Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Manual tracking in Parkinson's disease : implications for L-dopa-induced dyskinesias Stevenson, James 2011

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

Item Metadata

Download

Media
24-ubc_2011_fall_stevenson_james.pdf [ 3.16MB ]
Metadata
JSON: 24-1.0103332.json
JSON-LD: 24-1.0103332-ld.json
RDF/XML (Pretty): 24-1.0103332-rdf.xml
RDF/JSON: 24-1.0103332-rdf.json
Turtle: 24-1.0103332-turtle.txt
N-Triples: 24-1.0103332-rdf-ntriples.txt
Original Record: 24-1.0103332-source.json
Full Text
24-1.0103332-fulltext.txt
Citation
24-1.0103332.ris

Full Text

Manual tracking in Parkinson’s disease: implications for L-dopa-induced dyskinesias by  James Stevenson B.Sc., Dalhousie University, 2006  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (NEUROSCIENCE) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) May, 2011 © James Stevenson, 2011  ABSTRACT Though Parkinson’s disease (PD) is considered to be a prototypical basal ganglia disorder, it has become increasingly clear that this traditional view does not capture the complexity of the disease pathophysiology. For instance, imaging studies demonstrate altered cerebellar activity in PD that may compensate for and/or contribute to the symptoms of the disease. L-dopa-induced dyskinesias (LID) are involuntary writhing movements that commonly occur as a side effect of L-dopa therapy, and despite the prevalence of LID their underlying mechanisms are poorly understood. Altered cerebellar activity in PD may contribute to the pathophysiology of LID, and due to altered ‘forward models’ lead dyskinetic subjects to more heavily rely on ambiguous visual feedback. The principal aim of this thesis was to investigate the ability of PD subjects to deweight ambiguous visual feedback during motor performance, while examining this ability as well as subtle differences in motor performance across dyskinetic and non-dyskinetic PD subjects. To this end we designed a large-amplitude visually guided tracking task where the target ‘jittered’ about the desired trajectory, and used root mean square (RMS) error and linear dynamical system (LDS) models to quantify tracking performance. The three major findings of this work were: 1) in addition to their known susceptibility to speed, PD subjects off medication were significantly more susceptible to increasing visual uncertainty than control subjects, 2) despite similar RMS error during non-ambiguous tracking the damping ratio parameter of the LDS models was significantly lower for dyskinetic subjects off medication, and 3) dyskinetic PD subjects were significantly more susceptible to visual uncertainty than non-dyskinetic and control subjects, and though L-dopa improved their  ii  overall tracking ability, this came at the price of a greater response to and reliance on ambiguous visual feedback. From this work we conclude that PD subjects demonstrate an impaired ability to deweight ambiguous visual input, possibly due to inadequate forward models, and which may be specific to LID pathophysiology. The presence of motor abnormalities while dyskinetic subjects are off medication and not actively experiencing LID is suggestive of persistent neural plasticity. We argue these findings are related to altered cerebellar function in PD.  iii  PREFACE Chapter 2 of this thesis was conducted under the supervision of Dr. Martin J. McKeown, and in collaboration with Dr. Meeko K. Oishi and Dr. Edmond Cretu in the Electrical and Computer Engineering department at UBC, as well as Dr. Cretu’s Masters student Sara Farajian. A version of chapter 2 has been published: [Stevenson, JK], Oishi, MMK, Farajian, S, Cretu, E, Ty, E. and McKeown, MJ. (2011) Response to sensory uncertainty in Parkinson’s disease: a marker of cerebellar dysfunction? Eur J Neurosci, 33: 298-305. My role in the work of this chapter was to help design and implement the experimental paradigm, conduct the experiment and collect the data, perform the data analysis, draft the manuscript and revise it in collaboration with Drs. McKeown and Oishi. Dr. Edna Ty recruited the subjects, and helped to conduct the experiments. Sara Farajian and Dr. Cretu helped with the study implementation, and Sara Farajian helped with data collection. Chapters 3 and 4 were conducted under the supervision of Dr. Martin J. McKeown, and in collaboration with Dr. Meeko K. Oishi and her Masters student Pouria Talebifard. The experimental paradigm was the same as that used for chapter 2. We recruited four additional PD subjects for these chapters, and I conducted the experiments with the study subjects, collected the data, and drafted and edited the manuscripts. Data analysis was performed by Dr. Oishi, Pouria Talebifard, Dr. McKeown and myself. In collaboration with Drs. McKeown and Oishi, I revised and edited the manuscripts, and Dr. Edna Ty recruited all the study subjects and helped with conducting the experiments. A list of published and submitted material from this thesis is provided:  iv  Chapter 2:  A version of chapter 2 has been published: Stevenson JK, Oishi MMK, Farajian S, Cretu E, Ty E, & McKeown MJ. (2011) Response to sensory uncertainty in Parkinson’s disease: a marker of cerebellar dysfunction? Eur J Neurosci, 33: 298-305.  Chapter 3:  A version of chapter 3 is under review for publication: Stevenson JK, Talebifard P, Ty E, Oishi MMK, & McKeown MJ. Dyskinetic Parkinson’s disease patients demonstrate motor abnormalities off medication.  Chapter 4:  A version of chapter 4 has been submitted for publication: Stevenson JK, Talebifard P, Ty E, Oishi MMK, & McKeown MJ. Improvement at a price: Ldopa worsens sensitivity to unclear visual input in dyskinetic Parkinson’s.  All the work conducted in this thesis was under UBC Research Ethics Board and Vancouver Coastal Health Authority approval. The UBC Clinical Research Ethics Board Certificate number is H08-01402, and the Vancouver Coastal Health Authority Research number is V08-0188.  v  TABLE OF CONTENTS ABSTRACT........................................................................................................................................ ii	
   PREFACE .......................................................................................................................................... iv	
   TABLE	
  OF	
  CONTENTS .................................................................................................................. vi	
   LIST	
  OF	
  TABLES.......................................................................................................................... viii	
   LIST	
  OF	
  FIGURES ........................................................................................................................... ix	
   ACKNOWLEDGEMENTS.................................................................................................................x	
   CHAPTER	
  1	
  INTRODUCTION ...................................................................................................... 1	
   1.1	
   Parkinson’s	
  disease:	
  not	
  just	
  a	
  disease	
  of	
  the	
  basal	
  ganglia ..........................................1	
   1.1.1	
   Parkinson’s	
  disease	
  and	
  the	
  classic	
  model	
  of	
  the	
  basal	
  ganglia ......................................... 1	
   1.1.2	
   Pathology	
  beyond	
  the	
  basal	
  ganglia ............................................................................................... 3	
   1.2	
   L-­dopa-­induced	
  dyskinesias	
  in	
  PD..........................................................................................4	
   1.2.1	
   LID	
  severity	
  and	
  risk	
  factors.............................................................................................................. 5	
   1.2.2	
   Basal	
  ganglia	
  pathology	
  in	
  LID.......................................................................................................... 6	
   1.2.3	
   LID	
  pathology	
  beyond	
  the	
  basal	
  ganglia ....................................................................................... 8	
   1.3	
   Visually	
  guided	
  movements	
  and	
  the	
  cerebellum ............................................................ 10	
   1.3.1	
   Visual	
  afferents	
  to	
  the	
  cerebellum.................................................................................................10	
   1.3.2	
   Visually	
  guided	
  tracking	
  and	
  the	
  cerebellum ...........................................................................11	
   1.3.3	
   Cerebellar	
  lesions	
  and	
  visually	
  guided	
  movements...............................................................12	
   1.4	
   Forward	
  models,	
  the	
  cerebellum	
  and	
  motor	
  control .................................................... 13	
   1.4.1	
   The	
  forward	
  model	
  and	
  state	
  estimation ...................................................................................14	
   1.4.2	
   Uses	
  of	
  forward	
  models......................................................................................................................15	
   1.4.3	
   Forward	
  models	
  and	
  sensory	
  uncertainty.................................................................................16	
   1.4.4	
   Forward	
  models	
  of	
  the	
  external	
  environment .........................................................................19	
   1.4.5	
   The	
  cerebellum	
  as	
  an	
  anatomical	
  substrate	
  of	
  forward	
  models ......................................20	
   1.5	
   Parkinson’s	
  disease	
  and	
  visually	
  guided	
  tracking.......................................................... 23	
   1.6	
   Study	
  aims	
  and	
  hypotheses .................................................................................................... 24	
    CHAPTER	
  2	
  RESPONSE	
  TO	
  VISUAL	
  UNCERTAINTY	
  IN	
  PARKINSON’S	
  DISEASE:	
  A	
   MARKER	
  OF	
  CEREBELLAR	
  DYSFUNCTION?.........................................................................32	
   2.1	
   Synopsis ........................................................................................................................................ 32	
   2.2	
   Introduction................................................................................................................................. 33	
   2.3	
   Methods......................................................................................................................................... 34	
   2.3.1	
   Subjects .....................................................................................................................................................34	
   2.3.2	
   Study	
  paradigm......................................................................................................................................35	
   2.3.3	
   Quantification	
  of	
  manual	
  tracking.................................................................................................37	
   2.3.4	
   Signal	
  analysis ........................................................................................................................................37	
   2.3.5	
   Quantification	
  of	
  tracking	
  performance......................................................................................38	
   2.3.6	
   Statistical	
  analyses ...............................................................................................................................38	
   2.4	
   Results ........................................................................................................................................... 39	
   2.4.1	
   Tracking	
  performance ........................................................................................................................40	
    vi  2.4.2	
   Training	
  effect ........................................................................................................................................40	
   2.4.3	
   Effect	
  of	
  ambiguity	
  on	
  RMS	
  tracking	
  error ................................................................................40	
   2.4.4	
   Effect	
  of	
  medication	
  on	
  RMS	
  tracking	
  error..............................................................................41	
   2.4.5	
   Effect	
  of	
  tracking	
  speed	
  on	
  RMS	
  tracking	
  error.......................................................................41	
   2.4.6	
   Regression	
  analysis:	
  effect	
  of	
  target	
  ambiguity	
  and	
  speed	
  on	
  RMS	
  tracking	
  error..41	
   2.5	
   Discussion .................................................................................................................................... 42	
   2.5.1	
   Conclusion................................................................................................................................................48	
    CHAPTER	
  3	
  DYSKINETIC	
  PARKINSON’S	
  DISEASE	
  PATIENTS	
  DEMONSTRATE	
   MOTOR	
  ABNORMALITIES	
  OFF	
  MEDICATION......................................................................54	
   3.1	
   Synopsis ........................................................................................................................................ 54	
   3.2	
   Introduction................................................................................................................................. 55	
   3.3	
   Methods......................................................................................................................................... 56	
   3.3.1	
   Subjects .....................................................................................................................................................56	
   3.3.2	
   Study	
  paradigm......................................................................................................................................58	
   3.3.3	
   Quantification	
  of	
  tracking	
  performance......................................................................................58	
   3.3.4	
   Statistical	
  analyses ...............................................................................................................................59	
   3.4	
   Results ........................................................................................................................................... 59	
   3.5	
   Discussion .................................................................................................................................... 61	
    CHAPTER	
  4	
  IMPROVEMENT	
  AT	
  A	
  PRICE:	
  L-­DOPA	
  INDUCES	
  EXCESSIVE	
  RELIANCE	
   ON	
  AMBIGUOUS	
  VISUAL	
  INPUT	
  IN	
  PARKINSON’S	
  SUBJECTS	
  WITH	
  DYSKINESIA ...70	
   4.1	
   Synopsis ........................................................................................................................................ 70	
   4.2	
   Introduction................................................................................................................................. 71	
   4.3	
   Methods......................................................................................................................................... 74	
   4.3.1	
   Subjects .....................................................................................................................................................74	
   4.3.2	
   Study	
  paradigm......................................................................................................................................75	
   4.3.3	
   Quantification	
  of	
  tracking	
  performance......................................................................................75	
   4.3.4	
   Statistical	
  analyses ...............................................................................................................................76	
   4.4	
   Results ........................................................................................................................................... 76	
   4.5	
   Discussion .................................................................................................................................... 80	
    CHAPTER	
  5	
  CONCLUSION..........................................................................................................92	
   5.1	
   Introduction................................................................................................................................. 92	
   5.2	
   Summary	
  of	
  findings	
  and	
  interpretations ......................................................................... 93	
   5.3	
   Study	
  significance ....................................................................................................................100	
   5.4	
   Study	
  limitations ......................................................................................................................103	
   5.4.1	
   Subjects ..................................................................................................................................................103	
   5.4.2	
   The	
  off-­‐medication	
  state.................................................................................................................105	
   5.4.3	
   Visual	
  contrast	
  sensitivity..............................................................................................................106	
   5.5	
   Future	
  directions .....................................................................................................................106	
    BIBLIOGRAPHY ......................................................................................................................... 110	
    vii  LIST OF TABLES  Table 2–1. Description of PD subjects’ characteristics. .........................................................49	
   Table 3–1. Description of dyskinetic and non-dyskinetic subjects’ characteristics. .............66	
    viii  LIST OF FIGURES Figure 1-1. The classic model of the basal ganglia.................................................................29	
   Figure 1-2. The therapeutic window. ......................................................................................30	
   Figure 1-3. The forward model. ..............................................................................................31	
   Figure 2-1. The visually guided tracking task and performance. ...........................................50	
   Figure 2-2. The training effect. ...............................................................................................51	
   Figure 2-3. Effect of speed and ambiguity on RMS error. .....................................................52	
   Figure 3-1. Damping ratio by group. .....................................................................................67	
   Figure 3-2. LDS model simulation results: dyskinetic tracking overshoot. ...........................68	
   Figure 3-3. Damping ratio and UPDRS. ................................................................................69	
   Figure 4-1. The training effect for dykinetic and non-dyskinetic subjects. ............................88	
   Figure 4-2. Effect of speed and ambiguity on dyskinetic subjects’ RMS error.....................89	
   Figure 4-3. Mean decay rate by group. ..................................................................................90	
   Figure 4-4. LDS model simulation results: response to ambiguous visual feedback. ............91	
    ix  ACKNOWLEDGEMENTS I would like to thank my supervisor, Dr. Martin McKeown, for his continued guidance and support throughout the duration of my graduate studies. Your willingness and desire to teach serves as an inspiration to me, and has provided me with a strong research foundation on which to continue building. Moreover your kindness extended throughout my training has made this a wonderful experience. Thank you. I would like to thank our collaborator Dr. Meeko Oishi, and her student Pouria Talebifard, for their extensive help and contribution in the completion of this thesis. I would like to thank my supervisory committee for their advice and guidance through my research project, and for their time invested in helping me with my research training. I would like to thank Dr. Edna Ty for her unwavering help in recruiting subjects and conducting the experiments with me, and for being a wonderful person with which to work. I would also like to thank the members of the McKeown lab for their friendship and support. I could not have done this work without the gracious help of the patients and their families who selflessly participated in this research. You shall always be an inspiration to me.  x  CHAPTER 1 INTRODUCTION 1.1  Parkinson’s disease: not just a disease of the basal ganglia  	
    Parkinson’s disease (PD) is a neurodegenerative disease that affects more than one million people in North America (Lang & Lozano, 1998a). PD is diagnosed clinically, the hallmark symptoms of the disease being tremor, rigidity, akinesia and postural instability (Fahn, 2003). PD is associated with a significant increase in mortality as compared to an agematched healthy population (Bennett et al., 1996; Morens et al., 1996), and consequently significantly reduces patients’ life expectancy (Morens et al., 1996). 1.1.1  Parkinson’s disease and the classic model of the basal ganglia  The primary pathology of PD is the progressive degeneration of the dopaminergic neurons of the substantia nigra pars compacta (SNpc), resulting in a significant reduction of striatal dopamine levels (the major input of the basal ganglia) (Bergman & Deuschl, 2002) that is believed to result in rigidity and akinesia (Lang & Lozano, 1998a). Physiologically, the decrease in striatal dopamine in PD affects the basal ganglia circuitry via decreased activation of the striatal dopamine D1 and D2 receptors (Lang & Lozano, 1998a; Obeso et al., 2000). This ultimately leads to excessive inhibitory output of the Globus Pallidus Internal segment (GPi), resulting in excessive inhibition of the thalamo-cortical motor pathway that characterizes PD (Obeso et al., 2000). The basal ganglia are characterized by the direct and indirect pathways. In the normal state the direct pathway serves to inhibit the GPi (and thus facilitate movement), and the indirect pathway ultimately stimulates the GPi and inhibits movement (Lang & Lozano, 1998b). This dichotomy between the basal ganglia pathways has led to the notion of action selection and cessation, where the analogy of gas and brake pedals has been applied to the  1  direct and indirect pathways, respectively (Graybiel, 2000). Dopamine acts to modulate these basal ganglia pathways through the excitatory D1 receptor that is associated with the ‘direct’ pathway, and the inhibitory D2 receptor through the ‘indirect’ pathway (Lang & Lozano, 1998b). Due to the opposing effects of dopamine on the D1 and D2 receptors, dopamine always has a pro-movement modulating effect on the basal ganglia, where it increases activity of the direct pathway and decreases activity of the indirect pathway (Lang & Lozano, 1998b). Therefore, the effect of a reduction in dopamine in PD ultimately leads to decreased movement through overactivity of the indirect pathway and reduced activity of the direct pathway (Lang & Lozano, 1998b). As part of the indirect pathway, the subthalamic nucleus normally provides excitatory input to the GPi. However, markedly increased excitatory activity in the subthalamic nucleus and the resulting excessive inhibitory activity of the GPi are cardinal pathophysiological features of PD, and are believed to underlie many of the symptomatic features of the disease (Obeso et al., 2000). Deep brain surgery (DBS) is a procedure that may be offered to advanced PD patients, and DBS is most commonly applied to the GPi or the subthalamic nucleus, and can be very effective at alleviating parkinsonian symptoms (Obeso et al., 1997, Krack et al., 1998; Lang & Lozano, 1998b). Though this classic model of the basal ganglia circuitry is continually being adapted as progress is made to better understand its physiology, the model has been very useful in providing a conceptual framework of the basal ganglia and in providing targets for DBS surgery in PD (Obeso et al., 2008). Figure 1-1 demonstrates the classic model of the basal ganglia.  2  1.1.2  Pathology beyond the basal ganglia  Pathology in PD is not limited to the SNpc and striatum, as a host of other neuron populations are as well affected, including the nucleus basalis of Meynert, hypothalamic neurons, olfactory bulb neurons, and small cortical neurons (Lang & Lozano, 1998a). From a behavioural perspective, the involvement of dementia in PD is being increasingly recognized where studies indicate nearly 30% of PD patients are afflicted (Aarsland et al., 1996). A percentage of depression in PD is explained by cortical Lewy body pathology, though a significant fraction of the depression in PD is of unknown etiology beyond the PD itself (Hughes et al., 1993). Further evidence demonstrates pathology in the hippocampus (Churchyard & Lees, 1997) and the amygdaloid nucleus (Braak et al., 1994; Tandberg et al., 1996), and cognitive impairments have been demonstrated in PD involving frontostriatal circuitry leading to executive dysfunction (Zgaljardic et al., 2003). Thus it is clear a number of brain regions are affected in PD, likely contributing to the heterogeneity of the symptoms of the disease. Functional changes in the cerebellum and its connections have also been demonstrated in PD (Rascol et al., 1997; Lewis et al., 2007; Yu et al., 2007; Ballanger et al., 2008; Palmer et al., 2009; Ni et al., 2010), and it is hypothesized that the cerebellum may be recruited in PD to compensate for the affected basal ganglia (Glickstein & Stein, 1991). Concurrent cerebellar hyperactivity and putamen hypoactivity (an integral basal ganglia nucleus) has been demonstrated in PD, suggestive of neural compensation presumably by recruitment of the cerebello-thalamo-cortical (CTC) circuit (Yu et al., 2007). Further imaging studies have demonstrated cerebellar compensation in PD through increased CTC activity in a PD twin compared to the non-PD twin during a visually guided task (Lewis et al., 2007), and through  3  recruitment of novel areas of the cerebellum in PD subjects during a visually guided tracking task (Palmer et al., 2009). Anecdotal evidence has long supported the observation of paradoxical movements in PD, whereby patients can leap to safety from the threat of impending danger, despite their normally debilitating bradykinesia (Glickstein & Stein, 1991). More recent imaging provides evidence that ‘motor urgency’ in a visually guided reaching task in PD may be processed in the cerebellum (Ballanger et al., 2008). Thus it is clear that many neural systems are involved in PD, and though the hallmark of the disease remains the degeneration of the SNpc dopaminergic neurons with the subsequent decrease in striatal dopamine, there is a substantial amount of evidence indicating altered brain regions outside the basal ganglia in PD, including the cerebellum. 1.2  L-dopa-induced dyskinesias in PD  More than fifty years since its inception, L-dopa remains the gold standard of therapy in PD (Agid et al., 1999). However, despite its extensive use in PD treatment, L-dopa therapy is not without adverse side effects, one of which is levodopa-induced dyskinesias (LID). LID are involuntary, irregular, purposeless, movements that may take many different forms, for example stereotyped movements, chorea (dance like movements), ballism (violent flinging movements), dystonia (sustained muscular contractions) and myoclonus (brief muscle twitches) (Jankovic, 2005). Though dyskinesias can occur as L-dopa medication is wearing off (known as ‘off dyskinesias’ of which dystonia is commonly observed), and when an L-dopa dose is taken and while it is wearing off (known as ‘diphasic dyskinesias’), the most common type of dyskinesias are peak-dose LID that usually manifest about 30 minutes after taking an L-dopa dose and coincide with peak dopamine brain concentrations (Fahn,  4  2000). Peak-dose LID commonly consist of stereotypic, choreic or ballistic movements in the extremities and/or head and neck (Jankovic, 2005). 1.2.1  LID severity and risk factors  LID can range from mild to severe, and accordingly there is significant variability in the impact they have on patients’ quality of life. In their milder form LID may not be noticed by PD patients (Vitale et al., 2001), and may even be associated with an increased quality of life (Marras et al., 2004). At least initially, patients may prefer mild LID as a side effect of improved motor function to the return of parkinsonian symptoms experienced at lower prescribed L-dopa doses in attempt to minimize the occurrence of LID (Schrag & Quinn, 2000). However, in their more severe form LID can interfere with daily function and are associated with a reduced quality of life (Damiano et al., 2000; Pechevis et al., 2001; Pechevis et al., 2005). Pechevis & colleagues (2005) have found that LID can limit patients’ social lives leading to isolation, frustration, anger and depression, and are associated with increased health care costs. Thus, despite the frequently mild initial presentation of LID, as PD progresses LID can become intrusive and problematic for the patient. Several risk factors for the development of LID have been indentified and include age of onset of PD, duration of L-dopa treatment, daily L-dopa dose, and disease progression (Nyholm & Aquilonius, 2004; Jankovic, 2005). Interestingly, young-onset PD is associated with a higher incidence of LID, as 50% of patients with disease onset between 40-59 years of age are expected to develop LID within 5 years of starting L-dopa therapy (Kumar et al., 2005). By contrast, only 16% of those with disease onset at 70 years of age or greater will develop LID within 5 years of L-dopa treatment (Kumar et al., 2005). The finding that younger onset PD patients experience LID more than older patients may be reflective of  5  greater capacity for brain plasticity in response to L-dopa in the younger population. In addition to age of onset, a longer duration of levodopa treatment (Schrag & Quinn, 2000) and an increased L-dopa dose (Fahn, 1999; Fahn, 2002) have been found to increase the risk for developing LID. The ELLDOPA trial examined drug naïve PD patients who were assigned to placebo or to 150, 300, or 600 mg/day of L-dopa for 40 weeks, and demonstrated a significantly greater incidence of LID in those exposed to the highest dose of L-dopa (Fahn, 1999; Fahn 2002). Lastly, there is a relationship between disease progression and incidence of LID. Though it is generally believed that a ‘honeymoon’ period exists in the initial treatment of PD patients, as disease progresses it becomes progressively more difficult to treat PD symptoms without the development of LID, known as a shrinking of the therapeutic window – the range of possible L-dopa dose required to maintain PD patients symptom free and without side effects such as LID (Jankovic, 2005). The relationship between the therapeutic window and the progression of PD can be seen schematically in Figure 1-2. 1.2.2  Basal ganglia pathology in LID  Despite the prevalence of LID in PD their pathophysiological mechanisms are not well understood. Nonetheless, the dysregulation of striatal dopamine homeostasis that occurs with nigral denervation and subsequent L-dopa therapy represents a prevailing hypothesis regarding the development of LID (Chase, 1998). The pulsatile release of dopamine in PD brought about by SNpc degeneration and subsequent L-dopa therapy is in contrast to the relatively constant physiological release of nigral dopamine in the healthy individual, and is thought to prime the brain for the development of LID (Olanow & Obeso, 2000; Cenci, 2007). In support of this, evidence suggests that continuous infusion of L-dopa is beneficial  6  in the clinical management of PD patients to maintain more constant and physiological levels of striatal dopamine to reduce the development of LID (Nyholm & Aquilonius, 2004; Stocchi et al., 2005; Olanow et al., 2006). However, studies demonstrating a beneficial effect of more continuous delivery of dopaminergic agents on LID utilize a duodenal pump to achieve the result, requiring the invasiveness and costs of surgery. Furthermore, in a recent investigation of a pharmacological attempt at continuous dopaminergic therapy via oral administration, the study failed to demonstrate a decrease in the frequency and severity of LID (Stocchi et al., 2010). Moreover, kinematic studies of LID using accelerometers to characterize these movements have demonstrated that despite their random appearance, repeated and structured movement patterns exist that suggest LID are not the result of random pulsatile dopamine release (Gour et al., 2007). Therefore, abnormal dopamine firing within the basal ganglia may only represent part of the pathophysiological picture of LID. Persistent neural plasticity within the basal ganglia, subsequent to dopamine dysregulation, has been postulated to underlie the development of LID (Linazasoro, 2005), and such plasticity has been demonstrated in L-dopa treated parkinsonian animal models (Calon et al., 2000a; Calon et al., 2000b). Plasticity in the nigral-striatal nerve terminal has also been demonstrated in PD patients that experience LID. Troiano & colleagues (2009) have found decreased dopamine transporter expression in dyskinetic PD patients in comparison to non-dyskinetic PD patients and healthy control subjects. The dopamine transporter functions in the reuptake of dopamine from the synaptic cleft to maintain stable synaptic dopamine levels, and reduced expression found in early PD is thought to be a compensatory mechanism to preserve available dopamine in the face of nigral cell death  7  (Lee et al., 2000). As postulated by Troiano & colleagues, such compensatory mechanisms may be detrimental in the long run by contributing to the development of LID. 1.2.3  LID pathology beyond the basal ganglia  However neural plasticity in relation to LID is not limited to the basal ganglia, and has as well been found in other brain regions. Using a finger-to-thumb opposition motor task, Rascol et al. (1998) has found that dyskinetic PD subjects while on L-dopa demonstrate a significantly greater regional cerebral blood flow in the motor cortex and supplementary motor area than non-dyskinetic PD subjects. Morgante et al. (2006) have found differences in the ability of L-dopa to restore motor evoked potentials after transcranial magnetic stimulation (TMS) to the motor cortex, such that L-dopa did not restore motor cortex potentiation in dyskinetic subjects but did in non-dyskinetic subjects. The authors concluded that deficient and non-dopamine responsive motor cortex plasticity may lead to the formation of undesired motor patterns such as LID in PD. Interestingly, the cortical plasticity that differentiated dyskinetic from non-dyskinetic PD subjects described above was in response to L-dopa administration. In other words, it was not until taking L-dopa that these cortical changes differentiating dyskinetic from non-dyskinetic subjects were observed. However, if plasticity in the brain leading to LID were more permanent in nature, then such plasticity would presumably be evident while patients are off medication. Thus, while peak-dose LID manifest while subjects are on medication, plasticity in the off medication state may result in more subtle differences between dyskinetic and non-dyskinetic movements. There is increasing evidence demonstrating functional changes in the cerebellum and its connections in PD subjects while off L-dopa (Rascol et al., 1997; Yu et al., 2007; Lewis et al., 2007; Ballanger et al., 2008; Palmer et al., 2009; Ni et al., 2010), and further evidence  8  implicating the cerebellum in the pathology of LID (Nimura et al., 2004; Koch et al., 2009). Koch & colleagues (2009) demonstrated that cerebellar TMS was capable of decreasing the severity of LID, and the binding potential of cerebellar sigma receptors in PD patients before undergoing pallidotomies has been positively correlated with LID scores but not with disease severity (Nimura et al., 2004). Altered cerebellar activity in PD could contribute to the pathophysiology of LID by directly affecting cortical areas via cerebello-thalamo-cortical pathways, or by directly affecting the basal ganglia through pathways linking the cerebellum and basal ganglia. Recent evidence supports a direct interaction between the cerebellum and the basal ganglia via a disynaptic pathway from the cerebellar dentate nucleus to the striatum of the basal ganglia in primates (Hoshi et al., 2005), as well as a direct pathway between the subthalamic nucleus of the basal ganglia and the cerebellar cortex (Bostan et al., 2010). These pathways provide a direct neuroanatomical basis for functional changes occurring in the cerebellum to affect the basal ganglia, and vice versa. At the behavioural level, dyskinetic patients demonstrate similar responses to visual feedback during visually guided tracking as has been previously demonstrated in patients with cerebellar disease (Beppu et al., 1984; Beppu et al., 1987; Liu et al., 2001). In both the dyskinetic PD patients studied by Liu & colleagues and the cerebellar patients studied by Beppu & colleagues, an increased response to visual feedback was demonstrated. Interestingly, it was upon removal of either the cursor or target during tracking that performance of both subject groups became markedly smoother and the variation in tracking velocity decreased. By contrast, PD subjects of similar disease duration but without LID did not demonstrate increased responses to visual feedback (Liu et al., 1999).  9  Though the basal ganglia are undoubtedly involved in the pathophysiology of LID, there is evidence to suggest that this may incompletely explain LID and that other brain regions, notably the cerebellum, may have important roles in their development. 1.3  Visually guided movements and the cerebellum  The integration between sensory and motor systems is an integral part of motor performance, and as suggested by the significant portion of the cortex devoted to vision, there is a profound influence of vision on movement. At the cortical level visually guided movements have largely been attributed to the fiber tracts projecting from the visual striate cortex to the posterior parietal cortex (PPC). This occipito-parieto pathway is often referred to as the ‘dorsal’ or ‘where’ pathway, and has been differentiated from the occipto-temporal pathway, known as the ‘ventral’ or ‘what’ pathway, the distinction predominantly based on studies of patients with lesions to these pathways (Rossetti et al., 2003). Lesions to the dorsal occipito-parietal stream leads to the development of optic ataxia, a condition of deficits in visually guided movements despite intact perception of visual cues (Perenin & Vighetto, 1988). For example optic ataxia patients have difficulty utilizing appropriate grip aperture in a reaching task and yet can appropriately recognize the object being reached for (Jeannerod, 1986). By contrast, lesions to the occipito-temporal visual stream are associated with difficulties of perception, rendering these patients visually agnostic (Milner et al., 1991). Interestingly, lesions in the occipito-parietal fibers don’t interfere with motor commands or visual sensory input, but rather affect the integration of the two (Rossetti et al., 2003). 1.3.1  Visual afferents to the cerebellum  Major cortical projections are sent from the PPC via the superior longitudinal bundle to the frontal lobe, taking a circuitous route to the frontal eye fields and then to pre-motor  10  cortex (Stein & Glickstein, 1992). It has been noted by Stein & Glickstein (1992) that this is a lengthy route with many relays that is unlikely to solely govern the quick responses to vision during movement. Furthermore, damage to the superior longitudinal bundle while preserving the cortico-pontine fibers (pathways from the occipital and parietal cortices to the pontine nuclei) does not deteriorate visually guided movements (Myers et al., 1968), and consequently the contribution of subcortical structures to visually guided movements has long been recognized. Indeed the cerebellum receives significant afferent input, including visual, auditory, somatosensory and proprioceptive afferents (Stein & Glickstein, 1992), and the visual afferents to the cerebellum travel predominantly from the posterior parietal cortex via the pontine nuclei (Glickstein, 1985, Kamali et al., 2010). The pontine nuclei fibers then project as mossy fibers to the contralateral cerebellar cortex (Robinson et al., 1984). Mossy fiber afferents to the cerebellum are known to modulate purkinje cell activity, which in turn inhibit output from the cerebellar nuclei that project to a wide range of cortical targets including the motor cortex, supplementary motor area and the pre-motor cortex (Middleton & Strick, 1998). Thus, from a neuroanatomical perspective, major pathways exist from the PPC to the cerebellum that allow the cerebellum to gain access to visual afferents, and in turn project back to the cortex to influence visually guided movements. 1.3.2  Visually guided tracking and the cerebellum  Visually guided tracking tasks incorporate many aspects of cerebellar motor control. First, matching the position of a limb with the position of a desired target during tracking requires coordination that is known to degrade with cerebellar lesions (Beppu et al., 1984; Beppu et al., 1987). Secondly, the inherent delays in the visuomotor system of 100 to 300 ms (Miall et al., 1985; Saunders & Knill, 2005) highlight the necessary predictive nature of  11  visually guided tracking (Miall & Reckess, 2002). Interestingly, delaying the visual feedback provided to healthy subjects during tracking deteriorates their tracking to performance similar to that of cerebellar subjects, probably by disrupting the timing between the predicted and actual sensory outcome during tracking (Miall & Reckess, 2002). Work in recent years utilizing functional magnetic resonance imaging (fMRI) provides further evidence supporting the role of cerebellum during visually guided tracking (Miall et al., 2000; Miall et al., 2001). This work demonstrated that 1) during ocular tracking only there was significant activation of the cerebellar vermis and 2) during manual tracking only there was significant activation of the lateral cerebellar hemisphere, and 3) during combined visually guided tracking where the eyes and hand track the same target, the cerebellar hemisphere and vermis activity was significantly increased (Miall et al., 2000). Imaging therefore provides strong evidence supporting integral role of the cerebellum during visually guided tracking requiring predictive and coordinated control. 1.3.3  Cerebellar lesions and visually guided movements  Lesions to the cerebellum in humans give rise to deficits in visually guided movements, though they differ from the deficits observed in optic ataxia. The classic symptoms resulting from lesions to the cerebellum are ataxia, dysmetria and dysdiadochokinesia, grossly interpreted as uncoordinated movement (Holmes, 1917, 1922). In reaching towards a target, monkeys with lesions to the cerebellum demonstrate jerky intermittent segmental movements as opposed to the normal motor response characterized by continuous movements in a smooth trajectory (Stein & Glickstein, 1992). Humans with cerebellar lesions demonstrate similar deficits during visually guided movements, where normally smooth pursuit movements become jerky and intermittent, characterized by errors  12  in direction and velocity, and overshoot is often observed (Hallett et al., 1975; Becker et al., 1990). The fascinating finding that these jerky uncoordinated movements become smooth upon removing vision of the limb (Beppu et al., 1984; Beppu et al., 1987) demonstrates the heightened response of cerebellar patients to visual feedback and the continuous correction of movement that typifies their motor performance (Bastian, 2006). Thus cerebellar patients seem to be overly reliant on and responsive to visual feedback when it is available. The feedback driven motor performance of patients with lesions to the cerebellum during visually guided tasks has led to the hypothesis that the cerebellum is involved in predictive motor control. For example, in a visually guided tracking task the cerebellum may coordinate hand movements with the predicted future location of the target (Stein & Glickstein, 1992). This hypothesis would account for the jerky intermittent tracking behaviour demonstrated by cerebellar patients unable to provide that prediction, who therefore make many corrective movements by responding to the movement errors signaled through visual feedback. Lesion studies support the cerebellum’s involvement in predictive motor control, demonstrated in catching tasks where cerebellar subjects do not generate predictive hand forces in advance of catching a ball, but rather react after the ball is caught (Lang & Bastian, 1999). Thus, as visual afferents to the cerebellum from the PPC may provide information regarding target direction and velocity, the cerebellum may then predict the target’s future location in coordination with the motor response. 1.4  Forward models, the cerebellum and motor control  Occupying approximately 10% of total brain volume, and yet comprising more neurons than the entire rest of the brain, an overarching function of the cerebellum that accounts for its diverse and profound effect on motor control has long been a subject of great  13  interest. In recent years a theoretical framework of cerebellar function, known as the ‘forward model’, has attracted significant research attention and highlights the predictive nature of the cerebellum. 1.4.1  The forward model and state estimation  The forward model is a type of internal model, and in the context of sensorimotor control internal models represent the computations required to transform sensory coordinates into motor coordinates, and vice-versa. There are two types of internal models, known as the ‘forward’ and ‘inverse’ models (Wolpert & Ghahramani, 2000). Inverse models compute the motor command necessary to reach the desired sensory state, whereas forward models provide a prediction of the sensory consequences of movement (Wolpert et al., 1995; Miall & Wolpert, 1996). It is through the forward model that the central nervous system (CNS) is believed to internally represent our interactions with the outside world by predicting sensory feedback throughout any given movement. During motor performance, the CNS requires knowledge of its current state in order to arrive at any given desired state or goal. In other words, with movement of a limb to a new position in space, the CNS must determine the updated state of the limb on an ongoing basis throughout the movement. State estimation is the process of determining the most updated current state of the CNS (e.g., position and velocity of a particular limb during or after movement). The current state can be interpreted from sensory feedback systems (such as vision or proprioception), derived from forward models, or obtained through a weighted combination of these two inputs (Wolpert et al., 1995). Figure 1-3 illustrates the use of feedback, the forward model, and a comparison of the two to derive the most accurate representation of the current state.  14  1.4.2  Uses of forward models  Several reasons have been elucidated as to why the CNS would benefit from computing forward models (Miall & Wolpert, 1996). Perhaps the most obvious reason for the use of forward models in the CNS is to avoid the time delays associated with sensory feedback. With variations depending on the sensory modality, inherent delays exist in the transmission of sensory impulses from the periphery to the CNS. In the visual system evidence indicates that approximately 200 ms are required before visual sensory feedback can be used to update motor performance (van Sonderen et al., 1988). Such delays would lead to sluggish motor performance because the CNS would be continuously delayed in updating its current state. Therefore a major advantage to the use of forward models in motor control is the ability to avoid inherent delays of sensory information, as the CNS could have access to its current state with negligible delay (Miall et al., 1993; Wolpert et al., 1995; Miall & Wolpert, 1996). Forward modeling is furthermore of direct benefit to coordination and timing during motor tasks (Miall & Wolpert, 1996). In a task where subjects performed ocular tracking of a cursor controlled by joystick movements performed with their hand, there was no delay found between the eye and hand movements, implicating the use of a predictive forward model strategy (Vercher & Gauthier, 1988). Expanding on this, when a delay was imposed between the hand movements and the resulting cursor display, the eyes were found to precede the cursor and be in time with the hand motion (Vercher & Gauthier, 1992). Therefore forwards models seem to be integral in producing coordinated hand-eye pursuit movements.  15  1.4.3  Forward models and sensory uncertainty  In addition to avoiding time delays and improving coordination, forward models can provide the CNS with a method to contend with sensory uncertainty. Though sensory feedback can be informative and heavily relied upon in motor performance, sensory feedback can also be ambiguous. For example, sensory uncertainty may be a product of varying degrees of neural noise inherent to neural transmission, effectively degrading the signal transmitted to the CNS (Tomko et al., 1974; Shadlen and Newsome, 1998; van Beers et al., 2002). Furthermore, particular states of the CNS such as fatigue or disease may contribute to uncertainty in interpreting sensory input. This may be of particular importance in PD where decreased visual contrast sensitivity has been demonstrated (Bodis-Wollner et al., 1987; Price et al., 1992). In addition to uncertainty in the neural processing of sensory input, visual cues that guide motor performance can be ambiguous, for example in conditions of low or dim light. Furthermore, in conditions of excessive or extraneous visual input, some cues may be less informative than others. When sensory input becomes ambiguous and relatively non-informative, forward models are able to reduce the effect of sensory uncertainty on motor performance by providing an alternate modality on which to base state estimation and improve motor performance (Wolpert & Ghahramani, 2000; van Beers et al., 2002; Vaziri et al., 2006). There is significant evidence to support the use forward models by human subjects during motor performance (Kuo, 1995; Wolpert et al., 1995; Merfeld et al., 1999; van Beers et al., 1999; Vaziri et al., 2006; Gritsenko et al., 2009). In the seminal experiments carried out by Wolpert & colleagues (1995), after viewing the starting position of their hand healthy human subjects were asked to make reaching movements in the dark, and to subsequently  16  estimate the final position of their hand with the use of a visual cursor controlled by their opposite hand. The subjects’ state estimation data were then modeled by a Kalman filter, which is an ideal observer used in engineering that integrates sensory feedback with forward models to reduce the uncertainty of the state estimate (Wolpert & Ghahramani, 2000). The results provide evidence for the use of forward models, as the bias (the difference between the actual end position and estimated position) and variance of subjects’ estimate were well matched by a Kalman filter model. This study provides evidence that the CNS integrates sensorimotor inputs during motor performance and weights them according to their relative reliability. Vaziri & colleagues (2006) provide further compelling evidence that healthy human subjects integrate sensory feedback with predictive forward models in order to obtain optimal state estimates when faced with sensory uncertainty. To demonstrate this, Vaziri & colleagues had subjects reach to a target in three conditions – visual feedback, forward model, or combined. In the ‘visual feedback’ condition the target was displayed in the peripheral vision with respect to their gaze fixation point, thus visual feedback relating the target position and gaze fixation position was provided in retinal coordinates. In the ‘forward model’ condition subjects were required to make an eye saccade away from the target to a new fixation point, but because the target was extinguished upon making the saccade, this condition required a forward model to remap the visual coordinates of the target with respect to the new fixation position – in other words CNS is required to remap where the target will land on the retina after the saccade (Batista et al., 1999; Crawford et al., 2004). In a third ‘combined’ condition the subjects made a saccade to a fixation point away from the target (as in the forward model condition), but this time the target was illuminated in the peripheral  17  vision after the saccade for a variable amount of time before being extinguished. Therefore in the third condition subjects had access to both visual feedback and to their predictive forward model of the visual feedback. Furthermore, in this combined condition the final target position was either shifted from or at the same position as the pre-saccade target position. This allowed the authors to probe the weighting of the forward model and the visual feedback because the forward model would bias the reaches towards the pre-saccade target position whereas the visual feedback would bias the reaches towards the post-saccade target position. The results demonstrated a reduction in the variability of reach errors in the combined condition as compared to either the visual feedback or forward model conditions, leading to the conclusion that the brain integrates forward models with visual feedback to provide an optimal estimate of the target position. Furthermore, as the uncertainty of the peripheral visual input increased in the combined condition, by decreasing the amount of time the target was displayed in the periphery after the saccade, subjects reached more to the pre-saccade target than the shifted post-saccade target. This indicated that subjects weighted the forward model component of the estimate of the target position more than the visual feedback, when the visual feedback was uncertain. This study provides an elegant demonstration of the brain’s use of forward models of the occulo-motor command to predict the sensory consequences of retinal processing, and demonstrates that in conditions of increasing visual uncertainty the brain weights these forward models more heavily. Further studies using a computational approach have expanded on the idea that human subjects can flexibly alter the weight they attribute to sensory feedback, depending on its reliability, during motor performance (Baddeley et al., 2003; Kording & Wolpert, 2004; Wei & Kording, 2010). Kording & Wolpert (2004) demonstrate that human subjects internally  18  represent the degree of sensory uncertainty in a reaching task where the laterally displaced visual feedback of their hand position was blurred. Interestingly, in cases with increasing visual blurring (uncertainty) subjects were found to weight their estimate of the lateral displacement more on the mean of the prior distribution, and in cases of decreased blurring they weighted their estimate of the displacement more on the visual feedback provided from a given trial. These results demonstrate human subjects may use a Bayesian statistical approach to mitigate the effect of sensory uncertainty on motor performance. In a study by Baddeley & colleagues (2003), the idea of relative weighting of sensory feedback was expanded to show that human subjects maintain near optimal motor performance despite sensory uncertainty. Subjects made pointing movements where the feedback relating the target and cursor was distorted according to three levels of displacement, reflecting thee levels of uncertainty. The subjects’ performance was then compared to the ideal performance of a Kalman filter, which demonstrated that subjects’ performance was highly efficient and the efficiency did not vary across the three levels of sensory feedback distortion. In order to maintain efficient performance across increasing levels of distortion of visual feedback, the data demonstrated that human subjects adjusted their weighting of the sensory error. Thus, ample experimental evidence exists indicating that human subjects indeed take into account the reliability of available sensory input and adjust the weighting of this input depending on its uncertainty during motor performance. 1.4.4  Forward models of the external environment  In addition to predicting the sensory consequences of our own actions, there is evidence indicating the human subjects also compute forward models of the external environment (Merfeld et al., 1999; McIntyre et al., 2001; Zupan et al., 2002; Davidson &  19  Wolpert, 2003; Zago et al., 2004; Davidson & Wolpert 2005; Schubotz, 2007; Zago et al., 2009). Research using interception tasks demonstrate the necessary anticipatory nature of forward models in such a task due to delays in the visuomotor system (Zago et al., 2004; Zago et al., 2009). Zago & colleagues (2004) demonstrate that in addition to the brain’s ability to measure the kinematics of a moving target, it computes a predictive model of target dynamics to track its movement. In the case of intercepting a moving ball in the external environment, internal models of gravity (Merfeld et al., 1999; McIntyre et al., 2001) are used to model object dynamics. Predictive internal models of target movement have been demonstrated in tracking tasks where the target was occluded from vision for a period of time, followed a curvilinear path, and then re-emerged (Mrotek & Soechting, 2007). The subjects’ response was to continue tracking the target along the expected curved path by using an internal model to guide their ocular tracking. These studies demonstrate that the sensory consequences of movement of objects in the external environment, in addition to that of our own movements, are modeled within the CNS and used in tasks requiring predictive and anticipatory motor control. 1.4.5  The cerebellum as an anatomical substrate of forward models  With increasing evidence that the CNS makes use of forward models during motor performance, there has been keen interest in determining their anatomical substrate, and a substantial amount of evidence has accumulated indicating that forward models reside in the cerebellum (Diener et al., 1993; Gao et al., 1996; Jueptner et al., 1997; Kettner et al., 1997; Blakemore et al., 1998; Inoue et al., 1998; Tamada et al., 1999; Kitazawa et al., 1998; Kawato et al., 2003: Imamizu et al., 2003; Bastian, 2006; Ito, 2008; Synofzik et al., 2008). Though the cerebellum is traditionally thought of as part of the motor system, the sensory  20  function of the cerebellum has been more recently emphasized. Jueptner & colleagues (1997) demonstrated through the comparison of active and passive elbow flexion that approximately 90% of the cerebellar regional cerebral blood-flow was related to sensory rather than motor function. The sensory function of the cerebellum was also shown by Gao & colleagues (1996) through fMRI results demonstrating significant cerebellar activity in a purely sensory condition where subjects had sandpaper passively rubbed against their finger tips. Though the exclusively sensory portion of the task activated the cerebellum, it was the sensory discrimination task that demonstrated even greater cerebellar activation, for example when subjects were asked to discriminate between changes in sandpaper coarseness passively applied to subjects’ fingers. Perhaps the most interesting result of the study was that a motor task requiring rapid coordinated finger movements but not requiring sensory discrimination did not significantly depend on cerebellar activity, but the same task where subjects were additionally required to discriminate between objects resulted in the greatest amount of cerebellar activity. The results of these studies support a significant sensory function of the cerebellum, and importantly highlight the sensorimotor integration nature of cerebellar activity. Further investigations into the specific nature of the sensory function of the cerebellum has led to the idea that the cerebellum is tuned to the sensory error – the discrepancy between the predicted and actual sensory consequences of movement. This notion is supported by PET studies (Inoue et al., 1998), and is elegantly demonstrated in a study by Blakemore & colleagues (1998). By comparing tactile stimulation that was either externally produced by a robot or self-produced by the subjects’ hand, the authors demonstrated reduced cerebellar and somatosensory cortex activity using fMRI in the self-produced condition. The authors  21  reasoned that in the self-produced tactile condition the cerebellum is able to accurately predict the sensory consequences of their own movement, and the sensory error between the forward model and the actual sensory feedback is small, explaining the decreased cerebellar activity. This work was followed-up by Blakemore & colleagues (2001), in a study where subjects controlled a robotic arm to produce a tactile stimulus that was applied after various amounts of delay. Their finding that cerebellar activity positively correlated with the amount of delay offers further evidence that the cerebellum predicts the sensory consequences of action, and that the cerebellum is particularly active when there is a mismatch between the prediction and the actual sensory consequences, which in this case was provided by the delay. Electrophysiological studies support the idea of cerebellar sensory error signaling, and the climbing fibers have been proposed as the site of the sensory error signal (Oscarsson, 1980). Electrophysiological studies in the cat demonstrate greater climbing fiber activation in unexpected conditions in movement where there is a mismatch between the expected and actual sensory consequences (Gellman et al., 1985; Andersson et al., 1985; Andersson et al., 1987; Simpson et al., 1996), and further studies support the role of predictive sensorimotor error signaling in the climbing fiber-Purkinje cell synapse (Kettner et al., 1997; Kitazawa et al., 1998). As previously noted, the idea of forward modeling is integral to the process of state estimation (Miall & Wolpert, 1996). The forward model of the predicted sensory consequences of movement in essence provides information about the change in state of the system, and further evidence using TMS has demonstrated that state estimation is processed in the cerebellum (Miall et al., 2007). In a reaching task towards a visual target healthy human subjects had TMS applied to the lateral cerebellum, an effect that rendered reaching  22  errors to be 138 ms out of date, indicating that the temporary inability of the cerebellum to produce forward models led motor performance to rely on visual feedback. Thus, numerous computational studies provide evidence that healthy human subjects use forward models and actively alter the relative weighting of sensory feedback when faced with uncertain visual feedback (Wolpert et al., 1995; Wolpert & Ghahramani, 2000; van Beers et al., 2002; Baddeley et al., 2003; Kording & Wolpert, 2004; Vaziri et al., 2006; Wei & Kording, 2010), and further compelling evidence demonstrates the cerebellum’s essential role in sensorimotor integration and forward modeling, as provided by imaging and TMS studies (Diener et al., 1993; Gao et al., 1996; Jueptner et al., 1997; Blakemore et al., 1998; Inoue et al., 1998; Blakemore et al., 2001; Kawato et al., 2003; Miall et al., 2007) as well as by electrophysiological studies (Oscarsson, 1980; Gellman et al., 1985; Andersson et al., 1985; Andersson et al., 1987; Simpson et al., 1995; Kettner et al., 1997; Kitazawa et al., 1998). 1.5  Parkinson’s disease and visually guided tracking  Visually guided tracking has been previously used to investigate PD. Inadequate anticipation and prediction in PD subjects has been demonstrated during visually guided tracking (Flowers, 1978a). In the tracking task used by Flowers, gaps occurred where the target being tracked was extinguished and subjects were required to continue tracking the predicted position of the target. Interestingly, though PD subjects all reported being able to perceive the track pattern they weren’t able to use the knowledge to accurately predict the path, leaving them much more reliant on visual feedback. Another study by Flowers (1978b) demonstrated that the difference in tracking performance between PD subjects and healthy controls during sinusoidal tracking was greatest when the sinusoid was predictable, and it  23  was speculated that the control subjects were able to use this predictable information to control their movements based on an internal model rather than on visual feedback (Flowers, 1978b). Thus it seems that predictive motor control in PD may be compromised. As previously highlighted, visually guided tracking tasks have been used to investigate dyskinetic PD subjects (Liu et al., 2001), where the results suggested that inadequate predictive control may be a feature of subjects who experience LID, and further research into predictive deficits contributing to LID is warranted. 1.6  Study aims and hypotheses  In summary to the themes elaborated on in the preceding sections of the introduction, PD has long been considered a classic ‘basal ganglia’ disease (Stein & Aziz, 1999), though it is becoming increasingly evident that brain regions beyond the basal ganglia are as well affected in PD. Moreover, an increasing amount of evidence exists in support of functional changes occurring in the cerebellum and its connections in PD (Rascol et al., 1997; Yu et al., 2007; Lewis et al., 2007; Ballanger et al., 2008; Palmer et al., 2009; Koch et al., 2010; Ni et al., 2010), and further investigation into the effect of altered cerebellar activity in PD is warranted. The cerebellum is believed to be an anatomical substrate of the forward model (Blakemore et al., 1998; Inoue et al., 1998; Blakemore et al., 2003; Kawato et al., 2003; Imamizu et al., 2003; Bastian, 2006; Ito, 2008; Synofzik et al., 2008), providing an effective method to mitigate the effect of sensory uncertainty on motor performance (Wolpert & Ghahramani, 2000; van Beers et al., 2002; Vaziri et al., 2006). Extensive research has demonstrated that healthy human subjects can de-weight ambiguous visual feedback during motor performance by instead more heavily relying on internal models (Wolpert et al., 1995; Wolpert & Ghahramani, 2000; van Beers et al., 2002; Baddeley et al., 2003; Angelaki et al.,  24  2004; Kording & Wolpert, 2004; Vaziri et al., 2006; Wei & Kording, 2010). In order to examine the ability of parkinsonian subjects to de-weight ambiguous visual feedback, this thesis focuses on the visually guided tracking movements of PD subjects, as well that of dyskinetic and non-dyskinetic PD sub-groups, while using healthy age-matched subjects as a control group. Visually guided tracking tasks have been shown to strongly activate the cerebellum (Miall et al., 2000; Miall et al., 2001), and represent a useful manner in which to indirectly investigate cerebellar involvement during motor performance in PD. By contrast to visually guided tracking tasks frequently used that require only small movements about a single joint, we have chosen to employ a large amplitude visually guided tracking task requiring movement about the wrist, elbow and shoulder joints, where subjects track a target by standing in front of a large screen and pointing with their index finger. The movement required in our tracking task is 1) more representative of complex, larger scale every-day-life movements, and 2) more suitable for examining the motor performance of dyskinetic PD subjects, as the excessive involuntary movements associated with dyskinesias may preclude the use of fMRI in this subject population due to large head motion. Our task compares baseline conditions, where a circular target smoothly follows a Lissajous path projected on a large screen, with ambiguous conditions where the target ‘jitters’ about the path at various amplitudes while still maintaining the path’s overall trajectory. We have elected to analyze the motor performance of PD and control subjects using two methods. The first method examines the discrepancy between the position of the target being tracked along a Lissajous path and the actual position of the subjects’ index finger during tracking, commonly assessed as the root mean square (RMS) error. RMS error is a  25  reliable method for assessing motor performance and indicates how tracking performance is affected by varying levels of visual ambiguity. For the second method of analysis we compute linear dynamical system (LDS) models of subjects’ tracking as a sensitive marker of motor performance. LDS models provide a dynamical mapping of the sensorimotor transformation required in our task by using the desired tracking trajectory as the input, and subjects’ actual tracking as the output. LDS models therefore allow us to probe the dynamical state of subjects’ tracking behaviour (Cheng & Sabes, 2006). LDS models are being increasingly utilized in sensorimotor studies (Thoroughman & Shadmehr, 2000; Scheidt et al., 2001; Baddeley et al., 2003; Donchin et al., 2003), and prior work from our lab has demonstrated LDS models to be sensitive to differences in tracking performance between parkinsonian and control groups, as well as to the effects of L-dopa on tracking performance, even when significant differences did not exist in RMS error (Au et al., 2010). However, as this previous work did not dichotomize patients with respect to levodopa-induced dyskinesias, we sought to use LDS models as a sensitive assay for detecting differences in motor performance between dyskinetic and non-dyskinetic PD subjects. Furthermore, LDS models provide a rigorous method to test the relative reliance on feedback during tracking while subjects are exposed to increasingly ambiguous visual stimuli. This thesis has two specific aims. The first aim is to assess the susceptibility of parkinsonian subjects, and of dyskinetic and non-dyskinetic PD subjects, to increasingly ambiguous visual stimuli during visually guided tracking. The second aim is to characterize possible subtle differences in the motor performance of these PD sub-groups, while they are off medication, during visually guided tracking. Using LDS models we will specifically examine the damping ratio parameter to quantify the tendency to ‘overshoot’ a desired  26  tracking trajectory, and the decay rate parameter to quantify subjects’ responsiveness to and reliance on ambiguous visual feedback during tracking. Chapter 2 of this thesis addresses the first aim by examining the tracking performance of PD subjects using RMS error, without dichotomizing subjects with respect to LID, and using healthy age-matched subjects as a control group. We hypothesized that the RMS tracking error of PD subjects OFF medication would be significantly more adversely affected by increasing visual ambiguity than that of control subjects. We based this hypothesis on three lines of evidence: 1) the use of the forward model in healthy human subjects to deweight the effect of visual uncertainty on motor performance, 2) the integral role of the cerebellum in the forward model, and 3) the altered cerebellar activity in PD subjects off medication that may affect their use of forward models. Chapter 3 of this thesis uses LDS models to focus on the damping ratio parameter of dyskinetic and non-dyskinetic subjects’ motor performance during non-ambiguous tracking. In this chapter we focused on the differences in damping ratios between these PD sub-groups while they were off medication. This is a novel approach with which to investigate LID as the movement of dyskinetic and non-dyskinetic subjects is clearly differentiated when subjects are actively experiencing involuntary movements, but it was not previously known if their movements differ in the off medication state when LID are absent. Our lab’s prior work demonstrated that L-dopa significantly reduced the damping ratio of PD subjects’ tracking performance (Au et al., 2010). We hypothesized that dyskinetic subjects might already demonstrate significantly lower damping ratios than non-dyskinetic subjects in the off medication state, rendering them more sensitive to the effects of L-dopa medication. Chapter 4 of this thesis uses RMS tracking error in addition to LDS models to examine  27  the effect of uncertain visual feedback on dyskinetic and non-dyskinetic PD subjects, as well as on healthy control subjects. This chapter aims to rigorously quantify differences in the use of predictive and feedback motor control between these groups. As we believe that altered cerebellar activity may contribute to the pathophysiology of LID in PD, we hypothesized that dyskinetic subjects’ RMS error would be significantly more affected by increasing levels of ambiguous feedback, and that the decay rate parameter (indicative of responsiveness to visual feedback and a tendency to attempt to chase the jittering target) would be significantly greater for dyskinetic subjects, as compared to non-dyskinetic and control subjects.  28  Figure 1-1. The classic model of the basal ganglia. This classic model of the basal ganglia shows the direct and indirect striatopallidal pathways and that SNc and VTA neurons provide dopaminergic input to the striatum. SNc = substantia nigra pars compacta; VTA = ventral tegmental area; GPe = globus pallidus external; GPi = globus pallidus internal; SNr = substantia nigra pars reticulata; STN = subthalamic nucleus (from Yin & Knowlton, 2006).  29  Figure 1-2. The therapeutic window. The dark gray sideways triangle demonstrates the therapeutic window in PD. It can be seen from the sideways triangle depicting the therapeutic window that as disease progresses it becomes more difficult to keep patients symptom and side-effect free (from Jankovic et al., 2005).  30  Figure 1-3. The forward model. The forward model predicts the sensory consequences of movement, which is then compared with the actual sensory feedback, the difference of which is known as the sensory discrepancy (from Miall &Wolpert, 1996).  31  CHAPTER 2 RESPONSE TO VISUAL UNCERTAINTY IN PARKINSON’S DISEASE: A MARKER OF CEREBELLAR DYSFUNCTION? 2.1  Synopsis  Motor performance is profoundly influenced by sensory information, yet sensory input can be noisy and uncertain. The basal ganglia and the cerebellum are important in processing sensory uncertainty, as the basal ganglia incorporate the uncertainty of predictive reward cues to reinforce motor programs, and the cerebellum and its connections mitigate the effect of ambiguous sensory input on motor performance through the use of forward models. While Parkinson’s disease (PD) is classically considered a primary disease of the basal ganglia, alterations in cerebellar activation are also observed, which may have consequences for the processing of sensory uncertainty. The aim of this chapter is to investigate the effect of visual uncertainty on motor performance in 15 PD patients and 10 age-matched control subjects. In addressing this aim, subjects performed a visually guided tracking task, requiring large amplitude arm movements, by tracking with their index finger a moving target along a smooth trajectory. To induce visual uncertainty, the target position randomly jittered about the desired trajectory with increasing amplitudes. We found that tracking error was related to target ambiguity to a significantly greater degree in PD subjects off medication compared to control subjects, indicative of susceptibility to visual uncertainty in PD. L-dopa partially ameliorated this deficit. We interpret our findings as being suggestive of an inability of PD subjects to create adequate forward models and/or de-weight less informative visual input. As these computations are normally associated with the cerebellum and connections, we suggest that alterations in normal cerebellar functioning may be a significant contributor to altered motor performance in PD.  32  2.2  Introduction  While sensory input is an integral part of motor control allowing for guidance and feedback during movement, sensory input can be uncertain and difficult to interpret. For example, visual input can become less informative in poor lighting or in conditions of poor visual contrast. In addition, neural responses to sensory stimuli can be noisy, as demonstrated by variable neuronal responses to constant sensory input (Tomko & Crapper, 1974; Shadlen & Newsome, 1998). The central nervous system (CNS) must therefore adapt to varying degrees of sensory uncertainty. One proposed model for this adaptation is the “forward model”, used to predict the sensory consequences of action. Accurate forward models can provide a precise expectation of sensory input from a motor movement. There is ample evidence that the cerebellum and its connections perform calculations similar to that of forward models during motor performance (Miall et al., 1993; Gao et al., 1996; Jueptner et al., 1997; Inoue et al., 1998). Other studies support the notion that humans account for the uncertainty of available sensory input by actively altering the relative importance they attribute to uncertain sensory input and to sensory predictions from forward models (Wolpert et al., 1995; Baddeley et al., 2003; Kording & Wolpert, 2004; Wei & Kording, 2010). Another form of uncertainty faced by the CNS is related to reward and reward expectation. Reward uncertainty represents the reliability with which a particular sensory stimulus predicts reward, and an ambiguous relationship between stimulus and reward can make it more difficult to reinforce desired motor actions. This form of uncertainty is likely processed by the basal ganglia and their cortical connections via reinforcement learning (Doya, 2000). The dopaminergic reward response in the striatum, combined with an estimate  33  of the uncertainty of the predictive reward cues, is a powerful proposed mechanism to reinforce optimal motor actions leading to reward (Daw et al., 2005). Functional changes occurring in both the basal ganglia and the cerebellum in Parkinson’s disease (PD) suggest that PD subjects may have difficulty responding to ambiguous external stimuli, whether used to guide or reinforce motor performance. The classic neuropathology of PD, a neurodegenerative disorder characterized by the clinical symptoms of bradykinesia, resting tremor, rigidity, and postural instability (Calne et al., 1992), involves the degeneration of the substantia nigra pars compacta with a reduction in striatal dopamine levels (Bezard et al., 2001). Beyond dopaminergic circuits there is evidence to suggest that altered cerebellar function may have an important role in the pathophysiology of PD (Lewis et al., 2007; Yu et al., 2007, Koch et al., 2009). Conversely, change in cerebellar function may also be compensatory and/or adaptive in nature (Glickstein & Stein, 1991; Palmer et al., 2009). The main goal of the present study was to investigate the effect of visual uncertainty on tracking performance in PD. Using a visually guided tracking task we experimentally introduced ambiguity to the position of the target to determine if subjects can de-weight the partially non-informative visual stimuli and rely more heavily on forward models of their movements. 2.3 2.3.1  Methods Subjects  We recruited 15 patients with clinically-definite PD (12 male, 3 female, mean age 62.4 ± 6.3 years) and 10 age matched control subjects without active neurological disorders (2 male, 8 female, mean age 61.6 ± 7.9 years). All patients had mild to moderately severe PD  34  (Hoehn and Yahr stage 1-3) (Hoehn & Yahr, 1967) with mean symptom duration of 8.7 ± 4.9 years. All patients were taking levodopa (mean daily dose 746 ± 229.8 mg). Other PD medications included dopamine agonists (10 patients). All patients had overnight withdrawal of medications for at least 12 hours before the study for L-dopa, and 18 hours for dopamine agonists. The mean Unified Parkinson’s Disease Rating Scale (UPDRS) motor score OFFlevodopa was 37.9 ± 20.5. The converted L-dopa daily dosage was calculated as 100 mg Ldopa = 125 mg of controlled-release L-dopa. This was then added to the equivalents dopamine agonists to give the L-dopa equivalent daily dosage (LEDD), where 100 mg of Ldopa = 1 mg of pramipexole, 6 mg of ropinirole, 10 mg of bromocriptine, 75 mg of L-dopa plus entacapone. Subject characteristics are shown in Table 2-1. All subjects gave written informed consent. The study conforms with the Code of Ethics of the World Medical Association (Declaration of Helsinki) and was approved by the Ethics Board of the University of British Columbia. 2.3.2  Study paradigm  Subjects performed a visually guided tracking task by tracking a moving visual target with their index finger that was displayed on a large screen. Subjects stood with their feet at shoulder-width apart and approximately 55 cm in front of a rear projection screen measuring approximately 1.62 m by 1.22 m. Using Matlab (The MathWorks Inc, MA) and the Psychophysics Toolbox Version 3 (Brainard, 1997), a Lissajous figure was created and displayed on the screen with a red circular target (approximately 12 cm in diameter) that started at the center of the screen and figure (Fig. 2-1A). Subjects were asked to track the target with their index finger as close to the screen as possible but without dragging their finger along the screen surface. We visually monitored that subjects were not resting their  35  index finger on the screen throughout the experiment. The motion required to complete this tracking task required both elbow and shoulder movement. Subjects were instructed to utilize the visual feedback relating the position of their index finger and the target to follow the Lissajous figure as accurately as possible. We modified two target parameters throughout the course of the experiment: target speed and target ambiguity. Speed was tested at a SLOW speed (average speed of 56.2 cm/sec) and a FAST speed (average speed of 78.3 cm/sec). In the baseline conditions, the target exactly followed the smooth trajectory of the Lissajous figure displayed on the screen. In subsequent conditions, visual ambiguity was artificially added to the target such that it randomly ‘jittered’ about the Lissajous figure with a predetermined amount of variability. In such cases subjects were instructed not to chase the jittering target about the figure, but rather to attempt to neglect the visually uncertain jitter and to track the target as if it were following the smooth trajectory of the Lissajous figure. Visual ambiguity was set at {0, 0.03, 0.05, and 0.07} representing the amplitude of the jitter with respect to the screen height. There was thus a total of 2 speeds x 4 ambiguity levels = 8 tracking conditions. Careful piloting was done to ensure that PD subjects could successfully track in all conditions. The study consisted of 24 tracking trials, with 8 conditions repeated in three trial sets. Each tracking trial consisted of 30 seconds of tracking, a 12 second rest, followed by 30 seconds of tracking, where the pre and post rest tracking parameters were identical. In between each trial there was approximately a 30 second rest. Three breaks were scheduled throughout the study where the subjects sat down for approximately 5 minutes. In the first trial set, the baseline SLOW condition with no target ambiguity was presented first, followed  36  by the baseline FAST condition with no ambiguity. The remaining six conditions were presented in random order. This sequence was then repeated in trial set 2 and trial set 3. PD patients underwent two studies, one during ‘OFF’ state (PDOFF) and the other after the levodopa challenge (PDON). They were tested in the ‘OFF’ state after overnight withdrawal of L-dopa for at least 12 hours, and after withdrawal of dopamine agonists for at least 18 hours. Though 12 hours of withdrawal of L-dopa may not be sufficient to completely deplete L-dopa from the system, this method for testing PD patients in the OFF state was utilized to minimize the patients’ time spent off medication. At the end of the first study, they were given immediate release L-dopa at a dose equivalent to their usual morning dose of L-dopa. They then had a rest for approximately 45 minutes before repeating the second study, which consisted of the same 24 tracking trials that were completed ‘OFF’ medication. 2.3.3  Quantification of manual tracking  A Polhemus Fastrak (Polhemus, Colchester, VT, USA), a 6-degrees-of-freedom electromagnetic tracking system, was used to record subject tracking. A stylus sensor was used to record a time series for displacement (x, y, and z) and orientation (pitch, roll, and yaw). The stylus was held and secured in the subjects’ dominant hand so that the motion sensor was located at the tip of the index finger. 2.3.4  Signal analysis  The sensor recorded a position time series throughout the course of each trial. The sensor data were irregularly sampled at ~40 Hz, which was then interpolated to provide a 10 Hz time series. For each subject, a robust linear regression analysis was performed on the x and y positions from the index finger sensor during the non-ambiguous tracking conditions to determine the optimum affine transformation to map the sensor readings to the Lissajous  37  figure coordinates. Using the coefficients from baseline non-ambiguous tracking conditions, the same transformation was then applied to the ambiguous tracking conditions. 2.3.5  Quantification of tracking performance  Tracking performance was assessed in two dimensions. Root mean square (RMS) tracking error was calculated by subtracting the processed x and y sensor data of the index finger from the x and y target position along the baseline track, squaring the result for each time point, taking the mean for the squared values for each trial, and taking the square root of the result:  (1) where xi is the difference between the index finger position and baseline track position at time point i. Any potential training effect was assessed by comparing the RMS tracking errors from the three identical trial sets. There was a training effect between trial set 1 and trial set 2 that stabilized between trial set 2 and trial set 3 (Fig. 2-2), and we therefore omitted trial set 1 in all subsequent calculations. 2.3.6  Statistical analyses  RMS tracking error was compared for all groups (PDOFF, PDON, and Control) using two-way analysis of variance (ANOVA) (Group x Condition). Significance was declared at p<0.05. All PD patients were pooled and a Pearson’s correlation was used to determine the existence of a significant relationship between RMS tracking error and age, years since diagnosis, daily L-dopa dose, and UPDRS score. Furthermore a Pearson’s correlation was performed to ascertain any possible relationship between age and UPDRS. We also  38  performed a robust multivariate regression analysis to determine the relative influence of ambiguity amplitude and speed on RMS tracking error for each of the groups. The multivariate regression analysis was in the form of: (2) where Y is the RMS tracking error, X is the design matrix which contained a column of ones to capture the mean, columns representing tracking ambiguity and speed amplitudes, as well as additional columns dummy coding for individual subjects, β are the parameters to be estimated, and ε is a vector of residuals. The above equation was solved with Matlab’s “robustfit” function. 2.4  Results  Subject characteristics are shown in Table 2-1. There was no significant correlation between PD subjects’ age and UPDRS (Pearson’s coefficient = 0.37, p>0.05). Control subjects’ RMS tracking error was not significantly correlated with age in any tracking conditions (magnitude of Pearson’s coefficients < 0.42, p>0.05). PDOFF subjects’ RMS tracking error was not significantly correlated with age or daily L-dopa dose in any tracking conditions (magnitude of all Pearson’s coefficients < 0.5, p>0.05). PDOFF subjects’ RMS tracking error was significantly correlated with UPDRS in conditions {SLOW, 0.05 and 0.07 ambiguity} and {FAST, 0.05 ambiguity} (magnitude of all Pearson’s coefficients <0.60, p<0.05), and with years since diagnosis in tracking condition {SLOW, no ambiguity} (Pearson’s coefficient = 0.6, p<0.05). RMS tracking error of PDON subjects was not significantly correlated with age, UPDRS, daily L-dopa dose or year’s since diagnosis under any tracking conditions (magnitude of Pearson’s coefficients < 0.5, p>0.05).  39  2.4.1  Tracking performance  An example of a typical PD and Control subject’s tracking performance is illustrated in Figure 2-1B. Qualitatively, the tracking performance in condition {FAST, no ambiguity} is similar between the Control subject and PDOFF subject. The effect of ambiguity on this PDOFF subject’s tracking performance is illustrated by the trace from condition {FAST, 0.07 ambiguity}. By contrast, the control subject maintains a relatively good tracking performance despite the ambiguity. 2.4.2  Training effect  A training effect occurred from trial set 1 to trial set 2 for all groups, demonstrated by the small but statistically significant within-group decrease in RMS tracking error (Fig. 2-2). After the initial training effect between trial sets 1 and 2, the training effect stabilized from trial set 2 to trial set 3 as the RMS tracking errors between these trials were not significantly different (p>0.3) for any group. The comparison of group RMS tracking error between trial sets 1 and 2 revealed that PDOFF and Control subjects demonstrated a slightly greater decrease in RMS tracking error than PDON subjects (p<0.05), and PDOFF subjects demonstrated a slightly greater decrease in RMS tracking error than Control subjects (p<0.05). 2.4.3  Effect of ambiguity on RMS tracking error  The effect of visual ambiguity on RMS tracking error, for SLOW and FAST tracking speeds, is illustrated in Figure 2-3A. ANOVA revealed a Group effect (F = 40.99, df = 2, p<0.001) and a Condition effect (F = 83.23, df = 7, p<0.001). The Group x Condition interaction was not significant (F = 1.49, df = 14, p>0.05). Post hoc analyses revealed that all groups’ RMS tracking performance significantly worsened with the addition of ambiguity, as  40  the increase in RMS tracking error from condition {no ambiguity} to conditions {0.03, 0.05 and 0.07 ambiguity}, at SLOW and FAST speeds, was significant for all groups (p<0.05). When examining group differences, PDOFF subjects’ RMS tracking error was significantly worse than that of PDON subjects (paired t-test) (p<0.05) and Control subjects (p<0.05) in all tracking conditions except for condition {SLOW, no ambiguity}, where the differences were not significant (p>0.05). Though in all tracking conditions PDON subjects’ RMS tracking error was still larger than the RMS tracking error of Control subjects, the difference was only significant in condition {FAST, 0.05 ambiguity}. 2.4.4  Effect of medication on RMS tracking error  L-dopa therapy improved tracking performance by partially normalizing the RMS tracking error of PDOFF subjects to that of Control subjects. The medication effect of Ldopa was significant (p<0.05) in all conditions except for condition {SLOW, no ambiguity}, as found by a paired t-test. Figure 2-3B illustrates an example of the medication effect on each PD subject in condition {FAST, no ambiguity} and condition {FAST, 0.07 ambiguity}. 2.4.5  Effect of tracking speed on RMS tracking error  All groups tracking performance significantly worsened with the addition of speed. When comparing SLOW and FAST tracking conditions at the same level of ambiguity, all groups’ RMS tracking performance significantly worsened (p<0.05) in the FAST conditions. 2.4.6  Regression analysis: effect of target ambiguity and speed on RMS tracking error  The overall effects of increasing visual ambiguity and increased speed on RMS tracking error are summarized in Figure 2-3C. The regression analysis indicated that increasing visual ambiguity contributed significantly more to the RMS tracking error of both PDOFF and PDON subjects than Control subjects (p<0.05), indicating sensitivity to visual  41  ambiguity in PDOFF subjects that does not completely normalize with medication. Figure 23C also illustrates the contribution of tracking speed to RMS tracking error for all groups, where again both PDOFF and PDON were significantly (p<0.05) more sensitive to the faster tracking speed than Control subjects, and medication did not completely normalize this sensitivity. All between group differences in the ambiguity and speed regression coefficients were highly significant. 2.5  Discussion  We have investigated how visually guided tracking performance is affected by increasing target ambiguity and speed in 15 PD patients and 10 Control subjects. Various experimental methods of imposing artificial uncertainty on sensory stimuli have been used in sensorimotor tasks in the past, such as blurring of external cues or distorting the distance between cursor and target in pointing tasks (Baddeley et al., 2003; Kording & Wolpert, 2004; Wei & Kording, 2010). Since PD subjects may have varying difficulty with visual contrast, we did not introduce visual ambiguity using contrast or target blurring. Rather, we elected to use a visually guided tracking task where the target that guided the behavioural response of the subjects jittered at varying amplitudes around the desired trajectory. The addition of jitter to the target imposes an instantaneous uncertainty on the target, making it difficult to ascertain the target’s exact position at any given time during the tracking trial. The proposed design therefore probed the ability of subjects to de-weight less meaningfully visual stimuli. We found that in addition to the known sensitivity of tracking error to speed in PD, PD patients are susceptible to progressively ambiguous visual stimuli. Sensitivity to sensory uncertainty may be of critical importance in PD, where known deficits in visual contrast  42  sensitivity (Bodis-Wollner et al., 1987; Price et al., 1992) may significantly compound the problem. Though the addition of ambiguity in visually guided tracking degraded the tracking performance of all groups, PDOFF subjects were significantly more susceptible to ambiguous visual input than Control subjects. There was a trend towards a normalizing effect of L-dopa on sensitivity to ambiguity, though PDON subjects still demonstrated a significantly greater susceptibility than Control subjects. As expected, PDOFF subjects were also more susceptible to increased tracking speed than Control subjects, though L-dopa did not restore speed susceptibility to normal levels. Jointly investigating the effects of speed and ambiguity with a regression model suggests roughly equal effects of ambiguity and speed between normal and PDON subjects, but PDOFF subjects appeared particularly susceptible to ambiguity (Fig. 2-3C). A preliminary training effect was demonstrated by all groups, as indicated by a small but significant decrease in RMS tracking error between trial sets 1 and 2 (Fig. 2-2). This is not motor learning in the classic sense, as the jitter in the target position was pseudorandom, and therefore could not be learned. Nevertheless, RMS tracking error stabilized between trial sets 2 and 3 for all groups, and we therefore subsequently controlled for the preliminary training effect in the analyses by using data from only trials two and three. Our results are therefore reflective of sensitivity to sensory uncertainty in PD after sufficient training, rather than simply of a slower adaptation process to sensory uncertainty in PD. Altered reinforcement learning due to basal ganglia dysfunction in PD is unlikely to explain our results, as PDOFF subjects did not demonstrate deficits in training between trial sets 1 and 2 when compared to Control subjects (Fig. 2-2). In fact, before stabilizing between trials 2 and  43  3, PDOFF subjects demonstrated a slightly augmented training effect between trials 1 and 2, suggesting reinforcement learning in the basal ganglia was not significantly affected. Previous work has shown conflicting results when visual feedback is removed during the motor performance of PD subjects. Klockgether & Dichgans (1994) examined arm movements in PD while manipulating the visual feedback of the moving hand, the target, or both. They found that PD subjects were most impaired when they could not see their hand – consistent with an inability to accurately use forward models. This contrasts the work by Liu and colleagues (1999) who demonstrated that tracking performance of PD subjects did not degrade more than that of control subjects when either the cursor or target was removed during visually guided tracking. While our work is not directly comparable as we degraded visual input rather than completely eliminating it, our findings are most consistent with the finding of Klockgether & Dichgans that during a motor task requiring accurate forward modeling, Parkinsonian motor performance deteriorates to a greater degree than that of normal subjects. The large amplitude movements we utilized, while more realistic to everyday movements performed by PD subjects, are not conducive to brain imaging studies, where less realistic small-amplitude movements must be used to minimize head motion artifact. Nonetheless, during the various conditions of our task, subjects must relate the hand and visual target positions, requiring the coordination between eye and hand movements known to strongly activate the cerebellum (Miall et al., 2000; Miall et al., 2001). Thus, we interpret the susceptibility to sensory uncertainty that we have observed as indicative of altered cerebellar function in PD. Forward models are used by the CNS not only to avoid the inherent delays of sensory feedback, but also to provide a mechanism to contend  44  with the effect of sensory uncertainty on motor performance (for review see Wolpert & Ghahramani, 2000; van Beers et al., 2002). Briefly, a predictive forward model of the sensory feedback of a motor action is computed, and this predicted sensory feedback is subsequently compared to the actual sensory feedback, the discrepancy of which is assessed as sensory error. When confronted with large sensory uncertainty, a forward model provides an alternative input on which the CNS can optimize motor performance. Sensory errors are likely processed by the cerebellum, demonstrated by heightened cerebellar activity when there is a mismatch between actual and predicted sensory feedback resulting from an externally produced stimuli (Blakemore et al., 1998). Furthermore, evidence supports the role of the cerebellar climbing fiber-Purkinje cell synapse in signaling predictive error in sensorimotor tasks (Kettner et al., 1997; Kitazawa et al., 1998), and the cerebellum is widely thought to be a biological substrate of forward models (Diener et al., 1993; Miall et al., 1993; Gao et al., 1996; Jueptner et al., 1997; Inoue et al., 1998). Though the cerebellum has an integral role in timing, we note that the susceptibility to ambiguity was determined independently of speed within a regression framework (Fig. 23C). Moreover, though cerebellar patients have been shown to have deficits on tasks that require discrete timing events such as finger tapping, they do not demonstrate such deficits in timing tasks requiring smooth continuous movements (Spencer et al., 2003). We therefore do not interpret our results as related to deficits in cerebellar timing, but rather as that of impaired forward modeling. Adequate cerebellar forward models need to account for changes in properties of the limb such as inertia (Miall & Wolpert, 1996). In PD, limb rigidity will alter the effective mechanical impedance of the limb – the response to movement perturbation (Wang et al.,  45  2001). Changes in limb impedance and consequently the limb dynamics may provide the basis for inadequate cerebellar forward models in PD. An inability of the cerebellum in PD to incorporate the altered effective mechanical impedance into its forward models may lead to a mismatch between the actual limb dynamics and the forward models based on non-rigid limbs. Evidence demonstrates that PD subjects excessively rely on sensory feedback (Cooke et al., 1978), and inadequate forward models incapable of accurately guiding motor performance may render PD patients reliant on visual feedback and susceptible to visual ambiguity. Our results demonstrate that L-dopa partially counteracts susceptibility to visual ambiguity in PD. While it is possible that subjects could be considered sub-optimally treated after their dosage of L-dopa (they did not receive dopamine agonists between the ON and OFF studies even if they were normally treated with them), we believe that this would have a relatively minor impact. The use of dopamine agonists is usually for minimization of either developing or inducing L-dopa induced dyskinesias, rather than to enhance motor performance. Rather, we interpret the partial restoration of susceptibility to visual ambiguity to be on the basis of plastic changes such as compensatory mechanisms that may not be completely reversed upon optimal dopaminergic treatment. While the mechanisms through which L-dopa therapy may partially restore cerebellar function to partially normalize susceptibility to visual ambiguity is not known, we propose four possible mechanisms. Firstly, animal studies have demonstrated high densities of the dopamine D3 receptor in the cerebellum (Sokoloff et al., 1990; Bouthenet et al., 1991), providing a direct cerebellar site of action for L-dopa. Secondly, L-dopa may indirectly affect cerebellar function via the basal ganglia, as recent evidence demonstrates the existence  46  of disynaptic connections between the cerebellum and basal ganglia (Hoshi et al., 2005; Bostan et al., 2010). Bostan and colleagues (2010) have shown a direct pathway from the subthalamic nucleus of the basal ganglia to the cerebellar cortex, providing a neuroanatomical pathway by which affected basal ganglia activity in PD may alter cerebellar output such as forward models. L-dopa therapy may therefore correct aberrant basal ganglia output to the cerebellum, thereby restoring cerebellar function. Thirdly, evidence suggests the cerebellum may be recruited in a compensatory manner for the basal ganglia in PD (Lewis et al., 2007; Yu et al., 2007; Palmer et al., 2009). Thus L-dopa normalization of the basal ganglia may lead to decreased cerebellar compensatory demand, allowing for an indirect, partial return to normal cerebellar activity. Lastly, it may be that in restoring the mechanical properties of the limb from the Parkinsonian state to the normal state, L-dopa partially restores the match between cerebellar forward models and limb dynamics. The poor ability of PD subjects to de-weight ambiguous stimuli and/or generate accurate forward models is in contrast to control subjects who maintain this ability. The current results and other studies suggest that normal subjects can alter the amount of weight they place on uncertain sensory input during performance of sensorimotor tasks (Wolpert et al., 1995; Baddeley et al., 2003; Kording & Wolpert, 2004; Wei & Kording, 2010). By specifically adjusting the weighting of the sensory error in cases of increased sensory uncertainty, normal human subjects are able to maintain near optimal motor performance in a reaching task (Baddeley et al., 2003). Control-based Kalman filters (Kalman, 1960), used in engineering to adjust the weight attributed to feedback and feedforward circuits in order to obtain an optimal response, have been proposed as a mechanism through which humans mediate sensory uncertainty (Wolpert et al., 1995; Baddeley et al., 2003).  47  The Kalman filter may provide a conceptual model to consider altered cerebellar function in PD. The inability to de-weight uncertain sensory input and/or rely more on a predictive forward model in PD may be succinctly interpreted as an impaired Kalman filter. We have successfully used system identification analysis to assess the nature of Parkinsonian movement during a tracking task, demonstrating that model parameters are sensitive for detecting PD-associated abnormalities (Au et al., 2010). Further analysis using system identification models and Kalman filtering to examine the dynamics of PD tracking in response to uncertainty may allow for a direct quantification of the ability of PD subjects to devalue uncertain sensory input. 2.5.1  Conclusion  Our results indicate that tracking accuracy in PD subjects is not only sensitive to speed, but also susceptible to ambiguous visual stimuli. These results likely indicate a change in cerebellar function in PD, as the cerebellum plays a key role in processing sensory uncertainty through the use of forward models. We suggest that a Kalman filter scheme may prove a useful conceptual framework in which to assess altered cerebellar function in PD.  48  Years Since  Daily L-dopa  Other Parkinson’s  L-dopa Equivalent  Dosage (mg)  Medications  Dose (mg)  67  650  ropinirole, amantadine  750  7  42  880  entacapone, amantadine  1173.3  63  5  8  320  pramipexole  620  4  68  4  19  400  None  400  5  68  13  51  660  entacapone  880  6  64  9  69  860  None  860  7  59  9  14  740  None  740  8  45  4  11  780  None  780  9  65  9  51  640  entacapone, pramipexole  1003.3  10  65  15  57  720  entacapone  960  11  66  5  45  1020  None  1020  12  64  4  22  1280  pramipexole  1580  13  63  10  54  800  pramipexole  1000  14  51  7  37  800  bromocriptine  1000  15  66  7  22  640  ropinirole  673.3  62.4 ± 6.3  8.7 ± 4.9  37.9 ± 20.5  746 ± 229.8  Subject #  Age  1  65  22  2  64  3  PD (Mean ± S.D.) Control (Mean ± S.D.)  Diagnosis  UPDRS  896 ± 271.2  61.6 ± 7.9  Table 2–1. Description of PD subjects’ characteristics. (from Stevenson et al., 2011)  49  Figure 2-1. The visually guided tracking task and performance. A) The task. The target and Lissajous figure were displayed on the screen. The actual size of the screen was 1.62 m by 1.22 m. The task began with the red target (with scaled red, green, blue values specified as [1 0 0]) in the center of the gray Lissajous figure (specified as [0.973 0.973 0.973]) on a light background (specified as [0.957 0.957 0.957]). When the subjects were ready each trial began by initiating the task from a computer situated behind the subject. B) Tracking performance: An example of tracking performance of a Control subject and a PDOFF subject during tracking conditions {FAST, no ambiguity} and {FAST, 0.07 ambiguity}. The dashed line represents the desired trajectory and the solid lines represent the trajectory of the subject’s index finger. Top left: Tracking performance of Control subject 5 during condition {FAST, no ambiguity}, RMS tracking error = 5.6. Bottom left: Tracking performance of Control subject 5 during condition {FAST, 0.07 ambiguity}, RMS tracking error = 11.0. Top right: Tracking performance of PDOFF subject 5 during condition {FAST, no ambiguity}, RMS tracking error = 7.7. Bottom right: Tracking performance of PDOFF subject 5 during condition {FAST, 0.07 ambiguity}, RMS tracking error = 20.7 (from Stevenson et al., 2011).  50  Figure 2-2. The training effect. All groups demonstrated a small but significant (p<0.05) decrease in RMS tracking error between trial sets 1 and 2. There were no significant differences (p>0.3) in the RMS tracking error for any group between trials 2 and 3, indicating a stabilization of the training effect. PDOFF and Control subjects demonstrated a small but significantly (p<0.05) greater decrease in RMS tracking error between trials 1 and 2 than PDON subjects, where the decrease was slightly but statistically greater (p<0.05) for PDOFF subjects than for Control subjects. The error bars represent the standard errors (from Stevenson et al., 2011).  51  Figure 2-3. Effect of speed and ambiguity on RMS error. A) The effect of target ambiguity on RMS error. Left panel: SLOW tracking speed. Right panel: FAST tracking speed. The error bars represent the 95% confidence intervals. PDOFF subjects’ RMS tracking error  52  was significantly higher (p<0.05) than Control subjects’ RMS tracking error in all conditions except for condition {SLOW, no ambiguity}, where the difference was not significant (p>0.05). PDON subjects’ RMS tracking error was significantly higher (p<0.05) than Control subjects’ RMS tracking error only in condition {FAST, 0.05 ambiguity}. A paired t-test indicated a significant difference (p<0.05) between PDOFF and PDON subjects’ RMS tracking error for all tracking conditions except condition {SLOW, no ambiguity}. B) The effect of medication on RMS error. Left Panel: Tracking condition {FAST, no ambiguity}. Right panel: Tracking condition {FAST, 0.07 ambiguity}. Individual subjects’ RMS tracking errors are illustrated. C) The regression analysis demonstrating the effect of target speed and ambiguity on RMS tracking error. The group differences in ambiguity and speed regression coefficients are highly significant (p<0.05). PDOFF subjects, in addition to demonstrating sensitivity to increased target speed, also demonstrate heightened sensitivity to increasing target ambiguity. This sensitivity partially normalizes with L-dopa therapy. The vertical and horizontal radii of the ellipses represent the standard error of the regression coefficients (from Stevenson et al., 2011).  53  CHAPTER 3 DYSKINETIC PARKINSON’S DISEASE PATIENTS DEMONSTRATE MOTOR ABNORMALITIES OFF MEDICATION 3.1  Synopsis  The pathophysiology of L-dopa-induced dyskinesias (LID) in Parkinson’s disease (PD) remains poorly understood. The presence of superimposed LID clearly differentiates motor performance of dyskinetic from non-dyskinetic PD subjects when they are on medication, but in this chapter we investigate whether their respective motor performance differs while subjects are off L-dopa medication and LID are not apparent. We assessed the motor performance of 9 dyskinetic and 10 non-dyskinetic PD subjects off L-dopa, and of 10 age-matched control subjects, during a visually guided tracking task. As previous studies have suggested that linear dynamical system (LDS) models are useful to assess motor performance in PD, in addition to overall tracking error, we used LDS models to assess the damping ratio parameter of motor behaviour while controlling for disease severity. The results of this chapter show that while overall tracking error was similar across groups, dyskinetic PD subjects demonstrated a significantly decreased mean damping ratio compared to control and non-dyskinetic PD subjects. For both groups, greater disease severity significantly predicted a lower damping ratio, but even after controlling for disease severity, the damping ratio for dyskinetic subjects was significantly lower. We demonstrate somewhat counter-intuitively, that motor performance of dyskinetic and non-dyskinetic PD subjects differ, even off L-dopa when no dyskinesias are seen. A decreased damping ratio is indicative of a tendency to overshoot a target during motor performance, similar to the dysmetria found in cerebellar patients. We discuss the possibility of motor abnormalities in dyskinetic PD patients off medication in relation to altered functional cerebellar changes described in PD.  54  3.2  Introduction  Despite the common occurrence of peak-dose LID in PD (Fahn, 2000), their pathophysiological mechanisms remain unclear. There are, however, clinical observations that indirectly suggest LID may be associated with plastic changes in the brain. LID are more common in younger individuals (Golbe, 1991; Kumar et al., 2005), whose brains are presumably more likely to form adaptive (and maladaptive) connections. Furthermore, animal models of PD suggest that unnatural pulsatile stimulation of dopaminergic receptors, as may occur with intermittent dosing of L-dopa, may induce plastic changes and may be a contributing factor in the development of LID (Calon et al., 2000a; Calon et al., 2000b; Aubert et al., 2005), although recent clinical trials have yet to support this (Stocchi et al., 2010). While LID occur when patients are on medication, if persistent plastic changes contribute to their pathophysiology, it may be possible to detect subtle changes in motor performance between dyskinetic and non-dyskinetic subjects off medication, despite their virtually identical presentation in the “off” state. Previous work has suggested that during manual tracking tasks, PD subjects on and off medication may have similar overall tracking performance, but differ in the nature of their tracking (Au et al., 2010; Oishi et al., 2010). This has prompted the use of system identification techniques, which can be used to more accurately characterize tracking performance. In this case, human motor performance is considered a “black box”, where the desired trajectory is considered as the input, and actual motor performance is considered as the output. Even with this simple macroscopic approach, robust differences between PD and control subjects, and the effects of L-dopa medication, can be determined and represent an advance over standard, even more basic measures of tracking behaviour, such as the error between desired and actual performance (Au et al., 2010; Oishi et al., 2010). 55  The simplest and most common way to model the relationship between desired and actual tracking performance involves estimating the “damping ratio” and “natural frequency”. The damping ratio provides an indication of the amount overshoot in the response, with lower values indicating more overshoot, and the natural frequency roughly indicates how quickly the actual tracking oscillates around the desired trajectory. Using this method, Au et al., (2010) demonstrated that overall tracking performance (as assessed by tracking error) was not significantly different between PD subjects before and after L-dopa medication, but the damping ratio parameter was a sensitive marker of the motor effects of L-dopa medication. More specifically, L-dopa rendered PD subjects more underdamped during tracking performance, intuitively meaning that subjects were more likely to overshoot and oscillate around a desired target or trajectory. This previous work did not dichotomize dyskinetic and non-dyskinetic PD subjects, and therefore we wished to investigate whether dyskinetic PD subjects exhibit underdamped motor performance in a similar fashion to non-dyskinetic subjects. We hypothesized that dyskinetic subjects off medication would demonstrate subtle differences in motor performance, owing to persistent plasticity thought to contribute to LID (Troiano et al. 2009), that could be characterized by reduced damping ratio during a visually guided tracking task. 3.3 3.3.1  Methods Subjects  In addition to the 15 PD subjects and 10 age matched control subjects recruited for chapter 2, we recruited 4 more PD subjects for the completion of the work in this chapter. We note that the subjects who participated in chapter 2 did not repeat the tracking task for chapter 3, but rather we utilized their original raw tracking data set for this chapter. In total, 19 patients with clinically-definite PD (Hoehn and Yahr stage 1-3) (Hoehn & Yahr, 1967), 9 56  dyskinetic PD (DPD) subjects (7 male, 2 female, mean age 63.7 ± 7 years) and 10 nondyskinetic PD (NDPD) subjects (6 male, 4 female, mean age 61.4 ± 6.4 years), and 10 age matched control subjects without active neurological disorders (2 male, 8 female, mean age 61.6 ± 7.9 years), completed this study. Patients were recruited from the Pacific Parkinson’s Research Centre, a tertiary care centre specializing in movement disorders. Dyskinetic subjects were selected by the presence of LID noted by the treating neurologists (MJM) upon their regular scheduled visit to the centre, and in all subjects this was verified by the presence of predominantly choreiform dyskineisas upon administration of L-dopa during the testing phases (see below). All dyskinetic PD subjects were interpreted to have peak-dose, but not biphasic L-dopa-induced dyskinesias. The Ethics Board of the University of British Columbia approved the study, which has been performed in accordance with the ethical standards of the 1964 Declaration of Helsinki. All subjects gave written, informed consent. The converted L-dopa daily dosage was calculated as 100 mg L-dopa = 125 mg of controlled-release L-dopa. This was then added to the equivalents dopamine agonists to give the L-dopa equivalent daily dosage (LEDD), where 100 mg of L-dopa = 1 mg of pramipexole, 6 mg of ropinirole, 10 mg of bromocriptine, 75 mg of L-dopa plus entacapone. All patients had overnight withdrawal of medications for at least 12 hours before the study for L-dopa, and 18 hours for dopamine agonists. Disease severity was assessed according to the Unified Parkinson’s Disease Rating Scale (UPDRS) motor score. After completing the visually guided tracking task PD subjects took their equivalent morning dose of L-dopa in the immediate release form. Subjects were then monitored for up to 90 minutes for the presence of peak-dose LID. Subject characteristics are shown in Table 3-1.  57  3.3.2  Study paradigm  Subjects performed visually guided tracking in trials consisting of 30 seconds of tracking, a 12 second rest, followed by 30 seconds of tracking, in an identical manner and using the same task as described in the previous chapter. As this chapter of the thesis examines only the SLOW speed non-ambiguous tracking condition of our task, we subsequently analyzed 3 tracking trials for each subject, except for subject DPD 9 who complained of fatigue and completed only two tracking trials, and recorded subject’s arm position using the Polhemus Fastrak (Polhemus, Colchester, VT, USA), as previously described. 3.3.3  Quantification of tracking performance  As in chapter 2, root mean square (RMS) tracking error was calculated by subtracting the processed x and y sensor data of the index finger from the x and y target position along the baseline track, squaring the result for each time point, taking the mean for the squared values for each trial, and taking the square root of the result. As there was again a training effect specific to the non-ambiguous slow tracking condition between trial set 1 and trial set 2, that subsequently stabilized (as determined by significant changes in RMS tracking errors), we omitted trial set 1 in all subsequent calculations. Linear dynamical system (LDS) models of subjects’ tracking performance were computed using system identification techniques (Ljung, 1999) with the desired tracking trajectory as the input, and the subjects’ index finger position as the output. From these models the damping ratio and natural frequency parameters were extracted for each subject. Models were chosen by calculating a loss function (e.g., Akaike information criterion, or Akaike final prediction error) that indicate how well a model fits the data it was constructed  58  from, while penalizing the use of an excessive number of model parameters (to prevent overfitting). Lower loss function values indicate a relatively better fit to the experimental data and hence higher model quality, as there is less ‘loss’ of information from the experimental data. 3.3.4  Statistical analyses  RMS tracking error was compared for all groups (DPD, NDPD and Control) using analysis of variance (ANOVA). ANOVA was performed on the group damping ratios and natural frequencies to determine the statistical significance between group differences. Significance was declared at p<0.05 after correcting for multiple comparisons. In order to ascertain that the derived damping ratios were reflective of the dyskinetic state in PD, and not attributable to differences in LEDD, disease severity (UPDRS), or disease duration between groups, we performed a multivariate regression analysis on the contribution of these factors to the damping ratio values of the DPD and NDPD groups. 3.4  Results  There were no significant differences in mean age (p = 0.75), years since diagnosis (p = 0.16), UPDRS (p = 0.14) or LEDD (p = 0.062) between all groups, though converted daily L-dopa dosage was significantly higher in the DPD group than in the NDPD group (p = 0.047) (Table 3-1). ANOVA revealed that the RMS tracking error did not significantly differ between groups (Control, DPD, NDPD, F = 2.5, p = 0.1), indicating that despite the parkinsonian symptoms of both DPD and NDPD groups, PD subjects were able to perform the tracking task with the same level of overall accuracy as control subjects.  59  The mean and standard deviation of the estimated linear dynamical system models’ Akaike's Final Prediction Error (FPE) was 0.24 ± 0.11 for DPD subjects, 0.28 ± 0.33 for NDPD subjects and 0.20 ± 0.08 for Control subjects. The absolute value for the mean and standard deviation of the Akaike's Information Criterion (AIC) of the models was 1.48 ± 0.4 for DPD subjects, 1.57 ± 0.72 for NDPD subjects and 1.69 ± 0.4 for Control subjects. The low FPE and AIC values are indicative of high model quality/fit. ANOVA revealed a significant group difference in the mean damping ratio (Control, DPD, NDPD, F = 11.4, p = 3x10-4) (Fig. 3-1A). Post hoc analysis revealed that the DPD subjects’ mean damping ratio was significantly lower than that of NDPD subjects (p = 4.1x10-5) and Control subjects (p = 5.4x10-3). NDPD subjects’ mean damping ratio was higher than that of Control subjects, though not reaching significance (p = 0.14). Figure 3-2 demonstrates the relationship between the underdamped motor responses of DPD subjects and the tendency to overshoot the desired trajectory. ANOVA revealed that differences in mean natural frequency did not reach statistical significance between groups (Control, DPD, NDPD, F = 2.32, p = 0.12). Though the differences in disease severity, duration and LEDD were not significantly different between the DPD and NDPD groups, we nonetheless performed a regression analysis to further control for the contribution of these factors to the differences in mean damping ratios between groups. Regression analysis of these factors independent of one another indicated that neither disease duration (p = 0.91) nor LEDD (p = 0.50) significantly predicted damping ratio, whereas the presence of dyskinesias (p = 9.9x10-5) and UPDRS (p = 1.7x10-3) were significant predictors of damping ratio. Analyzing the effect of UPDRS on damping ratio separately for dyskinetic and non-dyskinetic groups demonstrated a significant  60  regression model in both cases, p = 8.4x10-3 for dyskinetic and p = 0.024 for non-dyskinetic groups (Fig. 3-3). Upper and lower panels of Figure 3-3 demonstrate that a higher UPDRS significantly predicts a lower damping ratio for both PD groups, and that a given UPDRS predicts a lower damping ratio for dyskinetic than for non-dyskinetic subjects. The lower panel in Figure 3-3 is for graphical purposes only to illustrate the effect of UPDRS on damping ratio when the mean UPDRS of the dyskinetic group has been ‘corrected’ to equal the mean UPDRS of the non-dyskinetic group, and it can be seen that the difference of damping ratio between dyskinetic and non-dyskinetic groups is a robust effect. 3.5  Discussion  Clinically, dyskinetic PD patients are easily differentiated from non-dyskinetic patients when they are on L-dopa medication. Kinematic studies utilizing accelerometers placed on the limbs of parkinsonian subjects have characterized the nature of LID, enabling dyskinetic movements to be distinguished from volitional movements through careful signal analysis (Hoff et al., 2001; Keijsers et al., 2003; Liu et al., 2005; Gour et al., 2007; Fenney et al., 2008; Chelaru et al., 2010; Mann et al., 2010). In contrast, off medication, both dyskinetic and non-dyskinetic PD patients demonstrate typical parkinsonian symptoms so that their movements appear grossly similar, if not identical. This is consistent with our results that off medication, tracking error of DPD and NDPD subjects did not differ significantly. Yet despite the similarity in overall accuracy of DPD and NDPD subjects’ motor performance off L-dopa, DPD subjects demonstrated significantly decreased damping ratios during tracking. Previous work has suggested that PD subjects off medication tend to be more damped than controls (Au et al., 2010), consistent with our finding that NDPD subjects  61  had a greater mean damping ratio than control subjects, though this did not reach statistical significance. In the current study, the decreased damping ratio of DPD subjects compared to control and to non-dyskinetic subjects even while off medication suggests motor abnormalities exist in the off-medication state in these individuals. Our results indicate very little overlap between the individual damping ratios of dyskinetic and non-dyskinetic subjects, aside from the slightly lower damping ratios of two non-dyskinetic subjects (Fig. 31B). As LID typically occur with disease progression, it may be the case that these two subjects who are non-dyskinetic at present may develop LID in the near future. Damping ratio is intuitively interpreted as the tendency to oscillate about a desired trajectory, where decreased damping ratios indicate more overshoot and oscillation. This is illustrated in Figure 3-2, in which modeled motor performance indicates dyskinetic subjects demonstrate greater overshoot to a step response than non-dyskinetic and control subjects. Conversely, a higher damping ratio leads to undershoot of a desired trajectory (Dancause et al., 2002), a characteristic of motor performance commonly observed in PD subjects off medication (Van Gemmert et al., 2003; Broderick et al., 2009). We propose two possible explanations for our results. Firstly, the abnormalities of dyskinetic tracking performance demonstrated in our study may relate to altered cerebellar activity. Cerebellar patients demonstrate underdamped motor performance (Morrice et al., 1990; Aisen et al., 1993; Manto et al., 2009), and overshoot is a commonly seen in cerebellar patients as dysmetria (Hore et al., 1991; Manto et al., 1995; Brown et al., 1999). Thus, we speculate abnormal cerebellar activity may lead to decreased damping (and overshoot) of the dyskinetic PD subjects tracking performance found here. Consistent with this hypothesis, there is growing indirect evidence implicating the cerebellum in the pathology of LID.  62  Repetitive transcranial magnetic stimulation to the cerebellum has been found to significantly reduce the severity of LID (Koch et al., 2009), and the binding potential of cerebellar sigma receptors in PD patients before undergoing pallidotomies has been positively correlated with LID scores but not with disease severity (Nimura et al., 2004). Moreover, cerebellar hyperactivity has been demonstrated in PD (Yu et al., 2007), which may be recruited to compensate for the affected basal ganglia (Lewis et al., 2007; Ballanger et al., 2008; Palmer et al., 2009), and further evidence demonstrates abnormal cerebellocoritcal connections in PD (Ni et al., 2010). Forward models used to predict the sensory consequences of movement, at least partially computed in the cerebellum (Miall et al., 1993; Jueptner et al., 1997), may be compromised in PD (chapter 2), further implicating altered cerebellar activity in PD. We hypothesize that maladaptive cerebellar plasticity in dyskinetic PD subjects while off medication may further predispose these subjects to an exaggerated motor effect of L-dopa. Disynaptic pathways between the cerebellum and basal ganglia have been identified (Hoshi et al., 2005; Bostan et al., 2010) providing a direct route for altered cerebellar plasticity to interact with the basal ganglia upon L-dopa administration. Alternatively, decreased damping in dyskinetic subjects may be related to functional changes in the basal ganglia. Co-contraction of agonist and antagonist muscles, as has been demonstrated in PD (Ohye et al., 1965) can alter the damping ratio (Kaji et al., 2005), and this may have differed between the DPD and NDPD groups. Though we did not measure electro-myographic activity, the rigidity sub-scores from the UPDRS, which would likely be altered with differing levels of co-contraction, did not significantly differ between PD groups (p = 0.16).  63  A potential limitation of this study is the trend towards higher UPDRS, disease duration and LEDD in the DPD group, though none of these differences were statistically different. We further do not suspect that these differences significantly impacted the damping ratio results as regression analysis indicated that neither the LEDDs nor the disease duration significantly contributed to the groups’ damping ratios. Regression analysis determined that higher UPDRS scores significantly predicted lower damping ratios for both dyskinetic and non-dyskinetic groups (Fig. 3-3), and it can be seen that a given UPDRS score predicts lower damping ratios for dyskinetic than non-dyskinetic subjects. Thus, the relationship between UPDRS and damping ratio differs in magnitude between dyskinetic and non-dyskinetic groups. Additionally, non-motor complications of PD were not quantified in this study, though off periods may be associated with significant non-motor complications (Chaudhuri & Schapira, 2009). Dyskinesias can be related to pain experienced by PD patients (Quinn et al., 1986), and such non-motor symptoms could potentially impact differences in motor performance between dyskinetic and non-dyskinetic groups. Patients were carefully monitored throughout testing to ensure the comfort (necessitating one subject to perform only 2 out of 3 trials because of fatigue). None of our patients complained of painful dystonia off of medication, and none indicated any discomfort that would warrant termination in the study. Though we studied PD subjects in the practically defined off medication state, with withdrawal of dopamine agonists and L-dopa for 18 hours and 12 hours respectively, we cannot exclude the possibility there were not still residual dopamine mediated effects in the off state. Nevertheless, clinically all PD subjects were in their “off” state. Withdrawal of medications for longer periods could probably not be justified on ethical grounds.  64  In conclusion, using linear dynamical system models as a sensitive assay of motor performance, we observed motor differences between dyskinetic and non-dyskinetic PD patients while both groups were off L-dopa medication, suggesting a fundamental functional restructuring of the motor system in dyskinetic subjects even when dyskinesias are not occurring. Dyskinetic subjects were found to have significantly lower damping ratios, leading to a tendency to overshoot the desired trajectory. As underdamped motor responses and overshooting a desired target are common cerebellar features, we suggest maladaptive cerebellar plasticity may be involved in the development of LID in PD. However, in order to fully test this hypothesis, simultaneous recording of tracking tasks with brain imaging methods (e.g. fMRI) will need to be performed.  65  Subject DPD D1 D2 D3 D4 D5 D6 D7 D8 D9 DPD (Mean ± S.D.) NDPD ND1 ND2 ND3 ND4 ND5 ND6 ND7 ND8 ND9 ND10 NDPD (Mean ± S.D.) Control (Mean ± S.D.) p value  Age  Years Since Diagnosis  UPDRS  Converted Daily L-dopa Dosage (mg)  65 64 68 65 66 64 51 55 75 63.7 ± 7  22 7 13 15 5 4 7 13 8 10.4 ± 5.8  65 42 51 57 45 22 37 40 47 45.1 ± 12.3  63 68 64 59 45 65 63 66 62 59  5 4 9 9 4 9 10 7 5 12  61.4 ± 6.4  Other Parkinson’s Medications  L-dopa Equivalent Dose (mg)  650 880 660 720 1020 1280 800 640 600 805.6 ± 223.3  ropinirole, amantadine entacapone, amantadine Entacapone entacapone None pramipexole bromocriptine pramipexole, amantadine none  750 1173.3 880 960 1020 1580 1000 665 600 958.7 ± 296.3  8 19 69 14 11 51 54 22 31 47  320 400 860 740 780 640 800 640 400 400  pramipexole None None None None entacapone, pramipexole pramipexole ropinirole None pramipexole  620 400 860 740 780 1003.3 1000 673.3 400 775  7.4 ± 2.8  32.6 ± 21.2  598 ± 200.3  725.2 ± 211.3  0.16  0.14  0.047  0.062  61.6 ± 7.9 0.75  Table 3–1. Description of dyskinetic and non-dyskinetic subjects’ characteristics.  66  #"  $"  1  !"  0.9  0.9  0.7  Damping Ratio  Damping Ratio  0.8  1  !"  0.6 0.5 0.4 0.3 0.2  0.8  0.7  0.6  0.5  0.1 0  Control  Dyskinetic  Nondyskinetic  0.4  Nondyskinetic  Dyskinetic  Figure 3-1. Damping ratio by group. A) Error bars represent the standard error and an asterisk indicates statistical significance (p < 0.05). DPD subjects had a significantly lower mean damping ratio than NDPD subjects (p = 4.1x10-5) and than Control subjects (p = 5.4x10-3). Though NDPD subjects demonstrated a greater mean damping ratio than Control subjects, this did not reach significance (p = 0.14). B) Individual damping ratios are shown for dyskinetic and non-dyskinetic subjects.  67  1.4 1.2  Amplitude  1 0.8 0.6 0.4 Control Dyskinetic Non-dyskinetic  0.2 0 -1  -0.5  0  0.5  1  1.5  2  2.5  3  3.5  4  Time (sec) Figure 3-2. LDS model simulation results: dyskinetic tracking overshoot. A key feature of dynamical system models is that they can predict the response to tracking any arbitrary trajectory, such as abrupt changes in tracked trajectory, as they are most likely to bring out undershoot and overshoot responses. If subjects were asked to track a target that had an abrupt change in position (solid black line), then the predicted finger position of different subject groups are shown. The error bars are derived from the variances of the parameters computed from the subject-specific models.  68  Legend! * Non-dyskinetic! o Dyskinetic! 1 0.8  Damping Ratio!  0.6  0.4  0  10  20  30  40  50  60  70  50  60  70  UPDRS (Raw data) 1 0.8  0.6  0.4  0  10  20  30  40  Mean-corrected UPDRS Figure 3-3. Damping ratio and UPDRS. Top panel: The effect of UPDRS on damping ratio on dyskinetic and non-dyskinetic subjects, where higher UPDRS significantly predicts lower damping ratio for dyskinetic (p = 8.4x10-3) and for non-dyskinetic (p = 0.024) subjects. Bottom panel: A graphical illustration of the relationship between UPDRS and damping ratio when the mean UPDRS for dyskinetic subjects has been corrected to be equal to the mean UPDRS of nondyskinetic subjects.  69  CHAPTER 4 IMPROVEMENT AT A PRICE: L-DOPA INDUCES EXCESSIVE RELIANCE ON AMBIGUOUS VISUAL INPUT IN PARKINSON’S SUBJECTS WITH DYSKINESIA 4.1  Synopsis  When faced with visual uncertainty during motor performance, humans rely more heavily on predictive forward models and proprioception and attribute lesser importance to ambiguous visual feedback. Conversely, deficits in predictive forward models lead to increased reliance on sensory feedback. Though disrupted predictive control is typical of patients with cerebellar disease, the involuntary nature of L-dopa-induced dyskinesias in Parkinson’s disease (PD) suggests that dyskinetic subjects’ may also demonstrate impaired predictive motor control. In chapter 4 we investigate the motor performance of 9 dyskinetic and 10 non-dyskinetic PD subjects on and off dopaminergic medication, and of 10 agematched control subjects, during a visually guided tracking task. We tested subjects’ ability to de-weight ambiguous visual feedback by introducing ‘jitter’ to a target that followed a Lissajous path. We calculated root mean square (RMS) tracking error, and performed a robust multivariate linear regression analysis to determine the contribution of tracking speed and ambiguity to RMS error. We further computed linear dynamical system (LDS) models of subjects’ tracking data to characterize the decay rate, an indicator of the relative reliance on the ambiguous jitter during tracking. The results of this chapter show that increasing target ambiguity and speed contributed significantly more to the RMS error of dyskinetic subjects off medication than all other groups. L-dopa improved the RMS tracking performance of both PD groups, while normalizing the RMS error of non-dyskinetic subjects. Yet despite decreases in RMS tracking error, medication paradoxically significantly increased the decay rate of dyskinetic compared to non-dyskinetic subjects’ tracking. In conclusion, the RMS  70  tracking performance of dyskinetic PD subjects off medication is significantly more affected by ambiguous visual feedback than that of non-dyskinetic and control subjects. L-dopa improves overall motor performance for both PD groups, but at the cost of inappropriate reliance on ambiguous visual feedback in dyskinetic PD subjects – a trade off that does not appear present in non-dyskinetic subjects. These results are indicative of inadequate weighting of predictive and feedback driven motor control in dyskinetic PD subjects confronted with visual uncertainty. 4.2  Introduction  Prediction is a fundamental component of motor control. For instance, when catching a baseball it is necessary to predict where the ball will be at a given instant and how much force its impact will generate in order to prepare the hand for the catch. Central to motor prediction is the forward model, which is a prediction of the sensory consequences of movement (Wolpert et al., 1995). Substantial evidence indicates that humans use forward models to predict the sensory consequences of their own actions (Kuo, 1995; Wolpert et al., 1995; Merfeld et al., 1999; van Beers et al., 1999; Vaziri et al., 2006; Gritsenko et al., 2009), as well as to predict the dynamics of objects in the external environment (Merfeld et al., 1999; McIntyre et al., 2001; Zupan et al., 2002; Davidson & Wolpert, 2003; Davidson & Wolpert 2005; Schubotz, 2007). Furthermore forward models of object dynamics are necessary to guide visuo-motor coordination tasks, and can even override observed kinematic feedback (Zago et al., 2004; Zago et al., 2009). Predictive forward modeling becomes even more imperative as the reliability of visual feedback is compromised, for example in dim lighting. Human subjects have been shown to both account for the degree of sensory uncertainty as well as to de-weight their  71  reliance on ambiguous sensory feedback during motor performance (Wolpert et al., 1995; Wolpert & Ghahramani, 2000; van Beers et al., 2002; Baddeley et al., 2003; Angelaki et al., 2004; Kording & Wolpert, 2004; Vaziri et al., 2006; Wei & Kording, 2010), and to do so utilizing predictive forward models (Wolpert & Ghahramani, 2000; van Beers et al., 2002; Vaziri et al., 2006). However, when subjects are unable to use predictive motor control, as has been shown in the case of patients with cerebellar degeneration, the motor response no longer anticipates sensory feedback but rather reacts to it in an uncoordinated manner (Müller & Dichgans 1994; Babin-Ratte et al., 1999; Nowak et al., 2002; Nowak & Hermsdorfer, 2005). Deficits in predictive motor control and subsequent increased reliance on feedback are classically seen in diseases of the cerebellum (Beppu et al., 1984; Stein, 1986; Beppu et al., 1987; Day et al., 1998; Lang et al., 1999; Bastian, 2006). Evidence from neuroimaging and computational studies is in accordance with this, indicating that predictive forward models likely reside in the cerebellum (Kettner et al., 1997; Blakemore et al., 1998; Inoue et al., 1998; Kitazawa et al., 1998; Tamada et al., 1999; Blakemore et al., 2003; Kawato et al., 2003; Imamizu et al., 2003; Bastian, 2006; Ito, 2008; Synofzik et al., 2008). However, deficits in predictive motor control have as well been described in Parkinson’s disease (PD) where subjects become overly reliant on visual feedback (Flowers, 1978a; Flowers, 1978b). L-dopa remains the gold standard of treatment in PD despite the relatively common occurrence of side effects such as L-dopa-induced dyskinesias (LID) (Agid et al., 1999). Peak-dose LID are excessive involuntary movements that occur as a side effect of L-dopa treatment in PD as dopamine reaches peak concentrations in the brain (Fahn, 2000). Deficits in sensorimotor control have been demonstrated in dyskinetic PD subjects, possibly relating  72  to inadequate predictive motor control (Moore, 1987). Moore found that dyskinetic subjects underestimate the distance their limb has moved, which he hypothesized was a result of an exaggerated ‘corollary discharge’, or efference copy of the motor command. Furthermore, a sensorimotor contribution to LID was demonstrated by the heightened motor responses of dyskinetic PD subjects to visual feedback (Liu et al., 2001). However, the authors concluded that deficits in predictive control were unlikely to explain their results, noting the basal ganglia pathology of PD and suggesting instead that the increased responses to feedback may have been in response to the excessive movements of the LID, or to the greater magnitude of tracking error resulting from involuntary movements. More recently there is a significant amount of evidence demonstrating altered cerebellar activity in PD (Rascol et al., 1997; Yu et al., 2007; Lewis et al., 2007; Ballanger et al., 2008; Palmer et al., 2009; Ni et al., 2010), and more specifically in dyskinetic PD (Koch et al., 2009), suggesting that cerebellar compensation and/or dysfunction may frequently occur in PD. Thus, possible cerebellar changes in dyskinetic PD, as well as the involuntary nature of LID, may impair predictive motor control in these subjects. In chapter 2 we demonstrated that PD subjects are susceptible to sensory uncertainty during visually guided tracking, but did not dichotomize dyskinetic and non-dyskinetic subjects. Therefore in chapter 4 we employ the same tracking task to assess the susceptibility to uncertain visual feedback of dyskinetic and non-dyskinetic PD subjects. As previous work has demonstrated linear dynamical system (LDS) models to be a sensitive marker of motor performance in PD (Au et al., 2010), here we used LDS models in addition to quantifying tracking error to assess tracking performance. By extracting the decay rate  73  parameter from the LDS models during ambiguous tracking, we quantified subjects’ relative reliance on uncertain visual feedback. 4.3 4.3.1  Methods Subjects  The work in this chapter utilizes the same raw tracking data set as that used in chapter 3. Therefore data from 19 patients with clinically-definite PD (Hoehn and Yahr stage 1-3) (Hoehn & Yahr, 1967), 9 of which were dyskinetic PD (DPD) subjects and 10 were nondyskinetic PD (NDPD) subjects, and 10 age-matched control subjects without active neurological disorders, were utilized in this chapter. However, here we examined both the ambiguous and non-ambiguous tracking conditions while subjects were off and on medication, and again dichotomized PD subjects with respect to presence or absence of LID. As noted in chapter 3, all patients had overnight withdrawal of medications for at least 12 hours before the study for L-dopa, and 18 hours for dopamine agonists. The converted Ldopa daily dosage was calculated as 100 mg L-dopa = 125 mg of controlled-release L-dopa. This was then added to the equivalents of dopamine agonists to give the L-dopa equivalent daily dosage (LEDD), where 100 mg of L-dopa = 1 mg of pramipexole, 6 mg of ropinirole, 10 mg of bromocriptine, 75 mg of L-dopa plus entacapone. The presence of peak-dose LID was assessed up to 1.5 hours after the L-dopa challenge, where subjects received the equivalent of their morning L-dopa dose given in the immediate release form. Peak-dose LID were defined by the presence of involuntary choreiform movements in any of the head/neck, trunk and limbs of variable duration, and may or may not have been accompanied by dystonia. LID severity was assessed according to the Goetz Dyskinesia Rating Scale (Goetz et al., 1994), and all DPD subjects had mild LID that were of minimal severity and did not  74  interfere with voluntary motor acts. Disease severity was assessed according to the Unified Parkinson’s Disease Rating Scale (UPDRS) motor score in the off medication state. The Ethics Board of the University of British Columbia approved the study and all subjects gave written, informed consent. 4.3.2  Study paradigm  The experimental paradigm used in this chapter is described in detail in chapter 2. Briefly, a Lissajous figure was presented on a screen measuring 1.62 m by 1.22 m with a red circular target (12 cm in diameter) in the center of the screen. Subjects stood approximately 55 cm in front of the screen, and tracked the moving target with their index finger, requiring movement about the wrist, elbow and shoulder joints. In the baseline trials the target smoothly followed the Lissajous path, either at a Slow tracking speed (average speed of 56.2 cm/sec) or a Fast tracking speed (average speed of 78.3 cm/sec). In subsequent visually ambiguous conditions, the target jittered about the path while maintaining the path’s overall trajectory. In the ambiguous tracking conditions, subjects were instructed to attempt not to chase the jitter, but rather to attempt to track the desired target’s position, which maintained the overall Lissajous trajectory. Four levels of visual ambiguity were tested, {0, 0.03, 0.05, 0.07} – representing the amplitude of jitter with respect to screen height, at two speeds, giving a total of 8 conditions. Each condition was tested in 3 different trials, where a trial consisted of 30 seconds of tracking, a 12 second rest, followed by 30 seconds of tracking. Subject DPD 9 was an exception and completed 2 trials of each condition due to fatigue. 4.3.3  Quantification of tracking performance  We calculated root mean square (RMS) tracking error as described in chapter 2. There was a training effect between trial set 1 and trial set 2 that subsequently stabilized (as  75  determined by changes in RMS tracking errors) (Fig. 4-1), and we therefore omitted trial set 1 in all subsequent calculations. We again computed linear dynamical system (LDS) models utilizing system identification techniques (Ljung, 1999), but in this chapter we focused on the decay rate parameter of subjects’ tracking during the ambiguous tracking conditions – which is indicative of the relative reliance on ambiguous visual feedback. 4.3.4  Statistical analyses  RMS tracking error and mean decay rate were compared for all groups using analysis of variance (ANOVA). Significance was declared at p<0.05 after correcting for multiple comparisons. A robust multivariate regression analysis was performed, using RMS error as the dependent variable, speed and ambiguity as the independent variables, and the B coefficients were obtained indicating the portion of RMS error that is explained by speed and ambiguity amplitude. We assessed the LDS models’ loss of function (e.g., Akaike information criterion, or Akaike final prediction error) in order to indicate how well a model fits the data it was constructed from, and overfitting was prevented by penalizing the use of an excessive number of model parameters. Higher model quality is indicated by lower loss function values, which represent a relatively better fit to the experimental data, as there is less ‘loss’ of information from the experimental data. 4.4  Results  For a description of subjects’ characteristics please refer to Table 3-1. Analysis of the RMS error between trials revealed that there was a training effect demonstrated between trials 1 and 2, indicated by a significant decrease (p<0.021) in RMS error for all groups,  76  except for the DPD group on medication (p>0.05). RMS error stabilized between trials 2 and 3 for all groups (p>0.39) (Fig. 4-1). Trial 1 data was therefore omitted from all subsequent data analysis to ensure we were not examining motor learning in our visually guided tracking task but rather the effect of visual uncertainty after learning had occurred and stabilized. ANOVA of RMS error revealed several findings. During non-ambiguous tracking conditions there were no significant differences in RMS error between all groups (F = 1.75, p = 0.16 for Slow tracking, F = 2.23, p = 0.082 for Fast tracking), indicating that even while both DPD and NDPD subjects were off medication they could perform the non-ambiguous tracking task as well as control subjects (Fig. 4-2A). However, between group comparisons of RMS error during ambiguous tracking revealed that at both Slow and Fast tracking speeds DPD OFF subjects had significantly greater RMS error than Control subjects at all levels of ambiguity (p<0.0021), and similarly NDPD OFF subjects had significantly greater RMS tracking error than Control subjects in all ambiguous conditions (p<0.047), except for in condition {Slow, 0.07 ambiguity} (p = 0.07). DPD OFF subjects demonstrated greater RMS error than NDPD OFF subjects in all ambiguous conditions, though this was not statistically different (p>0.15). The RMS error of DPD subjects on medication remained significantly higher than that of control subjects across all ambiguous tracking conditions (p<0.043) except for conditions {Slow, 0.05 ambiguity, p = 0.11} and {Fast tracking, 0.07 ambiguity, p = 0.0525}. NDPD ON subjects demonstrated a normalization of RMS error with medication, as their RMS error did not significantly differ with that of Control subjects in any of the ambiguous tracking conditions (p>0.25). As a result, the RMS error of DPD ON subjects significantly differed from that of NDPD ON subjects in conditions {Slow, 0.07  77  ambiguity} (p = 0.013) and {Fast, 0.07 ambiguity} (p = 0.048), but not in conditions {Slow, 0.03, 0.05 ambiguity} (p>0.055) and conditions {Fast, 0.03, 0.05 ambiguity} (p>0.053). ANOVA revealed significant within group differences in RMS error with the addition of target ambiguity for all groups. Specifically, the RMS error was significantly increased from the baseline non-ambiguous conditions (Slow and Fast speeds) to each of the ambiguous conditions at the respective tracking speed for all groups (p<3.9x10-6). Additionally, there were significant increases in RMS error between the conditions {0.03 ambiguity} and {0.07 ambiguity} at slow and fast speeds for DPD OFF subjects (p<0.033), for DPD ON subjects (p<0.03) and for control subjects (p<0.014). This significant difference was also observed in NDPD OFF and NDPD ON subjects at Fast tracking speed, respectively (p = 0.04 and p = 0.0055), but not in NDPD OFF and NDPD ON subjects at Slow tracking speed, respectively (p = 0.099 and p = 0.06). The differences in RMS error between non-ambiguous and maximum ambiguous tracking conditions were not significantly correlated with UPDRS scores at either tracking speed for either dyskinetic subjects (p>0.19) or for non-dyskinetic subjects and (p>0.1). This indicates that for both PD groups the increase in RMS error observed due to increasing ambiguous sensory input cannot be attributed to disease severity. The overall effect of increasing ambiguity and speed on tracking performance, and the L-dopa effect, is illustrated in Figure 4-2B. The regression analysis illustrates the relative contribution of increasing ambiguity and speed to RMS error by group. Increasing tracking speed and ambiguity contributed to the RMS error of DPD OFF subjects significantly more than for all other groups. Additionally, the susceptibility to speed and visual ambiguity is not normalized with medication for DPD ON subjects, but is roughly normalized for NDPD ON  78  subjects. The speed and ambiguity regression coefficients captured by the model are highly significant for all groups (p<10-4). Furthermore, the between group differences in both speed and ambiguity regression coefficients are as well highly significant (p<10-4). The Akaike's Final Prediction Error (FPE) and Akaike information criterion (AIC) used to assess the LDS models from ambiguous tracking conditions revealed robust tracking models. The means and standard deviations of the estimated linear dynamical system models’ FPE and AIC were < 3.1 ± 2.0 and < 1.8 ± 0.4 respectively, for all groups across all conditions, which is indicative of high model quality/fit. Furthermore, there were few outliers in the FPE and AIC values indicating high model validity. Figure 4-3 illustrates that DPD subjects demonstrated significantly greater mean decay rates while on medication. ANOVA revealed significant differences in mean decay rates between groups in conditions {Slow, 0.07 ambiguity} (p = 0.0027) and all fast tracking conditions {Fast, 0.03, 0.05, 0.07 ambiguity} (p<0.01). Medication tended to decrease or have no effect on the decay rate of NDPD subjects, but paradoxically increased the decay rate of DPD subjects, and the differing effect of medication on decay rate pooled across ambiguous tracking conditions was statistically significant between dyskinetic and nondyskinetic subjects (p = 8x10-4 for Fast tracking, p = 0.02 for Slow tracking). Furthermore, decay rate was not significantly correlated with UPDRS (p>0.14) for either of the PD groups in any of the ambiguous tracking conditions. In order to get an intuitive interpretation of the significantly different decay rates, we interrogated typical models from each group (i.e. models with eigenvalues closes to the mean for each group) with 1-dimensional sinusoidal inputs and additive noise similar to the experiment to determine the predicted tracking performance. Ideal tracking performance  79  would occur in systems that ignore the noisy input and faithfully maintain sinusoidal tracking. Consistent with the RMS error results, the sinusoidal tracking improved after medication in both dyskinetic and non-dyskinetic subjects. However, after L-dopa medication, dyskinetic subjects had a paradoxical worsening of their ability to ignore the noisy visual cue, and were excessively reliant on noisy ambiguous visual feedback (Fig. 44). 4.5  Discussion  We examined the ability of 9 dyskinetic and 10 non-dyskinetic PD subjects, as well as that of 10 age-matched control subjects, to de-weight uncertain sensory feedback and instead more heavily rely on predictive motor control during visually guided tracking. The relative contributions of increasing target ambiguity and speed to the RMS tracking error were the greatest for dyskinetic PD subjects off medication (Fig. 4-2B). As expected, Ldopa medication improved overall tracking performance for both PD groups, as evidenced by reduced RMS error with medication (Fig. 4-2) and the ability of fitted models to faithfully track a sinusoidal input (Fig. 4-4, right upper panel). However in dyskinetic subjects, improved overall tracking performance came at a price: they were also more responsive to and reliant on non-informative visual feedback (Fig. 4-4, right lower panel). We interpret our results in the context of established performance trade-offs in control theory, in which controllers that produce exceptionally fast, high-performance tracking under ideal circumstances are also extremely poor at disturbance rejection (that is, they experience high sensitivity to external or unmodeled noise processes), and in the biological system this leads subjects to more heavily rely on ambiguous sensory feedback and less on predictive motor control. In contrast to dyskinetic PD subjects, the decreased RMS error and decay rate of 80  non-dyskinetic subjects (off and on medication) and of control subjects during ambiguous tracking demonstrates that the ability to de-weight ambiguous sensory feedback was relatively preserved in these groups. Thus the improvement in motor performance with Ldopa does not come with the trade-off of greater reliance on ambiguous visual feedback in non-dyskinetic PD subjects. This is in agreement with other studies demonstrating that nondyskinetic PD subjects do not overly respond to visual feedback (Liu et al., 1999), and that healthy human subjects internally account for sensory uncertainty and de-weight uncertain feedback during motor performance (Wolpert et al., 1995; Baddeley et al., 2003; Kording & Wolpert, 2004; Wei & Kording, 2010), and do so using forward models (Wolpert & Ghahramani, 2000; van Beers et al., 2002; Vaziri et al., 2006). One possible explanation for dyskinetic PD subjects’ inability to de-weight ambiguous sensory feedback, is an increased response to the sensory discrepancy between actual and predicted sensory feedback. Forward models are used to predict sensory feedback, and the predicted feedback is subsequently compared to actual feedback when it becomes available after an inherent delay (van Sonderen et al., 1988; Miall & Wolpert, 1996). The difference between the actual and predicted sensory feedback is known as the sensory discrepancy or error, which is then used to update the forward model and in turn improve motor performance (Wolpert et al., 1995; Miall & Wolpert, 1996). During the nonambiguous conditions of our tracking task the sensory discrepancy is likely minimal as the predicted sensory feedback relating the subjects’ index finger position and the target position would be congruent, which is supported by the lack of differences in RMS tracking error between the groups in the baseline conditions (Fig. 4-2A). However, in the ambiguous tracking conditions the sensory discrepancy would be large due to the ambiguous jitter of  81  the target. In this manner control and non-dyskinetic PD subjects may neglect the sensory discrepancy and more heavily weight their forward model in conditions when sensory feedback is uncertain. By contrast, dyskinetic PD subjects may attempt to update their forward models based on ambiguous feedback, subsequently leading them to be more responsive to and reliant on ambiguous sensory feedback. Interestingly, evidence from neuroimaging studies demonstrates significant cerebellar activity in conditions of mismatch between predicted and actual feedback (Blakemore et al., 1998), and the degree of mismatch imposed by temporal delays has been correlated with cerebellar activity (Blakemore et al., 2001). Further evidence indicates that the cerebellar climbing fiber-Purkinje cell synapse may signal the error between the predicted and actual sensory feedback (Oscarsson 1980; Anderson & Armstrong, 1985; Gellman et al., 1985; Simpson et al., 1996; Kettner et al., 1997; Kitazawa et al., 1998). Accordingly, cerebellar hyperactivity in PD (Yu et al., 2007) may excessively sensitize PD subjects to sensory error, where the degree of sensitization may differ between dyskinetic and non-dyskinetic subjects. Alternatively, control and non-dyskinetic PD subjects may have been better able to predict the desired target position in the ambiguous tracking conditions. The concept of forward modeling has been extended to predicting the sensory consequences of input from the external environment (Merfeld et al., 1999; McIntyre et al., 2001; Zupan et al., 2002; Davidson & Wolpert, 2003; Zago et al., 2004; Davidson & Wolpert 2005; Schubotz, 2007; Zago et al., 2009). For example, evidence indicates that human subjects utilize forward models of visual cues (Zupan et al., 2002), of target motion during interception tasks (Zago et al., 2004; Zago et al., 2009), and of the physical laws of gravitational acceleration (Merfeld et al., 1999; McIntyre et al., 2001; Angelaki et al., 2004). Computationally, we  82  quantified RMS tracking error as the difference between the subjects’ index finger position and the target position along the smooth Lissajous path at any given time point, a position in the ambiguous conditions that corresponded to the mean of the jitter amplitude along the path. Human subjects have been shown to reliably predict the mean perturbation delivered from a variable distribution in reaching tasks (Scheidt et al., 2001; Takahashi et al., 2001), and to do so according to Bayesian inference (Kording & Wolpert, 2004). Though we did not explicitly test the use of Bayesian statistics in this study, the strategy of more heavily weighting the mean jitter amplitude and de-weighting the instantaneous uncertain jittering position of the target in order to predict the desired tracking position, corresponds to the optimal motor response in our task that minimizes RMS tracking error. Thus dyskinetic subjects may be less able predict the target location based on the mean jitter amplitude, rendering them susceptible to the ongoing ambiguous feedback. The cerebellum is known for its integral role in predictive motor control, and predictive deficits that lead subjects to excessively respond to feedback by making many corrective movements are typically seen in cerebellar disease (Beppu et al., 1984; Beppu et al., 1987; Day et al., 1998; Bastian, 2006), a motor response that is likely due to inadequate forward modeling (Babin-Ratte et al., 1999; Nowak et al., 2002; Bastian, 2006). Extensive evidence supports the use of forward models in human subjects (Kuo, 1995; Wolpert et al., 1995; Miall et al., 1996; Merfeld et al., 1999; van Beers et al., 1999; Vaziri et al., 2006; Gritsenko et al., 2009), and neuroimaging, electrophysiology and transcranial magnetic stimulation (TMS) studies provide strong evidence for the role of the cerebellum in forward modeling (Kettner et al., 1997; Blakemore et al., 1998; Inoue et al., 1998; Kitazawa et al., 1998; Tamada et al., 1999; Blakemore et al., 2003; Kawato et al., 2003; Imamizu et al.,  83  2003; Bastian, 2006; Miall et al., 2007; Ito, 2008; Synofzik et al., 2008). Thus, a possible neuroanatomical correlate of the inability of dyskinetic subjects to de-weight ambiguous visual feedback demonstrated in the present study is inadequate cerebellar forward modeling. The growing evidence of functional cerebellar changes that occur in PD supports this possibility (Rascol et al., 1997; Yu et al., 2007; Lewis et al., 2007; Ballanger et al., 2008; Palmer et al., 2009; Koch et al., 2009; Ni et al., 2010). However, in addition to the cerebellum the posterior parietal cortex (PPC) is believed to have an important role in predictive motor control (Blakemore & Sirigu, 2003). The role of the PPC in making online corrections (a process that requires forward models) during movement has been demonstrated in patients with lesions to this area and through the use of TMS (Desmurget et al., 1999; Pisella et al., 2000). TMS applied to the PPC of healthy human subjects prevented them from making fast on-line corrective movements to a target perturbation in a reaching task when vision of their arm was occluded, and they instead continued to reach to the initial target (Desmurget et al., 1999). As dyskinetic subjects in our study were found to be overly responsive to the ambiguous visual feedback (as opposed to unresponsive PPC subjects), this may argue against altered PPC function explaining our results. Moreover, PPC stimulation has been related to motor awareness (Desmurget et al., 2009), and interestingly dyskinetic PD patients can be unaware of their involuntary movements (Vitale et al., 2001). Nonetheless it is possible that altered PPC activity contributed to the impaired predictive motor control of dyskinetic subjects, and as the PPC and cerebellum have reciprocal neuroanatomical connections (Glickstein, 2000; Clower et al., 2001), it is likely that these two structures work together in using forward models to guide motor performance.  84  The overall RMS tracking error of dyskinetic PD subjects was most adversely affected by visual uncertainty while subjects were off dopaminergic medication. This is perhaps counterintuitive as the involuntary movements of LID that distinguish dyskinetic from non-dyskinetic subjects occur while they are on medication. However, differences in the off medication state may be explained by persistent neural plasticity in dyskinetic PD (Troiano et al., 2009). This is supported by animal models of PD suggesting that unnatural pulsatile stimulation of dopaminergic receptors, as occurs with intermittent dosing of Ldopa, may induce plastic changes that contribute to the development of LID (Calon et al. 2000a; Aubert et al. 2005). Interestingly, younger patients are more prone to developing LID (Kumar et al., 2005), and this may be related to a greater degree of plasticity occurring in the younger brain. Additionally, plasticity related to LID is not limited to the basal ganglia, as Nimura & colleagues (2004) demonstrated cerebellar plasticity particular to LID as the binding potential of the cerebellar sigma receptors was positively correlated with LID scores but not with disease severity of PD patients undergoing pallidotomies. Cerebellar TMS has been shown to lessen LID severity further suggesting altered cerebellar plasticity in the dyskinetic PD. We have furthermore found behavioural differences that differentiate dyskinetic from non-dyskinetic subjects in the off medication state that may be related to cerebellar plasticity (chapter 3). Direct neuroanatomcial pathways connecting the basal ganglia and the cerebellum have been found in primates (Hoshi et al., 2005; Bostan et al., 2010), providing a route for the administration of L-dopa to interact with altered persistent plasticity in these structures leading to excessive involuntary movements. Dyskinetic subjects on medication demonstrated fundamentally differing responses to non-informative ambiguous visual feedback compared to non-dyskinetic subjects on  85  medication. This not only suggests plastic changes of the motor system that differentiates these PD sub-groups, but also may directly contribute to the involuntary movements of LID. We propose three possible explanations for how these motor responses may contribute to involuntary movements. Firstly, contrast sensitivity is reduced in PD (Bodis-Wollner et al., 1987; Price et al., 1992), and possible differences between dyskinetic and non-dyskinetic subjects may effectively reduce the reliability with which dyskinetic subjects interpret the visual world. Deficits in contrast sensitivity in PD are more marked at higher spatial frequencies (Mestre et al., 1990; Jones et al., 1992), and as sensory ‘noise’ is characterized by higher frequencies, reduced contrast sensitivity may selectively amplify ambiguous sensory input perceived by dyskinetic subjects. Faced with a more uncertain world, a greater tendency to respond to the ongoing uncertainty as found in this study, may induce abnormal movements in dyskinetic subjects. Secondly, inadequate forward models in dyskinetic PD may produce a sensory discrepancy between the predicted and actual sensory outcomes of movement on a continual basis (and not only when artificially provided as in our tracking task), effectively serving as an endogenous ‘noisy’ error signal eliciting continual motor responses. Thirdly, L-dopa induces depressed proprioceptive sense in PD (O’Suilleabhain et al., 2001), and mismatches between visual and proprioceptive sensory input have as well been found in PD (Demirci et al., 1997). This may lead to a compensatory shift to greater reliance on visual input, which may lead dyskinetic subjects to attempt to ‘chase’ visual stimuli resulting in abnormal movements. In addressing potential limitations of this study, there was a trend towards greater disease severity of dyskinetic than non-dyskinetic PD subjects, though the difference was non-significant (Table 3-1). Nonetheless, in order to fully address this we examined the  86  relationship between UPDRS and the increase in RMS between the non-ambiguous and maximum ambiguous tracking conditions, and no significant correlation for either dyskinetic or non-dyskinetic subjects was found. Thus, the worsening of motor performance with increasing visual ambiguity was not associated with disease severity. Furthermore, UPDRS was not significantly correlated with decay rate for either PD groups. Secondly, we tested PD subjects in the practically defined off medication state with 12 hours of L-dopa withdrawal and 18 hours for dopamine agonists, and subjects were symptomatic upon study commencement. We note that this method of examining the practically defined off medication state in PD is universally utilized (Langston et al., 1992; Defer et al., 1999), though we acknowledge that this may not reflect a truly depleted dopaminergic state. In conclusion, we demonstrate that dyskinetic PD subjects are significantly more susceptible to visually ambiguous sensory input during a visually guided tracking task, and that the improvement in overall tracking performance with L-dopa medication comes at a price for dyskinetic PD subjects: an increased reliance on ambiguous visual feedback. The results indicate inadequate weighting of predictive motor control in dyskinetic PD, which may be a significant contributor to pathophysiology of LID. We discuss possible cerebellar dysfunction in dyskinetic PD as a neuroanatomical substrate of inadequate weighting of predictive motor control.  87  Figure 4-1. The training effect for dykinetic and non-dyskinetic subjects. All groups demonstrated a significant decrease in RMS error between trial 1 and trial 2 except for the DPD ON group (where the decrease was not significant), and the RMS error subsequently stabilized for all groups between trials 2 and 3. We therefore omitted trial 1 data in all subsequent calculations.  88  !"  SLOW TRACKING  18  Dyskinetic OFF Dyskinetic ON Non-dyskinetic OFF Non-dyskinetic ON Control  16  RMS TRACKING ERROR  FAST TRACKING  14 12 10 8 6 4 2  0  0.03  0.05  0  0.07  0.03  0.05  0.07  RELATIVE AMPLITUDE OF AMBIGUITY  #"  2.8  x 10  Dyskinetic OFF Dyskinetic ON Non-dyskinetic OFF Non-dyskinetic ON Control  2.6 AMBIGUITY REGRESSION COEFFICIENTS  -3  2.4  $%&'()" *+*,-"  2.2 2 1.8  $%&'()" *+*,-"  1.6 1.4 0.1  0.11  0.12  0.13  0.14  0.15  0.16  SPEED REGRESSION COEFFICIENTS  Figure 4-2. Effect of speed and ambiguity on dyskinetic subjects’ RMS error. A) RMS error by tracking condition. Differences in RMS error in the non-ambiguous tracking conditions between all groups did not significantly differ, and the effect of increasing ambiguity on RMS error at slow and fast tracking speeds by group is illustrated. Dyskinetic subjects off medication exhibit the greatest RMS error in the ambiguous tracking conditions. B) Regression analysis. The relative contribution of increasing target ambiguity and speed on RMS error is demonstrated. Dyskinetic subjects off medication demonstrate a significantly greater sensitivity to increasing ambiguity and speed than Control and nondyskinetic subjects off medication, which is not normalized with L-dopa medication for dyskinetic subjects.  89  SLOW TRACKING  1  FAST TRACKING  0.9 0.8  !"  DECAY RATE  0.7 0.6  !" !" !" !"  0.5 0.4  !"  0.3 0.2  !"!"  !" !"  !"  0.1 0  0.03!  0.05!  0.07!  0.03!  0.05!  0.07!  RELATIVE AMPLITUDE OF AMBIGUITY! Dyskinetic OFF Non-dyskinetic OFF Dyskinetic ON Non-dyskinetic ON Control  Figure 4-3. Mean decay rate by group. The asterisk indicates a significant difference (p < 0.05) between the Dyskinetic ON group and the indicated group.  90  Figure 4-4. LDS model simulation results: response to ambiguous visual feedback. Left panel: Tracking input is a combination of a smooth sinusoidal reference trajectory and bandpass filtered white noise. Right panel: Subjects modeled tracking output in a smooth non-ambiguous tracking condition (upper box) and in a ‘noisy’ ambiguous tracking condition (lower box). In the upper box the modeled tracking output of a representative subject from each group is shown. The subjects’ output in non-ambiguous conditions is similar across groups. In the lower box the effect of ‘noisy’ ambiguous input on modeled output of a dyskinetic and a non-dyskinetic subject is shown. The increased decay rate of dyskinetic subjects on medication leads them to attempt to track the noise significantly more. The noisy visual signal is reduced to 15% of the actual magnitude for visualization purposes.  91  CHAPTER 5 CONCLUSION 5.1  Introduction  The purpose of this thesis was to 1) investigate the ability of PD subjects, including that of dyskinetic and non-dyskinetic sub-groups, to de-weight ambiguous visual feedback during motor performance, and 2) to use a manual tracking task to investigate and characterize possible subtle differences between the motor performance of dyskinetic and non-dyskinetic PD sub-groups. To address this we designed a large-amplitude visually guided tracking task where the target ‘jittered’ about the desired trajectory rendering its instantaneous position uncertain, quantified motor performance using RMS tracking error, and computed LDS models to extract the damping ratio, natural frequency and decay rate parameters of the models. We found that 1) PD subjects off medication were significantly more susceptible to increasing visual uncertainty than control subjects, 2) dyskinetic subjects demonstrated significantly reduced damping ratios during non-ambiguous tracking compared to nondyskinetic subjects while both groups were off medication, and 3) dyskinetic subjects were significantly more susceptible to visual uncertainty off medication, and the L-dopa induced improvement in motor performance of both PD groups came at a price for dyskinetic subjects – a greater response to and reliance on ambiguous visual feedback, as demonstrated by increased decay rate with medication. This chapter will summarize the findings of this thesis and discuss the interpretations of our results, highlight the significance of the results while acknowledging the limitations of the experiments, and finally provide insight into future directions of study.  92  5.2  Summary of findings and interpretations  We used a visually guided tracking task to probe the motor response of PD subjects to ambiguous visual feedback, while comparing these responses as well as further subtleties in motor performance across dyskinetic and non-dyskinetic PD sub-groups. Collectively, the results of this thesis highlight the possible role of altered cerebellar function as a significant contributor to the pathophysiology of PD and LID. More specifically, the results from chapters 2 and 4 indicate that PD is marked by a susceptibility to and reliance on sensory uncertainty that is suggestive of impaired cerebellar forward modeling, and that this may be particular to dyskinetic PD. Chapter 3 supports the hypothesis that cerebellar plasticity may differentiate dyskinetic from non-dyskinetic PD subjects, as we found abnormalities in dyskinetic subjects’ tracking performance in the off medication state that are similar to abnormalities that are known to present in cerebellar patients. In chapter 2 we demonstrated that though PD subjects off medication were able to complete a large-amplitude tracking task to the same degree of accuracy as healthy age matched control subjects in the baseline non-ambiguous tracking condition, the RMS tracking error of PD subjects off medication was significantly more adversely affected by increasing target uncertainty and speed than that of control subjects (Fig. 2-3). Furthermore, L-dopa medication only partially normalized this result. We interpreted this susceptibility to increasing visual ambiguity of parkinsonian subjects as an inability to de-weight ambiguous visual feedback and instead more heavily rely on predictive motor control, a strategy that healthy human subjects have been shown to employ (Wolpert et al., 1995; Wolpert & Ghahramani, 2000; van Beers et al., 2002; Baddeley et al., 2003; Angelaki et al., 2004; Kording & Wolpert, 2004; Vaziri et al., 2006; Wei & Kording, 2010). We interpreted our  93  findings as suggestive of cerebellar dysfunction in PD, owing to the evidence highlighting the integral role of the cerebellum in forward modeling (Diener et al., 1993; Gao et al., 1996; Jueptner et al., 1997; Kettner et al., 1997; Blakemore et al., 1998; Inoue et al., 1998; Tamada et al., 1999; Kitazawa et al., 1998; Kawato et al., 2003: Imamizu et al., 2003; Bastian, 2006; Ito, 2008; Synofzik et al., 2008). In chapter 3 we used LDS models to examine the motor performance of dyskinetic and non-dyskinetic subjects off medication, as well as that of healthy control subjects, during the non-ambiguous slow tracking condition of our task. We demonstrated that though the RMS tracking error was not significantly different between these three groups, the damping ratio parameter extracted from the LDS models was significantly lower for dyskinetic subjects than for non-dyskinetic and control subjects (Fig. 3-1). This finding is of interest for two principal reasons. Firstly, this is a novel result demonstrating that motor performance differs between dyskinetic and non-dyskinetic groups even in the off medication state when LID are not present, which is a somewhat counterintuitive finding as PD symptoms of both groups appear similar if not identical in the off medication state. This finding is suggestive of persistent neural plasticity that may be fundamentally different between these PD subgroups, and in the dyskinetic case this may predispose these subjects to the development of LID. Secondly, underdamped motor performance leads to overshoot of a desired trajectory during motor performance (Fig. 3-2), a frequently observed phenomenon in patients with lesions to the cerebellum (Hore et al., 1991; Manto et al., 1995; Brown et al., 1999). We speculate that our findings are reflective of cerebellar plasticity in PD that may contribute to the pathophysiology of LID. This interpretation is supported by recent evidence demonstrating altered cerebellar plasticity in dyskinetic PD subjects (Nimura et al., 2004) as  94  well as results from a study demonstrating an improvement in LID severity with the use of cerebellar TMS (Koch et al., 2009). In chapter 4 we used RMS tracking error and LDS models to examine the effect of visual ambiguity and speed on the motor performance of dyskinetic, non-dyskinetic and healthy control subjects. Analysis of RMS error revealed that the tracking performance of dyskinetic PD subjects off medication was significantly more susceptible to increasing visual ambiguity than that of non-dyskinetic and control subjects (Fig. 4-2). As expected, L-dopa medication improved the RMS tracking error of both PD groups, while normalizing the RMS tracking performance of non-dyskinetic PD subjects. However, this improvement in overall motor performance was accompanied by an increase in the decay rate parameter of dyskinetic subjects’ tracking in ambiguous conditions, but not that of non-dyskinetic subjects (Fig. 4-3). The decay rate is indicative of the responsiveness to and reliance on the ambiguous jitter, and reflects the relative degree to which subjects were attempting to chase the jitter. Thus, the improvement in overall motor performance from L-dopa comes at a price for dyskinetic subjects, which is a greater tendency to respond to uncertain visual feedback (Fig. 4-4). Taken together, we interpret these findings as indicative of impaired predictive forward modeling in dyskinetic PD leading to an increased reliance on ambiguous visual feedback, and we suggest that this is reflective of altered cerebellar activity contributing to the pathophysiology of LID in PD. The need for predictive forward models in motor control has been demonstrated through studies indicating that human subjects anticipate changes in sensory information that would be otherwise precluded by delays in sensory feedback processing. This has notably been shown in grip force studies (Johansson & Cole 1992; Flanagan & Wing 1997; Kawato,  95  1999; Kawato et al., 2003), reaching tasks (Jordan & Rumelhart, 1992; Wolpert et al., 1995) and tracking tasks (Vercher & Gauthier, 1988; Vercher & Gauthier, 1992; Miall & Wolpert 1996). Furthermore, healthy human subjects have been shown use predictive motor control to mitigate the effect of sensory uncertainty on motor performance (Wolpert et al., 1995; Wolpert & Ghahramani, 2000; van Beers et al., 2002; Baddeley et al., 2003; Angelaki et al., 2004; Kording & Wolpert, 2004; Vaziri et al., 2006; Wei & Kording, 2010). In this manner, forward model predictions are compared with actual sensory feedback as it becomes available, and the discrepancy between the two is known as the sensory error, which is then used to update the forward model in order to improve motor performance (Wolpert & Ghahramani, 2000). In the case of less reliable sensory feedback however, the relative weighting of the sensory feedback can be decreased and the forward model more heavily relied upon in order to maintain near optimal motor performance (Wolpert & Ghahramani, 2000; van Beers et al., 2002; Baddeley et al., 2003; Angelaki et al., 2004; Vaziri et al., 2006). Therefore, one possible explanation for our results is that dyskinetic PD subjects are highly tuned to the sensory error, and detrimentally attempt to update their forward models and motor performance based on ambiguous feedback. This interpretation is supported by the decreased smoothness of movement observed in dyskinetic PD subjects when visual feedback is present (Liu et al., 2001). Hypersensitivity to the sensory discrepancy would lead to a susceptibility to uncertain visual feedback, as well as increased responsiveness to it, corresponding to our results in dyskinetic PD. An alternative but related possible explanation for our results is that forward models used by dyskinetic PD subjects were inadequate. If control and non-dyskinetic PD subjects were better able to predict the target’s actual position based on the mean amplitude of the  96  jitter, and to use this as the reference for the desired trajectory, then it is possible that even in the ambiguous tracking conditions these groups would have relatively minimal discrepancy between the actual and predicted feedback. The ability to use forward models to predict the target’s position is supported by studies demonstrating that humans use forward models of external objects (Merfeld et al., 1999; McIntyre et al., 2001; Zupan et al., 2002; Davidson & Wolpert, 2003; Zago et al., 2004; Davidson & Wolpert 2005; Schubotz, 2007; Zago et al., 2009). Furthermore, the possible ability of control and non-dyskinetic subjects to predict the mean jitter amplitude in our task is supported by studies demonstrating that healthy human subjects can predict the mean perturbation amplitude from a variable distribution during reaching tasks (Scheidt et al., 2001; Takahashi et al., 2001; Kording & Wolpert, 2004). This explanation contrasts the previous one given in that here the predictive forward models used by dyskinetic subjects would actually need to be impaired and/or inadequate, whereas for the explanation in the previous paragraph, it may not be that the forward models are inadequate, but rather the ability to de-weight the sensory error is what differs between dyskinetic and non-dyskinetic PD subjects. In either case, the results of this thesis argue that dyskinetic PD subjects demonstrate an imbalance between the relative weighting of predictive forward model and visual feedback during motor performance. The inability of dyskinetic PD subjects to de-weight ambiguous visual feedback is suggestive of altered cerebellar function. This is supported by numerous studies demonstrating altered activity in the cerebellum and its connections in PD (Rascol et al., 1997; Nimura et al., 2004; Yu et al., 2007; Lewis et al., 2007; Ballanger et al., 2008; Koch et al., 2009 Palmer et al., 2009; Ni et al., 2010), as well as by evidence indicating the cerebellum as a neuroanatomical correlate of the forward model (Diener et al., 1993; Gao et  97  al., 1996; Jueptner et al., 1997; Kettner et al., 1997; Blakemore et al., 1998; Inoue et al., 1998; Tamada et al., 1999; Kitazawa et al., 1998; Kawato et al., 2003: Imamizu et al., 2003; Bastian, 2006; Ito, 2008; Synofzik et al., 2008). An intriguing hypothesis is that dyskinetic PD subjects demonstrate significantly altered cerebellar function that sensitizes them to the sensory discrepancy between the actual and predicted feedback. Cerebellar involvement in signaling of the sensory discrepancy is supported by computational studies (Tseng et al., 2007), imaging studies (Blakemore et al., 1998; Blakemore et al., 2001), and by electrophysiological studies (Oscarsson 1980; Anderson & Armstrong, 1985; Gellman et al., 1985; Simpson et al., 1995; Kettner et al., 1997; Kitazawa et al., 1998). The cerebellar hyperactivity in PD found by Yu & colleagues (2007) might then correspond to increased sensitivity to sensory discrepancy that may be specific to dyskinetic PD. An alternative neuroanatomical correlate of the forward model possibly underlying our results is the posterior parietal cortex (PPC). Similarly to the cerebellum, evidence indicates that the PPC is integral in action prediction (Duhamel et al., 1992; Sirigu et al., 1996; Desmurget & Grafton, 2000; Blakemore & Sirigu 2003), though it may be that the PPC is more involved in higher level strategic planning related to conscious intention and motor awareness (Blakemore & Sirigu, 2003; Desmurget et al., 2009). PPC function seems to be essential for making fast online corrections during movement, where subjects with damage to the PPC or those who receive TMS to it do not respond to online target perturbations during reaching tasks (Desmurget et al., 1999; Pisella et al., 2000). We suspect this may be the opposite of what occurs in dyskinetic PD, where responses to visual perturbations are dramatically increased rather than reduced. Furthermore, PPC activity has been related to the awareness of motor activity, such that increased stimulation of the PPC leads subjects to  98  believe they have moved even when electromyography recordings indicate no muscle activity (Desmurget et al., 2009). This may further contrast dyskinetic PD subjects, where evidence indicates they are often unaware of their involuntary movements, at least initially when LID are in their milder form (Vitale et al., 2001). An intriguing finding from this work is that some of the motor abnormalities observed in dyskinetic PD subjects (susceptibility of RMS error to visual uncertainty, decreased damping ratio) were found while subjects were off medication. As peak dose LID occur while patients are on medication, the question arises of how these abnormalities may contribute to the involuntary dyskinetic movements themselves. First, it may be that altered cerebellar plasticity contributes to LID through an interaction with the basal ganglia upon Ldopa administration. Neuroanatomically this may occur via direct connections between the basal ganglia and cerebellum that have been found in primates (Hoshi et al., 2005; Bostan et al., 2010). Alternatively L-dopa treatment in PD may act directly on the cerebellum via the dopamine D3 receptor (Sokoloff et al., 1990; Bouthenet et al., 1991). Furthermore, though overall tracking performance was most susceptible to visual ambiguity off medication, it was not normalized with medication for dyskinetic subjects, suggesting cerebellar abnormalities that may underlie this result persist in the on medication state and thus may actively contribute to LID. In addition, L-dopa increased the decay rate parameter from the LDS models for dyskinetic subjects, demonstrating an increased reliance on ambiguous visual feedback in the on medication state that may directly contribute to the manifestation of involuntary movements. As our results suggest a sensorimotor contribution to the pathophysiology of LID, a hypothesis stemming from this work is that LID result from inadequate predictive motor control, combined with an L-dopa induced amplified reliance on  99  ambiguous visual feedback (in addition to its well known movement enhancing effect), that together induce involuntary movements in response to ongoing visual feedback. Overall the findings of this thesis suggest that altered sensorimotor control in PD may contribute to the development of LID. On the basis of the role of the cerebellum in forward modeling and predictive motor control, as well as the characteristic ‘overshoot’ seen in patients with lesions to the cerebellum, we interpret our findings as indicative of altered cerebellar function in PD that may contribute to the pathophysiology of LID. 5.3  Study significance  Interestingly, an overwhelming majority of studies examining LID have investigated functional changes in the basal ganglia and the striatal-thalamo-cortical pathway (Chase, 1998; Calon et al., 2000a; Olanow & Obeso, 2000; Nyholm & Aquilonius, 2004; Linazasoro, 2005; Stocchi et al., 2005; Cenci, 2007; Troiano et al., 2009; Calabresi et al., 2010). However, the complexity of PD pathophysiology suggests the basal ganglia may not be the sole contributor to LID. Rather than focus on the basal ganglia, this thesis expands on the idea of altered cerebellar activity in PD and indirectly investigates this as possible contributor to LID. Expanding on the evidence that microscopic cerebellar plasticity might contribute to LID (Nimura et al., 2004), we show behavioural differences between dyskinetic and non-dyskinetic subjects exist that may be the result of such altered cerebellar plasticity. As this thesis attempts to explore beyond basal ganglia pathophysiology in LID, it will hopefully inspire further work investigating the cerebellum and other potential neural contributors to the development of LID. The prevailing belief regarding the pathophysiology of LID is that the pulsatile stimulation of dopamine receptors subsequent to the altered basal ganglia state and treatment  100  in PD results in involuntary dyskinetic movements (Nyholm & Aquilonius, 2004; Stocchi et al., 2005; Olanow et al., 2006; Calabresi et al., 2010). Yet despite support for this hypothesis the mechanisms of LID remain poorly understood (Calabresi et al., 2010), suggesting that LID are more complex than random dopaminergic neural transmission in the striatum, a hypothesis that is supported by the deterministic as opposed to random movement patterns that characterize LID (Gour et al., 2007). Furthermore, attempts to pharmacologically stabilize the release of dopamine in the basal ganglia have failed at the clinical level to reduce LID and have even increased the risk of developing them (Stocchi et al., 2010). Our work here highlights not only a potential role for the cerebellum in LID, but moreover a potential role for sensory input to impact motor output in a manner that may contribute to involuntary movements. This work suggests that LID may not result solely from pharmacological instability, but additionally from altered sensorimotor function in PD. A novel finding of this thesis is that though the parkinsonian symptoms of dyskinetic and non-dyskinetic subjects appear similar if not identical when subjects are off medication, there are actually subtle differences in the characteristics of motor performance between these two disease states. This finding suggests that LID may be the result of a restructuring of the motor system, and not merely a lower threshold for an exaggerated response to L-dopa medication. Though LID manifest as a side effect of the medication, this result might argue that a fundamental difference exists in the disease itself between these sub-groups that allows the interaction with L-dopa to produce these involuntary movements. This result somewhat counter-intuitively highlights the suitability of future research investigating the off medication state of these PD sub-groups in order to better understand LID, despite the occurrence of this phenomenon in the on medication state.  101  An important finding from this thesis is that PD subjects demonstrate susceptibility to ambiguous visual input. This knowledge may prove useful in the design of non-invasive aid devices intended to improve motor performance in PD. For example, it may be possible to filter higher frequency visual stimuli as a way to reduce visual ‘noise’ that could prove beneficial to patients. Interesting anecdotal evidence from the clinic suggests that when dyskinetic patients who are actively experiencing LID close their eyes, LID severity is reduced and in some cases is completely attenuated. This observation, in addition to the findings of this thesis, suggest that reducing visual input, and in particular noisy or ambiguous visual input, may significantly improve the motor performance of dyskinetic PD subjects. Expanding on this, as contrast sensitivity can be reduced in PD (Mestre et al., 1990; Hutton et al., 1993), visual aids that improve contrast vision may as a result reduce the uncertainty of visual cues, and could underscore a possible approach to improving motor performance in PD. Thus the significant finding of susceptibility to visual uncertainty in PD may prove useful in directing further studies examining the effect of vision on motor performance in PD, and further providing insight into how we can alter visual input received by PD patients to improve their experience of the disease. State estimation is a current area of interest in motor control studies. Our work here begins to investigate how altering the state of a visual target in a tracking task, and thereby affecting patients’ knowledge of their own state in relation to the visual cue, affects motor performance in PD. This symbolizes an important contribution to motor control literature, as it was not previously known if PD subjects possessed the same ability as healthy human subjects to de-weight visual uncertainty during motor performance. Furthermore, in using system identification techniques we are able to detect subtle differences in motor  102  performance of PD subjects, indicating that LDS models can be utilized as an important tool with which to investigate PD. LDS models further provide great potential in investigating state estimation in PD, as parameters from the models can be extracted that may provide further insight into the weighting of feedforward and feedback components of the motor control system. In summation, the results of this thesis are significant in that they provide insight into potential cerebellar and sensorimotor contribution to the pathophysiology of LID, detect subtle differences between dyskinetic and non-dyskinetic motor performance in the off medication state, probe more specifically the interaction between vision and motor performance in PD, and successfully use novel methods to differentiate PD sub-groups, which collectively provide future research and non-invasive therapy directions. 5.4  Study limitations 5.4.1  Subjects  A potential limitation of this thesis arises from the comparison of dyskinetic and nondyskinetic subjects. PD is a heterogeneous disease, and it is therefore logical to sub-group subjects as a way to better understand the spectrum of the disease. However, when comparing across sub-groups, there are invariably differences between the groups beyond the primary difference being studied – in our case the presence or absence of dyskinesias. For example, it can be noted in Table 3-1 that the there were differences between dyskinetic and non-dyskinetic subjects in the means of disease severity (UPDRS), daily equivalent dopaminergic intake (LEDD), and duration of disease (years since diagnosis). In all cases these differences were not significantly different, though the dyskinetic subjects had higher values of these parameters than non-dyskinetic subjects. This finding was not unexpected, as  103  evidence indicates that LID frequently manifest with increased disease progression and dopaminergic dose (Nyholm & Aquilonius, 2004; Jankovic, 2005). Furthermore, we note that all PD subjects were classified as Hoehn & Yahr 1-3, indicating that all subjects were considered to have mild to moderate disease severity (Hoehn & Yahr, 1967). Nonetheless, in order to further address this issue, we performed regression analyses to account for a possible impact of these differences on our results. In Chapter 3 we demonstrated that LEDD and disease duration did not significantly predict damping ratio. As disease severity and dyskinetic state were found to be significant in the regression model, we subsequently examined the effect of disease severity on damping ratio in both dyskinetic and non-dyskinetic groups. This result demonstrated that a given UPDRS predicted a significantly lower damping ratio for dyskinetic than for non-dyskinetic subjects. Therefore, though there were differences in these disease parameters between PD sub-groups, by accounting for them in our analyses we don’t believe that they significantly impacted our primary result revealing subtle differences in damping ratio specific to the dyskinetic state in PD. To address this issue in chapter 4, we examined the relationship between differences in RMS error between the maximum ambiguous and non-ambiguous tracking conditions and UPDRS, for both dyskinetic and non-dyskinetic groups. We found that these parameters were not significantly correlated for either group. We further found no significant correlation between UPDRS and decay rate for either PD groups. This demonstrates that the primary result indicating that dyskinetic PD subjects were significantly more susceptible to and reliant on increasing visual uncertainty was not associated with disease severity. Nonetheless, it should be acknowledged that we cannot completely exclude the possibility  104  that differences in disease severity impacted our results. Future studies recruiting dyskinetic and non-dyskinetic subjects should attempt to match UPDRS, however this may be difficult to achieve, and we note previous studies comparing dyskinetic and non-dyskinetic subjects that had significantly greater differences between the groups than we had in our studies (Wenzelburger et al., 2002). 5.4.2  The off-medication state  The standard paradigm for investigating PD and the influence of L-dopa is to study PD subjects after 12 hour L-dopa withdrawal (and 18 hour dopamine agonist withdrawal). The primary reason for this method is the ethical concern for interfering with subjects’ treatment plans for longer time periods. The issue of overnight withdrawal of medication was discussed with all of our PD subjects at time of recruiting. This practically defined off medication state is universally adopted in PD research, and its validity is supported by the symptomatic state in which PD patients present at time of study commencement. Any further time off medication for PD patients would probably not be justifiable on ethical grounds. Despite the universal use of this research method in PD research, it is possible that after the wearing-off of medication (as evidenced by the return of symptoms of the disease) dopaminergic medications still have some residual effects that would require a longer withdrawal time to attenuate. Thus, when interpreting our results in the off medication state, it is important to note that this may be influenced by a lingering effect of L-dopa and not representative of brain function of a non-treated PD patient. The recruitment of non-treated PD subjects might provide a reasonable control group for future studies, however this may be a difficult recruiting strategy to achieve as the majority of patients begin prompt therapy upon diagnosis due to the impact of the symptoms.  105  5.4.3  Visual contrast sensitivity  A potential limitation of this thesis is that we did not measure differences in visual contrast sensitivity between dyskinetic and non-dyskinetic subjects, though they have been shown to occur in PD (Mestre et al., 1990; Hutton et al., 1993). However, we specifically designed our visually guided tracking task to modulate visual ambiguity without using differences in contrast, such as blurring of the visual target as has been used in other studies (Kording & Wolpert, 2004; Wei & Kording, 2010). Furthermore, we don’t suspect that visual contrast significantly differed between PD sub-groups, as RMS tracking performance did not significantly differ between dyskinetic and non-dyskinetic groups during nonambiguous tracking conditions. As we did not alter the contrast of our tracking task from the non-ambiguous to ambiguous conditions, we believe it is unlikely differences in contrast sensitivity existed between PD sub-groups, and furthermore that if differences existed that they impacted our results. Nonetheless, future studies examining dyskinetic and nondyskinetic subjects may consider measuring this parameter (see Future directions). 5.5  Future directions  The results of this thesis demonstrate susceptibility to visual uncertainty in PD, and suggest that cerebellar dysfunction may be a significant contributor to the pathophysiology of LID. A future direction from this work is to examine the behavioural results demonstrated in this thesis in conjunction with imaging of cerebellar activity. For instance, it would be of great interest to determine if cerebellar activity is correlated with parkinsonian motor performance during the ambiguous tracking conditions of our task, and furthermore if reduced damping demonstrated in dyskinetic subjects correlates with cerebellar activity from fMRI. Studying dyskinetic subjects with imaging is difficult as the involuntary movements  106  of LID create large head motion that may preclude the use of fMRI. However, as the results of this thesis demonstrate abnormalities in the off medication state in dyskinetic PD subjects, imaging these subjects even when LID are not apparent may prove insightful into the pathophysiology of LID. We have elucidated differences in the motor response of dyskinetic and non-dyskinetic PD subjects faced with ambiguous visual input, and we interpret these results as indicative of a fundamental difference in the function of the cerebellum differentiating these two groups. However, further questions remain as to whether these differences may cause or worsen the involuntary movements of LID. The next step of this research is to quantify the effect of ambiguous visual input during motor performance on the magnitude and severity of LID, and to determine if the decay rate is correlated with LID scores. This would address the question of whether reliance on ambiguous visual feedback may actually contribute to LID, in addition to underlying differences in the motor system between dyskinetic and nondyskinetic subjects. Further research examining the relative effects of ambiguous visual input versus extraneous visual input on the motor performance of PD subjects is warranted. In our tracking task the jitter of the target was random and too fast to accurately ascertain its actual position, implying that the target’s position at any given moment had to be predicted. This could be contrasted to a task where the target’s actual position is known, but extraneous visual input is provided that may distract subjects’ response. For example, a target that smoothly follows a Lissajous trajectory with an additional target superimposed with ambiguous jitter, would faithfully provide the desired tracking position in addition to providing distracting visual input that would need to be neglected. This could be contrasted  107  to a task testing only ambiguous visual input, by having multiple targets simultaneously provided with a variable distribution of the distance between them, analogous to a ‘swarm of bees’. This type of experiment would further address the issue of reliance on visual feedback due to excessive cueing versus inadequate predictive motor control. Dyskinetic PD subjects were differentiated from non-dyskinetic subjects in this thesis by decreased damping during visually guided tracking. A future direction of study is to elucidate whether or not decreased damping ratios during motor performance could act as a biomarker of LID. If it could be determined that reduced damping ratios accurately predict which patients will develop LID, it would be of great interest to follow non-dyskinetic subjects as the disease progresses to examine how their damping ratio changes over time. It may be that a significantly reduced damping ratio in non-dyskinetic subjects predicts the development of LID in the near future, knowledge of which would be of great clinical importance. Investigation into the relationship between contrast sensitivity and ambiguous visual feedback in PD is warranted. For instance, reduced visual contrast in PD (Bulens et al., 1987; Mestre et al., 1990; Jones et al., 1992; Hutton et al., 1993) may impart ambiguity to the perception of normally non-ambiguous visual cues in everyday life. However, contrast sensitivity is a function of spatial frequency (Hutton et al., 1993), and evidence suggests greater deficits in contrast sensitivity exist at higher spatial frequencies in PD (Mestre et al., 1990; Jones et al., 1992). Motion discrimination can also be affected in PD (Trick et al., 1994), and is dependent on contrast such that lower contrast visual stimuli appear slower (Stone & Thompson, 1992). Therefore, in order to probe possible differences in visual contrast sensitivity between dyskinetic and non-dyskinetic PD subjects, a future direction of  108  study is to manipulate the contrast, velocity and spatial frequency of a visual target during visually guided tracking. A first step to this research direction is to generate normative data on differences in the interaction of visual contrast sensitivity, spatial frequency and motion perception between dyskinetic and non-dyskinetic PD subjects. Rather than using contrast sensitivity charts, a computer program to sensitively test the dependent interactions between these parameters may prove highly informative. As our work suggest that forward models may be compromised in PD, future studies using paradigms specifically designed to test the use of forward models and state estimation in PD would be highly anticipated. For example, reaching tasks using TMS to the cerebellum during arm movement, but before directed movement to a target, have provided intriguing results in healthy human subjects (Miall et al., 2007). Subjects produced reaches that were 138 ms out of date when TMS was delivered when the hand was already in motion but not when it was stationary, indicating out of date state estimation that was reliant on feedback rather than predictive forward models. The use of such techniques in a PD population could provide further detailed accounts of forward modeling and state estimation in PD. This thesis demonstrates that LDS models are a useful and sensitive method with which to quantify motor performance in PD. As such, a future direction is to probe the Kalman gain parameter that can be obtained from LDS models, as this may provide a direct quantification of the weight attributed to the sensory discrepancy between the actual and predicted sensory feedback. Kalman filtering has been applied in motor control studies using healthy human subjects (Wolpert et al., 1995; Baddeley et al., 2003), but not in studies examining motor performance in PD, and thus may prove to be an informative research direction.  109  BIBLIOGRAPHY Aarsland, D., Tandberg, E., Larsen, J. P., & Cummings, J. L. (1996). Frequency of dementia in Parkinson disease. Arch Neurol, 53(6), 538-542. Agid, Y., Ahlskog, E., Albanese, A., Calne, D., Chase, T., De Yebenes, J., et al. (1999). Levodopa in the treatment of Parkinson's disease: a consensus meeting. Mov Disord, 14(6), 911-913. Aisen, M. L., Arnold, A., Baiges, I., Maxwell, S., & Rosen, M. (1993). The effect of mechanical damping loads on disabling action tremor. Neurology, 43(7), 1346-1350. Andersen, R. A., & Gnadt, J. W. (1989) Posterior parietal cortex. In: The Neurobiology of Saccadic Eye Movment. Reviews in Oculomotor Research, edited by R. Wurtz and M. Goldberg. Amsterdam: Elsevier 3, 315-355. Andersson, G., & Armstrong, D. M. (1985) Climbing fibre input to b zone Purkinje cells during locomotor perturbation in the cat. Neurosci Lett, Supp 22, 27. Andersson, G., & Armstrong, D. M. (1987). Complex spikes in Purkinje cells in the lateral vermis (b zone) of the cat cerebellum during locomotion. J Physiol, 385, 107-134. Angelaki, D. E., Shaikh, A. G., Green, A. M., & Dickman, J. D. (2004). Neurons compute internal models of the physical laws of motion. Nature, 430(6999), 560-564. Armstrong, R. A. (2008). Visual signs and symptoms of Parkinson's disease. Clin Exp Optom, 91(2), 129-138. Au, W. L., Lei, N., Oishi, M. M., & McKeown, M. J. (2010). L-dopa induces under-damped visually guided motor responses in Parkinson's disease. Exp Brain Res, 202(3), 553559. Babin-Ratte, S., Sirigu, A., Gilles, M., & Wing, A. (1999). Impaired anticipatory finger gripforce adjustments in a case of cerebellar degeneration. Exp Brain Res, 128(1-2), 8185. Baddeley, R. J., Ingram, H. A., & Miall, R. C. (2003). System identification applied to a visuomotor task: near-optimal human performance in a noisy changing task. J Neurosci, 23(7), 3066-3075. Ballanger, B., Baraduc, P., Broussolle, E., Le Bars, D., Desmurget, M., & Thobois, S. (2008). Motor urgency is mediated by the contralateral cerebellum in Parkinson's disease. J Neurol Neurosurg Psychiatry, 79(10), 1110-1116.  110  Bastian, A. J. (2006). Learning to predict the future: the cerebellum adapts feedforward movement control. Curr Opin Neurobiol, 16(6), 645-649. Batista, A. P., Buneo, C. A., Snyder, L. H., & Andersen, R. A. (1999). Reach plans in eyecentered coordinates. Science, 285(5425), 257-260. Bays, P. M., & Wolpert, D. M. (2007). Computational principles of sensorimotor control that minimize uncertainty and variability. J Physiol, 578(Pt 2), 387-396. Becker, W. J., Kunesch, E., & Freund, H. J. (1990). Coordination of a multi-joint movement in normal humans and in patients with cerebellar dysfunction. Can J Neurol Sci, 17(3), 264-274. Bennett, D. A., Beckett, L. A., Murray, A. M., Shannon, K. M., Goetz, C. G., Pilgrim, D. M., et al. (1996). Prevalence of parkinsonian signs and associated mortality in a community population of older people. N Engl J Med, 334(2), 71-76. Beppu, H., Nagaoka, M., & Tanaka, R. (1987). Analysis of cerebellar motor disorders by visually-guided elbow tracking movement. 2. Contribution of the visual cues on slow ramp pursuit. Brain, 110 ( Pt 1), 1-18. Beppu, H., Suda, M., & Tanaka, R. (1984). Analysis of cerebellar motor disorders by visually guided elbow tracking movement. Brain, 107 ( Pt 3), 787-809. Bergman, H., & Deuschl, G. (2002). Pathophysiology of Parkinson's disease: from clinical neurology to basic neuroscience and back. Mov Disord, 17 Suppl 3, S28-40. Bezard, E., Dovero, S., Prunier, C., Ravenscroft, P., Chalon, S., Guilloteau, D., et al. (2001). Relationship between the appearance of symptoms and the level of nigrostriatal degeneration in a progressive 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine-lesioned macaque model of Parkinson's disease. J Neurosci, 21(17), 6853-6861. Blakemore, S. J., Frith, C. D., & Wolpert, D. M. (2001). The cerebellum is involved in predicting the sensory consequences of action. Neuroreport, 12(9), 1879-1884. Blakemore, S. J., & Sirigu, A. (2003). Action prediction in the cerebellum and in the parietal lobe. Exp Brain Res, 153(2), 239-245. Blakemore, S. J., Wolpert, D. M., & Frith, C. D. (1998). Central cancellation of selfproduced tickle sensation. Nat Neurosci, 1(7), 635-640. Bodis-Wollner, I., Marx, M. S., Mitra, S., Bobak, P., Mylin, L., & Yahr, M. (1987). Visual dysfunction in Parkinson's disease. Loss in spatiotemporal contrast sensitivity. Brain, 110 ( Pt 6), 1675-1698.  111  Bostan, A. C., Dum, R. P., & Strick, P. L. (2010). The basal ganglia communicate with the cerebellum. Proc Natl Acad Sci U S A, 107(18), 8452-8456. Bouthenet, M. L., Souil, E., Martres, M. P., Sokoloff, P., Giros, B., & Schwartz, J. C. (1991). Localization of dopamine D3 receptor mRNA in the rat brain using in situ hybridization histochemistry: comparison with dopamine D2 receptor mRNA. Brain Res, 564(2), 203-219. Braak, H., Braak, E., Yilmazer, D., de Vos, R. A., Jansen, E. N., Bohl, J., et al. (1994). Amygdala pathology in Parkinson's disease. Acta Neuropathol, 88(6), 493-500. Brainard, D. H. (2009) The Psychophysics Toolbox. Spat Vis, 10, 433-436. Broderick, M. P., Van Gemmert, A. W., Shill, H. A., & Stelmach, G. E. (2009). Hypometria and bradykinesia during drawing movements in individuals with Parkinson's disease. Exp Brain Res, 197(3), 223-233. Brown, S. H., Hefter, H., Mertens, M., & Freund, H. J. (1990). Disturbances in human arm movement trajectory due to mild cerebellar dysfunction. J Neurol Neurosurg Psychiatry, 53(4), 306-313. Bulens, C., Meerwaldt, J. D., Van der Wildt, G. J., & Van Deursen, J. B. (1987). Effect of levodopa treatment on contrast sensitivity in Parkinson's disease. Ann Neurol, 22(3), 365-369. Calabresi, P., Di Filippo, M., Ghiglieri, V., Tambasco, N., & Picconi, B. (2010). Levodopainduced dyskinesias in patients with Parkinson's disease: filling the bench-to-bedside gap. Lancet Neurol, 9(11), 1106-1117. Calne, D. B., Snow, B. J., & Lee, C. (1992). Criteria for diagnosing Parkinson's disease. Ann Neurol, 32 Suppl, S125-127. Calon, F., Grondin, R., Morissette, M., Goulet, M., Blanchet, P. J., Di Paolo, T., et al. (2000a). Molecular basis of levodopa-induced dyskinesias. Ann Neurol, 47(4 Suppl 1), S70-78. Calon, F., Hadj Tahar, A., Blanchet, P. J., Morissette, M., Grondin, R., Goulet, M., et al. (2000b). Dopamine-receptor stimulation: biobehavioral and biochemical consequences. Trends Neurosci, 23(10 Suppl), S92-100. Cenci, M. A. (2007). Dopamine dysregulation of movement control in L-DOPA-induced dyskinesia. Trends Neurosci, 30(5), 236-243. Chase, T. N. (1998). Levodopa therapy: consequences of the nonphysiologic replacement of dopamine. Neurology, 50(5 Suppl 5), S17-25.  112  Chaudhuri, K. R., & Schapira, A. H. (2009). Non-motor symptoms of Parkinson's disease: dopaminergic pathophysiology and treatment. Lancet Neurol, 8(5), 464-474. Chelaru, M. I., Duval, C., & Jog, M. (2010). Levodopa-induced dyskinesias detection based on the complexity of involuntary movements. J Neurosci Methods, 186(1), 81-89. Cheng, S., & Sabes, P. N. (2006). Modeling sensorimotor learning with linear dynamical systems. Neural Comput, 18(4), 760-793. Churchyard, A., & Lees, A. J. (1997). The relationship between dementia and direct involvement of the hippocampus and amygdala in Parkinson's disease. Neurology, 49(6), 1570-1576. Clower, D. M., West, R. A., Lynch, J. C., & Strick, P. L. (2001). The inferior parietal lobule is the target of output from the superior colliculus, hippocampus, and cerebellum. J Neurosci, 21(16), 6283-6291. Cooke, J. D., Brown, J. D., & Brooks, V. B. (1978). Increased dependence on visual information for movement control in patients with Parkinson's disease. Can J Neurol Sci, 5(4), 413-415. Crawford, J. D., Medendorp, W. P., & Marotta, J. J. (2004). Spatial transformations for eyehand coordination. J Neurophysiol, 92(1), 10-19. Damiano, A. M., McGrath, M. M., Willian, M. K., Snyder, C. F., LeWitt, P. A., Reyes, P. F., et al. (2000). Evaluation of a measurement strategy for Parkinson's disease: assessing patient health-related quality of life. Qual Life Res, 9(1), 87-100. Dancause, N., Ptito, A., & Levin, M. F. (2002). Error correction strategies for motor behavior after unilateral brain damage: short-term motor learning processes. Neuropsychologia, 40(8), 1313-1323. Davidson, P. R., & Wolpert, D. M. (2003). Motor learning and prediction in a variable environment. Curr Opin Neurobiol, 13(2), 232-237. Davidson, P. R., & Wolpert, D. M. (2005). Widespread access to predictive models in the motor system: a short review. J Neural Eng, 2(3), S313-319. Daw, N. D., Niv, Y., & Dayan, P. (2005). Uncertainty-based competition between prefrontal and dorsolateral striatal systems for behavioral control. Nat Neurosci, 8(12), 17041711. Day, B. L., Thompson, P. D., Harding, A. E., & Marsden, C. D. (1998). Influence of vision on upper limb reaching movements in patients with cerebellar ataxia. Brain, 121 ( Pt 2), 357-372.  113  de la Fuente-Fernandez, R., & Stoessl, A. J. (2002). Parkinson's disease: imaging update. Curr Opin Neurol, 15(4), 477-482. Defer, G. L. (1999). [Controversial issues concerning the initial treatment of Parkinson's disease: L-Dopa or dopaminergic agonists?]. Rev Neurol (Paris), 155(1), 43-45. Demirci, M., Grill, S., McShane, L., & Hallett, M. (1997). A mismatch between kinesthetic and visual perception in Parkinson's disease. Ann Neurol, 41(6), 781-788. 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. Nat Neurosci, 2(6), 563-567. Desmurget, M., & Grafton, S. (2000). Forward modeling allows feedback control for fast reaching movements. Trends Cogn Sci, 4(11), 423-431. Desmurget, M., Reilly, K. T., Richard, N., Szathmari, A., Mottolese, C., & Sirigu, A. (2009). Movement intention after parietal cortex stimulation in humans. Science, 324(5928), 811-813. Diener, H. C., Hore, J., Ivry, R., & Dichgans, J. (1993). Cerebellar dysfunction of movement and perception. Can J Neurol Sci, 20 Suppl 3, S62-69. Donchin, O., Francis, J. T., & Shadmehr, R. (2003). Quantifying generalization from trialby-trial behavior of adaptive systems that learn with basis functions: theory and experiments in human motor control. J Neurosci, 23(27), 9032-9045. Doya, K. (2000). Complementary roles of basal ganglia and cerebellum in learning and motor control. Curr Opin Neurobiol, 10(6), 732-739. Duhamel, J. R., Colby, C. L., & Goldberg, M. E. (1992). The updating of the representation of visual space in parietal cortex by intended eye movements. Science, 255(5040), 90-92. Fahn, S. (1999). Parkinson disease, the effect of levodopa, and the ELLDOPA trial. Earlier vs Later L-DOPA. Arch Neurol, 56(5), 529-535. Fahn, S. (2000). The spectrum of levodopa-induced dyskinesias. Ann Neurol, 47(4 Suppl 1), S2-9; discussion S9-11. Fahn, S. (2002) Results of the ELLDOPA (earlier vs. later levodopa) study. Mov Disord, 17, S13. Fahn, S. (2003). Description of Parkinson's disease as a clinical syndrome. Ann N Y Acad Sci, 991, 1-14.  114  Fenney, A., Jog, M. S., & Duval, C. (2008). Short-term variability in amplitude and motor topography of whole-body involuntary movements in Parkinson's disease dyskinesias and in Huntington's chorea. Clin Neurol Neurosurg, 110(2), 160-167. Flanagan, J. R., & Wing, A. M. (1997). The role of internal models in motion planning and control: evidence from grip force adjustments during movements of hand-held loads. J Neurosci, 17(4), 1519-1528. Flowers, K. (1978a). Lack of prediction in the motor behaviour of Parkinsonism. Brain, 101(1), 35-52. Flowers, K. (1978b). Some frequency response characteristics of Parkinsonism on pursuit tracking. Brain, 101(1), 19-34. Gao, J. H., Parsons, L. M., Bower, J. M., Xiong, J., Li, J., & Fox, P. T. (1996). Cerebellum implicated in sensory acquisition and discrimination rather than motor control. Science, 272(5261), 545-547. Gellman, R., Gibson, A. R., & Houk, J. C. (1985). Inferior olivary neurons in the awake cat: detection of contact and passive body displacement. J Neurophysiol, 54(1), 40-60. Glickstein, M. (2000). How are visual areas of the brain connected to motor areas for the sensory guidance of movement? Trends Neurosci, 23(12), 613-617. Glickstein, M., May, J. G., 3rd, & Mercier, B. E. (1985). Corticopontine projection in the macaque: the distribution of labelled cortical cells after large injections of horseradish peroxidase in the pontine nuclei. J Comp Neurol, 235(3), 343-359. Glickstein, M., & Stein, J. (1991). Paradoxical movement in Parkinson's disease. Trends Neurosci, 14(11), 480-482. Goetz, C. G., Stebbins, G. T., Shale, H. M., Lang, A. E., Chernik, D. A., Chmura, T. A., et al. (1994). Utility of an objective dyskinesia rating scale for Parkinson's disease: inter- and intrarater reliability assessment. Mov Disord, 9(4), 390-394. Golbe, L. I. (1991). Young-onset Parkinson's disease: a clinical review. Neurology, 41(2 ( Pt 1)), 168-173. Gour, J., Edwards, R., Lemieux, S., Ghassemi, M., Jog, M., & Duval, C. (2007). Movement patterns of peak-dose levodopa-induced dyskinesias in patients with Parkinson's disease. Brain Res Bull, 74(1-3), 66-74. Graybiel, A. M. (2000). The basal ganglia. Curr Biol, 10(14), R509-511.  115  Gritsenko, V., Yakovenko, S., & Kalaska, J. F. (2009). Integration of predictive feedforward and sensory feedback signals for online control of visually guided movement. J Neurophysiol, 102(2), 914-930. Hallett, M., Shahani, B. T., & Young, R. R. (1975). EMG analysis of patients with cerebellar deficits. J Neurol Neurosurg Psychiatry, 38(12), 1163-1169. Hoehn, M. M., & Yahr, M. D. (1967) “Parkinsonism: onset, progression and mortality.” Trends Neurosci, 14, 427-442. Hoff, J. I., van den Plas, A. A., Wagemans, E. A., & van Hilten, J. J. (2001). Accelerometric assessment of levodopa-induced dyskinesias in Parkinson's disease. Mov Disord, 16(1), 58-61. Holmes, G. (1917) The symptoms of acute cerebellar injuries due to gunshot injuries. Brain, 40, 461-535. Holmes, G. (1922) Clinical symptoms of cerebellar disease and their interpretation. The Croonian lectures I. Lancet, 1, 1177-1182. Hore, J., Wild, B., & Diener, H. C. (1991). Cerebellar dysmetria at the elbow, wrist, and fingers. J Neurophysiol, 65(3), 563-571. Hoshi, E., Tremblay, L., Feger, J., Carras, P. L., & Strick, P. L. (2005). The cerebellum communicates with the basal ganglia. Nat Neurosci, 8(11), 1491-1493. Hughes, A. J., Daniel, S. E., Blankson, S., & Lees, A. J. (1993). A clinicopathologic study of 100 cases of Parkinson's disease. Arch Neurol, 50(2), 140-148. Hutton, J. T., Morris, J. L., & Elias, J. W. (1993). Levodopa improves spatial contrast sensitivity in Parkinson's disease. Arch Neurol, 50(7), 721-724. Imamizu, H., Kuroda, T., Miyauchi, S., Yoshioka, T., & Kawato, M. (2003). Modular organization of internal models of tools in the human cerebellum. Proc Natl Acad Sci U S A, 100(9), 5461-5466. Inoue, K., Kawashima, R., Satoh, K., Kinomura, S., Goto, R., Koyama, M., et al. (1998). PET study of pointing with visual feedback of moving hands. J Neurophysiol, 79(1), 117-125. Ito, M. (2008). Control of mental activities by internal models in the cerebellum. Nat Rev Neurosci, 9(4), 304-313. Jankovic, J. (2005). Motor fluctuations and dyskinesias in Parkinson's disease: clinical manifestations. Mov Disord, 20 Suppl 11, S11-16.  116  Jeannerod, M. (1986). Mechanisms of visuomotor coordination: a study in normal and braindamaged subjects. Neuropsychologia, 24(1), 41-78. Johansson, R. S., & Cole, K. J. (1992). Sensory-motor coordination during grasping and manipulative actions. Curr Opin Neurobiol, 2(6), 815-823. Jones, R. D., & Donaldson, I. M. (1995). Fractionation of visuoperceptual dysfunction in Parkinson's disease. J Neurol Sci, 131(1), 43-50. Jones, R. D., Donaldson, I. M., & Timmings, P. L. (1992). Impairment of high-contrast visual acuity in Parkinson's disease. Mov Disord, 7(3), 232-238. Jordan, M. I, & Rumelhart, D. E. (1992) Forward models: supervised learning with a distal teacher. Cogn Sci, 16, 307–54. Jueptner, M., Ottinger, S., Fellows, S. J., Adamschewski, J., Flerich, L., Muller, S. P., et al. (1997). The relevance of sensory input for the cerebellar control of movements. Neuroimage, 5(1), 41-48. Kaji, R., Urushihara, R., Murase, N., Shimazu, H., & Goto, S. (2005). Abnormal sensory gating in basal ganglia disorders. J Neurol, 252 Suppl 4, IV13-IV16. Kalman, R. E. (1960) A New Approach to Linear Filtering and Prediction Problems. ASMEJ Basic Engineering, 82 (Series D), 35-45. Kamali, A., Kramer, L. A., Frye, R. E., Butler, I. J., & Hasan, K. M. (2010). Diffusion tensor tractography of the human brain cortico-ponto-cerebellar pathways: a quantitative preliminary study. J Magn Reson Imaging, 32(4), 809-817. Kawato, M. (1999). Internal models for motor control and trajectory planning. Curr Opin Neurobiol, 9(6), 718-727. Kawato, M., Kuroda, T., Imamizu, H., Nakano, E., Miyauchi, S., & Yoshioka, T. (2003). Internal forward models in the cerebellum: fMRI study on grip force and load force coupling. Prog Brain Res, 142, 171-188. Keijsers, N. L., Horstink, M. W., & Gielen, S. C. (2003). Movement parameters that distinguish between voluntary movements and levodopa-induced dyskinesia in Parkinson's disease. Hum Mov Sci, 22(1), 67-89. Kettner, R. E., Mahamud, S., Leung, H. C., Sitkoff, N., Houk, J. C., Peterson, B. W., et al. (1997). Prediction of complex two-dimensional trajectories by a cerebellar model of smooth pursuit eye movement. J Neurophysiol, 77(4), 2115-2130. Kitazawa, S., Kimura, T., & Yin, P. B. (1998). Cerebellar complex spikes encode both destinations and errors in arm movements. Nature, 392(6675), 494-497.  117  Klockgether, T., & Dichgans, J. (1994). Visual control of arm movement in Parkinson's disease. Mov Disord, 9(1), 48-56. Koch, G., Brusa, L., Carrillo, F., Lo Gerfo, E., Torriero, S., Oliveri, M., et al. (2009). Cerebellar magnetic stimulation decreases levodopa-induced dyskinesias in Parkinson disease. Neurology, 73(2), 113-119. Kording, K. P., & Wolpert, D. M. (2004). Bayesian integration in sensorimotor learning. Nature, 427(6971), 244-247. Kording, K. P., & Wolpert, D. M. (2006). Bayesian decision theory in sensorimotor control. Trends Cogn Sci, 10(7), 319-326. Krack, P., Benazzouz, A., Pollak, P., Limousin, P., Piallat, B., Hoffmann, D., et al. (1998). Treatment of tremor in Parkinson's disease by subthalamic nucleus stimulation. Mov Disord, 13(6), 907-914. Kumar, N., Van Gerpen, J. A., Bower, J. H., & Ahlskog, J. E. (2005). Levodopa-dyskinesia incidence by age of Parkinson's disease onset. Mov Disord, 20(3), 342-344. Kuo, A. D. (1995). An optimal control model for analyzing human postural balance. IEEE Trans Biomed Eng, 42(1), 87-101. Lang, A. E., & Lozano, A. M. (1998a). Parkinson's disease. First of two parts. N Engl J Med, 339(15), 1044-1053. Lang, A. E., & Lozano, A. M. (1998b). Parkinson's disease. Second of two parts. N Engl J Med, 339(16), 1130-1143. Lang, C. E., & Bastian, A. J. (1999). Cerebellar subjects show impaired adaptation of anticipatory EMG during catching. J Neurophysiol, 82(5), 2108-2119. Langston, J. W., Widner, H., Goetz, C. G., Brooks, D., Fahn, S., Freeman, T., et al. (1992). Core assessment program for intracerebral transplantations (CAPIT). Mov Disord, 7(1), 2-13. Lee, C. S., Samii, A., Sossi, V., Ruth, T. J., Schulzer, M., Holden, J. E., et al. (2000). In vivo positron emission tomographic evidence for compensatory changes in presynaptic dopaminergic nerve terminals in Parkinson's disease. Ann Neurol, 47(4), 493-503. Lewis, G. N., Byblow, W. D., & Walt, S. E. (2000). Stride length regulation in Parkinson's disease: the use of extrinsic, visual cues. Brain, 123 ( Pt 10), 2077-2090. Lewis, M. M., Slagle, C. G., Smith, A. B., Truong, Y., Bai, P., McKeown, M. J., et al. (2007). Task specific influences of Parkinson's disease on the striato-thalamo-cortical and cerebello-thalamo-cortical motor circuitries. Neuroscience, 147(1), 224-235.  118  Linazasoro, G. (2005). New ideas on the origin of L-dopa-induced dyskinesias: age, genes and neural plasticity. Trends Pharmacol Sci, 26(8), 391-397. Liu, X., Carroll, C. B., Wang, S. Y., Zajicek, J., & Bain, P. G. (2005). Quantifying druginduced dyskinesias in the arms using digitised spiral-drawing tasks. J Neurosci Methods, 144(1), 47-52. Liu, X., Osterbauer, R., Aziz, T. Z., Miall, R. C., & Stein, J. F. (2001). Increased response to visual feedback of drug-induced dyskinetic movements in advanced Parkinson's disease. Neurosci Lett, 304(1-2), 25-28. Liu, X., Tubbesing, S. A., Aziz, T. Z., Miall, R. C., & Stein, J. F. (1999). Effects of visual feedback on manual tracking and action tremor in Parkinson's disease. Exp Brain Res, 129(3), 477-481. Ljung, L. (1999) System identification: theory for the user. Upper Saddle River NJ, Prentice Hall. Majsak, M. J., Kaminski, T., Gentile, A. M., & Flanagan, J. R. (1998). The reaching movements of patients with Parkinson's disease under self-determined maximal speed and visually cued conditions. Brain, 121 ( Pt 4), 755-766. Mann, R. K., Edwards, R., Zhou, J., Jog, M., & Duval, C. (2010). Intra- and inter-limb coherency during stance in non-dyskinetic and dyskinetic patients with Parkinson's disease. Clin Neurol Neurosurg, 112(5), 392-399. Manto, M. (2009). Mechanisms of human cerebellar dysmetria: experimental evidence and current conceptual bases. J Neuroeng Rehabil, 6, 10. Manto, M., Godaux, E., & Jacquy, J. (1995). Detection of silent cerebellar lesions by increasing the inertial load of the moving hand. Ann Neurol, 37(3), 344-350. Marras, C., Lang, A., Krahn, M., Tomlinson, G., & Naglie, G. (2004). Quality of life in early Parkinson's disease: impact of dyskinesias and motor fluctuations. Mov Disord, 19(1), 22-28. Marras, C., & Lang, A. E. (2003). Measuring motor complications in clinical trials for early Parkinson's disease. J Neurol Neurosurg Psychiatry, 74(2), 143-146. Marsden, C. D., & Parkes, J. D. (1982) Fluctuations of disability in Parkinson's disease -clinical aspects, in Movement Disorders, C.D. Marsden and S. Fahn, Editors. Butterworth: London, 96-122. Martin, J. P. (1967) Locomotion and the basal ganglia. In: Martin JP. The basal ganglia and posture. London: Pitman Medical.  119  McIntyre, J., Zago, M., Berthoz, A., & Lacquaniti, F. (2001). Does the brain model Newton's laws? Nat Neurosci, 4(7), 693-694. Merfeld, D. M., Zupan, L., & Peterka, R. J. (1999). Humans use internal models to estimate gravity and linear acceleration. Nature, 398(6728), 615-618. Mestre, D., Blin, O., Serratrice, G., & Pailhous, J. (1990). Spatiotemporal contrast sensitivity differs in normal aging and Parkinson's disease. Neurology, 40(11), 1710-1714. Miall, R. C., & Wolpert, D. M. (1996) Forward models for physiological motor control. Neural Netw, 9, 1265–1279. Miall, R. C., Christensen, L. O., Cain, O., & Stanley, J. (2007). Disruption of state estimation in the human lateral cerebellum. PLoS Biol, 5(11), e316. Miall, R. C., Imamizu, H., & Miyauchi, S. (2000). Activation of the cerebellum in coordinated eye and hand tracking movements: an fMRI study. Exp Brain Res, 135(1), 22-33. Miall, R. C., & Reckess, G. Z. (2002). The cerebellum and the timing of coordinated eye and hand tracking. Brain Cogn, 48(1), 212-226. Miall, R. C., Reckess, G. Z., & Imamizu, H. (2001). The cerebellum coordinates eye and hand tracking movements. Nat Neurosci, 4(6), 638-644. Miall, R. C., Weir, D. J., & Stein, J. F. (1985). Visuomotor tracking with delayed visual feedback. Neuroscience, 16(3), 511-520. Miall, R. C., Weir, D. J., Wolpert, D. M., & Stein, J. F. (1993). Is the cerebellum a smith predictor? J Mot Behav, 25(3), 203-216. Middleton, F. A., & Strick, P. L. (1998). Cerebellar output: motor and cognitive channels. Trends Cogn Sci, 2(9), 348-354. Milner, A. D., Perrett, D. I., Johnston, R. S., Benson, P. J., Jordan, T. R., Heeley, D. W., et al. (1991). Perception and action in 'visual form agnosia'. Brain, 114 ( Pt 1B), 405428. Moore, A. P. (1987). Impaired sensorimotor integration in parkinsonism and dyskinesia: a role for corollary discharges? J Neurol Neurosurg Psychiatry, 50(5), 544-552. Morens, D. M., Davis, J. W., Grandinetti, A., Ross, G. W., Popper, J. S., & White, L. R. (1996). Epidemiologic observations on Parkinson's disease: incidence and mortality in a prospective study of middle-aged men. Neurology, 46(4), 1044-1050.  120  Morgante, F., Espay, A. J., Gunraj, C., Lang, A. E., & Chen, R. (2006). Motor cortex plasticity in Parkinson's disease and levodopa-induced dyskinesias. Brain, 129(Pt 4), 1059-1069. Morrice, B. L., Becker, W. J., Hoffer, J. A., & Lee, R. G. (1990). Manual tracking performance in patients with cerebellar incoordination: effects of mechanical loading. Can J Neurol Sci, 17(3), 275-285. Morris, M. E., Iansek, R., Matyas, T. A., & Summers, J. J. (1994). The pathogenesis of gait hypokinesia in Parkinson's disease. Brain, 117 ( Pt 5), 1169-1181. Morris, M. E., Iansek, R., Matyas, T. A., & Summers, J. J. (1996). Stride length regulation in Parkinson's disease. Normalization strategies and underlying mechanisms. Brain, 119 ( Pt 2), 551-568. Mrotek, L. A., & Soechting, J. F. (2007). Predicting curvilinear target motion through an occlusion. Exp Brain Res, 178(1), 99-114. Muller, F., & Dichgans, J. (1994). Dyscoordination of pinch and lift forces during grasp in patients with cerebellar lesions. Exp Brain Res, 101(3), 485-492. Myers, M. J., Fuchs, A. F., & Wallman, J. (1968) Neural mechanisms in visual guidance of limb movements. Arch Neurol, 7, 195-202. Ni, Z., Pinto, A. D., Lang, A. E., & Chen, R. (2010). Involvement of the cerebellothalamocortical pathway in Parkinson disease. Ann Neurol, 68(6), 816-824. Nimura, T., Ando, T., Yamaguchi, K., Nakajima, T., Shirane, R., Itoh, M., et al. (2004). The role of sigma-receptors in levodopa-induced dyskinesia in patients with advanced Parkinson disease: a positron emission tomography study. J Neurosurg, 100(4), 606610. Nowak, D. A., & Hermsdorfer, J. (2005). Grip force behavior during object manipulation in neurological disorders: toward an objective evaluation of manual performance deficits. Mov Disord, 20(1), 11-25. Nowak, D. A., Hermsdorfer, J., Marquardt, C., & Fuchs, H. H. (2002). Grip and load force coupling during discrete vertical arm movements with a grasped object in cerebellar atrophy. Exp Brain Res, 145(1), 28-39. Nyholm, D., & Aquilonius, S. M. (2004). Levodopa infusion therapy in Parkinson disease: state of the art in 2004. Clin Neuropharmacol, 27(5), 245-256. O'Suilleabhain, P., Bullard, J., & Dewey, R. B. (2001). Proprioception in Parkinson's disease is acutely depressed by dopaminergic medications. J Neurol Neurosurg Psychiatry, 71(5), 607-610.  121  Obeso, J. A., Guridi, J., & DeLong, M. (1997). Surgery for Parkinson's disease. J Neurol Neurosurg Psychiatry, 62(1), 2-8. Obeso, J. A., Marin, C., Rodriguez-Oroz, C., Blesa, J., Benitez-Temino, B., Mena-Segovia, J., et al. (2008). The basal ganglia in Parkinson's disease: current concepts and unexplained observations. Ann Neurol, 64 Suppl 2, S30-46. Obeso, J. A., Rodriguez-Oroz, M. C., Rodriguez, M., Lanciego, J. L., Artieda, J., Gonzalo, N., et al. (2000). Pathophysiology of the basal ganglia in Parkinson's disease. Trends Neurosci, 23(10 Suppl), S8-19. Ohye, C., Tsukahara, N., & Narabayashi, H. (1965) Rigidity and disturbance of reciprocal innervation. Confin Neurol, 26, 24-40. Oishi, M. M. K., Ashoori, A., & McKeown, M. J. (2010) Mode Detection in Switched Pursuit Tracking Tasks: Hybrid Estimation to Measure Performance in Parkinson's Disease. 49th IEEE Conference on Decision and Control. Atlanta, GA. Olanow, C. W., & Obeso, J. A. (2000). Pulsatile stimulation of dopamine receptors and levodopa-induced motor complications in Parkinson's disease: implications for the early use of COMT inhibitors. Neurology, 55(11 Suppl 4), S72-77; discussion S7881. Olanow, C. W., Obeso, J. A., & Stocchi, F. (2006). Continuous dopamine-receptor treatment of Parkinson's disease: scientific rationale and clinical implications. Lancet Neurol, 5(8), 677-687. Oscarsson, O. (1980) The Inferior Olivary Nucleus: Anatomy and Physiology, eds Courville J, DeMontigny C, Lamarre Y. Raven, New York, 279–289. Palmer, S. J., Ng, B., Abugharbieh, R., Eigenraam, L., & McKeown, M. J. (2009). Motor reserve and novel area recruitment: amplitude and spatial characteristics of compensation in Parkinson's disease. Eur J Neurosci, 29(11), 2187-2196. Pechevis, M. (2001) Direct and indirect costs of Parkinson’s disease and L-dopa induced dyskinesias: A prospective European study. Parkinsonism Relat Disord, 7, 106. Pechevis, M., Clarke, C. E., Vieregge, P., Khoshnood, B., Deschaseaux-Voinet, C., Berdeaux, G., et al. (2005). Effects of dyskinesias in Parkinson's disease on quality of life and health-related costs: a prospective European study. Eur J Neurol, 12(12), 956-963. Perenin, M. T., & Vighetto, A. (1988). Optic ataxia: a specific disruption in visuomotor mechanisms. I. Different aspects of the deficit in reaching for objects. Brain, 111 ( Pt 3), 643-674.  122  Pisella, L., Grea, H., Tilikete, C., Vighetto, A., Desmurget, M., Rode, G., et al. (2000). An 'automatic pilot' for the hand in human posterior parietal cortex: toward reinterpreting optic ataxia. Nat Neurosci, 3(7), 729-736. Price, M. J., Feldman, R. G., Adelberg, D., & Kayne, H. (1992). Abnormalities in color vision and contrast sensitivity in Parkinson's disease. Neurology, 42(4), 887-890. Quinn, N. P., Koller, W. C., Lang, A. E., & Marsden, C. D. (1986). Painful Parkinson's disease. Lancet, 1(8494), 1366-1369. Rascol, O., Sabatini, U., Brefel, C., Fabre, N., Rai, S., Senard, J. M., et al. (1998). Cortical motor overactivation in parkinsonian patients with L-dopa-induced peak-dose dyskinesia. Brain, 121 ( Pt 3), 527-533. Rascol, O., Sabatini, U., Fabre, N., Brefel, C., Loubinoux, I., Celsis, P., et al. (1997). The ipsilateral cerebellar hemisphere is overactive during hand movements in akinetic parkinsonian patients. Brain, 120 ( Pt 1), 103-110. Robinson, F. R., Cohen, J. L., May, J., Sestokas, A. K., & Glickstein, M. (1984). Cerebellar targets of visual pontine cells in the cat. J Comp Neurol, 223(4), 471-482. Rossetti, Y., Pisella, L., & Vighetto, A. (2003). Optic ataxia revisited: visually guided action versus immediate visuomotor control. Exp Brain Res, 153(2), 171-179. Saunders, J. A., & Knill, D. C. (2005). Humans use continuous visual feedback from the hand to control both the direction and distance of pointing movements. Exp Brain Res, 162(4), 458-473. Scheidt, R. A., Dingwell, J. B., & Mussa-Ivaldi, F. A. (2001). Learning to move amid uncertainty. J Neurophysiol, 86(2), 971-985. Schrag, A., & Quinn, N. (2000). Dyskinesias and motor fluctuations in Parkinson's disease. A community-based study. Brain, 123 ( Pt 11), 2297-2305. Schubotz, R. I. (2007). Prediction of external events with our motor system: towards a new framework. Trends Cogn Sci, 11(5), 211-218. Shadlen, M. N., & Newsome, W. T. (1998). The variable discharge of cortical neurons: implications for connectivity, computation, and information coding. J Neurosci, 18(10), 3870-3896. Simpson, J. I., Wylie, D. R., & De Zeeuw, C. I. (1996) On climbing fiber signals and their consequences(s). Behav Brain Sci, 19, 368-383.  123  Sirigu, A., Duhamel, J. R., Cohen, L., Pillon, B., Dubois, B., & Agid, Y. (1996). The mental representation of hand movements after parietal cortex damage. Science, 273(5281), 1564-1568. Sokoloff, P., Giros, B., Martres, M. P., Bouthenet, M. L., & Schwartz, J. C. (1990). Molecular cloning and characterization of a novel dopamine receptor (D3) as a target for neuroleptics. Nature, 347(6289), 146-151. Spencer, R. M., Zelaznik, H. N., Diedrichsen, J., & Ivry, R. B. (2003). Disrupted timing of discontinuous but not continuous movements by cerebellar lesions. Science, 300(5624), 1437-1439. Stein, J. F. (1986). Role of the cerebellum in the visual guidance of movement. Nature, 323(6085), 217-221. Stein, J. F., & Aziz, T. Z. (1999). Does imbalance between basal ganglia and cerebellar outputs cause movement disorders? Curr Opin Neurol, 12(6), 667-669. Stein, J. F., & Glickstein, M. (1992). Role of the cerebellum in visual guidance of movement. Physiol Rev, 72(4), 967-1017. Stevenson,	
   J.	
   K.,	
   Oishi,	
   M.	
   M.,	
   Farajian,	
   S.,	
   Cretu,	
   E.,	
   Ty,	
   E.,	
   &	
   McKeown,	
   M.	
   J.	
   (2011).	
   Response	
  to	
  sensory	
  uncertainty	
  in	
  Parkinson's	
  disease:	
  a	
  marker	
  of	
  cerebellar	
   dysfunction?	
  Eur	
  J	
  Neurosci,	
  33(2),	
  298-­‐305. Stocchi, F., Rascol, O., Kieburtz, K., Poewe, W., Jankovic, J., Tolosa, E., et al. (2010). Initiating levodopa/carbidopa therapy with and without entacapone in early Parkinson disease: the STRIDE-PD study. Ann Neurol, 68(1), 18-27. Stocchi, F., Vacca, L., Ruggieri, S., & Olanow, C. W. (2005). Intermittent vs continuous levodopa administration in patients with advanced Parkinson disease: a clinical and pharmacokinetic study. Arch Neurol, 62(6), 905-910. Stone, L. S., & Thompson, P. (1992). Human speed perception is contrast dependent. Vision Res, 32(8), 1535-1549. Synofzik, M., Lindner, A., & Thier, P. (2008). The cerebellum updates predictions about the visual consequences of one's behavior. Curr Biol, 18(11), 814-818. Takahashi, C. D., Scheidt, R. A., & Reinkensmeyer, D. J. (2001). Impedance control and internal model formation when reaching in a randomly varying dynamical environment. J Neurophysiol, 86(2), 1047-1051. Tamada, T., Miyauchi, S., Imamizu, H., Yoshioka, T., & Kawato, M. (1999) Activation of the cerebellum in grip force and load force coordination: an fMRI study. Neuroimage, 6, S492.  124  Tandberg, E., Larsen, J. P., Aarsland, D., & Cummings, J. L. (1996). The occurrence of depression in Parkinson's disease. A community-based study. Arch Neurol, 53(2), 175-179. Thoroughman, K. A., & Shadmehr, R. (2000). Learning of action through adaptive combination of motor primitives. Nature, 407(6805), 742-747. Tomko, G. J., & Crapper, D. R. (1974). Neuronal variability: non-stationary responses to identical visual stimuli. Brain Res, 79(3), 405-418. Trick, G. L., Kaskie, B., & Steinman, S. B. (1994). Visual impairment in Parkinson's disease: deficits in orientation and motion discrimination. Optom Vis Sci, 71(4), 242-245. Troiano, A. R., de la Fuente-Fernandez, R., Sossi, V., Schulzer, M., Mak, E., Ruth, T. J., et al. (2009). PET demonstrates reduced dopamine transporter expression in PD with dyskinesias. Neurology, 72(14), 1211-1216. Tseng, Y. W., Diedrichsen, J., Krakauer, J. W., Shadmehr, R., & Bastian, A. J. (2007). Sensory prediction errors drive cerebellum-dependent adaptation of reaching. J Neurophysiol, 98(1), 54-62. van Beers, R. J., Baraduc, P., & Wolpert, D. M. (2002). Role of uncertainty in sensorimotor control. Philos Trans R Soc Lond B Biol Sci, 357(1424), 1137-1145. van Beers, R. J., Sittig, A. C., & Gon, J. J. (1999). Integration of proprioceptive and visual position-information: An experimentally supported model. J Neurophysiol, 81(3), 1355-1364. Van Gemmert, A. W., Adler, C. H., & Stelmach, G. E. (2003). Parkinson's disease patients undershoot target size in handwriting and similar tasks. J Neurol Neurosurg Psychiatry, 74(11), 1502-1508. van Sonderen, J. F., Denier van der Gon, J. J., & Gielen, C. C. (1988). Conditions determining early modification of motor programmes in response to changes in target location. Exp Brain Res, 71(2), 320-328. Vaziri, S., Diedrichsen, J., & Shadmehr, R. (2006). Why does the brain predict sensory consequences of oculomotor commands? Optimal integration of the predicted and the actual sensory feedback. J Neurosci, 26(16), 4188-4197. Vercher, J. L., & Gauthier, G. M. (1988). Cerebellar involvement in the coordination control of the oculo-manual tracking system: effects of cerebellar dentate nucleus lesion. Exp Brain Res, 73(1), 155-166.  125  Vercher, J. L., & Gauthier, G. M. (1992). Oculo-manual coordination control: ocular and manual tracking of visual targets with delayed visual feedback of the hand motion. Exp Brain Res, 90(3), 599-609. Vitale, C., Pellecchia, M. T., Grossi, D., Fragassi, N., Cuomo, T., Di Maio, L., et al. (2001). Unawareness of dyskinesias in Parkinson's and Huntington's diseases. Neurol Sci, 22(1), 105-106. Wang, T., Dordevic, G. S., & Shadmehr, R. (2001). Learning the dynamics of reaching movements results in the modification of arm impedance and long-latency perturbation responses. Biol Cybern, 85(6), 437-448. Wei, K., & Kording, K. (2010). Uncertainty of feedback and state estimation determines the speed of motor adaptation. Front Comput Neurosci, 4, 11. Wenzelburger, R., Zhang, B. R., Pohle, S., Klebe, S., Lorenz, D., Herzog, J., et al. (2002). Force overflow and levodopa-induced dyskinesias in Parkinson's disease. Brain, 125(Pt 4), 871-879. Wolpert, D. M., & Ghahramani, Z. (2000). Computational principles of movement neuroscience. Nat Neurosci, 3 Suppl, 1212-1217. Wolpert, D. M., Ghahramani, Z., & Jordan, M. I. (1995). An internal model for sensorimotor integration. Science, 269(5232), 1880-1882. Yin, H. H., & Knowlton, B. J. (2006). The role of the basal ganglia in habit formation. Nat Rev Neurosci, 7(6), 464-476. Yu, H., Sternad, D., Corcos, D. M., & Vaillancourt, D. E. (2007). Role of hyperactive cerebellum and motor cortex in Parkinson's disease. Neuroimage, 35(1), 222-233. Zago, M., Bosco, G., Maffei, V., Iosa, M., Ivanenko, Y. P., & Lacquaniti, F. (2004). Internal models of target motion: expected dynamics overrides measured kinematics in timing manual interceptions. J Neurophysiol, 91(4), 1620-1634. Zago, M., McIntyre, J., Senot, P., & Lacquaniti, F. (2009). Visuo-motor coordination and internal models for object interception. Exp Brain Res, 192(4), 571-604. Zgaljardic, D. J., Borod, J. C., Foldi, N. S., & Mattis, P. (2003). A review of the cognitive and behavioral sequelae of Parkinson's disease: relationship to frontostriatal circuitry. Cogn Behav Neurol, 16(4), 193-210. Zupan, L. H., Merfeld, D. M., & Darlot, C. (2002). Using sensory weighting to model the influence of canal, otolith and visual cues on spatial orientation and eye movements. Biol Cybern, 86(3), 209-230.  126  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0103332/manifest

Comment

Related Items