<%BANNER%>

An Event-Related Potential (ERP) Study Examining Reward Processing and Mood in Parkinson's Disease and Healthy Aging

Permanent Link: http://ufdc.ufl.edu/UFE0041902/00001

Material Information

Title: An Event-Related Potential (ERP) Study Examining Reward Processing and Mood in Parkinson's Disease and Healthy Aging
Physical Description: 1 online resource (144 p.)
Language: english
Creator: Kellison, Ida
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: aging, ern, event, executive, frn, parkinson
Clinical and Health Psychology -- Dissertations, Academic -- UF
Genre: Psychology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: AN EVENT-RELATED POTENTIAL (ERP) STUDY EXAMINING REWARD PROCESSING AND MOOD IN PARKINSON'S DISEASE AND HEALTHY AGING The ability to appropriately respond to environmental feedback is crucial for successful living. Positive and negative feedback differentially activate reinforcement learning systems in the brain and reliance on external feedback changes adaptively over the course of learning so that the probability of success or failure can be predicted prior to action. Despite significant advances in our understanding of the reinforcement learning system, questions remain, particularly regarding outcomes of damage to the system. Older age and Parkinson's disease (PD) are known to be associated with impairments in reinforcement learning and damage to the neural networks underlying the processing of errors and feedback. In order to better understand the nature of these impairments, two event-related potential (ERP) experiments were conducted in young adults, community-dwelling older adults, and PD patients. The experiments used a probabilistic learning task that manipulated valence and feedback probability in order to investigate differential valence-related reactivity, error detection, and feedback processing, as measured by response-locked and feedback-locked ERP components. Within the young adult control group, predictions from the reinforcement learning theory were generally upheld for both error-related and feedback-related processing, providing a basis of comparison for examination of age-related changes. Evidence for age-related changes was primarily found in error-related processing. Interpreted within the context of reinforcement learning theory, this was attributed to failure to develop internalized representations of responses with learning in the older adult controls. Consistent with predictions, some evidence was found for relative preservation of activity related to positive outcomes as compared to negative outcomes of actions. Contrary to expectations, patients with Parkinson's disease did not demonstrate impairments in reinforcement learning compared to older adult controls with respect to accuracy or ERP reflections of error- and feedback-processing. In fact, some evidence was found for increased response-related activity in PD patients versus controls. Based on valence-related patterns found in the data, hypotheses were generated based on reinforcement learning theory and the effects of aging, Parkinson's disease and dopaminergic medication on positive and negative feedback processing. Taken together, the results of these experiments provide support for the reinforcement learning theory of the ERN and bolster the validity of requests for inclusion of the role of positive information processing in the theory. Moreover, these results help to clarify the impact of aging and Parkinson's disease on learning and decision-making and have implications for designing and improving interventions for cognitive impairments.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Ida Kellison.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Perlstein, William.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041902:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041902/00001

Material Information

Title: An Event-Related Potential (ERP) Study Examining Reward Processing and Mood in Parkinson's Disease and Healthy Aging
Physical Description: 1 online resource (144 p.)
Language: english
Creator: Kellison, Ida
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: aging, ern, event, executive, frn, parkinson
Clinical and Health Psychology -- Dissertations, Academic -- UF
Genre: Psychology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: AN EVENT-RELATED POTENTIAL (ERP) STUDY EXAMINING REWARD PROCESSING AND MOOD IN PARKINSON'S DISEASE AND HEALTHY AGING The ability to appropriately respond to environmental feedback is crucial for successful living. Positive and negative feedback differentially activate reinforcement learning systems in the brain and reliance on external feedback changes adaptively over the course of learning so that the probability of success or failure can be predicted prior to action. Despite significant advances in our understanding of the reinforcement learning system, questions remain, particularly regarding outcomes of damage to the system. Older age and Parkinson's disease (PD) are known to be associated with impairments in reinforcement learning and damage to the neural networks underlying the processing of errors and feedback. In order to better understand the nature of these impairments, two event-related potential (ERP) experiments were conducted in young adults, community-dwelling older adults, and PD patients. The experiments used a probabilistic learning task that manipulated valence and feedback probability in order to investigate differential valence-related reactivity, error detection, and feedback processing, as measured by response-locked and feedback-locked ERP components. Within the young adult control group, predictions from the reinforcement learning theory were generally upheld for both error-related and feedback-related processing, providing a basis of comparison for examination of age-related changes. Evidence for age-related changes was primarily found in error-related processing. Interpreted within the context of reinforcement learning theory, this was attributed to failure to develop internalized representations of responses with learning in the older adult controls. Consistent with predictions, some evidence was found for relative preservation of activity related to positive outcomes as compared to negative outcomes of actions. Contrary to expectations, patients with Parkinson's disease did not demonstrate impairments in reinforcement learning compared to older adult controls with respect to accuracy or ERP reflections of error- and feedback-processing. In fact, some evidence was found for increased response-related activity in PD patients versus controls. Based on valence-related patterns found in the data, hypotheses were generated based on reinforcement learning theory and the effects of aging, Parkinson's disease and dopaminergic medication on positive and negative feedback processing. Taken together, the results of these experiments provide support for the reinforcement learning theory of the ERN and bolster the validity of requests for inclusion of the role of positive information processing in the theory. Moreover, these results help to clarify the impact of aging and Parkinson's disease on learning and decision-making and have implications for designing and improving interventions for cognitive impairments.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Ida Kellison.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Perlstein, William.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041902:00001


This item has the following downloads:


Full Text

PAGE 1

1 AN EVENT RELATED POTENTIAL (ERP) STUDY EXAMINING REWARD By IDA LILLIAN KELLISON A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Ida Lillian Kellison

PAGE 3

3 To those who might benefit from this project

PAGE 4

4 ACKNOWLEDGMENTS I would like to acknowledg e my dissertation chair and mentor, William M. Perlstein, Ph.D. for his encouragement and support of this project. I would also thank to thank my committee members, Russell M. Bauer, Ph.D., W. Keith Berg, Ph.D., Michael Marsiske, Ph.D., John C. Rosenbek, P h.D., and Frank Skidmore, M. D. for their helpful advice and feedback. Finally, I expre ss gratitude to Chris Sozda, M.S ., Sarah Key DeLyria, M.A., David Stigge Kaufman, Ph.D. and Michael Larson, Ph.D. for their invaluable assistance generosity humor, and kindness throughout this process. This research was supported by a scholarship awarded by the American Psychological Foundation.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ ........ 10 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 GENERAL INTRODUCTION AND METHODS ................................ ....................... 14 General Introduction ................................ ................................ ............................... 14 Neuroanatomical Model of Reinforcement Learning ................................ ........ 16 Ventral striatum and dopamine ................................ ................................ .. 16 Anterior cingulate cortex ................................ ................................ ............ 17 Orbitofro ntal cortex and amygdala ................................ ............................. 18 Electrophysiological Model of Reinforcement Learning ................................ .... 19 Measuring error detection and feedback proce ssing using ERPs .............. 20 Reinforcement learning theory of the ERN ................................ ................. 21 Predictions from the reinforcement learning theory ................................ .... 22 Executive Function and Reinforcement Learning in Aging ............................... 23 ERP Studies of Reinforcement Learning in Older Adults ................................ 24 ........ 26 ..... 29 ................................ .................. 31 Interactions Between Cognition and Emotion Importance of the ACC .......... 32 Summary and Rationale for the Present Study ................................ ................ 35 General Methods ................................ ................................ ................................ .... 38 ERP Task Stimuli and Procedures ................................ ................................ ... 3 8 Task stimuli ................................ ................................ ................................ 38 Trial procedure ................................ ................................ ........................... 39 Neuropsycho logical Measures ................................ ................................ ......... 40 Emotional Measures ................................ ................................ ......................... 41 EEG Data Acquisition and Reduction ................................ ............................... 41 Data Analysis ................................ ................................ ................................ ... 43 2 EXPERIMENT 1: THE EFFECT OF AGING ON ERROR DETECTION AND FEEDBACK PROCESSING ................................ ................................ .................... 46 Overview and Predictions ................................ ................................ ....................... 46 Methods ................................ ................................ ................................ .................. 47 Participants ................................ ................................ ................................ ....... 47 Procedures ................................ ................................ ................................ ....... 48

PAGE 6

6 Results ................................ ................................ ................................ .................... 48 Behavioral Data ................................ ................................ ................................ 48 Reinforcement learning task performance ................................ ........................ 48 Neuropsychological and emotional functioning ................................ ................ 50 Event Related Potential Data ................................ ................................ ........... 51 Response locked ERPs (ERN/CRN) ................................ ................................ 51 Feedback locked ERPs (FRN) ................................ ................................ ......... 56 Relationship to performance on neuropsychologic al tests ............................... 59 Discussion ................................ ................................ ................................ .............. 60 Specific Aim 1: Support for the Reinforcement Learning Theory ...................... 61 Behavioral data ................................ ................................ ................................ 61 Response locked ERPs (ERN/CRN) ................................ ................................ 62 Feedback locked ERPs (FRN) ................................ ................................ ......... 63 Specific Aim 2: Effects of Aging on ERP Reflections of Reinforcement Learning ................................ ................................ ................................ ........ 65 Response locked ERPs (ERN/CRN) ................................ ................................ 65 Feedback locked ERPs (FRN) ................................ ................................ ......... 67 Specific Aim 3: Relationship to Neuropsychological Test Performance ........... 69 3 DETECTION AND FEEDBACK PROCESSING ................................ ..................... 87 Overview and Predictions ................................ ................................ ....................... 87 Methods ................................ ................................ ................................ .................. 88 Participants ................................ ................................ ................................ ....... 88 Procedures ................................ ................................ ................................ ....... 89 Results ................................ ................................ ................................ .................... 90 Behavioral Data ................................ ................................ ................................ 90 Reinforcement learning task performance ................................ ........................ 90 Cognitive an d emotional functioning ................................ ................................ 91 Event Related Potential Data ................................ ................................ ........... 92 Response locked ERPs (ERN/CRN) ................................ ................................ 92 Feedback locked ERPs (FRN) ................................ ................................ ......... 95 Relationship to performance on neuropsychological tests and disease variables ................................ ................................ ................................ ........ 98 Discussion ................................ ................................ ................................ .............. 98 Effects of Aging and PD on ERP Reflections of Reinforcement Learning ........ 99 Behavioral data ................................ ................................ ................................ 99 Response locked ERPs (ERN/CRN) ................................ .............................. 100 Feedback locked ERPs (FRN) ................................ ................................ ....... 102 Relatio nship to Neuropsychological Test Performance ................................ .. 105 4 GENERAL DISCUSSION ................................ ................................ ..................... 123 Review and Conclusions ................................ ................................ ....................... 123 Strengths and Limitations ................................ ................................ ..................... 126 Implications and Future Directions ................................ ................................ ........ 127

PAGE 7

7 LIST OF REFERENCES ................................ ................................ ............................. 130 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 144

PAGE 8

8 LIST OF TABLES Table page 2 1 Means and standard deviations ( SD ) of demographic and neuropsychological data for younger and older participants. ................................ ............................. 71 2 2 Mean response time ( SD ) in the three validity conditions (100%, 80%, and 60%), displayed separately fo r the three bins and two age groups. ................... 71 2 3 Mean accuracy (SD) in the three validity conditions (100%, 80%, and 60%), displayed separately for the three bins and two age groups. .............................. 72 2 4 Mean ( SD ) number of trials per condition in each age group. ............................ 72 2 5 Mean amplitudes ( V) of the ERN in the three validity conditions (10 0%, 80%, and 60%), displayed separately for the three bins and two age groups. ... 72 2 6 Mean amplitudes ( V) of the CRN in the three validity conditions (100%, 80%, and 60% validity), displayed separately for the three bins and two age groups. ................................ ................................ ................................ ............... 72 2 7 Summary of the 2 Group x 2 Response Type x 3 Validity x 3 Bin repeated measures ANOVA conducted on the ERN/CRN mean amplitude data. ............. 73 2 8 Mean amplitudes ( V) of the non reward related FRN in the three validity conditions (100%, 80%, and 60% validity), displayed separately for the three bins and two age groups. ................................ ................................ ................... 73 2 9 Mean amplitudes ( V) of the reward related FRN in the three validity conditions (100%, 80%, and 60% validity), displayed separately for the three bins and two age groups. ................................ ................................ ................... 73 2 10 Summary of the 2 Group x 2 Feedback Type x 3 Validity x 3 Bin ANOVA conducted on the reward and non reward FRN mean amplitude data. ............. 74 2 11 Signif icant correlations between FRN difference waves and neuropsychological measures for younger and older groups combined. ............ 74 3 1 Demographic and neuropsychological data for older adults and patie nts with ................................ ................................ ......................... 108 3 2 Mean response time (SD) in the three validity conditions (100%, 80%, and 60%), displayed separately for the three bins and two groups. ........................ 109 3 3 Mean accuracy (SD) in the three validity conditions (100%, 80%, and 60%), displayed separately for the three bins and two groups. ................................ ... 109

PAGE 9

9 3 4 M ean ( SD ) number of trials per condition in each group. ................................ 109 3 5 Mean amplitudes ( V) of the ERN in the three validity conditions displayed separately for the three bins and two groups. ................................ ................... 109 3 6 Mean amplitudes ( V) of the CRN in the three validity conditions displayed separately for the three bins and two groups. ................................ ................... 109 3 7 Mean amplitudes ( V) of the non reward related FRN in the three validity conditions displayed separately for the three bins and two groups. .................. 110 3 8 Mean amplitudes ( V) of the reward related FRN in the three validity conditions (100%, 80%, and 60% validity), displayed separately for the three bins and two groups. ................................ ................................ ........................ 110 3 9 Significant correlations between FRN difference waves and ne uropsychological measures for PD and older control groups combined. ...... 110

PAGE 10

10 LIST OF FIGURES Figure page 1 1 Typical example of event related brain pote ntials associated with negative and positive feedback recorded from electrode FCz.. ................................ ........ 45 2 1 Accuracy over time in each of the three validity conditions displayed separately for A) younger and B ) older adults. ................................ ................... 75 2 2 Spherical spline voltage maps for the ERN CRN difference waves in both grou ps, taken at 60 ms ................................ ................................ ....................... 76 2 3 Am plitude of the ERN CRN difference wave in each condition over time displayed separately for A) younger and B) older adults. ................................ ... 77 2 4 Grand averaged response locked ERPs taken from electrode FCz displayed separately for each group in each validity condition collapsed across all three bins.. ................................ ................................ ................................ ................... 78 2 5 Grand averaged response locked ERPs at electrode FCz demonstrating learning related effe cts for each group in the 100% validity condition.. .............. 79 2 6 ERN amplitudes in each condition over time displayed separately for A) younger and B) older adults. CRN amplitudes in each condition over time displayed separately for C) younger and D) older adults ................................ .... 80 2 7 Spherical spline voltage maps for the FRN cFRN difference waves in both groups. ................................ ................................ ................................ ............... 81 2 8 Amplitude of the non reward minus reward FRN difference wave in each condition over time displayed separately for A) younger and B) older adults. .... 82 2 9 Grand averaged feedback locked ERPs taken from electrode FCz displayed separately for each group in each validity condition collapsed across the bins. ................................ ................................ ................................ .................... 83 2 10 Grand averaged feedback related ERPs t aken from electrode FCz demonstrating learning related effects for each group in the 100% validity condition. ................................ ................................ ................................ ............ 84 2 11 Non reward related FRN amplitudes in each condition over time displayed se parately for A) younger and B) older adults. Reward related FRN amplitudes in each condition over time displayed separately for C) younger and D) older a dults. ................................ ................................ ............................ 85

PAGE 11

11 2 12 A) Mean accuracy in each condition. B) Mean amplitude of the ERN CRN difference wave in each condition. C) Mean amplitude of the FRN to non reward minus the FRN to reward difference wave in each condition. ................ 86 3 1 Ac curacy over time in each of the three validity conditions displayed separately for A) older adult controls and B) patients with PD. ......................... 111 3 2 Spherical spline voltage maps for the ERN CRN diff erence waves in both groups. ................................ ................................ ................................ ............. 112 3 3 Amplitude of the ERN CRN difference wave in each condition over time displayed separately for A) older controls and B) patients with PD. ................. 113 3 4 Grand averaged response locked ERPs taken from electrode FCz displayed separately for each group in each validity condition collapsed across all three bins. ................................ ................................ ................................ .................. 114 3 5 Grand averaged response locked ERPs at electrode FCz demonstrating learning related effects for each group in the 100% validity condition. ............. 115 3 6 ERN amplitudes in each condition over time displayed separately for A) older controls and B) patients with PD. CRN amplitudes in each condition over time displayed separately for C) older controls and D) patients with PD.. ........ 116 3 7 Spherical spline voltage maps for the non reward minus reward difference waves in both groups. ................................ ................................ ....................... 117 3 8 Amplitude of the non reward minus reward FRN difference wave in each condition over time displayed separately for A) older controls and B) patients with PD. ................................ ................................ ................................ ............ 118 3 9 Grand averaged feedback locked ERPs taken from electrode FCz displayed separately for each grou p in each validity condition collapsed across the bins. ................................ ................................ ................................ .................. 119 3 10 Grand averaged feedback related ERPs taken from electrode FCz demonstrating learning related effects for each group in the 100% validity condition. ................................ ................................ ................................ .......... 120 3 11 Non reward related FRN amplitudes in each condition over time displayed separately for A) older controls and B) patients with PD. Reward related FRN amplitudes in each condition over time displayed separately for C) older controls and D) patients with PD. ................................ ................................ ..... 121 3 12 A) Mean accuracy in each condition. B) Mean amplitude of the ERN CRN difference wave in e ach condition. C) Mean amplitude of the non reward FRN minus reward FRN difference wave in eac h condition. ............................ 122

PAGE 12

12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in P artial Fulfillment of the Requirements for the Degree of Doctor of Philosophy AN EVENT RELATED POTENTIAL (ERP) STUDY EXAMINING REWARD By Ida Lillian Kellison August 2010 Chair: William M. Perl stein Major: Psychology The ability to appropriately respond to environmental feedback is crucial for successful living. Positive and negative feedback differentially activate reinforcement learning systems in the brain and reliance on external feedback changes adaptively over the course of learning so that the probability of success or failure can be predicted prior to action. Despite significant advances in our understanding of the reinforcement learning system questions remain particularly regarding outcomes of damage to the system Older age are known to be associated with impairment s in reinforcement learning and damage to the neural networks underlying the processing of errors and feedback In order to better understan d the nature of these impairment s two event related potential (ERP) experiments were conducted in young adults, community dwelling older adults and PD patients. The experiments used a probabilistic learning task that manipulated valence and feedback prob ability in order to investigate differential valence related reactivity, error detection, and feedback processing, as measured by response locked and feedback locked ERP components. Within the young adult control group, predictions from the reinforcement l earning theory were generally upheld for both error related and feedback related processing,

PAGE 13

13 providing a basis of comparison for examination of age related changes. Evidence for age related changes was primarily found in error related processing. Interpret ed within the context of reinforcement learning theory, this was attributed to failure to develop internalized representations of responses with learning in the older adult controls. Consistent with predictions, some evidence was found for relative preser vation of activity related to positive outcomes as compared to negative outcomes of actions. Contrary to expectations, patients with P disease did not demonstrate impairment s in reinforcement learning compared to older adult controls with respect to accuracy or ERP reflections of error and feedback processing. In fact, some evidence was found for increased response related activity in PD patients versus controls. Based on valence related patterns found in the data, hypotheses were generated base d on dopaminergic medication on positive and negative feedback processing. Taken together, the results of these experiments provide support for the reinforcement learning theor y of the ERN and bolster the validity of requests for inclusion of the role of positive information processing in the theory. Moreover, these decision making and have impl ications for designing and improving interventions for cognitive impairment s

PAGE 14

14 CHAPTER 1 GENERAL INTRODUCTION AND METHODS General Introduction Monitoring performance in order to appropriately and flexibly adapt decision making in response to environmenta l feedback is crucial for survival. The process of responding to both positive and negative environmental feedback has been called (Holroyd & Coles, 2002) ; the key principle behind this type of learning is that be havior followed by positive or rewarding outcomes is more likely to recur, whereas behavior followed by negative or punishing outcomes is less likely to recur. The process of predicting evaluating and responding to such rewards and punishments is complex and involves dopaminergic activity in frontal and subcortical regions of the brain (Schultz, Tremblay, & Hollerman, 2000) Unfortunately, d espite the apparent importance of this p rocess for learning and succe ssful living it has been hypothesized that frontal regions of the brain subserving them may be particularly susceptible to dysfunct ion as we age (West, 2000) ; in addition d ec lines in dopaminergic activity that occur in older age may negatively impact problem solving, decision making, and other related aspects of executive f unction (Band, Ridderinkhof, & Segalowitz, 2002; Woodruff Pak, 1997) Similarly, t he more pathological depletion of dopaminergic neurons in the substantia nigra that occurs in Parkinson disease (PD) also leads to dysfunction of frontal subco rtical circuitry resulting in deficits in frontal executive function (Zgaljardic et al., 2006) including aspects of performance monitoring such as error detection and feedback processing (Band et al., 2002; Schmitt Eliassen, Ferstl, Wiesner, Deuschl, & Witt, 2007; Schott et al., 2007) as well as mood symptoms (e.g.,

PAGE 15

15 depression and apathy) Reports of impairment s in reinforcement learning are inconsistent in both aging and PD Moreover, no studies have examined relationships between aspects of reinforcement learning (i.e., error detection and feedback monitoring) and emotional symptoms (i.e., apathy and depression) reported by patients despite the fa ct that associations have been found between these symptoms and other elements of executive dysfunction (Isella et al., 2002; Pluck & Brown 2002; Starkstein et al., 1992 ) T he overall goal of this dissertation stu dy is to examine the neuro physiological and neuropsychological underpinnings of error detection, feedback processing, and mood (i.e., apathy and depression) by comparing a community dwelling group of older adults with suspected subclinical dopaminergic dys function to a PD group with clinical dopaminergic dysfunction. The may provide a useful model for understanding relative levels of impairment and preserved ability in performance monitoring between these groups and how aspects of performance monitoring (i.e., error detection and reinforcement learning ) relat e to emotional symptoms Specifically, the following research questions will be addressed: 1. What effect do feedback valence and probability have on reinforcement learning (i.e., error detection and feedback processing ) over time? 2. detect errors and process positive and negative feedback? 3. How are error detection and feedback processing relat ed to other aspects of cognitive and emotional functioning in young adults, community dwelling older Although aspects of executive function such as performance monitoring (e.g., the ability to problem solve us ing examiner feedback) are commonly assessed clinically by the use of neuropsychological tests, behavioral assessment of performance monitoring

PAGE 16

16 is difficult and complicated by the fact that these functions requ ire multiple overlapping skills As a result, use of a more direct measurement instrument may be necessary to elucidate dysfunctional component processes that may contribute to impairment s in performance monitoring The field of cognitive neuroscience provides one such method, namely the use of high density electroencephalography to measure rapid changes in scalp recorded brain event related potentials (ERPs). ERPs have been used to examine two aspects of performance monitoring relevant to reinforcement learning (i.e., error detection and feedback mo nitoring). This chapter will discuss neuroanatomical and electro physiological model s of reinforcement learning, review the use of ERPs for studying error detection and feedback processing in aging and PD, and conclude with an introduction to the general m ethods used in this study. Neuroanatomical Model of Reinforcement Learning Structures involved in reward processing have been identified through human fMRI studies and animal studies using single cell recording. Areas consistently demonstrating activation include the ventral striatum, regions of the frontal lobe (i.e., the orbitofrontal cortex, or OFC, and the anterior cingulate cortex, or ACC), and the amygdala (Knutson & Cooper, 2005; Knutson, Fong, Adams, Varner, & Hommer, 2001) Ventral striatum and dopamine Based on animal studies, it has been proposed that d opaminergic (DA) signals from the ventral tegmental area and substantia nigra to the ventral striatum/nucleus accumbens aid in the prediction of rewards by increasing in response to received rewards and decreasing when an expected reward is not received. In this way, DA neurons are thought to create stimulus reward associations by

PAGE 17

17 ctations and actual outcomes, prediction error) than expected (Schultz, 2002; Schultz et al., 2000) Over repeated interactions a s stimulus reward associations are learned, the increased DA response in time so that it becomes associated with the conditioned st imulus rather than the reward, thereby aiding in the prediction of reward (Holroyd & Coles, 2002 ). Anterior cingulate cortex The dorsal anterior cingulate cortex (dACC) is believed to contribute to attention, motivation and cogniti ve control (Bokura, Yamaguchi, & Kobayashi, 2001; Botvinick, Cohen, & Carter, 2004; Kerns et al., 2004; Ridderinkhof, Nieuwenhuis, & Bashore, 2003) T he ACC is also involved in reward processing and error information (Amador, Schlag Rey, & Schlag, 2000; Shidara & Richmond, 2002) via its projection s fr om the mesencephalic DA system (Paus, 2001) It has been theorized that the commission of errors causes decreased dopaminergic activity that subsequently disinhibits neuronal activity in the ACC (Holroyd & Coles, 2002) ; however the precise contribution of the ACC to performance monitoring is still unclear. Recently, two positions on the topic have been outlined (Holroyd & Coles, 2008) : Those who believe th at the dACC has an evaluative function suggest that it monitors performance in order to deter mine success and detect errors (Dehaene, Kerszberg, & Changeux, 1998; Holroyd, Nieuwenhuis et al., 2004; Knutson, Westdorp, Kaiser, & Hommer, 2000) or conflict (Yeung, Botvinick, & Cohen, 2004) On the other hand, those who believe that the dACC is involved in response se lection hypothesize that performance monitoring occurs in other brain regions e.g., the basal ganglia (Holroyd, Nieuwenhuis, Yeung, & Cohen, 2003; Holroyd,

PAGE 18

18 Yeung, Coles, & Cohen, 2005) and the dACC plays a more dir ect role in the decision making process (Holroyd & Coles, 2002, 2008; Holroyd, Larsen, & Cohen, 2004) This more direct role involves the use of the reward prediction errors signaled by the midbrain dopami ne system for action selection (Gibson, 2006; Holroyd & Krigolson, 2007; Schultz, 2002; Schultz et al., 2000) implying the use of information from reward history to adjust future behavioral choices, rather than simply evalua ting them Consistent with the latter hypothesis, activations in the ACC have been seen following receipt of expected rewards (Knutson et al., 2 001) and dur ing the anticipation of reward (Knutson & Cooper, 2005) ; it has been proposed that the ACC codes for the probability of expected outcomes (Knutson & Coo per, 2005; Knutson, Taylor, Kaufman, Peterson, & Glover, 2005) The ventral portion (or affective subdivision) of the ACC is part of the limbic system making the ACC part of a circuit that regulates both cognitive and emotional processi ng, with these two types of inf ormation processed separately (Bush, Luu, & Posner, 2000) Because the ventral ACC is connected to the amygdala, nucleus accumbens, hypothalamus, anter ior insula, hippocampus and OFC, it is believed to be involved in the affective evaluation of the valence of feedback as positive or negative. Orbitofrontal cortex and amygdala The orbitofrontal cortex (OFC) and amygdala also contribute to the affect ive aspect of feedback processing. Specifically, the amygdala signa ls the level of arousal and reward intensity (Anderson et al., 2003) and the OFC which receives input from primary taste and olfactory cortices, determines the value of the reward and the strength of the behavioral response (Rolls, 2000)

PAGE 19

19 Electrophysiological Model of Reinforcement Learning Over the last four decades advances in electrophysiology have enabled new kinds of questions to be addressed regarding neural systems and processes that underlie cognition. Electroencephalography (EEG) record s volume conducted electrical activity of the brain using electrodes placed non invasi vely on the scalp (Davidson, Jackson, & Larson, 2000) Electrical activity recorded from EEG can be averaged in association with the presentation of specific events of interest. Initially, the event related response associated with the presentation of a stimulus is embedded in the ong oing EEG activity. A n event related potential (ERP) waveform associated with a specific stimulus is extracted by averaging multiple samples of the EEG that are time locked to repeated occurrences of the stimulus (Fabiani, Gratton, & Coles, 2000) Assuming that the underlying brain activity remai ns constant during the same conditions of an experiment, the benefit of averaging is that the ERPs should remain relatively consistent from trial to trial, while at the same time the ongoing random background EEG is averaged out of the resulting waveform (Otten & Rugg, 2005) ERPs are highly sensitive to changes in neural activity on the level of milliseconds with respect to temporal resolution (Fabiani et al., 2000) ERP waveforms usually consist of discrete voltage deflections that can either be positive or negative going. Specific n accordance with their polarity (positive or negative) and peak latency (in ms). Two examples of negative going ERP waveforms are the ERN (error related negativity) and the FRN (feedback related negativity).

PAGE 20

20 Measuring error detection and feedback proces sing using ERPs In the past ten to fifteen years, investigation of the mechanisms involved in performance monitoring has been informed by the study of ERP components sensitive to feedback FRN; (Holroyd & Krigolson, 2007; Miltner, Braun, & Coles, 1997) and error detection (i.e., the ERN; (Falkenstein, Hohnsbein, Hoormann, & Blanke, 1991; Falken stein, Hoormann, Christ, & Hohnsbein, 2000) These potentials are thought to be associated with reward prediction signals generated by dop amine signaling to the ACC. The FRN is maximal over frontal areas, has an average amplitude between 5 and 10 V, and a peak latency around 300 ms (Hajcak, Holroyd, Moser, & Simons, 2005; Nieuwenhuis, Holroyd, Mol, & Coles, 2004) F indings from ERP source modeling, functional magnetic resonance imaging, and single unit rec ording studies suggest t hat the FRN is generated in the medial frontal cortex, most likely the ACC (Amiez, Joseph, & Procyk, 2005; Brown & Braver, 2005; W. J. Gehring & Willoughby, 2002; Holroyd & Coles, 2002; Ridder inkhof et al., 2003; Shidara & Richmond, 2002) Although the FRN was initially used to study negative reinforcement, it is also reliabl y elicited by positive feedback (Hajcak et al., 2005; Holroyd & Coles, 2008; Oli veira, McDonald, & Goodman, 2007) when it appear s as a less negative going ERP deflection (compared to that elicited by negative feedback; see Figure 1 1 and Nieuwenhuis et al., 2004 for reviews) The FRN also appears to be modulated by feedback probab il ity such that lower probability events elicit higher amplitudes and higher probability events elicit lower amplitudes (Bellebaum & Daum, 2008; Hajcak, Moser, Holroyd, & Simons, 2007) The identification of the ACC as the source of the FRN suggests that it may be related to the error related n egativit y (ERN), a putative reflection of the evaluative

PAGE 21

21 process of performance monitoring that becomes more negative with awareness of errors. The ERN which is a response loc ked deflection, has a peak latency within 100 ms of an error response is distributed medial frontally, and also appears to be generated by the ACC (Holroyd, Larsen et al., 2004; Holroyd, Nieuwenhuis et al., 2004; Ye ung et al., 2004) The ERN was initially identified for its relationship to error detection, and it was suggested that the ERN was elicited when a mismatch occurred between the internal representation of a correct response and the actual response (Falkenstein et al., 1991; W.J. Gehring, Goss, Coles, Meyer, & Donchin, 1993) An alternate view developed positing that the ERN is a manifestation of conflict between competing responses (Yeung et al., 2004) that can also occur on corr ect trials ; on correct trials it is referred to as the correct response negativity or CRN (Ford, Roth, Menon, & Pfefferbaum, 1999) The CRN is less negative in amplitude than the ERN. It has been suggested that t he similar ity between the ERN and FRN indicates that they reflect activation of a reinforcement learning system that eval uates outcomes in order to direct future reward seeking behavior (Holroyd & Coles, 2002; Nieuwenhuis et al., 2004). This evaluation of outcomes as either better or worse than expected is coded as a reward prediction error (Schultz, 2002) Reinforceme n t learning theory of the ERN The reinforcement learning theory of the functional significance of the ERN has been gainin g increasing support ( Holroyd & Coles, 2002). This theory combines the role of the mesencephalic DA system in learning with the error d etection and feedback processing function of the ACC (see Nieuwenhuis et al., 2002 for review). Particularly relevant to a discussion of PD, this theory is also based on early research of the basal ganglia that indicates that the basal ganglia monitor and evaluate ongoing events and

PAGE 22

22 predict whether they will result in success or failure. When predictions are revised positively, the basal ganglia signal a phasic increase in dopaminergic activity, and when predictions are revised negatively, they signal a ph asic decrease in dopaminergic activity (see Schultz, 2002 for review). A decrease in DA causes a disinhibition of ACC neurons that generates a larger ERN, reflecting larger discrepancies between expectation and outcome (i.e., a negative prediction error), and increases in dopamine activity are associated with smaller ERNs (Holroyd & Coles, 2002). Prior to learning, only feedback (reflected in the FRN) provides information about the correctness of a response; w ith learning, however, the ERN should propagat e back in time and be elicited at the time of the response (Pietschmann, Simon, Endrass, & Kathmann, 2008) Thus the feedback processing system improve s future performance by using DA signals to learn the earliest predictors of reward or pun ishment Predictions from the reinforcement learning theory Importantly, the reinforcement learning theory is the only theory of the ERN that also makes predictions for changes in FRN amplitude (Nieuwenhuis et al., 2004). With learning, as individuals become better able to predict outcomes of their actions, the amplitude of the response ERN should increase (because errors become more whereas the F RN amplitude should decrease, as individuals rely less and less on feedback to shape their behavior (Holroyd & Coles, 2002). These changes in ERN and F RN amplitudes over time should also be related to behavioral changes (i.e., reduced errors over time).

PAGE 23

23 Executive Function and Reinforcement Learning in Aging Though it is a matter of much debate, there is some evi dence that the frontal regions are selecti vely and differentially affected by aging (West, 2000) Reduced blood flow (Rabbi tt et al., 2006) volume (Raz, Rodrigue, Head, Kennedy, & Acker, 2004) and met abolism (Greenwood, 2000) are reported in frontal areas in older adults M arkers of white matter integ rity are also reduced with maximal changes noted in anterior white matter; this l oss of white matter fibers may play an important role in cognitive dys (Shenkin et al., 2005) In fact such p hysiological changes in frontal regions have been shown to correlate with scores on tests of executive function in older adults (O'Sullivan, Barrick, Morris, Clark, & Markus, 2005; O'Sullivan et al., 2001) Executive dysfunction in older adults has been observed as impairment s in behavioral flexibility, or the ability to adapt behavior based o n rewards and punishments (Deakin, Aitken, Robbins, & Sahakian, 2004; MacPherson, Phillips, & Della Sala, 2002; Mell et al., 2005; Ridderinkhof, Span, & van der Molen, 2002) R eduction of behavioral flexibility in older age is likely caused by structural declines in the frontal regions just mentioned, as well as by chemical alterations of the reward system, including loss of serotonergic receptors (Wang et al., 1995) and dopaminergic midbrain neurons located in the ventral tegmenta l area (VTA), ventral striatum amygdala, and prefrontal cortex (Schultz et al., 2000) as well as the substa ntia nigra (Przuntek, Muller, & Riederer, 2004) to impairments in feedback processing, learning of stimulus reinforcement associations and r eward based decision making (Marschner et al., 2005)

PAGE 24

24 Of note, it appears that the integrity of the response to feedback in older adults may be altered based on the valence of the feedback For exampl e, Eppinger and colleagues (Eppinger, Kray, Mock, & Mecklinger, 2008) reported reduced amplitude ERPs reflecting negative feedback processing in older adults, and concl uded that older adults appear to be less affected by negative feedback than positive feedback as compared to young controls. In addition, a recent event related fMRI study measuring activation in mesolimbic regions during a monetary incentive delay task f ound evidence for intact activation of brain regions involved in gain anticipation, but a reduction in activation during loss anticipation in older adults compared to young adults (Samanez Larkin et al., 2007) Differential response to feedback based on valence may be due to disruption of distinct systems underlying processing of po sitive and negative feedback since i t has been sh own that p ositiv e feedback is processed in medial OFC and striatum (Liu et al., 2007; Nieuwenhuis, Slagter, von Geusau, Heslenfeld, & Holroyd, 2005; O'Doherty, Kringelbach, Rolls, Hornak, & Andrews, 2001) whereas n ega tive feedback is processed in lateral OFC ACC and insula (G. K. Frank et al., 2005; Liu et al., 2007; O'Doherty et al., 2001) ERP Studies of Reinforcement Learning in Older Adults As detailed above, behavioral investigations have concluded that older adults often demonstrate impairment s in aspects of reinforcement learning (Deakin et al., 2004; MacPherson et al., 2002; Mell et al., 2005; Ridderinkhof et al., 2002) ; howev er, ERP studies of error detection and feedback processing in healthy aging have not consistently supported these conclusions Based on the reinforcement learning theory, the amplitude of the ERN should increase with learning and the FRN should decrease with learning as participants rely

PAGE 25

25 less on feedback. Using a reinforcement learning paradigm that manipulated the validity of feedback, Nieuwenhuis and colleagues (2002) provided evidence for the theory when they found the expected increases in ERN amplit ude and decreases in FRN amplitude as learning progressed in young adults. They reported that older adults demonstrated reduced ERNs and FRNs when compared to healthy young controls, consistent with reports of a dysfunctional reinforcement learning system in older adults. Using a slightly different paradigm, Pietschmann and colleagues (2008) also found evidence for a dysfunctional reinforcement learning system in older adults when they reported no ERN amplitude increase over time in older adults despite sim ilar ERN amplitudes between groups at the begi nning of the learning process Importantly, they also found that FRN amplitudes decreased with learning in both age groups and were reduced in older relative to younger adults but only in response to negative feedback A ccording to Eppinger and colleagues (2008), t he Nieuwenhuis study was limited because it did not account for slowed processing speed in older adults, which placed greater time pressure on them, possibly impairing their ability to learn from fee dback Thus, i n order to account for age related changes in response speed Eppinger and colleagues (2008) used a probabilistic learning task in which validity of feedback was manipulated and response time was adjusted based o n performance differences. In contrast to the studies described above, they did not find evidence for an age related reduction of the ERN when controlling for performance differences between age groups, challenging the hypothesis that older adults are generall y impaired in error proce ssing. Like Pietschmann and colleagues (2008), Eppinger and colleagues (2008) reported a reduction of the non reward related FRN in the elderly. Eppinger and colleagues

PAGE 26

26 suggested that their study provided evidence for an age related asymmetry in the proces sing of feedback valence (with more attention paid to positive than negative feedback) and age related reductions in activity of structures involved in the processing of negative feedback (e.g., ACC and OFC). Interestingly, they explained these findings w cognitive control mechanisms to increase attention to positive information and de crease attention to negative information. R ecent fMRI findings (Samanez Larkin et al., 2007) using a gain and loss anticipation task seem to support this view by showing that older adults are less affected by potential losses than younger adults, whereas both age groups are equally affected by potential gains. Taken together, these studies suggest that older adults exhibit signs o f a dysfunctional reinforcement learning system characterized by valence dependent impairments in feedback processing; however, these studies do not provide consistent information on changes in the ERN with aging Executive Function and Reinforcement Learn Although PD is a movement disorder, it is important to recognize that it is not only physically debilitating ( causing slowed movements, postural instability, rigidity, muscle weakness and tremor), but also psychologically debilit ating ( causing increased prevalence of depression, anxiety, and apathy) (Fahn, 2003) and cognitively debilitating ( with increas ed prevalence of dementia and executive dysfunction). N europathological processes distinguishing cognitive and emotional dysfunction from motor dysf unction in PD have been clarified with the discovery of reciprocal and distinct circuits connecting cortical and subcortical areas (Alexander, DeLong, & Strick, 1986; Middleton & Strick, 2000) T hree non motor loop s originating in the dorsolateral prefrontal cortex, lateral

PAGE 27

27 orb itofrontal cortex, and anterior cingulate / medial orbitofrontal cortices have been identified Although d ysfunction in all of these areas is likely in PD (Zgaljardic et al., 2006) the loops originating in the lateral orbitofr ontal cortex and the ACC/medial orbitofrontal cortex are most relevant to the current project. The lateral loop is thought to be responsible for mediating reward processing and mood, whereas the medial loop is thought to be responsible for conflict monitor ing (Zgaljardic, Borod, Foldi, & Mattis, 2003) T here appears to be increasing evidence that the medial loop (including t he ACC) is also involved in emotional processing and the processing of both negative and positive feedback (i.e., rewards). I dentif ying relationships between the debilitating cognitive and emotional problem s associated with these circuits may be critical f or the development of appropriate treatments (McDonald, Richard, & DeLong, 2003) PD patients often demonstrate declines in executive functions similar to those seen in normal aging, such as inhibition, planning (Owen, 2004) sequencing and set shifting (Gotham, Brown, & Ma rsden, 1988; Owen, Iddon, Hodges, Summers, & Robbins, 1997) and response monitoring (Cooper, Sagar, Tidswell, & Jordan, 1994) These various aspects of executive dysfunction may contribute to difficulties initiating goal directed behavior and impairment s in performance monitoring. The mechanism responsible for these cognitive deficits in PD is still unclear but likely include s structures such as the striatum, areas of the frontal cortex, or connections between the two. Functional neuroimaging studies examining cognitiv e deficits in PD point to disruption of the major dopaminergic pathways including the nigrostriatal (involved in the production of movement), mesolimbic (involved in reward), and mesocortical (involved in motivation and emotion) pathways (Cools, Stefanova, Barker, Robbins, & Owen, 2002; Dagher,

PAGE 28

28 Owen, Boecker, & Brooks, 2001; Goldman Rakic, Bates, & Chafee, 1992; Mattay et al., 2002) Due to the disruption of the mesolimbic dopaminergic pathway, it is not surprising t hat patients with PD demonstrate executive dysfunction in the area of reward processing (i.e., reinforcement learning) ; however, the precise nature of these impairment s is still unclear. As in older adults, patients with PD are more impaired than healthy young adults at learning stimulus reward associations (M. J. Frank, Seeberger, & O'Reilly R, 2004; Marschner et al., 2005) Non demented, non depressed and medicated PD patients have also been found to make disadvan tageous choices on a test used to measure the ability to adjust performance and learn stimulus response associations based on monetary feedback (Iowa Gambling Task; (Kobayakawa, Koyama, Mimura, & Kawamura, 2008; Pago nabarraga et al., 2007) Although this suggests a failure to appropriately learn from feedback, it is not known whether this impairment is due to dysfunction of the mesolimbic reward prediction system, which involves accurate detection of errors, or to dy sfunction in processing the rewards/feedback themselves (Schott et al., 2007) D ysfunction in either of these areas is possible in PD since ventral striata l areas are important for the prediction and anticipation of reward, and ventromedial frontal cortex is important for the processing of rewards; either (or both) of these regions may be dysfunctional in PD (Knutson et al., 2001) As described above, these two aspects of performance monitoring are thought to be dissociable using EEG. In light of the fact that feedback processing systems depend on d opaminergic transmission, the medication status of patients with PD is important to consider during

PAGE 29

29 assessment of feedback based learning. Studies examining dopaminergic medication effects on feedback processing are contradictory or counterintuitive, with some studies reporting impairments in feedback based learning when patients are on medication versus off medication (Cools, Altamirano, & D'Esposito, 2006; Czernecki et al., 2002; Swainson et al., 2006) and others reporting contrasting results (M. J. Frank et al., 2004; Shohamy, Myers, Grossman, Sage, & Gluck, 2005; Swainson et al., 2006) T hese differences may be related to fa ctors such as demographic variables, disease vari ables or task difficulty or task demands (Cools, Barker, Sahakian, & Robbins, 2001; Mattay et al., 2002; Swainson et al., 2006) Alternatively it has been hypothesized that conflicting results demonstrated by med icated and unmedicated patients on feedback based tasks are due to differences in the processing of positive versus negative feedback (Cools et al., 2006; M. J. Frank et al., 2004) but a consistent pattern of resul ts has not been found. Some studies have reported that patients ON medication demonstrate impairment s on feedback based tasks when they are required to learn from negative feedback or outcomes; in contrast, when learning based on unexpected rewards or posi tive outcomes, patients ON medications perform as well as patients OFF medications and healthy older adults (Cools et al., 2006; M. J. Frank et al., 2004) Other investigators have reported that patients OF F medicat ion demonstrate impairment s when learning from positi ve feedback (Schott et al., 2007) ERP and fMRI se Some EEG studies suggest that patients with PD demonstrate dysfunction in the error detection system. As in healthy aging, examinations of the ERN in patients with PD have reported conflicting results: Some studies of medicated and unmedicated PD pa tients have reported reduced amplitude ERNs when compared to older adult controls

PAGE 30

30 (Falkenstein et al., 2001; Stemmer, Segalowitz, Dywan, Panisset, & Melmed, 2007) and others have found no difference between medicate d patients (tested off medication) and older adult controls (Holroyd, Praamstra, Plat, & Coles, 2002) Recently, reduced ERN amplitudes were observed in both on and off medication conditions in the same group of patients when compare d to older adult controls (Willemssen, Muller, Schwarz, Hohnsbein, & Falkenstein, 2008) Consistent with reduced ERN amplitudes in patients with PD another EEG study found amplitude reductions of a brain potential reflectin g reward anticipation (i.e., the stimulus preceding negativity) in PD patients when compared to older adult controls (Mattox, Valle Inclan, & Hackley, 2006) Similarly, a recent fMRI study showed that, w hereas young adults exhibit ed midbrain and ventral striatal activation during rewa rd prediction and no mesolimbic response to the reward itself, unmedicated patients with Parkinson disease demonstrate d no mesolimbic activity during reward prediction, but rather during feedback processing (Schott et al., 2007). T his seems to suggest th at PD patients demonstrate impairment in learning the predictive value of the rewards despite intact ability to process rewards themselves S urprisingly, no studies to date have exploited the temporal sensitivity of electrophysiological reflections to di fferentiate whether impairment s occur in reward anticipation or in the ability to process feedback itself (i.e., by measuring the FRN ) to bolster these findings in PD. Because there are no published studies examining the FRN in PD, in order to identify reasonable predictions regarding the FRN in this population it might be useful to consider ERP evidence of error detection and reward processing impairment s in patients diagnosed with schizophrenia, another neurological population characterized

PAGE 31

31 by a dysfu nctional dopaminergic system. Like PD patients, p atients with schizophrenia demonstrate diminished ERN amplitude compared to healthy subjects in a variety of experimental tasks (Alain, McNeely, He, Christensen, & W est, 2002; Kopp & Rist, 1999; Mathalon et al., 2002; Morris, Heerey, Gold, & Holroyd, 2008; Morris, Yee, & Nuechterlein, 2006) In the only study examining FRN amplitude in patients with schizophrenia Morris and colleagues (Morris et al., 2008) reported reduced FRN amplitude during early trials of a probabilistic learning task in which feedback was necessary for accurate performance. As in studies of older adults, t his reduction was seen only in response to negativ e feedback. Due to disruption of mesocortical pathways involved in motivation and emotion, it is not surprising that mood symptoms are common among PD patients. In comparison to the estimated prevalence of depress ion in the general population (15% ) the prevalence of depression in PD is quite high, with approximately 40 50% of PD pat ients reporting these symptoms (Cummings, 1992; McDonald et al., 2003) Depressive symptom s reported in PD may be explained as a psychological reaction to stress and loss of function associated with the disease. On the other hand, these symptoms may be explained by physio logical outcomes resulting from degeneration of brain regions including t he substantia nigra and ventral tegmental area (McDonald et al., 2003) ; reduced metabolism in the caudate, orbitofrontal cortex, and medial frontal lobes (Mayberg et al., 1990; Ring et al., 1994) ; or cha nges in neurotransmitter levels such as serotonin (Maye ux, Stern, Cote, & Williams, 1984) In addition to depression, an estimated 15 42 % of PD patients report symptoms of ap athy (Cummings, 1997; Zgaljardic et al., 2003) An apathy syndrome has been

PAGE 32

32 defined as a pri mary lack of motivation exhibited in behavioral, co gnitive, and emotional domains (Marin, 1991) A ccording to this model the behavioral domain includes symptoms s uch as lack of effort, lack of productivity, and lack of initiative (or diminished goal directed behavior); the cognitive domain includes symptoms such as loss of interest in new experience s affect ive domain includes symptoms of flattened affect and lack of response t o positive or negative events. Historically, differentiation of depressive symptoms from apathy symptoms has been difficult; however, more recently, the importance of this distinctio n has become the focus of research. For example, it has been reported that up to 29% of PD patients meet criteria for apathy without significant depression (Kirsch Darrow, Fernandez, Marsiske, Okun, & Bowers, 2006 ; Starkstein et al., 1992 ) suggesting that different mechanisms may underlie apathy, (e.g., orbitomedial or ACC/VTA connections) (Tekin & Cummings, 2002) and depression (e.g., orbitofrontal/subcortical connections) (Cummings & Masterman, 1999) The identific ation of different mechanisms underlying these disorders may signal the need for different treatments. Unfortunately, there are currently no pharmacologic tr eatments for apathy. L ike executive dysfunction, however, apathy may respond to dopaminergic agen ts, implying a similar mechanism associated with the loss of DA (Pedersen et al.) Thus, hypothesized mechanisms of apathy in PD may be similar to the mechanisms of executive dysfunction, consistent with Interac tions Between Cognition and Emotion Importance of the ACC Indeed, studies have reported correlations between apathy and executive functioning in PD such that increasing apathy is related to worse executive functioning

PAGE 33

33 (Isella et al., 2002; Pluck & Brown 2002; Starkstein et al., 1992 ) The mechanism underlying this relationship could be dysfunction of the ACC. Lesions of the ACC have produced a host of symptoms, including apathy, dysregulation of autonomic funct ions, akinetic mutism and emotional instability (Bush et al., 2000) ; as described in the preceding review, impairments in performance monitoring have also been linked with dysfunction of the ACC. In addition, patients with disease who exhibit apathy are more likely to show damage to the medial frontal and ACC regions (Craig et al., 1996; Migneco et al., 2001; Ro sen et al., 2005) given the close connections between brain regions subserving emotional symptoms of apathy and regions subserving cognitive executive functions such as per formance monitoring and reinforcement learning, it seems plausible that apathy may be related to decreased sensitivity to reinforcement. In fact, a relationship has been found between negative symptom severity and reduced FRN amplitude (Morris et al., 2008) in a sample of patients with schizophrenia (a population also aff ected by dysregulated dopamine). In addition, reduced ventral striatal activation during reward anticipation was associated with an increase in se verity of the negative symptoms (e.g., apathy) in an unmedicated sample of patients with schizophrenia (Juckel et al., 2006) B ecause PD also negatively impacts thes e regions patients with Parkinson disease represent an ideal population for examining relationships between these cognitive and affective symptoms. Surprisingly, one previous study comparing non demented and non depressed patients with PD on and off dopa minergic medications found no relationship between mood and stimulus reward learning (using a task similar to the Wisconsin Card Sorting Task and a

PAGE 34

34 gambling task ) (Czernecki et al., 2002) These investig ators found that levodopa medication improved subjectively reported motivation (i.e., apathy), but increased perseverative errors, suggesting a positive influence of medication on the subjective evaluation of motivation, but negative effects on feedback se nsitivity. Unfortunately, the measures used in this study did not allow for a precise examination of component processes of stimulus reward learning, such as error detection and feedback processing, which could be assessed using EEG. The negative effects o f depression on cognition are well documented. Depressive symptoms are correlated with an increased risk of dementia (McDonald et al., 2003) and memory as measured by the Dementia Rating Scale has been found to be worse in depressed patients wi th PD than in non depressed patients with PD or healthy controls (Norman, Troster, Fields, & Brooks, 2002) There is a vast literature on the association between executive dys function and depressive symptoms in otherwise healthy individuals (e.g., (Veiel, 1997) A relationship between executive impairment (e.g., in problem sol ving, set shifting, phonemic fluency) and depression has also been reported in PD, but it is not clear if the deficits are specific to depression in PD, the disease itself, or simply an aspect of global cognitive deterioration, or overall clinical severity (Troster et al., 1995) Studies of the impact of depression on aspects of pe rformance moni toring such as error detection and feedback processing are generally consistent with theories that depression is accompanied by increased attention to errors (Davidson, 1998) Bas ed on reports of enhanced reactivity to errors evidenced by elevated ERN amplitude in patients with major depressive disord er, it has been suggested that these patients may

PAGE 35

35 have exaggerated e arly error detection processes (Chiu & Deldin, 2007) Depressed individuals are seen as being highly sensitive to negative feedback, demonstrated by inc reased activation in the anterior cingulate cortex in response to negative stimuli (Anand et al., 2005) In addition depressed and dysphoric individuals are less responsive to positive stimuli and rewards such as monetary incentives (Naranjo, Tremblay, & Busto, 20 01) Studies of the responsivity to negative feedback using EEG to examine the FRN are less clear, and report enhanced amplitudes in moderately depressed patie nts and attenuated amplitudes in severely depressed patients (Tucker, Luu, Frishkoff, Quiring, & Poulsen, 2003) Summary and Rationale for the Present Study Impairments in executive functioning and aspects of performance monitoring including error detection and feedback proce ssing have been reported in both healthy aging and Parkinson disease. These impairments likely result from varying degrees of decline in the integrity of frontal lobe structures including the ACC and OFC; dysfunctional connections between these frontal structures and subcortical structures (e.g., the striatum); and disruptions in dopaminergic transmission. Failure to appropriately adapt decision making in response to environmental feedback may result in an increase in negative outcomes associated with b ehavioral choices and a decrease in positive outcomes, which in turn may lead to increased negative emotion (e.g., depression) and amotivation (i.e., apathy). Alternatively, depression and apathy may exacerbate difficulties with learning from feedback, cr eating a cycle of maladaptive behavior choices. Although progress has been made with respect to understanding the brain structures and neurotransmitters involv ed in feedback related decision making,

PAGE 36

36 questions remain regarding the effects of aging and Park inson disease on feedback processing, and regarding relationships between cognitive and affective components of this process. Because evaluating component processes of feedback based decision making using traditional neurop sychological tests is difficult e vent related potentials may provide a more direct method of assessing the neural processes involved. Increased attention is being paid to the differential effects of valence on the processing of feedback in older adults, with converging evidence pointin g to specific impairments in the processing of negative feedback. Continued support for the finding of preserved positive feedback processing could have implications for cognitive rehabilitation. Moreover, despite strong support for modulation of the ERN a nd FRN components by affect (i.e., depression), only one study (Morris et al., 2008) has examined the relationship between negative symptoms (including apathy) and the FRN. Because PD patients exhibit relatively high rates of apathy, understanding the deg ree to which this motivational disorder is associated with ERP reflections of impairment s in error detection and decreased sensitivity to reinforcement may also be important for intervention planning Thi s research could demonstrate a relationship between affective processing and performance monitoring in PD, adding to our understanding of m odels of reinforcement learning and dopamine related aging processes in general. This dissertation will investigate the following specific aims and hypotheses : 1) Spe cific Aim 1: Provide support for the reinforcement learning theory by examining the effe cts of manipulation of feedback probability and v alence on ERP components during the course of the experiment Hypothesis 1: T he ERN will be of greater (more negative) amplitude than the CRN in all probability conditions and that t he greatest (most negative) ERN amplitude will be observed in the least ambiguous (i.e., most probable) condition reflecting greater awareness of errors in this condition

PAGE 37

37 Hypothesis 2 : ERN a mplitude should become more negative over time reflecting the idea that awareness of errors increases with learning ; this change w ill be most apparent in the least ambiguous condition Hypothesis 3 : FRN amplitude will be greater in response to negative fee dback than to positive feedback in all probab ility conditions; t he most negative FRN ampli tude will be observed in the most ambiguous condition reflecting the idea that reliance on feedback increases under conditions of greater ambiguity Hypothesis 4 : FRN amplitude will decrease (become less negative) over time reflecting the idea that reliance on feedback decreases with learning 2 ) Specific Aim 2: Examine the effects of aging and Parkinson disease on reinforcement learning (i.e., error detection and feedback processing) Hypothesis 5 : At the start of the experiment, older controls will demonstrate smaller ERN amplitud es than young er controls ; however, over subsequent trial blocks, ERN amplitudes of older controls will be similar to young er contr ols Hypothesis 6 : ERN amplitudes of older adults and patients with PD will not differ at the start of the experiment; however, over subsequent blocks, ERN amplitudes of patients with PD will be smaller than those of older controls. Hypothesis 7 : With re spect to the amplitude of the FRN following positive feedback, older adult controls will not differ from younger controls Hypothesis 8 : I n turn, the amplitude of the FRN following positive feedback for patients with PD will not differ from that of older c ontrols Hypothesis 9 : With respect to the processing of negative feedback, older adult control s will exhibit reduced amplitude FRNs compared to younger controls across blocks. Hypothesis 10 : In turn, PD patients will exhibit reduced amplitude FRNs compa red to older adult control s 3) Specific Aim 3 : Exploratory analyses will investigate the relationships between feedback processing, emotional symptoms and other aspects of execu tive dysfunction Hypothesis 11 : Participants from all groups will demonst rate relationships between measures of executive functioning and ERP reflections of negative feedback processing after controlling for positive feedback processing In addition, feedback processing will be associated with emotional symptoms particularly a pathy

PAGE 38

38 General Methods Because the testing procedure and ERP methods were identical across the studies, the task stimuli and procedures, as well as the ERP acquisition, reduction, and analysis methods are described below. ERP Task Stimuli and Procedures T ask stimuli Stimuli and procedures for the task were modified from Eppinger et al (2008 ). Stimuli consisted of 18 black and white images of objects taken from the Snodgrass and Vanderwart picture database (Snodgrass & Vanderwart, 1980) Each stimulus belonged to one of three learning conditions in which validity of feedback was ondition (i.e., feedback i s always valid), stimulus A was ma pped to the right r esponse key and stimulus B to the left response key. Thus, if participants pressed the right button in response to stimulus A they always received positiv e feedback; if they responded to stimulus A with a left button press, they always received negative feedback (an d the reverse for stimulus B). Two different stimuli (C and D) 80% validity cond ition. If participants responded to C with a left button press, they received positive feedback 80% of the time and negative fe edback 20% of the time I f they responded with a right button press, they receive d negative feedb ack 80% of the time and positive feedback 20% of the time (and vice versa for Stimulus D). I were presented in mixed order (i.e., not in blocks) Prior to starting the experiment, participants completed one practice block of als and one practice block of 70 trials from all three validity

PAGE 39

39 conditions in order to familiarize themselves with the experiment. These practice stimuli were different from those used in the experimental blocks. There were three blocks in the experiment ; each block used a new set o f stimuli from the six s timulus categories. In each experimental block each of the six stimuli was presented 40 times in random o rder. Thus, e ach participant completed 240 trials per block, for a total of 720 trials Partic ipants were instructed to press one of two keys on a keyboard as quickly as possible in response to each stimulus and to determine the correct stimulus response associations using trial and error based on feedback that they were given. Positive feedback wa When a participant responded too slowly, they saw the words too slow To increase motivation and interest in the task, they were informed that they could win up to ten extra dollars for their participation, depending on their performance Following completion of the ERP task, participants completed a post task questi onnaire assessing their level of engagement in the task and t heir perceived emotional reactivity to the task. Trial procedure At the beginning of each tria l, a fixation cross appeared for 500 ms, followed by a stimulus picture for 500 ms. In an effort to ensure an equal number of trials from which to learn, we adj usted for age and disease related slowing by adapting the response deadline in 100 ms steps in a range of 600 1000 ms depending on the proportion of time out trials relative to performed trials (s ee Eppinger et al., 2008 ). Each participant began with a response deadline of 7 00 ms. Afte r the first trial, an algorithm monitored the proportion of time out trials (number of time out trials relative to the trials performed). If the proportion of time out trials was smaller than 2%, a response deadline

PAGE 40

40 of 600 ms was applied. For every time out rate increase of 2%, the response deadline was increased by 100 ms, with a maximum deadline of 1000 ms (for over 8% of time out trials). This was done in order to make sure that all subjects produced a similar proportion of time out trials After responding to the stimulus picture, a blank screen appeared for 500 ms followed by the feedb ack stimulus for 500 ms. Then participants began the next trial. Neuropsychological Measures Participants complete d a short batter y of n europsychological tests focused primarily on the assessment of executive functions. The Mini Mental State Exam (MMSE; (Folstein, Folstein, & McHugh, 1975) and the Dementia Rating Scale (DRS; (Mattis, 1988) were used to screen for dementia and estimate overall cognitive functioning. All participants completed the MMSE, but only older adults (controls and patients with PD) completed the DRS. Full scale intelligence quot ients (FSIQs) were estimated using the National Adult Reading Test. Participants also completed Trail Making Tests A and B (Reitan & Wolfson, 1995) the Digit S pan subtest from the Wechsler Adult Intelligence Scale Third Edition (WAIS III; (Wechsler, 1997) the Controlled Oral Word Association Test (COWA; (Benton & Hamsher, 1989) the Str oop Color and Word Test (Golden, 1978) and the Wisconsin Card Sorting Test (Heaton, Chelu ne, Talley, Kay, & Curtiss, 1993) Each of these tests measures a slightly different form of cognitive/ executive functioning. Trails A and B provide a measure of attention and cognitive flexibility, Digit Span measures attention and working memory, the Stroop Color and Word Test measures inhibition, selective attention, and response conflict, and the Wisconsin Card Sorting Test measures abstract problem solving and set shifting using examiner feedback (Lezak, Howieson, & Loring, 2004) COWA is a

PAGE 41

41 measure of speeded verbal output, an ability that has been shown to be impaired in PD and is related to ACC functioning (Abrahams et al., 2003) Emotional Measures Participants also complete d self report questionnaires assessing emotional functioning. Depression was assessed using both the Beck Depression Inventory Second Edition (BDI II; (Beck, 1996) and the Geriatric Depression Scale (GDS; (Yesavage et al., 1983) The GDS was used in addition to the BDI II because it accounts for somatic symptoms often experienced by older adults that may artificially inflate the likeliho od of depression diagnosis (Blazer, 2002) A modifie d version of the Apathy Evaluation Scale (Starkstein et al., 1992) as well as the Lille Apathy Rating Scale (LARS; (Sockeel et al., 2006) were used to assess apathy symptoms The State Trait Anxiety I nventory (STAI; (Speilberger, Gorusch, Lushene, Vagg, & Jacobs, 1983) manifest ed in temporary states of distress and more lo ng term personality traits. EEG Data Acquisition and Reduction EEG was recorded from 64 scalp sites using a 64 channel geodesic sensor net and Electrical Geodesics Incorporated (EGI ; Eugene, Oregon, USA ) amplifier system ( amplification 20K, nominal bandpa ss .10 100Hz). EEG was referenced to Cz and digitized continuously at 250 Hz with a 16 bi t analog to digital converter. Electrode placements enable d recording vertical and horizontal eye movements reflected in electro oculographic (EOG ) activity. A ri ght posterior electrode approximately two inches behind the right mastoid serve d as common ground. Electrode impedance was generally maintained manufacturer.

PAGE 42

42 EEG data were analyzed using Brain Electric Source Analysis (BESA) software (MEGIS Software, GmbH, Grfelfing, Germany). Eyeblink artifacts were identified with a template based method (Ille, Berg, & Scherg, 200 2) and corrected using the adaptive artifact correction (Ille et al., 2002) Continuous EEG was then segmented into condition related epochs. Using the BESA artifact scan tool, single trial epochs were discarded using threshold criteria that maximized the number of trials accept ed from each individual. The average voltage threshold that was used for excluding trials was 121.7 150) Point to point transitions were not allowed to Single trial EEG was then digitally re referenced to an average reference (Bertrand, Perrin, & Pernier, 1985) Prior to analyses, EEG was digitally low pass filtered at 30 Hz (zero phase). The individual subject event related potentia ls (ERPs) were extracted and av eraged from the ongoing EEG recording in discrete temporal windows that coincided with response and feedback onset to obtain respon se and feedback locked ERPs respectively. Individual subject response locked averages were d erived separately for correct (CRN) and incorrect (ERN) trials for the 100%, 80% and 60% conditions spanning 200ms prior to and 500ms following response and baseline corrected using the 200ms pre response window. To ensure accurate characterization of ER N/CRN amplitude and prevent spurious findings that might result from potential groupwise latency differences, the response locked components were measured as the mean amplitude within a 60 ms time window centered on the peak of the ERN/CRN at electrode FCz in each group. ERP trials conta ining errors of omission were excluded from averages.

PAGE 43

43 Individual subject feedback locked ERPs were derived separately for reward and non reward trials for the 100%, 80% and 60% feedback blocks spanning 100 ms before and 80 0 ms following feedback and were baseline corrected using the 100 ms pre feedback stimulus window. In light of previous f indings that measurement of the FRN can be confounded by potential overlap with other components (e.g., P300; (Holroyd, Larsen et al., 2004; Holroyd et al., 2003) initial quantification of the FR N will be completed by calculating difference waves subtracting the mean amplitude of the ERP associated with reward feedback from the ERP associated with non reward feedback. probability conditions The feedback locked components were measured as the mean amplitude within a 60 ms time window centered on the peak of the FRN at electrode F Cz in each group As mentioned above, the ERP components of interest ERN/CRN and FRN were quantified at electrode FCz. This location was chosen based on previous studies showing that they are maximal at this medio frontal site (Hajcak et al., 2005; Holroyd, Larsen et al., 2004; Holroyd et al., 2003; Larson, Kelly, Stigge Kaufman, Schmalfuss, & Perlstein, 2007) and because both components were largest at that site on examination of grand averaged w avef orms Data Analysis Independent samples t tests were performed to examine between group differences on neuropsychological data. The accuracy data from the feedback based learning task was analyzed by averaging mean accuracy rates individually for each subject and validity c 0 trials), refle cting the three learning blocks Trials for which no response was given were removed before calculating

PAGE 44

44 accuracy. The mean accuracy rates (% co rrect) as well as ERP component mean amp litudes, were analyzed using separate repeated measures analyse s of variance (ANOVA s ). ANOVAs included the between subjects factor of group (younger vs older ; older vs PD ) and within subject factors of validity (100%, 80%, 60% ), bin (1, 2, 3) and valen ce (positive or negative) P artial eta 2 i s reported as a measure of effect size. In accordance with a priori hypotheses, planned contrasts with polynomial comparisons were used to decompose interaction effects. In addition to planned contrasts, follow up post hoc comparisons were made using Bonferroni corrections for multiple comparisons. The Greenhouse Geisser epsilon adjustment (Greenhouse & Geisser, 1959) was applied for ANOVAs with more than two l evels of a within subject factor to correct for possible violations of sphericity ; corrected p values are reported where the assumption of sphericity is violated. Pearson product correlation coefficients were used to examine the relationships between diff erence wave amplitudes and neuropsychological data of interest.

PAGE 45

45 Figure 1 1 Typical example of event related brain potentials associated with negative and positive feedback recorded from electrode FCz. Note that here the FRN Adapted from Nieuwenhuis et al. ( 2002).

PAGE 46

46 CHAPTER 2 EXPERIMENT 1: THE EFFECT OF AGING ON ERROR DETECTION AND FEEDBACK PROCESSING Overview and Predictions Experiment 1 was conducted in order to replicate the results reported by Eppinger and colleagues ( 2008 ) using a modified version of their task. More specifically, this experiment examined whether electrophysiological correlat es of error detection ( as reflected in the fronto central ERN ) and feedback processing ( as refl ected in the fronto central FRN) reflect changes in reinforcement learning abilities and differential sensitivity to feedback valence in younger versus ol der adults by using a probabilisti c learning task in which feedback validity was mani pulated In addition, exploratory analyses exami ned relationships between ERP amplitudes, neuropsychological test results, and self reported emot ional symptoms For the young adult group, in accordance with the predictions from the reinforcement learning theory of the ERN/FRN (Holroyd & Coles, 2002; Nieuwenhuis et al., 2004) it was expected that the amplitude of the ERN would be larger than the CRN across learning conditions and that the amplitude of the ERN would increase over time th learning. The increase was expected to be most apparent in the 100% condition, followed by the 80% condition. Thus the amplitude of the incorrect minus correct difference wave for the ERN should be greatest the more valid the feedback. Similar ly, t he amplitude of the FRN for negative fe edback (i.e., non reward) was expected to be larger than the amplitude for positive feedback (i.e., reward) but the amplitude of the FRN should decrease over time as participants rely less and less on feedback. T he amplitude of the incorrect minus correct difference wave for the FRN should increase the more invalid the feedback in both groups

PAGE 47

47 In accordance with previous results in an older adult sample (Eppinger et al., 2008) it was predicted that older adults would also demonstrate the largest ERN /CRN s in the 100% condition, followed by the 80% and 60% condi tions. Although larger incorrect minus correct difference waves were expected with inc reasing validity for the response related activity it was predicted that these increased amplitudes would be more pronounced in the younger than older participants O lder subjects sh ould demonstrate reductions of the ERN/CRN early in the course of the exp eriment; however, by the end of the experiment their amplitudes should not differ from younger adults, suggesting that older adults are not impaired in error detect ion ; rather, they are slower to learn stimulus response mappings The FRNs for both groups w ere expected to be maximal in the 60% condition, followed by the 80% and 100% conditions. A ge related reductions in the FRN were predicted for incorrect trials but not for correct trials, regardless of validity condition, suggesting that older adults rely more on positive than negative feedback valence during learning. Finally in exploratory analyses, participants were expected to show correlations betwee n FRN in correct minus correct difference waves and scores on cognitive measures thought to be dependent on functioning of frontal brain regions. In particular, relationships were expected between FRN difference waves and performance on neuropsychological tests of problem solving using feedback (WCST ) and working memory (Digits Backwards ). Methods Participan ts Forty three participants were recruited from the community for this study. Three younger and two older adults were excluded from data analysis due to technical difficulties during data acquisition. The final sample included n ineteen young

PAGE 48

48 participants (ages 19 35) and nineteen older participants (ages 56 76 ). Table 2 1 provides demographic and neuropsychological data for the participants. The two groups were matched for gender ( 2 = 1.31, p > .20) and handedness ( 2 = .36 p > .50). Although older su bjects had a higher mean level of education compared to those in the young er group, t (36) = 3.3, p = .002 the two groups were well matched with respect to premorbid estimated FS IQ, t (36) = .84, p > .40 All participants completed all cognitive measure s. All participants denied histories of learning disability, traumatic brain injury, and neurological disease Procedures After informed consent was obtained, participants began an experimental session which lasted approximately 3 hours. All individuals completed all tasks in one session. Subjects rec eived financial compensation ($2 0 + $10 bonus ) for their participation. Results Behavioral Data Reinforcement learning task performance Although an attempt was made t o equate the number of no response trial s and both groups had a relatively low percentage of such trials ( M = .01 SD = .01, for younger adults, M = .02 SD = .02 for older adults), this difference was significant such that older adults had a greater number of timed out trials in the 80% [ F (1, 36) = 6.3, p < .05] and 60% [ F (1,36) = 5.0, p < .05] validity conditions compared to the younger adults T here were no significant correlations between number of timed out trials and accuracy in any of the validity conditions. See Table 2 2 for a summary of median response times.

PAGE 49

49 Accuracy data for each group in the three validity conditions and three bins are presented in Table 2 3 and displayed in Figure 2 1. The accuracy data were analyzed using an ANOVA with a between subject factor of age group (you ng er old er ) and within subjects factor s of validity condition (100%, 80% and 6 0%) and bin (1, 2, 3) The ANOVA revealed a significant main effect of age group, F (1, 36) = 9.89, p < 0.01 reflecting increased accuracy in the younger compared to older adult s There was also a significant main effect of validity condition F ( 2, 72) = 162.5 4, p < .00 1, partial eta 2 = .82, and a significant interaction between age group and validity, F (2, 72) = 8.15, p < 0.01, partial eta 2 = .19. Contrasts for each level of the validity factor showed a higher accuracy for the 100% compared to the 80% validity condition and for the 80% compared to the 60% validity condition ( p s < .001 ). Separate ANOVAs for each of the validity conditions revealed a significa nt age difference only in the 100% validity condition, F (1, 36) = 15.14, p < .001, such that older adults also performed worse than younger adults in this condition. The difference between groups was marginally significant for the 80% condition F (1, 36) = 3.02, p = .09; older a dults tended to perform worse than younger adults in this condition. Examination of effects of learning over time revealed a significant main effect of bin F (2, 72) = 45.41 p < .0001, partial eta 2 = .56 Post hoc pairwise comparisons of accuracy during each bin revealed significantly better performance in the last bin compared to the first bin and in the second bin compared to the first bin ( p s < .0001); however, the improvement between the second and third bins was only marginally significant ( p = .06). T here were no significant bin x group [ F (2, 72) = .45, p = .64, partial eta 2 f = .11] or validity x bin x group [ F (4,

PAGE 50

50 144) = 1.63, p = .18, partial eta 2 f = .21] interactions. Sens itivity analysis of the within interactions (alpha = .05, power = .80, mean correlation between repeated measures = .55) revealed that this design should be able to detect effects up to f = .20. T here was a significant interaction between validity and bin F (4, 144 ) = 10.47 p < .0001, partial eta 2 = .23 In the 100% valid condition, both groups improved significantly from bin 1 to bin 2 ( p s < .0001) but not from bin 2 to bin 3 ( p s > .05) In the 80% condition, the young er group demonstrated significant improvements from bin 1 to bin 2 ( p < .05) and marginally significant improvements from bin 2 to bin 3 ( p = .09); likewise for the older group, significant improvements in accuracy could be seen from bin 1 to bin 2 ( p < .0 5), but not from bin 2 to bin 3. As expected, there were no significant learning related changes in the 60% condition for either group T hese results are generally consistent with those reported previously by Eppinger and colleagues ( 2008 ), with two im portant differences: 1) The current study observed a significant difference in accuracy between age groups at all time points (i.e., bins) in the 100% condition, despite a similar number of trials attempted in this condition and 2) this study observed the opposite effects of learning over time in the 80% condition such that older adults demonstrated similar accuracy to younger adults early in the experiment in this condition, but did not improve from the second to third bins resulting in worse performance in the third bin Neuropsychological and emotional functioning The younger and older groups performed comparably on the MMSE, NART FSIQ, Digit Span, and the Wisconsin Card Sorting Test, and endorsed similar levels of apathy and depression symptoms. Olde r adults out performed younger sub jects on the

PAGE 51

51 Boston Naming Test. Not surprisingly, younger adults out performed the older adults on timed tests requiring speeded responses, including phonemic and semantic verbal fluency, Trails A and B 1 and Stroop Color Word Naming (ps < .05). Although no individuals met diagnostic criteria for any psychiatric disorder, one member of the young er group obtained a score on the AES that was above the conventional clinical cutoff for apathy (14), one member of the older gro up obtained a score on the BDI II that was in the range for mild depression (>14) and the young subjects endorsed a greater level of trait anxiety (p < .05). Event Related Potential Data See Table 2 4 for the number of trials comprising the ERP waveforms in each condition for the two groups. In line with the accuracy data, t he two age groups differed in numbers of trials per waveform in the 100% condition such that the ERN and non reward FRN waveforms ( p < .0001) for older adults contained more trials tha n those of younger adults, and the CRN waveforms ( p < .01) for younger adults had more trials than those of older adults. No significant group differences were observed in the other probability conditions. Response locked ERPs (ERN /CRN ) In the first step, incorrect minus correct difference waves were calculated. T hese difference waves showed maximal negative amplitudes wit h a latency of approximately 60 ms in both groups Spline interpolated voltage maps are shown in Figure 2 2 This roughly corresponds to the latencies of the peak amplitude seen in the original ERN and CRN waveforms. Mean amplitudes of these difference waves were calculated usin g a 1 When Trails A was used as a covariate in order to control for the effects of generalized slowing, the difference between groups for Trails B was no longer significant ( p = .46).

PAGE 52

52 time window from 30 9 0 ms. A comparison of the mean amplitudes of the difference waves was conducted using a repeated measures ANOVA (2 group x 3 validity condition x 3 bin). There was a significant main effect of group F (1, 35) = 15.67 p < .001 reflecting greater amplitudes for the younger versus the older adults There was also a significant main effect of v alidity F (2, 70 ) = 9.77 p = .001, partial eta 2 = .22 and a significant interaction between age group and validity F (2, 70) = 6.50 p = .003 partial eta 2 = .16 Contrasts for each of the validity conditions showed that the mean amplitude of the differenc e wave was greater for the 100% condition compared to the 80% and 60% conditions ( p s < .05 ); however, we did not detect a significant difference in amplitud e in the 80% condition compared to the 60% condition ( p = .19 ). T he younger group exhibited greater amplitudes of the difference waves than older adults in the 100% and 80% conditions ( p s < .01 ) but not the 60% condition Analyses of each group separately revealed that the younger group showed the expected declines in amplitude with decreasing probabili ty ( p s < .05) whereas no significant differences in amplitude between conditions were found in the older group Results of the repeated measures a nalysis of amplitudes of the difference waves also revealed significant learning related effects across bins (see Figure 2 3) There was a main effect of bin F ( 2, 70) = 3.86 p < .05 partial eta 2 = .10 such that the amplitude of the difference wave i ncreased significantly from bin 1 to bin 3 ( p < .05), bu t not from bin 1 to bin 2 or from bin 2 to bin 3 Ther e were also significant interactions between group and bin F (2, 70) = 4.05 p < .05 partial eta 2 = .10 validity and bin F (4 140 ) = 3.02 p < .05 partial eta 2 = .08 and group, validity, and bin F (4, 140) = 3.45 p = .01 partial eta 2 = .09 Separate a nalyses for the factor validity showed that for the

PAGE 53

53 younger adults, the amplitude of the difference wave in the 100% condition increased from bin 2 to bin 3 ( p < .05) and bin 1 to bin 3 ( p = .002 ), but not from bin 1 to bin 2 ( p = .08 ). Y ounger adults fail ed to demonstrate significant increases in amplitude across bins in the 80% or 60% condition s Interestingly although the older adults demonstrated learning effects across bins in terms of increased accuracy, this was not reflected in increased amp litudes of the difference waves in any validity condition (all p s > .25 ). Follow up ANOVAs of the amplitude of the difference wave in each validity condition and each bin showed that the amplitude of the difference wave was significantly more negative for the yo unger adults compared to the older adults in bins 2 and 3 for the 100% condition ( p p < .05). There were also trends for the younger adults to exhibit more negative amplitudes than older adults in bins 1 ( p = .0 6) and 3 ( p < .06) for the 80% condition. As expected, there were no significant differences for the 60% condition. In the second step, the response locked components were examined separately in order to examine the effects of age and validity condition on changes in mean amplitude by response type (correct or incorrect). The response locked components were measured as the mean amplitude within a 60 ms time window centered on the peak of the ERN at electrode FCz. A 2 (age group) x 2 (respons e type) x 3 (va lidity) x 3 (bin) repeated measures ANOVA was conducted. One younger adult was excluded from the ERN/CRN analysis because he did not commit any errors in the third bin for the 100% valid condition. Response locked ERP waveforms from the probabilistic learn ing task can be seen in Figure 2 4 Mean ERP amplitude data are presented in Tables 2 5 and 2 6. In line with Eppinger et al (2008) we did not find a main effect of group [ F (1, 36) =

PAGE 54

54 2.67, p = .11, partial eta 2 f = .2 7 ] Sensitivity analysis indicated that this design should be able to detect effects up to f = .34 (alpha = .05, power = .80, mean correlation between repeated measures = .28). As expected there was a significant main effect of response type F (1,35) = 17 .54 p = .001, such that the ERN was more negative than the CRN This main effect was qualified by interactions between group and response type F (1, 35) = 19.77 p < .001, partial eta 2 = .36 validity and response type F (2, 70 ) = 14.75 p < .0 0 1, partial eta 2 = .2 9 and group, validity, and response type F (2, 70) = 7.57 p = 0 01 partial eta 2 = .17 The validity x response type interaction reflected linear relationships such that the mean amplitude of the E RN was largest in the 10 0% condi tion, followed by the 80% and 6 0% conditions; in contrast the mean amplitude of the C RN was largest in the 6 0% condition fo llowed by the 80% and 10 0% conditions. These linear relationships were observed in the younger adult group, but not the older adult group. There wer e no significant quadratic relationships Follow up ANOVAs of the group x validity x response type interaction revealed significant differences such that the mean amplitude of the CRN was smaller for the younger adults compared to the older adults in the 100% and 60% conditions ( p ERN in any condition. S ee Table 2 7 for a summary of the ANOVA results W h e n examining effects of learning over time, we did not find a main e ffect of bin [ F (2, 70) = 1.66, p = .20, partial eta 2 f = .22]. Sensitivity analysis indicated that this design should be able to detect significant effects up to f = .19 (alpha = .05, power = .80, mean correlation betw een repeated measures = .59). W e did find a significant bin x

PAGE 55

55 group interaction F (2, 70) = 9.70, p < .001 partial eta 2 = .22 and a significant bin x response type interaction F (2, 70) = 4.77 p < .05 partial eta 2 = .12 The re was also a significant gro up x response type x bin interaction F (2, 70) = 6.00 p < 01, partial eta 2 = .15 a significant grou p x validity x bin interaction F (4, 140) = 2.96, p < .05, partial eta 2 = .08, a significant response type x validity x bin interaction F (4, 140 ) = 4.36, p < .01, partial eta 2 = .11 and a sign ificant group x response type x validity x bin interaction F (4, 140) = 3.63, p < .01 partial eta 2 = .09 Follow up analyses for the group x response type x validity x bin interaction revealed that during bin 2, the mea n amplitudes of the CRN were smaller for the younger adults than the older adults in all conditions. During bin 3, they were smaller in the 100% ( p < .001) and 60% ( p < .05) condition s only In contrast, during bin 2, the mean amplitude of the ERN was larg er for the older adults compared to the younger adults for the 80% and 60% conditions ( p s < .05) During bin 3, the mean amplitude of the ERN was larger for the younger adults compared to the older adults in the 100% condition only ( p < .01). Within group analyses showed that for the younger adults, the mean amplitude of the CRN became significantly smaller from bin 1 to bin 2 ( p = .002) and bin 1 to bin 3 ( p < .001) in the 100% condition. In the 80% condition, the mean amplitude of the CRN became smaller from bin 1 to bin 2 ( p < .001) and from bin 1 to bin 3 ( p < .05), but became larger from bin 2 to bin 3 ( p = .002). F or the older adults, the mean amplitude of the CRN be came significantly smaller from bin 2 to bin 3 in the 100% condition ( p < .01). F or t he younger adults, t he mean amplitude of the ERN significan tly increased from bin 2 to bin 3 and from bin 1 to bin 3 in the 100% condition ( < .05 ; see Figure 2 5 ). In the 80% condition, the mean amplitude of the ERN decreased from bin 1 to bin 2

PAGE 56

56 ( p = .003) and increased from bin 2 to bin 3 ( p < .05). The older adults did not show any learning related changes in the mean amplitude of the ERN (see Figure 2 6). Feedback locked ERPs (FRN) In the first step, non reward minus reward difference waves were ca lculated. These difference waves showed maximal negative amplitudes wit h a latency of approximately 280 ms for younger adults and 290 ms for older adults as shown in Figure 2 7 This roughly corresponds with latencies of the peak amplitu de seen in the ori ginal reward and non reward related FRN waveforms. Mean amplitudes of these difference waves were calculated usin g a time window from 250 310 ms for younger adults and 260 320 ms for older adults As with the error related activity, a comparison of the mean amplitudes of the difference waves for feedback related activity was conducted using a repeated measures ANOVA (2 group s x 3 validity condition s x 3 bin s ). In contrast to the results of Eppinger et al., 2007, a significant main effect of group was no t detected [ F (1, 36) = .001, p = .98, partial eta 2 f < .01 ] Sensitivity analysis suggested that this design should be able to detect significant effects up to f = .40 (alpha = .05, power = .80, mean correlation betwee n repeated measures = .61). There was a significant main effect of validity F (2, 72 ) = 11.6 0, p < .001, partial eta 2 = .24 and a significant interaction between age group and validity F (2, 72) = 7.85 p = .001, partial eta 2 = .18 Contrasts for each of t he validity conditions showed a significant linear trend such that the mean amplitude of the difference wave was greater (more negative) for the 6 0% condition compared to the 100% and 8 0% conditions ( p s < .05 ); in turn, the mean amplitude of the difference wave for the 80% condition was greater than that for the 10 0% condition ( p < .05 ). Consistent with Eppinger et al. (2008) separate

PAGE 57

57 within group analyses showed that these relationships were observed for the younger adults ( p s < .05 ), but not for older ad ults ( p s > .5 0) Analysis of amplitudes of the difference waves also revealed significant learning related effects across bins (see Figure 2 8) There was a main effect of bin F (2, 72) = 3.99 p < .05, partial eta 2 = .10 ; pairwise comparisons showed trend s toward decreased a mplitude of the difference wave from bin 1 to bin 2 ( p = .09) and from bin 1 to bin 3 ( p = .06 ) but n ot from bin 2 to bin 3 There was also a significant quadratic interaction between group and bin F (2, 72) = 5.08 p < .01, partial eta 2 = .12 and a marginally significant interaction between validity and bin F (4, 140 ) = 2.47 p = .07 partial eta 2 = .06 Separate within groups analyses showed that for the younger group, the amplitude of the difference wave decreased as expected from bin 1 to bin 2 in both the 100% and 80% conditions (ps < .05). For the older adult group, significant decreases in FRN amplitude over time could not be detected although there was a marginally significant decrease from bin 1 to bin 3 in the 100% condition ( p = .07). In the second step, the feedback locked components were examined separately in order to examine the effects of age on changes in mean amplitude by feedback type (reward or non reward). The feedback locked components were measured as the mean ampl itude within a 60 ms time window centered on the peak of the FRN at electrode FCz (238 ms in younger adults and 260 ms in older adults). A 2 (age group) x 2 ( feedback type) x 3 (validity) x 3 (bin) repeated measures ANOVA was conducted. Feedback locked ER P waveforms from the probabilistic learning task can be seen in Figure 2 9 Mean ERP amplitude data are presented in Tables 2 8 and 2 9 In contrast to Eppinger et al, we did not find a ma in effect of group [ F (1, 36) = .99, p = .33, partial

PAGE 58

58 eta 2 = .03, ob f = .17] or a significant interaction between group and feedback type [ F (1, 36) = .45, p = .51, partial eta 2 = .01, observed power = f = .10] Sensitivity analyses indicated that this design should be able to detec t significant between group effects up to f = .44 and a significant interaction up to f = .11 (alpha = .05, power = .80, mean correlation between repeated measures = .86). T here was a significant main effect of validity F (2, 72) = 3.60, p < .05 partial eta 2 = .09. Pairwise comparisons revealed a trend towards greater amplitudes for the 100% condition compared to the 6 0% condition ( p = .07). There was also a main effect of feedback type F (1, 36) = 8.89 p < .01 such that the FRN related to non reward wa s more negative than the FRN related to reward. These main effect s were qualified by interactions between validity and feedback type F (2, 72 ) = 6.52 p < .01 partial eta 2 = .15 and group, validity, and feedback type F (2, 72) = 5.33 p < .01, partial eta 2 = .13 Separate within group analyses showed that for the younger adults, the FRN to reward was larger in the 10 0% condition compared to the 80% condition ( p = .01) and in the 80% condition compared to the 60% condition ( p = .001). There were no significa nt differences in mean amplitude for the FRN to non reward in any condition. When comparing the mean amplitudes between valences and within condition, there was a trend for the FRN to non reward to be larger than the FRN to reward in the 80% condition ( p = .08). Similarly, in the 60% condition, the FRN to non reward was larger than the FRN to reward ( p = .001). In the older adult group, no significant differences could be found. With respect to effects of learning over time, we did not find a main effect of bin [ F (2, 72) = 1.03, p = .36, partial eta 2 f = .17] or a

PAGE 59

59 significant interaction between feedback type, validity, and bin ( see Table 2 10) Sensitivity analysis suggested that this design should be able to detect a si gnificant effect of bin up to f = .20 (alpha = .05, power = .80, mean correlation between repeated measures = .60). T here was a significant interaction of bin by feedback type F (2, 72) = 3.22 p < .05, partial eta 2 = .08, which was qualified by a signific ant group x feedback type x bin interaction F (2, 72 ) = 7.77 p = .001 partial eta 2 = 1 8 When the two groups were analyzed separately, the younger adults exhibited significantly larger mean amplitudes of the FRN to reward for the 100% and 80% conditions in bin 2 compared to bin 1 (ps < .01) and in bin 3 compared to bin 1 (ps < .05). For older adults, the ampli tude of the reward related FRN increased from bin 2 to bin 3 in the 100% condition ( p < .01). There were no significant changes over time for the me an amplitude of the FRN to non reward in any validity condition for either group (see Figure s 2 10 and 2 11 ) Follow up ANOVAs showed group differences such that the mean amplitude of the FRN to reward was larger for the younger adults compared to the o lder adults in the 100% condition during bin 2 (p < .05). The mean amplitude of the FRN to non reward was larger for the younger adults compared to the older adults in the 80% condition during bin 1 (p < .05). Relationship to performance on neuropsychologi cal tests Several demographic and neuropsychological variables corre lated with FRN differen ce wave amplitudes ( see Table 2 1 1 ) In the 100% valid condition, scores on the LARS were negatively correlated with the FRN difference wave in bin 2 only ( r = .351 p < .05 ) and BNT scores were negatively correlated with the FRN difference wave in bin 1 only ( r = .373, p < .05 ) In the 80% valid condition, phonemic fluency (i.e., COWA)

PAGE 60

60 performance was positively correlated with the FRN difference wave for bin 2 on ly ( r = .361, p < .05) and the Trails B raw score was negatively correlated with the difference wave in bin 2 ( r = .330, p < .05) In the 60% valid condition, the number of errors on Trails A was positively correlated with the difference wave in bin 1 ( r = .339, p < .05) Discussion This study had three primary goals The first was to verify predictions from the reinforcement learning theory by measuring the effects of feedback valence and probability on ERP reflections of error detection and f eedback proc essing over time; the second was to investigate how and whether aging affects reinforcement learning (i.e., error detection and feedback processing abilities ) believed to be mediated by the mesencephalic dopamine system ; and the third was to explore how ER P reflections of error detection and feedback processing are related to other aspects of cognitive functioning thought to be dependent on frontal brain regions Although some evidence has suggested that older adults exhibit impairment s in error detection t hat negatively impact lear ning (Nieuwenhuis et al., 2002) recent counter evidence has demon s trated that when older adult s a re given equivalent opportunity to learn from errors, r eflections of error detection a re not affected by aging; rather, older adults demonstrate valence specific differences in feedback processing such that they are less affected by negative than positi ve feedback (Eppinger et al., 2008), consistent with theory of aging older adults use emotion regulation and cognitive control to increase attention to positive information and decrease attention to negat ive information (Mather & Carstensen, 2005) The curren t study intended to replicate the latter findings using a modified version of a probabil istic learning paradigm in which positive and negative

PAGE 61

61 feedback were provided in three validity conditions (100%, 80%, and 60%) and an attempt was made to equate the n umber of trials contributing to learning in both group s Differences in reinforcement learning were measured as a function of changes in ERP amplitude as well as changes in performance (i.e ., accuracy) over time. In addition, exploratory analyses investig ated relationships between electrophysiological refle ctions of feedback processing and neuropsychological measures of theoretically related cognitive constructs. Specific Aim 1 : Support for the Reinforcement Learning Theory Behavioral d ata Although we wer e unsuccessful in our attempt to equate the number of trials retained between groups by using an adaptive response deadline in the 80% and 60% conditions, a similar number of trials was retained for the 100% condition. Despite an equivalent number of tria ls attempted in the 100% condition, the younger adults out performed the older adults; in addition, they demonstrated a trend toward better performance in the 80% condition, consistent with previous reports suggesting that older adults exhibit impairment s in reinforcement learning (Deakin et al., 2004; MacPherson et al., 2002; Mell et al., 2005; Ridderinkhof et al., 2002; Nieuwenhuis et al., 2002; Pietschmann et al., 2008). Despite the evidence for an impaired reinforcement learning system in older adults provided above, both groups demonstrated better performance in the 100% condition compared to the 80% condition and in the 80% condition compared to the 60% condition, suggesting that older adults benefit to some extent from increasing condition validity. As expected, both groups improved over time in the two learning conditions (i.e., 100% and 80%), but did not exhibit learning over time in the 60%

PAGE 62

62 condition. Importantly, in both groups, the significant changes in learning occurred from bin 1 to bin 2; le arning appeared to plateau from bin 2 to bin 3 such that no significa nt learning effects were found. Based on this trend, it is unclear whether older adults would have been able to improve to the level of younger adults if given more time, as was previousl y reported (Eppinger et al., 2008). Response locked ERPs (ERN/CRN) Consistent with predictions derived from the reinforcement learning theory (Eppinger et al., 2008; Holroyd & Coles, 2002; Nieuwenhui s et al., 2002), for the younger control group, the ERN was larger than the CRN in all validity condi tions. T he mea n amplitude of the ERN increased with increasing validity, such that t he largest amplitude was observed in the 100% valid condition reflecting the idea that awa reness of error commission is greate st under conditions in which correctness of response is the least ambiguous Support for learning related changes in the response related ERPs was most clear when examining the mean amplitude of the ERN CRN difference waves. Within the younger group, the amplitude of the difference wave increased over time in the least ambiguous learning condition, providing support for the idea that as internal representations of correct and incorrect responses become more fixed with learning, greater conflict occurs when erroneous responses are made Learning related effects were also found for young er adults in the 100% condition when looking at the mean amplitud e of the ERN and CRN separately such that the mean amplitude of the ERN increased over time and the mean amplitude of the CRN decreased over time Learning effects in the 80% condition were less clear, but essentially paralleled those from the 100% condition I t is possible that ERN amplitudes in the 80% condition were

PAGE 63

63 affected to some degree by the use of a variable response deadline. Younger adults may not have experience d necessary time pressure to elicit greater amplitude ERNs in the more ambiguous condition Overall t he s e results support predictions from the reinforcement learning theory, including re cent suggestions that correct response activity (i.e., positive prediction errors) in addition to error activity (i.e., negative prediction errors) contribute s to reinforcement learning (Eppinger et al., 2008 ; Pietschmann et al., 2008 ) Although conflic ting theories have been generated regarding the function of the CRN, these results suggest that the CRN is a correct response analogue to the ERN; in other correct respon CRN. Feedback locked ERPs ( FRN) Anal ysis of the feedback locked difference waves within the younger control group also provided some support for the reinforcemen t learning theo ry. As expected, t he mean amplitude of the non reward reward difference wave decreased with increasing feedback validity, reflecting the idea that reliance on feedback is greatest in the most ambiguous situations (Eppinger et al., 2008; Holroyd & Coles, 2002) As previously reported (Eppinger et al., 2008 ; Nieuwenhuis et al., 2002 ) older adults in the present study did not exhibit variations in amplitude based on feedback validity Learning related effects on th e difference waves appeared to be consistent with the theory; the mean amplitude of the difference wave decreased from bin 1 to bin 2 in both learning conditions providing some evidence that reliance on feedback decreases with learning (Holroyd & Coles, 2 002)

PAGE 64

64 Separate a nalyse s of the FRN to non reward and the FRN to reward revealed results that were somewhat con tra ry to expectations. Although the non reward related FRN was larger than the reward related FR N in the 80% and 60% conditions as expected, we we re unable to find differences in the amplitude of the non reward related FRN across conditions Although there were no predictions from the reinforcement learning theory with respect to changes in reward related ac tivity, the reward related FRN became larg er with increasing feedback validity such that it was most negative in the 100% condition, followe d by the 80% and 60% conditions, contrary to predictions for the non reward related activity. Similar results have been reported previously in an older adult group (Nieuwenhuis et al., 2002). Learning r elated effects observed in the reward related FRN for younger ad ults were also unexpected: the reward related FRN became more negative over time in both the 100% and 80% conditions, suggesting that reactivity to positive feedback did not decrease over time as hypothesized There were no learning related effects observed for the non reward related FRN in keeping with prior research (Eppinger et al., 2008) These results are interesting and may indicate preferentia l reliance on positive feedback over time even in healthy young adults, particularly in less ambiguous learning condi tions In summary the younger adult data support both condition related and learning related predictions from the reinforce ment learning theory with respect to the ERN/CRN Predictions related to feedback locked activity were generally supported when examining difference waves; however, separate analysis of reward related activity yielded unexpected results

PAGE 65

65 Specific Aim 2 : Effects of Aging on ERP Reflections of Reinforcement Learning Response locked ERPs (ERN/CRN) In consideration of predictions from the electrophysiological model of the reinforcement learning theory, we attempted to discern whether the behavioral impairment s observed in o lder adults were a result of dysfunction in the error detection phase of learning, the feedback processing phase, or both. Consistent with Eppinger and colleagues (2008), our younger control group exhi bited greater amplitude ERN CRN difference waves than those of o lder adults in both learning conditions providing initial support for the hypothesis that older adults exhibit impairment s in the error detection phase of learning In order to investigate whether these impairment s are valence specific, as repo rted previously (Eppinger et al., 2008; Pietschmann et al., 2008) we examined ERN and CRN amplitudes separately In line with these two previous reports, we found no difference between groups in the amplitude of th e ERN in any condition when all time points (i.e., bins across trials) were combined. These results conflict with several previous studies that reported smaller ERN amplitudes in older adults (Band & Kok, 2000; Falke nstein et al., 2001; Mathewson, Dywan, & Segalowitz, 2005; Nieuwenhuis et al., 2002; Themanson, Hillman, & Curtin, 2006) One salient difference between studies finding c omparable ERNs between groups and studies finding differences is that the former eit her reduced time pressure on older adults (Eppinger et al., 2008 ) or did not apply time pressure at all (Pietschmann et al., 2008 ) Although reduced time pressure was the intended aspect of the current study it is possible that relieving the time pressure on both groups resulted in less resp onse conflict in younger adults, artificially reducing their ERN amplitudes

PAGE 66

66 W ith regard to learning related changes of the ERN it was predicted that earlier in the experiment, older controls would demonstrate smaller ERN amplitudes than younger controls, but that these differences would not be observed later in the experiment, suggesting that older adults require more time to develop internal representations of appropriate responses. Instead, ERNs did not differ betwe en groups during bin 1, and the ERN of older adults was smaller than that of younger adults by the end of the experiment in the least ambiguous condition. Based on the re inforcement learning theory, it might be conclude d that older adults failed to create an internal representation of the appropriate response, or developed a weak representation resulting in uncertainty and reduced amplitude ERNs (Band & Kok, 2000) W e anticipated that the amplitude of the CRN would also be equivalent between groups (Falkenstein et al., 2001); however, the amplitude of the CRN was unexpectedly smaller for the younger adult s compared to the older adults in the 100% and 60 % conditions. Within the older adult group, the amplitude of the CRN decreased from bin 2 to bin 3 in the least ambiguous condition; sim ilarly, within the younger adult group, the amplitude of the CRN decreased over time in both learning conditions. Importantly, initial interpretations of the error ivity on correct trials, except that it was subtracted from incorrect trials to form difference waves as an internal control. Although this view is increasingly being challenged (Band & Kok, 2000; Bartholow et al., 2 005; Falkenstein et al., 2001; W. J. Gehring & Knight, 2000; Mathewson et al., 2005) the function of the CRN is still a matter of much debate. One theory is that the CRN increases when uncertainty about the correct response is high. It seems reasonable

PAGE 67

67 then, that the CRN should decrease when uncertainty is low (e.g., as learning progresses). This hypothesis fits well with the current study; however, another group who found significant reductions of the CRN with learning in a young control g roup (Pietsch mann et al. 2008) concluded that CRN reductions signified that performance monitoring becomes error specific with advanced learning. It has also been surmised evoked components in the response locked ERP (Coles, Scheffers, & Holroyd, 20 01) In general for the older participants, ERN/CRN mean amplitudes were not altered by validity condition and changed very little over the course of the entire experiment, consistent with previous research (Pietschmann et al., 2008). Feedback locked ERP s (FRN ) In contrast to predictions from the reinforcement learning theory, we were unable to find group differences in t he mean amplitudes of the FRN difference wav es in any condition. In line with our hypotheses when we compa red the mean amplitudes of t he reward related FRN between groups, collapsing across bins, th ere were no group differences; however, c ontrary to expectations, we also found little evidence for reduced non reward related FRNs in the olde r adults compared to younger adults It has previ ously been reported that, whereas older adults do not demonstrate a reduction in reward related FRN amplitude, they do exhibit a reduction in reactivity to negative feedback, contributing to the theory that older adults preferentially rely on positive feed back for learning (Eppinger et al., 2008; Pietschmann et al., 2008). Reasons for these disparate results may be related to differences in study design. For example, one study did not manipulate feedback validity ( Pietschmann et al., 2008) which may have r educed the need for reliance on feedback, resulting in reductions in the non reward

PAGE 68

68 related FRN (Pietschmann et al., 2008). This explanation seems unlikely, however, since reductions in the reward related FRN would also have been expected and were not foun d. It is also possible that high functioning in our group of older adults served as a protective factor and related changes seen in structures underlying the processing of negative feedback (to be discussed in the next chapter) Overa ll, our data suggest that the feedback processing system is relatively intact as compared to the error processing system in this group of older adults ; however, it should be noted that, as with the error related activity, older adults demonstrated very lit tle expected variability as a function of probability condition. We also failed to find evidence for learning related effects for the non reward related FRN in either group, consistent with previous reports (Eppinger et al 2008; Nieuwenhuis et al 2002 ). T he reward related FRN increased over time for younger adults as has been reported previously (Pietschmann et al ., 2008 ) Interestingly, for the older adults, the reward related FRN increased as the CRN decreased from bin 2 to bin 3 in the 100% conditio n only Taken together, the most striking pattern found in these data is that the younger adults exhibited at least some degree of condition valence and learning related modulations in electrophysiological reflections of error detection and feedback p rocessing consistent with the reinforcement learning theory but the older adults generally did not in keeping with their poorer behavioral performance Importantly, we found few er differences between groups with respect to feedback processing; however, older adults exhibited clear impairmen ts with respect to error and correct response detection with learning as compared to younger adults (see Figure 2 12) Interpreted

PAGE 69

69 within the context of the reinforcement learning theory, we could conclude dopaminergi c signaling is relatively intact early in the course of learning (i.e., when learning is dependent on feedback); however, there is a failure of the system to both appropriately modulate the signal based on probability of reinforcement propagate ck in time so that it becomes associated with the response, rather than the stimulus. Thus, an internalized representation of the response is not created, resulting in greater uncertainty regarding whether the outcome is better or worse than expected. On e notable exception to the lack of learning related change seen in the older data wa s the increase in reward related FRN amplitude over time and decrease in CRN amplitude at the end of the experiment in the least ambiguous condition. Perhaps this provides some support for the contention that the positive feedback processing system is relatively intact in older adults as compared to the negative feedback processing system. In any case, it is difficult to see how reports of decreased non reward rela ted FRNs in the context of similar reward related FRNs has contributed rather than an argument for dysregulated reactivity to punishment This is especially unclear in light of the fact that this positivity effect appears to be viewed as an adaptive cognitive control mechanism despite evidence for poorer performance on reinforcement learning tasks S pecific Aim 3: Relationship to Neuropsychological Test P erformance Feedback related activity corr elated with estimated IQ, verbal fluency, visuomotor sequencing and set switching, symptoms of apathy, and performance on a language measure. W e expected larger ( more negative ) values of the FRN difference waves to be correlated with higher scores on meas ures of executive function Unfortunately, several of the relationships were in the counter intuitive direction For example better

PAGE 70

70 verbal fluency and set switching performance s were related to smaller ( more positive ) amplit u des of the FRN difference wave as were higher IQ and lower levels of apathy. In contrast, better naming ability and fewer errors on an attention measure were related to more negative amplitudes. The reason for unexpected correlation patterns is not clear but it is possible that perfor mance on the probabilistic learning task did not generalize well to performance on other tests of frontal executive functioning. Like other ex ecutive functions feedback processing is likely dependent on extremely complex cognitive systems not encompasse d by these exploratory hypotheses.

PAGE 71

71 Table 2 1. Means and standard deviations ( SD ) of demographic and neuropsychological data for younger and older participants. Young er (n = 19) Old er (n = 19) Mean ( SD ) Mean ( SD ) p Demographics Age (years) 23.9 (5.2) 66.9 (6.7) < .001 Education (years) 14.4 (1.8) 16.5 (2.1) .002 Female (%) 15.8 -31.6 -ns Right Handed (%) 89.5 -94.7 -ns Cognitive Functioning MMSE 28.2 (1.5) 28.3 (1 .7) ns NART FSIQ 112.9 (6.5) 114.8 (7.4) ns Boston Naming Test 55.8 (2.2) 56.8 (5.0) < .01 COWA (FAS) 47.7 (11.7) 39.0 (8.9) < .05 Semantic Fluency (Animals) 25.5 (4.0) 20.4 (3.7) < .001 Digit Sp an Forward 12.4 (2.0) 11.0 (2.6) ns Digit Span Backward 7.9 (2.6) 6.7 (2.1) ns Trails A (sec) 18.2 (5.7) 31.8 (9.1) < .001 Trails B (sec) 41.5 (14.2) 71.0 (26.8) < .001 Stroop Word Reading 113 .8 (19.1) 108.1 (17.1) ns Stroop Color Naming 80.6 (16.5) 74.8 (12.4) ns Stroop Color Word Naming 53.5 (12.8) 45.5 (11.0) < .05 Stroop Interference 5.2 (8.9) 2.5 (8.9) ns WCST Categories Completed 5.2 (1.5) 5.6 (1.0) ns WCST Perseverative Errors 12.3 (10.4) 12.5 (9.6) ns WCST Failure to Maintain Set .5 (1.1) .6 (.7) ns Emotional Functioning BDI II 5.2 (3.9) 3.6 (4.9) ns AES 6.7 (4.8) 5.7 (3.0) ns LARS 26.6 (6.0) 24.8 (3.6) ns STAI State 29.4 (9.0) 26.9 (9.4) ns STAI Trait 33.9 (9.1) 27.6 (8.3) < .05 Table 2 2 Mean response time ( SD ) in the three validity condition s (100%, 80%, and 60%) displayed separately for the three bins and two age groups. Bin Response time in milliseconds validity Young Adults (n = 19) Older Adults (n = 19) 100% 80% 60% 100% 80% 60% 1 522 ( 133 ) 525 (143 ) 539 (136 ) 595 (165 ) 591 (167 ) 5 86 (157 ) 2 510 (119 ) 518 (133 ) 518 (144 ) 577 (154 ) 579 (157 ) 580 (164 ) 3 511 (122 ) 513 (126 ) 527 (124 ) 573 (164 ) 587 (150 ) 591 (155 )

PAGE 72

72 Table 2 3 Mean accuracy ( SD ) in the three validity conditions (100%, 80%, and 60%), displayed separately for the thre e bins and two age groups Bin Accuracy in % correct, validity Young Adults (n = 19) Older Adults (n = 19) 100% 80% 60% 100% 80% 60% 1 .74 ( .09) .61 ( .08) .47 ( .06) .63 ( .10) .57 ( .10) .47 ( .06) 2 .85 ( .10) .65 ( .10) .49 ( .06) .72 ( .10) .61 ( .09) .48 ( .06) 3 .86 ( .09) .68 ( .08) .50 ( .05) .75 ( .12) .60 ( .11) .52 ( .06) Table 2 4. Mean ( SD ) number of trials per condition in each age group. ERP Young Adults (n = 19) Older Adults (n = 19) 100% 80% 60% 100% 80% 60% ERN 38 (16) 76 (15) 109 (14) 6 2 (21) 85 (20) 108 (11) CRN 174 (26) 136 (23) 103 (12) 149 (26) 127 (25) 107 (14) FRN(neg) 35 (17) 72 (19) 103 (22) 60 (23) 84 (25) 105 (18) FRN(pos) 161 (35) 126 (30) 96 (18) 143 (33) 122 (26) 100 (20) Table 2 5 Mean amplitude s ( V ) of t he ERN in the three validity conditions (100%, 80%, and 60%), displayed separately for the three bins and two age groups Bin Young Adults (n = 18) Older Adults (n = 19) 100% 80% 60% 100% 80% 60% 1 .01 ( 1.9 ) .2 2 ( 1.6 ) .63 ( 1.8 ) .09 ( 1.3 ) 1 8 ( 1.2 ) .15 ( 1.4 ) 2 .19 ( 3.1 ) .67 ( 1.6 ) .80 ( 1. 6) .03 ( 1.0 ) .11 ( 1.2 ) .07 ( 1.3 ) 3 2.0 ( 3.0 ) 0 1 ( 1.6 ) .59 ( 1.4 ) .23 ( 1. 4 ) .18 ( 1.2 ) .04 ( 1.8 ) Table 2 6 Mean amplitudes ( V ) of the CRN in the three validity conditions (100% 80%, and 60% validity), displayed separately for the three bins and two age groups Bin Young Adults (n = 18) Older Adults (n = 19) 100% 80% 60% 100% 80% 60% 1 .77 (1.6 ) .19 (1.8 ) .29 (1.6 ) .21 (1.3 ) .11 (1.4 ) .13 (1.4 ) 2 1.65 (1.3 ) 1.31 (1.8 ) .83 (1. 7 ) .17 (1.2 ) .37 (1.8 ) .32 (1.5 ) 3 2.0 5 (1.2 ) .75 (1.5 ) .73 (1.4 ) .47 (1.3 ) .15 (1.7 ) .25 (1.1 )

PAGE 73

73 Ta ble 2 7 Summary of the 2 Group x 2 Response Type x 3 Validity x 3 Bin repeated measures ANOVA conducted on the ERN /CRN mea n amplitude data. F p partial eta 2 Group a Response Type a 17 5 .001 27 Validity b Bin b Group x Response Type a 19 8 < .001 .3 6 Group x Validity b Group x Bin b 9.7 < .0 0 1 .22 Response Type x Validity b 14 8 < .001 .2 9 Respo nse Type x Bin b 4.8 < .05 .12 Validity x Bin c Group x Response Type x Validity b 7. 6 .001 .17 Group x Response Type x Bin b 6.0 < .0 1 .15 Group x Validity x Bin c 3.0 < .05 .08 Response Type x Validity x Bin c 4.4 < .0 1 .11 Group x Response Type x Validity x Bin c 3 6 < .0 1 .09 a df = 1,35 b df = 2,70 c df = 4,140 Ta ble 2 8 Mean amplitudes ( V ) of the non reward related FRN in the three validity conditions (100%, 80%, and 60% validity), displayed separately for the three bins and two age groups Bin Young Adults (n = 19 ) Older Adults (n = 19) 100% 80% 60% 100% 80% 60% 1 1.0 (2.8 ) 1.2 (3.2 ) 1.0 (3.6 ) .3 (2.4 ) .7 (2.2 ) .3 (1.9 ) 2 .1 (4.2 ) .6 (2.5 ) .7 (2 6) .04 (2.2 ) .01 (2.5 ) .3 (3.3 ) 3 .3 (2.9 ) .7 (2.5 ) 1.3 (2.7 ) .01 (2.3 ) .1 (2.8 ) .2 (3.0 ) Table 2 9 Mean amplitudes ( V ) of the reward related FRN in the three validity conditions (100%, 80%, and 60% validity), displayed separately for the three bins an d two age groups Bin Young Adults (n = 19) Older Adults (n = 19) 100% 80% 60% 100% 80% 60% 1 .2 (2.2 ) .4 (2.3 ) .8 (2.4 ) .4 (2.6 ) .5 (2.8 ) .4 (4.1 ) 2 1.2 (2.3 ) .7 (2.4 ) .5 (2. 6) .6 (2.4 ) .6 (3.0 ) .8 (3.0 ) 3 1.3 (2.2 ) .7 (2.1 ) .4 (2.2 ) .4 (3.2 ) .6 (2.5 ) .2 (3.9 )

PAGE 74

74 Table 2 10 Summary of the 2 Group x 2 Feedback Type x 3 Validity x 3 Bin ANOVA conducted on the reward and non reward FRN mean amplitude data. F p partial eta 2 Group a Feedback Type a 8 9 < .0 1 .15 Val idity b 3.6 < .05 Bin b Group x Feedback Type a Group x Validity b Group x Bin b Feedback Type x Validity b 6 .5 < .0 1 .1 5 Feedback Type x Bin b 3.2 < .05 .08 Validity x Bin c Group x Feedback Type x Validity b 5 3 < .0 1 .1 3 Group x Feedback Type x Bin b 7.8 .0 0 1 .18 Group x Validity x Bin c Feedback Type x Validity x Bin c Group x Feedback Type x Validity x Bin c a df = 1,36 b df = 2,72 c df = 4, 144 Table 2 11 Significant correlations between FRN difference waves and neuropsychological measures for younger and older groups combined. 100% 80% 60% Bin 1 Bin 2 Bin 3 Bin 1 Bin 2 Bin 3 Bin 1 Bin 2 Bin 3 NART .33* COWA .36* BNT .37 Trails A Errors .34* Trails B .33* LARS .35* p < .05 (two tailed)

PAGE 75

75 Figure 2 1 Accuracy over time in each of the three validity conditions displayed separately for A) younger and B) older adults.

PAGE 76

76 Figure 2 2 Spherical spline voltage maps for the ERN CRN difference waves in both groups, taken at 60 ms. (Note different voltage scale ranges for the different probability conditions.)

PAGE 77

77 Figure 2 3 Amplitude of the ERN CRN difference wave in each condition over time displayed separately for A) younger and B) older adults. Error bars represent standard error of the mean (SEM). (Note that y axis scales differ between groups.)

PAGE 78

78 Figure 2 4 Grand averaged response locked ERPs taken from electrode FCz displayed separately for each group in each validity c ondition collapsed across all three bins Arrows indicate approximate location of the ERP component. Microvolts on the y axis, milliseconds on the x axis. Negative is plotted up by convention.

PAGE 79

79 Figure 2 5 Grand averaged response locked ERPs at elect rode FCz demonstrating learning related effects for each group in the 100% validity condition Microvolts on the y axis, milliseconds on the x axis. Negative is plotted up by convention.

PAGE 80

80 Figure 2 6 ERN amplitudes in each condition over time displayed separately for A) younger and B) older adults CRN amplitudes in each condition over time displayed separately for C) younger and D) older adults Error bars represent SEM. (Note that y axis scales differ between groups.)

PAGE 81

81 Figure 2 7 Spherical spline v oltage maps for the FRN cFRN difference waves in both groups. (cFRN = FRN to reward. Note different voltage scale ranges for the different probability conditions.)

PAGE 82

82 Figure 2 8 Amplitude of the non reward minus reward FRN difference wave in each condit ion over time displayed separately for A) younger and B) older adults. Error bars represent standard error of the mean (SEM). (Note that y axis scales differ between groups.)

PAGE 83

83 Figure 2 9 Grand averaged feedback locked ERPs taken from electrode FCz dis played separately for each group in each validity condition collapsed across the bins. Arrows indicate approximate location of the ERP component. Microvolts on the y axis, milliseconds on the x axis. Negative is plotted up by convention. (CFRN = reward rel ated FRN.)

PAGE 84

84 Figure 2 10 Grand averaged feedback related ERPs taken from electrode FCz demonstrating learning related effects for each group in the 100% validity condition Microvolts on the y axis, milliseconds on the x axis. (CFRN = reward related FR N.)

PAGE 85

85 Figure 2 11 Non reward related FRN amplitudes in each condition over time displayed separately for A) younger and B) older adults. Reward related FRN amplitudes in each condition over time displayed separately for C) younger and D) older adults. E rror bars represent SEM. (Note that y axis scales differ between groups.)

PAGE 86

86 Figure 2 12 A) Mean accuracy in each condition B) Mean amplitude of the ERN CRN difference wave in each condition. C) Mean amplitude of the FRN to non reward minus the FRN to reward difference wave in each condition. Error bars represent SEM.

PAGE 87

87 CHAPTER 3 DETECTION AND FEEDBACK PROCESSING Overview and Predictions Experiment 2 was conducted in order to examine the effects o disease on ERP reflections of error detection and feedback processing. Because these aspects of cognitive functioning are thought to be related to integrity of the frontal lobe and its connections to other brain regions, it was expected that the greater degree of fronto striatal dysfunction in P D would result in greater impairmen ts than those seen in community dwelling older adults who presumably exhibit a lesser degree of disruption to these regions. Thus it was predicted that medicated pat ients with PD would perform more poorly on the probabilistic learning task Consistent with some previous reports (Falkenstein et al., 2001; Stemmer et al., 2007) it was initially anticipated th at PD patients would exhibit decreas ed ERNs compared to older adult controls and that the older adult controls would exhibit greater increases in error related activity over time compa red to medicated patients with PD ; thus group differences in ERP reflections of error detect ion should be most apparent at the end of the experiment Because ERN amplitude increases over time were not detected in the older adults during Ex periment 1, however, predictions were revised to reflect similar ERN amplitudes in medicated PD patients comp ared to older adult controls as reported elsewhere (Holroyd et al., 2002) In light of previous reports of intact positive feedback processing in medicated PD patients using purely behavioral measures (Cools et al., 2006; Frank et al. 2004) it was expect ed that patients with PD would demonstrate similar amplitudes as older controls for the FRN to reward ; however, in keeping with the prediction of greater disruption to frontally mediated cognitive functioning in PD, it was anticipate d that patients with PD

PAGE 88

88 would demonstrate reductions in the amplitude of the FRN to non reward compared to older controls. In ex ploratory analyses, relationships between ERP component amplitudes, performance on neuropsychological tests, and self reported mo od symptoms were examined In addition to the expected relationships between FRN difference waves and performance on measures of executive functioning, part icularly Digits Backward verbal fluency (COWA) and the Wisconsin Card Sorting Test, it was expected that FRN amplitudes would correlate with self reported symptoms of apathy (AES, LARS). Methods Participants Data from the older adult sample recruited for Experiment 1 were used in Experiment 2. PD patients for Experiment 2 were recruited through the com munity and through the Neurology Clinic at the Malcom Randall VA Medical Center in Gainesville, Florida. Of th e t hirty five participants recruited, t wo older adults and four patients with PD were excluded from data analysis due to technical difficulties during data acquisition or for excessive noise in the EEG data Exclusionary criteria for the control participants included a history of learning disability, neurological disease, or head injury. To be included in the PD group, patients had to be non deme nted and meet diagnostic criteria for PD. Clinical criteria for PD diagnosis included at least two of four cardinal motor signs (akinesia, bradykinesia, resting tremor, and rigidity; (Hughes, Ben Shlomo, Daniel, & Lees, 1992) and a history of demonstrated therapeutic response to dopamine replacement therapy as measured by improvement in motor signs on the United Parkinson Disease Rating Scale (UPDRS; (Fahn & Elton, 1987)

PAGE 89

89 In addition to the nineteen older adult controls (ages 56 76) recruited for Experiment 1, t he final sample for Experiment 2 included ten age matched patients with Pa (ages 57 78) The two groups did not differ with respect to age t (27) = .08, p > .90, gender ( 2 = .44 p > .5 0) handedness ( 2 = 1.53 p > 21), educatio n t (27) = .14 p > .88 or premorbid estimated FSIQ, t (27 ) = .84, p > .16 Due to the small number of participants in the PD group, sensitivity analyses and post hoc power analyses were conduc ted using G Power software (Faul, 2007) By convention, the following interpretations of effect sizes will be used: f = .1 ( small ) f = .25 ( medium ) and f = .4 ( large ) (Cohen, 1992) All participants in this study obtained scores of 24 or higher on the MMSE (Folstein et al., 1975) and 130 or higher on the Dementia Rating Scal e (Mattis, 1988) in this study. Al l pati ents were in Hoehn Yahr stage s 1 3 when tested on medication. Of the ten participants with PD, one received a Hoehn Yahr score of 1, five were given the score of 2 two were scored as 2.5, and two were scored as 3. Demographic and neuropsychological data for the participants are shown in Table 3 1. Procedures After informed consent was obtained, participants began an experimental sess ion which lasted approximately three hours. All individuals completed all tasks in one session. Participants receive d financial compensation ($30) for their participation.

PAGE 90

90 Results Behavioral Data Reinforcement learning task performance T he number of timed out trials did not differ between groups ( M = .02 SD = .01 for older controls M = .03 S D = .03, for PD patients ) See Table 3 2 for a summary of mean response times. Accuracy data for each group in the three validity conditions and three bins is presented in Table 3 3 and Figure 3 1 The accuracy data were initially analyzed with an ANOVA design with the factors group ( older control, PD ) and validity (100%, 80% and 60% validity). Results of a sensitivity analysis (alpha = .05, power = .80, mean correlation between repeated measures = .27) found that this design should be able to detect significant between group e ffects up to f = .49 and significant group x validity interactions up to f = 42 The ANOVA revealed no significant main effect of group, F ( 1 27) = .38 p = .54 partial eta squared = .014, observed f = .11 T here was a significant ma in effect of validity, F (2, 54) = 51.36 p < .001, partial eta squared = .66 Contrasts for each level of the validity factor showed a higher accuracy for the 100% compared to the 80% validity condition and for the 80% compared to the 60% validity conditio n ( p s < .0 0 01). There was no significant interaction between group an d validity F (2, 54 ) = .10, p = .89, partial eta squared = .004, observed power = .06, f = 07 We conducted a sensitivity analysis for the main effect of bin and the interaction between bin and group (alpha = .05, power = .80, mean correlation between repeated measures = 51 ) and found that this d esign should be able to detect effect s up to f = 31 Examination of effects of learning over time using a repeated measures ANOVA (2

PAGE 91

91 g roup x 3 conditions x 3 bins) revealed a significant main effect of bin F (2, 54) = 19.48 p < .0001, partial eta squared = .42 Post hoc pairwise comparisons of accuracy during each bin revealed significantly better performance in the last bin compared to the first bin and in the second bin compared to the first bin ( p s < .0001); however, the improvement between the s econd and third bins was not significant. There was a significant interaction between validity and bin F (2.74, 74.10) = 3.1 5 p < .05, partia l eta squared = .11 observed power = .68 In the 100% valid condition, the older adult control group improved significantly from bin 1 to bin 2 and from bin 1 to bin 3 ( p s < .0001), and demonstrated a marginally significant improvement from bin 2 to bin 3 ( p = .08). In contrast, the PD group only demonstrated significant improvement in accuracy from bin 1 to bin 3 ( p < .05). In the 80% condition, the older adult control group showed significant improv ements in accuracy from bin 1 to bin 2 ( p < .05), but n ot from bin 2 to bin 3. PD patients did not show significant improvements between any of the bins in this condition. As expected, there were no significant learning related changes in the 60% condition for either group. There were no significant bin x gro up [ F (2, 54) = .03, p = .97, partial eta squared = .001, observed power = .06 f = .06 ] or valid ity x bin x group interactions. Cognitive and emotional functioning The groups performed comparably on all neuropsychological tests except Digit Span f orward and Stroop Word Reading (see Table 3 1) The patients with PD performed more poorly than the older adult control participants on these two tasks ( ps < .05 ) The PD patients also demonstrated trends toward slower performance on Stroop Color Reading ( p = .09) and more errors committed on Trails B ( p = .07) Both groups endorsed similar levels of symptoms of anxiety and depression. PD patients endorsed

PAGE 92

92 a higher level of apathy symptoms than older adult control subjects on the AES; however, they endorse d similar levels of apathy symptomatology on the LARS. Although no individuals met diagnostic criteria for any psychiatric disorder currently three member s of the PD group obtained score s on the AES above the conventional clinical cutoff for apathy (14), and one member of the older group and two members of the PD group obtained score s on the BDI II in the ra nge for mild depression (>14) Event Related Potential Data Table 3 4 presents the number of trials comprising the ERP waveforms in each condition fo r the two groups. The two groups differed in numbers of trials per waveform in the 6 0% condition such that the incorrect feedback FRN waveforms ( p < .0 5 ) for PD patients had more tri als than those of controls No significant group differences were observe d in the other probability conditions. Response locked ERPs (ERN/CRN) In the first step, incorrect minus correct difference waves were calculated. These difference waves showed maximal negative amplitudes with a latency of approximately 45 ms in the PD gr oup and 60 ms in the control group shown in Figure 3 2 This roughly corresponds to the latencies of the peak amplitude seen in the original ERN and CRN waveforms. Mean amplitudes of these difference waves were calculated using a time window from 15 75 ms for the PD group and 30 90 ms for the control group with each window centered on the peak amplitude as observed in the grand means for each group A n initial comparison of the mean amplitudes of the difference waves was conducted using a repeated measur es ANOVA (2 grou p x 3 validity condition x 3 bin ). We also conducted a sensitivity analysis (alpha = .05, power = .80, mean correlation between repeated measures = .36 ) and found that this design should be able to detect

PAGE 93

93 significant between group effects u p to f = 52, significant effect of condition up to f = .36 and significant g roup x validity interactions up to f = 3 8 W e did not detect any significant main effects or interactions. [Main effect of group: F (1, 27) = .03, p = .87, partial eta squared = .001, observ f = .0 3 ; main effect of validity condition: F (2, 54) = .49 p = .6 1 partial eta squared = .018, observed power = .1 3 f = 14 ; interaction: F (2, 54) = .33, p = .72, partial eta squared = .012, observed power = f = .1 1 ]. We also failed to detect any significant learning related effects on amplitude across bins [ F ( 2 54) = 1.0 p = 38 partial eta squared = .036 observed power = 22 f = .1 9 ] (see Figure 3 3) Sensitivity analysis of the wit hin and associated interactions (alpha = .05, power = .80, mean correlation between repeated measures = .12) revealed that this design should be able to detect significant effects of f = 32 There was no significant bin x group inter action [ F (2, 54) = .84, p = .44, partial eta squared = .03 observed power = .1 18 ] or bin x condition interaction [ F (4, 108) = .55 p = .70 partial eta squared = .02, observed power = .18 f = .14 ] In the second step, the respons e locked components were examined separately in order to examine the effects of disease status on changes in mean amplitude by response type (correct or incorrect). The response locked components were measured as the mean amplitude within a 60 ms time wind ow centered on the peak of the ERN at electrode FCz. A 2 (group) x 2 (response type) x 3 (validity) x 3 (bin) repeated measures ANOVA was conducted. Response locked ERP waveforms from the probabilistic learning task can be seen in Figure 3 4 Mean ERP amp litude and latency

PAGE 94

94 data are pr esented in Tables 3 5 through 3 6. S ensitivity analysis (alpha = .05, power = .80, mean correlatio n between repeated measures = .76 for valence and .80 for condition ) revealed that this design should be able to detect a s igni ficant group main effect up to f = 64 main effect of valence up to f = 24 main effect of condition up to 20 a significa nt group x valence interaction up to f = 19 and a significant group x validity interaction up to f = 16 In line with the result s from the analysis of difference waves, we detected no significan t main eff ects or interactions [ main effect of group: F (1, 27) = 2.58 p = .12 partial eta squared = .09 observed power = .34 f = 31 ; main effect of valence: F (1, 27) = .18, p = .68, partial eta squared < .01 observed power = f = 08 ; main effect of validity condition: F (2, 54) = .45 p = .6 4, partial eta squa red = .016 observed power = .12 f = 13 ; group x valence interaction: F (1, 27) = .009 p = .93 par tial eta squared < .0001 observed f = 0 1 ; group x validity condition interaction: (2, 54) = 1.37 p = .26 partial eta squared = .05 observed power = .28, f = .2 3 ] E xamination of tests of within subjects contrasts revealed a marginally significant group x condition interaction ( p = .09). Follow up ANOVAs demonstrated that these differences occurred in the 100% condition for both the ERN [ F (1, 27) = 3.34, p < .09] and the CRN [ F (1, 27) = 4.03, p < .06] such that the mean amp litudes of the response locked ERPs were larger for the PD group than for the control group. W e were unable to detect any significant effects of learning over time [ main effect of bin : [ F ( 2 54) = 1.74 p = .19 partial eta squared = .06 observed power = .3 5 f = .25 ] see Figure 3 5 and 3 6 Sensitivity analys e s of the within subjects (alpha = .05, power = .80, mean correlation

PAGE 95

95 between repeated measures in the 100% condition = .59 ) revealed that this design should be able to detect significant effects up to f = 2 3 We detected no significant bin x group [ F ( 2 54) = 1.25 p = .29 partial eta squared = .044, observed power = .26 f = 21 ] bin x valence [ F ( 2 54) = .99 p = .38 partial eta squared = .04 observed power = .21 f = 19 ] or bin x validity condition [ F ( 4 108) = .55 p = .70 partial eta squared = .02, observed power = .18 f = 14 ] interactions. Feedback locked ERPs (FRN) In the first step, non reward minus reward differ ence waves were calculated. These difference waves showed maximal negative amplitudes with a latency of approximately 290 ms for older controls and 312 ms for PD patients as shown in Figure 3 7. This roughly corresponds with latencies of the peak amplitude seen in the original reward and non reward related FRN waveforms. Mean amplitudes of these difference waves were calculated using a time window from 260 320 ms for older controls and 282 342 ms for PD patients As with the response related activity, t he mean amplitudes of the difference waves for feedback related activity were compared using a repeated measures ANOVA (2 group x 3 validity condition) We conducted a sensitivity analysis (alpha = .05, power = .80, mean correlation between repeated measur es = 64 ) and found that this design should be able to detect significant between group effects up to f = .59, a significant effect of validity condition up to f = .31 and a significa nt group x validity condition interaction up to f = .2 3 N o significant m ain effect of group was detected [ F (1, 27) = .63 p = .44 partial eta 2 = .02, observ f = .15]. There was no significant main effect of condition [ F (2, 54 ) = 2.41 p = .12 partial eta 2 = .08 observed power = .40 f = .30]. The group x validity condition interaction was not

PAGE 96

96 significant [ F (2, 54) = .9 9, p = .38 partial eta 2 f = .19 ] Sensitivity analys e s of the within (alpha = .05, power = .80, mean correlation between repeated measures = .15 ) revealed that this design should be able to detect significant effects up to f = 33 Examination of effects of learning over time revealed no main effect of bin [ F (2, 54) = 1.27 p = .2 9, partial et a 2 = .05, observed f = .22 ] There was a m arginally significant bin x group interaction [ F (2, 54) = 2.49, p = .09, partial eta 2 = .08 f = .30 ] This was a quadratic trend such that PD patients exhibited the largest amplitudes in the first bin, followed by the third bin, then the second bin In contrast, the control group exhibited the largest amplitudes in the second bin followed by the first bin, then the third bin (see Figure 3 8) A significant bin x condition interaction was not detected [ F (4, 108 ) = 1.50 p = .22 partial eta 2 = .05 f = 23 ] In the second step, the feedback locked components were examined separately in order to investigate the effects of group on chang es in mean amplitude by feedback type (reward or non reward). The feedback locked components were measured as the mean amplitude within a 60 ms time window centered on the peak of the FRN at electrode FCz (260 m s in older controls and 282 for PD patients ). A 2 (group) x 2 (feedback type) x 3 (validity) repeated measures ANOVA was conducted as well as a sensitivity analysis (alpha = .05, power = .80, mean correlati on between the repeated measure valence = 89; mean correlation between the repeated measure c ondition = .90 ). Based on the analysis, t his design should be able to detect significant between

PAGE 97

97 group effects up to f = .52 significant within gr oup effects up to f = 32 and significa nt within between interaction s up to f = .17 Feedback locked ERP wave forms from the probabilistic learning task can be seen in Figures 3 9 and 3 10 Mean ERP amplitude data are presented in Tables 3 7 through 3 8 Although w e did not find main effect s of group [ F (1,27) = .009 p = .92 partial eta 2 < .0001, observed power = .05 s f < .03 ] or validity condition [ F (2, 54 ) = 2.08 p = .1 5, partial eta 2 = .07 2 observed power = .35 f = .28 ], there was a significant main effect of feedback type [ F (1,27) = 6.50 p < .05, partial eta 2 = .19, observed power = .69 f = .48 ] such that the FRN to non reward was more negative than the FRN related to reward. We were unable to detect any significant group x validity [ F (2, 54) = 1.51 p = .23 partial eta 2 = .05 s f = .24 ] group x fee dback type [ F ( 1 27) = .57 p = .46 partial eta 2 = .021, observed power = .11 s f = .15 ] or validity x feedback type interactions [ F (2, 54 ) = .86 p = .40 partial eta 2 = .03 observed power = .17 s f = .15 ] When we added the effect of bi n, we were unable to detect any significant main effect of bin [ F ( 2 54 ) = .27 p = .71 partial eta squared = .01, observed power = .09 s f = .10 ] We were also unable to detect significant bin x group [ F ( 2, 54) = .22 p = .81 partial eta 2 = .01 observed power = .08, s f = .09 ] bin x validity condition [ F (4, 108 ) = .93 p = .42 partial eta 2 = .033, observed power = .35, f = 18 ] or bin x feedback type [ F (2, 54 ) = 1.04 p = .35 partial eta 2 = .037, observed power = 21, f = .2 0 ] interactions but t he main effect of feedback type was qualified by a marginally signific ant interaction between group, bin and feedback type [ F (2, 54) =

PAGE 98

98 2.95 p = .06 partial eta 2 = .10 observed power = .55 f = .33 ] (see Figure 3 11) R elationship to performance on neuropsychological tests and disease variables When bot h groups were combined, several neuropsychological variables correlated with FRN difference wave amplitudes (see Table 3 9 ). In the 100 % valid condition, performance on Di git Span Backwards was positively correlated with the FRN difference wave in bin 1 ( r = .37 p < .05) ; time to complete Trails B was positively correlated with the FRN difference wave in bin 2 ( r = 40 p < .05 ); number of categories completed on the WCST was negatively correlated with the FRN difference wave in bin 1 ( r = .37 p < .05) ; number of perseverative errors on the WCST was positively correlated with the FRN difference wave in bin 2 ( r = 43 p < .05 ), and performance on the BNT was negatively co rrelated with the FRN difference wave in bins 1 and 2 ( r = .45, p < .05) In the 80% valid condition, the amplitude of the FRN difference wave in bin 3 was positively correlated with age ( r = .37, p < .05 ) and time to complete Trails B ( r = 38 p < .05 ) and negatively correlated with number of categories completed on the WCST ( r = 31 p < .05 ). Within the PD group duration of PD symptoms was negatively correlated with the FRN difference wave in the 100% condition across the entire experiment ( r = 6 3 p < .05 .) and Hoehn Yahr stage was positively correlated with the FRN difference wave in the 80% condition in bin 3 ( r = 86 p < .001). Discussion The purpose of Experiment 2 was to use the same experimental paradigm used in Experiment 1 to investigate feedback processing. Of particular interest was the question of whether differences

PAGE 99

99 existed with respect to valence. Thus, in addition to feedback probability, feedback valence was manipulated and ERP reflections of error detection and feedback processing were compared between and age matched controls. As in the first experiment, a secondary (exploratory) goal was to explore relationships between feedback processin g, emotional sympt oms and executive functioning. Effects of Aging and PD on ERP Reflections of Reinforcement Learning Behavioral data T he groups performed comparably on the reinforceme nt learning task and demonstrated better performance in the 100% condit ion compared to the 80% condition and in the 80% condition compared to the 60% condition. Both groups generally improved over time in the least ambiguous (i.e., 100%) learning condition ; however, only the older adult control group improved in the more amb iguous (i.e., 80% ) condition from the first to second bins. As expected, neither group exhibited learning in the 60% condition. Although we predicted that PD patients would perform more poorly than the control group equivalent performance on feedback base d learning tasks between older adult controls and medicated patients with PD has been reported previously (Shohamy et al., 2005; Swainson et al., 2006) and could be due to factors such as demographic and disease var iables, or task difficulty and demands. For example, both groups in our sample were premorbidly high functioning, and PD patients were generally in mild to moderate stages of disease. Moreover, in the current study the use of an adaptive response deadline successfully equated the number of trials attempted between groups, providing equivalent opportunity to learn from feedback In addition, it has previously been reported that PD patients perform comparably to controls on similar tasks

PAGE 100

100 requiring learning under ambiguous conditions; this has been attributed to relative sparing of limbic dependent processes as compared to processes more dependent on dorsolateral prefrontal function (Euteneuer et al., 2009) Response l ocked ERPs (ERN/CRN) In line with the revised hypothesis that accounted for the unexpectedly reduce d ERN amplitudes in older adult controls found in Experiment 1 significant group differences were not found in the amplitude of the ERN CRN difference wave nor did we find amplitude differences by condition, or increased amplitude over time These null findings are in contrast to studies reporting reduced amplitude ERNs in medicated (Falkenstein et al., 2001; Ito & Ki tagawa, 2006) and unmedicated (drug nave) (Falkenstein et al., 2001; Stemmer et al., 2007) patients with PD as we ll as in the same patients test ed both on and off medication (Willemss en et al., 2008) when compared to older adult controls P revious reports of reduced am plitude ERNs in PD patients fit well with the dopamine hypothesis of the reinforcement learning theory, by suggesting that degeneration of dopaminergic neurons leads to disrupted signaling and reduced activation of the ACC ; however, as with behavioral data, reasons for disparate results between studies could be due to several factors. For example, it is possible that the ERNs of our older adults were unusually small, cont ributing to the appearance of he amplitude of ERP waveforms can be measured in many ways, often with disparate effects. ERN amplitude has been measured variously using peak to peak scoring, difference wave calculations, mean amplitudes, and base to peak scoring approaches. In the present study, mean amplitudes were used because it has been suggested that this method help s to account for artificially increased maximum amplitud e that can occur

PAGE 101

101 to unequal number of trials per waveform ( Luck, 2005) Difference wave calculations were also used based on the work of previous investigators (e.g., Eppinger et al., 2008). D ifferences in paradigm de sign may also impact results, such as use of tasks that are more complex and/or contain a smaller number of trials (Falkenstein et al., 2001; Stemmer et al., 2007), which may not allow patients to learn and develop increased ERN amplitudes. Finally, diffe rences in patient sample characteristics, such as inclusion of patients in more severe stages of disease (e.g., Ito & Kitagawa, 2006) may also contribute to differences in results In support of our data Holroyd and colleagues (Holr oyd et al., 2002) have previously reported spared error related potentials in a group of PD patients in mild to moderate stages of disease suggesting perhaps that early stage r When mean amplitudes of the ERN an d CRN were measured separately results were surprising T hough findings of the current study should be interpreted with caution in light of the small sampl e size and relative lack of power PD patients exhibited greater amplitude ERNs and CRNs than controls in the least ambiguous condition Moreover, although we did not find significant learning related changes in amplitude examination of data patterns sugg ested that the ERN and CRN amplitudes of the PD patients began to dissociate (i.e., the ERN became larger as the CRN became smaller) over time as has been observed in a group of healthy young controls (Pietschmann et al., 2008 ). To our knowledge, these re lationships have not been reported previously in a group of PD patients though similar CRN amplitudes have been found between patients and

PAGE 102

102 controls (Falkenstein et al., 2001; Willemssen et al., 2008) Interpreted within the dopamine hypothesis of the reinforcement learning theory, perhaps dopaminergic medication was able to restore the error aging. Because patients in published studies are often taking a variety of dif ferent reports suggest that dopamine agonists increase ERN amplitude (de Bruijn, Hulstijn, Verkes, Ruigt, & Sabbe, 2004) it would be worthwhile to test whether different medications affect the erro r processing system in different ways. The results of the current study are interesting in light of suggestions that older adu lts with dopaminergic agents might improve their performance on rei nforcement learning tasks O f note, even if dopaminergic medication contributes to appropriate signaling to the ACC (and corresponding generation of ERNs and CRNs) it is not clear that this activ ation improve s behavioral outcomes i n terms of increased accuracy ; many previous reports describe la ck of cognitive enhancement with dopaminergic medication (Cools, 2008) Feedback locked ERPs (FRN) A small amount of support for FRN r elated predictions of the reinforcement learning theory was found i n the data from the older adult controls and PD patients. Marginally significant results suggested greater amplitudes of the difference waves in the more ambiguous learning condition compa red to the least ambiguous learning condition. In addit ion, generally decreasing amplitudes of the difference wave were observed over time, though decline was not clearly demonstrated because it was a quadratic trend One of the most important hypotheses o f the current study was the prediction that aging and PD would have differential effect s on the processing of feedback depending

PAGE 103

103 on valence such that PD patients would exhibit similar FRN amplitudes to reward, but reduced amplitude FRNs to non reward comp ared to controls In line with our hypotheses, no differences between groups were found with respect to the amplitude of the FRN to reward. I f increased sample size and power were to uphold these findings, the current study would provide support for spare d positive feedback processing (Cools et al., 2006) Consistent with widely held accounts of the relative amplitude of the two feedback related components, t he amplitude of the FRN to non reward was greater than the amplitude of the FRN to reward. Although this outcome is positive since it suggests that the feedback processing system is relat ively intact in these two groups it also contributed to our failure to find predicted group differences in the ability to proce ss negative feedback It is possible that low statistical power contributed to these null findings since support for impaired reactivity to negative feedback in medicated PD patients is mounting For example, a recent behavioral study concluded that reacti vity to posit ive and negative feedback is differentially affected by disease and by particular dopaminergic medications used to treat its symptoms (Bodi et al., 2009) Bodi and colleagues (2009) found th at treatment nave patients demonstrated selective deficits in positive feedback (i.e., reward) processing but when patients were treated with dopamine agonists (pramipexole and ropinerole) these deficits were remediated and the ability to process negati ve feedback became impaired In addition, evidence has been found for blunted electrodermal responses to losses, but not to gains, in medicated patients with PD compared to older controls (Euteneuer et al., 2009).

PAGE 104

104 An explanation for these differential va lence effects b ased on recent neuroimaging findings suggests that PD patients recruit a compensatory system which shifts reward processing activity from more dysfunctional striatal systems to relatively better functioning frontal regions (Keitz et al., 2008) In healthy controls, increased activation in the left putamen is observed when provided monetary feedback compared to positive and neutral feedback; however, medicated PD patients exhibit increased activation i n the left putamen in response to neutral feedback compared to both types of positive feedback (Keitz et al., 2008; Kunig et al., 2000). In addition, PD patients exhibit increased activation in medial prefrontal cortex (Keitz et al., 2008; Schott et al., 2 007), dorsolateral prefrontal cortex (Kunig et al., 2000) and ACC (Kunig et al., 2000; Schott et al., 2007) in positive feedback conditions, which is not observed in controls. An alternative hypothesis was previously outlined explaining the differential response to feedback based on valence. Frank and colleagues (M. J. Frank & Claus, 2006; M. J. Frank et al., 2004) developed a n elegant neurobio logically based computational model of the role of the basal ganglia in learning and decision making This model suggests that phasic dopamine bursts generated in response to rewards increase synaptic plasticity in the direct pathway ( which connects the st riatum to the substantia nigra pars reticulata and the internal segment of the globus pallidus) and decrease activity in the indirect pathway ( which connects the striatum via the external segment of the globus pallidus to the substantia nigra pars reticula ta and the internal segment of the globus pallidus). Thus, increased synaptic plasticity in the direct pathway is thought to reinforce rewarding behavior In contrast, in response to negative or punishing feedback, phasic decreases in dopamine have the

PAGE 105

105 opp osite effect, leading to avoidance of harmful or bad decisions. Based on this model, it is thought that decreased dopamine in regions underlying processing of positive feedback (e.g., striatum) (Liu et al., 2007; Ni euwenhuis et al., 2005; O'Doherty et al., 2001) leads to impaired reward related learning in non medicated patients. On the other hand, when patients are treated with dopaminergic medication, these regions are y to positive feedback; however, reactivity to negative feedback is impaired, either because the medication blocks the effects of normal dopamine dips Euteneuer et al. 2009), or because regions underlying negative feedback processing (e.g., lateral OFC ACC, and insula) (G. K. Frank et al., 2005; Liu et al., 2007; O'Doherty et al., 2001) become In light of these hypotheses it is possible that impairment s in positive or negative fee dback processing were not detected because medication levels were appropriately balanced in our sample such that frontal regions were not overdosed Alternatively as suggested in the discussion of Experiment 1, it is possible that feedback processing mech anisms were relatively spared in this high functioning group of PD patients. Relationship to Neur opsychological Test Performance Consistent with expectations, f eedbac k related activity correlated with age and performance on measures of executive functionin g B ecause the reinforcement learning task required problem solving using feedback, it is not surprising that FRN amplitude was associated with better performance on the WCST as measured by a greater number of categories completed and fewer perseverative errors. I nterestingly it has previously been reported that performance on the WCST was not related to ERN amplitude in PD patients and older adul t controls (Falkenstein et al. 2001 ; Willems s en et al., 2008 ) perhaps suggesting that performance on the WC ST relies more on

PAGE 106

106 feedback than error processing Intact feedback pro cessing was also related to visuomotor sequencing and set switching, which may reflect reliance on cognitive flexibility for success on the reinforcement learning task and subsequent app ropriate changes in FRN amplitude Increased age was associated with smaller amplitudes of the difference wave. As in Experiment 1, better performance on a language measure was ass ociated with intact feedback processing The fact that this relationship ap peared in both experiments is somewhat surprising considering the restricted range of scores resulting from universally high performance on this measure. Counter intuitively, better working memory was associated with smaller amplitudes of the difference wa ve. This result is particularly unexpected since a wealth of evidence supports a relationship between dopaminergic signaling and working memory; however, it is possible (as hypothesized above), that dopaminergic medication resulted in ap propriate signaling as reflected by the FRN but the medication caused an (Cools, Gibbs, Miyakawa, Jagust, & D'Esposito, 2008; Cools, Lewis, Clark, Barker, & Robbins, 2007) Unfortunately, as in Experiment 1, we did not find expected relationships between feedback processing and emotional symptoms ( e.g., apathy) This null finding was likely due to the small sample size, as well as to a small range of scores on these measures. The inclusion of p atients in more advanced stages of disease would be necessary for required levels of apathy. Within the PD group, feedback processing was relat ed to several disease variables. S maller FRN amplitudes were related to greater disease severity as measured by Hoehn Yahr stage in keeping with well known relationships between dopamine uptake and disease severity (Benamer et al., 2000; Berding et al., 2003)

PAGE 107

107 There was a trend toward relationships between larger FRN amplitu des and levodopa equivalent dosage, which might support the hypothesis that appropriate medication management in our sample contributed to evidence for intact feedback processing. Somewhat surprisingly, shorter duration of symptoms was related to smal ler a mplitude difference waves; however, duration of symptoms is often an inaccurate marker of disease severity since variations exist not only with respect to progression of symptoms, but also with respect to how long symptoms were present prior to their ident ification.

PAGE 108

1 08 Table 3 1. D emographic and neuropsychological data for older adults and patients with Older (n = 19) PD (n = 10 ) Mean ( SD ) Mean ( SD ) p Demographics Age (years) 66.9 (6.7) 67.1 (6.5) ns Educatio n (years) 16.5 (2.1) 16.6 (2.7) ns Female (%) 31.6 -20.0 -ns Right Handed (%) 94.7 -80.0 -ns Cognitive Functioning MMSE 28.3 (1.7) 29.0 (1.0) ns Dementia Rating Scale 140.7 (3.1) 140.1 (3.9) ns NART FSIQ 114.8 (7.4) 109.4 (13.0) ns Boston Naming Test 56.8 (5.0) 57.5 (2.8) ns COWA (FAS) 39.0 (8.9) 35.7 (11.8) ns Semantic Fluency (Animals) 20.4 (3.7) 20.5 (6.1) ns Digit Span Forward 11.0 (2.6) 8.4 (2.3) < .05 Digit Span Backward 6.7 (2.1) 6.9 (2.6) ns Trails A (sec) 31.8 (9.1) 33.8 (11.1) ns Trails A (errors) .1 (.3) .1 (.3) ns Trails B (sec) 71.0 (26.8) 85.7 (33.2) ns Trails B (errors) .3 (.7) .9 (1.0) .07 Stroop Word Reading 108.1 (17.1) 93.1 (9.0) < .05 Stroop Color Naming 74.8 (12.4) 67.2 (8.3) .09 Stroop Color Word Naming 45.5 (11.0 ) 45.7 (9.0) ns Stroop Interference 2.5 (8.9) 3.8 (5.8) ns WCST Categories Completed 5.6 (1.0) 5.1 (1.9) ns WCST Perseverative Errors 12.5 (9.6) 11.5 (10.5) ns WCST Failure to Maintain Set .6 (.7) .7 (1.9) ns Emotional Functioning BDI II 3.6 (4.9) 5.7 (6.0) ns GDS 1.9 (3.6) 4.0 (6.3) ns AES 5.7 (3.0) 10.1 (6.0) < .05 LARS 24.8 (3.6) 26.9 (4.9) ns STAI State 26.9 (9.4) 25.8 (4.5) ns STAI Trait 27.6 (8.3) 29.7 (7.7) ns Disease Characteristics Duration of Symptoms (years) --7.9 (4.0) -UPDRS Motor On Meds --25.4 (10.2) -Hoehn Yahr Scale On Meds --2.2 (.6) -Levodopa equivalent dosage (mg) --754 4 ( 251.9 ) -Antidepressant Medications (%) 5 -4 0 -< .05

PAGE 109

109 Table 3 2. Mean response time (SD) in the three validity conditions (100%, 80%, and 60%) display ed separately for the three bins and two groups. Bin Response time in milliseconds validity Older Adults (n = 19) PD Patients (n = 10) 100% 80% 60% 100% 80% 60% 1 595 (165 ) 591 (167 ) 586 (157 ) 656 (170 ) 663 (170 ) 659 (170 ) 2 577 (154 ) 579 (157 ) 58 0 (164 ) 641 (166 ) 634 (156 ) 657 (185 ) 3 573 (164 ) 587 (150 ) 591 (155 ) 629 (166 ) 654 (157 ) 657 (156 ) Table 3 3. Mean accuracy ( SD ) in the three validity conditions (100%, 80%, and 60%), displayed separately for the three bins and two groups Bin Accu racy in each validity condition Older Adults (n = 19) PD Patients (n = 10 ) 100% 80% 60% 100% 80% 60% 1 .63 ( .10) .57 ( .10) .47 ( .06) .64 (.08) .54 (.05) .47 (.06) 2 .72 ( .10) .61 ( .09) .48 ( .06) 69 (.16) .59 (.10) .50 (.05) 3 .75 ( .12) .60 ( .11) .5 2 ( .06) .72 (.18) .61 (.15) .49 (.05) Table 3 4. Mean ( SD ) number of trials per condition in each group. ERP Older Adults (n = 19) PD Patients (n = 10 ) 100% 80% 60% 100% 80% 60% ERN 62 (21) 85 (20) 108 (11) 69 (29) 92 (17) 111 (11) CRN 149 ( 26) 127 (25) 107 (14) 154 (36) 126 (24) 105 (11) FRN(neg) 60 (23) 84 (25) 105 (18) 69 (31) 95 (20) 117 (9) FRN(pos) 143 (33) 122 (26) 100 (20) 148 (37) 124 (22) 103 (12) Table 3 5. Mean amplitudes ( V ) of the ERN in the three validity conditi ons displayed separa tely for the three bins and two groups Bin Older Adults (n = 19) PD Patients (n = 10 ) 100% 80% 60% 100% 80% 60% 1 .09 (1.3 ) .18 (1.2 ) .15 (1.4 ) .51 (1.2 ) .37 (1.8) .38 (.9) 2 .03 (1.0 ) .11 (1.2 ) .07 (1.3 ) .63 (1 .3 ) .56 (1.4) .69 (.8) 3 .23 (1.4 ) .18 (1.2 ) .04 (1.8 ) .95 (1.5 ) 1.1 (1.5) .49 (.8) Table 3 6 Mean amplitudes ( V ) of the CRN in the three validity conditions displayed separately for the three bins and two groups Bin Older Adults (n = 19 ) PD Patients (n = 10 ) 100% 80% 60% 100% 80% 60% 1 .21 (1.3 ) .11 (1.4 ) .13 (1.4 ) .90 (.8) .53 (1.4) .56 (1.4) 2 .17 (1.2 ) .37 (1.8 ) .32 (1.5 ) .74 (1.4) .68 (1.3) .93 (1.1) 3 .47 (1.3 ) .15 (1.7 ) .25 (1.1 ) .44 (.7) .57 (.7) .64 (1.1)

PAGE 110

110 Table 3 7. Mean amplitudes ( V ) of the non reward related FRN in the three validity conditions displayed separately for the three bins and two groups Bin Older Adults (n = 19) PD Patients (n = 10) 100% 80% 60% 100% 80% 60% 1 .3 (2.4) .7 (2.2) .3 (1.9) .2 (2.4) .7 (2.5) .1 (2.6) 2 .04 (2.2) .01 (2.5) .3 (3.3) .2 (2.5) .2 (2.1) .4 (2.5) 3 .01 (2.3) .1 (2.8) .2 (3.0) .1 (2.4) .5 (1.8) .6 (2.3) Table 3 8. Mean amplitudes ( V ) of t he reward related FRN in the three validity conditions (100%, 80%, and 60% validity), displayed separately for the three bins and two groups Bin Older Adults (n = 19) PD Patients (n = 10) 100% 80% 60% 100% 80% 60% 1 .4 (2.6) .5 (2.8) .4 (4.1) 7 (2.2) .7 (2.5) 1.1 (2.6) 2 .6 (2.4) .6 (3.0) .8 (3.0) .4 (2.0) .5 (1.6) .8 (2.3) 3 .4 (3.2) .6 (2.5) .2 (3.9) .5 (1.9) .6 (1.7) 1.5 (3.2) Table 3 9 Significant correlations between FRN difference waves and neuropsychological measures for PD and older control groups combined. 100% 80% 60% Bin 1 Bin 2 Bin 3 Bin 1 Bin 2 Bin 3 Bin 1 Bin 2 Bin 3 Age .37* Digits Backward .37* Trails B (raw) .40* .38* BNT .45* .45* WCST Categories .37* .3 1* WCST Pers Err .43* p < .05 ** p <.01 (two tailed)

PAGE 111

111 Figure 3 1 Accuracy over time in each of the three validity conditions displayed separately for A) older adult controls and B) patients with PD

PAGE 112

112 Figure 3 2 Spherical spline voltag e maps for the ERN CRN difference waves in both gro ups (Note different voltage scale ranges for the different probability conditions.)

PAGE 113

113 Figure 3 3 Amplitude of the ERN CRN difference wave in each condition over time displayed separately for A) olde r controls and B) patients with PD. Error bars represent standard error of the mean (SEM).

PAGE 114

114 Figure 3 4 Grand averaged response locked ERPs taken from electrode FCz displayed separately for each group in each validity condition collapsed across all thr ee bins Arrows indicate approximate location of the ERP component. Microvolts on the y axis, milliseconds on the x axis. Negative is plotted up by convention.

PAGE 115

115 Figure 3 5 Grand averaged response locked ERPs at electrode FCz demonstrating learning re lated effects for each group in the 100% validity condition Microvolts on the y axis, milliseconds on the x axis. Negative is plotted up by convention.

PAGE 116

116 Figure 3 6 ERN amplitudes in each condition over time displayed separately for A) older controls and B) patients with PD. CRN amplitudes in each condition over time displayed separately for C) older controls and D) patients with PD Error bars represent SEM.

PAGE 117

117 Figure 3 7 Spherical spl ine voltage maps for the non reward minus reward dif ference wave s in both groups. (Note different voltage scale ranges for the different probability conditions.)

PAGE 118

118 Figure 3 8 Amplitude of the non reward minus reward FRN difference wave in each condition over time displayed separately for A) older controls and B) pa tients with PD. Error bars represent standard error of the mean (SEM).

PAGE 119

119 Figure 3 9 Grand averaged feedback locked ERPs taken from electrode FCz displayed separately for each group in each validity condition collapsed across the bins. Arrows indicate ap proximate location of the ERP component. Microvolts on the y axis, milliseconds on the x axis. (CFRN = reward related FRN.)

PAGE 120

120 Figure 3 10 Grand averaged feedback related ERPs taken from electrode FCz demonstrating learning related effects for each grou p in the 100% validity condition Microvolts on the y axis, milliseconds on the x axis. (CFRN = reward related FRN.)

PAGE 121

121 Figure 3 11 Non reward related FRN amplitudes in each condition over time displayed separately for A) older controls and B) patients w ith PD. Reward related FRN amplitudes in each condition over time displayed separately for C) older controls and D) patients with PD. Error bars represent SEM.

PAGE 122

122 Figure 3 12 A) Mean accuracy in each condition. B) Mean amplitude of the ERN CRN differen ce wave in each condition. C) Mean amplitude of the non reward FRN minus reward FRN difference wave in each condition. Error bars represent SEM.

PAGE 123

123 CHAPTER 4 GENERAL DISCUSSION Review and Conclusions These two experiments compared error detection and feedb ack processing age matched controls. The first aim of this project was to provide support for predictions from the reinforcement learning theory by examining the ef fects of manipulation of feedback probability and valence on ERP reflections of error detection a nd feedback processing in a sample of healthy young adults. As predicted, the ERN was larger than the CRN and increased w ith increasing validity, reflecting t he idea that awareness of error commission is greatest under co nditions in which response selection is the least ambiguous. Also in line with predictions, the mean amplitude of the ERN and the ERN CRN difference wave increased over time, particularly in th e least ambiguous condition, providing support for the idea that as internal representations of correct and incorrect responses become more fixed with learning, greater conflict occurs when erroneous responses are made. Notably the increase in the differe nce wave was due, in part, to the fact that the CRN decreased as the ERN increased over time. This finding is important with regard to questions about differential processing of positive and negative valence information and lends support to recent recommen dations for the incorporation of a role for correct response activity in reinforcement learning and theories of the ERN and FRN (Eppinger et al., 2008; Pietschmann et al., 2008) Predictions from the theory were also supported by analyses of the feedback related activity. As expected, the non reward related FRN was larger than the reward

PAGE 124

124 related FRN. In addition, m ean amplitudes of the difference wave decreased with increasing feedback validity, reflecting the idea that reliance on feedback is greatest in the most ambiguous situations (Eppinger et al., 2008; Holroyd & Coles, 2002) L earning related effects on the difference waves were also consistent with the theory such that mean amplitude s generally decreased over time in both learning conditions, provid ing evidence that reliance on external feedback decreases with learning as representations of correct responses are internalized (Holroyd & Coles, 2002). When feedback related components were examined separately for positive and negative activity, it ap peared that predictions from the reinforcement learning theory with respect to non reward related activity were not supported. Although the theory did not provide predictions fo r reward related activity, the reward related FRN increased with increasing fe edback validity and increased over time, in direct contrast to expectations for non reward related activity E ach of these results has been reported previously; however, they have not been reported simultaneously ( Eppinger et al., 2008; Nieuwenhuis et al. 2002). These results also support the role for positive feedback processing in reinforcement learning and, when combined with the results from analysis of response locked activity, may suggest that reactivity to positive feedback increases over time as response certainty increases in healthy young adults. The second aim of the project disease on reinforcement learning (i.e., error and feedback processing), with a primary s disease differentially affect processing of n egative versus positive feedback Broadly dysfunction in the reinforcement learning system in both groups was most clear when examining error related activity F or the older adult

PAGE 125

125 controls ERN/CRN mean amp litudes were not altered by validity condition and changed very little over the course of the entire experiment, consistent with previous research (Pietschmann et al., 2008). The only exception to this observation was relatively intact correct response ac tivity in the least ambiguous condition at the end of the experiment. Interestingly, PD patients tended to exhibit greater amplitudes of response related activity compared to older adult controls, and amplitudes changed appropriately with learning. Group differences in feedback processing were not detected, contrary to expectations. These results were interpreted within the context of reinforcement learning theory and discussed in relation to medication and disease related effects on brain structures u nde rlying the processing of positive and negative information An exploratory third aim of this dissertation project was to examine relationships between feedback processing and other aspects of cognitive as well as emotional functioning. In Experiment 1, yo ung and older adults did not demonstrate relationships between measures of executive function and feedback processing. Instead, the most meaningful relationship was between intact feedback processing and attention. In Experiment 2, when the PD patients and older controls were combined expected relationships emerged between feedback pr ocessing and performance on tests of executive function, p articularly problem solving using feedback and cognitive flexibility. Interestingly, better performance on a language measure was associated with intact feedback processing in both experiments. An unexpected relationship between better working memory and smaller amplitudes of the difference wave was found in ect of dopaminergi c medication on working memory Within the PD group, smaller feedback related

PAGE 126

126 amplitudes were related to greater disease severity Unfortunately, we did not find expected relationships between feedback processing and emot ional symptoms (e.g., apathy), likely due to the small sample size, as well as to a small range of scores on these measures. The inclusion of patients in more advanced stages of disease would be necessary for required levels of apathy. T he identification of relationshi ps between feedback processing and psychological functioning could be important for designing interventions and should be explored in future research. Strengths and Limitations Unfortunately, recruitment difficulties faced throughout this project not only led to small sample size and decreased power, they also contributed to lack of generalizability of results to the larger population since members of all group s tended to be highly educated. Moreover, although it was likely necessary to modify this paradigm for use in a population who were expected to suffer from increased fatigability this adaptation may have clouded interpretation of results. For example, reducing the number of trials may have prevented the development of larger ERNs over time ; in additio n, reducing the number of bins may have reduced our power to detect learning changes in the latter part of the experiment; and the use of a 60% condition (rather than 50%) may have ambi guous (i.e. 80%) learning condition. In addition, as is often the case when conducting research in aging and in the latter case, there is a large degree of variability in symp tom presentation, which likely added to noise in the data and difficulty finding differences between groups

PAGE 127

127 A methodological strength of this study was its use of a parad igm that manipulated not only feedback validity, but also valence, which allowed fo r the testing of predictions from the reinforcement learning theory as well as investigation of distinct mechanisms underlying the processing of reward and punishment. Another strength of the study was its use of an adaptive response deadline in an attempt to equate the number of trials utilized by participants for learning. Although it is possible that the use of the adaptive deadline artificially reduced the amplitude of the ERN component in younger adults by reducing time pressure, it allowed us to flex ibly use the same paradigm in two groups for whom cognitive and/or motor slowing were likely to negatively impact performance, thereby potentially artificially d ecreasing their ERN amplitudes. Finally, this study combined methodology from the disciplines of neuroscience and neuropsychology in an function s ERPs were used to distinguish error detection and feedback processing r often indistinguishable using paper and pencil based tests alone. Although ERPs have been used in previous studies testing predictions from the reinforcement learning theory, those studies did not attempt to correlate electrophysiological reflections of processes mediated by frontal regions of the brain with more commonly used neuropsychological measures of similar functions. Such correlations may be important when making clinical interpretations of Implications and Fut ure Directions T he broad goal of this project was to use EEG to examine the brain mechanisms processing. The res ults of

PAGE 128

128 these two experiments have provided evidence that correct response activity plays an important role in reinforcement learning theory, and disruption of frontal striatal circuitry underlying reinforcement learn ing causes differential impairment s not only with respect to valence (i.e., correct versus error response positive versus negative feedback ), but also with respect to component abilities (i.e., error and feedback processing) contributing to learning. Because of the apparent complexity underly ing these systems, further research is needed in order to clarify the neurological and psychological factors that contribute to impairment s in reinforcement learning. Ideally, such research should employ tightly controlled experiments conducted with a larg e sample of patients closely matched on disease and demographic variables as well as medication regimens. A comparison of patients in different stages of disease (e.g., mild to moderate to severe) might allow for the discovery of informative relationships among demographic, disease, psychological and cognitive variables. Elaboration of comparisons currently being conducted on deep brain stimulation) on reinforcement learning could also prove useful for individualized treatment planning. Most importantly, continued research on the effects of valence on reinforcement learning in these populations will be crucial, not only for improved characterization of the functional significance of these ERP components, but also for determining how impairment s in reinforcement learning might be improved perhaps by using behavioral interventions focused on use of positive environmental feedback In light of recent reports that decisi on making impairments contribute to world

PAGE 129

129 functional implications of impaired reinforcement learning in everyday activities would be useful for aiding in the design of su ch interventions (Delazer et al., 2009)

PAGE 130

130 LIST OF REFERENCE S Abrahams, S., Goldstein, L. H., Simmons, A., Brammer, M. J., Williams, S. C., Giampietro, V. P., et al. (2003). Functional magnetic resonance imaging of verbal fluency and confrontation naming using compressed image acquisition to per mit overt responses. Hum Brain Mapp, 20 (1), 29 40. Alain, C., McNeely, H. E., He, Y., Christensen, B. K., & West, R. (2002). Neurophysiological evidence of error monitoring deficits in patients with schizophrenia. Cereb Cortex, 12 (8), 840 846. Alexander, G E., DeLong, M. R., & Strick, P. L. (1986). Parallel organization of functionally segregated circuits linking basal ganglia and cortex. Annu Rev Neurosci, 9 357 381. Amador, N., Schlag Rey, M., & Schlag, J. (2000). Reward predicting and reward detecting neuronal activity in the primate supplementary eye field. J Neurophysiol, 84 (4), 2166 2170. Amiez, C., Joseph, J. P., & Procyk, E. (2005). Anterior cingulate error related activity is modulated by predicted reward. Eur J Neurosci, 21 (12), 3447 3452. Anand, A., Li, Y., Wang, Y., Wu, J., Gao, S., Bukhari, L., et al. (2005). Activity and connectivity of brain mood regulating circuit in depression: a functional magnetic resonance study. Biol Psychiatry, 57 (10), 1079 1088. Anderson, A. K., Christoff, K., Stappen I., Panitz, D., Ghahremani, D. G., Glover, G., et al. (2003). Dissociated neural representations of intensity and valence in human olfaction. Nat Neurosci, 6 (2), 196 202. Band, G. P., & Kok, A. (2000). Age effects on response monitoring in a mental rotat ion task. Biol Psychol, 51 (2 3), 201 221. Band, G. P., Ridderinkhof, K. R., & Segalowitz, S. (2002). Explaining neurocognitive aging: is one factor enough? Brain Cogn, 49 (3), 259 267. Bartholow, B. D., Pearson, M. A., Dickter, C. L., Sher, K. J., Fabiani, M., & Gratton, G. (2005). Strategic control and medial frontal negativity: beyond errors and response conflict. Psychophysiology, 42 (1), 33 42. Beck, A. T. (1996). Beck Depression Inventory Second Edition (BDI II) USA: The Psychological Corporation. Bel lebaum, C., & Daum, I. (2008). Learning related changes in reward expectancy are reflected in the feedback related negativity. Eur J Neurosci, 27 (7), 1823 1835.

PAGE 131

131 Benamer, H. T., Patterson, J., Wyper, D. J., Hadley, D. M., Macphee, G. J., & Grosset, D. G. (2 000). Correlation of Parkinson's disease severity and duration with 123I FP CIT SPECT striatal uptake. Mov Disord, 15 (4), 692 698. Benton, A. L., & Hamsher, K. (1989). Multilingual Aphasia Examination Iowa City: AHA Associates. Berding, G., Brucke, T., Od in, P., Brooks, D. J., Kolbe, H., Gielow, P., et al. (2003). [[123I]beta CIT SPECT imaging of dopamine and serotonin transporters in Parkinson's disease and multiple system atrophy. Nuklearmedizin, 42 (1), 31 38. Bertrand, O., Perrin, F., & Pernier, J. (198 5). A theoretical justification of the average reference in topographic evoked potential studies. Electroencephalogr Clin Neurophysiol, 62 (6), 462 464. Blazer, D. (2002). Symptoms and signs. In Depression in late life (pp. 35 57). New York, NY: Springer. B odi, N., Keri, S., Nagy, H., Moustafa, A., Myers, C. E., Daw, N., et al. (2009). Reward learning and the novelty seeking personality: a between and within subjects study of the effects of dopamine agonists on young Parkinson's patients. Brain, 132 (Pt 9), 2385 2395. Bokura, H., Yamaguchi, S., & Kobayashi, S. (2001). Electrophysiological correlates for response inhibition in a Go/NoGo task. Clin Neurophysiol, 112 (12), 2224 2232. Botvinick, M. M., Cohen, J. D., & Carter, C. S. (2004). Conflict monitoring and anterior cingulate cortex: an update. Trends Cogn Sci, 8 (12), 539 546. Brown, J. W., & Braver, T. S. (2005). Learned predictions of error likelihood in the anterior cingulate cortex. Science, 307 (5712), 1118 1121. Bush, G., Luu, P., & Posner, M. I. (2000). Cognitive and emotional influences in anterior cingulate cortex. Trends Cogn Sci, 4 (6), 215 222. Chiu, P. H., & Deldin, P. J. (2007). Neural evidence for enhanced error detection in major depressive disorder. Am J Psychiatry, 164 (4), 608 616. Cohen, J. (1 992). A power primer. Psychol Bull, 112 (1), 155 159. Coles, M. G., Scheffers, M. K., & Holroyd, C. B. (2001). Why is there an ERN/Ne on correct trials? Response representations, stimulus related components, and the theory of error processing. Biol Psychol, 56 (3), 173 189. Cools, R. (2008). Role of dopamine in the motivational and cognitive control of behavior. Neuroscientist, 14 (4), 381 395.

PAGE 132

132 Cools, R., Altamirano, L., & D'Esposito, M. (2006). Reversal learning in Parkinson's disease depends on medication st atus and outcome valence. Neuropsychologia, 44 (10), 1663 1673. Cools, R., Barker, R. A., Sahakian, B. J., & Robbins, T. W. (2001). Mechanisms of cognitive set flexibility in Parkinson's disease. Brain, 124 (Pt 12), 2503 2512. Cools, R., Gibbs, S. E., Miyaka wa, A., Jagust, W., & D'Esposito, M. (2008). Working memory capacity predicts dopamine synthesis capacity in the human striatum. J Neurosci, 28 (5), 1208 1212. Cools, R., Lewis, S. J., Clark, L., Barker, R. A., & Robbins, T. W. (2007). L DOPA disrupts activ ity in the nucleus accumbens during reversal learning in Parkinson's disease. Neuropsychopharmacology, 32 (1), 180 189. Cools, R., Stefanova, E., Barker, R. A., Robbins, T. W., & Owen, A. M. (2002). Dopaminergic modulation of high level cognition in Parkins on's disease: the role of the prefrontal cortex revealed by PET. Brain, 125 (Pt 3), 584 594. Cooper, J. A., Sagar, H. J., Tidswell, P., & Jordan, N. (1994). Slowed central processing in simple and go/no go reaction time tasks in Parkinson's disease. Brain, 117 ( Pt 3) 517 529. Craig, A. H., Cummings, J. L., Fairbanks, L., Itti, L., Miller, B. L., Li, J., et al. (1996). Cerebral blood flow correlates of apathy in Alzheimer disease. Arch Neurol, 53 (11), 1116 1120. Cummings, J. L. (1992). Depression and Parkin son's disease: a review. Am J Psychiatry, 149 (4), 443 454. Cummings, J. L. (1997). The Neuropsychiatric Inventory: assessing psychopathology in dementia patients. Neurology, 48 (5 Suppl 6), S10 16. Cummings, J. L., & Masterman, D. L. (1999). Depression in p atients with Parkinson's disease. Int J Geriatr Psychiatry, 14 (9), 711 718. Czernecki, V., Pillon, B., Houeto, J. L., Pochon, J. B., Levy, R., & Dubois, B. (2002). Motivation, reward, and Parkinson's disease: influence of dopatherapy. Neuropsychologia, 40 ( 13), 2257 2267. Dagher, A., Owen, A. M., Boecker, H., & Brooks, D. J. (2001). The role of the striatum and hippocampus in planning: a PET activation study in Parkinson's disease. Brain, 124 (Pt 5), 1020 1032. Davidson, R. J. (1998). Anterior electrophysiolo gical asymmetries, emotion, and depression: conceptual and methodological conundrums. Psychophysiology, 35 (5), 607 614.

PAGE 133

133 Davidson, R. J., Jackson, D. C., & Larson, C. L. (2000). Human electroencephalography. In J. T. Cacioppo, L. G. Tassinary & G. G. Bernst on (Eds.), Handbook of psychophysiology (pp. 27 52). Cambridge, UK: Cambridge University Press. de Bruijn, E. R., Hulstijn, W., Verkes, R. J., Ruigt, G. S., & Sabbe, B. G. (2004). Drug induced stimulation and suppression of action monitoring in healthy vol unteers. Psychopharmacology (Berl), 177 (1 2), 151 160. Deakin, J., Aitken, M., Robbins, T., & Sahakian, B. J. (2004). Risk taking during decision making in normal volunteers changes with age. J Int Neuropsychol Soc, 10 (4), 590 598. Dehaene, S., Kerszberg, M., & Changeux, J. P. (1998). A neuronal model of a global workspace in effortful cognitive tasks. Proc Natl Acad Sci U S A, 95 (24), 14529 14534. Delazer, M., Sinz, H., Zamarian, L., Stockner, H., Seppi, K., Wenning, G. K., et al. (2009). Decision making u nder risk and under ambiguity in Parkinson's disease. Neuropsychologia, 47 (8 9), 1901 1908. Eppinger, B., Kray, J., Mock, B., & Mecklinger, A. (2008). Better or worse than expected? Aging, learning, and the ERN. Neuropsychologia, 46 (2), 521 539. Euteneuer, F., Schaefer, F., Stuermer, R., Boucsein, W., Timmermann, L., Barbe, M. T., et al. (2009). Dissociation of decision making under ambiguity and decision making under risk in patients with Parkinson's disease: a neuropsychological and psychophysiological st udy. Neuropsychologia, 47 (13), 2882 2890. Fabiani, M., Gratton, G., & Coles, M. G. H. (2000). Event related brain potentials: Methods, theory, and applications. In J. T. Cacioppo, L. G. Tassinary & G. G. Bernston (Eds.), Handbook of Psychophysiology (pp. 5 3 84). Cambridge, UK: Cambridge University Press. Fahn, S. (2003). Description of Parkinson's disease as a clinical syndrome. Ann N Y Acad Sci, 991 1 14. Fahn, S., & Elton, R. L. (1987). Unified Parkinson's Disease Rating Scale. In C. D. M. S. Fahn, M. Go ldstein, & D.B. Calne (Ed.), Recent developments in Parkinson's disease (Vol. 2, pp. 153 163). Florham Park, NJ: MacMillan Healthcare Information. Falkenstein, M., Hielscher, H., Dziobek, I., Schwarzenau, P., Hoormann, J., Sunderman, B., et al. (2001). Act ion monitoring, error detection, and the basal ganglia: an ERP study. Neuroreport, 12 (1), 157 161.

PAGE 134

134 Falkenstein, M., Hohnsbein, J., Hoormann, J., & Blanke, L. (1991). Effects of crossmodal divided attention on late ERP components. II. Error processing in ch oice reaction tasks. Electroencephalogr Clin Neurophysiol, 78 (6), 447 455. Falkenstein, M., Hoormann, J., Christ, S., & Hohnsbein, J. (2000). ERP components on reaction errors and their functional significance: a tutorial. Biol Psychol, 51 (2 3), 87 107. Fa ul, F. E., E., Lang, A.G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Meth (39), 175 191. Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). "Mini ment al state". A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res, 12 (3), 189 198. Ford, J. M., Roth, W. T., Menon, V., & Pfefferbaum, A. (1999). Failures of automatic and strategic processing in schizophrenia: co mparisons of event related brain potential and startle blink modification. Schizophr Res, 37 (2), 149 163. Frank, G. K., Bailer, U. F., Henry, S. E., Drevets, W., Meltzer, C. C., Price, J. C., et al. (2005). Increased dopamine D2/D3 receptor binding after r ecovery from anorexia nervosa measured by positron emission tomography and [11c]raclopride. Biol Psychiatry, 58 (11), 908 912. Frank, M. J., & Claus, E. D. (2006). Anatomy of a decision: striato orbitofrontal interactions in reinforcement learning, decision making, and reversal. Psychol Rev, 113 (2), 300 326. Frank, M. J., Seeberger, L. C., & O'Reilly R, C. (2004). By carrot or by stick: cognitive reinforcement learning in parkinsonism. Science, 306 (5703), 1940 1943. Gehring, W. J., Goss, B., Coles, M. G. H., Meyer, D. E., & Donchin, E. (1993). A neural system for error detection and compensation. Pscyhological Science, 4 385 390. Gehring, W. J., & Knight, R. T. (2000). Prefrontal cingulate interactions in action monitoring. Nat Neurosci, 3 (5), 516 520. Gehri ng, W. J., & Willoughby, A. R. (2002). The medial frontal cortex and the rapid processing of monetary gains and losses. Science, 295 (5563), 2279 2282. Gibson, E. L. (2006). Emotional influences on food choice: sensory, physiological and psychological pathw ays. Physiol Behav, 89 (1), 53 61. Golden, C. J. (1978). Stroop Color and Word Test Chicago: Stoelting. Goldman Rakic, P. S., Bates, J. F., & Chafee, M. V. (1992). The prefrontal cortex and internally generated motor acts. Curr Opin Neurobiol, 2 (6), 830 83 5.

PAGE 135

135 Gotham, A. M., Brown, R. G., & Marsden, C. D. (1988). 'Frontal' cognitive function in patients with Parkinson's disease 'on' and 'off' levodopa. Brain, 111 ( Pt 2) 299 321. Greenhouse, S. W., & Geisser, S. (1959). On the methods in the analysis of prof ile data. Psychometrika, 24 95 112. Greenwood, P. M. (2000). The frontal aging hypothesis evaluated. J Int Neuropsychol Soc, 6 (6), 705 726. Hajcak, G., Holroyd, C. B., Moser, J. S., & Simons, R. F. (2005). Brain potentials associated with expected and une xpected good and bad outcomes. Psychophysiology, 42 (2), 161 170. Hajcak, G., Moser, J. S., Holroyd, C. B., & Simons, R. F. (2007). It's worse than you thought: the feedback negativity and violations of reward prediction in gambling tasks. Psychophysiology, 44 (6), 905 912. Heaton, R. K., Chelune, G. J., Talley, J. L., Kay, G. G., & Curtiss, G. (1993). Wisconsin Card Sorting Test. Manual revised and expanded. Odessa, FL: Psychological Assessment Resources. Holroyd, C. B., & Coles, M. G. (2002). The neural ba sis of human error processing: reinforcement learning, dopamine, and the error related negativity. Psychol Rev, 109 (4), 679 709. Holroyd, C. B., & Coles, M. G. (2008). Dorsal anterior cingulate cortex integrates reinforcement history to guide voluntary beh avior. Cortex, 44 (5), 548 559. Holroyd, C. B., & Krigolson, O. E. (2007). Reward prediction error signals associated with a modified time estimation task. Psychophysiology, 44 (6), 913 917. Holroyd, C. B., Larsen, J. T., & Cohen, J. D. (2004). Context depen dence of the event related brain potential associated with reward and punishment. Psychophysiology, 41 (2), 245 253. Holroyd, C. B., Nieuwenhuis, S., Yeung, N., & Cohen, J. D. (2003). Errors in reward prediction are reflected in the event related brain pote ntial. Neuroreport, 14 (18), 2481 2484. Holroyd, C. B., Nieuwenhuis, S., Yeung, N., Nystrom, L., Mars, R. B., Coles, M. G., et al. (2004). Dorsal anterior cingulate cortex shows fMRI response to internal and external error signals. Nat Neurosci, 7 (5), 497 4 98. Holroyd, C. B., Praamstra, P., Plat, E., & Coles, M. G. (2002). Spared error related potentials in mild to moderate Parkinson's disease. Neuropsychologia, 40 (12), 2116 2124.

PAGE 136

136 Holroyd, C. B., Yeung, N., Coles, M. G., & Cohen, J. D. (2005). A mechanism fo r error detection in speeded response time tasks. J Exp Psychol Gen, 134 (2), 163 191. Hughes, A. J., Ben Shlomo, Y., Daniel, S. E., & Lees, A. J. (1992). What features improve the accuracy of clinical diagnosis in Parkinson's disease: a clinicopathologic s tudy. Neurology, 42 (6), 1142 1146. Ille, N., Berg, P., & Scherg, M. (2002). Artifact correction of the ongoing EEG using spatial filters based on artifact and brain signal topographies. J Clin Neurophysiol, 19 (2), 113 124. Isella, V., Melzi, P., Grimaldi, M., Iurlaro, S., Piolti, R., Ferrarese, C., et al. (2002). Clinical, neuropsychological, and morphometric correlates of apathy in Parkinson's disease. Mov Disord, 17 (2), 366 371. Ito, J., & Kitagawa, J. (2006). Performance monitoring and error processing d uring a lexical decision task in patients with Parkinson's disease. J Geriatr Psychiatry Neurol, 19 (1), 46 54. Juckel, G., Schlagenhauf, F., Koslowski, M., Wustenberg, T., Villringer, A., Knutson, B., et al. (2006). Dysfunction of ventral striatal reward p rediction in schizophrenia. Neuroimage, 29 (2), 409 416. Keitz, M., Koerts, J., Kortekaas, R., Renken, R., de Jong, B. M., & Leenders, K. L. (2008). Prefrontal cortex and striatal activation by feedback in Parkinson's disease. Brain Res, 1236 225 233. Kern s, J. G., Cohen, J. D., MacDonald, A. W., 3rd, Cho, R. Y., Stenger, V. A., & Carter, C. S. (2004). Anterior cingulate conflict monitoring and adjustments in control. Science, 303 (5660), 1023 1026. Kirsch Darrow, L., Fernandez, H. H., Marsiske, M., Okun, M. S., & Bowers, D. (2006). Dissociating apathy and depression in Parkinson disease. Neurology, 67 (1), 33 38. Knutson, B., & Cooper, J. C. (2005). Functional magnetic resonance imaging of reward prediction. Curr Opin Neurol, 18 (4), 411 417. Knutson, B., Fong G. W., Adams, C. M., Varner, J. L., & Hommer, D. (2001). Dissociation of reward anticipation and outcome with event related fMRI. Neuroreport, 12 (17), 3683 3687. Knutson, B., Taylor, J., Kaufman, M., Peterson, R., & Glover, G. (2005). Distributed neural representation of expected value. J Neurosci, 25 (19), 4806 4812. Knutson, B., Westdorp, A., Kaiser, E., & Hommer, D. (2000). FMRI visualization of brain activity during a monetary incentive delay task. Neuroimage, 12 (1), 20 27.

PAGE 137

137 Kobayakawa, M., Koyama, S., Mimura, M., & Kawamura, M. (2008). Decision making in Parkinson's disease: Analysis of behavioral and physiological patterns in the Iowa gambling task. Mov Disord, 23 (4), 547 552. Kopp, B., & Rist, F. (1999). An event related brain potential substrate of d isturbed response monitoring in paranoid schizophrenic patients. J Abnorm Psychol, 108 (2), 337 346. Kunig, G., Leenders, K. L., Martin Solch, C., Missimer, J., Magyar, S., & Schultz, W. (2000). Reduced reward processing in the brains of Parkinsonian patien ts. Neuroreport, 11 (17), 3681 3687. Larson, M. J., Kelly, K. G., Stigge Kaufman, D. A., Schmalfuss, I. M., & Perlstein, W. M. (2007). Reward context sensitivity impairment following severe TBI: an event related potential investigation. J Int Neuropsychol S oc, 13 (4), 615 625. Lezak, M. D., Howieson, D. B., & Loring, D. W. (2004). Neuropsychological Assessment (4th Edition ed.). New York: Oxford Press. Liu, X., Powell, D. K., Wang, H., Gold, B. T., Corbly, C. R., & Joseph, J. E. (2007). Functional dissociatio n in frontal and striatal areas for processing of positive and negative reward information. J Neurosci, 27 (17), 4587 4597. Luck, S. J. (2005). Ten simple rules for designing ERP experiments. In T. C. Handy (Ed.), Event related potentials: A methods handboo k. (pp. 17 32). Cambridge, MA: Massachusetts Institute of Technology. MacPherson, S. E., Phillips, L. H., & Della Sala, S. (2002). Age, executive function, and social decision making: a dorsolateral prefrontal theory of cognitive aging. Psychol Aging, 17 (4 ), 598 609. Marin, R. S. (1991). Apathy: a neuropsychiatric syndrome. J Neuropsychiatry Clin Neurosci, 3 (3), 243 254. Marschner, A., Mell, T., Wartenburger, I., Villringer, A., Reischies, F. M., & Heekeren, H. R. (2005). Reward based decision making and ag ing. Brain Res Bull, 67 (5), 382 390. Mathalon, D. H., Fedor, M., Faustman, W. O., Gray, M., Askari, N., & Ford, J. M. (2002). Response monitoring dysfunction in schizophrenia: an event related brain potential study. J Abnorm Psychol, 111 (1), 22 41. Mather, M., & Carstensen, L. L. (2005). Aging and motivated cognition: the positivity effect in attention and memory. Trends Cogn Sci, 9 (10), 496 502. Mathewson, K. J., Dywan, J., & Segalowitz, S. J. (2005). Brain bases of error related ERPs as influenced by age and task. Biol Psychol, 70 (2), 88 104.

PAGE 138

138 Mattay, V. S., Tessitore, A., Callicott, J. H., Bertolino, A., Goldberg, T. E., Chase, T. N., et al. (2002). Dopaminergic modulation of cortical function in patients with Parkinson's disease. Ann Neurol, 51 (2), 156 16 4. Mattis, S. (1988). Dementia Rating Scale (DRS) Odessa, FL: Psychological Assessment Resources. Mattox, S. T., Valle Inclan, F., & Hackley, S. A. (2006). Psychophysiological evidence for impaired reward anticipation in Parkinson's disease. Clin Neurophy siol, 117 (10), 2144 2153. Mayberg, H. S., Starkstein, S. E., Sadzot, B., Preziosi, T., Andrezejewski, P. L., Dannals, R. F., et al. (1990). Selective hypometabolism in the inferior frontal lobe in depressed patients with Parkinson's disease. Ann Neurol, 28 (1), 57 64. Mayeux, R., Stern, Y., Cote, L., & Williams, J. B. (1984). Altered serotonin metabolism in depressed patients with parkinson's disease. Neurology, 34 (5), 642 646. McDonald, W. M., Richard, I. H., & DeLong, M. R. (2003). Prevalence, etiology, an d treatment of depression in Parkinson's disease. Biol Psychiatry, 54 (3), 363 375. Mell, T., Heekeren, H. R., Marschner, A., Wartenburger, I., Villringer, A., & Reischies, F. M. (2005). Effect of aging on stimulus reward association learning. Neuropsycholo gia, 43 (4), 554 563. Middleton, F. A., & Strick, P. L. (2000). Basal ganglia and cerebellar loops: motor and cognitive circuits. Brain Res Brain Res Rev, 31 (2 3), 236 250. Migneco, O., Benoit, M., Koulibaly, P. M., Dygai, I., Bertogliati, C., Desvignes, P. et al. (2001). Perfusion brain SPECT and statistical parametric mapping analysis indicate that apathy is a cingulate syndrome: a study in Alzheimer's disease and nondemented patients. Neuroimage, 13 (5), 896 902. Miltner, W. H., Braun, C. H., & Coles, M. G. H. (1997). Event related brain potentials following incorrect feedback in a time estimation task: Evidence for a "generic" neural system for error detection. Journal of Cognitive Neuroscience, 9 (6), 788 798. Morris, S. E., Heerey, E. A., Gold, J. M., & Holroyd, C. B. (2008). Learning related changes in brain activity following errors and performance feedback in schizophrenia. Schizophr Res, 99 (1 3), 274 285. Morris, S. E., Yee, C. M., & Nuechterlein, K. H. (2006). Electrophysiological analysis of error m onitoring in schizophrenia. J Abnorm Psychol, 115 (2), 239 250. Naranjo, C. A., Tremblay, L. K., & Busto, U. E. (2001). The role of the brain reward system in depression. Prog Neuropsychopharmacol Biol Psychiatry, 25 (4), 781 823.

PAGE 139

139 Nieuwenhuis, S., Holroyd, C B., Mol, N., & Coles, M. G. (2004). Reinforcement related brain potentials from medial frontal cortex: origins and functional significance. Neurosci Biobehav Rev, 28 (4), 441 448. Nieuwenhuis, S., Ridderinkhof, K. R., Talsma, D., Coles, M. G., Holroyd, C. B., Kok, A., et al. (2002). A computational account of altered error processing in older age: dopamine and the error related negativity. Cogn Affect Behav Neurosci, 2 (1), 19 36. Nieuwenhuis, S., Slagter, H. A., von Geusau, N. J., Heslenfeld, D. J., & Holr oyd, C. B. (2005). Knowing good from bad: differential activation of human cortical areas by positive and negative outcomes. Eur J Neurosci, 21 (11), 3161 3168. Norman, S., Troster, A. I., Fields, J. A., & Brooks, R. (2002). Effects of depression and Parkin son's disease on cognitive functioning. J Neuropsychiatry Clin Neurosci, 14 (1), 31 36. O'Doherty, J., Kringelbach, M. L., Rolls, E. T., Hornak, J., & Andrews, C. (2001). Abstract reward and punishment representations in the human orbitofrontal cortex. Nat Neurosci, 4 (1), 95 102. O'Sullivan, M., Barrick, T. R., Morris, R. G., Clark, C. A., & Markus, H. S. (2005). Damage within a network of white matter regions underlies executive dysfunction in CADASIL. Neurology, 65 (10), 1584 1590. O'Sullivan, M., Jones, D. K., Summers, P. E., Morris, R. G., Williams, S. C., & Markus, H. S. (2001). Evidence for cortical "disconnection" as a mechanism of age related cognitive decline. Neurology, 57 (4), 632 638. Oliveira, F. T., McDonald, J. J., & Goodman, D. (2007). Performan ce monitoring in the anterior cingulate is not all error related: expectancy deviation and the representation of action outcome associations. J Cogn Neurosci, 19 (12), 1994 2004. Otten, L. J., & Rugg, M. D. (2005). Interpreting event related potentials. In T. C. Handy (Ed.), Event related potentials: A methods handbook Cambridge, MA: The MIT Press. Owen, A. M. (2004). Cognitive dysfunction in Parkinson's disease: the role of frontostriatal circuitry. Neuroscientist, 10 (6), 525 537. Owen, A. M., Iddon, J. L. Hodges, J. R., Summers, B. A., & Robbins, T. W. (1997). Spatial and non spatial working memory at different stages of Parkinson's disease. Neuropsychologia, 35 (4), 519 532. Pagonabarraga, J., Garcia Sanchez, C., Llebaria, G., Pascual Sedano, B., Gironell A., & Kulisevsky, J. (2007). Controlled study of decision making and cognitive impairment in Parkinson's disease. Mov Disord, 22 (10), 1430 1435.

PAGE 140

140 Paus, T. (2001). Primate anterior cingulate cortex: where motor control, drive and cognition interface. Nat R ev Neurosci, 2 (6), 417 424. Pedersen, K. F., Alves, G., Bronnick, K., Aarsland, D., Tysnes, O. B., & Larsen, J. P. Apathy in drug naive patients with incident Parkinson's disease: the Norwegian ParkWest study. J Neurol, 257 (2), 217 223. Pietschmann, M., Si mon, K., Endrass, T., & Kathmann, N. (2008). Changes of performance monitoring with learning in older and younger adults. Psychophysiology, 45 (4), 559 568. Pluck, G. C., & Brown, R. G. (2002). Apathy in Parkinson's disease. J Neurol Neurosurg Psychiatry, 7 3 (6), 636 642. Przuntek, H., Muller, T., & Riederer, P. (2004). Diagnostic staging of Parkinson's disease: conceptual aspects. J Neural Transm, 111 (2), 201 216. Rabbitt, P., Scott, M., Thacker, N., Lowe, C., Jackson, A., Horan, M., et al. (2006). Losses in gross brain volume and cerebral blood flow account for age related differences in speed but not in fluid intelligence. Neuropsychology, 20 (5), 549 557. Raz, N., Rodrigue, K. M., Head, D., Kennedy, K. M., & Acker, J. D. (2004). Differential aging of the me dial temporal lobe: a study of a five year change. Neurology, 62 (3), 433 438. Reitan, R. M., & Wolfson, D. (1995). Category Test and Trail Making Test as measures of frontal lobe functions. The Clinical Neuropsychologist, 9 50 56. Ridderinkhof, K. R., Nie uwenhuis, S., & Bashore, T. R. (2003). Errors are foreshadowed in brain potentials associated with action monitoring in cingulate cortex in humans. Neurosci Lett, 348 (1), 1 4. Ridderinkhof, K. R., Span, M. M., & van der Molen, M. W. (2002). Perseverative b ehavior and adaptive control in older adults: performance monitoring, rule induction, and set shifting. Brain Cogn, 49 (3), 382 401. Ring, H. A., Bench, C. J., Trimble, M. R., Brooks, D. J., Frackowiak, R. S., & Dolan, R. J. (1994). Depression in Parkinson' s disease. A positron emission study. Br J Psychiatry, 165 (3), 333 339. Rolls, E. T. (2000). The orbitofrontal cortex and reward. Cereb Cortex, 10 (3), 284 294. Rosen, H. J., Allison, S. C., Schauer, G. F., Gorno Tempini, M. L., Weiner, M. W., & Miller, B. L. (2005). Neuroanatomical correlates of behavioural disorders in dementia. Brain, 128 (Pt 11), 2612 2625.

PAGE 141

141 Samanez Larkin, G. R., Gibbs, S. E., Khanna, K., Nielsen, L., Carstensen, L. L., & Knutson, B. (2007). Anticipation of monetary gain but not loss in h ealthy older adults. Nat Neurosci, 10 (6), 787 791. Schmitt Eliassen, J., Ferstl, R., Wiesner, C., Deuschl, G., & Witt, K. (2007). Feedback based versus observational classification learning in healthy aging and Parkinson's disease. Brain Res, 1142 178 188 Schott, B. H., Niehaus, L., Wittmann, B. C., Schutze, H., Seidenbecher, C. I., Heinze, H. J., et al. (2007). Ageing and early stage Parkinson's disease affect separable neural mechanisms of mesolimbic reward processing. Brain, 130 (Pt 9), 2412 2424. Schul tz, W. (2002). Getting formal with dopamine and reward. Neuron, 36 (2), 241 263. Schultz, W., Tremblay, L., & Hollerman, J. R. (2000). Reward processing in primate orbitofrontal cortex and basal ganglia. Cereb Cortex, 10 (3), 272 284. Shenkin, S. D., Bastin, M. E., Macgillivray, T. J., Deary, I. J., Starr, J. M., Rivers, C. S., et al. (2005). Cognitive correlates of cerebral white matter lesions and water diffusion tensor parameters in community dwelling older people. Cerebrovasc Dis, 20 (5), 310 318. Shidara, M., & Richmond, B. J. (2002). Anterior cingulate: single neuronal signals related to degree of reward expectancy. Science, 296 (5573), 1709 1711. Shohamy, D., Myers, C. E., Grossman, S., Sage, J., & Gluck, M. A. (2005). The role of dopamine in cognitive se quence learning: evidence from Parkinson's disease. Behav Brain Res, 156 (2), 191 199. Snodgrass, J. G., & Vanderwart, M. (1980). A standardized set of 260 pictures: Norms for name agreement, image agreement, familiarity, and visual complexity. Journal of E xperimental Psychology: Human Learning and Memory, 6 174 215. Sockeel, P., Dujardin, K., Devos, D., Deneve, C., Destee, A., & Defebvre, L. (2006). The Lille apathy rating scale (LARS), a new instrument for detecting and quantifying apathy: validation in P arkinson's disease. J Neurol Neurosurg Psychiatry, 77 (5), 579 584. Speilberger, C. D., Gorusch, R. L., Lushene, R., Vagg, P. R., & Jacobs, G. A. (1983). Manual for the State Trait Anxiety Inventory Palo Alto, CA: Consulting Psychologists Press. Starkstein S. E., Mayberg, H. S., Preziosi, T. J., Andrezejewski, P., Leiguar da, R., & Robinson, R. G. (1992 ). Reliability, validity, and clinical correlates of apathy in Parkinson's disease. The Journal of Neuropsychiatry and Clinical Neurosciences, 4 (2), 134 139.

PAGE 142

142 Stemmer, B., Segalowitz, S. J., Dywan, J., Panisset, M., & Melmed, C. (2007). The error negativity in nonmedicated and medicated patients with Parkinson's disease. Clin Neurophysiol, 118 (6), 1223 1229. Swainson, R., SenGupta, D., Shetty, T., Watkins, L. H ., Summers, B. A., Sahakian, B. J., et al. (2006). Impaired dimensional selection but intact use of reward feedback during visual discrimination learning in Parkinson's disease. Neuropsychologia, 44 (8), 1290 1304. Tekin, S., & Cummings, J. L. (2002). Front al subcortical neuronal circuits and clinical neuropsychiatry: an update. J Psychosom Res, 53 (2), 647 654. Themanson, J. R., Hillman, C. H., & Curtin, J. J. (2006). Age and physical activity influences on action monitoring during task switching. Neurobiol Aging, 27 (9), 1335 1345. Troster, A. I., Paolo, A. M., Lyons, K. E., Glatt, S. L., Hubble, J. P., & Koller, W. C. (1995). The influence of depression on cognition in Parkinson's disease: a pattern of impairment distinguishable from Alzheimer's disease. Neu rology, 45 (4), 672 676. Tucker, D. M., Luu, P., Frishkoff, G., Quiring, J., & Poulsen, C. (2003). Frontolimbic response to negative feedback in clinical depression. J Abnorm Psychol, 112 (4), 667 678. Veiel, H. O. (1997). A preliminary profile of neuropsych ological deficits associated with major depression. J Clin Exp Neuropsychol, 19 (4), 587 603. Wang, G. J., Volkow, N. D., Logan, J., Fowler, J. S., Schlyer, D., MacGregor, R. R., et al. (1995). Evaluation of age related changes in serotonin 5 HT2 and dopami ne D2 receptor availability in healthy human subjects. Life Sci, 56 (14), PL249 253. Wechsler, D. (1997). Wechsler Adult Intelligence Scale Third Edition manual New York: The Psychological Corporation. West, R. (2000). In defense of the frontal lobe hypo thesis of cognitive aging. J Int Neuropsychol Soc, 6 (6), 727 729; discussion 730. Willemssen, R., Muller, T., Schwarz, M., Hohnsbein, J., & Falkenstein, M. (2008). Error processing in patients with Parkinson's disease: the influence of medication state. J Neural Transm, 115 (3), 461 468. Woodruff Pak, D. S. (1997). Normal aging of the brain. In The Neuropsychology of Aging (pp. 86 108). Malden, MA: Blackwell. Yesavage, J. A., Brink, T. L., Rose, T. L., Lum, O., Huang, V., Adey, M., et al. (1983). Development and validation of a Geriatric Depression Screening Scale: A preliminary report. Journal of Psychiatric Research, 17 (1), 37 49.

PAGE 143

143 Yeung, N., Botvinick, M. M., & Cohen, J. D. (2004). The neural basis of error detection: conflict monitoring and the error related negativity. Psychol Rev, 111 (4), 931 959. 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 fro ntostriatal circuitry. Cogn Behav Neurol, 16 (4), 193 210. Zgaljardic, D. J., Borod, J. C., Foldi, N. S., Mattis, P. J., Gordon, M. F., Feigin, A., et al. (2006). An examination of executive dysfunction associated with frontostriatal circuitry in Parkinson' s disease. J Clin Exp Neuropsychol, 28 (7), 1127 1144.

PAGE 144

144 BIOGRAPHICAL SKETCH Ida Lillian Kellison completed a Bachelor of Arts degree from the University of Iowa in Iowa City, graduating with distinction majoring in psychology, philosoph y, and religious studies Aft er working as a research coordinator in the Neurology Clinic at the University of Iowa Hospitals and Clinics, she began her doctoral studies in clinical psychology at the University of Florida. She earned a Master of Science degree in 2006, and a Ph.D. in clinical psychology with specialization in neuropsychology in 2010, following completion of her internship at the Boston Consortium.