanc, Desmurget, & Gréa, 2003).
13
! The!MultipleGProcesses!Model!To account for early online corrections, the multiple-processes model was proposed by Elliott et
al. (2010). Similar to Woodworth’s original model, the multiple-processes model outlines two
types of control. The first type of online control is impulse-regulation and the second is limb-
target regulation. Impulse-regulation occurs early in the movement and is based on egocentric
representations including comparisons of the visual target location and the efference copy. Limb-
target regulation occurs later in the movement, and is based allocentric information such as the
comparison between visual information about the target position and proprioceptive information
about the hand position (see also Elliott et al., 2017).
Overall, the multiple-processes model is based on the premise that visually-guided
movement unfolds in the following manner: First, prior to movement initiation, the initial
movement vector is programmed based on expected sensory consequences and prior knowledge.
Second, when the signal to initiate the movement is received, the muscle commands and an efferent
copy are produced. The efference copy is used to predict the sensory consequences of the
movement based on the motor command. When the limb begins to move, impulse regulation
begins, and corrections occur by evaluating the difference between the observed and expected
(predicted) sensory consequences. Early in the trajectory (prior to peak velocity), the visual system
gathers sensory information to be used in limb-target regulation. When the primary sub-movement
is completed (i.e., approximately at peak deceleration) limb-target regulation begins based on
sensory information obtained throughout the trajectory (see Figure 3).
Overall the information obtained and constructed by the visual system is used in two ways
for goal-directed actions. The first is to define and plan a trajectory, and the second is to control
an ongoing movement.
14
&The!multipleGprocesses!model!of!limb!control.!The!planning!stage!occurs!before!movement!start.!The!initial!movement!vector!is!planned!based!on!sensory!information!and!a!priori!knowledge!from!previous!trials!and!the!predicted!sensory!consequences.!After!peak!acceleration!(PAcc)!impulse!regulation!processes!begin.!Corrections!during!impulse!regulation!are!based!on!the!contrast!differences!between!expected!and!obtained!sensory!consequences.!! Impulse!regulation!processes!continue!until!after!peak!deceleration!(PDecc)!and!the!end!of!the!primary!subGmovement.!LimbGtarget! regulation!processes!start!at!PDecc!and!continue!until!movement!end.!LimbGtarget!correction!processes!are!based!on!visual!and!proprioceptive!feedback!obtained!from!prior!to!peak!velocity!(PV)!to!the!end!of!the! trajectory.!Figure!adapted!from!Elliot!et!al.!(2010).!
15
! The!Somatosensory!System!As mentioned above, the dorsal visual stream constructs a representation of the external
environment based on one’s intended actions. These visual representations are then used to amend
the limb’s trajectory at both the early and later phases of a visually-guided movement. Without
vision of the limb, other sensory systems capable monitoring limb position are necessary in order
to engage in corrective mechanisms. One sensory system that could be effective for the formation
of action-centered representations is the somatosensory system. For experiments in this thesis and
other work (see: Section 2.4.3), somatosensory information also served to define target positions.
The following discussions of somatosensory information will be focused on two modalities of
sensory information pertaining to somatosensation: tactile information emerging from cutaneous
receptors in the skin and proprioception emerging from receptors in the muscles and joints.
! Somatosensory!Sensory!Receptors!
Somatosensation, similar to kinesthesia (Sherrington, 1916) and often referred to as
proprioception, is the sense of one’s own body position and body motion in space. There has been
some debate as to the functional necessity of certain somatosensory receptors for the monitoring
of kinematic and kinetic parameters associated with changes in body position (for review see
Proske & Gandevia, 2012). Nonetheless, the prominent discourse in the present literature is that
somatosensory information emerges from a combination of inputs from numerous sensory organs
including: skin receptors, joint receptors, muscle spindles, and Golgi tendon organs.
It has been demonstrated that skin afferent receptors play a substantial role in one’s sense
of body position (Edin & Abbs, 1991; Edin, 2004). For example, both the slow-acting and fast-
acting mechanoreceptors in non-glabrous skin have been shown to respond to voluntary finger and
hand movements. Fast-acting receptors produce rapid graded responses depending on the extent
of the movement-induced skin stretch at proximal joints. Slow-acting receptors respond
directionally by increasing discharge during proximal-joint flexion and decreasing discharge
during proximal-joint extension. Furthermore, it has been demonstrated that the activity of neurons
responsive to skin stretch in the primary somatosensory cortex is correlated with their activity
16
levels when encoding kinematic information about joint positions (Cohen, Prud’homme, &
Kalaska, 1994). These results suggest that skin receptors could contribute to limb localization
during both static postures and limb movements.
Joint receptors are located within joint capsules and provide information about the
articulation of a joint and its associated ligaments. The range of activation of joint receptors is
limited as most receptors respond to end ranges of flexion and extension in a non-direction specific
manner (Burgess & Clark, 1969). However, it has been shown that, in the absence of other
proprioceptive receptors (muscle spindles, cutaneous receptors, and Golgi tendon organs), joint
receptors respond in a position dependent manner to low velocity finger movements (Gandevia &
McCloskey, 1976). Further research in humans is required to truly determine the extent to which
joint receptors contribute to proprioception; however, as it stands, the contribution of joint
receptors appears to be minor.
Muscle spindles are receptors that are located within muscle tissue. Muscle spindles
respond to changes in muscle length, and thus provide information about the position, velocity,
and acceleration of the limb during both self-generated and externally-triggered movements
(Proske & Gandevia, 2012). Muscle spindles’ signals project to the somatosensory cortex and the
activity of these spindles can help distinguish various kinematic features across a wide range of
joint angles. The role of muscle spindles as the primary proprioceptive sensory organ has been
supported by work examining joint replacements (Goodwin, McCloskey, & Matthews, 1972),
dorsal column lesions (Lee & Tatton, 1975), and tendon vibration (Redon, Hay, & Velay, 1991).
Golgi tendon organs are proprioceptive receptors located in the muscle-tendon junction.
These organs were originally thought to respond to changes in muscle force (Proske & Gandevia,
2012). However, mounting experimental evidence shows that Golgi tendon organs may not sense
tensile forces but rather respond to changes in tendon length (Kistemaker, Van Soest, Wong,
Kurtzer, & Gribble, 2012). Despite these new findings, it is agreed that Golgi tendon organs, along
with muscle spindles are the main proprioceptive receptors responsible for the accurate
performance of goal-directed actions in the absence of vision.
17
! Somatosensory!Information!Processing:!From!Spinal!Cord!to!the!Cortex!
Proprioceptive and tactile information emerging from mechanoreceptors the in innervated regions
of the skin (dermatomes) and muscles (myotomes) are relayed to the central nervous system via
peripheral nerve fibres that vary in diameter, myelination, and conduction velocity. Information
from most tactile and proprioceptive afferents are relayed via large diameter (12 – 20 µm) or
medium diameter (6 – 12 µm) myelinated dorsal root ganglion cells (Type I and II afferents). It is
hypothesized that the fast conduction velocities of the larger diameter cells (70- 120 m/s) make it
possible for this information to be obtained during the course of slow and rapid actions. Nerve
fibers carrying proprioceptive and tactile information ascend to the cerebellum via the
spinocerebellar tract or to the brainstem through the dorsal column of the spinal cord (the dorsal–
medial – lemniscal Pathway; see Figure 4).
Fibres from the lower body regions ascend in the medial portion of the dorsal column,
while fibres from the upper regions ascend in the more lateral portions. The rostral third of the
dorsal column contains neurons responsible for the relay of proprioceptive information whereas
and the middle relays primarily tactile and cutaneous information. These tracts either ascend
ipsilaterally and terminate in the cerebellum (spinocerebellar tract) or terminate in the lower
medulla of the brainstem where, after synapsing, the fibers cross and ascend through the medial
lemniscus. Due to the the crossing of fibers at the sensory decussation (see Figure 4), afferent
information from proprioceptive and tactile receptors in the right side of the body is processed in
the left cerebral hemisphere.
Information from the dorsal– medial – lemniscal pathway enters the thalamus before being
relayed to cortical areas. It has been argued that there are two main regions of somatosensory
processing in the thalamus (Kaas, Merzenich, & Killackey, 1983). The ventral posterior nucleus
that receives information from cutaneous inputs, and the ventral posterior superior nucleus that
receives inputs from fibers carrying proprioceptive information. The ventral posterior area sends
projections to primary somatosensory area 3b (i.e., Broadmann’s areas) and the ventral posterior
superior area sends proprioceptive information primarily to area 3a. The degree of specialization
of these two somatosensory processing areas remains an active area of investigation (Disbrow,
Roberts, & Krubitzer, 2000; Disbrow, Roberts, Poeppel, & Krubitzer, 2001; Hinkley, Krubitzer,
18
Nagarajan, & Disbrow, 2007). Although previous studies have reported that integration of
proprioceptive and tactile information occurs in other areas (e.g., areas 1 and 2), work in non-
human primates has revealed neurons in both areas 3a and 3b respond to both proprioceptive and
tactile stimulation (Kim et al., 2015).
How somatosensory information conveyed through these pathways give rise to a sense of
body position during passive and dynamic actions is still an active area of inquiry for researchers
interested in human movement science. Currently, there is no agreed-upon model with the
explanatory power to account for all possible functions of the somatosensory system. Converging
fields of research have, however, produced models that are capable of explaining how
somatosensory cortical networks, which are active during movement planning and movement
control, might interact with proprioceptive receptors for the production and correction of action
(Bullock, Cisek, & Grossberg, 1998; Cisek, Grossberg, & Bullock, 1998).
19
&The! dorsalGmedialGlemniscal! pathway! for! the! transmission! of! proprioception! and! touch! information! to! the!somatosensory!cortex.!Information!from!muscle!spindles!and!cutaneous!receptors!enters!through!the!dorsal!roots!of!the!spinal!cord!then!ascends!ipsilaterally!to!the!medulla.!These!fibers!synapse!then!cross!(decussate)!at!the!medulla!and!ascend!toward!to!the!ventral!posterolateral!nucleus!of!the!thalamus!and!then!to!the!cortex!(adapted!from:!Noback,!Ruggiero,!Demarest,!&!Strominger,!2005).!!
20
! Somatosensory!Representations!During!Action!–!Vector!Integration!to!Endpoint!Model!
One sensorimotor model pertinent to this thesis, which describes the role of somatosensory
information and cortical areas in goal-directed actions, is the Vector Integration To Endpoint
(VITE) model proposed by Bullock and colleagues in 1996. This model outlines how the
somatosensory system participates in the generation and control of reaching movements under
various task constraints and is primarily based on single and population neuron activity recordings
of somatosensory regions in the cortex.
According to the VITE model, before movement begins, a difference vector between the
the estimated position of the limb at the target, and the limb’s current position is calculated. This
difference vector, could be based on predictions or simulations (Blakemore, Wolpert, & Frith,
2000), and is characterized by a buildup of activity in somatosensory portions of area 5 based on
input from other parietal areas. Recall that the parietal areas receive projections from the dorsal
stream and could be a possible area for interactions between the visual and somatosensory systems
(Chuchland, Santhanam, & Shenoy, 2006; Reichenbach, Thielscher, Peer, Bülthoff, & Bresciani,
2014). At the same time, other neural populations within area 5 monitor the limb’s position in
space. The activity of these monitoring neurons is tonic in nature as they receive information about
static limb positions from sensory fibers (i.e., Type Ia and Type II muscle spindles and Golgi
tendon organs).
Information about the limb’s position and the prediction of the location of the target
contribute to the development of a movement command. In the VITE model, the output of the
movement is regulated by a velocity vector characterized by phasic-tonic neural activity in area 4.
The output of this velocity vector is modulated by a volitional “GO” signal, most likely emerging
from the higher centers of the brain (i.e., frontal regions). The neurons that generate the velocity
vector project to alpha motor neurons, which control extrafusal (i.e., force generating) muscle
fibers. When the command is sent, and the limb begins to move, information received from the
sensory fibers modulates firing patterns until the limb position vector matches the desired output.
This cortico-spinal VITE model has been tested using simulations, and these parameters
were sufficient to explain proprioceptive adjustments in the contexts of limb perturbation, target
21
perturbation, and Coriolis fields (Bullock, Cisek, and Grossberg, 1998). Although not
comprehensive, this model is indicative of the possible role proprioception has in the planning and
completion of goal direction actions (see Figure 5).
22
&The!Vector!Integration!to!Endpoint!Model!(VITE)!of!proprioceptive!online!control.!Before!movement!starts,!a!difference!vector!(DV)!is!calculated!from!a!comparison!of!the!target!position!vector!(TPV)!and!the!current!position!of!the!arm!(PPV).!Movement!commences!with!a!voluntary!and!scalable!GO!signal!that!scales!a!velocityGbased!command!(DVV)!that!activates!dynamic!gamma!motor!neurons!(γd).!The!DVV!produces!an!outflow!position!vector!(OPV)!which!activate!static!gamma!(γs)!and!alpha!motor!neurons!(α).!The!activity!of!the!OPV,!which!is!driven!by!movement,!and!is!altered!by!sensory!information!from!type!Ia!and!II!afferents,!changes!in!the!PPV!as!arm!position!changes.!This!feeds!back!into!the!desired!velocity!vector!and!alters!the!movement!of!the!arm.!Movement!ends!when!the!PPV!matches!the!DV.!This!Figure!was!adapted!from!Bullock!et!al.!(1996)!
!
23
! Sensorimotor!Transformations!for!Movement!Planning!and!Control!
Both the multiple-processes and VITE models provide a framework for interpreting how visual
and somatosensory information is used during movement. However, both models do not fully
explain how these two senses might interact during the planning and control of goal-directed
actions. To address this question, the following section outlines current theories and experimental
evidence investigating multisensory integration processes as they are related to the performance
of goal-directed actions. Furthermore, whether these processes are flexible and adjust to
environmental constraints is also discussed.
It is important to acknowledge that late action selection and early programming
mechanisms do also influence sensory integration process, neural activities, and behavioral
performance related to goal- directed aiming movements (Allport, 1987; Cisek, 2007; Hommel,
2004, 2009; Prinz, 1997; Welsh, 2011). Although these concepts are relevant, to focus this
introduction, only sensory integration processes related to the later stages of movement planning
(i.e., after the action required to obtain the goal is already determined) were reviewed.
! Multisensory!Combination!and!Integration!for!Movement:!General!Principles!
Research on multisensory integration during movement has traditionally focused on how sensory
information obtained from multiple sources is converted and used to guide action. In the context
of action planning and control, sensory information is often discussed with regard to frames of
reference and sensorimotor transformations. Simply put, a sensory reference frame is a coordinate
system wherein the locations or quantities of objects are specified (e.g., Alais, Newell, &
Mamassian, 2010; Ernst & Bülthoff, 2004; Pritchett, Carnevale, & Harris, 2012). These reference
frames are associated with a specific sense or sensory organ and are linked to the receptive fields
of the neurons they subserve. For example, visual receptive fields are anchored to the neuronal cell
features in the retina (see Section 2.2.1 above). Thus, if retinal sensory information is used to
localize positions or features of objects, information emerging from these receptors is said to be
24
mapped in a gaze-centred or retinotopic reference frame1. In addition, changes in the position or
orientation of the sensory organ or effector will change how locations or features of the perceived
objects are encoded (Batista and Newsome, 2000; see Figure 8). For this thesis, a visual or gaze-
centred reference frame was used to describe stimuli or movements if they were coded relative to
gaze-position. A non-visual or somatosensory reference frame was used to describe stimuli or
movements coded independent of gaze-position. For some studies below, the results are indicative
of limb-centred coding, which means that stimuli or movements were encoded with respect to
effector positions (see Figure 8).
The production of goal-directed actions requires the transformation of information from
many different reference frames into commands to be sent to the muscles to produce movements.
Prior to the motor command (or when a correction is needed), previous research has suggested that
both the limb and the target must first be represented in the same sensory map (Jeannerod, 1991;
Reichenbach, Thielscher, Peer, Bülthoff, & Bresciani, 2009). These sensorimotor transformations
are thought to involve numerous multisensory integration and combination processes.
! Multisensory!Combination!and!Integration!!Signals acquired from numerous sensory sources must be combined and integrated prior to being
used for sensorimotor transformations. Multisensory combination is the process of extracting
complementary (but not redundant) information from sensory afferences that are coded in different
reference frames. For example, when determining the position of a reachable object, multisensory
combination processes could involve the evaluation of: the eye position relative to the head; the
head position relative to the trunk; visual information about the goal object in relation to the hand;
hand position relative to the trunk; and the learned experience of the actor (Ernst & Bülthoff,
2004).
Sensory integration processes refer to the incorporation of these various sources of
1 Gaze-centred and retinotopic reference frames may not always refer to the same coordinate system. A Gaze-centred reference frame could refer to a reference frame that considers the position of the eyes in the head whereas a retinotopic reference frame could refer to a reference frame based solely on retinal activations. Thus, although related, they are not always the same (see Cohen & Andersen, 2002).
25
information with the goal of forming a useable percept or representation. Presently, the most
accepted model of multisensory integration suggests that each source of sensory information is
integrated based on a maximum likelihood or minimum variance estimate. That is, each sensory
source is converted and given a weight based on its variance. This weighting determines the degree
to which it contributes to the final percept (Ernst & Bülthoff, 2004). This model has been useful
for predicting performance in visual-haptic (Ernst & Banks, 2002) and audio-visual experimental
paradigms (Shams, Ma, & Beierholm, 2005), and has received support from neurological data
examining representations of sensory probability distributions in neural populations (Fetsch,
Pouget, DeAngelis, & Angelaki, 2012).
In the case of goal-directed actions, sensory integration processes could be viewed as the
transformation of position signals derived from visual and somatosensory senses about the location
of the arm, and the location of a goal object, to construct or amend a movement trajectory. When
considering this process, often an intermediate step is proposed whereby sources are converted
into a common reference frame (Ambrosini et al., 2012; Beurze, Van Pelt, & Medendorp, 2006;
Cohen & Andersen, 2002; Jeannerod, 1991; Mueller & Fiehler, 2014b; Pouget, Ducom, Torri, &
Bavelier, 2002). It is hypothesized that the establishment of this common reference frame involves
designating a space common to all inputs and transforming the locations of sensory signals into a
common coordinate system. One goal of this thesis was also to examine how task requirements
and changes in sensory information about the target alter these transformation processes prior to
and during goal-directed actions.
! Multisensory!Combination!and!Integration!During!Movement!Planning!and!Control!
Presently, there is a debate within the literature as to the composition (or even existence) of a
common reference frame for sensorimotor transformations (Beurze et al., 2006; Cohen &
Andersen, 2002; Feldman & Levin, 1996; McGuire & Sabes, 2009; Pritchett et al., 2012). On one
side, it has been argued that the sensory signals used for defining the initial movement vector, as
well as those used for movement control are mapped exclusively in gaze-centred coordinates. In
contrast, a growing body of literature has also built the case for the use of mixed or somatosensory
reference frames.
26
! The!Common!Reference!Frame:!Visual!Reference!Frame!for!Movement!Planning!and!Control!!
The idea of a common reference frame for motor planning and control can be traced back to early
experimentalists such as Karl Lashley, whom, in his work The Problem of Serial Order in
Behaviour, posited that the central nervous system’s ability to effectively control rapid actions was
made possible, by comparing sensory signals in “reference to a space system” (Lashley, 1951, p
126). Furthermore, he theorized that “perceptions from distance receptors, vision, hearing, and
touch are also constantly modified and referred to the same space coordinates” (Lashley, 1951, p
126). Lashley, a behaviourist by training, had constructed this idea using his own observations
from his experiments on cats, humans, and non-human primates (Lashley, 1917). He was
particularly fascinated by the ability of humans and animals with damaged somatosensory
afferents to still generate accurate movements at varying speeds. Although Lashley did not directly
implicate a specific sensory modality, his observations that movements can be made without
proprioception served as the basis for future work on visual-sensorimotor transformations
(Desmurget, Pélisson, Rossetti, & Prablanc, 1998). Since Lashley’s initial observations, the idea
of a common visual reference frame for action related sensory information processing has received
copious amounts of support from both psychophysical experiments examining human behavior,
and neurophysiological experiments looking a single and population neuronal activities
(Ambrosini et al., 2012; Cohen & Andersen, 2000; Crawford, Medendorp, & Marotta, 2004;
Henriques, Klier, Smith, Lowy, & Crawford, 1998; Pouget et al., 2002; Stetson & Andersen, 2014;
Stricanne, Andersen, & Mazzoni, 1996; Zipser & Andersen, 1988).
!Behavioural!Evidence!for!a!GazeGCentred!Reference!Frame!for!Movement!Planning!and!Control!
Evidence for the use of a gaze-centred reference frame for movement planning and control is
primarily derived from studies looking at the effect of target-eccentricity and gaze-direction on
movement endpoint errors. For example, Bock (1986) noted that when participants performed
reaching movements to visual targets presented in the retinal periphery, without vision of their
reaching limb, they “reached too far to the right if targets were presented in the right hemispace,
and too far to the left if targets were presented in the left hemispace” (Bock, 1986, p. 478).
Moreover, in a second experiment, wherein targets were only presented in the right hemispace (as
27
participants maintained central fixation), the author noted a stable and systematic bias in horizontal
reach endpoints towards the right of the targets. Because this bias was related to gaze-direction,
the author concluded that upper-limb reaching movements were planned and controlled in gaze-
centred coordinates (see also Bard, Hay, & Fleury, 1985; Prablanc et al., 1979).
Henriques et al. (1998) provided further evidence for a gaze-centred reference frame during
movement planning when they sought to determine if eye movements after target disappearance
affected aiming performance. In their experiments, participants were presented with a target
located at 0 degrees with respect to the participant’s right eye. After the target disappeared,
participants performed a saccade to a point located 15 degrees to the left of the original
(disappeared) target location and subsequently performed a reaching movement to the memorized
target position with the upper-limb. The authors noted that reaching endpoints were biased toward
the right of the target, replicating what was found by Bock (1986). These authors concluded that,
in the primary stages of motor planning, the information used to construct the initial movement
vector is stored in a dynamic gaze-centred map and this vector is remapped after each eye
movement (see also Medendorp & Crawford, 2002).
Evidence for gaze-centred mapping during movement planning and control has also been
demonstrated in experiments looking at the effect of target-to-body versus target-to-gaze position
(Ambrosini et al., 2012). Movement endpoint variability and the degree of endpoint bias
significantly increased with retinal eccentricity of the target. Also, this relationship was
independent of body position. Furthermore, when individuals are performing a visuomotor task,
differences between perceived and actual body position at one spatial location led to movement
control errors representative of a gaze-centred remapping of the entire workspace (Thompson,
Byrne, & Henriques, 2014; Vetter, Goodbody, & Wolpert, 1999). These findings suggest that
movement control processes occur in also accordance with a gaze-centred reference frame.
Although, much of the work mentioned above focused on visual targets, there has been
experiments looking at how other sensory modalities could be mapped in visual coordinates. For
example, building off of the result that arm movements made to either seen or memorized visual
targets are affected by gaze position (e.g., Henriques et al., 1998; Medendorp & Crawford, 2002),
Pouget et al. (2002) sought to determine if this gaze-centred mapping could be used for targets of
28
different sensory modalities (see Figure 6). To this end, the authors conducted an experiment
wherein human participants performed aiming movements to visual targets, auditory targets (the
perceived source projected sounds), proprioceptive targets (right foot), and imaginary targets.
Similar to the previous experiments, the retinal locations of the targets were altered by changing
the ocular fixation points and movements were performed without vision of the limb. It was found
that changes in fixation position resulted in endpoint bias for all types of targets, regardless of their
sensory modality (see also Blangero, Rossetti, Honoré, & Pisella, 2005; Jones & Henriques,
2010). Thus, the authors concluded that prior to goal-directed actions, the target location and initial
movement vector are represented in gaze-centered coordinates.
29
30
&Figure!adapted!from!Pouget!et!al.!(2002).!A)!The!experimental!paradigm,!participants!fixated!on!a!peripheral!fixation!point!(FP)!prior!to!making!a!movement!to!a!central!target.!Targets!were!either!visual!(led),!auditory!(sound!from!a!speaker)!or!proprioceptive!(projected!position!of!the!participant’s!right!foot).!Note!that,!in!all!conditions,!participants!pointed!towards!the!targets!without!actually!touching!them!B)!GazeGdependent!reaching!errors!for!each!target!modality.!Movement!endpoints!to!auditory,!visual,!and!proprioceptive!targets!were!all!significantly!biased!by!gaze!direction.!!
31
!Behavioural!Evidence!for!Visual!Space!Remapping!After!Target!Perturbations:!Target!Perturbation!Paradigms!
Experiments examining the impact of the relative positions of gaze and the body on reaching
performance show that a visual reference frame may be used for the control of ongoing actions
(e.g., Ambrosini et al., 2012; Thompson et al., 2014). Other experiments have also assessed online
control more directly using target perturbation paradigms (also known as the double-step
paradigm). In the double-step paradigm the initial target location of a goal-directed reaching
movement is shifted (i.e., perturbed) either during movement planning (e.g., Flash & Henis, 1991)
or during online control (e.g., Komilis, Pélisson, & Prablanc, 1993; Reichenbach et al., 2009;
Sarlegna & Blouin, 2010). In the former scenario, participants can adjust their movement plan to
successfully complete the task, whereas in the latter scenario participants must compute and
implement changes to their hand trajectory as the ongoing movement unfolds. Previous studies
have hypothesized that the computation of online trajectory amendments is based on visual
information about both the new target location and reaching limb position (Desmurget et al., 1999;
Komilis et al., 1993; Reichenbach et al., 2009). Thus, when there is no vision of the reaching limb,
sensorimotor transformations would be required to obtain a visual estimate of limb position prior
to the implementation of corrections.
Evidence that online corrections occur in a visual reference frame are derived from studies
employing the double-step paradigm without vision of the reaching limb. Komilis et al. (1993),
performed an experiment wherein participants reached with or without vision of their limb to
targets that were stable (75% of trials), or perturbed at movement onset or peak velocity (25% of
trials). In contrast to other double-step paradigms (e.g., Goodale et al., 1986; Pélisson, Prablanc,
Goodale, & Jeannerod, 1986) participants in Komilis et al. (1993) were fully aware that the target
had changed position. In response to perturbations at movement onset, the authors found no
differences in movement endpoint errors between trials with the limb visible and trials with no
vision of the reaching limb. In contrast, for perturbations at peak velocity, the authors found that
participants only partially corrected for the perturbation. Furthermore, participants corrected to a
greater extent (i.e., 38% of perturbation distance) when they had vision of their limb compared to
when they had no vision of their reaching limb (i.e., 28% of perturbation distance). The authors
32
stated that, when there was vision of the limb, movement corrections were programmed based on
the retinal information about the new target location and current limb position. However, when the
limb was not visible, corrections were programmed based on retinal information about the target
location and visual estimates of the limb position derived from somatosensory and efferent
information. The authors concluded that the processes required to transform somatosensory
information onto a visual reference frame prior to these computations takes time, and thus leaves
less time available to implement trajectory corrections. This increase in error therefore was
expected as there were no between vision condition differences in movement times (note that times
ranged from 460 to 516 ms for the perturbed trials).
The findings that somatosensory information must be transformed into a visual reference
frame during online control were also supported by the work of Reichenbach et al. (2009). In their
study, participants performed reaching movements to visual targets located 20 cm away from a
home position using a 4 degree-of-freedom haptic manipulandum. The experimental protocol
consisted of perturbations to both the visual targets (i.e., shift in target position 7.5 degrees to the
left or to the right of the original target) and the reaching limb (10 N force the left or to the right
applied to the arm by the manipulandum) which occurred either early (1 cm from starting position)
or late (5 cm from starting position) in the trajectory. Correction latencies to perturbations were
computed using both kinematic measures and muscle activations and were compared between
limb-visible and limb-occluded trials. In support of the conclusions of Komilis et al. (1993),
Reichenbach et al. (2009) found that correction latencies in response to early visual target
perturbations were longer when movements were performed without vision of the limb compared
to when they were performed with vision of the limb (see Figure 7). The authors concluded that
this increase in correction time was associated with the remapping of somatosensory information
onto a visual reference frame prior to corrections.
The conclusions of the previous studies suggest that online control takes place in a visual
reference frame. One limitation of the extant literature, however, is that this type of paradigm has
seldom been used to examine movements to targets of other sensory modalities. This question will
be addressed in Study B of this thesis which investigates the online control of movements to
perturbed somatosensory targets.
33
&Experimental!setup!and!results!of!visual!target!condition!adapted!from!Reichenbach!et!al.!(2009).!Participants!made!movements!to!visual!targets!that!either!remained!stable!or!were!shifted!early!or!late!in!the!trajectory.!Movements!without!vision!of!the! limb!had! larger!endpoint!errors!and! longer!correction! latencies!than!movements!made!without!vision!of!the!limb.!
34
!Neurophysiological!Evidence!for!a!GazeGcentred!Reference!Frame!for!Movement!Planning!and!Control!
Neurophysiological evidence supporting the use of a gaze-centred reference frame for goal-
directed actions has been derived from the observed activity of single neurons and neuronal
populations prior to, and during reaching tasks. Overall, this literature indicates that the patterns
of neuronal activity in parietal, premotor, and motor areas support the use of a visual reference
frame for movement planning and control.
The main cortical area implicated in the transformation of sensory signals for planning and
control of goal-directed actions is the posterior parietal cortex (PPC). Most studies on the motor
related activities of the PPC have found evidence of gaze-centred mapping. The seminal studies
touting the PPC as the neurophysiological conduit of the visual reference frame also revealed that
the gain of PPC neuron activities are modulated by changes in eye position (Andersen &
Mountcastle, 1983; Mountcastle, Lynch, Georgopoulos, Sakata, & Acuna, 1975). Specifically,
PPC neurons were found to fire if the angle of an illuminated target was in the preferred direction
of the retinal receptive field. Also, when the eyes moved, the intensity of firing was altered as a
function of the target’s distance from neurons’ preferred directions. Furthermore, it has been
demonstrated that, prior to reaching movements to visual targets, activations patterns of PPC are
altered by gaze position (Batista & Newsome, 2000). This gaze-dependent gain shift in neuronal
activity has also been shown for movements to auditory targets and in delayed reaching paradigms
(Gillman, Cohen, & Groh, 2005; Stricanne et al., 1996). In agreement with the aforesaid
behavioural experiments (Section: 2.4.3.1.2), these findings suggest that similar networks are used
for gaze-centred transformations prior to reaches toward auditory, visual, and memorized target
locations. Furthermore, network analyses of neuronal population activity in the PPC has
demonstrated that it is possible to derive target locations in the external world based on neural gain
field activities (Burnod et al., 1999; Zipser & Andersen, 1988). The above-mentioned results,
together with PPC’s role in the planning and control of saccades and limb movements (for reviews
see Andersen et al., 1997; Cohen & Andersen, 2002), have led researchers to conclude that PPC
gain responses represent a gaze-centred transformation network used for the planning and control
of goal-directed actions.
35
Gaze-centred activations in PPC have also been hypothesized to drive other areas
associated with movement planning. Although studies examining neural activity in premotor and
motor cortex have revealed that their receptive fields code positions in hand-centred coordinates
(see Battaglia-Mayer, 2003; Bremner & Andersen, 2012; Burnod et al., 1999), researchers have
found that both eye position and retinal limb position do modulate neuronal firing in some
premotor areas (Boussaoud & Bremmer, 1999; Rozzi, Ferrari, Bonini, Rizzolatti, & Fogassi,
2008). For example, Boussaoud et al. (1998) examined the effect of gaze position and retinal target
location on dorsal premotor cortex activity as monkeys made movements to visual targets. It was
found that dorsal premotor activity was modulated principally by three factors: limb movement
direction, retinal target location, and eye position in their orbit. These results were corroborated
by other experiments examining similarities in activation patterns between the PPC, ventral
premotor, and prefrontal cortex (Boussaoud, Barth, & Wise, 1993; Mushiake, Tanatsugu, & Tanji,
1997). In addition, the discovery of interactions with other cortical areas during movement
planning as well as the discovery of kinematic-feature specific coding in PPC neurons (Chen,
Reitzen, Kohlenstein, & Gardner, 2009) have led to the idea that PPC serves as a sensorimotor
interface.
Support for PPC as a gaze-centred sensorimotor interface for movement related
sensorimotor transformations comes primarily from data about regional differences in PPC neuron
activity and virtual lesion studies. As mentioned above, PPC neural activity in response to different
target modalities is representative of a gaze-centred reference frame. Theoretical and
computational models propose that all sensory information important for the movement enters the
PPC via the SC and extrastriate visual areas (Andersen, Essick, & Siegel, 1985; Buneo &
Andersen, 2006; Zipser & Andersen, 1988). Subsequently, this sensory information is used to
adjust the gain of dorsal PPC neurons in the inferior parietal lobule (IPL) and intraparietal sulcus
(IPS) before being transformed into more hand-centred coordinates in the superior parietal lobule
(SPL).
Coherent with the above-mentioned empirical evidence, movement related activity in IPL
and IPS neurons are found to code target and effector positions in gaze coordinates (Andersen et
al., 1997; Cohen & Andersen, 2002). These include sensory information emerging from multiple
sources, which is pertinent to the reaching movement (Andersen et al., 1997). As information
36
travels to the SPL, two different types of cells are noted. First, in the dorsal areas of the SPL, closer
to the IPS, are neurons with firing patterns similar to the firing patterns of gaze-centred neurons
mentioned above (see Figure 5 in Buneo & Andersen, 2006). In contrast, neurons in the ventral
SPL appear to code effector and target positions similar to the hand-centred neurons of premotor
and motor cortex (Ashe & Georgopoulos, 1994; Bremner & Andersen, 2012; Fabbri, Caramazza,
& Lingnau, 2010). The anatomical arrangement of these cells, their interactions with premotor and
motor areas, as well as the notable absence of cells with mixed reference frames argue for a direct
transformation of gaze-tuned representations of target and hand positions to motor representations
for action (Andersen et al., 1985; Andersen et al., 1997; Batista & Newsome, 2000; Buneo &
Andersen, 2006; Cohen & Andersen, 2000). Furthermore, this direct transformation hypothesis is
supported by the role of the PPC in the online control of reaching based on forward models and
sensory feedback. It has been demonstrated PPC lesions lead to deficits in predicting arm position
(Wolpert, Goodbody, & Husain, 1998) and virtual lesions to the PPC induced by TMS result in
online control impairments when participants perform reaches in the double-step paradigm
(Desmurget et al., 1999).
Overall, the work presented above argues for a gaze-centred reference frame for movement
planning and control. Sensory information from many sources is used to build a map of the
environment in visual coordinates. The gain of neuronal activity related to the visual mapping of
sensory information may be modified by inputs from other sensory modalities (see Figure 9).
37
&Differences! between! handGcentred! and! gazeGcentred! neural! receptive! fields! (RF).! HandGcentred! neural!activations!shift!with!the!position!of!the!hand!in!space,!and!gaze!centred!activations!shift!with!the!position!of!the!eyes.!Both!types!of!representations!have!been!noted!in!the!PPC.!Adapted!from!Batista!and!Newsome!(2000).!!
38
‘
&The! common! eyeGcentred! reference! frame! gain! field! hypothesis.! A)! Transformations! in! the! PPC.! The!distribution!of!reach!related!activities!in!the!PPC!is!arranged!from!primarily!eyeGcentred!in!the!inferior!parietal!lobule!(IPL)!to!heterogeneously!limbGcentred!and!eyeGcentred!in!the!superior!parietal!lobule!(SPL).!For!this!reason,!the!PPC!is!viewed!as!the!sensorimotor!interface!for!transformations!related!to!the!planning!and!control!of!goalGdirected!actions.!Sensory! information!about!target!and!effector!positions!from!pertinent!sources!modify!the!reach!related!activities!of!eyeGcentred!neurons.!Response!patterns!of!neurons!in!the!SPL!suggest!that!information!is!then!transformed!directly!into!limbGcentred!coordinates.!!B)!Modulations!of!reach!related!neurons!in!the!PPC.!Neurons!in!the!PPC!fire!with!respect!to!changes!in!eye!position.!The!gain!of!these!responses!can!be!modulated!by!visual!information.!Figure!representations!adapted!from!data!described!in!Buneo!&!Andersen!2006,!and!Cohen!&!Andersen!2002.!
39
! The!Common!Reference!Frame:!!Somatosensory!Reference!Frames!for!Movement!Planning!and!Control!!
As outlined above, there is ample evidence supporting the existence of a visual reference frame
for sensorimotor transformations occurring prior to, and during goal-directed actions. A growing
body of evidence has called into question the idea that sensorimotor transformation processes are
performed exclusively in this way. Based on the maximum likelihood model of sensory integration
briefly described in Section 2.4.2 (see also Ernst & Banks, 2002), this emerging axiom of literature
purports that the sensorimotor transformation strategies employed by the central nervous system
are dependent on both the availability of sensory information and the relevance of this information
to the planning and completion of the action (Burnod et al., 1999; McGuire & Sabes, 2009;
Saradjian, 2015; Sober & Sabes, 2005).
One of the first studies to examine the relative importance of retinal versus other sensory
signals for motor control was an observational study conducted on behaving monkeys (Cohen,
1961). In this experiment, monkeys had either their extra-ocular muscles severed and had their
ciliary muscle paralyzed, or had their neck muscles paralyzed, before being observed in an
interactive environment. Cohen observed how the loss of these different types of sensory
information affected their ability to interact with their environment. Almost to the surprise of the
author, monkeys with severed extra-ocular muscles showed no severe deficits in behaviour, while
monkeys lacking neck proprioception demonstrated severe motor control deficits akin to those of
serious vestibular damage. The conclusion of this study was that information about the head
relative to the body location provided by neck proprioceptive sensations contributed significantly
to behaviour, and extra-ocular (or eye) signals may play less of a crucial role than previously
thought.
The idea that somatosensory information could be as crucial as eye position information
for behaviour was later built on by studies on goal-directed reaching. These studies have
demonstrated that somatosensory information is important limb movement control processes in
visually-guided actions (Ghez, Gordon, Ghilardi, & Sainburg, 1995). Neurologically intact
participants and participants with large fibre neuropathy performed aiming movements to targets
varying in amplitude and direction. Vision of the reaching limb was also precluded systematically
for each target, while limb kinematics and movement performance variables were recorded.
40
Patterns of movement errors, limb accelerations, and trajectory profiles revealed that vision
contributed more to movement direction planning. In contrast, somatosensory information was
used to update the internal model of the limb position. Based on these results the authors proposed
a model wherein target positions are defined primarily in visual coordinates (an interpretation
influenced by the neurophysiological work described in Section 2.4.3.1.3 above), and initial limb
position is primarily defined in somatosensory coordinates. Furthermore, the authors stated that
differences in error patterns between patient groups and vision conditions suggest that both sensory
modalities contribute to online control processes.
The proposition that modifying vision and proprioception produced modality-specific
changes in movement errors and trajectory kinematics gave rise to the hypothesis that each sense
may have differing effects on sensorimotor transformations (McGuire & Sabes, 2009; Sober &
Sabes, 2005). Sober and Sabes (2005) conducted a seminal study to specifically test this
hypothesis. The authors employed a reaching task to visual and proprioceptive targets and
systematically altered proprioceptive and visual feedback of the limb’s position. To ascertain the
effect of target modality on sensorimotor transformation processes, two error variables were
computed. The first was a movement vector error that represents the error in the computed vectors
of the visual target and initial hand position. The second was an inverse model error that
represented the differences in planned direction of the movement converted into muscle or joint
based motor commands (see Sober & Sabes, 2005 for details). Performance data from multiple
experiments were used to construct a mathematical model that predicted the contributions of each
sense to the measured error outcomes. Vision of the limb was found to more accurately predict
movement vector error performance for aiming movements to visual targets, whereas limb
proprioceptive sensory information had a stronger predictive power for movements to
proprioceptive targets. These results were in contrast to the visual reference frame hypothesis, as
this would predict vision should have an equally important role for both types of targets.
Overall the above results suggest that early sensorimotor integration processes can be
modulated by task constraints, and perhaps a different type of transformation is used when
performing movements to proprioceptive targets.
41
!Behavioural!Evidence!for!a!Somatosensory!Reference!Frame!in!Movement!Planning!and!Control!
Since Sober and Sabes (2005), an accumulating body of evidence has emerged supporting the
existence of an alternative reference frame for sensorimotor transformations. Much of this support
is drawn from the observed differences between movements made to visual and somatosensory
targets. Overall, these studies provide further evidence that sensorimotor transformations used for
movements to somatosensory targets are coded in an somatosensory reference frame and this
coding could be a strategy to reduce neuromotor noise (Bernier, Gauthier, & Blouin, 2007;
Sarlegna & Bernier, 2010; Sarlegna, Przybyla, & Sainburg, 2009; Sarlegna & Sainburg, 2007).
Further support for sensorimotor transformations in a non-visual reference frame can be
taken from experiments looking at adaptations and limb kinematics. For example, in Bernier,
Gauthier, & Blouin, (2007), two groups of participants underwent a prismatic adaptation protocol
where they practiced aiming movements to visual targets. After the adaptation period, one group
of participants performed a post-test using only visual targets, while another other group was tested
using peripheral visual targets and a somatosensory target (i.e., the fingertip position of the non-
reaching hand). The authors observed a prismatic after-effect for movements to visual targets but
found no evidence of such a shift when participants performed post-test trials to the somatosensory
target (see Figure 10 A). The authors concluded that separate representations could be used for
each target modality. Thus, the somatosensory transformations used for somatosensory targets
were not biased by the visual prismatic adaptation protocol.
Studies that altered visual and proprioceptive information about initial limb position also
reported that sensorimotor transformations to somatosensory targets occur in a non-visual
reference frame. In Sarlegna & Sainburg, (2007), Two groups of participants performed reaching
movements to either proprioceptive (i.e., again reaching to the fingertip of the non-reaching hand)
or visual targets. The experimenters varied visual position of the reaching limb before movement
onset creating a visual-proprioceptive mismatch. Amplitude and direction errors as well as
movement kinematics were computed to gauge the contributions of each modality to the motor
plan. When reaching to visual targets, movement distance error was dependent on initial visual
information but not proprioceptive information. In contrast, for movements to proprioceptive
targets, it was found that visual information contributed less to movement distance errors than
42
proprioceptive information. Overall, these results contrast with the classical two-stage model of
motor planning where visual information is used to plan the vector and proprioception is used to
adjust the visual gain response (see “Visual Reference Frame” discussions above in Section
2.4.3.1.3). Instead, these results lend further support for the existence of a different, possibly more
direct, sensorimotor transformation process for movements made to somatosensory targets.
One reason for these differential transformation strategies could be to reduce the noise that
emerges from reference frame conversions. During motor planning, different sources of sensory
information are represented in many reference frames and conversion from one frame to another
could incur a time cost and induce errors, thus hampering performance. This hypothesis was tested
by having individuals perform reaching movements to either proprioceptive or visual targets
without vision of their reaching limb (Sarlegna et al., 2009). Inverse dynamics analysis of shoulder
and elbow torques was used to assess if the transformation of information between reference
frames affected performance. For movements to visual targets, participants exhibited more curved
movement trajectories compared to movements to proprioceptive targets. Also, inverse dynamics
analyses revealed shoulder and elbow muscles were initially counteractive to straight movement
directions for visual targets. In contrast, movements to proprioceptive targets showed less
curvature and less overall muscle torques around both joints (see Figure 10B). These results
indicate that there are more efficient coordination patterns for proprioceptive targets performed
without vision of the reaching limb. Overall, the authors concluded that sensorimotor
transformations occurring between different reference frames produces errors in motor planning.
43
&A)!Adapted!from!Bernier!et!al.!(2007).!Participants!exhibited!lower!errors!due!to!prismatic!adaptation!when!reaching!to!a!proprioceptively!defined!target!than!a!visual!target.!!B)!Adapted!from!Sarlegna!et!al.!(2007),!movements!to!somatosensory!targets!were!less!curved!than!movements!to!visual!targets!in!the!same!position.!Both!of!these!results!argue! that! different! sensorimotor! transformation! mechanisms! underlie! movements! to! visual! and! somatosensory!targets.!!
44
! The!Common!Reference!Frame:!!Neurophysiological!Evidence!for!Somatosensory!Reference!Frames!for!Movement!Planning!and!Control!!
Ample support for sensorimotor transformations using non-visual reference frames has also
emerged from experimental neurophysiology. Studies implementing a wide range of techniques
including: virology, immunohistochemistry, electroencephalography (EEG), functional magnetic
resonance imaging (fMRI), and transcranial magnetic stimulation (TMS), have all uncovered
possible networks capable of performing planning and control processes in somatosensory
coordinates. Overall, these studies argue that sensorimotor transformation processes during
movement planning and control can occur in a non-visual reference frame.
Anatomical investigations of the brains of monkeys and cats were the first to theorize the
existence of a network for sensorimotor transformation using somatosensory information (Babb,
Waters, & Asanuma, 1984; Jones & Powell, 1969; Jones & Powell, 1968; Langren, Silfventus, &
Wolsk, 1967). Stimulations, lesions, and categorizations of cellular architecture revealed that the
somatosensory cortex, motor cortex, and the PPC were interconnected and functioned reciprocally
to produce behaviour (Jones & Powell, 1969). It was hypothesized that information obtained by
the somatosensory areas from thalamocortical neurons could be directly projected to the PPC
(Babb et al., 1984; Langren et al., 1967). These connected areas could “well be part of a short
latency feedback of joint position information to motor cortex for the production of movements”
(Babb et al., 1984, p.483; see also Disbrow et al., 2000, 2001; Schmahmann & Pandya, 1990).
From the aforesaid studies, however, it was not possible to state the exact nature of these
connections because sensory regions of the thalamus were viewed as heterogeneous. Thus, it was
unclear whether networks carrying somatosensory information, either from the somatosensory
cortices or thalamus, had direct projections to movement planning areas (PPC). As a result, as
discussed above (Section 2.4.3.1.3), the interactions between proprioceptive information and
neural activity in the PPC during sensorimotor transformations were hypothesized to be
exclusively through visual gain field adjustments (Andersen et al., 1997).
Since the gain field hypothesis, experiments using immunohistochemistry have provided
compelling evidence for the direct transmission of somatosensory signals to the PPC via
thalamocortical neurons and neurons in somatosensory cortical areas (Prevosto, Graf, & Ugolini,
45
2010, 2011). To study these projections, researchers injected the PPC of macaque monkeys with a
viral (i.e., rabies) neural tracer (i.e., cholera toxin B). A monoclonal antibody2 directed at the rabies
virus was then used to visualize the neural pathways connected to each area. It was revealed that
areas of the PPC, specifically the ventral lateral (LIPv) and medial intraparietal areas (MIP),
received disynapatic inputs directly from somatosensory thalamo-cortical neurons. These neural
projections were also similar to those observed in somatosensory areas, providing evidence for the
direct transmission of information about neck and arm proprioception to the PPC. In addition, the
MIP also received inputs from somatosensory areas associated with the processing of
proprioceptive signals from muscle spindles and joint receptors (Prevosto et al., 2011). Taken
together with cytoarchitectural studies revealing somatotopic arrangement for somatosensory
signals in PPC (Rozzi et al., 2008), these results support the existence of a PPC network with the
physiological features to directly integrate proprioceptive signals.
The existence of a specific PPC network for the transformation of somatosensory signals
is also supported by studies looking at brain activity in humans performing reaches to
somatosensory targets. Bernier and Grafton (2010) examined the possible functions of this network
by looking at how different PPC regions (i.e., precunate nucleus vs. parieto-occipital junction),
respond to movements toward somatosensory targets. In addition to the sensory modality of the
target, the authors varied target location with respect to both body and gaze position. Neural
responses were measured by fMRI to determine what networks were active for each target location
change. Results indicated that anterior precuneus activity was activated to different extents
depending on target modality. Precuneus activations were greater for changes related to gaze
position when visual targets were present, and greater for changes related to body positions when
proprioceptive targets were present. Areas of the premotor and somatosensory cortices also
showed a greater activation for body-position changes when proprioceptive targets were presented.
The authors concluded that these results could represent interactions between PPC and
2 Monoclonal antibodies are proteins produced by lymphocytes that bind foreign protein molecules, such as those produced by viruses (see Alberts, Johson, & Lewis, 2002)
46
somatosensory areas to facilitate body-centred transformations during aiming movements to
proprioceptive targets.
Further support for a network facilitating non-visual transformations in PPC was provided
by experiments looking brain stimulation during reaching movements. For example, it has been
shown that TMS applied to the dorsal lateral PPC prior to reaching to memorized targets disrupts
initial estimates of unseen hand position rather than visual estimates of the reach vector (Davare,
Zénon, Pourtois, Desmurget, & Olivier, 2012; Vesia, Yan, Henriques, Sergio, & Crawford, 2008).
Furthermore, TMS delivered to particular region of the PPC (i.e., the posterior mIPS), when vision
of the hand was unavailable, impaired the efficacy of somatosensory-based online corrections, but
not visual-based online corrections (Reichenbach et al., 2014).
Overall, the experiments presented above argue for a network capable of using
somatosensory information for sensorimotor transformations during the planning and control of
goal-directed actions.
! Literature!Review!Conclusions,!Gaps,!and!Further!Questions!To summarize, both visual and somatosensory information contribute to the planning and control
of goal-directed actions. Sensory receptors from both senses are capable of coding limb and target
positions. Although theoretical models based on feedback and feedforward mechanisms have been
proposed to account experimental observations, no cohesive model presently exists.
An important remaining point of contention in motor control studies concerns the frame of
reference in which somatosensory information is processed during the planning and online control
of movement. This ongoing debate is based on observations that somatosensory targets can be
encoded in both visual and non-visual reference frames. The dominant hypothesis regarding the
selection of frame of reference posits that all movements are planned and controlled in visual
coordinates (see Section 2.4.3.1). The main underlying assumption of this scenario is that
somatosensory inputs from the reaching limb needs to undergo a series of transformations prior to
movement onset. The alternative hypothesis is that the encoding of somatosensory information is
flexible and depends on the sensory context and task goals (see Section 2.4.3.2).
47
The goal of the research presented in this thesis was to shed light on the ongoing debate
regarding the sensorimotor transformation processes underlying the planning and control of
movements directed towards somatosensory targets. In these experiments, participants were asked
to reach to specific fingers of the opposite hand (i.e., somatosensory targets) that were identified
either by tactile or auditory cues. In the first experiment, we used an established method to
determine if participants planned their movements using an exteroceptive (i.e., visual) or
interoceptive body representations by measuring gaze-dependent reaching errors. It was
hypothesized that if an exogenous cue (i.e., auditory) was used to signal the somatosensory target
locations, movements would be planned in a visual reference frame. In the second experiment, the
cortical mechanisms underlying the remapping of the somatosensory target position into retinal
coordinates was examined. It was hypothesized that the neural activities underlying sensorimotor
transformations into a visual reference frame would require greater visual processing in areas
involved in visuomotor transformations and gaze-dependent movement coding (i.e., PPC, lateral
occipital cortex) compared to conditions where no conversion of somatosensory information is
required (e.g., reaches to auditory-cued visual targets).
The third experiment of the present thesis examined the reference frame used for the online
control of movements toward somatosensory targets. This question has been much less
investigated than those related to the sensorimotor processes underlying movement planning. As
discussed in Section 2.4.3.1.2, the online control of actions has been considered in the literature to
be performed in a visual reference frame. However, the existence of fronto-parietal networks
capable of relaying somatosensory information to areas associated with online control (e.g., PPC;
Desmurget et al., 1999; Prablanc et al., 2003), and the existences of fast correction mechanisms
(e.g., Scott, 2004), lead to the possibility that movements to somatosensory targets can be
controlled in a non-visual reference frame. To test this hypothesis a study was conducted wherein
participants made reaching movement to visual and somatosensory targets that were displaced
after movement onsets.
48
&Outline!of!the!experiments!conducted!in!this!thesis,!Experiment!A1!and!A2!investigated!the!sensorimotor!transformations!used!for!planning!movements!to!somatosensory!targets,!whereas!Experiment!B!investigated!the!role!of!these!transformations!in!the!online!control!of!goalGdirected!actions.!!!
!
49
!!
! General!Methodology!
! Participants!The number of participants for each study was selected based on a power analysis done in
G*power. Mean and median (see Schimmack, 2012) population effect sizes, were derived from a
literature review of relevant studies (see Appendix 1 for further descriptions and analyses). For
every experiment, participants were right-handed, and had normal or corrected-to-normal vision.
Informed consent was obtained from each participant prior to participation. Depending on
the location of the experiments, a local ethics committee at either Aix-Marseille University, or the
University of Toronto approved the experimental protocol.
! General!Procedures!The apparatuses and material used varied substantially by experiment, thus each is described in
detail in the methods section presented in each study. In general, all tasks, to some extent, involved
an upper-limb reaching movement. The involvement of motor networks in higher cognitive
processes, sensory processes, and the continuous and parallel features of the activations underlying
reaching movements, make upper-limb reaches ideal for the study of complex phenomena,
specifically with regard to sensorimotor integration (Allport, 1987; Bekkering & Neggers, 2002;
Cisek & Kalaska, 2005; Craighero, Fadiga, Rizzolatti, & Umiltà, 1999; Hommel, 2004).
Competition between choices (Cisek & Kalaska, 2005; Welsh, Elliott, & Weeks, 1999), spatial
recognition and attention (Craighero, Fadiga, Rizzolatti, & Umiltà, 1999; Tipper, Lortie, & Baylis,
1992), and even language processing capabilities (Song & Nakayama, 2009) can be distinguished
by looking at trajectories mapped onto Cartesian space. Furthermore, analysis of the kinetic and
kinematic features of reaching behaviours can delineate the contributions of planning and online
control processes during action (Chua & Elliott, 1993; Elliott, Helsen, & Chua, 2001; Elliott et al.,
2010; Flash & Hogan, 1985; Heath, 2005).
50
! Behavioural!Variables!Recording!and!Analysis!Data about the reaching trajectory were collected either with an Optotrak Certus (Northern Digital
Inc., Waterloo, ON, Canada) sampling at 200 Hz, or a Flock of Birds electromagnetic sensor
(Ascension, Burlington) sampling at 100 Hz. Data from both devices were analyzed using custom
Matlab (The Mathworks Inc.) scripts and custom software (Analyse, Marcel Kaszap, Canada).
! Temporal!Measures!
Reaction time (RT), movement time (MT), were the main behavioural temporal variables
computed in this thesis. RT was defined as the time from the movement go signal, to the start of
the limb movement. RT was identified by either the resultant velocity of the limb above 30 mm/s
for 2 samples in a row or the sample where a micro switch was released (see experimental methods
for further details). MT was calculated as the time between the start of the movement start and
movement end, which was defined as the first sample that the limb velocity went below 30 mm/s.
! Error!Measures!
The main error measures used in Study A were radial error, defined as the absolute distance
between target and the finger at movement offset, and movement vector error (i.e., angular error),
defined as the angular difference between the start to target vector and the start to movement end
vector. In Experiment A1 where ocular fixation points were altered, and performance was
compared to that of a central target (see Experiment A1 methods below for details), a population
Z score transformation was used to normalize performance to the control condition(s).
For Study B, constant error and variable error in the movement amplitude and movement
direction axes were used as measures of performance. Constant error was calculated as the bias in
endpoint position and the participants’ perceived target positions. For the amplitude axis, positive
constant values meant the participant exhibited an overshooting bias when aiming towards the
target whereas negative constant error values indicated that participants undershot the target
location. For the direction axis, positive and negative constant error were defined in relation to a
coordinate system (see methods of Study B for details).
51
Variable error was calculated as standard deviation of movement endpoints. Constant error
provides a measure of endpoint bias and whereas variable error provided an idea of the distribution
of endpoint positions.
! Latency!of!Online!Corrections!
Quantifying the online control of goal-directed actions is an ongoing challenge for researchers
interested in human movement science (see Khan et al., 2006 for a review). The methods employed
to measure online control vary greatly depending on the experimental apparatus and the
manipulations used to induce online corrections. As mentioned in Section 2.4.3.2, the use of target
perturbations is a prominent method for examining online control. The degree of online control
using these methods is quantified by contrasting characteristics of movements to perturbed targets
and of movements to stationary targets. The characteristics used for these contrasts have included:
muscle activations (Carlton & Carlton, 1987; Reichenbach et al., 2009); temporal-kinematic
features of movement trajectories (Oostwoud Wijdenes et al., 2013; Prablanc & Martin, 1992;
Saunders & Knill, 2003; Veerman, Brenner, & Smeets, 2008); and oscillation frequencies of
reaching limb (de Grosbois & Tremblay, 2016, 2018). The identification of the onset of corrective
responses (correction latency) is often of particular interest. For this computation, a variety of
approaches, including: thresholds (Reichenbach et al., 2009, 2014); statistical tests (Brenner &
Smeets, 1997; Heath, 2005; Saunders & Knill, 2003); and more contemporarily linear
extrapolation (Oostwoud Wijdenes et al., 2013; Oostwoud Wijdenes, Brenner, & Smeets, 2014;
Veerman et al., 2008); have been applied to position, velocity, and acceleration traces to generate
estimates (see the introduction section of: Oostwoud Wijdenes et al., 2014 for a brief overview) .
Although there is no universally accepted standard for computing correction latency, previous
work has revealed that reliable and accurate methods of detection do exist (Oostwoud Wijdenes et
al., 2014).
In the present thesis, the method of determining the latency in response to target
perturbations was adapted from Oostwoud Wijdenes et al. (2013). In their study, participants
performed 90 cm reaching movements to visual targets that could be perturbed 5 cm in the
movement direction, movement amplitude, or both axes. To compute correction latency in
response to perturbations in target direction, the difference the acceleration profiles between
52
perturbed and unperturbed trials in the axis of perturbation was computed. The author’s first
identified the maximum difference in the acceleration occurring after perturbation onset. Then, a
line was drawn between the points on the acceleration profile corresponding to 25% and 80% of
the maximum difference. The correction latency was defined as the time interval between the
perturbation and where this line crossed 0 (i.e., y = 0; see Figure 12D for an application of this
method to a subset of data in Study B). Using simulated movement data, Oostwoud Wijdenes et
al. (2014) examined the efficacy of different assessments of online control, the extrapolation
method was deemed to be the most accurate and precise method for detecting correction latencies
(see Figure 12).
!
53
&Figure! outlining! the! extrapolation!method! for! determining! correction! latency! as! described! by!Oostwoud!Wijdenes! et! al.! (2014).! The! authors! tested! the! predictive! ability! of! different! computations! of! correction! latency! to!estimate!a!100!ms!latency!to!target!perturbations!of!1!to!4!cm!in!simulated!movements!of!300!–!500!ms.!The!methods!included!threshold,!confidence!interval!and!the!extrapolation!method.!Panel!A)!shows!the!main!findings!of!Oostwoud!Wijdenes!et!al.!(2014)!for!the!3!cm!perturbation!stimulations,!the!extrapolation!method!applied!to!averaged!acceleration!data! yielded!a! very! accurate! prediction! of! correction! latency! across!movement! times.!Panel!B)!shows! the!method!applied!to!data!in!the!Study!B!of!this!thesis.!Averaged!acceleration!profiles!for!the!perturbation!curve!showed!in!panel!B!were!derived!from!movements!to!somatosensory!targets!perturbed!away!from!the!body.!!!
54
! Robotic!Guidance!Apparatus,!Development,!and!Testing!An Epson Selectively Compliant Assembly Robot Arm (SCARA; Epson E2L853, Seiko Epson
Corp.) capable of moving in four degrees of freedom with a 0.02 mm spatial repeatability was used
to deliver target perturbations in thesis Study B. This robot was used in a previous study to deliver
guidance trajectories (Manson et al., 2014).
To perform the real-time manipulations reported in Study B, a new real-time
communication system was developed between the motion capture apparatus mentioned above
(Optotrak) and the robotic guidance device. There were two problems that needed to be addressed.
The first was the dramatic decrease sampling resolution that occurred when both the robot and
motion capture systems were run on the same computer (EPSON RC420: Microsoft Windows XP:
Intel Celeron processor, 851 Mhz: 1GB RAM). The second, and related issue, was the magnitude
and variability of the lag time between the reception of real-time data from the motion capture
system and movement of the robot (see Appendix 2 for initial values).
To solve these problems, control of the motion capture system was moved to another
computer (Microsoft Windows XP: Intel Pentium ® 3.00 GHz; 2.00 GB RAM). Syncing of the
motion capture with the robot was achieved by creating a local network using an ethernet cross-
over cable, and two custom made parallel output boards. Communication between the motion
capture system and robot guidance was done using both electrical signals and network-shared
variables outputted from a custom MATLAB script on both computers (version 7.10, R2010a and
version 7.2 R2006a) and a custom SPEL+ script (version 4.2) interfaced with MATLAB through
custom dynamic linked libraries.
The sampling frequency, perturbation times, and perturbation speed (acceleration and
deceleration values) used in Study B were chosen partially as a result of output and input frequency
tests for each custom parallel port i/o board, transfer time from motion capture to network variable
writing, and lag time from variable reading to detectable robot motion (for all tests, analyses, and
code used to establish communications: see Appendix 2).
55
! Electroencephalography!Apparatus,!Recordings,!and!Analyses!
Electroencephalography (EEG) was used in Experiment A1 to measure the brain’s responses to
visual stimuli during sensorimotor transformations. Overall, EEG measures changes in electrical
activity occurring in the extracellular spaces of the brain over time. Although numerous cellular
activities contribute to the charge expressed in the extracellular matrix (see Table 1), activities
measured by EEG are commonly associated with post-synaptic potentials generated by the
depolarization and hyperpolarization of pyramidal neurons in layers IV and V of the cerebral
cortex (Buzsáki, Anastassiou, & Koch, 2012; Katznelson, 1981).
56
Mechanism Description of Activity Effect on Extracellular Matrix/ EEG Recording
Fast acting potentials Synchronous activation of a neuronal network creating rapid (< 2ms) large amplitude electric fields.
Contributes to high frequency oscillations measured by EEG.
Calcium (Ca2+) spikes Flow of calcium into the cell mediated by opening of voltage gated channels. Can be independent of synaptic activity.
Long lasting (up to 100ms) effects similar an excitatory post-synaptic potential.
Intrinsic current resonance
Voltage dependent membrane responses of neurons activated in synchrony.
Effects neuronal oscillation frequency.
After-hyperpolarization potentials (AHP)
The influx of other ions into the extracellular matrix by ligand gated channels. Usually activated after neuronal hyperpolarization. Activities can be widespread or tightly localized.
Effect on the electrical field potentials if temporally synchronous. AHPs are hypothesized to have a role in the Berietschaftpotential.
Gap junctions Electrical communication between neural synapses.
Has an effect on oscillations if synchronous.
Epiphasic effects Electrical gradients generated in extracellular matrix by the activation of a nearby network of neurons.
Hypothesized to have an effect on the activation or likelihood of activation of nearby neurons (akin to external transcranial direct stimulation).
Table&1.& Descriptions&of&additional&cellular&processes&contributing&to&local&field&potentials&
57
! Apparatus!!
In Experiment A2, when used, EEG was recorded continuously from 64 pre-amplified Ag-AgCl
(ActiveTwo, Biosemi, Amsterdam, Netherlands) electrodes embedded in an elastic cap mapped to
the extended 10-20 system (see Koessler et al., 2009 for comparisons with newer systems). Two
electrodes, a Common Mode Sense (CMS) active electrode and a Driven Right Leg (DRL) passive
electrode served as a feedback loop driving the average potential of the measured signal to levels
as close as possible to the analog-to-digital converter reference voltage. EEG signals were digitized
at a sampling rate of 2048Hz (DC 268, 3dB/octave).
! Visual!Evoked!Potentials!Features,!Calculations,!and!Analysis!
With regard to EEG, the main dependent variables used in this thesis were the magnitudes and
latencies of the different components of the visual evoked potential (VEP). VEPs are event related
potentials (ERPs) locked to the onset or offset of a visual stimulus and are a reliable way of
assessing the efficacy of retinal and visual cortical pathways (Celesia, 1984; Courchesne, Hillyard,
& Galambos, 1975; Phurailatpam, 2014). Furthermore, and important considering the goals of this
thesis it has been shown that VEPs increase with baseline neural activity related to visual
processing (Chawla, Lumer, & Friston, 2000; Chawla, Rees, & Friston, 1999), and thus are an
adequate way to examine cortical areas involved in visual-sensorimotor processes (e.g., Lebar,
Bernier, Guillaume, Mouchnino, & Blouin, 2015).
For Experiment A2, VEPs were obtained by averaging all epochs time-locked (~200 ms to
~400 ms) to the onset of the visual stimulus (each visual stimulus is described in the methods
associated with the experiment), with an average amplitude of -100 ms to – 5 ms serving as the
pre-stimulus baseline (see Figure 13 for graphic ERP averaging procedure). Raw recordings were
re-referenced to the average of the mastoid electrodes and data was visually inspected to remove
segments with artifacts.
58
&
&&To!calculate! an!ERP,! continuous!EEG!data!were! preprocessed!and!artifacts! due! to!external! noise! and!interference!were!removed.!After,!the!data!were!segmented!into!defined!time!windows!centered!around!the!event!of!interest.!Finally,!by!averaging!all!trials!for!certain!conditions,!an!ERP!waveform!could!be!observed!for!each!electrode!of!interest!(see!below).!!!
In general, VEPs were characterized by the following deflections occurring at different
points in time with respect to the visual stimulus: a negative deflection occurring at 80- 100 ms
after; followed by a positive deflection occurring at 100- 120 ms; followed by yet another negative
deflection occurring at 140-160 ms; and finally, a second positive deflection occurring 180-220
ms (see Figure 14). Although, there has been evidence to suggest different components of the
VEP are representative of activities localized in different areas of the cortex (Brodeur et al., 2008;
Clark & Hillyard, 1996), the respective contribution of each above-noted component to
sensorimotor processes remains largely unknown. For the purposes of this thesis, the peak to peak
amplitude (delta) of the earlier identifiable components (P100 and N150) were investigated for
each of the electrodes of interest (see Figure 15).
59
&Characteristic!peaks!of!a!visual!evoked!potential.!A!negative!deflection!N1!occurs!at!around!80!ms!after!the!visual!stimulation.!The!P1!occurs!at!about!100!ms!after!that,!followed!by!the!N2!at!around!140!ms.!Lastly!there!is!a!P2!that!occurs!about!200!ms!after!the!initial!stimulation.!For!the!EEG!analyses!in!the!present!thesis,!the!peak!to!peak!amplitude!between!the!P100!and!N150!was!used!for!analysis!(shown!in!red),!as!this!was!the!most!recognizable!peaks!across!participants.!!
&Activities!of!associated!with!visual!–sensorimotor!processing!were!analyzed!using!the!electrodes!overlapping!the! occipital! cortex! and! the! occipitoGparietal! junction! of! the! left! hemisphere! (electrodes! PO7,! PO3,! O1).! These!electrodes!were!selected!based!on!findings!in!previous!studies!(Lebar!et!al.,!2015).!
!
60
! Current!Source!Density:!The!Surface!Laplacian!
To enhance the spatial and temporal resolution of the EEG data, current source density (CSD)
calculations were employed (Hjorth, 1975; Law, Rohrbaugh, Adams, & Eckardt, 1993; Perrin,
Bertrand, & Pernier, 1987; Perrin, Pernier, Bertrand, & Echallier, 1989). Signals were interpolated
using the spherical spline interpolation (order of splines: 3; maximal degree of Legendre
polynomials: 10; approximation parameter λ: 1.0e – 005) procedure (Perrin et al., 1987), as
implemented Brain Vision Analyzer 2.0 (Brain Product GMBH; Gilching, Germany). Current
source density calculations are independent of the reference electrode and are lesser affected by
far-field generators, monopolar recordings, and other artifacts (see Kayser & Tenke, 2015; Vidal
et al., 2015, for reviews). Analyses of peak amplitudes and latencies of VEPs were done using the
data obtained from CSD computations.When comparing CSD amplitudes between participants and
conditions, values were expressed as relative to a control condition using a log2 transformation.
! Source!Localization!Procedures!and!Analysis!!
In addition to analysis of CSDs, source analyses were also conducted to determine cortical
generators of the observed activities noted at the level of the scalp. Overall, source analyses attempt
to rectify what is known as the inverse problem of EEG. That is, for any recorded scalp potential,
many different combinations of generator activities are possible (Michel et al., 2004). By using
volume conduction models, and mathematical assumptions based on Maxwell’s laws (Tenke &
Kayser, 2012a; van den Broek, Reinders, Donderwinkel, & Peters, 1998), it is possible to compute
inverse solutions and estimate activated areas associated with recorded scalp activity.
For this thesis, source localization for EEG data was conducted using the minimum-norm
techniques implemented in the Brainstorm software package (Tadel, Baillet, Mosher, Pantazis, &
Leahy, 2011). Data from all sensors were imported, processed, and averaged for each participant.
Forward models for source analysis were computed using the boundary element method (BEM,
Gramfort, Papadopoulo, Olivi, & Clerc, 2010) on the anatomical MRI brain template from the
Montreal Neurological Institute (MNI Colin 27). This method uses triangulations of interfaces
between cerebral compartments of equal isotropic conductivities as a geometrical model. The main
benefit of head models constructed with the boundary element methods is the substantial buffering
61
of external noise (Michel et al., 2004). Cortical sources for the VEPs were examined at the
latencies wherein the peaks of the CSDs were identified. Mean source activity were then submitted
to statistical contrasts (see methods of Experiment A1).
! Electrooculography!(EOG)!
! EOG!for!EEG!
EOG measures the changes in the standing corneo-fundal potential that primarily emerges from
the retinal pigment epithelium. As a consequence of the polarity between the front and back of
the eye, the potential recorded by electrodes placed on the skin changes when the eye moves
from side to side (Brown et al., 2006).
In Experiment A2, EOG electrodes integrated with the EEG system were used to monitor eye
movements and positions of gaze. The primary purpose of monitoring eye movements during EEG
collection was to subtract artifacts that could have occurred due to oculomotor related noise.
Because eye movements can be both spontaneous and reflexive (Bahill, Clark, & Stark, 1975;
Becker, 1991), the amplitude of artifacts created by different types of eye movements can vary.
Artifacts caused by eye movements and blinks were removed from recordings using independent
components analysis (ICA) s as implemented in Brain Vision Analyzer 2.0.
! EOG!for!Eye!Tracking!and!Gaze!Direction!Measurements!
For experiments not involving EEG, EOG was recorded from using an electroculogram
(Coulbourn Instruments, Lablinc Inc., Lehigh Valley, PA) sampling at 1000Hz. For Study A, eye
positions were monitored to ensure participants complied with instructions and task demands (i.e.,
to ensure participants were maintained fixation).
62
!!Flexibility!in!the!Encoding!of!Reaching!Movements!to!
Somatosensory!Targets:!Behavioural!and!Electrophysiological!Experiments!!
A revised version of Study A has been submitted to: PloS ONE
63
! Study!A!!
! Abstract!Prior to goal-directed actions, somatosensory target positions can be localized using either an
exteroceptive or an interoceptive body representation. The goal of the present study was to
investigate if the body representation selected to plan reaches to somatosensory targets is
influenced by the sensory modality of the cue indicating the target’s location. In the first
experiment, participants reached to somatosensory targets prompted by either an auditory or a
vibrotactile cue. As a baseline condition, participants also performed reaches to a visual target
prompted by an auditory-cue. Gaze-dependent reaching errors were measured to determine the
contribution of the exteroceptive representation to motor planning processes. The results showed
that reaches to both auditory-cued somatosensory and auditory-cued visual targets exhibited larger
gaze-dependent reaching errors than reaches to vibrotactile-cued somatosensory targets. Thus, an
exteroceptive body representation was likely used to plan reaches to auditory-cued somatosensory
targets. In the second experiment, the effect of using an exteroceptive body representation to
encode somatosensory target positions on pre-movement sensorimotor transformation processes
was examined. The cortical response to a task-irrelevant visual flash was measured as participants
planned movements to either auditory-cued somatosensory or auditory-cued visual targets. Larger
visual-evoked potentials were found when participants planned movements towards
somatosensory vs. visual targets. Furthermore, source analyses revealed that these activities were
localized to the left occipital and posterior parietal areas. These results indicated that visual and
visuomotor processing areas were more engaged when using an exteroceptive body representation
to plan movements to a somatosensory target, than when planning movements to an external visual
target.
Keywords: Body representation, somatosensory targets, proprioception, visually- evoked
potential
64
! Experiment!A1!
! Introduction!!
In a game of “Simon Says”, spoken instructions such as “touch your elbow” can prompt
movements towards a specified body location. Similarly, the feeling of a mosquito landing on the
elbow could also prompt movements to this same position. Although the goal of both actions is to
reach to a common body location (i.e., hereafter referred to as a somatosensory target), the manner
in which this position is identified could influence the processes used to determine the target’s
coordinates. The purpose of the present study was to examine if the modality of the stimulus
indicating a somatosensory target’s position influences the representation used to plan movements.
Evidence that movements to somatosensory targets could be planned using multiple
representations is drawn from studies of autotopagnosia, a rare nervous system disorder
characterized by the inability to localize and orient one’s own body parts (Buxbaum & Coslett,
2001; Olsen & Ruby, 1941; Pick, 1922; Sirigu, Grafman, Bressler, & Sunderland, 1991). In a case
study by Sirigu et al. (1991), a patient was tested on her ability to point to her own body positions
in response to different instructional cues. The authors found that the patient was inaccurate when
pointing to body parts in response to verbal instructions but was able to accurately point to the felt
location of small objects placed on these same body positions (see also Ogden, 1985). Based on
these findings, the authors hypothesized that an exteroceptive, visually-based, body representation
was used to plan movements to somatosensory targets in response to verbal cues, whereas an
interoceptive, somatosensory-based, body representation was used to derive somatosensory target
positions in response to tactile stimulation (see de Vignemont, 2010 for a review of different body
representation taxonomies; and Berlucchi & Aglioti, 2010 and Paillard, 1991, 1980 for additional
theoretical discussions).
It is difficult to determine if the context-dependent mapping of somatosensory targets in
patients with autotopagnosia emerges as an adaptation to their neurological impairment or whether
such processes could also occur in healthy individuals. Furthermore, the verbal instructions used
in these studies were much more complex than the tactile cues, such that the additional brain
processes to encode words may offer less than optimal support for the mechanistic explanation. In
the present study, the contribution of the exteroceptive and interoceptive body representations to
65
movement planning was examined in healthy individuals by evaluating reaching movements
towards somatosensory targets cued by simple auditory and vibrotactile stimuli.
To examine the sensorimotor transformations used to prepare movements to
somatosensory targets, Experiment A1 involved measuring gaze-dependent reaching errors (Bock,
1986; Henriques et al., 1998; Mueller & Fiehler, 2014a, 2016; Pouget et al., 2002; Schütz,
Henriques, & Fiehler, 2013). Large gaze-dependent reaching errors are indicative of a reliance on
visual information for movement planning processes (i.e., gaze-dependent coding, see Blangero et
al., 2005; Henriques et al., 1998; Pouget et al., 2002). Conversely, movements are termed gaze-
independent if the peripheral gaze-shift does not produce significant increases in error (Mueller &
Fiehler, 2014a). Studies examining movements to visual targets have shown that gaze-dependent
errors increase in magnitude with peripheral fixation eccentricity (Bock, 1986; Henriques et al.,
1998). In contrast, previous studies examining movements to somatosensory targets have shown
evidence for both gaze-dependent and gaze-independent coding (Jones & Henriques, 2010;
Mueller & Fiehler, 2014a, 2016).
Although there is evidence for flexibility in the encoding of somatosensory targets (Mueller
& Fiehler, 2014a, 2016), it is not clear if the cue used to indicate somatosensory target’s location
influences the body representation and sensorimotor transformations used for motor planning. The
present study investigated the influence of the cue modality by directly comparing the effect of a
gaze shift on reaching movements towards one of three possible somatosensory targets (i.e., the
index, ring, and middle fingers of the non-reaching hand). Somatosensory targets were identified
either by an exogenous auditory cue (i.e., intensities of sound indicating which finger to reach to)
or a direct tactile stimulation. To provide a reference of gaze-dependent reaching errors,
participants also performed movements to visual targets cued by an auditory stimulus.
! Methods!!
! Participants!!
Ten participants (8 females; mean age: 24.4 ± 2.6 years; range: 21-28) volunteered to take part in
the experiment. All participants were right-handed with normal or corrected-to-normal vision. The
experiment took 1.5 hours to complete and participants were compensated with 10€. Informed
66
consent was obtained prior to the beginning of the experiment and the local research ethics
committee approved all procedures.
! Apparatus!
A drawn representation of the experimental setup is presented in Figure 16. The experiment took
place in a dark room where participants were seated comfortably in front of a custom-built aiming
apparatus. Once seated, participants placed their forehead on a stabilizing headrest located above
the aiming apparatus. From this position, participants viewed the aiming apparatus through a semi-
reflective glass mounted 30 cm above the aiming surface. Positioned 30 cm above the semi-
reflective glass was a liquid crystal display monitor (HP Compaq LA1956x, Palo Alto, CA) used
to project images onto the glass.
67
&Panel!A)!A!drawn!representation!of!the!apparatus!used!in!Experiment!1.!Participants!sat!facing!an!immersive!
display!comprised!of!a!computer!monitor,!a!semiGreflective!glass!surface,!and!a!custom!aiming!apparatus!(not!to!scale).!Participants!made!movements!from!an!unseen!home!position!(microswitch)!to!either!visual!targets!projected!onto!the!surface! of! the! aiming! console! or! the! perceived! position! of! their! fingers.! In! the! somatosensoryGtarget! conditions,!participants!positioned!their!target!fingers!beneath!a!plastic!case!and!performed!movements!to!the!perceived!position!of!their!fingers!as!if!projected!onto!the!case’s!surface.!Note!that!the!visual!items!projected!on!the!semiGreflective!glass!were!perceived!by!the!participants!as!appearing!on!the!table!surface.!Panel!B)!A!representation!(not!to!scale)!of!the!aiming!surface!in!the!visual!target!condition.!Piezoelectric!buzzers!were!positioned!to!the!left!of!the!aiming!surface,!and!provided!the!imperative!signals!in!the!auditoryGcued!aiming!conditions! !
68
The aiming apparatus was also equipped with a microswitch positioned 32 cm directly in
front of the participant. The switch was used as the starting position for the reaching finger. Three
piezoelectric buzzers (frequency: 4,000 Hz; KPEG158-P5H, Kingsgate Electric Corp, Taipei-
Hsien, Taiwan) were positioned 25 cm to the left of the microswitch. The input to the buzzers was
adjusted using potentiometers to produce a loud (i.e., 70 dB), a medium (i.e., 58 dB) or a soft (i.e.,
48 dB) sound, which signaled the location of targets in the auditory-cued conditions (see below).
Three circular indentations were made between the participant and the microswitch, at 16.5
cm, 15 cm, and 15.5 cm from the microswitch (see Figure 16) and -3.5 deg, 0 deg, and 4 deg
relative to the participant’s cyclopean eye, respectively. These indentations served as placeholders
for the somatosensory targets in the somatosensory-target conditions (see below). A black plastic
case (40 cm x 9 cm x 3 cm) with an opening towards the participants was used to cover the three
indentations and prevent the reaching finger from making contact with the target fingers (i.e.,
precluding any tactile feedback about endpoint accuracy). Within each indentation the stimulation
surface of a solenoid vibrator (type 347-652, Johnson Electric, Shatin, Hong Kong) was affixed.
These solenoid vibrators delivered brief vibrations (50 ms at 80Hz, with a 2 mm amplitude) that
signaled the position of the target finger in the vibrotactile-cued somatosensory conditions (see
below).
Fixation positions and visual targets (used in the visual target condition) were projected
onto the aiming surface using a custom MATLAB program (The Mathworks Inc., Natick, MA)
and Psychtoolbox-3 (Brainard, 1997). Three fixation positions were projected as blue circles,
measuring 0.45 deg of visual angle in diameter. They were located 4.5 cm distal to the microswitch,
at 0 deg, 10 deg to the right, and 10 deg to the left of the participant’s cyclopean eye. In the visual
target condition, targets of 1 deg of visual angle were projected onto the same spatial locations as
the indentations utilized for the somatosensory targets (see above).
A white light emitting diode (LED, 4 mm diameter) attached to the participant’s right index
finger was visible through the semi-reflective glass and provided visual feedback of the initial
fingertip position in the visual-target condition (see below). Also, a black plastic shield was placed
beneath the glass surface to occlude vision of the limb during the aiming movement. The position
69
of the occluding surface was adjusted for each participant such that they could see the illuminated
LED on their finger when they were at the home position.
Eye positions were monitored with electro-oculography (EOG) (Coulbourn Instruments,
Lablinc Inc., Lehigh Valley, PA), sampled at 1000Hz. The position of the index finger was tracked
using an electromagnetic sensor secured to the tip of the right index finger (Flock of Birds,
Ascension Technology Corp., Burlington, VT), sampled at 100 Hz.
Overall, the participants’ task was to maintain gaze fixation while performing accurate
reaching movements with the index finger of the right hand to targets located between the home
position and their body (see Figure 16). Participants reached to somatosensory targets in response
to either an auditory (AUD-SOMA) or a vibrotactile (TACT-SOMA) cue. To obtain reference
values for gaze-dependent reaching errors, participants also performed reaches to visual targets in
response to an auditory cue (AUD-VIS). The presentation order of these conditions was
counterbalanced across participants.
! CueGtarget!conditions!!
In the AUD-VIS condition, participants positioned their left hand in a comfortable position, either
on their lap or on the table surface to the left of the aiming apparatus. The trial sequence was
initiated once the participant depressed the home position microswitch with their right index
finger. This action triggered the illumination of the finger LED and the presentation of the current
fixation point (i.e., -10 deg, 0 deg, or +10 deg). Two seconds later, participants were presented
with a soft, medium, or loud sound. For half of the participants, the loud and soft sounds
corresponded to right and left targets, respectively. This correspondence was reversed for the other
half of the participants. At movement onset, the finger LED was extinguished, therefore
participants only saw the position of their reaching finger relative to the visual target prior to
reaching.
The AUD-SOMA condition was similar to the AUD-VIS condition, but instead of visual
targets, participants placed the three middle fingers of their left hand (i.e., the ring, middle, and
index) into the three circular indentations beneath the plastic casing to serve as the target positions.
Participants made movements to target positions in response to the same auditory cues as in the
70
AUD-VIS condition. Participants were instructed to “aim to the perceived position of their
fingernail as if it was projected onto the surface of the plastic case”. In contrast to the AUD-VIS
condition, participants did not receive visual information about the position of their reaching hand
prior to movement initiation.
The TACT-SOMA condition was similar to the AUD-SOMA condition, except that a
vibrotactile stimulus indicated the location of the target finger (i.e., index, middle, or ring finger).
Similar to the auditory-cue, the tactile stimulation of the target finger occurred 2 s after the
participant depressed the microswitch (i.e., the home position).
! Familiarization!trials!
Before each experimental condition, participants performed 2 sets of familiarization trials. For the
AUD-VIS and AUD-SOMA conditions, the first set of familiarization trials consisted of 3 blocks
of 3 reaching trials, wherein the same auditory-cue was presented in each block (e.g., 3 trials with
the “loud” sound). Subsequently, participants performed 1 block of 15 trials wherein auditory cues
were presented in random order. After each trial in this familiarization block, participants were
asked to report which auditory cue they had heard (e.g., “loud”, “medium”, or “soft”) and they
were very accurate at distinguishing between the different levels of sound (mean accuracy: 97%;
SD = 0.5%). Participants were also presented with 2 sets of familiarization trials in the TACT-
SOMA condition. Auditory cues were replaced by brief tactile stimulation applied to the reaching
finger. The participants’ task in second set of familiarization trials was to report which target finger
had been stimulated and they were all perfectly accurate in this task.
! Experimental!trials!
In all experimental conditions, participants performed 10 trials in each fixation-target combination
(i.e., 3 fixation [-10 deg, 0 deg, +10 deg] by 3 possible targets [left, middle, right], yielding 90
trials per condition) for each of the 3 cue-target conditions (i.e., AUD-VIS, AUD-SOMA, TACT-
SOMA) yielding a total of 270 total experimental trials.
! Data!analysis!and!reduction!!
Only data from movements to the centre target were analyzed because gaze-shifts were equidistant
with respect to this target (i.e., 10 deg on either side) and a consistent sound level (i.e., medium
71
sound) was used to signal the target position across conditions. All finger and eye movement
recordings were exported to custom software (Analyse, Marcel Kazsap, QC, Canada) and
processed offline. Trials wherein the participants failed to maintain their fixation position prior to
or during the movement and trials where reaction times or movement times were higher or lower
than 2.5 times the within condition standard deviation were excluded from the analyses. Overall,
less than 9% of trials (to the centre target) were excluded (an average of 2.6 ± 1.6 trials per
participant). Statistical analyses were performed with the Statistical Package for Social Scientists
(version 21: SPSS Inc., Chicago, IL). Post-hoc-comparisons were performed using R (version
3.02, R Development Core Team).
Normalized reaction times (RTs), movement times (MTs), and directional reaching errors
were computed to investigate the effect of peripheral fixation on movement performance. RT was
defined as the time that elapsed between the go signal (i.e., either the sound or the finger vibration)
and the release of the home position microswitch. MT was defined as the time that elapsed between
the release of the home position microswitch and movement end, which was defined as the first
sample at which the limb fell below 30 mm/s. Directional reaching error was defined as the angular
difference (in degrees) between the home-target position vector and the home-movement end
position vector. Directional errors were computed for each fixation condition (i.e., left, middle,
and right). The within-participant mean and standard deviation of the directional errors in the
centre fixation position were used to calculate a population z-score value for each trial in the left
and right fixation condition. Negative and positive values indicate movements biased to the left
and right of the centre position, respectively. The resulting normalization thus took into
consideration the variability in participant’s performance in the centre fixation condition. This
normalization procedure was also completed for the reaction times and movement times. Larger
positive z-scores represent longer times whereas larger negative z-scores are indicative of shorter
times, relative to the centre fixation condition.
All z-score values were submitted to a 2-fixation direction (Left, Right) x 3 cue-target
condition (AUD-VIS, AUD-SOMA, TACT-SOMA) repeated-measures ANOVA. Alpha level
was set at 0.05 for all statistical contrasts, and effect sizes were reported using partial eta squared.
Post-hoc tests for any significant interactions were completed using pairwise t-tests with the
Bonferonni correction. Additionally, for horizontal reaching errors, one-sample t-tests were used
72
to examine if peripheral fixation elicited a significant bias (e.g., different than 0 deg) for each
fixation direction, effect sizes are reported using Cohen’s dz.
! Results!!
! Directional!reaching!errors!
The ANOVA performed on the directional reaching errors revealed a significant main effect of
fixation direction (F(1,9) = 64.7, p < .001 ηP2 = 0.89) and a significant interaction between fixation
direction and cue-target condition (F(2,18) = 5.1, p < 0.05, ηP2 = 0.36; see Fig 17 A). Critically,
decomposing the interaction (Bonferonni corrected alpha of p = .006) revealed that the differences
in gaze-dependent errors for the left and right fixation directions were higher in both the AUD-
SOMA (M = 1.32, SD = 0.44), and AUD-VIS target conditions (M = 1.31, SD = 0.63), compared
to the TACT-SOMA condition (M = 0.69, SD = 0.69). Furthermore, one sample t-tests revealed
that the magnitude of the directional reaching errors in both left and right fixation conditions in
the AUD-SOMA (left: t(9) = 4.2, p < .001 dz = 1.3; right: t(9) = -4.1, p < .001, dz =1.3 ) and AUD-
VIS conditions (left: t(9) = 2.7, p <. 05, dz = 0.9; right: t(9) = -4.0, p <. 01, dz = 1.3) were
significantly greater than zero. In contrast, neither left or right fixation directions in the TACT-
SOMA condition were significantly greater than zero (left: t(9) = 1.8, p = 0.10, dz = 0.6 ; right:
t(9) = -1.5, p = 0.17, dz = 0.5). These results reveal gaze-dependent coding for movements to
auditory-cued visual and auditory-cued somatosensory targets (see Fig 17 B and Fig 17 C).
! Normalized!RT!and!MT!!
Analysis of RT yielded a significant main effect of cue-target condition (F(2,18) = 5.4, p < 0.01,
ηP2 = 0.38). Post-hoc tests (Bonferonni corrected alpha of p = 0.016) revealed that, in the AUD-
SOMA condition, participants exhibited significantly longer RTs (M = 0.44, SD = 0.43) compared
to both the AUD-VIS (M = -0.02, SD = 0.58), and the TACT-SOMA conditions (M = 0.09, SD =
0.37: see Fig 17 B). In contrast, the analyses of MTs yielded no significant main effects or
interactions (Fs < 1.96, ps > 0.16).
73
&Panel! A)! Averaged! normalized! reach! trajectories! for! all! participants! in! each! cueGtarget! condition! (error!patches!represent!the!betweenGparticipant!standard!error!of!the!mean).!Panel!B)!Average!normalized!directional!error!for!each!cueGtarget!condition!(error!bars!represent!the!between!participant!standard!deviation).!Participant’s!reaching!errors!were!significantly!more!influenced!by!gaze!fixation!position!(i.e.,!Right!Fix!vs.!Left!Fix)!in!the!auditoryGcued!target!conditions!(GVIS!and!–SOMA)!than!in!the!vibrotactileGcued!target!conditions.!Panel!C)!Mean!normalized!reaction!times!for!participants!in!each!experimental!condition!(error!bars!represent!the!betweenGsubjects!standard!deviation).!Reaction!times!of!gazeGshifted!movements!were!significantly! longer!when!participants!aimed!to!auditoryGcued!somatosensory!targets!compared!to!both!auditoryGcued!visual!targets!and!vibrotactileGcued!somatosensory!targets.!
74
! Discussion!
In Experiment A1, participants performed reaching movements to one of three somatosensory
target locations cued by either an auditory or vibrotactile stimulus. To provide a comparison for
gaze-dependent reaching errors participants also performed movements to visual targets cued by
an auditory stimulus. Overall, participants exhibited larger gaze-dependent endpoint errors when
reaching to auditory-cued somatosensory targets compared to vibrotactile-cued somatosensory
targets. Also, gaze-dependent errors for reaches to auditory-cued somatosensory targets were no
different than those observed for reaches to auditory-cued visual targets. However, normalized
reaction times were longer for reaches to auditory-cued somatosensory targets compared to
auditory-cued visual targets. Taken together, these findings suggest that more complex
sensorimotor transformation processes were required when using an exteroceptive representation
to plan movements to somatosensory targets than when planning movements to auditory-cued
visual and vibrotactile-cued somatosensory targets.
Participants exhibited larger gaze-dependent reaching errors in both auditory-cued target
conditions compared to the vibrotactile-cued somatosensory-target condition. These results are
congruent with earlier studies that found movements to somatosensory targets were planned in
gaze-independent coordinates if the eyes and target positions remained stable prior to movement
onset (Mueller & Fiehler, 2014a, 2016). According to the classical model of somatosensory
processing, cutaneous and proprioceptive inputs project to different areas (cutaneous: areas 3b and
1; proprioceptive: areas 3a and 2) of the post-central cortex (Friedman, Murray, O’Neill, &
Mishkin, 1986; Kaas et al., 1983). These areas are reciprocally connected and recent investigations
have shown that these areas respond to both cutaneous and proprioceptive stimuli (see Kim et al.,
2015). Given the close link between tactile stimulation and non-visual body representations (see
Serino & Haggard, 2010), the integration of spatially congruent somatosensory inputs in the
TACT-SOMA condition may have favoured the use of an interoceptive body representation for
encoding target position and planning the reaching vector.
In contrast, when the site of the tactile stimulation does not correspond to the location of
the target, planning processes could rely on an exteroceptive, more cognitive type of
representation. Consistent with this notion and similar to previous studies (Blangero et al., 2005;
75
Jones, Fiehler, & Henriques, 2012; Pouget et al., 2002), gaze-dependent coding for somatosensory
targets was found in the AUD-SOMA condition. The absence of tactile stimulation in the AUD-
SOMA condition might have precluded direct encoding of the finger position and thus, access to
the gaze-independent coordinates when planning the reaching movement. These findings are
consistent with previous studies proposing that if an exteroceptive representation is used to define
a body position, the location of this position is more biased by environmental visual information
(Schwoebel, Coslett, & Buxbaum, 2001).
Although the analyses did not yield significant differences in gaze-dependent errors
between the AUD-VIS and AUD-SOMA conditions, the longer reaction times in AUD-SOMA
compared to the AUD-VIS conditions suggest that more complex sensorimotor transformation
processes are required when planning movements to somatosensory targets using the exteroceptive
body representation. This result is in line with previous studies showing that distinct sensorimotor
transformation processes are used for planning movements to somatosensory targets (Bernier,
Burle, Hasbroucq, & Blouin, 2009; Bernier & Grafton, 2010; Blouin, Saradjian, Lebar, Guillaume,
& Mouchnino, 2014; Sarlegna & Sainburg, 2007). For example, Bernier et al. (Bernier et al., 2009)
found that reach-related activities occurred earlier and were more pronounced in posterior parietal
cortex (PPC), when participants planned movements to visual targets compared to when they
planned movements to somatosensory targets. The authors also noted that movements to
somatosensory targets produced earlier and greater activations in premotor cortex compared to
movements to visual targets.
In the present study, because movements to auditory-cued somatosensory targets showed
gaze-dependent reaching errors, it is possible that the initial movement vector was computed in
visual coordinates. This would require a sensorimotor reference frame transformation wherein
somatosensory information about the target and effector are used to compute visual estimates of
their positions prior to the start of the movement (see Neggers & Bekkering, 2001). This remapping
of the somatosensory target positions would likely require greater visual processing in areas
involved in visuomotor transformations and gaze-dependent movement coding (Batista &
Newsome, 2000; Bernier & Grafton, 2010; Darling, Seitz, Peltier, Tellmann, & Butler, 2007;
Medendorp, Beurze, Van Pelt, & Van Der Werf, 2008) compared to conditions where this
76
remapping is not required (e.g., reaches to visual targets). A second experiment was designed to
test this prediction.
! Experiment!A2!
! Introduction!
In Experiment A1, movement preparation times were longer when participants reached toward
auditory-cued somatosensory targets compared to when they reached toward auditory-cued visual
targets. This result was attributed to the additional transformation processes required to shift from
a gaze-independent (interoceptive) representation to a gaze-dependent (exteroceptive)
representation. The purpose of Experiment 2 was to examine the neural mechanisms associated
with the gaze-dependent encoding of somatosensory targets.
Although behavioural evidence for the remapping of somatosensory targets onto a gaze-
dependent reference frame has been reported (e.g., Mueller & Fiehler, 2014a, 2016; Pouget et al.,
2002), the neural processes underlying these sensorimotor transformations remain largely
unknown. Previous studies examining motor planning have implicated the occipito-parietal
network in the gaze-dependent coding of effector and target positions (Batista & Newsome, 2000;
Bernier & Grafton, 2010; Medendorp et al., 2008). Moreover, studies on autotopagnosia have
shown that the inability to use the exteroceptive body representation is linked to damage to the left
posterior parietal and parietal-occipital cortices (Buxbaum & Coslett, 2001; Corradi-Dell’Acqua
& Rumiati, 2007; de Vignemont, 2010; Ogden, 1985; Olsen & Ruby, 1941; Sirigu, Grafman,
Bressler, & Sunderland, 1991). Given these findings, it was hypothesized that there would be
greater activation in parieto-occipital networks when preparing movements to somatosensory
targets encoded in gaze-dependent coordinates as compared to external visual targets. Keep in
mind that for external visual targets, no remapping is needed as both hand and target positions are
already encoded in an visual coordinate system (Buneo & Andersen, 2006; Cohen & Andersen,
2002; Reichenbach et al., 2009).
To test these predictions, the cortical responses to a task-irrelevant visual stimulus (i.e., the
visual-evoked potential or VEP) was measured as participants planned movements to both
auditory-cued visual and auditory-cued somatosensory targets. Because baseline neural activity in
77
the extrastriate cortex is a marker of visual processing and the amplitude of visually-evoked
responses increases along with baseline activity (Chawla et al., 2000, 1999), comparing changes
in VEP amplitudes should provide a good proxy for the engagement of visual and visuomotor
cortical networks.
! Methods!
! Participants!!
Ten participants (5 female, mean age: 26 ± 4.7; range 20-35), who did not participate in
Experiment 1, were recruited for Experiment 2. All participants self-reported being right-handed
and had normal or corrected-to-normal vision. Informed consent was obtained prior to data
collection and a local ethics committee at Aix-Marseille University approved all procedures. In
total, the experiment took place over 2 sessions lasting 1.5 to 2 hours each. There was at least 1
but no more than 10 days between sessions (average = 4 ± 3.5 days).
! Apparatus!
To accommodate the use of electroencephalography (EEG) and reduce electrical noise, all visual
stimuli were generated using LEDs positioned on the aiming surface (see Fig 18) instead of the
LCD monitor used in Experiment A1. Two additional microswitches and 3 yellow LEDs were
added to the aiming surface. Each additional microswitch served as a different possible starting
position and each LED, placed 0.5 cm distal and 0.5 cm to the left of each microswitch, served as
a possible fixation location (see procedures below). Thus, when participants fixated on any of the
LEDs and placed their finger on the corresponding microswitch, the finger was positioned in the
same relative retinal location, in the lower right visual field.
Three orange LEDs were placed at the same positions as the circular indentations that
served as the positions for somatosensory targets in Experiment A1. These LEDs were used as
visual targets in the AUD-VIS condition. To prevent participants from receiving tactile feedback
when reaching to visual targets, a thin piece (39 cm x 9 cm x ~0.3 cm) of transparent laminated
Bristol board was placed over the LEDs. In addition, the participant’s right index finger was
equipped with two LEDs: the same white LED that was used in Experiment 1 to indicate the initial
78
position of the pointing finger; and a green LED (width 5mm, beam angle 40°, luminance 60000
mcd, 22.7 lm) attached to the fingernail generated the stimuli for the VEPs.
79
&Panel!A)&A!drawn!representation!of!the!experimental!apparatus!for!Experiment!2!(not!to!scale).!Participants!fixated!on!one!of!three!fixation!locations!and!began!their!movements!from!the!associated!home!position.!In!the!auditoryGcued! somatosensoryGtarget! condition! (represented! above),! participants! performed!movements! to! one! of! the! three!middle!fingers!of!their!nonGreaching!limb.!Panel!B)!A!representation!of!the!aiming!surface!in!the!visual!target!condition.!Participants!fixated!on!of!three!target!positions!and!placed!their!fingers!on!the!corresponding!microswitch!to!begin!each!trials.!Piezoelectric!buzzers!to!left!again!provided!the!imperative!signals!to!initiate!movement!in!both!the!AUDGVIS!and!AUDGSOMA!conditions.!
80
! Trial!procedures!
In contrast to Experiment A1, no gaze-shift stimulus was presented prior to the reaching
movement. As the purpose of Experiment A2 was to examine the differences cortical network
activation underlying reaches to auditory-cued somatosensory and visual targets, it was imperative
that visual stimulus was presented in the same retinal location prior to the start of the reaching
movement.
Each trial began when one of the three possible fixation positions was illuminated.
Participants fixated on the yellow LED and placed their right index finger onto the corresponding
microswitch. As in Experiment A1, a soft, medium, or loud sound was presented 2 seconds after
the participant placed the finger on the microswitch. That auditory cue once again either indicated
the spatial location of the somatosensory or visual target while also serving as the imperative signal
to begin the pointing movement. Then, 100 ms after the presentation of the auditory cue, during
the participant’s reaction time (i.e., movement planning), the green LED located on the pointing
finger generated a 50 ms flash. This presentation time was chosen based on previous studies
demonstrating significant modulations in evoked responses (Blouin et al., 2014) and reach-related
activity (Bernier et al., 2009) for both movements to somatosensory and visual targets (i.e.,
between 100-150 ms after the go signal). Participants were instructed to reach to the target as
precisely as possible and to stay at the target location until the next fixation point was illuminated.
Only one cue-target condition was used per experimental session (AUD-VIS or AUD-
SOMA), and the order of target modality presentation was counterbalanced across participants.
As in Experiment A1, the sound-target mapping was counterbalanced between participants and
remained the same for both experimental sessions. Movements were also conducted in darkness
and thus, participants had no visual feedback of the reaching limb during the movements.
!CueGtarget!conditions!
In the AUD-VIS condition, when participants placed their finger on the home microswitch, the
finger and all target LEDs were illuminated. When participants released the microswitch to begin
their reaching movement the finger LED was extinguished, but all target LEDs remained
illuminated throughout the trajectory.
81
In the AUD-SOMA condition, participants performed reaches to the fingernail of one of
the three middle fingers on their left hand (see Section 2.5). As in the AUD-SOMA condition of
Experiment A1, participants did not receive visual feedback of their reaching finger during
movement planning. Because differences in endpoint biases were not informative in Experiment
A2, participants could make physical contact with the target finger, allowing terminal feedback
about their movement endpoint.
!Control!condition!
The magnitude of VEPs varies considerably between participants (e.g., due to difference in
impedance and thickness of the skull; see Allison, Wood, & Goff, 1983). For this reason, a series
of trials were performed in a control condition before both experimental sessions to normalize
VEPs for comparison across participants and sessions. The main difference between the control
and experimental trials was the absence of reaching movements.
For the control AUD-VIS condition, when the microswitch was depressed, the finger and
target LEDs were illuminated. Two seconds later, one of the three sounds were presented (i.e.,
soft, medium or loud). This sound was followed, 100 ms later, by the 50ms-flash of the green LED
of their index finger. Participants remained on the home position until the fixation stimulus was
turned off (i.e., ~3s after it was turned on). The same trial procedure was used for the control AUD-
SOMA condition, except the finger and target LEDs were not illuminated.
! Data!recording!
EEG data were recorded continuously from 64 pre-amplified Ag-AgCl (ActiveTwo, Biosemi,
Amsterdam, Netherlands) electrodes embedded in an elastic cap mapped to the extended 10-20
system. Two electrodes, a Common Mode Sense (CMS) active electrode and a Driven Right Leg
(DRL) passive electrode specific to the Biosemi system, which served as a feedback loop driving
the average potential of the measured signal to levels as close as possible to the analog-to-digital
converter reference voltage. Electrooculographic (EOG) activity was recorded bipolarly with
surface electrodes placed near both outer canthi as well as under and above the orbit of the right
eye. EOG recordings were used to identify ocular artefacts (see below) and also allowed the
experimenters to verify that participants fixated on the fixation LEDs during movement planning
and execution. The EEG and EOG signals were digitized (sampling rate 1,024 Hz, DC, 268 Hz, 3
82
dB/octave) and band-pass filtered (0.1– 45 Hz, notch filter applied at 50Hz, 12 dB/octave). As in
Experiment 1, kinematic data of the reaching finger were recorded by tracking the positions of an
electromagnetic sensor fixed on the top of the index finger and sampled at a rate of 100 Hz.
! Data!analysis!
VEPs were obtained by averaging the time-locked (-200 to +300 ms) responses to the onset of the
50 ms green flash. These epochs were averaged for each participant and for each condition. The
mean amplitude of the -200 ms to -5 ms segment of each epoch served as the pre-stimulus baseline.
The monopolar recordings were referenced to the average of the right and left mastoid electrodes.
Recordings were visually inspected and epochs with eye movements or artefacts were rejected.
The average number of traces used for each condition varied from 172 to 179 (out of a possible
180), and there were no significant differences between the amount of trials rejected between
conditions as revealed by a 2 phase (control, experimental) by 2 cue-target condition (AUD-VIS
and AUD-SOMA) repeated-measures ANOVA. The result of this analysis indicates that the signal-
to-noise ratio was not different conditions.
A current source density (CSD) analysis was employed to increase the spatial resolution of
EEG signals (Babiloni et al., 1996; Law, Rohrbaugh, Adams, & Eckardt, 1993; Perrin et al., 1989).
The signal was interpolated with a spherical spline interpolation procedure to compute the second-
order derivatives in two-dimensional space (order of splices: 3; maximal degree of Legendre
polynomials: 10; approximation parameter lambda; 1.0e-005: see Perrin et al., 1989). CSD
measurements represent a reference-free estimate of activity at each electrode and are less affected
by far-field generators than monopolar recordings (Bradshaw & Wikswo, 2001; Law et al., 1993;
Nunez et al., 1994). Thus, CSD analyses were deemed to yield measures that better reflected the
underlying cortical activities and local sources (Kayser & Tenke, 2015; Tenke & Kayser, 2012b).
CSD-VEPs computed from the left occipital and left occipital-parietal electrodes sites (O1,
PO3, PO7) were used for the main analyses. These sites, contralateral to both the flash stimuli and
the reaching hand, were chosen based on studies showing increased processing of visual
information as a result of sensorimotor transformation processes, and studies implicating the
underlying cortical areas in the use of the exteroceptive body representation (Bernier & Grafton,
83
2010; Corradi-Dell’Acqua, Hesse, Rumiati, & Fink, 2008; Corradi-Dell’Acqua, Tomasino, &
Fink, 2009; Felician et al., 2004; Lebar et al., 2015; Medendorp et al., 2008).
CSD-VEPs were assessed, for each electrode, by measuring the peak-to-peak amplitude
between the first positive and negative deflections that could be identified in all participants and
conditions after stimulus onset (see Figure 19). The latencies of the positive and negative
deflections (i.e., ~100 ms and ~150 ms, respectively) are reported for each electrode in Table 2.
The amplitude of the P100-N150 (hereafter referred to as the VEP) was expressed as the ratio of
the CSD-VEP amplitude measured in the Control and Experimental conditions (CSD-VEP ratio =
CSD-VEP experimental conditions/CSD-VEP control conditions). CSD-VEPs were represented
by a log2 transformation to reduce the nonlinearity of ratios. Normalized CSD-VEP amplitudes
were then used for statistical contrasts. An increase in normalized CSD-VEP amplitude was
considered indicative of additional visual processing in the early stages of sensorimotor
transformations.
The neural sources of the P100-N150 in all experimental and control conditions were
estimated using minimum norm technique as implemented in the Brainstorm software (Tadel et
al., 2011). To resolve the inverse problem and estimate the cortical sources of the VEP, data were
imported from all sensors, processed, and averaged for each participant, condition, and electrode.
The forward model was computed for each condition using a symmetric boundary element method
(BEM, Gramfort et al., 2010) on the anatomical MRI brain template from the Montreal
Neurological Institute (MNI Colin27). Sources of the averaged absolute activity were estimated
using the dynamic statistical parametric maps (dSPM) technique (Dale et al., 2000).
Both behavioural temporal measures (i.e., movement time and reaction time) as well as
normalized CSD-VEP amplitudes were submitted to paired samples t-tests. Effects sizes are
reported with Cohen’s dz.
84
&Grand!average!VEPs!for!each!electrode!in!the!AUDGVIS!and!AUDGSOMA!conditions!(error!patches!represent!the!betweenGparticipant!standard!deviation).!The!peakGtoGpeak!amplitudes!between!P100!and!N150!as!determined!by!a!current!source!density!analysis!were!used!for!statistical!contrasts.!!
85
The!latencies!of!the!CSDGVEPs!were!analyzed!using!a!2Gphase!(control,!experimental)!by!2Gcondition!(AUDGVIS,!AUDGSOMA)! repeatedGmeasures! ANOVA! for! each! electrode! of! interest.! The! analysis! of! latencies! did! not! reveal! any!significant!main!effects!or!interactions!between!fixation!direction!and!cueGtarget!conditions!for!either!the!P100!or!N150!component!for!any!electrode!(ps!>.05,!Fs!<!5.0).!
AUD-VIS AUD-SOMA
Control Experimental Control Experimental
Electrodes P100 N150 P100 N150 P100 N150 P100 N150
Po3 111 (23) 151 (26) 102 (16) 147 (14) 103 (18) 145 (10) 107 (14) 147 (12)
Po7 103 (7) 144 (13) 101 (14) 156 (16) 106 (13) 151 (14) 109 (8) 152 (17)
O1 101 (15) 140 (13) 104 (10) 153 (16) 112 (13) 148 (12) 106 (11) 143 (15)
Table&2.& &Mean&(and&standard&deviation)&of&latencies&(in&ms)&for&the&peaks&used&in&the&CSD@&VEP&calculation.&&
86
! Results!!
! Behavioural!Variables!
Overall, there were no significant differences between any behavioral variables related to the
temporal aspects of movement performance between the AUD-VIS and AUD-SOMA conditions.
Paired samples t-tests did not reveal significant differences in movement times (overall mean 737
ms ± 136 ms; t(9) = - 0.14 , p = .88, dz = 0.04 ) or reaction times (t(9) = -2.22, p = .06, dz = 0.7),
although reaction times tended to be longer for somatosensory targets (M = 520 ms, SD = 70 ms)
compared to visual targets (M = 480 ms, SD = 60 ms), as in Experiment A1.
! CSDGVEPs!!
The analysis of CSD-VEPs revealed larger responses to the visual stimulus when reaching to
auditory-cued somatosensory targets compared to auditory-cued visual targets in both the occipital
(O1: see Figure 20A) and occipital-parietal (Po7: Figure 20B) electrodes. At the O1 electrode,
normalized (log2) VEP amplitudes of the P100-N150 component measured in the AUD-SOMA
condition were significantly larger than those observed in the AUD-VIS condition (M = 0.8, SD =
1.2 vs. M = 0.16, SD = 0.85, t(9) = 3.2, p = 0.05, dz = 0.7). Comparable VEP amplitude differences
were found at the Po7 electrode, as the amplitude of the P100-N150 component was larger when
planning movements in the AUD-SOMA condition as compared to the AUD-VIS condition (M =
0.43, SD = 0.90 vs. M = 0.79, SD = 2.6, t(9)= 3.2, p = 0.02, dz = 1.0). In contrast, there were no
significant CSD-VEP differences at the Po3 electrode (t(9) = -0.395, p = 0.70, dz = 0.13).
Figure 21 shows the average source activity between 100 and 150 ms (i.e., between the
peaks of the P100-N150 used for the VEP calculation) projected on a cortical template for all
experimental and control conditions in both the AUD-SOMA and AUD-VIS conditions. The
source analyses revealed a greater response to the flash in left occipito-parietal areas in the AUD-
SOMA compared to the AUD-VIS condition. Interestingly, contrasting the mean activity between
the AUD-SOMA condition and its control condition revealed that movement planning led to
significantly greater visual-related activation of the left occipital and posterior parietal cortices.
87
Conversely, no significant increases in activity were found in these regions when contrasting the
AUD-VIS condition with its control condition.
! !
88
!
!
&CSD!normalized!VEPs!for!the!occipital!(O1)!and!occipitoGparietal!(PO7)!electrode!(error!bars!represent!the!standard!error!of!the!mean).!For!both!electrodes,!CSDG!VEPs!were!significantly!larger!in!the!AUDGSOMA!compared!to!the!AUDGVIS!condition.!!
89
!
&Grand!average!source!activity!for!each!condition!between!the!P100!and!N150!latency!time.!Source!activity!(color!maps! reveal! activation! levels)!was! localized! in!parietal! and!occipital!areas! in! both! experimental! and! control!conditions.!Statistical!contrasts!(pairedGsamples!tGtests,!alpha!set!to!0.05,!tGvalue!map!indicates!direction!of!effects)!revealed!significantly!more!activity!in!the!left!parietal!and!parietoGoccipital!regions!(as!indicated!by!the!shade!of!red)!for!the!AUDGSOMA!experimental!condition!compared!to!the!AUDGSOMA!control!condition.!
! !
90
! Discussion!!In Experiment A2, the cortical response to a task-irrelevant visual stimulus was measured as
participants performed reaches to both auditory-cued visual and auditory-cued somatosensory
targets. No differences in movement times or reaction times were found between the two
conditions. Because the goal of Experiment A2 was to contrast cortical activation during
movement preparation between target modalities, it was important that the VEP stimulus be
presented at roughly the same time (within 150 ms, see Bernier et al., 2009) during movement
planning in both conditions. Although there was a non-significant trend for longer reaction times
for the AUD-SOMA vs. AUD-VIS conditions, the average magnitude of these differences was 37
ms (SD = 52 ms). Thus, it is likely that the visual probe was presented at a similar time in the
motor preparation process for both the AUD-VIS and AUD-SOMA conditions.
Overall, the results of both the CSD and source analyses revealed greater response to visual
input with the AUD-SOMA condition versus AUD-VIS condition. These findings are consistent
with the hypothesis that the gaze-dependent remapping of somatosensory target positions, using
an exteroceptive body representation, employs cortical networks associated with visual processing
and visuomotor transformations. The increased sensitivity of occipital networks to visual inputs
might be linked to a re-weighting of sensory information prior to engagement in higher-order
processes. For example, because occipital areas have been implicated in early multisensory
integration (see Murray et al., 2016 for a review) the increased activation observed in the current
study could be indicative of the role of these cortical networks in encoding body positions in visual
coordinates. This idea is supported by neuroimaging studies showing that hand-selective cells of
the lateral occipital cortex are strongly lateralized to the left hemisphere (Bracci, Cavina-Pratesi,
Ietswaart, Caramazza, & Peelen, 2012; Bracci, Ietswaart, Peelen, & Cavina-Pratesi, 2010).
In addition to enhanced activity in occipital areas, increased visual processing was found
in the left PPC when planning movements to auditory-cued somatosensory targets. This result
supports the findings of several studies showing that the left PPC is associated with the formation
of exteroceptive body representations (Buxbaum & Coslett, 2001; de Vignemont, 2010; Felician
et al., 2004; Olsen & Ruby, 1941; Sirigu et al., 1991b). Previous research has revealed that PPC
neurons are also responsible for the gaze-dependent encoding of the hand and target positions
91
(Andersen, Essick, & Siegel, 1985; Buneo & Andersen, 2006; Medendorp et al., 2008;
Mountcastle et al., 1975). Although these studies have mostly used visual targets and visible body
positions, there is some evidence that gaze-dependent coding occurs in conditions without visual
inputs (Darling et al., 2007; Filimon, 2010; Medendorp et al., 2008). In a PET functional brain
imaging study, Darling et al. (2007) found that additional neural activation in the occipital and
posterior parietal lobes when their participants reached to a memorized somatosensory target (i.e.,
position held by the unseen hand before being passively displaced). Based on these findings, the
authors suggested that networks employed to guide reaches to visual targets were also used when
reaching to memorized somatosensory targets. These findings therefore build on those of Darling
et al. (2007) by showing that visual processes are also involved when planning movements to the
actual, non-memorized, position of kinesthetic targets.
Previous studies have shown that, in some cases, visual information is attenuated (i.e.,
gated, see Chapman, Bushnell, Miron, Duncan, & Lund, 1987) during movement planning to
somatosensory targets (Bernier, Gauthier, & Blouin, 2007; Blouin et al., 2014; Sarlegna &
Sainburg, 2007; Sober & Sabes, 2005). Evidence from both behavioural and neurophysiological
experiments have indicated that, when target locations are encoded in somatosensory coordinates
(e.g., fingers of the non-reaching hand), visual information about the reaching limb does not
contribute to the same extent to computation of the reach vector (Sober & Sabes, 2005). Although
there is substantial evidence that transformation networks exist to support direct conversions of
somatosensory information into a movement vector, conflicting evidence about the contexts
wherein these networks are employed have led to uncertainty about their role in motor planning
processes (Battaglia-Mayer et al., 2003; Bernier et al., 2009; Buneo & Andersen, 2006; Prevosto
et al., 2010; Saradjian, 2015). Based on the results of Study A, one contextual factor that should
be considered when evaluating the involvement of these networks is the body representation used
to define somatosensory target locations.
Lastly, it should be noted that the results of the present study do not necessarily support a
disengagement of the occipital cortex and PPC when planning movements to auditory-cued visual
targets or that visual information was not relevant motor planning to visual targets. Rather, the
results of the present study suggest that planning movements when vision of the reaching hand and
target is available does not require additional processing from areas sensitive to visual inputs,
92
compared to the control condition, at least at the time when the visual probe was presented (i.e.,
100 ms after target cue presentation).
! Conclusion!In summary, it was found that the sensory modality employed to identify the position of a
somatosensory target impacts the body representation used for movement planning processes.
Preparing reaching movements towards auditory-cued somatosensory targets prompted the use of
an exteroceptive representation of the body. Conversely, an interoceptive body representation was
used when the position of the somatosensory targets was indicated by tactile stimulation.
Furthermore, it was found that the time required to initiate movement towards auditory-cued
somatosensory targets is longer than when the targets are indicated by tactile cues. This delay
likely indicates that the use of an exteroceptive representation to encode body position involves
more complex sensorimotor transformation processes as compared to conditions where the
transformation of target and effector positions is not required. By measuring the cortical response
to a visual stimulus, it was found that the additional sensorimotor transformation processes elicited
by the use of an exteroceptive body representation were associated with increased visual
processing in occipital and posterior parietal areas. Taken together, the findings of the present
study indicate that the body representation used to derive somatosensory target locations impacts
sensorimotor integration processes during movement planning. And more specifically, auditory
cues indicating the somatosensory target to be reached invokes visual sensorimotor processes and
an exteroceptive body representation.
93
!!Rapid!online!corrections!for!upperGlimb!reaches!to!perturbed!somatosensory!targets:!Evidence!for!nonGvisual!sensorimotor!
transformation!processes!!
A version of STUDY B has been accepted for publication in the journal: Experimental Brain
Research
94
! Study!B!
! Abstract!
When performing upper-limb reaches, the sensorimotor system can adjust to changes in target
location even if the reaching limb is not visible. To accomplish this task, sensory information about
the new target location and the current position of the unseen limb are used to program online
corrections. Previous researchers have argued that, prior to the initiation of corrections,
somatosensory information from the unseen limb must be transformed into a visual reference
frame. However, most of these previous studies involved movements to visual targets. The purpose
of the present study was to determine if visual sensorimotor transformations are also necessary for
the online control of movements to somatosensory targets. Participants performed reaches towards
somatosensory and visual targets without vision of their reaching limb. Target positions were
either stationary, or perturbed before (~450 ms), or after movement onset (~100 ms or ~200 ms).
In response to target perturbations after movement onset, participants exhibited shorter correction
latencies, larger correction magnitudes, and smaller movement endpoint errors when they reached
to somatosensory targets as compared to visual targets. Because reference frame transformations
have been shown to increase both processing time and errors, these results indicate that hand
position was not transformed into visual reference frame during online corrections for movements
to somatosensory targets. These findings support the idea that different sensorimotor
transformations are used for the online control of movements to somatosensory and visual targets.
Keywords: reaching, online-control, somatosensory targets, sensorimotor transformations,
double-step
95
! Introduction!
When performing movements towards visual targets, the motor system can successfully adapt to
changes in target location (Day and Lyon 2000; Johnson et al. 2002; Sarlegna and Mutha 2015;
Smeets et al. 1990). These corrections can occur even when perturbations are not detected or when
there is no vision of the reaching limb (Goodale et al. 1986; Heath 2005; Komilis et al. 1993;
Pélisson et al. 1986; Reichenbach et al. 2009; Saunders and Knill 2003). When the reaching limb
is not visible, visual information about the new target location and somatosensory information
about the limb’s position can be used to perform trajectory amendments. Because movements to
visual targets are hypothesized to be planned in a visual reference frame (Ambrosini et al. 2012;
Buneo and Andersen 2006; Thompson et al. 2012, 2014), previous studies have argued that the
reaching limb’s position must be converted into visual coordinates prior to the initiation of
corrections (Prablanc & Martin, 1992; Reichenbach et al., 2009). It is unknown if such online
sensorimotor transformation processes also occur for reaches to non-visual targets. The purpose
of the present study was to investigate the online sensorimotor transformation processes involved
in the online control of reaches to somatosensory targets performed without vision of the reaching
limb.
! OnlineG!visuomotor!transformation!processes!for!movements!to!visual!targets!
Previous studies have found that rapid reaches to perturbed targets without vision of the limb are
less accurate and have longer correction latencies when there is no vision of the reaching limb
compared to when there is vision of the limb. For example, Komilis et al. (1993) had participants
perform reaching movements to visual targets that were perturbed either at movement onset or at
peak velocity. Movements could be performed either with or without vision of the limb. Similar
to previous studies (Goodale et al., 1986), the authors found that participants were able to
completely correct their movement trajectories in response to target perturbations at movement
onset, regardless of whether or not the limb was visible. However, for perturbations at peak limb
velocity, movements performed without vision of the limb were slower and slightly less accurate
than movements performed with vision of the limb (see also Heath 2005). Reichenbach et al.
96
(2009) also noted that correction latencies were longer (+10 ms based on EMG and +30 ms based
on limb kinematics) when participants performed reaches to visual targets without vision of their
limb compared to when they reached with vision of their limb. It was reasoned that, without vision
of the limb, corrections are programmed based on the updated visual target location, and visual
estimates of the current limb position derived from somatosensory inputs and efferent information
(see also. Reichenbach and colleagues concluded that when the limb is not visible additional time
is required for the transformation of somatosensory information about the limb position to a visual
estimate prior to the programming of the correction. Together, the results of these studies support
the idea that a common visual reference frame is used for the online control of actions irrespective
of the sensory modality used to encode hand (see also Buneo and Andersen 2006; Buneo et al.
2008).
! Sensorimotor!transformation!processes!for!movements!to!somatosensory!targets:!planning!and!online!control!
The idea that a common visual reference frame is used for the online control of goal-directed
actions is based primarily on studies of movements to visual targets. Very little is known about
online sensorimotor control processes for movements to somatosensory targets. Research on
movement planning processes, however, has revealed that the references frame used to plan
movements to somatosensory targets could be visual ( Blangero et al. 2005; Jones and Henriques
2010; Pouget et al. 2002) or non-visual (Bernier et al. 2007, 2009; Blouin et al. 2014; McGuire
and Sabes 2009; Sarlegna and Sainburg 2007; Sober and Sabes 2005).
Both Pouget et al. (2002) and Jones and Henriques (2010) found that when participants
shifted their gaze position prior to reaching towards somatosensory targets, participants’ endpoint
errors were biased in the direction opposite to the gaze-shift (see also Blangero et al., 2005). The
effect of gaze position on movement accuracy was similar to that observed when participants
reached to visual targets (see Bock, 1986; Henriques et al., 1998). As mentioned in the above
Section 2.4.3.1.1 (Behavioural Evidence for a Gaze-Dependent Reference Frame), movements to
external visual targets are planned used in visual coordinates. Because these studies have shown
similar errors for movements to targets of both modalities, the authors concluded that movements
to somatosensory targets were also planned in a visual coordinate system (see also Mueller and
Fiehler 2014).
97
In contrast, other studies have found that movements to somatosensory targets can be
planned in a non-visual coordinate system. For example, Sarlegna and Sainburg (2007) found that
altering visual information about the initial hand position hand no effect on endpoint errors when
participants reached to somatosensory targets but did hinder endpoint performance when they
reached to visual targets (see also McGuire and Sabes 2009; Sober and Sabes 2005). Because both
the limb and the target can be represented in somatosensory coordinates when there was no vision
of the limb (Battaglia-Mayer et al., 2003), and movements were unaffected by visual perturbations,
the authors concluded that the computation of the movement vector occurred in a non-visual
reference frame. Other studies have also suggested that using non-visual sensorimotor
transformations for movement planning to somatosensory targets may avoid errors which are
associated with the conversion from a somatosensory coordinate system to a visual coordinate
system (Sarlegna et al., 2009).
Although sensorimotor transformations in both visual and non-visual coordinate systems
have been found for the planning of movements to somatosensory targets, the type of sensorimotor
transformation employed for online trajectory amendments remains unclear. In the present study,
the latency and magnitude of online trajectory corrections to perturbed visual and somatosensory
targets were assessed to investigate the sensorimotor transformation processes used for the online
control of an unseen reaching limb. It was hypothesized that if a visual reference frame is
employed, longer correction latencies and smaller corrections should be observed for reaches to
somatosensory targets as compared to reaches to visual targets. This is because somatosensory
cues from both the perturbed somatosensory target position and reaching limb would have to be
converted into visual coordinates prior to the initiation of corrections. For movements to visual
targets, only the reaching limb position would require a conversion into a visual coordinate system.
In contrast, if corrections are programmed in a non-visual reference frame then we would expect
faster and more accurate corrections in response to somatosensory target perturbations compared
to visual target perturbations. Corrections in a non-visual reference frame would be programmed
using the new target and limb position in somatosensory coordinates, without the need for a
reference frame transformation.
98
! Methods!
! Participants!
Fourteen participants (10 women, aged 20 – 33 M = 25, SD = 4) took part in the experiment. All
participants were right-handed (assessed by the Edinburgh handedness questionnaire, adapted
from Oldfield 1971), self-declared neurologically healthy, and had normal or corrected-to-normal
vision.
Written informed consent was obtained prior to the experimental protocol and the
University of Toronto’s Office of Research Ethics approved all procedures. Including the informed
consent, breaks, and debriefing, the experiment lasted approximately 3 hours and participants were
compensated with $20 CAD.
! Apparatus!
A drawn representation of the experimental setup is shown in Figure 22. The experiment was
conducted in a completely dark room, where participants were seated comfortably on a kneeling
chair facing a protective cage. Participants placed their head on a headrest that was positioned on
the outside of the cage. They interacted with the experimental materials that were inside the cage
through a window (80 cm high). A small microphone (FBA_4330948551, Phantom YoYo) that
was used for the vocal response time protocol (see below) was placed near the headrest position.
Inside the cage was a Selectively Compliant Assembly Robot Arm (SCARA; Epson
E2L853, Seiko Epson Corp.) that was used to position target stimuli in both the visual and
somatosensory target conditions (see Manson et al. 2014 and appendix 2 for details on the robotic
device). Located directly below the robot was a table with a custom-built aiming apparatus placed
on the table’s surface.
The aiming apparatus included a black tinted Plexiglas aiming surface (60 cm wide by 45
cm long by 0.5 cm thick) mounted 12 cm above a wooden base. Underneath the aiming surface,
there was a textured home position (2 cm by 2 cm). On the top of the aiming surface was a blue
light emitting diode (LED, 2 mm diameter) that served as the gaze fixation point. The LED was
aligned with the participant’s midline and was located ~65 cm from the participant’s eye position.
99
In the robot’s neutral position (i.e., for no-perturbation trials), the custom end-effector
attached to the robot arm was positioned 0.5 cm above the aiming surface and 35 cm to the left of
the home position. Participants grasped the robot’s end effector with their left hand and were
instructed to depress a micro-switch (B3F-1100, OMRON) with their index finger. The micro-
switch served as both a reference for the somatosensory target location and a safety mechanism
that shuts off the robot if the button was released (note: no participants released the switch during
the study). In the visual target condition, a green LED (~6mm diameter) was attached to the robot’s
end effector (i.e., at the microswitch location) and served as the target.
The participant’s reaching fingertip and the robot’s end effector were both affixed with an
infrared light emitting diode (IRED). An Optotrak Certus (Northern Digital Inc.) motion tracking
system was used to track movements of the robotic arm and the reaching finger at 200Hz. A
custom MATLAB (The Mathworks Inc.) program was used to gather data from the Optotrak and
microphone, as well as send outputs to the aiming console and robotic effector. A piezoelectric
buzzer (SC628, Mallory Sonalert Products Inc.) was used to provide brief auditory cues to the
participants.
100
&!A!drawn!representation!of!the!experimental!setup!(not!to!scale)!is!shown!in!panels!(a)!and!(b).!Panel!(c)!is!a!representation!of!the!aiming!console!and!stimuli! for!each!target!modality.! !Participants!sat!comfortably!facing!the!aiming! apparatus! in! a! completely! dark! room.! A! robotic! device! was! used! to! deliver! target! perturbations! in! both!somatosensory! (left! finger)! and! visual! (LED)! targets.!Participants! performed! leftward! arm!movements! to! reach! the!position!of!the!target!as!if!it!was!projected!onto!the!aiming!surface.!!
101
! Procedure!
Participants completed the experimental tasks over two sessions, one for each target modality. The
presentation of target modality was counterbalanced across participants, and the time between
sessions was between 5 and 14 days (M = 10.5 days) for most participants.
Each session consisted of two protocols: a vocal response time protocol, and a reaching
protocol. In the first protocol, participants were asked to make a vocal response to the perturbation
of the target stimulus. In the second protocol, participants were asked to reach to the target stimulus
as accurately as possible within a movement time bandwidth (i.e., 450 – 600 ms from movement
start to movement end). Participants were given the instructions for the reaching protocol only
once they had completed the vocal response time protocol.
! Vocal!Response!Time!Protocol:!
The goal of the vocal response time protocol was to examine whether the modality of the target
could alter the time taken to detect the motion of the target. For the somatosensory target session,
participants responded to perturbations of the left limb placed on the robot’s end effector. For the
visual target session, participants responded to the perturbation of the target LED placed on the
robot’s end effector. During both the sessions, participants placed their right index finger on the
home position located underneath the aiming surface (see Fig. 22a and 22b).
At the beginning of each trial, the fixation light was turned on. After the experimenter
verified that participants were on the home position, the trial was started. After a random
foreperiod, the robot perturbed the visual or somatosensory target either 3 cm toward or away from
the participants, or 3 cm to the left of them (i.e., “catch trials”). The duration (M = 200 ms, SD =
3.8) and velocity of the target perturbations were the same in both the vocal response time protocol
and the reaching protocol. Participants were instructed to verbally respond with “Yo!” as soon as
the target stimulus moved either toward or away from their body, and to not respond when the
target moved to their left. Note that participants did react on 0.02% of the catch trials, however
none of the participants reacted to more than 1 catch trial. Following the recording of the vocal
response, the fixation light turned off and the robot returned to the neutral position.
102
Participants were first presented with a set of familiarization trials (6 trials) wherein each possible
perturbation direction (e.g., away, towards, and catch) was presented twice in a row. Following
the familiarization trials, participants were presented with a randomized order of 58 perturbation
trials (24 away, 24 toward, and 10 catch).Vocal response times were computed as the time
difference between the onset of target displacement (i.e., when the velocity of the robot surpassed
30 mm/s) and the response time recorded by the microphone (see Fig. 23).
! Reaching!Protocol:!
After participants completed the vocal response time protocol, the experimenter explained the
reaching protocol and trial procedures. The reaching protocol consisted of 4 phases:
familiarization, perception of target position in the pre-test, reaching trials, and perception of target
position in the post-test. All 4 phases were performed for both target modalities.
There were two kinds of reaching trials: no-perturbation and perturbation reaching trials.
In the no-perturbation trials, participants performed movements from the home position to the
neutral target position within a movement time bandwidth of 450 – 600 ms. Each trial started with
the illumination of the fixation LED. In the visual target session, the target LED was also
illuminated at the same time as the fixation LED. Four hundred milliseconds after the fixation
LED was turned on, an auditory go signal (50 ms beep) cued participants to begin their reaching
movement. The time instant at which finger velocity raised above and fell below 30 mm/s for more
than 10 ms marked movement start and end, respectively. Once the reaching movement was
completed, participants received auditory feedback about their movement time and the fixation
LED was turned off. The auditory feedback indicated to the participants whether or not their
movement time was within the time bandwidth. Participants were presented with two short (50
ms) beeps if their movement time was within the bandwidth; one long (100 ms) beep if their
movement time was too shorter than the lower limit of the bandwidth; or three short (50 ms) beeps
if their movement time was longer than the upper limit of the bandwidth. The beeps also served
as a signal to move back to the home position to begin the next trial.
In the perturbation trials, the target was shifted 3 cm away from or towards the participant
either 300 ms before the go signal, or ~100 ms or ~200 ms after the movement onset. On average,
these perturbations occurred 450 ms (SD = 73 ms) before movement onset, or 93 ms (SD = 4 ms)
103
or 190 ms (SD = 4 ms) after movement onset. (see Figure 23 for velocity profiles of hand and
robot movements in each perturbation condition). The change of target position required a change
in movement direction (with consideration that movements are planned as a vector defined in terms
of amplitude and direction; Buneo and Andersen 2006; Dadarlat et al. 2015, Desmurget et al.
1998). Offline analyses showed that the 100 ms perturbation time occurred prior to peak velocity,
at a time during which visual information has been found to be important for online corrections
(Kennedy et al., 2015; Tremblay et al., 2017). The 200 ms time roughly corresponded with the
peak velocity of the aiming movements. This time may be too late to use visual feedback
effectively (e.g., Kennedy et al., 2015) but may still be viable for the use of somatosensory
information (Goodman et al. 2018; Redon et al. 1991). Perturbations before movement onset were
included as a control condition to compare the corrections that resulted from planning and online
control processes between somatosensory and visual conditions.
Participants were asked to reach to the position of the target stimulus (e.g., the surface area
of the finger on the button, or target LED) as if projected onto the underneath of the aiming surface.
To perform the reaching task, participants performed an “underhanded” reaching movement,
primarily with muscles in their shoulder joints. Participants' wrists remained supinated throughout
the trajectory. To discourage wrist movements, participants also wore a wrist orthotic (Champion-
C218, Champion Health and Sports Supports, Cincinnati, USA). Participants performed 180
reaching trials for each target modality (360 trials total). These trials consisted of 90 no-
perturbation trials and 90 perturbation trials (15 trials in each of the 6 perturbation conditions).
None of the perturbation conditions were repeated more than twice consecutively.
104
&!Examples!of!velocity!profiles!of! the!reaching!hand!and!the!robotic!effector!(Target)!for!each!perturbation!time.!Panel!(a)!shows!a!perturbation!occurring!before!movement!onsetl!panel!(b)!shows!a!perturbation!occurring!~100!ms!after!movement!onset,!and!panel!(c)!shows!a!perturbation!occurring!~200!ms!after!movement!onset.!
105
!Familiarization!
After receiving the instructions for the reaching task, participants performed 30 trials to familiarize
themselves with the experimental task, the auditory feedback, and the movement time bandwidth.
Participants were presented with 18 no-perturbation trials followed by 12 perturbation trials (2
trials in every perturbation condition).
!Perception!of!Target!Position!Pre!and!Post!Tests!
To record their perceptions of target positions, participants were asked to reach to where they
perceived each target position was as if it were projected onto the underneath surface of the
Plexiglas. Participants first reached to the centre target and adjusted their index finger until they
felt it matched the target’s position. Participants then verbally indicated to the experimenter when
their hands were on the target position, and this position was recorded. Once the reaching hand
was returned to the home position, the robotic effector was moved to the ‘away’ target position
and the response procedure was repeated. Finally, the entire sequence was then repeated for the
‘toward’ target position. Participants’ perceived target locations were recorded twice during each
session, once after the familiarization trials (pre) and again after the completion of the reaching
trials (post). The perceived target locations recorded during these trials were used for endpoint
error calculations (see constant and variable error sections in results).
! Data!Analysis!
!Vocal!Response!Time!Protocol!
For vocal response time data, trials with a response time more than 3 standard deviations higher
or lower than the participant’s mean were removed. These accounted for 5.2% of all vocal response
trials. Vocal response data were submitted to a 2-perturbation direction (away, toward) by 2-target
modality (somatosensory, visual) repeated-measures ANOVA.
!Reaching!Protocol!
Trials with movement times, reaction times, or endpoint errors in both the movement
amplitude axis (initial movement axis) and movement direction axis (axis of perturbation) that
106
were more than 3 standard deviations from the mean were excluded from the analyses. This
resulted in the exclusion of 4.5% of all reaching trials. The main dependent variables for this
experiment were constant error, variable error, correction magnitude, and the latency of online
corrections.
!Constant!and!Variable!Error!!
Constant error was calculated as the bias in endpoint position relative to the participant’s averaged
perceived target position (calculated using the pre- and post- target perception trials). Constant
error was computed for both the amplitude and the direction axes (hereafter referred to as
amplitude constant errors and direction constant errors, respectively). Variable errors were
computed by calculating the standard deviation of these constant errors (hereafter referred to as
amplitude variable error and direction variable error).
For amplitude constant errors, positive values indicated an overshoot relative to the target location
whereas negative values indicated an undershoot relative to the target location. Similarly, for
direction constant errors, positive values represent an over-correction relative to the new target’s
position, whereas negative values represent an under-correction relative to the new target’s
position.
Amplitude and direction constant and variable errors were submitted to separate 2-target modality
(somatosensory, visual) by 2-perturbation direction (away from the body and toward the body) by
3-perturbation time (before, 100 ms, 200 ms) repeated-measures ANOVAs.
!Correction!Magnitude!!
Correction magnitude was calculated as the average of the absolute difference between the average
end position of the perturbation trials (e.g., before, 100 ms, and 200 ms) and the average end
position of the no-perturbation trials. This measure was only computed for the direction axis (i.e.,
axis of the perturbations). It is worth noting that there were no differences in overall endpoint
variability between somatosensory and visual target conditions in the no-perturbation conditions
t(13) = -1.35 p = 0.20 (see results section: Comparison of no-perturbation trials, and Table 2.
Correction magnitudes were submitted to a 2- target modality (somatosensory, visual) by 2-
107
perturbation direction (away, toward) by 3-perturbation time (before, 100 ms, and 200 ms)
repeated-measures ANOVA.
!Latency!of!Online!Corrections!
The method of determining the latency of online corrections was adapted from Oostwoud
Wijdenes et al. (2013). Using this method, correction latency was computed based on a linear
extrapolation of the differences in the average acceleration profiles in the movement direction axis
(axis of the perturbation) between no-perturbation and perturbation trials (see also Oostwoud
Wijdenes, Brenner, & Smeets, 2011; Veerman et al., 2008). When tested on simulated data, this
extrapolation method was deemed to be the most accurate and precise method for detecting
correction latencies (Oostwoud Wijdenes et al., 2014).
Accelerations profiles in the movement direction axis were computed by a double
differentiation of the displacement data obtained from sampling the finger IRED and subsequently
low pass-filtering these time-series with a second-order recursive bidirectional Butterworth filter
at 50Hz. For each participant, the difference in the average acceleration profiles were computed
between no-perturbation and 100 ms perturbation trials, and no-perturbation and 200 ms
perturbation trials. These difference profiles were then used to compute the correction latency.
The no-perturbation trials used for the computation of the average acceleration profile were
selected based on the distribution of movement times used to compute the profile in the
perturbation condition. The number of control trials used to compute the average trajectories thus
varied for each condition within each participant. The number of control trials used ranged from
25 to 85 (M = 59 trials, SD = 14). Overall, movement times between perturbation trials (M = 544
ms, SD = 17) and control trials (M = 543 ms, SD = 16) were not significantly different, as indicated
by a paired-samples t-test, t(13) = 0.86, p = 0.41.
To determine correction latency, the maximum acceleration value occurring after
perturbation was first identified. Second, a line was drawn between the points on the acceleration
profile corresponding to 25% and 75% of the maximum acceleration. Response latency was
defined as the difference between the time of perturbation and the time where this line crossed
108
zero (i.e., y value of zero; see Oostwoud Wijdenes et al. 2013, 2014; Veerman et al. 2008) and
Figure 24 for a graphical representation of the method applied to participants’ data from the present
study).
Correction latencies were submitted to a 2-target modality (somatosensory, visual) by 2-
perturbation time (100 ms, 200 ms) by 2-perturbation direction (away, toward) repeated-measures
ANOVA. Note that corrections to the target perturbations that occurred before movement onset
likely occurred during movement planning and are therefore no reflective of online control
processes. Furthermore, accurate correction latencies for the “before” perturbation time are
unlikely to be detected by this method. Because of the above-stated reasons, the before
perturbation condition was not included in the statistical design.
It is also worth noting that for this analysis, the absolute values of correction latency were
not as important as the between-modality differences. In the present study, because of technical
limitations between the syncing of the Optotrak and robotic apparatuses, reaching data were
sampled at a lower frequency (200 Hz) than what was used in previous studies (500 Hz: Oostwoud
Wijdenes et al. 2011, 2014). Also, this was the first time such a method was applied to reaching
movements using performed with a supinated wrist posture to a somatosensory target. Thus, some
contrasts between our values and those commonly found in the literature were expected.
109
&The!extrapolation!method! for!determining! the! latency!of!online!corrections.!For!each!participant!average!acceleration! profiles! in! the! direction! axis! were! computed! for! both! perturbation! and! noGperturbation! trials.! The!acceleration! difference! between! these! profiles! (Accel! Difference)! was! then! plotted! to! calculate! correction! latency.!Correction!latencies!were!computed!by!drawing!a! line!(Extrapolation!Line)!between!75%!and!25%!of!the!maximum!difference!in!the!Accel!Difference!profile!(Extrapolation!points)!and!extrapolating!the!line!to!the!first!zero!crossing.!The!time!between!the!perturbation!and!the!zero!crossing!was!defined!as!the!correction!latency.!Panel!(a)!shows!this!method!applied!to!averaged!data!for!somatosensory!target!perturbations!and!panel!(b)!shows!the!method!applied!to!averaged!data!for!visual!target!perturbations.!!
110
!Comparison!of!noGperturbation!trials!!
To examine the effect of target modality on reaching performance and kinematics, paired samples
t-tests were performed (effect sizes reported with Cohen’s dz) on reaction time, movement time,
total movement amplitude, amplitude and direction constant and variable errors, as well as on time
to, and time after peak velocity. Note that for these comparisons, direction constant errors were
defined using a coordinate system relative to the home and target positions. Negative values
indicated deviations closer to the body with respect to the target, and positive values indicated
endpoints further away from the body with respect to the target.
!Statistics!and!postGhoc!tests!!
All statistical analyses were performed using the Statistical Package for Social Scientists (SPSS:
IBM Inc. version 20). For all t-tests and repeated-measures ANOVA’s, alpha was set to p = 0.05.
Only the significant main effects and interactions were reported. Also, when main effects could be
solely explained by a higher order interaction, only the break-down of the interaction was reported.
The Hyunh-Feldt correction was used to correct the degrees of freedom (corrected to 1 decimal
place) when the assumption of sphericity was violated. The Tukey’s Honestly Significant
Differences (HSD) test was used to decompose all significant interactions.
! Results!!
! Vocal!Response!Protocol!!
For target shift detection times, the analysis yielded a significant main effect of stimulus modality,
F(1,13) = 5.04, p < 0.05, ηP2 = 0.28 and a significant stimulus modality by perturbation direction
interaction, F(1,13) = 5.56, p < 0.05, ηP2 = 0.30, HSD = 22 ms. Overall, participants detected
perturbations to somatosensory targets (M = 452 ms, SD = 95) faster than perturbations to visual
targets (M = 486 ms, SD = 104). Breaking down the interaction between modality and perturbation
direction revealed that, when the stimulus was perturbed away from the body, there was a larger
difference in detection times between somatosensory and visual stimuli (somatosensory away: M
= 445 ms, SD = 100; visual away: M = 491ms, SD = 102) than when the stimuli were perturbed
toward the body (somatosensory towards: M = 459 ms, SD = 92; visual towards: M = 481 ms, SD
= 108).
111
! Reaching!Protocol!!
Normalized trajectory profiles for each condition in the reaching protocol are displayed in Figure
25. Also, the overall kinematic and temporal features of the movements are shown in Table 3
(visual targets) and Table 4 (somatosensory targets).
! Comparison!of!noGperturbation!trials!
First, no-perturbation trials were analyzed to determine whether the sensory modality of the target
had significant effects on the different reaching variables. The analyses revealed significant effects
of target modality on the reaction time, t(13) = 2.98, p < 0.05, dz = 0.8 movement amplitude, t(13)
= 4.12, p < 0.001, dz = 1.1 and constant error in the movement direction axis, t(13) = -3.15, p <
0.01, dz = 0.8 (for all values see Table 1 and Table 2). Participants took more time to initiate
movements to somatosensory targets (M = 330 ms, SD = 60) compared to visual targets (M = 253
ms, SD = 71). Also, participants had larger movement amplitudes when reaching to somatosensory
targets (M = 35.2 ms, SD = 3.4) as compared to visual targets (M = 31.2 ms, SD = 5.0). Finally,
participant’s endpoints were distributed further away from the body when reaching to visual targets
(M = 1.31 cm, SD = 1.19) compared to somatosensory targets (M = 0.16 cm, SD = 1.07).
112
&&Average!reaching!trajectories!for!each!condition!in!the!reaching!protocol.!Panels!(a),!(b),!depict!perturbations!occurring!before!movement!onset.!Panels!(c)!and!(d)!depict!perturbations!100!ms!after!movement!onset.!Panels!(e)!and! (f)! depict! perturbations!occurring! 200!ms!after!movement! onset.! Trajectories!were! normalized!with! each!point!representing!2%!of!movement!duration.!Error!bars!indicated!the!betweenGsubject!standard!deviation!of!spatial!position!
113
Table&3.& Mean%and%standard%deviation%for%the%temporal%and%kinematic%variables%of%movements%to%somatosensory%(somato)%and%visual%targets.%
Perturbation Conditions
Variables Targets
Before
100ms
200ms
No Perturbation
Away Toward
Away Toward
Away Toward
Reaction Time (ms) Visual
253 (71)
119 (69) 105 (62)
254 (75) 253 (93)
257 (77) 257 (77)
Somato
330 (60)
127(83) 135 (75)
326 (48) 324 (60)
332 (61) 320 (68)
Movement Time (ms) Visual
521 (22)
518 (27) 506 (32)
559 (40) 544 (32)
533 (24) 559 (30)
Somato
522 (14)
515 (20) 519 (32)
554 (22) 525 (33)
550 (30) 536 (30)
Robot-Hand Start Difference (ms) Visual
-
-449 (71) -435 (69)
91 (2) 91 (2)
192 (1) 191 (2)
Somato
-
-453 (85) -459 (73)
95 (3) 95 (4)
193 (2) 192 (2)
Time to Peak Velocity (%) Visual
39 (5)
36 (5) 37 (4)
37 (5) 37 (5)
37 (6) 37 (4)
Somato
38 (5)
36 (5) 37 (5)
36 (5) 40 (7)
37 (6) 38 (6)
Time after Peak Velocity (%) Visual
61 (5)
64 (5) 63 (4)
63 (5) 63 (5)
63 (6) 63 (4)
Somato
62 (5)
64 (5) 63 (5)
64 (5) 60 (7)
63 (6) 62 (6)
Peak Velocity (m/s) Visual
1.12 (0.18)
1.13 (0.18) 1.24 (0.20)
1.12 (0.14) 1.11 (0.18)
1.13 (0.18) 1.11 (0.18)
Somato 1.31 (0.17) 1.33 (0.19) 1.45 (0.21) 1.27 (0.18) 1.34 (0.15) 1.28 (0.17) 1.30 (0.18)
114
%
Table&4.& Mean%and%standard%deviation%(in%cm)%for%the%accuracy%variables%of%movements%to%somatosensory%(somato)%and%visual%targets.%
Perturbation Conditions
Variables Targets
Before
100ms
200ms
No Perturbation
Away Toward
Away Toward
Away Toward
Movement Amplitude Visual
31.2 (5)
30.9 (5) 31.4 (4.6)
30.9 (4.8) 31.6 (4.8)
31.2 (5.0) 31.6 (4.8)
Somato
35.2 (3.4)
34.7 (3.5) 35.7 (3.4)
34.6 (3.6) 36.1 (3.4)
34.7 (3.5) 36.2 (3.4)
Amplitude Constant Error Visual
-0.43 (2.22)
-0.02 (1.97) 0.19 (2.33)
0.02 (1.83) 0.40 (2.47)
0.23 (1.91) 0.39 (2.50)
Somato
0.38 (1.49)
0.48 (2.14) 0.66 (2.18)
0.38 (2.12) 1.07 (2.30)
0.44 (1.97) 1.14 (2.45)
Direction Constant Error Visual
1.32 (1.20)
0.47 (1.82) -0.10 (1.63)
-0.14 (2.02) -0.80 (2.11)
-1.62 (1.45) -2.03 (2.03)
Somato
0.16 (1.07)
-0.33 (1.30) -0.07 (1.69)
-0.06 (1.21) 0.60 (1.33)
-1.03 (1.35) -0.73 (1.24)
Amplitude Variable Error Visual
1.47 (0.35)
1.60 (0.92) 1.48 (0.71)
1.55 (0.93) 1.23 (0.30)
1.51 (0.63) 1.43 (0.78)
Somato
1.55 (0.49)
1.53 (0.61) 1.55 (0.38)
1.62 (0.52) 1.60 (0.63)
1.65 (0.67) 1.55 (0.54)
Direction Variable Error Visual
1.39 (0.35)
1.20 (0.37) 1.28 (0.47)
1.56 (0.64) 1.64 (0.41)
1.71 (0.55) 1.82 (0.48)
Somato
1.28 (0.23)
0.99 (0.28) 1.34 (0.38)
1.12 (0.28) 1.70 (0.69)
1.37 (0.55) 1.94 (0.76)
Correction Magnitude Visual
-
3.46 (0.94) 3.71 (0.81)
3.08 (0.97) 3.10 (0.70)
1.83 (0.57) 2.08 (0.72)
Somato - 4.18 (0.99) 4.68 (1.53) 4.45 (0.86) 5.34 (1.23) 3.53 (0.90) 4.06 (1.02)
115
! Constant!and!Variable!Errors!
For direction constant error, the ANOVA yielded a significant main effect of perturbation time,
F(2,26) = 40.85, p < .001, ηP2 = 0.76, HSD = 0.41 cm, and a significant modality by perturbation
time interaction, F(2,26) = 11.94, p < .001, ηP2 = 0.48, HSD = 0.64 cm. Breaking down the 2-way
interaction revealed that participants showed smaller direction constant errors in response to target
perturbations in both the 100 ms and 200 ms perturbation conditions when they reached to
somatosensory targets as compared to when they reached to visual targets. No difference in
direction constant error between target modalities was observed in response to perturbations before
movement onset (see Fig. 26a and Table 3 & 4). For the trials with target perturbations, the
analyses yielded no significant main effects or interactions for amplitude constant and amplitude
variable errors (Fs > 0.09 & ps > 0.098).
For direction variable error, the analysis yielded a significant main effect of perturbation direction,
F(1,13) = 16.80, p < .001, ηP2 = 0.62; perturbation time, F(2,26) = 21.00, p < .001, ηP2 = 0.56,
HSD = 0.20; and a target modality by perturbation direction interaction, F(1,13) = 11.54, p <.01,
, ηP2 = 0.47, HSD = 0.25 cm. Participants’ endpoint variability was significantly higher when they
reached to targets perturbed 200 ms after movement onset (M = 1.71 cm, SD = 0.62) compared to
when they reached to targets perturbed 100 ms after movement onset (M = 1.51 cm, SD= 0.56).
Direction variable errors were also significantly higher in response to both 100 ms and 200 ms
perturbations than in response to perturbations before movement onset (M = 1.20 cm, SD = 0.39).
Breaking down the target modality by perturbation direction interaction revealed that, when
moving to a somatosensory target, participants' direction variable errors were significantly lower
when the target was perturbed towards the body (M = 1.16 cm, SD = 0.37) compared to when the
target was perturbed away from the body (M =1.66 cm, SD = 0.56). There were no differences
between perturbation directions for reaches to visual targets (global M = 1.62 cm SD = 0.58).
! Correction!Magnitude!
The analysis of correction magnitude yielded a significant main effect of target modality, F(1,13)
= 70.07, p <.001, ηP2 = 0.84 perturbation time, F(2,26) = 39.81, p <.001, ηP2 = 0.75 HSD = 0.63
cm, perturbation direction, F(1,13) = 7.71, p <.05, ηP2 = 0.37 and a target modality by perturbation
time interaction, F(2,26) = 10.04, p < .001, ηP2 = 0.44, HSD = 0.55 cm. Participants exhibited
116
larger corrections in response to targets perturbed toward the body (M = 3.83 cm, SD = 1.47)
compared to targets perturbed away from the body (M = 3.42 cm, SD = 1.20). Decomposing the
target modality by perturbation time interaction revealed that, overall participants performed larger
corrections in response to somatosensory target perturbations compared to visual target
perturbations and that, compared to the before condition, the increase in correction magnitudes
observed in the 200 ms condition was significantly larger for movement to visual targets (1.63 cm
increase) compared to movements to somatosensory targets (0.64 cm increase: see Fig. 26b).
117
!(a)!Direction!constant!error.!Participants!were!more!accurate!when!performing!reaches!to!somatosensory!targets!perturbed!after!movement!onset!compared!to!when!performing!reaches!to!visual!targets.!When!aiming!to!visual!targets!participants!exhibited!a!larger!underBcorrection!relative!to!when!making!movements!to!somatosensory!targets.!(b)!Correction!magnitude.!Participants!exhibited!larger!corrections!in!response!to!somatosensory!target!perturbations!than!in!response!to!visual!target!perturbations!at!all!perturbation!times.!Furthermore,!for!both!modalities,!participants!exhibited! smaller! corrections! in! response! to! perturbations! at! 200!ms! than! in! response! to! the! before! and! 100!ms!perturbation!times.!!!!
118
! Latency!of!Online!Corrections!Following!Target!Perturbation!
The analysis of correction latencies yielded a significant main effect of target modality, F(1,13) =
501.00, p < .001, ηP2 = 0.96, perturbation direction, F(1,13) = 18.11, p < .001, ηP2 = 0.58 and a
target modality by perturbation direction interaction, F(1,13) = 11.62, p <.01, ηP2 = 0.47, HSD =
30 ms. The main effect of target modality revealed that correction latencies in response to
somatosensory target perturbations were significantly shorter (M = 68 ms, SD = 20) than correction
latencies to visual target perturbations (M = 188 ms SD = 46). Breaking down the interaction
revealed that, for visual targets, correction latencies were significantly shorter when the target was
perturbed away from the body (M = 164 ms, SD = 30) as compared to when the target was
perturbed towards the body (M = 213 ms, SD = 46). There were no significant differences in
correction latency between directions for movements to somatosensory targets (see Fig. 27).
A supplementary analysis was performed to investigate if the differences between
modalities in correction latencies could be explained by the differences in perturbation detection
times (see Appendix 3). It was found that the differences in vocal response times between visual
and somatosensory target modalities (M = 34 ms, SD = 61) were largely and significantly lower
than differences between modalities in correction latency (M = 120 ms, SD = 41). Overall, the
results of this analysis revealed that between modality differences in target shift detection could
not fully explain the differences in correction latency observed between the visual and
somatosensory target conditions.
119
!Correction! latencies! in! response! to! target! perturbations.! Overall,! participants! corrected! faster! for!somatosensory! target! perturbations! than! for! visual! target! perturbations.! Furthermore,! for! visual! targets,! correction!latencies!were!longer!in!response!to!targets!perturbed!toward!the!body.!!!
120
! Discussion!
The goal of the present study was to investigate whether the online sensorimotor transformations
for movements to somatosensory targets, performed without vision of the limb, occurred in a visual
or non-visual reference frame. Participants performed reaches to both somatosensory and visual
targets that were either stationary or perturbed either before (~ 450 ms) or after (~ 100 ms or ~200
ms) movement onset. If sensorimotor transformations for the online control of movements to
perturbed somatosensory targets employed a visual reference frame, then higher endpoint errors
and longer correction latencies were expected in response to such perturbations. In contrast to this
hypothesis, participants produced larger corrections and were more accurate when reaching to
perturbed somatosensory targets compared to when reaching to perturbed visual targets. Also,
correction latencies were shorter in response to somatosensory target perturbations than in
response to visual target perturbations that occurred after movement onset. Taken together, these
results provide evidence that non-visual sensorimotor transformations are employed for the online
control of movements to somatosensory targets when reaching with an unseen limb.
Participants were able to implement adjustments in response to somatosensory target
perturbations (average correction latency = 68 ms) more rapidly than in response to visual target
perturbations (average correction latency = 188 ms). These differences in latency were observed
even though the amplitude, the speed and timing of target displacements were the same for both
conditions. Correction latencies in response to the shift of visual target position (e.g., 120 – 300
ms) were in the range of what is typically found in other studies that examined reaching movements
performed without vision of the reaching limb (Day and Lyon 2000; Komilis et al. 1993; Prablanc
and Martin 1992; Reichenbach et al. 2009; Saunders and Knill 2003). Furthermore, the noted
correction latencies were longer than the online visual feedback processing times observed when
vision of the reaching limb is available (e.g., ~100 ms: Carlton 1992; Oostwoud Wijdenes et al.
2013; Zelaznik et al. 1983). These findings suggest that, similar to previous studies, corrections in
response to perturbed visual targets likely involved the online remapping of the unseen reaching
hand position to a visual reference frame.
121
In contrast, correction latencies in response to somatosensory target perturbations were
much shorter than those commonly found when examining movements to visual targets. As
mentioned above (see above Section: 5.1.3.6.5 Latency of Online Corrections), technical
differences between the current study and previous work could account for some of these
discrepancies. However, the much shorter latencies in response to somatosensory target
perturbations as compared to visual target perturbations does provide evidence against the
hypothesis that the online control of reaching movements occur in a common visual reference
frame. For upper-limb reaches to perturbed somatosensory targets, both the unseen target location
and the unseen limb position would have to be remapped onto an extrinsic coordinate system prior
to the initiation of online trajectory amendments. These transformations would likely require more
time (Reichenbach et al. 2009) and result in greater errors (Sarlegna et al., 2009) than
transformations required when reaching to a visual target with the unseen hand position, where
only the latter must be remapped in visual coordinates. Thus, in the present study, it is likely that
online sensorimotor transformations occurred in a non-visual reference frame for planning and
correcting reaches to somatosensory targets.
The use of non-visual sensorimotor transformations for reaches to somatosensory targets
with the unseen hand would support the more rapid and more accurate corrections found in the
present study. For this type of sensorimotor transformation, corrections would be computed in
somatosensory coordinates based on inputs from both the target and the reaching limb (Battaglia-
Mayer et al. 2003; Burnod et al. 1999; Sarlegna and Sainburg 2007). Previous studies investigating
connections of the posterior parietal cortex in macaque monkeys have revealed that neural
networks capable of performing computations in somatosensory coordinates (Prevosto et al.,
2011). Specifically, the medial intraparietal area of the posterior parietal cortex, which is
implicated in sensorimotor transformations during the planning and control of arm movements
(Buneo and Andersen 2006; Desmurget et al. 1999; Reichenbach et al. 2014) was shown to receive
direct projections from both the somatosensory cortex (area 3a) and the dorsal column nuclei
(Prevosto et al., 2010, 2011). It is thus possible that when reaching to a somatosensory target,
updates to target location and reaching limb positions are processed directly through these network
connections. Some support for this hypothesis could be drawn from previous studies which showed
that disrupting processing in medial intraparietal area through transcranial magnetic stimulation
122
impaired somatosensory-based corrections during the control of goal-directed actions
(Reichenbach et al., 2014).
The short correction latencies noted when participants reached to somatosensory targets
may also suggest the use of predictive mechanisms for correcting the movements. The fact that the
target finger was always displaced by 3 cm may have contributed to the speed of movement
corrections. That is, a priori knowledge of the final finger target position might have facilitated
processes responsible for the trajectory corrections and thus reducing the correction latency.
Although the visual target was always similarly displaced by 3 cm, the latencies for correcting the
movements were much longer in the visual target condition. Together, these observations might
suggest that predictive mechanisms are facilitated for controlling goal-directed arm movements
online when both the target and the reaching hand can be encoded in a common sensory modality.
Similar to previous studies (e.g., Saunders and Knill, 2003; Reichenbach et al., 2009) there
was no effect of perturbation time on correction latency for movements to both visual and
somatosensory targets. This finding supports those of previous studies that showed that online
corrections occurred at roughly the same time relative to the onset of visual target perturbations
(i.e., 163 ms after) in response to both early (25% of movement distance) and mid (50% of
movement distance) perturbation conditions (Saunders & Knill, 2003). This result was taken as
evidence for the pseudo-continuous use of visual feedback throughout the reaching trajectory (see
also Elliott et al. 1991). In the present study, our observation that correction latencies were not
significantly altered by perturbation time when reaching to both visual and somatosensory targets
may indicate that somatosensory information can be used in a pseudo-continuous manner at least
in the first 200 ms of a movement lasting at least 500 ms (see also Tremblay et al. 2017).
It is also important to note that the aforesaid differences in correction latencies between
movements to visual and somatosensory targets were not attributable to differences in the detection
of the target displacement. Although participants detected perturbations to somatosensory targets
faster than they detected perturbations to visual targets, the between modality differences in
detection time was much lower than the between modality differences in correction latencies (see
Appendix 3).
123
The response times in both detection tasks (i.e., visual and somatosensory) may appear
longer (i.e., >440 ms) compared to previous studies. However, it is important to note that, in the
present study, participants had to respond only if the visual or somatosensory targets moved toward
or away from them and had to refrain from responding when the targets moved to their left.
Therefore, the detection tasks used here can be considered as the go/no-go tasks which are known
to increase the latency of the go response compared to the response latency obtained through
simple reaction tasks (Miller and Low 2001). Moreover, the longer detection times in the visual
target condition could also be explained by the fact that it takes less time to detect motion onset
than to detect motion direction (Blouin et al. 2010).
In agreement with the correction latency results, the analyses of direction constant errors
revealed that participants were more effective at implementing corrections in response to perturbed
somatosensory targets compared to perturbed visual targets. Participants exhibited larger
corrections when reaching to perturbed somatosensory targets compared to perturbed visual targets
for all perturbation times. Moreover, the constant error analyses revealed that participants were
more accurate when correcting for somatosensory target perturbations as compared to visual target
perturbations after movement onset (i.e., 100 ms and 200 ms conditions). No differences in
accuracy were noted with respect to target modality for target perturbations that occurred before
movement onset.
Similar to previous studies (Goodale et al., 1986; Komilis et al., 1993), the current results
showed that, if the target was perturbed during movement planning, participants were able to fully
correct for changes in target location. In the present study, the time available to implement
trajectory amendments in response to perturbations that occurred before movement onset (> 900
ms) could explain the absence of differences in movement accuracy between target modalities.
In contrast, when target perturbations occurred after movement onset, movement endpoints
were more accurate for reaches to somatosensory targets compared to visual targets. This finding
could have two possible explanations. First, because corrections latencies were ~120 ms shorter in
the somatosensory target condition, participants had more time to implement corrections when
reaching to somatosensory targets than to visual targets. Second, when both the target and the hand
are mapped in the same sensory coordinate system, sensorimotor transformations leading to the
124
movements would be more accurate compared to when a reference frame conversion is necessary
(Blouin et al. 2014; Sarlegna et al. 2009; Sober and Sabes 2005). To investigate whether target
information obtained from somatosensory sources provides a better estimate of new target
position, future studies should examine how disruptions in the accuracy of somatosensory
information from the target limb (e.g., via tendon vibration) affects endpoint accuracy. If more
accurate reference frame transformations are responsible for endpoint accuracy, then disrupting
somatosensory target information should decrease movement accuracy. This decrease in accuracy
would also be independent of the time required to implement corrections.
In the present study, differences were found when targets were perturbed away vs. towards
the body, and that was the case for both the detection and control processes. Participants were
quicker at detecting perturbed somatosensory targets compared to perturbed visual targets, but the
difference between the two detection times was smaller when the targets were perturbed toward
the body. This result could be explained by the physical attributes of the experimental setup. In the
somatosensory target condition, the limb was already in slight extension. Thus, further extension
may be more easily sensed than flexion due to the increased loading of muscle spindles (Hulliger,
1984). A further extension also shifts the limb further away from the body’s centre of gravity likely
invoking a greater postural response that could have been more salient due to greater activation of
vestibular and cutaneous receptors (Lacquaniti & Soechting, 1986). For the visual target, the
physical setup of the experiment could have also played a role. Because of the position of the eyes
and the fixation LED, the angular displacement of the target resulting from the 3 cm perturbations
was greater when the target moved towards than away from the body. Even though all
perturbations took place in the lower visual field, it is possible that this wider change in angle
could have facilitated perturbation detection.
For somatosensory targets, the shorter detection times appeared to have a positive impact
on performance as movements to somatosensory targets perturbed away from the body were
significantly more precise than movements to somatosensory targets perturbed towards the body.
In contrast, the pattern of results in detection times was not consistent with the results obtained for
correction latency for visual targets. For example, it was found that correction latency was
significantly longer for movements to visual targets perturbed towards the body compared to visual
targets perturbed away from the body. These findings are therefore consistent with the hypothesis
125
that detection processes have very little influence on correction processes during ongoing reaching
movements (Smeets, Oostwoud Wijdenes, & Brenner, 2016).
! Conclusions!
The present study demonstrated that, without vision of the reaching limb, movements to perturbed
somatosensory targets are corrected for faster and more accurately than movements to visual
targets. Thus, in contrast to movements to external visual targets, movements to somatosensory
targets are likely controlled using a non-visual transformation process based on somatosensory
information about the reaching limb and target positions. These findings lend support to the idea
that, different sensorimotor transformations, and perhaps different cortical networks are
responsible for the online control of movements to somatosensory and visual targets performed
without vision of the limb.
126
!!General!Discussion!
! Thesis!Findings,!and!Future!Directions!
! Summary!of!Thesis!Findings!The current dissertation provides new knowledge pertaining to the sensorimotor transformation
processes underlying the planning and control of movements to somatosensory targets. Experiment
A1 revealed that endpoint errors for movements to auditory-cued somatosensory targets were more
biased by gaze position than movements to tactile-cued somatosensory targets. Also, shifts in gaze-
position resulted in increases in reaction time for movements to auditory-cued somatosensory
targets relative to both auditory-cued visual targets and tactile-cued somatosensory targets. These
results provide evidence that movements to auditory-cued somatosensory targets were planned in
a visual reference frame, and perhaps required more complex sensorimotor transformations.
Experiment A2 investigated the neural basis of sensorimotor transformations for planning
movements to auditory-cued somatosensory targets versus auditory-cued visual targets. It was
found that there was a facilitation of visual information processing when planning movements to
auditory-cued somatosensory compared to auditory-cued visual targets. These results provide
evidence that cortical visual networks play a role in the sensorimotor transformation processes
responsible for the remapping of movements to somatosensory targets into a visual reference
frame. Study B examined if online corrections for movements to somatosensory targets are
programmed in a visual reference frame. Participants showed larger correction magnitudes, more
accurate movement endpoints, and shorter correction latencies when reaching with an unseen limb
to perturbed somatosensory targets as compared to when reaching to perturbed visual targets.
These results provide evidence that non-visual sensorimotor transformations were used for the
online control of movements to perturbed somatosensory targets. The aforesaid findings were
discussed with for relevance to other experimental studies, clinical work, and theories of motor
control.
127
! GazeB!(in)dependent!Encoding!of!Somatosensory!Targets!are!Influenced!by!the!Modality!of!the!Imperative!Stimulus!
In Experiment A1, gaze-independent coding was observed for movements to somatosensory
targets cued by a vibrotactile stimulus applied to the target position. This result was consistent
with previous studies that showed that if the eyes and somatosensory target positions remain stable
during movement planning, endpoint errors for reaches to vibrotactile-cued somatosensory targets
are not biased by gaze-position (e.g., Mueller & Fiehler, 2014a). Although the mechanistic
explanation for the gaze-independent mapping of somatosensory targets remains speculative,
previous studies suggest gaze-independent coding could be facilitated by the close links between
tactile stimulus processing and the somatotopic maps associated with the interoceptive body
representation (see section 3.2.2).
Studies in primates have found that somatosensory areas in the cortex contain somatotopic
maps of body positions that respond to electrical stimulation of peripheral nerves in the mapped
regions of the body (Kopietz et al., 2009; Sato et al., 2005). Specifically, when tactile stimulation
is applied to different digits, neural activation patterns corresponding to the stimulated digit could
be derived (Sato et al., 2005). Importantly, networks for different digits could be distinguished
even though the stimulated regions were close in proximity. Because these areas contain reciprocal
connections to premotor, motor, and parietal areas, it is possible that tactile stimulation could
promote movement planning using target coordinates derived from the somatotopic map of body
(see Cisek and Kalaska, 2002; Rushworth et al., 1998). Overall, the absence of tactile stimulation
in the auditory-cued conditions may limit access to somatosensory coordinates when planning
voluntary arm movements. In this case, a more cognitive, exteroceptive, representation of the
target and limb positions may be used for movement planning.
The idea that an exteroceptive body representation could be used for planning movements
to auditory-cued somatosensory targets is supported by previous studies showing that movements
to somatosensory targets are planned in a visual reference frame (Blangero et al., 2005; Cohen &
Andersen, 2002; Jones & Henriques, 2010; Pouget et al., 2002). This finding, however, contrasts
with other studies arguing that the reference frame selected for planning movements to
somatosensory targets can be non-visual (Bernier et al., 2009; Bernier & Grafton, 2010; McGuire
& Sabes, 2009; Sober & Sabes, 2003, 2005). Studies examining movements to targets of different
128
modalities have revealed that the sensorimotor transformation processes prior to movements to
somatosensory targets are distinct from the transformation processes used for movements to visual
targets. For example, somatosensory information was found to contribute more to movement errors
when planning movements to somatosensory targets than visual targets (Bernier et al., 2007;
Sarlegna & Sainburg, 2007; Sober & Sabes, 2003, 2005). Also, other studies have shown that the
cortical processes and activation patterns associated with movement planning are also different
for movements to visual and somatosensory targets (Bernier et al., 2009; Bernier & Grafton, 2010;
Blouin et al., 2014). Although the results of Experiment A1 are not necessarily in conflict with
target modality-dependent transformation hypotheses, the results do suggest that these hypotheses
should be modified. The results of the Experiment A1 extend the literature on modality-dependent
processing by providing evidence that factors other than target modality could influence
integration processes prior to movement onset. This hypothesis coincides with a growing body of
literature, suggesting that somatosensory target positions be represented concurrently in multiple
reference frames, and the reference frame selected is based on task-dependent sensory re-
weighting processes (Badde, Röder, & Heed, 2014; Battaglia-Mayer et al., 2003; Liu & Ando,
2018).
Previous studies have found that the reference frame used to identify body positions is
dependent on the goal of the tasks and the sensory information available. One prominent example
is the crossed limb temporal order judgement task (Yamamoto & Kitazawa, 2001). In this task,
participants are asked to judge the relative onset of two tactile stimuli applied to the right and left
limb with their eyes closed. Participants have more difficulty reporting the correct side of the first
stimulation when adopting a crossed limb posture compared to an uncrossed limb posture (Badde
et al., 2014; Noel & Wallace, 2016; Overvliet, Azañon, & Soto-Faraco, 2011; Yamamoto &
Kitazawa, 2001). In contrast to tactile detection tasks, the adoption of a crossed limb posture has
no effect on errors when participants are asked to reach to the location of the tactile stimulus
(Brandes & Heed, 2015). Analysis of reach trajectories reveal however, that trajectories of hand
movements toward tactile locations are corrected later when moving to tactile targets located on
crossed limbs versus uncrossed limbs (Brandes & Heed, 2015). Furthermore, this difference
between crossed and uncrossed occurs only when multiple target locations on different limbs are
used. In the crossed limb posture, the exteroceptive representation of the body is in a spatial
129
conflict with somatosensory senses as the right limb is located on the left side of the body and the
left limb is located on the right side of the body. Thus, for tasks involving the identification of
body positions, or determining the location of stimuli applied to body positions, it is possible that
an exteroceptive, visual representation is used.
In Study A, the participants used auditory information to derive spatial locations of body
positions. Thus, participants had to use auditory information as cue to derive the spatial location
of a body position. Even though participants were capable of accurately completing these
localizations, our results suggest that similar to the crossed-limb judgement and reaching tasks,
body positions were localized in visual coordinates prior to the programming of the movement.
! Activation!of!Cortical!Networks!Associated!with!Visuomotor!Transformations!for!Movements!to!AuditoryBCued!Somatosensory!Targets!
In Experiment A2, there was an increased cortical response to a task-irrelevant visual stimulus
during the planning of movements to auditory-cued somatosensory targets versus auditory-cued
visual targets. The sources of this response were localized to occipital and parietal-occipital areas,
which are typically associated with visuomotor processing and sensorimotor transformations
(Buneo & Andersen, 2006; Cohen & Andersen, 2002; Davare et al., 2012; Reichenbach et al.,
2014; Vesia et al., 2008). This increase in cortical visual information processing indicates that
visual and visuomotor processing networks were recruited when movements to somatosensory
target positions were planned in a visual reference frame. These networks could therefore be
implicated in the sensorimotor transformations responsible for converting somatosensory
information about the unseen hand and target locations into visual coordinates.
In agreement with the results of Experiment A2, previous studies have also implicated the
left occipital and posterior parietal cortices in the visual mapping of somatosensory information.
For example, studies on autotpagnosia have revealed that lesions in the left occipital and parietal
cortices result in the inability to identify the spatial location of body positions (Buxbaum and
Coslett, 2001; Felican et al., 2004; Sirigu et al., 1991b) and studies on movement control have also
found that virtual lesions to posterior parietal areas are associated with the impairments in the
online transformation of somatosensory information into visual coordinates (Reichenbach et al.,
130
2014). The results of Experiment A2 builds on these previous findings by demonstrating that these
networks are also involved visual remapping of somatosensory targets prior to upper-limb reaching
movements. Furthermore, in accordance with Experiment A1 the results of Experiment A2
suggests that alterations in cortical sensory information processing are not exclusively driven by
target modality, but the relevance of the sensory information to sensorimotor transformations.
The assertion that cortical processing of sensory information is influenced by factors
beyond target modality is supported studies examining environmental alterations in sensory cue
relevance. Previous research has demonstrated that the late cortical response to proprioceptive
stimulation (i.e., vibration of the lower limbs) was larger when participants prepared stepping
movements versus when standing still in normal gravity. Importantly, no differences in the cortical
response to somatosensory stimuli between standing and stepping were found in microgravity
(Saradjian, Tremblay, Perrier, Blouin, & Mouchnino, 2013). This facilitation in the cortical
processing of somatosensory information in normal gravity is likely related to equilibrium
constraints when shifting the centre of pressure to unload the stepping leg. The absence of a
facilitation in somatosensory information processing in microgravity could thus be related to the
lack of equilibrium constraints and centre of pressure shifts, as observed by Mouchnino et al.
(1996). Changes in visual sensory information processing have been shown in studies examining
mirror-reversed drawing (Lebar et al., 2015). It was found that accurate performers had enhanced
responses to task-irrelevant visual stimuli during mirror reversed drawing compared to rest. In
mirror reversed drawing, accurate performance is more dependent on visual information about
limb position than somatosensory information (Balslev et al., 2004; Lajoie et al., 1992). Therefore,
the accurate performance of movements would require a greater weighting of visual relative to
somatosensory inputs. Taken together, the results of these studies, Experiment A2, and others (see
Saradjian, 2015 for review), provide evidence that the cortical processing of sensory information
depends not only on target modality, but are flexible. In the case of Study A, our results indicate
the processing of visual information may be enhanced because of its relevance to the processes
required to transform the unseen hand and somatosensory targets locations into visual coordinates.
131
! Fast!Correction!Latencies!Suggest!Control!Based!on!Optimized!Feedback!use!for!Movements!to!Perturbed!Somatosensory!Targets!
The observation that the online control of arm movements to perturbed somatosensory targets is
facilitated by non-visual sensorimotor transformation processes is also a key finding of this
dissertation. Previous studies have argued that online corrections are programmed in a visual
reference frame (e.g., Komilis et al., 1993; Prablanc & Martin, 1992; Reichenbach et al., 2009;
Thompson et al., 2014). One limitation of this previous work is that primarily movements to visual
targets were investigated. If online control were to occur in a visual reference frame for movements
to somatosensory targets, both the target position and reaching limb position would have to be
remapped unto a visual coordinate system prior to the programming of corrections. These
transformation processes would require time, and likely result in larger errors than movements
where no transformation is required (Sarlegna et al., 2009; and Study A). In Study B, it was found
that corrections were indeed initiated faster, and were more accurate when participants corrected
their movements to perturbed somatosensory targets compared to perturbed visual targets. These
results provided evidence that non-visual transformation processes were used for the online control
of reaches to somatosensory targets.
The use of a non-visual reference frame for trajectory corrections would require
computation of a corrective movement based on the updated target location and the unseen limb
position in somatosensory coordinates. Although neural connectivity studies have revealed that
the PPC and premotor areas both contain reciprocal connections and inputs from thalamocortical
neurons, thus relaying dynamic arm information (Prevosto et al., 2010, 2011), there is very little
evidence that these reciprocal networks are employed during online control (Reichenbach et al.,
2014). Also, given the rapid speed of adjustments to somatosensory perturbations in Study B (~70
ms) and others (see Pruszynski & Scott, 2012; Rothwell, Traub, & Marsden, 1980) it is likely that
subcortical (e.g., Nagy, Kruse, Rottmann, Dannenberg, & Hoffmann, 2006; Werner, 1993) and
optimal feedback mechanisms were also employed.
The results of Study B revealed significantly shorter correction latencies in response to
perturbations to somatosensory targets as compared to visual targets. With the methodological
limitations of the computation of correction latencies taken into consideration, somatosensory
correction times were still much lower than what would predicted solely by feedback control
132
mechanisms. Recent advances in optimal control theory (see Scott, 2004) have proposed that rapid
motor responses (such as the long latency stretch reflex) could contribute to upper-limb control
based primarily on somatosensory information. In a review, Pruszynski and Scott (2012) argue
that voluntary control processes may employ rapid motor responses (~50 ms) to adjust to
mechanical perturbations delivered to the limbs. To support this claim the authors showed that
rapid motor responses (“reflex responses”) to mechanical perturbations share features with
voluntary responses in that they can be modified by task instructions (Hammond, 1956), features
of the perturbation (Pruszynski & Scott, 2012), and movement goals (Pruszynski, Kurtzer, & Scott,
2008). Furthermore, the neural circuitries underlying rapid motor responses (such as the long
latency stretch reflex) may be accessible to voluntary feedback networks (Pruszynski & Scott,
2012). A fast feedback system based on the recruitment of rapid motor response networks could
serve as a mechanism for rapid adjustments to somatosensory targets perturbation seen in Study
B.
The fast correction times could also be explained by the involvement of subcortical
movement control networks (Alstermark et al., 2007; Azim et al., 2015; Courjon et al., 2015;
Lünenburger et al., 2008). Previous studies on non-human primates have found that subcortical
neuron populations in the superior colliculus code for arm movements independent of gaze-
position (Werner, 1993; Werner, Dannenberg, & Hoffmann, 1997). Neurons in the superior
colliculus also project to motor networks in the spinal cord providing an anatomical basis for a
subcortical movement control pathway (Gandhi & Katnani, 2011). Electrical stimulation of these
pathways during reaching movements in cats have been shown to elicit gaze-independent
deviations in the trajectory in as little as 38 ms (Courjon et al., 2015). Although much more work
in primates is needed to understand if these networks are involved in voluntary movement control,
the speed of these subcortical networks and the capability to produce fast corrections represent a
possible mechanism to explain corrections to somatosensory target perturbations.
Although further work is required to determine the exact neural substrates used for online
corrections in somatosensory coordinates, the findings of Study B extend previous work by
providing evidence for the existence of a rapid and distinct online control mechanisms for
movements to somatosensory targets.
133
! Future!Directions!and!Perspectives!The conclusions drawn from these thesis experiments serve as a basis for future research, and in
the case of Study B, a new protocol for investigating potential new ideas.
For the study of sensorimotor transformations prior to action, the next step could be to
examine whether the remapping processes associated with the use of exogenous cues are specific
to the auditory modality, or whether other exogenous cue modalities also facilitate the conversion
of somatosensory target positions into a visual reference frame. For example, what would happen
if tactile cues indicating the somatosensory target position were not delivered to the target
position? Would indirect mapping of tactile cues facilitate the use of an exteroceptive
representation? Studies about conflicting body representations (e.g., Brandes & Heed, 2015)
suggest that this could be the case.
With regard to the findings of Study B, the next step would be to investigate whether
providing visual information about initial hand location affects correction times for movements to
visual and somatosensory targets. Previous studies have revealed that movements to visual targets
are more accurate when vision of the limb is provided compared to when it is not (Prablanc et al.,
1979). Following the logic of Study B, if both the limb and target are initially coded in a visual
reference frame then correction times should be shorter and more accurate than when a
sensorimotor transformation is necessary (i.e., with an unseen limb).
134
!!Conclusions!
Everyday goal-directed reaches include movements towards our own body. The planning of these
movements may use a visual representation, depending on the cue employed to identify the
somatosensory target location. In contrast, the control of an ongoing reaching action towards a
displaced somatosensory target can rely on the use of a somatosensory representation that does not
entail remapping onto a visual representation. Together, these results extend previous work done
on the impact of sensory contexts on goal-directed actions and open up new possibilities for further
investigations of these mechanisms in movements to somatosensory targets.
135
References!
Alais, D., Newell, F. N., & Mamassian, P. (2010). Multisensory processing in review: From
physiology to behaviour. Seeing and Perceiving, 23(1), 3–38.
http://doi.org/10.1371/journal.pone.0011283
Alberts, B., Johson, A., & Lewis, J. (2002). Analyzing protein structure and function. In
Molecular Biology of the cell (4th ed.). New York: Garland Science. Retrieved from
https://www.ncbi.nlm.nih.gov/books/NBK26820/
Allison, T., Wood, C. C., & Goff, W. R. (1983). Brain stem auditory, pattern-reversal visual, and
short-latency somatosensory evoked potentials: Latencies in relation to age, sex, and brain
and body size. Electroencephalography and Clinical Neurophysiology, 55(6), 619–636.
http://doi.org/10.1016/0013-4694(83)90272-9
Allport, A. (1987). Selection for action: Some behavioural and neurophysiological
considerations of attention and action. In H. Heuer & A. F. Sanders (Eds.), Perspectives on
Perception and Action (pp. 395–419). Hillsdale, NJ: Lawrence Eribraum Associates.
Ambrosini, E., Ciavarro, M., Pelle, G., Perrucci, M. G., Galati, G., Fattori, P., … Committeri, G.
(2012). Behavioral investigation on the frames of reference involved in visuomotor
transformations during peripheral arm reaching. PLoS ONE, 7(12), 1–8.
http://doi.org/10.1371/journal.pone.0051856
Andersen, R. A., Essick, G. K., & Siegel, R. M. (1985). Encoding of spatial location by posterior
parietal neurons. Science, 230(4724), 456–458. http://doi.org/10.1126/science.4048942
Andersen, R. A., & Mountcastle, V. B. (1983). The influence of the angle of gaze upon the
excitability of the light-sensitive neurons of the posterior parietal cortex. Journal of
Neuroscience, 3(3), 532–548.
Andersen, R. A., Snyder, L. H., Bradley, D., & Xing, J. (1997). Multimodal representation of
space in the posterior parietal cortex and its use in planning movements. Annual Review of
136
Neuroscience, 20(1), 303–330. http://doi.org/10.1146/annurev.neuro.20.1.303
Ashe, J., & Georgopoulos, A. (1994). Movement parameters and neural activity in motor cortex
and area 5. Cerebral Cortex, 4(6), 590–600. http://doi.org/10.1093/cercor/4.6.590
Babb, R. S., Waters, R. S., & Asanuma, H. (1984). Corticocortical connections to the motor
cortex from the posterior parietal lobe (areas 5a, 5b, 7) in the cat demonstrated by the
retrograde axonal transport of horseradish peroxidase. Experimental Brain Research, 54(3),
476–484. http://doi.org/10.1007/BF00235473
Babiloni, F., Babiloni, C., Carducci, F., Fattorini, L., Onorati, P., & Urbano, A. (1996). Spline
Laplacian estimate of EEG potentials over a realistic magnetic resonance-constructed scalp
surface model. Electroencephalography and Clinical Neurophysiology, 98, 363–373.
http://doi.org/10.1016/0013-4694(96)00284-2
Badde, S., Röder, B., & Heed, T. (2014). Multiple spatial representations determine touch
localization on the fingers. Journal of Experimental Psychology. Human Perception and
Performance, 40(2), 784–801. http://doi.org/10.1037/a0034690
Bahill, A. T., Clark, M. R., & Stark, L. (1975). The main sequence, a tool for studying human
eye movements. Mathematical Biosciences, 24(3), 191–204. http://doi.org/10.1016/0025-
5564(75)90075-9
Balslev, D., Christensen, L. O. D., Lee, J.-H., Law, I., Paulson, O. B., & Miall, R. C. (2004).
Enhanced accuracy in novel mirror drawing after repetitive transcranial magnetic
stimulation-induced proprioceptive deafferentation. The Journal of Neuroscience, 24(43),
9698 LP-9702. http://doi.org/10.1523/JNEUROSCI.1738-04.2004
Bard, C., Hay, L., & Fleury, M. (1985). Role of peripheral vision in the directional control of
rapid aiming movements. Canadian Journal of Psychology, 39(1), 151–161.
http://doi.org/10.1037/h0080120
Bard, C., Turrell, Y., Fleury, M., Teasdale, N., Lamarre, Y., & Martin, O. (1999).
Deafferentation and pointing with visual double-step perturbations. Experimental Brain
137
Research, 125(4), 410–416. http://doi.org/10.1007/s002210050697
Barlow, H. B., Blakemore, C., & Pettigrew, J. D. (1967). The neural mechanism of binocular
depth discrimination. Journal of Physiology, 193(2), 327.
Batista, A. P., & Newsome, W. T. (2000). Visuo-motor control: Giving the brain a hand. Current
Biology, 10(4), 145–148. http://doi.org/10.1016/S0960-9822(00)00327-4
Battaglia-Mayer, A., Caminiti, R., Lacquaniti, F., & Zago, M. (2003). Multiple levels of
representation of reaching in the parieto-frontal network. Cerebral Cortex, 13(10), 1009–
1022. http://doi.org/10.1093/cercor/13.10.1009
Becker, W. (1991). Saccades. In Eye Movements (pp. 95–117). London: Macmillan Press.
Bekkering, H., & Neggers, S. F. W. (2002). Visual search is modulated by action intentions.
Psychological Science, 13(4), 370–374. http://doi.org/10.1111/j.0956-7976.2002.00466.x
Berkeley, G. (1709). Essay Towards a New Theory of Vision.
Berlucchi, G., & Aglioti, S. M. (2010). The body in the brain revisited. Experimental Brain
Research, 200(1), 25–35. http://doi.org/10.1007/s00221-009-1970-7
Bernier, P.-M., Burle, B., Hasbroucq, T., & Blouin, J. (2009). Spatio-temporal dynamics of
reach-related neural activity for visual and somatosensory targets. NeuroImage, 47(4),
1767–1777. http://doi.org/10.1016/j.neuroimage.2009.05.028
Bernier, P.-M., Gauthier, G. M., & Blouin, J. (2007). Evidence for distinct, differentially
adaptable sensorimotor transformations for reaches to visual and proprioceptive targets.
Journal of Neurophysiology, 98(3), 1815–1819. http://doi.org/10.1152/jn.00570.2007
Bernier, P.-M., & Grafton, S. T. (2010). Human posterior parietal cortex flexibly determines
reference frames for reaching based on sensory context. Neuron, 68(4), 776–788.
http://doi.org/10.1016/j.neuron.2010.11.002
Beurze, S. M., Van Pelt, S., & Medendorp, W. P. (2006). Behavioral reference frames for
planning human reaching movements. Journal of Neurophysiology, 96(1), 352–362.
138
http://doi.org/10.1152/jn.01362.2005
Blakemore, S. J., Wolpert, D. M., & Frith, C. (2000). Why can’t you tickle yourself?
Neuroreport, 11(11), 11–16. doi: 10.1097/00001756-200008030-00002
Blangero, A., Rossetti, Y., Honoré, J., & Pisella, L. (2005). Influence of gaze direction on
pointing to unseen proprioceptive targets. Advances in Cognitive Psychology, 1(1), 9–16.
http://doi.org/10.2478/v10053-008-0039-7
Blouin, J., Saradjian, A. H., Lebar, N., Guillaume, A., & Mouchnino, L. (2014). Opposed
optimal strategies of weighting somatosensory inputs for planning reaching movements
toward visual and proprioceptive targets. Journal of Neurophysiology, 112(9), 2290–2301.
http://doi.org/10.1152/jn.00857.2013
Blouin, J., Teasdale, N., Bard, C., & Fleury, M. (1993). Directional control of rapid arm
movements: the role of the kinetic visual feedback system. Canadian Journal of
Experimental Psychology /Revue Canadienne de Psychologie Expérimentale, 47(4), 678–
696. http://doi.org/10.1037/h0078869
Bock, O. (1986). Contribution of retinal versus extraretinal signals towards visual localization in
goal-directed movements. Experimental Brain Research, 64(3), 476–482.
http://doi.org/10.1007/BF00340484
Boussaoud, D., Barth, T. M., & Wise, S. P. (1993). Effects of gaze on apparent visual responses
of frontal cortex neurons. Experimental Brain Research, 93(3), 423–434.
Boussaoud, D., & Bremmer, F. (1999). Gaze effects in the cerebral cortex: Reference frames for
space coding and action. Experimental Brain Research, 128(1), 170–180.
http://doi.org/10.1007/s002210050832
Bracci, S., Cavina-Pratesi, C., Ietswaart, M., Caramazza, A., & Peelen, M. V. (2012). Closely
overlapping responses to tools and hands in left lateral occipitotemporal cortex. Journal of
Neurophysiology, 107(5), 1443–1456. http://doi.org/10.1152/jn.00619.2011
Bracci, S., Ietswaart, M., Peelen, M. V., & Cavina-Pratesi, C. (2010). Dissociable neural
139
responses to hands and non-hand body parts in human left extrastriate visual cortex. Journal
of Neurophysiology, 103(6), 3389–3397. http://doi.org/10.1152/jn.00215.2010
Bradshaw, L. A., & Wikswo, J. P. (2001). Spatial filter approach for evaluation of the surface
laplacian of the electroencephalogram and magnetoencephalogram. Annals of Biomedical
Engineering, 29(3), 202–213. http://doi.org/10.1114/1.1352642
Brainard, D. H. (1997). The Psychophysics Toolbox. Spatial Vision, 10(4), 433–436.
http://doi.org/10.1163/156856897X00357
Brandes, J., & Heed, T. (2015). Reach trajectories characterize tactile localization for
sensorimotor decision making. The Journal of Neuroscience, 35(40), 13648 LP-13658.
http://doi.org/10.1523/JNEUROSCI.1873-14.2015
Bremner, L., & Andersen, R. A. (2012). Coding of the reach vector in parietal area 5d. Neuron,
29(2), 997–1003. http://doi.org/10.1016/j.biotechadv.2011.08.021.Secreted
Brenner, E., & Smeets, J. B. J. (1997). Fast responses of the human hand to changes in target
position. Journal of Motor Behavior, 29(4), 297–310.
http://doi.org/10.1080/00222899709600017
Bridgeman, B., Hendry, D., & Stark, L. (1975). Failure to detect displacement of the visual
world during saccadic eye movements. Vision Research, 15(6), 719–722.
http://doi.org/10.1016/0042-6989(75)90290-4
Brodeur, M., Bacon, B. A., Renoult, L., Prevost, M., Lepage, M., & Debruille, J. B. (2008). On
the functional significance of the P1 and N1 effects to illusory figures in the notch mode of
presentation. PLoS ONE, 3(10). http://doi.org/10.1371/journal.pone.0003505
Brown, M., Marmor, M., Vaegan, Zrenner, E., Brigell, M., & Bach, M. (2006). ISCEV standard
for clinical electro-oculography (EOG) (2006). Documenta Ophthalmologica, 113(3), 205–
212. http://doi.org/10.1007/s10633-006-9030-0
Budisavljevic, S., Dell’Acqua, F., & Castiello, U. (2018). Cross-talk connections underlying
dorsal and ventral stream integration during hand actions. Cortex, 103, 224–239.
140
http://doi.org/https://doi.org/10.1016/j.cortex.2018.02.016
Bullock, D., Cisek, P., & Grossberg, S. (1998). Cortical networks for control of voluntary arm
movements under variable force conditions. Cerebral Cortex, 8, 48–62.
http://doi.org/10.1093/cercor/8.1.48
Buneo, C. A., & Andersen, R. A. (2006). The posterior parietal cortex: Sensorimotor interface
for the planning and online control of visually guided movements. Neuropsychologia,
44(13), 2594–2606. http://doi.org/10.1016/j.neuropsychologia.2005.10.011
Buneo, C. A., Batista, A. P., Jarvis, M. R., & Andersen, R. A. (2008). Time-invariant reference
frames for parietal reach activity. Experimental Brain Research, 188(1), 77–89.
http://doi.org/10.1007/s00221-008-1340-x
Burgess, P. R., & Clark, F. J. (1969). Characteristics of knee joint receptors in the cat. Journal of
Physiology, 203, 317–335.
Burnod, Y., Baraduc, P., Battaglia-Mayer, A., Guigon, E., Koechlin, E., Ferraina, S., Caminiti,
R. (1999). Parieto-frontal coding of reaching: an integrated framework. Experimental Brain
Research, 129(3), 325–346. http://doi.org/10.1007/s002210050902
Buxbaum, L. J., & Coslett, H. B. (2001). Specialised structural descriptions for human body
parts: Evidence from autotopagnosia. Cognitive Neuropsychology, 18(4), 289–306.
http://doi.org/10.1080/02643290126172
Buzsáki, G., Anastassiou, C. A., & Koch, C. (2012). The origin of extracellular fields and
currents — EEG, ECoG, LFP and spikes. Nature Reviews Neuroscience, 13(6), 407–420.
http://doi.org/10.1038/nrn3241
Carlton, L. G. (1992). Visual processing time and the control of movement. In L. Proteau, & D.
Elliott. (Eds) Vision and motor control. (pp. 3–31). Oxford, England: North-Holland.
http://doi.org/10.1016/S0166-4115(08)62008-7
Carlton, L. G., & Carlton, M. J. (1987). Response amendment latencies during discrete arm
movements. Journal of Motor Behavior, 19(2), 227–239.
141
http://doi.org/10.1080/00222895.1987.10735409
Celesia, G. G. (1984). Evoked potential techniques in the evaluation of visual function. Journal
of Clinical Neurophysiology, 1(1), 55–76.
Chapman, C. E., Bushnell, M. C., Miron, D., Duncan, G. H., & Lund, J. P. (1987). Sensory
perception during movement in man. Experimental Brain Research, 68(3), 516–524.
http://doi.org/10.1007/BF00249795
Chawla, D., Lumer, E. D., & Friston, K. J. (2000). Relating macroscopic measures of brain
activity to fast, dynamic neuronal interactions. Neural Computation, 12, 2805–2821.
http://doi.org/10.1162/089976600300014737
Chawla, D., Rees, G., & Friston, K. J. (1999). The physiological basis of attentional modulation
in extrastriate visual areas. Nature Neuroscience, 2(7), 671–676.
http://doi.org/10.1038/10230
Chen, J., Reitzen, S. D., Kohlenstein, J. B., & Gardner, E. P. (2009). Neural representation of
hand kinematics during prehension in posterior parietal cortex of the macaque monkey.
Journal of Neurophysiology, 102(6), 3310–3328. http://doi.org/10.1152/jn.90942.2008
Chua, R., & Elliott, D. (1993). Visual regulation of manual aiming. Human Movement Science,
12(4), 365–401. http://doi.org/10.1016/0167-9457(93)90026-L
Chuchland, M. M., Santhanam, G., & Shenoy, K. (2006). Preparatory acitivity in premotor and
motor cortex reflects the speed of the upcoming reach. Journal of Neurophysiology, 96(6),
3130–3146. http://doi.org/10.1152/jn.00307.2006
Cisek, P. (2007). Cortical mechanisms of action selection: the affordance competition
hypothesis. Philosophical Transactions of the Royal Society of London. Series B, Biological
Sciences, 362(1485), 1585–1599. http://doi.org/10.1098/rstb.2007.2054
Cisek, P., Grossberg, S., & Bullock, D. (1998). A cortico-spinal model of reaching and
proprioception under multiple task constraints. Journal of Cognitive Neuroscience, 10(4),
425–444. http://doi.org/10.1162/089892998562852
142
Cisek, P., & Kalaska, J. F. (2005). Neural correlates of reaching decisions in dorsal premotor
cortex: Specification of multiple direction choices and final selection of action. Neuron,
45(5), 801–814. http://doi.org/10.1016/j.neuron.2005.01.027
Clark, V. P., & Hillyard, S. A. (1996). Spatial selective attention affects early extrastriate but not
striate components of the visual evoked potential. Journal of Cognitive Neuroscience, 8(5),
387–402. http://doi.org/10.1162/jocn.1996.8.5.387
Cohen, D. A. D., Prud’homme, M. J., & Kalaska, J. F. (1994). Tactile activity in primate primary
somatosensory cortex during active arm movements: correlation with receptive field
properties. Journal of Neurophysiology, 71(1), 161–72.
http://doi.org/10.1152/jn.1994.71.1.161
Cohen, J. (1992). A power primer. Psychological Bulletin, 112(1), 155–159.
http://doi.org/10.1037/0033-2909.112.1.155
Cohen, L. A. (1961). Role of eye and neck proprioceptive mechanisms in body orientation and
motor coordination. Journal of Neurophysiology, 24, 1–11.
http://doi.org/10.1152/jn.1961.24.1.1
Cohen, Y. E., & Andersen, R. A. (2000). Reaches to sounds encoded in an eye-centered
reference frame. Neuron, 27(3), 647–652. http://doi.org/10.1016/S0896-6273(00)00073-8
Cohen, Y. E., & Andersen, R. A. (2002). A common reference frame for movement plans in the
posterior parietal cortex. Nature Reviews Neuroscience, 3(7), 553–562.
http://doi.org/10.1038/nrn873
Corradi-Dell’Acqua, C., Hesse, M. D., Rumiati, R. I., & Fink, G. R. (2008). Where is a nose with
respect to a foot? The left posterior parietal cortex processes spatial relationships among
body parts. Cerebral Cortex, 18(12), 2879–2890. http://doi.org/10.1093/cercor/bhn046
Corradi-Dell’Acqua, C., & Rumiati, R. I. (2007). What the brain knows about the body: evidence
for dissociable representations. In Santonianni, F. & Sabatano, C. Brain development in
learning environments: Embodied and perceptual advancement (pp. 50–64). Newcastle:
143
Cambridge Scholars Publishing.
Corradi-Dell’Acqua, C., Tomasino, B., & Fink, G. R. (2009). What is the position of an arm
relative to the body? Neural correlates of body schema and body structural description. J
Neurosci, 29(13), 4162–4171. http://doi.org/10.1523/JNEUROSCI.4861-08.2009
Courchesne, E., Hillyard, S. A., & Galambos, R. (1975). Stimulus novelty, task relevance and
the visual evoked potential in man. Electroencephalography and Clinical Neurophysiology,
39(2), 131–143. http://doi.org/10.1016/0013-4694(75)90003-6
Courjon, J.-H., Zénon, A., Clément, G., Urquizar, C., Olivier, E., & Pélisson, D. (2015).
Electrical stimulation of the superior colliculus induces non-topographically organized
perturbation of reaching movements in cats. Frontiers in Systems Neuroscience, 9, 109.
http://doi.org/10.3389/fnsys.2015.00109
Craighero, L., Fadiga, L., Rizzolatti, G., & Umiltà, C. (1999). Action for perception: a motor-
visual attentional effect. Journal of Experimental Psychology-Human Perception and
Performance, 25(6), 1673–1692. http://doi.org/10.1037/0096-1523.25.6.1673
Crawford, J. D., Medendorp, W. P., & Marotta, J. J. (2004). Spatial transformations for eye −
hand coordination. Journal of Neurophysiology, 92(1), 10–19.
http://doi.org/10.1152/jn.00117.2004
Dale, A. M., Liu, A. K., Fischl, B. R., Buckner, R. L., Belliveau, J. W., Lewine, J. D., &
Halgren, E. (2000). Dynamic statistical parametric mapping: Combining fMRI and MEG
for high-resolution imaging of cortical activity. Neuron, 26(1), 55–67.
http://doi.org/10.1016/S0896-6273(00)81138-1
Darling, W. G., Seitz, R. J., Peltier, S., Tellmann, L., & Butler, A. J. (2007). Visual cortex
activation in kinesthetic guidance of reaching. Experimental Brain Research, 179(4), 607–
619. http://doi.org/10.1007/s00221-006-0815-x
Davare, M., Zénon, A., Pourtois, G., Desmurget, M., & Olivier, E. (2012). Role of the medial
part of the intraparietal sulcus in implementing movement direction. Cerebral Cortex,
144
22(6), 1382–1394. http://doi.org/10.1093/cercor/bhr210
Day, B. L., & Lyon, I. N. (2000). Voluntary modification of automatic arm movements evoked
by motion of a visual target. Experimental Brain Research, 130(2), 159–168.
http://doi.org/10.1007/s002219900218
de Grosbois, J., & Tremblay, L. (2016). Quantifying online visuomotor feedback utilization in
the frequency domain. Behavior Research Methods, 48(4), 1653–1666.
http://doi.org/10.3758/s13428-015-0682-0
de Grosbois, J., & Tremblay, L. (2018). Which measures of online control are least sensitive to
offline processes? Motor Control, 22(3), 358–376. http://doi.org/10.1123/mc.2017-0014
de Vignemont, F. (2010). Body schema and body image-pros and cons. Neuropsychologia,
48(3), 669–680. http://doi.org/10.1016/j.neuropsychologia.2009.09.022
Desmurget, M., Epstein, C. M., Turner, R. S., Prablanc, C., Alexander, G. E., & Grafton, S. T.
(1999). Role of the posterior parietal cortex in updating reaching movements to a visual
target. Nature Neuroscience, 2(6), 563–567. http://doi.org/10.1038/9219
Desmurget, M., Pélisson, D., Rossetti, Y., & Prablanc, C. (1998). From eye to hand: Planning
goal-directed movements. Neuroscience & Biobehavioral Reviews, 22(6), 761–788.
http://doi.org/10.1016/S0149-7634(98)00004-9
Disbrow, E., Roberts, T., & Krubitzer, L. (2000). Somatotopic organization of cortical fields in
the lateral sulcus of Homo sapiens: evidence for SII and PV. Journal of Comparative
Neurology, 418(1), 1–21. http://doi.org/10.1002/(SICI)1096-
9861(20000228)418:1<1::AID-CNE1>3.0.CO;2-P
Disbrow, E., Roberts, T., Poeppel, D., & Krubitzer, L. (2001). Evidence for interhemispheric
processing of inputs from the hands in human S2 and PV. Journa of Neurophysiology,
85(5), 2236–2244. http://doi.org/10.1152/jn.2001.85.5.2236
Edin, B. B. (2001). Quantitative analyses of dynamic strain sensitivity in human skin
mechanoreceptors. Journal of Physiology, 65(3), 3233–3243.
145
http://doi.org/10.1152/jn.00628.2004.
Edin, B. B., & Abbs, J. H. (1991). Finger movement responses of cutaneous mechanoreceptors in
the dorsal skin of the human hand. Journal of Neurophysiology, 65(3), 657–70.
http://doi.org/ 10.1152/jn.1991.65.3.657
Elliott, D., Hansen, S., Grierson, L. E. M., Lyons, J., Bennett, S. J., & Hayes, S. J. (2010). Goal-
directed aiming: Two components but multiple processes. Psychological Bulletin, 136(6),
1023–1044. http://doi.org/10.1037/a0020958
Elliott, D., Helsen, W. F., & Chua, R. (2001). A century later: Woodworth’s (1899) two-
component model of goal-directed aiming. Psychological Bulletin, 127(3), 342–357.
http://doi.org/10.1037/0033-2909.127.3.342
Elliott, D., Lyons, J., Hayes, S. J., Burkitt, J. J., Roberts, J. W., Grierson, L. E. M., … Bennett, S.
J. (2017). The multiple process model of goal-directed reaching revisited. Neuroscience &
Biobehavioral Reviews, 72, 95–110.
http://doi.org/https://doi.org/10.1016/j.neubiorev.2016.11.016
Ernst, M. O., & Banks, M. S. (2002). Humans integrate visual and haptic information in a
statistically optimal fashion. Nature, 415(6870), 429–433. http://doi.org/10.1038/415429a
Ernst, M. O., & Bülthoff, H. H. (2004). Merging the senses into a robust percept. Trends in
Cognitive Sciences, 8(4), 162–169. http://doi.org/10.1016/j.tics.2004.02.002
Fabbri, S., Caramazza, A., & Lingnau, A. (2010). Tuning curves for movement direction in the
human visuomotor system. Journal of Neuroscience, 30(40), 13488–13498.
http://doi.org/10.1523/JNEUROSCI.2571-10.2010
Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses using
G*Power 3.1: Tests for correlation and regression analyses. Behavioral Research Methods,
41(4), 1149–60. http://doi.org/10.3758/BRM.41.4.1149
Feldman, A. G., & Levin, M. F. (1996). Grasping cerebellar function depends on our
understanding the principles of sensorimotor integration: The frame of reference hypothesis.
146
Behavioral and Brain Sciences, 19(03), 442–445.
Felician, O., Romaiguère, P., Anton, J.-L. L., Nazarian, B., Roth, M., Poncet, M., … Roll, J.-P.
P. (2004). The role of human left superior parietal lobule in body part localization. Annals
of Neurology, 55(5), 749–751. http://doi.org/10.1002/ana.20109
Fetsch, C. R., Pouget, A., DeAngelis, G. C., & Angelaki, D. E. (2012). Neural correlates of
reliability-based cue weighting during multisensory integration. Nature Neuroscience,
15(1), 146–54. http://doi.org/10.1038/nn.2983
Filimon, F. (2010). Human cortical control of hand movements: parietofrontal networks for
reaching, grasping, and pointing. The Neuroscientist, 16, 388–407.
http://doi.org/10.1177/1073858410375468
Flash, T., & Henis, E. (1991). Arm trajectory modifications during reaching towards visual
targets. Journal of Cognitive Neuroscience, 3(3), 220–230.
http://doi.org/10.1162/jocn.1991.3.3.220
Flash, T., & Hogan, N. (1985). The coordination of arm movements: An experimentally
confirmed mathematical model. Journal of Neuroscience, 5(7), 1688–1703. Retrieved from
http://www.jneurosci.org/content/5/7/1688
Friedman, D. P., Murray, E. A., O’Neill, J. B., & Mishkin, M. (1986). Cortical connections of
the somatosensory fields of the lateral sulcus of macaques: Evidence for a corticolimbic
pathway for touch. Journal of Comparative Neurology, 252(3), 323–347.
http://doi.org/10.1002/cne.902520304
Gandevia, S. C., & McCloskey, D. I. (1976). Joint sense, muscle sense, and their combination as
position sense, measured at the distal interphalangeal joint of the middle finger. Journal of
Physiology, 260, 387–407. http://doi.org/10.1113/jphysiol.1976.sp011521
Gandhi, N. J., & Katnani, H. A. (2011). Motor functions of the superior colliculus. Annual
Review of Neuroscience, 34(1), 205–231. http://doi.org/10.1146/annurev-neuro-061010-
113728
147
Georgopoulos, A. P. (1990). Neurophysiology of Reaching. In M. Jeannerod (Ed.), Attention and
Motor Performance XIII:Motor represntation and control (pp. 227–263). Hillsdale, NJ:
Lawrence Eribraum Associates.
Ghez, C., Gordon, J., Ghilardi, M. F., & Sainburg, R. L. (1995). Contribution of vision and
proprioception to accuracy in limb movements. In M. S. Gazziniga (Ed.), The Cognitive
Neurosciences (pp. 549–564). Cambridge: MIT Press.
Gilbert, C. D. (2013a). Intermediate-level visual processing and visual primitives. In E. Kandel,
J. Schwartz, T. Jessell, S. Siegelbaum, & A. J. Hudspeth (Eds.), Principles of Neural
Science (5th ed., pp. 602–619). New York: McGraw-Hill.
Gilbert, C. D. (2013b). The constructive nature of visual processing. In E. Kandel, J. Schwartz,
T. Jessell, S. Siegelbaum, & A. J. Hudspeth (Eds.), Principles of Neural Science (5th ed.,
pp. 556–576). New York: McGraw-Hill.
Gillman, D., Cohen, Y. E., & Groh, J. M. (2005). Eye-centered, head-centered, and complex
coding of visual and auditory targets in the intraparietal sulcus. Journal of Neurophysiology,
94(4), 2259–2260. http://doi.org/10.1152/jn.00021.2005.
Goodale, M. A., & Milner, A. D. (1992). Separate visual pathways for perception and action.
Trends in Neurosciences, 15(1), 20–25. http://doi.org/10.1016/0166-2236(92)90344-8
Goodale, M. A., Pélisson, D., & Prablanc, C. (1986). Large adjustments in visually guided
reaching do not depend on vision of the hand or perception of target displacement. Nature,
319(30), 402–403.
Goodman, R., Crainic, V. A., Bested, S. R., Wijeyaratnam, D. O., de Grosbois, J., & Tremblay,
L. (2018). Amending ongoing upper-limb reaches: Visual and proprioceptive contributions?
Multisensory Research, 31(5), 455–480. http://doi.org/https://doi.org/10.1163/22134808-
00002615
Goodwin, G. M., McCloskey, D. I., & Matthews, P. B. C. (1972). The persistence of appreciable
kinesthesia after paralysing joint afferents but preserving muscle afferents. Brain Research,
148
37(2), 326–329. http://doi.org/10.1016/0006-8993(72)90679-8
Gramfort, A., Papadopoulo, T., Olivi, E., & Clerc, M. (2010). OpenMEEG: opensource software
for quasistatic bioelectromagnetics. Biomedical Engineering Online, 9(1), 1.
http://doi.org/10.1186/1475-925X-9-45
Green, D. G. (1970). Regional variations in the visual acuity for interference fringes on the
retina. The Journal of Physiology, 207(2), 351–356.
Hacking, I. (1983). Representing and Intervening: Introductory Topics in the Philosophy of
Natural Science (1st ed.). Cambridge: Cambridge University Press.
Hammond, P. H. (1956). The influence of prior instruction to the subject on an apparently
involuntary neuro-muscular response. The Journal of Physiology, 132(1), 17P–18P.
Hansen, S., Tremblay, L., & Elliott, D. (2008). Real-time manipulation of visual displacement
during manual aiming. Human Movement Science, 27(1), 1–11.
http://doi.org/10.1016/j.humov.2007.09.001
Heath, M. (2005). Role of limb and target vision in the online control of memory-guided reaches.
Motor Control, 9(3), 281–311. http://doi.org/ 10.1123/mcj.9.3.281
Held, R., & Hein, A. (1963). Movement-produced stimulation in the development of visually
guided behavior. Journal of Comparative and Physiological Psychology, 56(5), 872–876.
http://doi.org/10.1037/h0040546
Henriques, D. Y. P., Klier, E. M., Smith, M. A., Lowy, D., & Crawford, J. D. (1998). Gaze-
centered remapping of remembered visual space in an open-loop pointing task. Journal of
Neuroscience, 18(4), 1583–1594. http://doi.org/10.1523/JNEUROSCI.18-04-01583.1998
Hinkley, L. B., Krubitzer, L. A., Nagarajan, S. S., & Disbrow, E. (2007). Sensorimotor
integration in S2, PV, and parietal rostroventral areas of the human sylvian fissure. Journal
of Neurophysiology, 97(2), 1288–1297. http://doi.org/10.1152/jn.00733.2006
Hjorth, B. (1975). An online transformation of eeg scalp potentials into orthogonal source
149
derivations. Electroencephalography and Clinical Neurophysiology, 526–530.
Hommel, B. (2004). Event files: Feature binding in and across perception and action. Trends in
Cognitive Sciences, 8(11), 494–500. http://doi.org/10.1016/j.tics.2004.08.007
Hommel, B. (2009). Action control according to TEC (theory of event coding). Psychological
Research, 73(4), 512–26. http://doi.org/10.1007/s00426-009-0234-2
Howarth, C. I., Beggs, W. D. A., & Bowden, J. M. (1971). The relationship between speed and
accuracy of movement aimed at a target. Acta Psychologica, 35, 207–218.
http://doi.org/10.1016/0001-6918(71)90022-9
Hubel, D. H., & Wiesel, T. N. (1962). Receptive fields, binocular interaction and functional
architecture in the cat’s visual cortex. The Journal of Physiology, 160(1), 106–154.
http://doi.org/10.1113/jphysiol.1962.sp006837
Hubel, D. H., & Wiesel, T. N. (1968). Receptive fields and functional architecture of monkey
striate cortex. The Journal of Physiology, 195(1), 215–243.
http://doi.org/10.1113/jphysiol.1968.sp008455
Hulliger, M. (1984). The mammalian muscle spindle and its central control. In Reviews of
Physiology, Biochemistry and Pharmacology (pp. 1–110). Berlin, Heidelberg: Springer.
http://doi.org/10.1007/BFb0027694
Jeannerod, M. (1991). A neurophysiological model for the directional coding of reaching
movements. Brain and Space, 49–69.
Johnson, H., Van Beers, R. J., & Haggard, P. (2002). Action and awareness in pointing tasks.
Experimental Brain Research, 146(4), 451–459. http://doi.org/10.1007/s00221-002-1200-z
Jones, E. G., & Powell, T. P. S. (1968). The ipsilateral cortical connexions of the somatic
sensory areas in the cat. Brain Research, 9(1), 71–94. http://doi.org/0006-8993(68)90258-8
[pii]
Jones, E. G., & Powell, T. P. S. (1969). Connexions of the somatic sensory cortex of the rhesus
150
monkey. Brain, 92(3), 477–502. http://doi.org/10.1093/brain/92.3.477
Jones, S. A. H., Fiehler, K., & Henriques, D. Y. P. (2012). A task-dependent effect of memory
and hand-target on proprioceptive localization. Neuropsychologia, 50(7), 1462–1470.
http://doi.org/10.1016/j.neuropsychologia.2012.02.031
Jones, S. A. H., & Henriques, D. Y. P. (2010). Memory for proprioceptive and multisensory
targets is partially coded relative to gaze. Neuropsychologia, 48(13), 3782–3792.
http://doi.org/10.1016/j.neuropsychologia.2010.10.001
Kaas, J. H., Merzenich, M. M., & Killackey, H. P. (1983). The reorganization of somatosensory
cortex following peripheral nerve damage in adult and developing mammals. Annual
Review of Neuroscience, 6(1), 325–356.
http://doi.org/10.1146/annurev.ne.06.030183.001545
Katznelson, R. (1981). EEG recording, electrode placement, and aspects of generator
localization. In P. Nunez (Ed.), Electric Fields of The Brain (1st ed., pp. 176–213). New
York: Oxford University Press.
Kayser, J., & Tenke, C. E. (2015). On the benefits of using surface Laplacian (current source
density) methodology in electrophysiology. International Journal of Psychophysiology,
97(3), 171–173. http://doi.org/10.1016/j.ijpsycho.2015.06.001
Keele, S. W., & Posner, M. I. (1968). Processing of visual feedback in rapid movements. Journal
of Experimental Psychology, 77(1). http://doi.org/ 10.1037/h0025754
Kennedy, A., Bhattacharjee, A., Hansen, S., Reid, C., & Tremblay, L. (2015). Online vision as a
function of real-time limb velocity: Another case for optimal windows. Journal of Motor
Behavior, 47(6), 465–75. http://doi.org/10.1080/00222895.2015.1012579
Khan, M. A., Franks, I. M., Elliott, D., Grierson, L. E. M., Chua, R., Bernier, P.-M., … Weeks,
D. J. (2006). Inferring online and offline processing of visual feedback in target directed
movements from kinematic data. Neuroscience and Biobehavioral Reviews, 30(8), 1106–
1121. http://doi.org/https://doi.org/10.1016/j.neubiorev.2006.05.002
151
Kim, S., Callier, T., Tabot, G. A., Gaunt, R. A., Tenore, F. V., & Bensmaia, S. J. (2015).
Behavioral assessment of sensitivity to intracortical microstimulation of primate
somatosensory cortex. Proceedings of the National Academy of Sciences, 112(49), 15202–
15207. http://doi.org/10.1073/pnas.1509265112
Kistemaker, D., Van Soest, A., Wong, J., Kurtzer, I., & Gribble, P. (2012). Control of position
and movement is simplified by combined muscle spindle and Golgi tendon organ feedback.
Journal of Neurophysiology, 109(4), 1126–1139. http://doi.org/10.1152/jn.00751.2012
Koessler, L., Maillard, L., Benhadid, A., Vignal, J. P., Felblinger, J., Vespignani, H., & Braun,
M. (2009). Automated cortical projection of EEG sensors: Anatomical correlation via the
international 10-10 system. NeuroImage, 46(1), 64–72.
http://doi.org/10.1016/j.neuroimage.2009.02.006
Komilis, E., Pélisson, D., & Prablanc, C. (1993). Error processing in pointing at randomly
feedback-induced double-step stimuli. Journal of Motor Behavior, 25(4), 299–308.
http://doi.org/10.1080/00222895.1993.9941651
Kopietz, R., Sakar, V., Albrecht, J., Kleemann, A., Schöpf, V., Yousry, I., … Wiesmann, M.
(2009). Activation of primary and secondary somatosensory regions following tactile
ttimulation of the face. Clinical Neuroradiology, 19(2), 135–144.
http://doi.org/10.1007/s00062-009-8022-3
Lacquaniti, F., & Soechting, J. F. (1986). EMG responses to load perturbations of the upper
limb: effect of dynamic coupling between shoulder and elbow motion. Experimental Brain
Research, 61(3), 482–496. http://doi.org/10.1007/BF00237573
Lajoie, Y., Paillard, J., Teasdale, N., Bard, C., Fleury, M., Forget, R., & Lamarre, Y. (1992).
Mirror drawing in a deafferented patient and normal subjects: visuoproprioceptive conflict.
Neurology, 42(5), 1104–1106.
Langren, S., Silfventus, H., & Wolsk, D. (1967). Somato-sensory paths to the second cortical
projection area of group I muscle afferents. Journal of Physiology, 191, 543–559.
152
Lashley, K. S. (1917). The accuracy of movement in the absence of excitation from the moving
organ. American Journal of Physiology, 43, 169–194.
Lashley, K. S. (1951). The problem of serial order behaviour. In L. Jeffress (Ed.), Cerebral
Mechanisms in Behaviour (pp. 112–136). New York: Hafner Publishing Co
Law, S. K., Rohrbaugh, J. W., Adams, C. M., & Eckardt, M. J. (1993). Improving spatial and
temporal resolution in evoked EEG responses using surface Laplacians.
Electroencephalography and Clinical Neurophysiology - Evoked Potentials, 88, 309–322.
http://doi.org/10.1016/0168-5597(93)90055-T
Lebar, N., Bernier, P.-M., Guillaume, A., Mouchnino, L., & Blouin, J. (2015). Neural correlates
for task-relevant facilitation of visual inputs during visually-guided hand movements.
NeuroImage, 121, 39–50. http://doi.org/10.1016/j.neuroimage.2015.07.033
Lee, R. G., & Tatton, W. G. (1975). Motor responses to sudden limb displacements in primates
with specific CNS lesions and in human patients with motor system disorders. The
Canadian Journal of Neurological Sciences., 2(3), 285–293.
http://doi.org/10.1017/s0317167100020382
Liu, J., & Ando, H. (2018). Response Modality vs. Target Modality: Sensory Transformations
and Comparisons in Cross-modal Slant Matching Tasks. Scientific Reports, 8(1), 11068.
http://doi.org/10.1038/s41598-018-29375-w
Lünenburger, L., Kutz, D. F., & Hoffmann, K.-P. (2008). Influence of arm movements on
saccades in humans. European Journal of Neuroscience, 12(11), 4107–4116.
http://doi.org/10.1046/j.1460-9568.2000.00298.x
Manson, G. A., Alekhina, M., Srubiski, S. L., Williams, C. K., Bhattacharjee, A., & Tremblay,
L. (2014). Effects of robotic guidance on sensorimotor control: Planning vs. online control?
NeuroRehabilitation, 35(4), 689–700. http://doi.org/10.3233/NRE-141168
Maunsell, J. H., & Van Essen, D. C. (1983). Functional properties of neurons in middle temporal
visual area of the macaque monkey. I. Selectivity for stimulus direction, speed, and
153
orientation. Journal of Neurophysiology, 49(5), 1127–1147.
McGuire, L. M. M., & Sabes, P. N. (2009). Sensory transformations and the use of multiple
reference frames for reach planning. Nature Neuroscience, 12(8), 1056–1061.
http://doi.org/10.1038/nn.2357
Medendorp, W. P., Beurze, S. M., Van Pelt, S., & Van Der Werf, J. (2008). Behavioral and
cortical mechanisms for spatial coding and action planning. Cortex, 44(5), 587–597.
http://doi.org/10.1016/j.cortex.2007.06.001
Medendorp, W. P., & Crawford, J. D. (2002). Visuospatial updating of reaching targets in near
and far space. Neuroreport, 13(5), 633–636. http://doi.org/10.1097/00001756-200204160-
00019
Meister, M., & Tessier, L. (2013). Low-level visual procesing in the retina. In E. Kandel, J.
Schwartz, T. Jessell, S. Siegelbaum, & A. J. Hudspeth (Eds.), Principles of Neural Science
(5th ed., pp. 577–600). New York: McGraw-Hill.
Michel, C. M., Murray, M. M., Lantz, G., Gonzalez, S., Spinelli, L., & Grave De Peralta, R.
(2004). EEG source imaging. Clinical Neurophysiology, 115(10), 2195–2222.
http://doi.org/10.1016/j.clinph.2004.06.001
Mishkin, M., Ungerleider, L. G., & Macko, K. A. (1983). Object vision and spatial vision: two
cortical pathways. Trends in Neuroscience, 6, 414–417. http://doi.org/10.1016/0166-
2236(83)90190-X
Mountcastle, V. B., Lynch, J. C., Georgopoulos, A., Sakata, H., & Acuna, C. (1975). Posterior
parietal association cortex of the monkey: command functions for operations within
extrapersonal space. Journal of Neurophysiology, 38(4), 871–908.
http://doi.org/10.1152/jn.1975.38.4.871
Mueller, S., & Fiehler, K. (2014a). Effector movement triggers gaze-dependent spatial coding of
tactile and proprioceptive-tactile reach targets. Neuropsychologia, 62(1), 184–193.
http://doi.org/10.1016/j.neuropsychologia.2014.07.025
154
Mueller, S., & Fiehler, K. (2014b). Gaze-dependent spatial updating of tactile targets in a
localization task. Frontiers in Psychology, 5, 66–66.
http://doi.org/10.3389/fpsyg.2014.00066
Mueller, S., & Fiehler, K. (2016). Mixed body-and gazed-centered coding of proprioceptive
reach targets after effector movement. Neuropsychologia, 87(1), 63–73.
http://doi.org/10.1016/j.neuropsychologia.2014.07.025
Murray, M. M., Thelen, A., Thut, G., Romei, V., Martuzzi, R., & Matusz, P. J. (2016). The
multisensory function of the human primary visual cortex. Neuropsychologia, 83, 161–169.
http://doi.org/10.1016/j.neuropsychologia.2015.08.011
Mushiake, H., Tanatsugu, Y., & Tanji, J. (1997). Neuronal activity in the ventral part of
premotor cortex during target-reach movement is modulated by direction of gaze. Journal
of Neurophysiology, 78(1), 567–571.
Nagy, A., Kruse, W., Rottmann, S., Dannenberg, S., & Hoffmann, K.-P. (2006). Somatosensory-
motor neuronal activity in the superior colliculus of the primate. Neuron, 52(3), 525–534.
http://doi.org/https://doi.org/10.1016/j.neuron.2006.08.010
Neggers, S. F. W., & Bekkering, H. (2001). Gaze anchoring to a pointing target is present during
the entire pointing movement and is driven by a non-visual signal. Journal of
Neurophysiology, 86(2), 961–970. http://doi.org/10.1152/jn.2001.86.2.961
Nerlich, B., & Clarke, D. D. (1996). Language, Action and Context: The Early History of
Pragmatics in Europe and America (Vol. 80). John Benjamins Publishing.
Noback, C. R., Ruggiero, D. A., Demarest, R. J., & Strominger, N. L. (2005). The Human
Nervous System!: Structure and Function. Totowa: Humana Press.
Noel, J.-P., & Wallace, M. (2016). Relative contributions of visual and auditory spatial
representations to tactile localization. Neuropsychologia, 82, 84–90.
http://doi.org/10.1016/j.neuropsychologia.2016.01.005
Nunez, P. L., Silberstein, R. B., Cadusch, P. J., Wijesinghe, R. S., Westdorp, A. F., & Srinivasan,
155
R. (1994). A theoretical and experimental-study of high-resolution EEG based on surface
laplacians and cortical imaging. Electroencephalography and Clinical Neurophysiology,
90(1), 40–57.
Ogden, J. A. (1985). Autotopagnosia occurence in a patient without nominal aphasia and with an
intact ability to point to animals and objects. Brain, 108(4), 1009–1022.
http://doi.org/10.1093/brain/108.4.1009
Olsen, C. W., & Ruby, C. (1941). Anosognosia and autotopagnosia. Archives of Neurology and
Psychiatry, 46(2), 340–344. http://doi.org/10.1001/archneurpsyc.1941.02280200146008
Oostwoud Wijdenes, L., Brenner, E., & Smeets, J. B. J. (2011). Fast and fine-tuned corrections
when the target of a hand movement is displaced. Experimental Brain Research, 214(3),
453–462. http://doi.org/10.1007/s00221-011-2843-4
Oostwoud Wijdenes, L., Brenner, E., & Smeets, J. B. J. (2013). Comparing online adjustments to
distance and direction in fast pointing movements. Journal of Motor Behavior, 45(5), 395–
404. http://doi.org/10.1080/00222895.2013.815150
Oostwoud Wijdenes, L., Brenner, E., & Smeets, J. B. J. (2014). Analysis of methods to
determine the latency of online movement adjustments. Behavior Research Methods, 46(1),
131–139. http://doi.org/10.3758/s13428-013-0349-7
Osterberg, G. (1935). Topography of the layer of rods and cones in the human retina. Acta
Ophthalmologica, (13), 1–102.
Overvliet, K. E., Azañon, E., & Soto-Faraco, S. (2011). Somatosensory saccades reveal the
timing of tactile spatial remapping. Neuropsychologia, 49(11), 3046–3052.
http://doi.org/10.1016/j.neuropsychologia.2011.07.005
Paillard, J. (1980). Le corps situé et le corps identifié. Rev. Méd. Suisse Romande, 100(2), 129–
141.
Paillard, J. (1991). Motor and representational framing of space. In J. Paillard (Ed.), Brain and
space (pp. 163-182). New York: Oxford University Press.
156
Pélisson, D., Prablanc, C., Goodale, M. A., & Jeannerod, M. (1986). Visual control of reaching
movements without vision of the limb - II. Evidence of fast unconscious processes
correcting the trajectory of the hand to the final position of a double-step stimulus.
Experimental Brain Research, 62(2), 303–311. http://doi.org/10.1007/BF00238849
Perrin, F., Bertrand, O., & Pernier, J. (1987). Scalp current density mapping: value and
estimation from potential data. IEEE Transactions on Bio-Medical Engineering, 34(4),
283–288. http://doi.org/10.1109/TBME.1987.326089
Perrin, F., Pernier, J., Bertrand, O., & Echallier, J. F. (1989). Spherical splines for scalp potential
and current density mapping. Electroencephalography and Clinical Neurophysiology, 72,
184–187. http://doi.org/10.1016/0013-4694(89)90180-6
Phurailatpam, J. (2014). Evoked potentials: Visual evoked potentials (VEPs): Clinical uses,
origin, and confounding parameters. Journal of Medical Society, 28(3), 140–144.
http://doi.org/10.4103/0972-4958.148494
Pick, A. (1922). Störung der Orientierung am eigenen Körper. Pyschologische Forschung,
1(303), 318.
Pouget, A., Ducom, J. C., Torri, J., & Bavelier, D. (2002). Multisensory spatial representations in
eye-centered coordinates for reaching. Cognition, 83(1), 1–11.
http://doi.org/10.1016/S0010-0277(01)00163-9
Prablanc, C., Desmurget, M., & Gréa, H. (2003). Neural control of on-line guidance of hand
reaching movements. In C. Prablanc, D. Pélisson, & Y. Rossetti (Eds.), Progress in Brain
Research (Vol. 142, pp. 155–170). Elsevier. https://doi.org/10.1016/S0079-6123(03)42012-
8
Prablanc, C., Echallier, J. E., Jeannerod, M., & Komilis, E. (1979). Optimal response of eye and
hand motor systems in pointing at a visual target. II. Static and dynamic visual cues in the
control of hand movement. Biological Cybernetics, 35(3), 183–187.
http://doi.org/10.1007/BF00337436
157
Prablanc, C., & Martin, O. (1992). Automatic control during hand reaching at undetected two-
dimensional target displacements. Journal of Neurophysiology, 67(2), 455–69.
http://doi.org/10.1152/jn.1992.67.2.455
Prevosto, V., Graf, W., & Ugolini, G. (2010). Cerebellar inputs to intraparietal cortex areas LIP
and MIP: Functional frameworks for adaptive control of eye movements, reaching, and
arm/eye/head movement coordination. Cerebral Cortex, 20(1), 214–228.
http://doi.org/10.1093/cercor/bhp091
Prevosto, V., Graf, W., & Ugolini, G. (2011). Proprioceptive pathways to posterior parietal areas
MIP and LIPv from the dorsal column nuclei and the postcentral somatosensory cortex.
European Journal of Neuroscience, 33(3), 444–460. http://doi.org/10.1111/j.1460-
9568.2010.07541.x
Prinz, W. (1997). Perception and action planning. European Journal of Cognitive Psychology,
9(2), 129–154. http://doi.org/10.1080/713752551
Pritchett, L. M., Carnevale, M. J., & Harris, L. R. (2012). Reference frames for coding touch
location depend on the task. Experimental Brain Research, 222(4), 437–45.
http://doi.org/10.1007/s00221-012-3231-4
Proske, U., & Gandevia, S. C. (2012). The proprioceptive senses: Their roles in signaling body
shape, Body position and movement, and muscle force. Physiology Reviews, 92(4), 1651–
1697. http://doi.org/10.1152/physrev.00048.2011
Pruszynski, J. A., Kurtzer, I., & Scott, S. H. (2008). Rapid motor responses are appropriately
tuned to the metrics of a visuospatial task. Journal of Neurophysiology, 100(1), 224–238.
http://doi.org/10.1152/jn.90262.2008
Pruszynski, J. A., & Scott, S. H. (2012). Optimal feedback control and the long-latency stretch
response. Experimental Brain Research, 218(3), 341–359. http://doi.org/10.1007/s00221-
012-3041-8
Purves, D., Augustine, G. J., Fitzpatrick, D., Katz, L. C., LaMantia, A.-S., McNamara, J. O., &
158
Williams, S. M. (2001). Anatomical Distribution of Rods and Cones.
https://www.ncbi.nlm.nih.gov/books/NBK10848/
Redon, C., Hay, L., & Velay, J. L. (1991). Proprioceptive control of goal-directed movements in
man, studied by means of vibratory muscle tendon stimulation. Journal of Motor Behavior,
23(2), 101–108. http://doi.org/10.1080/00222895.1991.9942027
Reichenbach, A., Thielscher, A., Peer, A., Bülthoff, H. H., & Bresciani, J.-P. (2009). Seeing the
hand while reaching speeds up on-line responses to a sudden change in target position. The
Journal of Physiology, 587(19), 4605–4616. http://doi.org/10.1113/jphysiol.2009.176362
Reichenbach, A., Thielscher, A., Peer, A., Bülthoff, H. H., & Bresciani, J.-P. (2014). A key
region in the human parietal cortex for processing proprioceptive hand feedback during
reaching movements. NeuroImage, 84, 615–625.
http://doi.org/10.1016/j.neuroimage.2013.09.024
Rosenbaum, D. A., Cohen, R. G., Jax, S. A., Weeks, D. J., & van der Wel, R. (2007). The
problem of serial order in behavior: Lashley’s legacy. Human Movement Science, 26(4),
525–554. http://doi.org/10.1016/j.humov.2007.04.001
Rothwell, J. C., Traub, M. M., & Marsden, C. D. (1980). Influence of voluntary intent on the
human long-latency stretch reflex. Nature, 286, 496. http://doi.org/10.1038/286496a0
Rozzi, S., Ferrari, P. F., Bonini, L., Rizzolatti, G., & Fogassi, L. (2008). Functional organization
of inferior parietal lobule convexity in the macaque monkey: Electrophysiological
characterization of motor, sensory and mirror responses and their correlation with
cytoarchitectonic areas. European Journal of Neuroscience, 28, 1569–1588.
http://doi.org/10.1111/j.1460-9568.2008.06395.x
Saradjian, A. H. (2015). Sensory modulation of movement, posture and locomotion. Clinical
Neurophysiology, 45(4), 255–267. http://doi.org/10.1016/j.neucli.2015.09.004
Saradjian, A. H., Tremblay, L., Perrier, J., Blouin, J., & Mouchnino, L. (2013). Cortical
facilitation of proprioceptive inputs related to gravitational balance constraints during step
159
preparation. Journal of Neurophysiology, 110(2), 397–407.
http://doi.org/10.1152/jn.00905.2012
Sarlegna, F. R., & Bernier, P.-M. (2010). On the link between sensorimotor adaptation and
sensory recalibration. Journal of Neuroscience, 30(35), 11555–11557.
http://doi.org/10.1523/JNEUROSCI.3040-10.2010
Sarlegna, F. R., & Blouin, J. (2010). Visual guidance of arm reaching: Online adjustments of
movement direction are impaired by amplitude control. Journal of Vision, 10(5), 24.
http://doi.org/10.1167/10.5.24
Sarlegna, F. R., & Mutha, P. K. (2015). The influence of visual target information on the online
control of movements. Vision Research, 110, 144–154.
http://doi.org/10.1016/j.visres.2014.07.001
Sarlegna, F. R., Przybyla, A., & Sainburg, R. L. (2009). The influence of target sensory modality
on motor planning may reflect errors in sensori-motor transformations. Journal of
Neuroscience, 164(2), 597–610. http://doi.org/10.1016/j.biotechadv.2011.08.021.Secreted
Sarlegna, F. R., & Sainburg, R. L. (2007). The effect of target modality on visual and
proprioceptive contributions to the control of movement distance. Experimental Brain
Research, 176(2), 267–280. http://doi.org/10.1007/s00221-006-0613-5
Sato, K., Nariai, T., Tanaka, Y., Maehara, T., Miyakawa, N., Sasaki, S., … Ohno, K. (2005).
Functional representation of the finger and face in the human somatosensory cortex:
Intraoperative intrinsic optical imaging. NeuroImage, 25(4), 1292–1301.
http://doi.org/10.1016/j.neuroimage.2004.12.049
Saunders, J. A., & Knill, D. C. (2003). Humans use continuous visual feedback from the hand to
control fast reaching movements. Experimental Brain Research, 152(3), 341–352.
http://doi.org/10.1007/s00221-003-1525-2
Schimmack, U. (2012). The ironic effect of significant results on the credibility of multiple-study
articles. Psychological Methods, 17(4), 551–566. http://doi.org/10.1037/a0029487
160
Schmahmann, J. D., & Pandya, D. N. (1990). Anatomical investigation of projections from
thalamus to posterior parietal cortex in the rhesus monkey: A WGA-HRP and fluorescent
tracer study. The Journal of Comparative Neurology, 295(2), 299–326.
http://doi.org/10.1002/cne.902950212
Schmidt, R. A., Sherwood, D. E., Zelaznik, H. N., & Leikind, B. J. (1985). Speed-accuracy
trade-offs in motor behavior: Theories of impulse variability. In Motor Behavior (pp. 79–
123). Springer.
Schütz, I., Henriques, D. Y. P., & Fiehler, K. (2013). Gaze-centered spatial updating in delayed
reaching even in the presence of landmarks. Vision Research, 87, 46–52.
http://doi.org/10.1016/j.visres.2013.06.001
Schwoebel, J., Coslett, H. B., & Buxbaum, L. J. (2001). Compensatory coding of body part
location in autotopagnosia: Evidence for extrinsic egocentric coding. Cognitive
Neuropsychology, 18(4), 363–381. http://doi.org/10.1080/02643290126218
Scott, S. H. (2004). Optimal feedback control and the neural basis of volitional motor control.
Nature Reviews Neuroscience, 5(7), 532–545. http://doi.org/10.1038/nrn1427
Serino, A., & Haggard, P. (2010). Touch and the body. Neuroscience and Biobehavioral
Reviews, 34(2), 224–236. http://doi.org/10.1016/j.neubiorev.2009.04.004
Shams, L., Ma, W. J., & Beierholm, U. (2005). Sound-induced flash illusion as an optimal
percept. Neuroreport, 16(17), 1923–1927.
http://doi.org/10.1097/01.wnr.0000187634.68504.bb
Sherrington, C. S. (1911). The integrative action of the nervous system.
Shlaer, S. (1937). The relation between visual acuity and illumination. The Journal of General
Physiology, 21(2), 165–188. http://doi.org/10.1085/jgp.21.2.165
Sirigu, A., Grafman, J., Bressler, K., & Sunderland, T. (1991b). Multiple Representations
Contribute To Body Knowledge Processing. Brain, 114(1), 629–642.
http://doi.org/10.1093/brain/114.1.629
161
Smeets, J. B. J., Erkelens, C. J., & van der Gon Denier, J. J. (1990). Adjustments of fast goal-
directed movements in response to an unexpected inertial load. Experimental Brain
Research, 81(2), 303–312. http://doi.org/10.1007/BF00228120
Smeets, J. B. J., Oostwoud Wijdenes, L., & Brenner, E. (2016). Movement Adjustments Have
Short Latencies because There Is No Need to Detect Anything. Motor Control, 20(2), 137–
148. http://doi.org/10.1123/mc.2014-0064
Sober, S. J., & Sabes, P. N. (2003). Multisensory Integration during Motor Planning. The
Journal of Neuroscience, 23(18), 6982 LP-6992. http://doi.org/10.1523/JNEUROSCI.23-
18-06982.2003
Sober, S. J., & Sabes, P. N. (2005). Flexible strategies for sensory integration during motor
planning. Nature Neuroscience, 8(4), 490–497. http://doi.org/10.1038/nn1427
Song, J. H., & Nakayama, K. (2009). Hidden cognitive states revealed in choice reaching tasks.
Trends in Cognitive Sciences, 13, 360–366. http://doi.org/10.1016/j.tics.2009.04.009
Stetson, C., & Andersen, R. A. (2014). The parietal reach region selectively anti-synchronizes
with dorsal premotor cortex during planning, 34(36), 11948–11958.
http://doi.org/10.1523/JNEUROSCI.0097-14.2014
Stricanne, B., Andersen, R. A., & Mazzoni, P. (1996). Eye-centered, head-centered, and
intermediate coding of remembered sound locations in area LIP. Journal of
Neurophysiology, 76(3), 2071–2076. http://doi.org/10.1021/je9001366
Tadel, F., Baillet, S., Mosher, J. C., Pantazis, D., & Leahy, R. M. (2011). Brainstorm: A user-
friendly application for MEG/EEG analysis. Computational Intelligence and Neuroscience,
2011. http://doi.org/10.1155/2011/879716
Tenke, C. E., & Kayser, J. (2012a). Generator localization by current source density (CSD):
Implications of volume conduction and field closure at intracranial and scalp resolutions.
Clinical Neurophysiology, 123(12), 2328–2345. http://doi.org/10.1016/j.clinph.2012.06.005
Tenke, C. E., & Kayser, J. (2012b). Generator localization by current source density (CSD):
162
Implications of volume conduction and field closure at intracranial and scalp resolutions.
Clinical Neurophysiology, 123(12), 2328–2345. http://doi.org/10.1016/j.clinph.2012.06.005
Thompson, A. A., Byrne, P. A., & Henriques, D. Y. P. (2014). Visual targets aren’t irreversibly
converted to motor coordinates: Eye-centered updating of visuospatial memory in online
reach control. PLoS ONE, 9(3), 1–10. http://doi.org/10.1371/journal.pone.0092455
Thompson, A. A., Glover, C. V., & Henriques, D. Y. P. (2012). Allocentrically implied target
locations are updated in an eye-centred reference frame. Neuroscience Letters, 514(2), 214–
218. http://doi.org/10.1016/j.neulet.2012.03.004
Tipper, S. P., Lortie, C., & Baylis, G. (1992). Selective reaching: Evidence for action centered
attention. Journal of Experimental Psychology: Human Perception and Performance, 18(4),
891–905.
Tremblay, L., Crainic, V. A., de Grosbois, J., Bhattacharjee, A., Kennedy, A., Hansen, S., &
Welsh, T. N. (2017). An optimal velocity for online limb-target regulation processes?
Experimental Brain Research, 235(1), 29–40. http://doi.org/10.1007/s00221-016-4770-x
Tremblay, L., Hansen, S., Kennedy, A., & Cheng, D. T. (2013). The utility of vision during
action: Multiple visuomotor processes? Journal of Motor Behavior, 45(2), 91–99.
http://doi.org/10.1080/00222895.2012.747483
Treue, S. (2001). Neural correlated of attention in the primate visual cortex. Trends in
Neurosciences, 24(5), 295–300.
van den Broek, S. ., Reinders, F., Donderwinkel, M., & Peters, M. (1998). Volume conduction
effects in EEG and MEG. Electroencephalography and Clinical Neurophysiology, 106(6),
522–534. http://doi.org/10.1016/S0013-4694(97)00147-8
Veerman, M. M., Brenner, E., & Smeets, J. B. J. (2008). The latency for correcting a movement
depends on the visual attribute that defines the target. Experimental Brain Research, 187(2),
219–228. http://doi.org/10.1007/s00221-008-1296-x
Vesia, M., Yan, X., Henriques, D. Y. P., Sergio, L. E., & Crawford, J. D. (2008). Transcranial
163
magnetic stimulation over human dorsal-lateral posterior parietal cortex disrupts integration
of hand position signals into the reach plan. Journal of Neurophysiology, 100(4), 2005–
2014. http://doi.org/10.1152/jn.90519.2008
Vetter, P., Goodbody, S. J., & Wolpert, D. M. (1999). Evidence for an eye-centered spherical
representation of the visuomotor map. Journal of Neurophysiology, 81(2), 935–939.
Welsh, T. N. (2011). The relationship between attentional capture and deviations in movement
trajectories in a selective reaching task. Acta Psychologica, 137(3), 300–308.
http://doi.org/10.1016/j.actpsy.2011.03.011
Welsh, T. N., Elliott, D., & Weeks, D. J. (1999). Hand deviations toward distractors. Evidence
for response competition. Experimental Brain Research, 127(2), 207–212.
http://doi.org/10.1007/s002210050790
Werner, W. (1993). Neurons in the primate superior colliculus are active before and during arm
movements to visual targets. European Journal of Neuroscience, 5(4), 335–340.
http://doi.org/10.1111/j.1460-9568.1993.tb00501.x
Werner, W., Dannenberg, S., & Hoffmann, K.-P. (1997). Arm-movement-related neurons in the
primate superior colliculus and underlying reticular formation: comparison of neuronal
activity with EMGs of muscles of the shoulder, arm and trunk during reaching.
Experimental Brain Research, 115(2), 191–205. http://doi.org/10.1007/PL00005690
Wienbar, S., & Schwartz, G. W. (2018). The dynamic receptive fields of retinal ganglion cells.
Progress in Retinal and Eye Research. http://doi.org/10.1016/J.PRETEYERES.2018.06.003
Wolpert, D. M., Goodbody, S. J., & Husain, M. (1998). Maintaining internal representations: the
role of the human superior\n parietal lobe. Nature Neuroscience, 1(6), 529–533.
http://doi.org/10.1038/2245
Woodworth, R. S. (1899). Accuracy of voluntary movement. Psychological Review
Monographs.
Yamamoto, S., & Kitazawa, S. (2001). Reversal of subjective temporal order due to arm
164
crossing. Nature Neuroscience, 4(7), 759–765. http://doi.org/10.1038/89559
Zelaznik, H. N., Hawkins, B., & Kisselburgh, L. (1983). Rapid visual feedback processing in
single-aiming movements, 15(3), 217–236.
http://doi.org/10.1080/00222895.1983.10735298
Zipser, D., & Andersen, R. A. (1988). A back-propagation programmed network that simulates
response properties of a subset of posterior parietal neurons. Nature, 331, 679–684.
http://doi.org/10.1038/331679a0
165
Appendix!1:!Power!Analyses!for!Experimental!Studies!
! Power!Analyses!for!Experimental!Studies!
! Study!A:!
! Experiment!A1!
After designing the experiment, a power analysis was conducted to determine what sample size
would be required for the statistical tests. The population effect size was estimated by calculating
the average and median effect sizes (see Schimmack 2012) of horizontal reaching error differences
in relevant experiments (see Table 5). Both the median and mean effect sizes were around 0.65.
To estimate the correlation among the proposed repeated-measures (i.e., sensory condition and
fixation direction), the relationship between different target modality experiments in Pouget et al.,
2002 was relied upon. Overall, similar to Pouget et al. (2002), a high positive correlation between
modalities and gaze directions (i.e., high negative correlation for gaze directions) was expected.
Due to the results of Pouget et al. (2002), it was also hypothesized that measures may be less
correlated between visual and proprioceptive target modalities. Thus, two estimates a lower
correlation coefficient 0.15 and a higher correlation coefficient of 0.5 were used for the analysis.
Alpha was set to 0.05 and a nonsphericity correction of 1 was used. All values were submitted to
GPower 3.1 (Faul, Erdfelder, Buchner, & Lang, 2009) and it was determined that 6-9 participants
would be sufficient to achieve 95 percent power, with a critical F between 2.48 and 2.60.
166
Experiment*and*Comparison! N! Effect!Size!F!
Henriques!et!al.,!1998!!(Experiment!1=!left=right)! 9" 0.82"
Pouget!et!al!2002!Experiment!a!(static=!left=0)! 9" 0.65"
Pouget!et!al!2002!Experiment!a!(static=!right=0)! 9" 0.6"
Pouget!et!al!2002!Experiment!b!(static=!left=!0)! 9" 0.73"
Pouget!et!al!2002!Experiment!b!(static!=right!0)! 9" 0.71"
Pouget!et!al!2002!Experiment!c!(static!=!right!=!0)! 9" 0.51"
Pouget!et!al!2002!Experiment!c!(static=!left!=!0)! 9" 0.46"
Table!5.! Effect!sizes!of!relevant!studies!
167
! !Experiment!A2!Power!Analysis!
Similar to Experiment A1, an analysis was conducted to determine how many participants were
needed to achieve the power necessary for statistical analyses. However, because few studies have
investigated evoked potentials during movement planning it was difficult to get an estimate of
what the expected population effect size would be for these tests. Blouin et al., 2014 investigated
somatosensory evoked potentials during movement planning to somatosensory and visual targets.
They recruited 10 participants and correlation analyses employed yielded strong positive
correlation between SEP amplitude and directional errors for proprioceptive targets [r = 0.61, p =
.05, yielding a d =1.5 (Cohen, 1992)]. The authors also noted a strong negative correlation between
SEP amplitude and directional errors with visual targets [r = -0.80, p = 0.05, yielding a d = 2.6].
Thus, given the large correlation effect size, and the uncertainty in the literature, the lower of the
two effect sizes was used to compute the estimated power (1.5). The statistical model used was the
pairwise t-test and alpha was set to .05. All values were submitted to Gpower 3.1 and the analysis
revealed that a total of 8 participants would achieve 95 percent power, with a critical t of 2.3.
Due to the uncertainty of the estimates for the population effect size, a greater number of
participants (10) were recruited. This number would also match previous studies looking at evoked
potentials during movement planning (Blouin et al., 2014).
! Study!B!Power!Analysis!
In Study B, a novel paradigm was employed to investigate latency of online corrections under
different sensory conditions. Our power analysis for this study was supposed to be based on
previous studies that investigated the latency of online corrections when reaching to visual targets.
The number of participants of these experiments is outlined in Table 6. However, because of a lack
of information with regard to the exact values of the effect sizes, an a priori power analysis could
not be computed.
168
Study Number of Participants
Komilis et al. (1993) Exp1: 10; Exp2: 5
Reichenbach et al. (2009) 11
Saunders and Knill (2003) 12
Sarlegna and Blouin (2010) Exp1: 8; Exp2: 5
A post-hoc power analysis was performed on the key variables of constant error and latency of
online corrections to verify that the tests used in Study B were sufficiently powered to find between
modality differences. As G*power is still limited to the assumption of one within subject factor
(Faul et al., 2009) and between modality differences were the focus of the experiment, power
analyses were conducted using the main effects of modality. For the main effect of modality, the
parameters inputted into G*power were as follows:
effect size f (U) was calculated as 1.78; (using the partial eta of 0.76 option: as in SPSS)
alpha: 0.05
Sample size: 14
Number of groups: 1
Number of measurements: 2 (modalities visual and somatosensory)
Nonsphericity correction ! = 1
Power was estimated at 99% with a critical F of 4.67.
For the main effect of modality on correction latency the following values were used:
Table!6.! Number!of!participants!in!relevant!studies!
169
effect size f (U) was calculated as 4.89; (using the partial eta of 0.96 option: as in SPSS)
alpha: 0.05
Sample size: 14
Number of groups: 1
Number of measurements: 2 (modalities visual and somatosensory)
Nonsphericity correction ! = 1
Power was estimated at 100% with a critical F of 4.67.
Based on these power analyses, the studies in the present thesis were likely sufficiently powered.
These results also suggest that this effect could be seen with smaller sample sizes. A liberal
estimate by G*power reveals that based on our data you should reliable results with samples as
low as 6 participants (as low as 3 for correction latency results only).
170
Appendix!2:!Supplementary!Analysis!Study!A!
! Robot!Apparatus!Development!and!Testing!
The robot perturbation device used in Study B was developed based on previously a existing
apparatus used for training upper-limb reaching movements (Manson et al., 2014). The device is
an Epson SCARA E2L853 (Seiko Epson Corp). The robot moves with 4 degrees of freedom and
is capable of reproducing highly accurate spatial trajectories (0.02 mm error) with loads of up to 5
kilograms.
To develop a protocol capable of producing real-time perturbations during the movement required
shifting the computational workload required by the motion tracking system to another device.
This solution was tested as numerous other possibilities using the same device for both motion
tracking and robotic guidance (as in Manson et al., 2014) failed to produce the desired result (see
a Figure .
There were two specific challenges: the significant lag and variability in the time between the
processing of the arm trajectory from the optotrak and the starting of the perturbation (see Figure
28). The second and related challenge was the loss of resolution that occurred when both the
optorak and robotic trajectories were recorded (see Figure 29).
171
!!Reaction!of!the!robot!perturbation!movement!when!signaled!by!motion!tracking!processed!on!the!same!computer.!
!Number! of! lost! samples! in! 500! 1Bsecond! recordings.! Sixty! (60)! percent! of! data! were! lost! samples,! this!severely!decreased!the!amount!and!resolution!of!the!useable!recordings.!Also!55%!during!the!robot’s!trajectory.!!
Lost Samples
172
To resolve these issues, numerous solutions were proposed and tested. For brevity only the solution
implemented for Study B will be presented. The proposed solution involved the use of a network
crossover cable to transform the single station robotic guidance setup into a dual station setup (see
Figure 30). To accomplish this, trajectories from both the robotic arm and the reaching arm were
tracked by the Optotrak on a separate work station. Using MATLAB’s data acquisition toolbox
and a combination of a hardwired direct ethernet connection, and custom parallel port input-output
devices attached to both robot controller and the Optotrak workstation realtime signals about the
limb’s trajectory (movement start, movement end) were used to signal robot perturbations. Using
this solution resulted in faster and less variable robot reaction times (see Figure 31) and less loss
due to dropped samples (see Figure 32). These values were computed using a longer sample time
(3s) to make it closer to the time used in the actual experiment.
173
!!Differences!between!single!and!dual!setups!for!the!recording!of!perturbation!data.!The!dual!station!setup!was!found!to!be!the!best!solution!for!both!maximizing!resolution!of!trajectory!data!and!reducing!variability!in!the!speed!of!perturbations.!
!Most!reaction!times!for!the!dual!station!setup!fell!within!60B!75!milliseconds!after!the!go!signal.!This!reduction!perturbation!variability!provided!a!basis!for!the!protocol.!!
!
174
!
!Overall!the!number!of!lost!samples!was!reduced!from!60%!to!10%!with!only!4%!of!lost!samples!occurring!during!the!robot’s!movements.!!
Lost Samples
175
Appendix!3:!Supplementary!Analysis!Study!B!
! Motion!Detection!Time!versus!Latency!of!Online!Corrections!
An additional analysis was conducted to examine if the differences in the time needed to detect
target motion was sufficient to explain the differences in correction latencies, between the two
target modalities. To answer this question, the differences in detection times were compared to the
differences in correction latencies. If the differences in correction times were attributed solely to
differences in the detection of target perturbations, then the between-modality differences in
detection times should not be different than the between-modality difference in correction times.
Between-modality differences in detection time (DTdiff)were computed for each of the perturbation
directions in the vocal response time protocol (see equation below).
DTdiff = DTvis – DTsoma
The between- modality differences in correction latencies (CTsoma) were computed by averaging
correction latencies from both the 100 ms and 200 ms conditions (see equations below) and then
subtracting the average differences between modalities for each of the perturbation directions.
CTvis = (CTvis 100 ms + CTvis 200 ms) / 2
CTsoma = (CTsoma 100 ms + CTsoma 200 ms) / 2
CTdiff = CTvis - CTsoma
Both CTdiff and DTdiff times were submitted to a 2 measure (detection time, correction time) by 2
perturbation direction (away, toward) repeated-measures ANOVA. The ANOVA revealed a
significant main effect of measure, F(1,13) = 25.34, p < 0.001, and a significant interaction
between measure and perturbation direction, F(1,13) = 15.45, p <.01, HSD = 39 ms. Overall, the
between modality differences in detections time were significantly lower (M = 34 ms, SD = 61)
than the between modality differences in correction time (M = 120 ms, SD = 41). Breaking down
the interaction revealed that only the between-modality differences in correction latencies were
larger for the targets perturbed towards the body (M = 144 ms, SD = 37) vs. away from the body
176
(M = 95 ms, SD = 30). Also, these differences in limb trajectory correction latencies were larger
than the between-modality differences in detection times (towards: M = 22 ms, SD = 50; away: M
= 47 ms, SD = 70), which did not differ across perturbation directions (see Figure 33).
!Contrasts!between!the!target!modality!differences!in!correction!time!and!detection!time.!The!difference!between!the!two!target!modality!conditions!is!significantly!larger!incorrection!time!compared!to!detection!time.!
!
! Normalized!Trajectory!Deviations!To further understand the differences in correction latency, a supplementary analysis on trajectory
deviations was also conducted. To do this, liberal t-tests were conducted between normalized
trajectory positions of perturbed and control trials in the direction axis. Overall, the results of these
analyses show that participants corrected for target significantly more in the somatosensory target
condition compared to the visual target condition. Significant trajectory differences were found
later in the trajectory between movements to somatosensory targets perturbed before and 100 ms
after movement onset and unperturbed trajectories. In fact, no reliable differences were found
between movements to visual targets perturbed after movement onset (see Figure 34).
Detection Time Correction Time0
50
100
150
200
Measure
Betw
een-
mod
ality
tim
e di
ffere
nce
(ms)
Visual- Somatosensory Differences for Response and Correction Tasks
Away
Toward
*
177
The results of this analysis, while supporting the primary hypothesis, does not completely coincide
with the results of our analyses in Study B. Early adjustments to somatosensory target
perturbations and later adjustments (but adjustments nonetheless) to movements to perturbed
visual targets were found. The present analysis could be viewed as evidence that participants were
unable to correct to visual target perturbations. These conflicting results are not unexpected as
studies examining collection latency methods (Oostwoud Wijdenes et al., 2014) has shown spatial
trajectory analyses are the most conservative measure of corrections.
178
!Analysis!of!trajectory!deviations.!Each!curve!represents!a!temporally!normalized!averaged!trajectory!of!each!condition!used!in!the!experiment.!Error!patches!represent!the!between!subject!standard!deviation.!Targets!were!either!perturbed!Before,!100!ms!or!200!ms!after!movement!onset!or!were!unperturbed!(control).!Participants!adjusted!their!trajectory! to! somatosensory! targets! perturbed! 100! ms! after! movement! onset! but! no! differences! between! spatial!trajectories! were! found! for! visual! targets.! Participants! were! capable! to! fully! adjusting! their! movements! to! targets!perturbed!before!movement!start.!!
179
Appendix!4:!LaplacianBTransformed!VEPs!for!Study!A!
!!Grand!average!of!the!Laplacian!transformed!VEPs!for!each!electrode!of!interest!in!the!AUDBVIS!and!AUDBSOMA!conditions.!The!peakBtoBpeak!amplitudes!between!P1!and!N1!were!used!for!the!normalization!and!comparisons!made!in!Experiment!A1.!! !
180
Appendix!5:!Experimental!Questionnaires!!
! Prior!to!Participation!Questionaires!
! Hand!Dominance!Test!
Hand dominance test (adapted from Oldfield, 1971)
Please indicate which hand you would use for the following activities:
Writing right left
Throwing right left
Scissors right left
Toothbrush right left
Drawing right left
***Participants answering right to 4 items or more are deemed to be right hand dominant. ***
Oldfield, R.C. 1971. The assessment and analysis of handedness: the Edinburgh
inventory. Neuropsychologia, 9, 97-113.
181
! Eye!Dominance!Test!
To perform the Miles (1930) test, participants will be asked to extend both arms in front of
themselves. They are then asked to bring both hands together to create a small opening and then
view a distant object through the opening. The experimenter will then ask the participant to close
right eye. If the viewed the object is no longer visible, the participant will be deemed to be right-
eye dominant.
!!Adapted!from!Miles,!W.R.!(1930).!Ocular!dominance!in!human!adults.!The$Journal$of$General$Psychology,$3,!412B430.!
182
! Brief!Neurological!Questionnaire!!
How often do you experience the following?
Headaches Never Seldom Often
Light-headed or dizziness Never Seldom Often
Numbness or tingling Never Seldom Often
Tremor Never Seldom Often
Paralysis Never Seldom Often
Convulsions or seizures Never Seldom Often
Stroke Never Seldom Often
Sensory impairment Never Seldom Often
To be considered neurologically intact, participants cannot tick more than one “often” box in the
first four categories and must tick “never” in the last four categories