1 NEURAL CORRELATES OF SEMANTIC INHIBITION IN OLDER AND YOUNGER ADULTS By ILANA FAWN LEVY 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 20 1 2
2 201 2 Ilana Fawn Levy
3 To my father Michael Jonathan Levy for the strength and balance that comes from tou ch during your life
4 ACKNOWLEDGMENTS I would like to thank Dr. Bruce Crosson for his guidance and mentorship, my dissertation committee (Drs. Catherine Price, Jamie Reilly, Tom Kerkhoff) for their intellectual contributions Stephen Towler for answerin g incessant questions at all hours of the day and night, and Viki Solt without whom I would not have succeeded in navigating the numerous logistical hurdles of graduate school. I would also like to thank my mother, Marcia Levy for the neve r ending encour agement and love; my step father Sheldon Richter for the steady support and willingne ss to help at the drop of a hat; my siblings Melissa Levy, Jennifer Richter, Byron Levy and Michael Richter, f or inspiration and distractions; my grandparents Ruby and Aaron Eisenbach and Diana and Morris Levy for their unwavering belief in my abilities; and Ohad Paris for keeping me laughing and well fed on even the most stressful days.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS ................................ ................................ ............................. 9 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 12 Changes in Frontal Lobe Function in Aging and Pathology ................................ .... 12 The Role of the Basal Ganglia in Complex Language Functions ............................ 15 Basal Ganglia Loops ................................ ................................ ........................ 16 Evidence of Complex Language Problems in Basal Ganglia Disorders ........... 17 Inhibitory Mechanisms in Language ................................ ................................ ....... 20 The Hayling Test ................................ ................................ .............................. 21 Use of Neuroimaging as a Measure of Inhibition ................................ .............. 23 2 HYPOTHESES ................................ ................................ ................................ ....... 26 Initiation versus Inhibition Hypothesis ................................ ................................ ..... 27 Older versus Younger Hypothesis ................................ ................................ .......... 27 Relationship Between f MRI and Behavioral Performance ................................ ...... 28 3 METHODS ................................ ................................ ................................ .............. 31 Participants ................................ ................................ ................................ ............. 31 Mate rials and Procedures ................................ ................................ ....................... 32 Dementia Screening ................................ ................................ ......................... 32 fMRI Task Design ................................ ................................ ............................. 32 B rain Image Acquisition ................................ ................................ .................... 36 Behavioral Data Analysis ................................ ................................ .................. 36 FMRI Data Analysis ................................ ................................ .......................... 37 4 RESULTS ................................ ................................ ................................ ............... 42 Behavioral Results ................................ ................................ ................................ .. 42 FMRI Results ................................ ................................ ................................ .......... 44 Effect of Task (Task vs. Rest) ................................ ................................ .......... 44 Unique Effects of Inhibition (Task B minus Task A) by Group .......................... 45
6 Effect of Age (Old vs. Young) by Task ................................ ............................. 45 Effect of Age (Old vs. Young) on Inhibition (Voxel by Voxel B A) ..................... 46 Correlational Analysis (Brain vs. Behavior) ................................ ...................... 46 5 DISCUSSION ................................ ................................ ................................ ......... 57 Effects of Age on Inhibition ................................ ................................ ..................... 57 Strengths and Weaknesses ................................ ................................ .................... 61 APPENDIX A SENTENCE STEMS INCLUDED IN TASK ................................ ............................. 65 B INSTRUCTIONS FOR SENTENCE COMPLETION TASK ................................ ..... 5 9 Outside Scanner Instructions ................................ ................................ .................. 59 Inside Scanner Instructions ................................ ................................ ..................... 60 C M ONTAGES OF FMRI CLUSTER ACTIVITY WITH ASSOC IATED AVERAGE HRFS ................................ ................................ ................................ ...................... 61 Effect of Task (Task vs. Baseline) ................................ ................................ .......... 61 Effect of Inhibition (Task B vs. Task A) for Young (none for Old) ............................ 63 Effect of Age (Old vs. Young) ................................ ................................ ................. 65 LIST OF REFERENCES ................................ ................................ ............................... 71 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 75
7 LIST OF TABLES Table page 3 1 Participant Demographics ................................ ................................ .................. 40 3 2 Error characte rater classification guide (1996) ................................ ................................ .......... 40 4 1 Average group behavioral performance by task. ................................ ................ 50 4 2 classification guide (1996). ................................ ................................ ................. 51 4 3 Effect of Task: Cluster locations by group based on one sam ple t tests ............ 51 4 4 Effect of Inhibition (task B task A): Cluster locations for Young group based on paired sample t tests ................................ ................................ ..................... 52 4 5 E ffect of Age (Old Young) during Task A: Cluster locations by task based on independent samples t tests ................................ ................................ ............... 53 4 6 Effect of Age (Old Young) during Task B: Cluster locations by task based on indepen dent samples t tests ................................ ................................ ............... 54 4 7 Age differences in activity associated with inhibition based on independent samples t tests ................................ ................................ ................................ ... 56
8 LIST OF FIGURES Figure page 1 1 Cartoon of Basal Ganglia loops ................................ ................................ .......... 25 2 1 Model of Cognitive Processing Driving Activity in fMRI Voxels. .......................... 30 3 1 Schematic of task design ................................ ................................ .................... 41 4 1 Mean r eaction times by group and task ................................ .............................. 48 4 2 Mean behavioral performance by group and task. ................................ .............. 48 4 3 Average RT by group plotted across time ................................ .......................... 49 4 4 Correlations between average AUC a ctivity for older participants during task B and RT fo r correct responses in task B ................................ ........................... 50 4 5 Effect of Inhibition (task B task A) for young group based on paired sample t tests ................................ ................................ ................................ .................... 52 4 6 Effect of age (Old Young) on task A based on independent sample t tests. ...... 53 4 7 Effect of age (Old Young) on task B base d on independent samp le t tests ....... 54 4 8 Effect of inhibition (B A) compared between groups based on independent sample t tests ................................ ................................ ................................ ..... 55 4 9 Age differences in acti vity associated with inhibition ................................ .......... 55
9 LIST OF ABBREVIATION S AC C Anterior Cingulate Cortex aMCI Amnestic Mild Cognitive Impairment AUC Area Under the Curve BOLD Blood Oxygenation Level D ependent DA Dopamine DBS Deep Brai n Stimulation DLPFC Dorsolateral Prefrontal Cortex dPM D orsal P remotor C ortex fMRI F unctional M agnetic R esonance I maging HAROLD H emispheric S ymmetry R eduction in Older A dults IFG I nferior frontal gyrus IFIS Integrated Functional Imaging System M1 Primar y Motor Cortex MMSE Mini Mental State Exam MNI Montreal Neurological Institute PD PET Positron Emission Tomography Pre SMA Pre supplementary Motor Area ROI Region of I nterest RT Reaction Time rTMS R epetitive T ranscranial M agnetic S timu lation SD Standard Deviation SOAs Stimulus Onset Asynchronies STN Subthalamic Nucleus
10 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Docto r of Philosophy NEURAL CORRELATES OF SEMANTIC INHIBITION IN OLDER AND YOUNGER ADULTS By Ilana Fawn Levy May 201 2 Chair: Bruce Crosson Major: Psychology Neuroimaging and behavioral studies have indicated frontal cortical areas and the basal ganglia as playing key roles in modulating inhibitory processes (Castner et al., 2007) This study examines inh ibitory language processes and age related changes through use of a sentence completion task similar to the Hayling test (Burgess & Shallice, 1997) Old (n=15; Mean age=76) and young (n=15; mean age=23) healthy adults underwent a neuropsychological assessment and fMRI while performing a sentence completion task. Brain activity (i.e., fMRI) and behavior al responses (i.e., reaction time (RT), and response accuracy) while completing the sentence with an appropriate word (i.e. initiation task) was compared to performance when providing an unconnected response thought to require inhibition of the related wor ds (i.e. inhibition task). Analysis involved both within group (Initiation vs. Inhibition task) and between group (Old vs. Young) comparisons. Results of behavioral comparisons revealed that the old group performed significantly slower than the young g roup for the inhibition task although not for the initiation task and made significantly more errors than the young group during the
11 inhibition task demonstrating that the older participants had greater difficulty generating unrelated responses and inhibit ing responses related to the sentences as compared to the young group Brain imaging results revealed frontal (DLPFC, SFG, OFC), parietal (precuneus, angular gyrus) and temporal (SMG) regions significantly associated with inhibition in the young group bu t no significant task differences in the older group. Further analysis revealed significant age related differences on both tasks thought to be the result of a combination of brain related changes with age and differences in the way the groups were perfor ming the task. This pattern of performance suggests that the networks involved in inhibitory language tasks change with age though more research is needed to characterize these differences.
12 CHAPTER 1 INTRODUCTION Changes in Frontal Lobe Function in Agin g and Pathology As the cohort of older adults in our society increases and the number of individuals surviving into old age continues to rise, research in aging has flourished. Differences between older and younger adult brains have been a focus of signifi cant interest in the literature and led to further investigation as to what these differences may mean for behavioral performance and development across the lifespan. One area which has been functionally implicated across many cognitive domains is the fron tal lobes. The literature has continued to provide evidence for differences in frontal lobe function in older adults as compared to younger adults. Specifically the phenomenon of hemispheric asymmetry reduction in older adults (HAROLD) states that while pe rforming similar cognitive tasks, prefrontal activity tends to be less lateralized in older adults than in younger adults. The HAROLD model has been reported across domains of episodic memory, semantic memory, working memory, perception, and inhibitory con trol. In spite of the accumulation of evidence supporting this model, a debate remains as to whether these age related hemispheric asymmetry reductions serve a compensatory function or whether they reflect difficulty in recruiting specialized neural mechan isms, possibly due to a reduction in inhibition of unrelated areas, referred to as dedifferentiation ( Cabeza, 2002) Another important consideration in aging, is that although normal aging is not in itself pathological, there are aging related changes in brain structure and chemistry which bear some similarity to pathological conditions. For example, the
13 disease (PD), is also negatively affected in aging by what appears to be a significant and steady decline of brain dopamine levels with older age (Kaasinen & Rinne, 2002) Over the last decade there has been a steady progression in our understanding of the many roles of the frontal lobes. This headway has been aided significantly by studies that comprehensively examined patients with focal frontal lesions. This research has led to improved knowledge of the anatomical and behavioral specifi city within the frontal lobes (Stuss et al., 2002) The assumption is that individuals with damage to a particular region will show comparable impairment s if the damaged region is involved in a specific function, even if additional surrounding tissue is also affected. Major neuro anatomical areas have been differentiated through these studies including right and left dorsolateral frontal regions, and super ior medial and inferior medial frontal regions. Neuropsychological assessments have demonstrated that when each of these areas was damaged, impairments in similar cognitive processes resulted Specifically of interest to this study is that both the right d orsolateral frontal and the inferior medial frontal regions were found to be involved during tasks requiring response inhibition (Stuss et al., 2002) Additional evidence from functional magnetic resonance imaging (fMRI) and repetitive transcranial magnetic stimulation (rTMS) have implicated similar areas in the right hemisphere (Chambers et al., 2007; Garavan, Ross, Murphy, Roche, & Stein, 2002; Kelly et al., 2004) during the performance of frontal inhibitory tasks. Chambers and colleagues (2007) cleverly dissociated the role of the right dorsal premotor cortex (dPM) in the execution of a task from the role of the right inferior frontal gyrus (IFG) in response inhibit ion. The task involved a flanker/stop signal paradigm in which
14 participants were required to respond to the direction of a central arrow, flanked by either congruent or incongruent distracters, with the corresponding hand. The target was periodically inter rupted by a stop signal which required them to cancel their response. When this task was performed following rTMS of the IFG or dPM they found that stimulation of the right, but not left, dPM was found to facilitate RT performance for both hands in all con ditions regardless of the degree of competition involved. In contrast, stimulation of the right, but not left, IFG was associated with impairments in stop signal inhibition but only when competition existed during response selection (i.e. Incongruent trial s). These data seem to suggest that the right IFG is especially important for inhibition during the suppression of a competing response. It has also been suggested that different neuroanatomical networks may underlie the process of response inhibition dep ending on the difficulty or speed of the task. One model proposed that a right hemisphere network involving the dorsolateral prefrontal and right inferior parietal cortices may be involved during an easier task requiring slower response times. However, mor e activity in the superior medial region, including the anterior cingulate cortex (ACC ) and pre supplementary motor area (pre SMA), may be expected with increased response speed during a more difficult task (Garavan et al., 2002) This raises a question about the influence of the length of preparation time on the type of network recruited and the relation ship between controlled and automatic processing. To address this issue, another model was proposed using fMRI and controlling for stimulus presentation rate and behavioral performance as well as manipulating response stimulus intervals. Results demonstrat ed that during a faster go no go task there were significantly higher levels of activation in the right dorsolateral
15 prefrontal cortex (DLPFC), and left putamen, caudate, precuneus and fusiform gyrus. This may be due to the higher demand on response select ion processes with little preparation time. In contrast, the slower condition evidenced greater activity in the anterior right superior and middle frontal gyri and left inferior parietal cortex, perhaps representing a more deliberate, controlled response s election and inhibitory process associated with the increased preparation time (Kelly et al., 2004) Although task differences likely contributed to these conflicting results, the existence of two frontal networks, a right lateral frontal network and a right medial frontal network, seem apparent during motor inhibition tasks. The basal ganglia also seem to play an important role in these networks and will be discussed in more detail below. The Role of the Basal Ganglia in Complex Language Functions Most of the research on the basal ganglia has revolved around their role in movement. It is widely accepted that the basal ganglia play a subtle role in movement, not by driving the movement exactly, but by enhancing selected movements and suppressing competing movements. A similar concept can be applied to linguistic activities, such as word selection or synt ax production Evidence for the role of the basal ganglia in language is supported by the cortico basal ganglia thalamic loops that have been studied, the numerous connections found and proposed between the basal ganglia and frontal/pre frontal cortical re gions, the evidence of complex language deficits in individuals with basal ganglia damage, as well as increasing fMRI evidence of basal ganglia involvement in language tasks (Crosson, Benjamin, & Levy, 2007) area and the basal gan glia (Ford et al., unpublished).
16 Basal Ganglia Loops The basal ganglia form a complex group of structures deep within the cerebral hemispheres that are thought to influence movement and cognition through their interactions with various cortical and thalami c regions (Alexander, DeLong, & Strick, 1986; Middleton & Strick, 2000) In general, frontal striatal circuits originate in frontal cortical regions such as primary motor cortex, premotor cortex, supplementary and pre supplementary motor areas, anterior cingulate cortex, dorsolateral prefrontal cortex, and orbitofronta l cortex. Each circuit operates in a closed loop fashion, beginning and ending at the same target cortical region (Figure 1 1). Each basal ganglia circuit can be decomposed into three separate loops (Nambu, Tokuno, & Takada, 2002; Nambu et al., 2000) enh ance selected actions and cognitions (Gerfen, 1992; Mink, 1996; Penney & Young, 1986) compete with those selected. described (Nambu et al., 2002, 2000) In thi s loop, unlike the direct and indirect loops, the cortical component projects directly to its subthalamic nucleus (STN), bypassing the striatum. Action in the hyperdirect loop is thought to have a suppressing effect on actions and cognitions. Nambu and co lleagues (2000) stimulated primary motor cortex (M1) in macaque monkeys and me asured effects on the basal ganglia downstream in the medial globus pallidus. Although the exact time scale certainly does not apply to cognition, Crosson, Benjamin and Levy (2007) suggested the general sequence of events does. Activity in the medial globus pallidus indicates that after M1 stimulation there is an initial suppression of behavior/cognition followed by an intermediate enhancement of
17 behavior, finally followed by a late suppression of behavior. Nambu and colleagues (2002) suggested that the initial wave of suppression resets the system by allowing it to change ongoing behavior/cognitions to new ones. The intermediate wave of enhancement raises the probability of initiating selected behaviors/cognitions. Finally, the late wave of suppression decreases the probability of the occurrence of competing behaviors/cognitions. Additional neurotransmitter influences occur from dopaminergic projections via th e substantia nigra pars compacta to the striatal component of the frontal striatal circuitry. While dopaminergic actions are less straightforward than the actions of neurotransmitters elsewhere in the frontal striatal circuitry, dopamine is thought to infl uence both the direct and indirect basal ganglia pathways The rationale for outlining basal ganglia circuitry and neurotransmission is that a disease or injury at various points along this circuitry should display predictable neurocognitive deficits. Thu s, certain disease states or brain injuries should theoretically involvement of the direct basal ganglia circuitry, while others may produce dysfunction inuum via involvement of the hyperdirect and/or indirect circuitry and related functions. An understanding of basal ganglia pathways is important in the context of this study as a primary goal is to better characterize inhibitory phenomena in the normal ag ing process in the hopes that this knowledge will inform our understanding of potential neuropathological processes and likely consequences Evidence of Complex Language Problems in Basal Ganglia Disorders Semantic priming studies have commonly been used t o examine subtle differences in language process es among patient groups and as a way of better
18 understanding the time course of language processing in healthy subjects. The idea that a person should respond faster to a lexical decision about a target when that target is preceded by a semantically related word ( i.e., the prime), is based on the cognitive model of spread of activation in lexical and semantic network s In this model, at the time the prime is activated, this activity will also begin to spread t o nearby (i.e. related) nodes in the lexical and semantic network s It follows that if the target i s already somewhat activated a person should be able to respond to it faster than an unrelated word outside of the network that has not been primed (Chenery, Angwin, & Copland, 2008) The tricky part is determining the duration and timing of the effects of t he semantic network. Some very clever experiments have provided us with a way to examine these effects. In a study of direct (e.g., tiger stripes) and indirect (e.g., lion stripes) priming it was found that healthy controls showed faster RT to direct condi tions than indirect conditions, which intuitively makes sense since the indirect condition would involve a more mediation by an intermediate node in the semantic network. Priming effects (i.e. faster RT to related than unrelated targets) were found at a sh ort delay between prime and target as well as a longer delay. Interestingly, when healthy subjects were given L Dopa prior to participation, their priming effects differed systematically compared to controls (Angwin et al., 2004; Kischka et al., 1996) A lthough direct and indirect priming effects we re still found at the short delay between prime and target, priming effects disappeared more quickly than for controls not given L Dopa, at the longer delay. These data suggest that DA modulates aspects of semantic activation possibly by reducing the sprea d of activation, effectively focusing activation (Kischka et al., 1996) or by speeding up the rate of decay of semantic activation t hereby changing the time course of priming
19 effects (Angwin et al., 2004) In other words DA may facili tat e selection of the salient stimulus initially, but due to the early facilitation, the spread of activation diminishes earlier than in the controls not administered L Dopa. These results provide valuable information about the effects of damage to the DA system. Angwin and colleagues (2005) also found that individuals with PD demo nstrated a different pattern of priming across stimulus onset asynchronies (SOAs). In other words, how much time was allowed between the prime and the target affected whether or not the semantic network was primed. However when they separated those with p oor auditory comprehension from those with intact auditory comprehension, they found that the most significant differences in priming were in those with poor auditory comprehension suggesting that perhaps the language problems were due to slower processing speed caused by changes to the DA system. Copland (2003) found that priming effects on a lexical ambiguity task with word pairs that had either dominant (bank money) or subordinate (bank river) meanings differed for controls as compared to those with PD, non thalamic subcortical lesions and cortical lesions. All groups demonstrate d priming effects for the dominant condition at both the short and long inter stimulus intervals (ISI). However, the controls demonstrated priming for the subordinate meaning only at the short ISI and no priming at the longer ISI. In comparison, the PD and non thalamic subcortical lesion groups continued to demonstrate priming for both conditions at the long ISI suggesting a lack of normal suppression and implying damage to the indirect loop. Crosson and colleagues (2007) suggested that these studies may elucidate the time course of basal ganglia activity contributing to cognition a nd specifically language
20 production and provide evidence for a cognitive model similar to the structure of the model proposed by Nambu and colleagues (2000) for movement. In this model, suppression via the hyperdirect pathway initially resets the system prior to initiating a new cognition, followed by enhancement of the selected cog nition via the direct pathway and finally suppression of competing alternatives via the indirect pathway. It is the indirect pathway that likely plays an important role in the inhibitory processes that will be addressed below. Inhibitory Mechanisms in Lang uage The lesion model teaches us the benefits of studying patients with focal damage in order to explore brain behavior connections. Impairments in inhibitory processes are commonly observed in individuals with PD suggesting that disruption to the DA syste m Deep brain stimulation (DBS) of the STN has recently become a common treatment for the debilitating motor symptoms in PD. While DBS has been shown to significantly improve motor impairments associated wit h PD, reports on its effects on cognitive processes have been more varied in their results. The majority of studies that have found deficits in inhibitory processing during STN stimulation us ing Stroop interference type tasks and reported delayed RTs or gr eater numbers of errors during STN stimulation as compared to preoperative or off stimulation conditions (Alegret et al., 2001; Moretti et al., 2003) These differential results suggest that the STN may be involved in inhibitory control through the modulation of basal ganglia thalamocortical circuit s. Therefore, to investigate the role of the basal ganglia in inhibitory processing, it is useful to first look at studies of individuals with PD who have undergone DBS of the STN.
21 Declines in verbal fluency performance have also commonly been reported following STN DBS providing further evidence for the role of basal ganglia thalamocortical circuitry in modulation of lexical semantic processes such as word or meaning retrieval. Verbal fluency tasks require two main cognitive processes: response initiati on and suppression of the current response in order to generate a novel response (Bouquet, Bonnaud, & Gil, 2003; Burgess & Shallice, 1996) The declines in performance on both interference tasks and fluency tasks following STN DBS, provides additional support for the involvement of b asal g anglia and associated circuitry in inhibitory language processes. The Hayling Test One test which allows for the differentiation between components of initiation and inhibition during verbal word generation is the Hayling test (Burgess & Shallice, 1996, 1997) The test consists of sentences read aloud in whic h the last word is missing. In part A of the test, the examinee is asked to respond as quickly as they can with a word that completes the sentence. In part B they are asked to respond with a word that is completely unrelated to the sentence or the word wh ich would complete the sentence. Burgess and Shallice (1996) found that when this test was given to individuals with frontal lobe lesions and healthy controls all groups demonstrated significantly longer RTs during part B as compared to part A. By subtracting RTs for section A for each individual from RTs for section B, one can co mpute a measure representative of the simply completing the sentence This subtraction also allows for a removal of the potential confound of initiation problems during the task. Indiv iduals with frontal lobe lesions were found to perform both parts of the test slower than the healthy individuals and patients
22 with lesions in other parts of the brain suggesting significant deficits in response initiation and inhibition Additionally, tho se with frontal lesions made significantly more errors than the other groups. When this same task was administered to individuals with PD following STN DBS, interesting performance differences were noted on versus off stimulation. Overall, the PD group had slower RTs than the control group on both tasks. However, when on and off stimulation conditions were compared separately, the PD group in the on stimulation condition was found to have significantly slower RTs than controls or the off stimulation group i n task A suggesting response initiation problems related to STN stimulation. In contrast, the PD group in the on stimulation condition performed more like controls in task B and on the B A measure and only the those in the off stimulation condition demons trated significantly slower RTs than controls (Castner et al., 2007) The use of the Hayling test all owed researchers to isolate inhibitory lexical semantic mechanisms in PD and determine the modulatory effect of basal ganglia thalamocortical circuits though observation of the effects of STN stimulation on task performance. These data seem to suggest that STN stimulation may increase response initiation problems but improve response i nhibition in PD. The mechanisms by which STN stimulation work are still under considerable debate However, one hypothesis is that the inhibition deficits in PD participants i n the off stimulation condition are due to hyperactivity of the STN and disruptions of the indirect pathway leading to excessive inhibit ion (Castner et al., 2007) STN stimulation may serve to modulate this disruption in the indirect pathway leading to more normal inhibitory processes.
23 Initiation and inhibitory processes are considered generally associated with frontal lobe activity and have been shown to decline with normal aging while other, non executive cognitive functions remain more stable (Bielak, Mansueti, Strauss, & Dixon, 2006) When norms on the Hayling test were collected from a typical aging population it was found that increasing age was associated with slower respondi ng overall. However, this e ffect was greater for task B than A. The number of errors was also significantly affected by age with younger individuals providing fewer error responses than older participants (Bielak et al., 2006) Taken together, these data suggest two questions. First, what is the nature of the involvement of the neuroanatomical structures or networks implicated in inhibitory processing in PD? Second, how are these st ructures or networks affected by normal aging? Use of Neuroimaging as a Measure of Inhibition Only a few studies have used neuroimaging to examine the neuroanatomical networks involved in inhibitory language processes by administering a version of the Hayling test (Allen et al., 2008; Collette et al., 2001; Nathaniel James & Frith, 2002; Nathaniel James, Fletcher, & Frith, 1997) Positron emission tomography (PET) studies have reported that the Hayling test elicits activity in prefrontal and lateral temporal regions in the language network (Nathaniel James et al., 1997) and that response initiation (i.e. task A ) is associated with incre ased activity in the left IFG, left middle and superior temporal gyri and bilateral inferior parietal lobes. These are areas that have been implicated in storage and retrieval of semantic information. During response inhibition (i.e. task B ) greater activi ty was found in areas associated with manipulation of information, planning and inhibitory control including the left dorsolateral and orbital prefrontal cortices (Collette et al., 2001; Nathaniel James & Frith, 2002) Similarly, when
24 Allen and colleagues (2008) administered a version of the Hayling test during fMRI, they reported that response i nhibition was associated with greater activity in the left prefrontal cortex including the dorsolateral and orbito frontal regions as well as the precuneus bilaterally as compared to response initiation. The current study aims to extend this line of resea rch to determine how these networks, thought to be involved in inhibitory processes, may be affected by the aging process
2 5 Figure 1 1. Cartoon of Basal Ganglia loops adapted from Crosson, Benjamin, & Levy (2007) GPi = Internal Globus Pallidus; GPe = External Globus Pallidus; STN = Subthalamic Nucleus; GABA = G amma A minobutyric A cid ; DA = Dopamine.
26 CHAPTER 2 HYPOTHESES Although previous studies have begun to investigate the neural correlates of inhibitory language processes by using versions of th e Hayling test, the neuroanatomical networks are still not well understood. While behavioral studies have indicated that the basal ganglia play a key role in modulating these processes (Castner et al., 2007) the specific means of involvement has yet to be characterized using functional neuroimaging. The current study use d fMRI and a version of the Hayling test (Burgess & Shallice, 1997) to investigate the effects of aging on the neural correlates of inhi bitory semantic processing. The present study is based on the premise that each component of the Hayling test is associated with a different set of cognitive processes ( Figure 2 1). In order to successfully complete task A one needs to use language proces ses including those involved in reading (e.g., lexical semantic, grammatical) as well as initiation of a response including retrieval of the correct word to complete the sentence and speech production of that word. In order to successfully complete task B, the same principles are involved however instead of simple retrieval of the word which completes the sentence, one must first inhibit the related words and instead produce a novel response which is unconnected to the sentence. Therefore, a subtraction o f the processes (e.g., RT, fMRI activity) used to complete task A from those used to complete task B should reveal the unique processes used to produce inhibition. The current study was designed with the following hypotheses in mind :
27 Initiation versus Inhi bition Hypothesis Previous lesion and imaging studies le a d us to believe that irrespective of age group we will find greater left frontal cortical activity in the dorsolateral, inferior medial frontal, and orbito frontal regions (Allen et al., 2008; Stuss et al., 2002) during inhibitory processing a s measured by a task B minus task A comparison to remove the contribution of initiation. Due to the nature of task B requiring inhibition of an expected response, and the proposed role of the STN in inhibitory processing through its modulation of the indir ect loop we also expect to find greater basal ganglia involvement, including STN and caudate nucleus, during performance of the inhibitory condition as compared to the initiation condition (Castner et al., 2007; Kelly et al., 2004) Older versus Younger Hypothesis During the response initiation + language processes condition (i.e., task A) which is a relatively simple sentence completion task, we do not expect to find significant group differences in brain activity, although we predict slower RTs for the older group as compared to the younger grou p based on the aging norms collected by Bielak and colleagues (2006) In task B as compared to task A, we expect to find increased left DLPFC activity for the older group due to the increased effort involved in inhibiting the expected response. In contrast we expect decreased basal ganglia involvement in the older adult group due to the effects of age dependent decline of brain dopamine levels (Elsinger et al., 2003; Kaasinen & Rinne, 2002) Although this effect is not pathological, it would also be consistent with reductions in dopamine activity in the basal ganglia for individuals with PD (Elsinger et al., 2003)
28 Relationship Between fMRI and Behavioral Performance Although no subjects with pathological cognitive functions were included in this study, our hypotheses are based on the assumption that some differences exist between brain activity in younger and older adult s during an inhibition task. Regardless of where in the brain these differences occur, it will be useful t o characterize the relationship between regional brain activity and behavio ral performance on the language task. C orrelation s will be calculated between average amount of activity in a given region of interest and behavioral performance differences as defi ned by reaction time and number of correct responses on the modified Hayling test Findings of increased brain activity in older as compared to younger adults during word retrieval tasks (Cabeza, 2002; C. E. Wierenga, Benjamin, et al., 2008) even in the context of equivalent task performance, have provided evidence that heightened activity in some areas may be compensatory while in o thers it may be nonbeneficial reduction in the ability to inhibit activity in regions of the brain unnecessary for task completion The aging norms on the Hayling test (Bielak et al., 2006) demonstrated an association between older age and increased RT which is even slower on task B than A These r esults suggest the need to determine whether there are brain regions which are associated with improved inhibition in older adults or conversely regions which are associated with poorer performance Using functionally defined anatomical regions of interest based on areas differentially active in the old and young group s, we expect to find a positive correlation between latency on task B and brain activity for the older adults (i.e., increased RT associated with greater activity ) With regards to accuracy we expect the opposite
29 correlation such that a greater number of correct responses will be associated with less activity in those same regions of interest.
30 Figure 2 1. Model of Cognitive Processing Driving Activity in fMRI Voxels. The components of the H ayling test are used as measures of the cognitive processes thought to be involved in their completion. Task A involves various language processes as well as initiation of a response, while task B involves the former processes as well as inhibition. A subt raction of the processes involved in performing task A from those involved in task B is used as a measure of inhibition.
31 CHAPTER 3 METHODS Participants Sixteen (8 male, 8 female) neurologically normal young adults (age 18 35 years), and 16 (8 male, 8 fem ale) neurologically normal older adults (age 65 80 years) participate d in the study. One participant from the older group was excluded from final analyses due to a discovered brain abnormality and one was excluded from the younger group due to not followin g task instructions and responding before the cue making the data unusable in the group analyses Thus, each group included fifteen participants Older and younger participants were matched for gender and level of education (Table 3 1 ) and there were no s ignificant group differences on these variables. All participants were native English speakers recruited from the University of Florida faculty, staff, and students, the Malcom Randall VAMC and from the Gainesville, Florida community. All participants w ere strongly right handed as determined by a score of greater than 50 on the Edinburgh Handedness Inventory (Oldfield, 1971) Potential participants w ere excluded if they report ed any of the following medical problems: history of neurological disorder (e.g., stroke), or head trauma, dementia or Mild Cognitive Impairment, learning disability (e.g., dyslexia), psychiatric disorder (e.g., schizophrenia), substance abuse, or chronic medical conditions likely to impair cognition (e.g., cardiovascular disease, renal or hepatic failure). Since many prescription psychoactive medications may affect cognition, subjects on these classes of medications w ere also excluded. In addition, due to the fMRI component of the task, participants were excluded if they ha d metal in their body other than dental fillings and metal firmly anchored to bone or any other conditions contrai ndicated for MRI. Female participants were excluded if the y were pregnant or
32 trying to become pregnant. On the day of the scan, participants w ere instructed to abstain from caffeine. Informed consent w as obtained from participants according to guidelines e stablished by the Health Science Center Institutional Review Board at the University of Florida. Participants w ere compensated for participation in the study. Materials and Procedures Dementia Screening Prior to enrollment in the fMRI study, participants w ere administered a few brief cognitive tests in order to ascertain the presence of dementia or Mild Cognitive Impairment amnestic type. As the goals of the study were to examine normal aging, it was decided to exclude those with Amnestic Mild Cognitive Imp airment (aMCI), a precursor to pathological aging which has been indicated as a high risk factor for the for aMCI include: a subjective memory complaint, preferably co rroborated by an informant, impaired objective memory function for age and education, as compared to preserved general cognitive functioning, intact activities of daily living, and not meeting criteria for dementia (Petersen et al., 2001). Individuals who report ed subjective memory complaints and who score d below 27 out of 30 on the Mini Mental State Exam (MMSE) (Folstein et al., 1975) w ere excluded due to possible dementia. Individuals who obtain ed scores 1.5 SD or more below the age corrected norms on the Long Delay Free Recall and Recognition Memory of the California Verbal Learning Test II (CVLT II; Delis, Kramer, Kaplan, & Ober, 2000) w ere excluded due to possible aMCI. fMRI Task Design The Hayling test (Burgess & Shallice, 1997) was adapted for use in this fMRI experiment. The stimuli consist ed of 60 sentence stems selected from those given by
33 Bloom a nd Fischler (1980 Appendix A ) A primary consideration was the cloze probabil ity (CP) for each sentence stem. CP is defined as the probability that a particular word will be used to complete a specific sentence in a given subject sample (Kutas & Hillyard, 1984) Highly constrained sentence stems lead to a high probability that a particular word will be used to complete it, while low constraint stems can be completed with a wide ra nge of semantically appropriate words. Since responses for sentences with high CPs (CP> .6) involve a more restricted word choice pool, (e.g., Father carved the turkey with a ____ .) these sentence stems likely elicit less inhibitory process ing during a re sponse than those with moderate or low CPs. Sentence stems with low CPs (CP<.3) increase task difficulty due to the wider word choice pool (e.g. They went to see the famous ____ .) A lthough previous research (Allen et al., 2008 ) does not indicate significa nt differences in patterns of activity dependent on low versus high CPs, we chose to use o nly sentences with a moderate probability of one particular response (moderate constraint sentences: CP between 0.3 and 0.6 ) in this study both to limit variability a nd balance the need to elicit inhibition with the desire to maintain the Sentence stems consist ed of six or seven words. The 60 sentence stems were divided into two lists with equal C Ps and numbers of 6 and 7 word sentence stems in each list. T he MRC Psycholinguistic Database (Wilson, 1988) was used to match the critical word in each sentence for length, frequency, concreteness, and imageability and each list was balanced on these characteristics One list of sentences w as used in the response initiation condition (i.e. task A) which required complet ion of the sentence with a congruent response (i.e., Jeff was
34 sent to bed without DINNER ), while the other list was used in a response inhibition condition, in which an incongruent completion was required (i.e., The cup of tea felt very TEL EPHONE ). The 30 sentence stems in each of the two congruency conditions were arranged into 6 blocks containing 5 sentence stems each for a total of 12 blocks which were alternated in an ABABAB ABABA B block design (see Figure 3 1 for a schematic of the tas k design) The blocks were divided into 4 scanner runs which were either composed of one initiation and two inhibition block s or one inhibition block and two initiation blocks Stimulus presentation order was pseudo randomized such that each participant re ceived one of 8 different run sequences. The pseudorandomization as opposed to true randomization, was used to preserve the alternating task design. Prior to each block a two second cue appear ed on the screen with either the to inform the participant whether their task was to provide a congruent response or an incongruent respo nse. Pilot data using a variety of cue words suggested that these were the most effective and least confusing cue words and did not appear to influence word choice in any way. Sentence stems were presen ted visually in the MRI scanner by projection onto a an Eloquence System ( Invivo ) designed to present stimuli and collect data in fMRI experiments and E Prim e version 1 software. Although auditory presentation was considered, variables including scanner noise, differences in reading speed as compared to auditory processing and synchronization of stimuli led to the decision to present stimuli visually. To contr ol for inter subject reading speed and reduce the potential for creating an accidental working memory task each word in the sentence stem appear ed on the screen one at a time at
35 an interval of 500 m illiseconds (ms). Words appear ed from left to right with all words in the sentence stem remaining on the screen after they were presented for the remainder of the trial Participants were then cued to articulate their verbal response by the appearance of a n underscored blank (e.g., ________) following the last w ord of the sentence stem which remain ed on the screen for a further 4 sec to allow for the In order to align responses of the 6 word (3 second duration) and 7 word (3.5 second duration ) sentences and line up the appearance of the blank with the scanner image acquisition ( TR =2 seconds ). A variable rest time precede d each item (either 500ms or 1000ms) so that the total stimulus presen tation time for each item equaled 4 seconds Each functional run began with a n initial 12 seconds of rest d uring which the participant was instructed to focus on a fixation cross in the center of the screen. Each trial last ed 8 seconds followed by a resting block consisting of a fixation point presented for a variable duration of either 10, 12, or 14 seconds t o allow for the hemodynamic response to return to baseline prior to the start of the next trial. In total each block last ed between 1 min and 40 seconds to 1 min and 44 seconds and each run ( either one initiation and two inhibition block s or the reverse ) last ed 5 minutes and 28 seconds Before scanning, participants receive d training on the task which emphasiz ed that in the inhibition condition responses should not be semantically related to the preceding sentence stem. In the training phase, participants w ere presented with 7 novel sentence stems for response initiation and 7 for inhibition; which were not later u sed during scanning. Once inside the scanner, participants were given standardized instructions before each run reminding them of their task for both the initiation and inhibition blocks.
36 Brain Image Acquisition Images were acquire d on a 3 T esla Phillips whole body scanner. I mages were collected with an 8 channel SENSE radio frequency head coil at the McKnight Brain Institute at the University of Florida Functional images were obtained with a 1 shot gradient echo EPI scan: 24 cm FOV, 80 80 matrix, 3 mm 3 mm in plane resolution, TR = 2000 ms, TE = 30 ms, flip angle = 8 38 3 mm thick axial slices covering the whole brain were acquired. A high resolu tion T1 weighted 3D scan (TE = 3. 69 ms; TR = 8. 06 ms; FOV= 24 cm; matrix size = 240x240 ; 1 8 0 1.0mm slices) was obtained to provide anatomic reference. Head motion was minimized using foam padding. Behavioral Data Analysis Overt verbal responses were coded during the scan as well as digitally recorded for analysis of error rates and response times. A digital audio edito r program, Audacity (Ash et al., 1999) was used to remove scanner noise interference in the digital recordings and two raters reviewed the files to ensure correct documentation of responses, calculation of response times and coding of error types. Response times are de fined as the duration between the presentation of the blank following the last word of the sentence stem Average RTs for each condition for each group were compared along with number and type of errors. S tatistical comparisons were analyzed using PASW statistics 18, Release version 18.0.0 An error analyses based on the rater classification guide described by Burgess and Shallice (1996 Table 3 2) was conducted on r esponses to the inhibition task ( t ask B) which divided errors into two main error types, Correct completions (C) meaning the participant was unable to inhibit as directed by the task instructions and provided a response which correctly completed the sentence and responses semantically related to
37 the correct response. These semantic responses were further broken down into 4 categories, a semantic opposite (SO), a word semantically related to the subject of the sentence (SA), a word semantically related to the correct response (SB) and a vague or ludicrous completion (SC). Responses were coded by two independent raters and dif ferences were discussed until agreement could be reached. Behavioral data were analyzed through two repeated measures ANOVAs. T he effect of group and task on response latency was explored using RT as the dependent variable (DV), and the effect of group and task on accuracy was explored using the n umber of correct responses as the DV. FMRI Data Analysis FMRI data were analyzed and overlaid onto structural images with the Analysis of Functional Neuroimaging (AFNI) program from the National Institutes of Healt h (Cox, 1996) The first 6 images of all the functional runs w ere discarded to ensure attainment of steady state. To control for the effects of head motion, time series images were spatially registered in three dimensional space. Images w ere visually inspected for gross artifacts and quality control procedures applied to the data to detect residual motion or susceptibility artifact. Fo r each of the subjects, the 4 imaging runs for the FMRI task w ere detrended of low frequency signal drifts (Birn, Saad, & Bandettini, 2001) and concatenated into a single time series. Prior to deconvolution and functional analyses, functional images were spatially smoothed with a Gaussian kernel of 5mm full width at half maximum. For each voxel, the obser ved fMRI intensity time series was modeled as the convolution of the experimental stimulus vector (comprised of 30 initiation trials and 30 inhibition trials) and the estimated best fit of the impulse response
38 during the 12 seconds following stimulus prese ntatio n allowing the hemodynamic response to return to baseline. Following coding of the behavioral data, incorrect items (e.g., task A items which did not complete the sentence and task B items which correctly completed the sentence) and missed items we so that only correct responses were included in the final analyses. T he area under the curve (AUC) of the deconvolved HDR was used as t he dependent variable for analyses AUC w as calculated by adding the deconvolved ima ge intensity at each deconvolved time point of the impulse response. The T1 weighted anatomic images and the deconvolved functional activation maps w ere then warped to the co ordinates of a standard brain in Montreal Neurological Institute (MNI) space and resampled at a 1 mm 3 resolution using FMRIB Software Library ( FSL ) version 4.1 tools (Smith et al., 2004) Warping individual brains into standard space is a method widely used as a technique for comparing individual subjects whose data naturally contains variability in size and location of major sulci and gyri and was acquired in slightly different coordinates. The standard brain used was one that has been created from an average of 152 MRI scans of normal control subjects. Cluster thresholding was 3dClustSim program which determined the minimum number of voxels necessary in a given clu ster to achieve a family wise error corrected alpha level of .05. For data i n MNI ( space with voxel size of 1mm x 1mm x 1 mm, adjusted for 5 mm smoothing, and assuming a cluster type that requires that at least the corners of each voxel included are
39 touchin g the corner of another voxel in the cluster an uncorrected p value of .0001 requi res a minimum cluster size of 215 voxels. Three sets of fMRI statistical analyses were completed to explore the neural correlates of task, inhibition, and age 3dttest program For each of these analyses average AUC of the HDR was the dependent variable Within group, within task, voxel wise one t tests were used to examine the effect of task activity as compared to activity during the baseline r esting condition. Within group, paired samples t tests were used to compare activity during task A to activity during task B in order to examine the effect of inhibition. Between group, independent samples t tests were used to explore the effect of age by comparing activity in the young group to that of the old group on each task condition. A fourth correlational analysis explored the relationship between cerebral responses and behavioral performance (RT) using For this analysis, group cluster reports generated by the regions of difference between the old and young group s were overlaid as masks on individual subject data using the AFNI program 3dMaskAve to determine the average level of activity within a given regio n of interest (ROI) for each subject. Correlations were performed between level of activity in each ROI and behavioral perf ormance on the modified Hayling Test subjects performed while in the scanner.
40 Table 3 1 Participant Demographics Group Means(SD) N M/F Age Education (years) MMSE CVLT LDFR CVLT R Young 15 8/7 23. 3(3.8) 16 (2.1) 29 .4(1.2) 0.67(0.99) 0.13(0.51) Old 15 7/8 76.7 (5.1) 15.8(2.9) 28. 8(0.9) 0.5(0.88) 0.06(1.14) Notes: Demographic data summarized above includes only participants whose data was included in the final analyses. N = number of participants; M/F=male/female ; CVLT LDFR=CVLT II long delay free recall z score; CVLT R=CVLT II recognition z score Table 3 2 Error characterization of task B responses based on the Burgess & Shalli rater classification guide (1996). Error Type Description Classification 1. Does the word reasonably complete the sentence (i.e. it's a word you might give yourself if asked to provide a word that would fit at the end). C 2. Is the word an oppos ite of what you might expect as an answer? SO 3. Is the word obviously semantically connected to the subject of the sentence? SA 4. Is the word obviously semantically connected to the expected response? SB 5. Does the word vaguely fit at the end of the sentence, but in a way that makes the sense of the sentence ludicrous or is the word a slang semi obscenity? SC
41 Figure 3 1. Schematic of task design. A) Components and timing of an individual item presentation. B) Sample block of 5 sentence stems. Prio r to each block, Sample scanner run which starts by reminding the participant of the task instructions and is followed by 3 blocks of 5 sentence stem. D) Model of the whole task design i ncluding practice trials containing novel sentences. Between each run subjects were reminded of the task instructions.
42 CHAPTER 4 RESULTS Behavioral Results Results of the behavioral analyses revealed significant difference s between groups ( Figure 4 1, Fi gure 4 2, Table 4 1 Table 4 2 ) Two repeated measures ANOVAs were conducted to compare the effect of task on 1) response latency (DV= RT ) and 2) accuracy (DV= number of correct responses ) in both the old and young group s. There was a significant effect of task on RT such that across groups, RT was significantly slower on t ask B than task A ( F (1,28)=22.768, p <.001 ) Similarly, across groups, the number of correct responses was significantly less on task B than task A ( F (1,28)=11.268, p <.01 ) In other wor ds, across age groups, participants performed slower and produced fewer correct items during task B as compared to task A. Across tasks, there were also significant between groups effect s of age on RT ( F (1,28)=11.2 53 p <.01 ) and age on number of correct responses ( F (1,28)= 7 555 p <.01 ) In other words, the older group performed slower and produced more errors across tasks than the younger group. Results also revealed a significant interaction between task and group for RT ( F (1,28)=8.516, p <.01), meani ng that the pattern of RT differences by task differed for each age group Similarly there was a significant interaction between task and group for the number of correct responses ( F (1,28)=13.991, p <.001), such that the differences between number of corre ct responses generated in each task varied by group. In order to further clari fy this interaction and to investigate our hypotheses regarding the effect of age and task on performance, group means were compared using t t ests Overall the Old group respon ded significantly slower than the Young
43 group across task conditions ( t = 2.7 79 p <.0 5 ) Task comparisons examining only correct responses revealed no significant RT differences between group s on task A though on task B, the Old group performed significant ly slower than the Young group ( t = 3. 524 p <.01 ) One question which we did not address directly was whether performance improved over time and if this varied by group. To investigate this somewhat indirectly, these RT differences were broken down and exa mined across runs in the scanner (Figure 4 3). F or both groups, RT on task A did not differ significantly over time. Interestingly, o n task B, RT for the young dropped after the first run, likely due to effective use of strategy, and remained constant for the remaining runs. In contrast, RT for the old group remained constant until the last run when RT suddenly decreased. These trends suggest that the young are benefitting much earlier than the old from effective use of strategy to complete task B and may p rovide evidence that the task is being performed differently by each group. Results of the error analysis can be found in Table 4 2. A response was coded as an error in the initiation condition (task A) when a participant provided a response that did not c omplete the preceding sentence stem in an expected or sensible way. In the inhibition condition (task B) a response was defined as an e rror when it completed the sentence in a sensible way or had an obvious connection in meaning to the preceding sentence s tem. These errors were further broken down into five error types as described in Table 3 2. The Young group provided significantly more correct responses than the Old group on t ask B ( t =3. 496 p <.005) while group performance was not significantly
44 differe nt on task A The Old group missed or did not provide an answer to significantly more items on task B than the young group ( t (28) = 2.752, p <.05) though again no significant group differences were discovered when comparing performance on task A The error analysis of incorrect responses to the inhibition task revealed that the Old group provided a significantly greater number of type C errors ( i.e., c orrect completions of the sentence stem ) than the Young group ( t = 2.093, p <.05) No significant group diffe rences were found for the semantically related error types ( i.e., SO, SA, SB, SC). F MRI Results Effect of Task (Task vs. Rest) Separate group analyses were completed to compare the mean activity during each task to the rest condition (Table 4 3) In both t he young and old group s i ncreased blood oxygenation level dependent (BOLD) contrast in task A as compared to r est was associated with activity in middle frontal gyrus and precentral gyrus as well as middle temporal gyrus, and cuneus Unique areas of activity in the young group relative to the old group include d the anterior cingulate sulcus, the anterior cerebellum and the left dorsomedial and left lateral thalamus. Unique areas of activity i n the old group included superior and inferior parietal cortices, superior temporal gyrus, pulvinar and putamen. In task B when compared to the resting condition, both groups showed similar patterns of frontal and parietal activity to task A with the addi tion of DLPFC activity which was unique to task B for both groups. In the young group as compared to the old group during task B activity was uniquely observed in precuneus, the left anterior nucleus and right dorsomedial nucleus of the thalamus and the l eft body of the caudate
45 I n the old group unique activity compared to the young group consisted of superior and inferior parietal regions angular gyrus, superior temporal gyrus, pulvinar and right puta men. Across groups and tasks activity in the cingula te gyrus and cingulate sulcus anterior to the corpus callosum showed negative activity (i.e. greater activity during rest than task) and this activity was greater in the young than the old group s P rior studies (Hafkemeijer, van der Grond, & Rombouts, 2011; Koch et al., 2010) have indicated cingula te activity anterior to the corpus callosum are part of the default mode network (DMN) and that interruptions in the DMN occur with aging leading to decreased activation Additional negative activity was found in subcortical regions for the old group incl uding the anterior nuclei of the thalamus and right head of the caudate. Unique Effects of Inhibition (Task B minus Task A) by Group In the young group the inhibition comparison was associated with greater activity bilaterally in dorsolateral prefrontal c ortex (DLPFC), superior frontal gyrus (SFG), orbitofrontal cortex (OFC), precuneus, anterior cerebellum and the posterior cingul ate gyrus, as well as activity in right angular gyrus and left supramarginal gyrus ( Figure 4 5, Table 4 4 Appendix C ). No areas showed greater activity for the initiation task than the inhibition task. In contrast, analysis of the old group did not reveal any areas associated with significant task differences in either direction. In other words no areas were found to be significan tly more active in one task than the other for the old group Effect of Age (Old vs. Young) by Task During Task A the old group demonstrated greater activity than the young group in left precentral gyrus, left supramarginal gyrus, right angular gyrus and right parieto occipital sulcus as well as bilateral activity in the paracentral lobule, inferior parietal
46 cort ices, postcentral gyri and sulci, precuneus and cingulate cortex ( Figure 4 6, Table 4 5, Appendix C ) No areas were found to be associated with g reater activity in the young than old group on task A Results differed for the inhibition condition. The old group showed greater activity than the young group in superior and medial frontal gyri, left precentral gyrus, bilateral ly in superior and inferi or parietal cortices, and posterior cingulate gyrus. The old group also showed greater negative activity in the right head of the caudate and prefrontal cortex than the young group ( Figure 4 7, Table 4 6 Appendix C ) Effect of Age (Old vs. Young) on Inhib ition (Voxel by Voxel B A) To further describe the results of the group and task differences, the AUC during task A was subtracted from the AUC during task B for each voxel of each participant. Using these voxel by voxel models of inhibitory activity, an i ndependent samples t test between old and young groups revealed consistently greater activity in the young than the old in the left cuneus (BA 7), cingulate (BA 23 and 32), left superior parietal lobe (BA 5/7), right orbitofrontal cortex (BA 47/11), and ri ght DLPFC (BA 9) (Figures 4 8, 4 9; Table 4 6). Correlational Analysis (Brain vs. Behavior) To clarify the relationship between brain activity differences in older adults and behavioral performance on the task, average RT for each participant in the old g roup during task B was correlated with brain activity as measured by average AUC in the clusters generated by the old vs young comparison Although originally planned, correlations were not completed using the number of correct responses due to the fact t hat the data were non normal ly distributed with an average number of correct responses of 26.6 out of 30 or 88% correct. For the old group significant positive
47 correlations were found between increased RT during task B and increased bilateral midline acti vity in the m edial and superior frontal gyri (MF G /SFG) ( r s =.518, p <.05 ); the left superior parietal lobe/ post central gyrus ( r s =.668, p <.01 ) and right head of the caudate r s =.59, p <.05 ; Figure 4 4 ) Although it may seem counterintuitive since the activit y in the caudate was negative, the correlation with RT revealed that greater negative activity is associated with faster responding.
48 Figure 4 1. Mean r eaction ti mes by g roup and t ask Indicates significant group difference ( p <.05) ; RT = reaction ti me. Figure 4 2. Mean behavioral performance by g roup and t ask. 0 500 1000 1500 2000 2500 3000 Task A Task B Task B-Task A Reaction Time in milliSeconds Mean Reaction Times by Group Young Old
49 Figure 4 3. Average RT by group plotted across time with standard error bars representing variability For both groups RT on task A did not differ significantly over time. On t ask B, RT for the young dropped after the first run, likely due to effe ctive use of strategy, and remained constant for the remaining runs. RT for the old group remained constant until the last run when RT suddenly decreased. These trends suggest that the young are benefitting much earlier than the old from effective use of strategy to complete task B and may provide evidence that the task is being performed differ ently by each group.
50 Figure 4 4 Correlations between average AUC activity for older p articipants during task as follows A: r s =.518, p<.05 ; B: r s =.668, p<.01 ; C: r s =.59, p<.05 Table 4 1. Average group behavioral performance by task Group A RT B RT A Correct B C orrect A NR B NR Mean (SD) Young 12 68 ( 363 ) 1 48 2 (47 3 ) 2 8 93 ( 0 88 ) 29. 07 ( 0 .9 6 ) 0.13( 0 .35) 0.40( 0 .91) Old 1 45 3 ( 325 ) 2 34 4 (82 0 ) 29.07( 0 .88) 26.60(2.56) 0.40( 0 .63) 2.13(2.26) Notes : = Significant difference between group means p<.05 ; A RT = reaction time o n task A in ms ; B RT = RT on task B in ms ; Correct=total correct responses out of 30 possible items; NR = number of unanswered items ; SD = standard deviation.
51 Table 4 2 Error a nalysis of t ask B r classificatio n guide (1996). Group C Error SO Error SA Error SB Error SC Error Mean (SD) Young 0. 53 ( 0 .64) 0.13( 0 .52) 0. 87 ( 0 92 ) 0. 87 (1.5 1 ) 1. 2 0(1. 15 ) Old 1.27(1.33) 0.33(1.05) 0.93(1.62) 0.80(1.26) 1.13(1.13) Notes: = Significant difference between group means p< .05; C Error = the word reasonably completes the sentence; SO Error = the word is an opposite of what you might expect; SA Error = the word is semantically connected to the subject of the sentence; SB Error = the word is semantically connected to the expec ted response ; SC Error = word vaguely fits in a way that makes the sentence ludicrous Table 4 3. Effect of Task: Cluster locations by group based on one sample t tests Cluster Location Young A Young B Old A Old B Cortical Subcortical Fro ntal Parietal Temporal Occipital Cerebellum Cingulate Thalamus Striatum Broca IFG MFG PreCG SMG MTG Cuneus Ant.lobe AC S AC G L DMN L LN Broca IFG MFG PreCG DLPFC preCuneus MTG Cuneus Ant.lobe AC S AC G L AN DMN L Caudate B Broca IFG MFG PreCG ~ SPL ~ IPL SMG ~ STG MTG Cuneus AC G ~ Pulvinar ~ Putamen Broca IFG MFG PreCG DLPFC ~ SPL ~ IPL SMG ~ AG ~ STG MTG Cuneus AC G AN /VA ~ DMN ~ R Caudate H ~ R Putamen Notes: Clusters size 215 voxels thresholded at an uncorrected p value of .001 to achieve a family wise error corrected alpha level of .05. Activity is bilateral unless specified ; R=right, L=left; (pars opercularis, pars triangularis); IFG=Inferior Frontal Gyrus; MFG=Medial Frontal Gyrus; DLPFC=dorsolateral prefrontal cortex; PreCG=PreCentral Gyrus; AC S =Anterior Cingulate Sulcus; AC G =Anterior Cingulate Gyrus; SPL=Superior Parietal Lobe, IPL=Inferior Parietal Lobe; SMG=Supramarginal Gyrus ; AG=Angular Gyrus ; S TG=Superior Temporal Gyrus; MTG=Medial Temporal Gyrus; AN=Anterior nucleus of the t halamus; VA=Ventral anterior nucleus of the thalamus DMN=Dorsomedial nucleus of the Thalamus; LN=Lateral nuclei of the Thalmus; Caudate H =Head of Caudate; Caudate B =Body of Ca udate ; = negative BOLD ; ~ =unique area in Old
52 Figure 4 5 Effect of Inhibition (task B task A) for young group based on paired sample t tests P<.0001. All areas are greater AUC for task B than A. Table 4 4 contains descriptions of clusters. Table 4 4 Effect of Inhibition ( task B task A) : Cluster locations for Young group based on paired sample t tests Cluster Location BA #Voxels X Y Z Frontal R DLPFC 46,11,10,9 8 22642 27.6 34.4 30.3 L DLPFC 46 1584 41.3 29.4 21.8 R SFG 8/9 330 3.3 4 0.9 51.4 R OFC 25 601 28.4 21.1 18.2 L OFC 10 2694 27.9 50.1 13.7 L OFC 11 1909 24.2 59.5 9.7 Parietal P recuneus 7 15322 0.2 68.4 38.3 R A ngular 39 4163 46 50.8 38.9 L SMG 40 2696 34.7 58.4 40.9 Cerebellar L Ant. lobe 985 31 61.9 36 R Ant.lobe 667 28.1 66.5 38.9 Cingulate P ostCinG 23 4026 0.6 33 21.4 Notes : Clusters included contained at least 215 voxels and were thresholded at an uncorrected p value of .001 to achieve a family wise error corrected alpha level of .05. BA=Brodm ann Area; DLPFC=dorsolateral prefrontal cortex; SFG=Superior frontal gyrus; OFC= Orbitofrontal cortex; SMG=Supramarginal Gyrus ; Ant. Lobe=anterior lobe ; P ost C in G=Posterior Cingulate g yrus B>A A>B L R
53 Figure 4 6 Effect of age (Old Young) on task A bas ed on independent sample t tests p<.00009. Table 4 5 contains descriptions of clusters. Table 4 5 Effect of Age (Old Young) during Task A : Cluster locations by task based on independent sample s t tests Cluster Location BA # Voxels x y z Old>Young Frontal L S uperior PreCG 4 540 26.2 14.4 59.7 4 246 29.4 11.2 48 P arace ntral lobule 5/3,1,2 1688 4.8 26 61.2 Parietal L IPL 43/40 1957 53.9 23.5 27.3 R IPL 43/40 338 59.2 14.9 22.8 L SMG 40 1056 38.1 47.4 36.1 R P ostCS 929 45.5 27 43.4 L P ostCG (superior) 3,1,2 249 47.8 23.6 45.3 R A ngular gyrus 39 239 36.7 59.3 40.3 R P recuneus 7 309 10.6 57 43.1 L P recuneus 7 716 9 68.6 32.9 7 485 11.7 54.6 54.4 Cingulate R CinG 23 227 5.4 23.6 37.8 31 781 13 41 34.7 L PostCinG 31 336 12.2 42.1 34.9 Occipital R Parieto occipital sulcus 262 17.3 71.1 40.9 Notes: Clusters included contained at least 215 voxels and were thresholded at an uncorrected p val ue of .00 009 to achieve a family wise error corrected alpha level of 01 PreCG=precentral gyrus; IPL=inferior parietal lobe, SMG=supramarginal gyrus; postCS=postcentral sulcus; postCG=postcentral gyrus; CinG=cingulated gyrus; P ost C in G= p osterior c ingulate g yrus Old > Young Old < Young L R
54 Figure 4 7 Effect of age (Old Young) on task B based on independent sample t tests p<.00009. Table 4 6 contains descriptions of clusters. Table 4 6 Effect of Age (Old Young) during Task B : Cluster locations by task based on independe nt sample s t tests Cluster Location BA # Voxels x y z Old>Young Frontal MFG, SFG 4/5/3 1 2 6426 5.4 22.6 57.6 L S uperior PreCG 4 607 27.2 18 .0 58.5 Parietal R post CG 3 1 2 1552 34.7 35.5 53 .0 L post CS / IPL 3 1 2/40 1038 53.1 25.1 32.2 R post CS/IPL 3 1 2/40 818 6 1 20.2 25.1 L S uperior post CG 3 1 2 566 31.7 34.5 47.2 R SPL /post CG 3 1 2 563 18.4 43.9 61.2 L SPL /post CG 3 1 2 232 17.8 41.3 60.3 Cingulate R PostC in G 23 689 11.2 38.7 34.4 L PostC in G 23 365 11.2 37.4 32.3 Young > O ld R Caudate H 221 8.2 17.8 4.1 Medial PFC / CinS 32 351 2.4 39.6 21.8 Notes: Clusters included contained at least 215 voxels and were thres holded at an uncorrected p value of .0 0009 to achieve a family wise error corrected alpha level of .0 1 SFG=Superior frontal gyrus; OFC= Orbitofrontal cortex; MFG=Medial Frontal Gyrus; PreCG=Pre Central Gyrus; PostCG=Post Central Gyrus, postCS=Post Central Sulcus; CinS=Cingulate Sulcus; P ost CinG=Posterior Cingulate Gyrus ; IPL=inferior parietal lobe ; PFC=prefrontal cortex O ld > Young Old < Young L R L
55 Figure 4 8 Effect of inhibition (B A) compared between groups based on independent sample t tests p<.0001. Clusters represent areas of significant age related activity differences associated with inhibition. Figure 4 8 and Table 4 7 contain descriptions of clusters. Figure 4 9 Age differences in activity associated with i nhibition Average AUC differences (B A) in c lusters found to be significantly different between old and young groups. L=left; R=right; Sup=superior; OFC=orbitofrontal cortex; DLPFC=dorsolateral prefrontal cortex. -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 L cuneus (BA 7) Cingulate (BA 32) Cingulate (BA 23) L Sup Parietal (BA 5/7) R OFC (BA 47/11) R DLPFC (BA 9) Average AUC differences (B A) Regions of significant group differences between task related activity Young Old Old > Young Old < Young
56 Table 4 7 Age differences in activity associated with inhibition based on independent samples t tests. Cluster Location BA # Voxels x y z Young > Old Left Cuneus 7 3633 3.2 67.9 50.0 Cingulate (bilateral act.) 32 1290 1.9 40.6 23.2 Cingulate (bilateral act.) 23 846 0.4 41.1 19.0 Left Superior Parietal 5/7 812 34.4 56.3 41.3 Right Orbitofrontal 47/11 727 24.7 21.8 16.4 Right DLPFC 9 615 38.7 49.2 24.8
57 CHAPTER 5 DISCUSSION This study explored the neural correlates of inhibitory proc esses in language and how they differ in older and younger adults Using a modified version of the Hayling test (Burgess & Shallice, 1997) during fMRI we were able to examine the neural correlates of differences in task performance between healthy older adults and healthy younger adults and compare related brain activity to b ehavioral performance. While sentence completion tasks have been used in a handful of imaging studies in order to examine aspects of inhibitory processes (Allen et al., 2008; Collette et al., 2001; Nathaniel James et al., 1997) none have examined the effe cts of age on inhibitory neural correlates. Effect s of Age on Inhibition In order to explore this complex question, a few levels of analyses were completed. To facilitate interpretation of these results, we wi ll explore the most complex analysis first and then attempt to understand it using results from the other analyses. The primary purpose of this study was to examine age related changes in the neural correlates of inhibition. Thus, we examined a voxel by voxel comparison of task related activity differ ences (i.e., activity during task A subtracted from activity during task B) for each participant to model inhibition in each voxel T hen we compared the resulting activity differences across groups (Figure 4 8, Table, 4 7). Significant clusters in this an alysis therefore represent regions in which there were group (old vs. young ) differences in activity required to support inhibition in task B over activity required to support task A.
58 Regions found to have greater task differences between groups (e.g., left cuneus (BA 7), cingulate cortex (BA 23 and 32), left superior parietal lobe (BA 5/7), right orbitofrontal cortex (BA 47/11), and right DLPFC (BA 9) ) all revealed greater activity differences in the young group than the old group (Figure 4 9). These re sults were unexpected given that the older adults performed task B significantly slower than the young adults and showed greater RT differences between tasks (Figure 4 1) which we had hypothesized would lead to increased activity differences. In order to comprehend these seemingly conflicting results we must turn to our earlier analyses. Most importantly, the results examining the effect of inhibition within each group (Figure 4 5, Table 4 4, Appendix C) informed us that though significant regions of activ ity (bilateral DLPFC, SFG, OFC, precuneus, anterior cerebellum, posterior cingulate gyrus, right angular gyrus and left supramarginal gyrus) were associated with inhibition in the young group, no such differences were found in the old group. Lack of signif icant task differences in the old group suggests that similar networks were being utilized to perform both task A and task B, but this comparison does not tell us what areas the old group used to perform tasks A and B Again we can turn to an earlier ana lysis for clarification. The comparison of activity associated with the effect of task (i.e., task vs. baseline) confirms that there was in fact significant task related activity (Appendix C, Table 4 3) for the old group during both tasks A and B and that the regions of activity were remarkably similar. It appears that those regions recruited by the young group during task B were already in use during task A for the old adults and therefore could not be differentially engage d in task B.
59 This explanation cl arifies why we did not find areas of greater task related differences in the old group in our overall analysis of the effect of age on inhibition, but it does not completely explain how the behavioral results relate to functional imaging results A review of the analysis exploring the effect of age (i.e., old vs. young) found that during both tasks, regions of group differences predominantly demonstrated greater activity in the old than the young group (Figure 4 6, Figure 4 7, Table 4 5, Table 4 6). There a re two possible explanations for greater activity in old than young adults: (1) Old adults may need an activity increase to support performance of tasks that have become more difficult with age. (2) There may be a loss of the ability of one brain area to s uppress other brain areas during task performance, resulting in less differentiated activity in the old than the young group. Under the latter hypothesis additional areas of activation could cause interference and decrease the efficiency of performance. For the current data, brain regions engaged during both tasks were similar in spite of behavioral results indicating that task B was more challenging for old adults ; i.e. old adults performed task B more slowly and produced more errors than young adults. T he theory of compensation is insufficient to explain this age related activity difference. If it had been compensatory old adults would have shown greater activity during task B than A in order to compensate for the difference in task difficulty. Instead these data tend to support the loss of inhibition theory because of the greater activity for old adults than young adults in task A, which was less difficult than task B and behaviorally showed similar reaction times and accuracy between the groups ( Figu re 4 2) However this is not the whole story. While most clusters during task B showed greater activity in old than young adults two exceptions were found : T he right head of
60 the caudate nucleus and the medial PFC/ anterior cingulate sulcus both demonst rated greater activity in the young than the old group These data generate two additional questions. First, what is the meaning of the two regions for which the young adults demonstrated greater activity during task B than the old adults ? S econd, are any of these regions associated with group behavioral performance differences on task B (i.e., RT) ? To begin, we can review the hemodynamic response functions (Appendix C) for the se two regions. In most of the clusters of activity, the pattern of activity is similar in both groups though the activity is greater in old than young adults In contrast, in both of these regions what we find is positive activity in the young and negative activity in the old. In other words, one cannot simply state that the young ar e showing greater task B related activity in those areas ; it also must be noted that the old adults are demonstrating suppression of activity. This is unique for old adults relative to previous studies in our laboratory. Specifically, McGregor et al. (2011) showed a loss of suppression of motor cortex ipsilateral to unilateral hand movement in old adults, and M einzer et al. (in press) and Zlatar (unpublished) showed a loss of suppression of default mode network activity in old adults during word generation. Thus, it is not true that old adults always lose the ability to sup press brain activity during an active task. T his phenomenon may be specific to the task at hand rather than a general characteristic of the older brain. To complicate things even more, our correlation analyses (Figure 4 4) found significant positive corre lations between the activity in the right head of the caudate and response latency during task B for the older adults. In other words the more negative the activity in the right head of the caudate, the faster the participant performed task B.
61 This finding indicates that suppression of right caudate head activity may be important for optimal task performance for old adults. Why suppressed caudate performance would be helpful in task B performance for old adults is unclear. The medial frontal cortex, which a lso showed increased activity in young adults and decreased activity in old adults, did not demonstrate any correlation between task related activity and behavior. One final finding is worth discussing. For some of the areas showing activity differences b etween groups, the associated time courses (Appendix C ) include only the declining portion of the hemodynamic response function suggesting that the activity is related to an event starting prior to the time period we modeled i.e., giving the word at the end of the sentence This activity is most likely associated with the reading that occurred before the response we modeled Although we attempted to control for reading speed by presenting the words of the sentence stems one at a time, other aspects of rea ding may have introduced additional variability. We designed the study so that the words of the sentence stem remained on the screen during the response period in order to remove potential task difficulty due to working memory issues H owever this may als o have encouraged old participant s to reread the words in the sentence stem while simultaneously anticipating the response to be given at the end of the sentence Strengths and Weaknesses A general difficulty with this study and the goal of investigating t he neural correlates of inhibitory processes is that the concept of inhibition itself is somewhat vague In a motor paradigm it is much easier to define the initiation and inhibition of a given task and to determine when the inhibition likely begins and e nds. In the extended cognitive paradigm used in this study (i.e., sentence reading and completion), the
62 inhibitory event may be less discrete and probably involves a few simultaneous or sequenti al processes. For example, in task B our analyses were perform ed on the fMRI images which occurred immediately after the cue to respond at the end of the sentence stem. Participants were instructed not to respond prior to that cue and our assumption was that after reading all the words of the sentence stem, a partici pant would engage the semantic network needed for a correct response, inhibit that response and retrieve a new unrelated response before initiating a verbal response By presenting the words of the sentence stems one at a time, we hoped to prevent the inhi bition from beginning until just after the last word of the sentence stem was read. However, in reality, reading each word of the sentence stem may bring a participant one step closer to knowing how the sentence will end and it is entirely possible that th e inhibitory semantic processes are co occurring while the participant is reading the sentence This may explain why it was difficult to find evidence of typical HDR functions in the time period for which we analyzed these data. The argument can be made th at a comparison of fMRI activity during task A to that of task B should eliminate the confounds of reading and the general processing which leads up to generating a word which completes the sentence ; however it seems clear from our data that it is not tha t simple. Semantic inhibition may be better thought of as a process which occurs over time, in stages, as information is gathered that can help inform the selection of a desired response or cognition. This way of thinking about inhibition means that for fu ture studies extreme care should be taken to design an extremely simple yet elegant task which allows for a time limited characterization of inhibitory processes.
63 In addition, it is unclear if old and young participants may have differentially benefitted from the use of strategies to complete the inhibition task and thereby introduced a confound into the inhibitory process we were trying to examine (Figure 4 3) In order to further investigate aging related changes to the neural correlates of semantic inhi bition it will be important to control for strategy use possibly by increasing the task difficulty, providing a strategy, or designing a task which does not allow fo r consistent use of strategies. In summary, a s improvements in health care increase longev ity, thereby facilitating growth of the aging population, the scientific community is increasingly motivated to study the cognitive changes which accompany biological aging. The literature is currently inundated with aging research A lthough huge strides h ave been made in our understanding of the neural correlates of cognitive changes associated with pathological aging for a variety of disorders, our understanding of normal, non pathological aging is still lagging behind. As normal aging is a process that n one of us can escape, and heavily impacts other pathological processes, it is essential that we continue attempt ing to understand the aging brain as the baseline state from which pathological aging departs It is well known that consistent d ifferences exis t between older and younger adult brains though the reasons for these differences and what they mean for behavioral performance and development across the lifespan is an area of study that needs considerably more investigation. Specifically, in the area o f language studies, we have assumed for a long time that the younger brain is an adequate model for diseases failed to take into account increases in right hemisphere a nd other activity during
64 language tasks for older adults when we interpret findings in studies of aphasia. Previous findings (Meinzer et al., in press; Wierenga et al., 2008) indicate that we can no longer ignore the effects of aging on brain language func tions. Findings from the current study have extended these previous findings and also suggest that we are still unraveling the complexities of age
65 APPENDIX A S ENTENCE STEMS INCLUD ED IN TASK A B 1. Jef f was sent to bed without 2. He had to fill his truck with 3. Scotty licked the bottom of the 4. Every month Rick had to clean his 5. The girl was advanced for her 6. Dan caught the ball with his 7. Diane slowly sank into the hot 8. Suzy liked to play with her toy 9. Did you want to go to the 10. He wondered if the storm would be 11. They were startled by the sudden 12. My uncle gave my mother a big 13. Plants will not grow in dry 14. Some of the ashes dropped on the 15. Her dress was made of very fine 16. The wooded lake made a pretty 17. In the morning Jak e took out the 18. The cup of tea felt very 19. Tim would often sleep during his lunch 20. The final score of the game was 21. No one wanted to accuse him of 22. Most students prefer to work during the 23. Don found that he had no spare 24. She locked the valuables in the 25. You can't buy anything for a 26. John poured himself a glass of 27. After speaking Allen left the noisy 28. Even their friends were left in the 29. The choir sang hymns while the people 30. Joan showed her friend a new card 31. The thick mud stuck to her 32. Barry wisely chose to pay the 33. Sur gery was needed to repair his failing 34. He was miles off the main 35. Our new green car blocked the narrow 36. The earth is shaped like a 37. Helen reached up to dust the 38. Matt was wild when he was 39. He smiled and sat down at the 40. My aunt likes to read the daily 41. They took short trips during the 42. He put his feet up on the 43. Jim had learned the special passage by 44. Jack hit his horse with a 45. Bob thought she had a friendly 46. Larry chose not to join the 47. The apple pie had a delicious 48. What you find depends on where you 49. Paul has always w anted to be a 50. In the park the hippie touched the 51. She cleaned the dirt from her 52. He drove the nail into the 53. The little girl was afraid of the 54. My father and mother are getting 55. Harriet sang while my brother played the 56. Stan slowed down going around the 57. Few nat ions are now ruled by a 58. Jill looked back through the open 59. Before jogging, it's a good idea to 60. Not even the cast liked the
59 APPENDIX B INSTRUCTIONS FOR SEN TENCE COMPLETION TAS K O utside Scanner Instructions Practice t rials explain each one to you and let you have a chance to practice before you go in the Bring up practice items: from it. For the first task I want you to carefully read each sentence and when the blank appears, your job is to give me a word which completes the sentence. Do you understand? Run through two examples. But the important thing is that I want you to give me your answer as quickly as you can, Run through the rest of practice for part A. second task. Again, you will see a series of sentences with the last word missing just like the ones you have already done, but this time I want you to give me a word which does not fit at the end of the sentence, I want the word you give me to be complet ely unconnected to the sentence in every way. Do you understand? Run through two examples. must be unconnected to the sentence and that it is important for you to give me your Run through the rest of practice for part B Issues to c orrect if they come up during practice trials: 1. If a subject gives an answer consisting of three words or more, say: 2. The first time a subject gives a related word for the second part say: e DOES fit at the end of the sentence. The sentence [ read sentence with completed word instance if you have given me is unconnected to the sentence [ read sentence ]. So can you think of a word which you could have given me instead of [
60 Let the subject try again, then move on to the next practice item, regardless of whether they produce an unconnected word. If the response is acceptable, just say: If the response is not acceptable (i.e. strongly connected to the sentence), say: 3. Any other time the subject gives a related response for second part say: d remember, what I want 4. If subject repeats responses say: good way of approaching this te 5. If a subject asks about other strategies, say: (BUT do not u nder any circumstances give examples of other strategies they might use). A fter the practice items: There will blocks of sentences where your task will be to complete the sentence, and other blocks where your task will be to give unconnected words. completes th give is unconnected to the sentence in every way. Do you understand? Any I nside Scanner Instructions Just before the scanner task: mpletion task. For each group of sentences, remember that your job is to read the sentences carefully and say a word based on the give are unconnected to the sentences in every way. It is also important for you to give Before a new run: Remember, unconnected to the sentences in every way. And please give your answers as quickly
61 APPENDIX C MONTAGES OF FMRI CLUSTER ACTI VITY WITH ASSOCIATED AVERAGE HEMODYNAMIC RES PONSE FUNCTIONS Effect of Task (Task vs. Baseline) Young: Task A vs. Baseline Young: Task B vs. Baseline Task A > Baseline Task A < Baseline Task B > Baseline Task B < Baseline L R L R
62 Old: Task A vs. Baseline Old: Task B vs. Baseline Task B > Baseline Task B < Baseline Task A > Baseline Task A < Baseline L R L R
63 Effect of Inhibition (Task B vs. Task A) for Young (none for Old) 1 2 3 4 5 6
64 7 8 9 10 11 12
65 Effect of Age (Old vs. Young) Task A (16 clusters) 1 2 3 4 5
66 6 7 8 9 10 11
67 12 13 14 15 16
68 Task B (12 Clusters) 1 2 3 4 5
69 6 7 8 9 10 11
71 LIST OF REFERENCES Alegret, M., Junque, C., Valldeoriola, F., Vendrell, P., Pilleri, M., Rumia, J., & Tolosa, E. (2001). Effects of Bilatera l Subthalamic Stimulation on Cognitive Function in Parkinson Disease. Arch Neurol 58 (8), 1223 1227. doi:10.1001/archneur.58.8.1223 Allen, P., Mechelli, A., Stephan, K. E., Day, F., Dalton, J., Williams, S., & McGuire, P. K. (2008). Fronto temporal Interac tions during Overt Verbal Initiation and Suppression. Journal of Cognitive Neuroscience 20 (9), 1656 1669. doi:10.1162/jocn.2008.20107 Angwin, A. J., Chenery, H. J., Copland, D. A., Arnott, W. L., Murdoch, B. E., & Silburn, P. A. (2004). Dopamine and seman tic activation: an investigation of masked direct and indirect priming. J Int Neuropsychol Soc 10 (1), 15 25. Angwin, A. J., Chenery, H. J., Copland, D. A., Murdoch, B. E., & Silburn, P. A. (2005). Summation of semantic priming and complex sentence compreh ension in Brain Res Cogn Brain Res 25 (1), 78 89. Ash, R., Chinen,, M., Crook, J., Ijbulatov, R., Johnson, V., Mazzoni, D., Shaw, M., et al. (1999). Audacity Retrieved from http://audacity.sourceforge.net/ Bielak, A. A. M., Mansueti, L., Strauss, E., & Dixon, R. A. (2006). Performance on the Hayling and Brixton tests in older adults: Norms and correlates. Archives of Clinical Neuropsychology 21 (2), 141 149. doi:10.1016/j.acn.2005.08.006 Birn, R. M., Saad, Z. S., & Bandettini, P. A. (2 001). Spatial heterogeneity of the nonlinear dynamics in the FMRI BOLD response. NeuroImage 14 (4), 817 826. doi:10.1006/nimg.2001.0873 Bloom, P. A., & Fischler, I. (1980). Completion norms for 329 sentence contexts. Memory & Cognition 8 (6), 631 642. Bouq uet, C. A., Bonnaud, V., & Gil, R. (2003). Investigation of supervisory attentional Journal of Clinical and Experimental Neuropsychology 25 (6), 751 760. Burgess, P. W., & Shalli ce, T. (1996). Response suppression, initiation and strategy use following frontal lobe lesions. Neuropsychologia 34 (4), 263 272. doi:10.1016/0028 3932(95)00104 2 Burgess, P. W., & Shallice, T. (1997). The Hayling and Brixton tests Bury St Edmunds, UK: T hames Valley Test Company Limited. Cabeza, R. (2002). Hemispheric asymmetry reduction in older adults: the HAROLD model. Psychol Aging 17 (1), 85 100.
72 Castner, J. E., Copland, D. A., Silburn, P. A., Coyne, T. J., Sinclair, F., & Chenery, H. J. (2007). Lexi cal function of subthalamic stimulation. Neuropsychologia 45 (14), 3167 77. Chambers, C. D., Bellgrove, M. A., Gould, I. C., English, T., Garavan, H., McNaught, E., Kamke, M., et al. (2007). Dissoc iable Mechanisms of Cognitive Control in Prefrontal and Premotor Cortex. J Neurophysiol 98 (6), 3638 3647. doi:10.1152/jn.00685.2007 Chenery, H. J., Angwin, A. J., & Copland, D. (2008). The basal ganglia circuits, dopamine, and ambiguious word processing : A neurobiological account of Journal of the International Neuropsychological Society 14 in press. Collette, F., Van der Linden, M., Delfiore, G., Degueldre, C., Luxen, A., & Salmon, E. (2001). The Functional Anat omy of Inhibition Processes Investigated with the Hayling Task. NeuroImage 14 (2), 258 267. doi:10.1006/nimg.2001.0846 Copland, D. (2003). The basal ganglia and semantic engagement: potential insights from semantic priming in individuals with subcortical v ascular lesions, J Int Neuropsychol Soc 9 (7), 1041 52. Cox, R. W. (1996). AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Computers and Biomedical Research, an Internat ional Journal 29 (3), 162 173. Crosson, Bruce, Benjamin, M., & Levy, I. (2007). Role of the basal ganglia in language and semantics: Supporting cast. In J. H. Jr. & M. A. Kraut (Eds.), Neural Basis of Semantic Memory New York: Cambridge University Press. Elsinger, C. L., Rao, S. M., Zimbelman, J. L., Reynolds, N. C., Blindaur, K. A., & Hoffmann, R. G. (2003). Neural Basis for Impaired Time Reproduction in Journal of the International Neuropsychological Society 9 (07), 10 88 1098. doi:10.1017/S1355617703970123 Garavan, H., Ross, T. J., Murphy, K., Roche, R. A. P., & Stein, E. A. (2002). Dissociable Executive Functions in the Dynamic Control of Behavior: Inhibition, Error Detection, and Correction. NeuroImage 17 (4), 1820 18 29. doi:10.1006/nimg.2002.1326 Gerfen, C. R. (1992). The Neostriatal Mosaic: Multiple Levels of Compartmental Organization in the Basal Ganglia. Annual Review of Neuroscience 15 (1), 285 320. doi:10.1146/annurev.ne.15.030192.001441 Hafkemeijer, A., van der Grond, J., & Rombouts, S. A. R. B. (2011). Imaging the default mode network in aging and dementia. Biochimica Et Biophysica Acta doi:10.1016/j.bbadis.2011.07.008
73 Kaasinen, V., & Rinne, J. O. (2002). Functional imaging studies of dopamine system and cogni Neuroscience & Biobehavioral Reviews 26 (7), 785 793. doi:10.1016/S0149 7634(02)00065 9 Kelly, A. M. C., Hester, R., Murphy, K., Javitt, D. C., Foxe, J. J., & Garavan, H. (2004). Prefrontal subcortical dissocia tions underlying inhibitory control revealed by event related fMRI. European Journal of Neuroscience 19 (11), 3105 3112. doi:10.1111/j.0953 816X.2004.03429.x Kischka, U., Kammer, T., Maier, S., Weisbrod, M., Thimm, M., & Spitzer, M. (1996). Dopaminergic mo dulation of semantic network activation. Neuropsychologia 34 (11), 1107 1113. Koch, W., Teipel, S., Mueller, S., Buerger, K., Bokde, A. L. W., Hampel, H., Coates, U., et al. (2010). Effects of aging on default mode network activity in resting state fMRI: d oes the method of analysis matter? NeuroImage 51 (1), 280 287. doi:10.1016/j.neuroimage.2009.12.008 McGregor, K. M., Zlatar, Z., Kleim, E., Sudhyadhom, A., Bauer, A., Phan, S., Seeds, L., et al. (2011). Physical activity and neural correlates of aging: a c ombined TMS/fMRI study. Behavioural Brain Research 222 (1), 158 168. doi:10.1016/j.bbr.2011.03.042 Meinzer, M., Seeds, L., Flaisch, T., Harnish, S., Cohen, M. L., McGregor, K., Conway, T., et al. (in press). Impact of changed positive and negative task rel ated brain activity on word retrieval in aging. Neurobiology of Aging doi:10.1016/j.neurobiolaging.2010.06.020 Mink, J. W. (1996). The Basal Ganglia: Focused Selection and Inhibition of Competing Motor Programs. Progress in Neurobiology 50 381 425. More tti, R., Torre, P., Antonello, R. M., Capus, L., Marsala, S. Z., Cattaruzza, T., Cazzato, G., et al. (2003). Neuropsychological changes after subthalamic nucleus stimulation: a 12 month follow disease. Parkinsonism Rela t Disord 10 (2), 73 9. Nambu, A., Tokuno, H., & Takada, M. (2002). Functional significance of the cortico subthalamo Neurosci Res 43 (2), 111 7. Nambu, A., Tokuno, H., Takada, M., Hamada, I., Kita, H., Imanishi, M., Akazawa, T., et al. (2000). Excitatory Cortical Inputs to Pallidal Neurons Via the Subthalamic Nucleus in the Monkey. The Journal of Neurophysiology 84 289 300. Nathaniel James, D. A., Fletcher, P., & Frith, C. D. (1997). The functional anatomy of verbal initiation and suppression us ing the Hayling Test. Neuropsychologia 35 (4), 559 566. doi:10.1016/S0028 3932(96)00104 2
74 Nathaniel James, D. A., & Frith, C. D. (2002). The Role of the Dorsolateral Prefrontal Cortex: Eviden ce from the Effects of Contextual Constraint in a Sentence Completion Task. NeuroImage 16 (4), 1094 1102. doi:10.1006/nimg.2002.1167 Oldfield, R. C. (1971). The Assessment And Analysis of Handedness: The Edinburgh Inventory. Neuropsychologia 9 97 113. Penney, J. B., & Young, A. B. (1986). Striata l inhomogeneities and basal ganglia function. Movement Disorders: Official Journal of the Movement Disorder Society 1 (1), 3 15. doi:10.1002/mds.870010102 Smith, S. M., Jenkinson, M., Woolrich, M. W., Beckmann, C. F., Behrens, T. E. J., Johansen Berg, H., Bannister, P. R., et al. (2004). Advances in functional and structural MR image analysis and implementation as FSL. NeuroImage 23 Suppl 1 S208 219. doi:10.1016/j.neuroimage.2004.07.051 Stuss, D.T., Alexander, M. P., Floden, D., Binns, M. A., Levine, B., Mcintosh, A. R., Rajah, N., et al. (2002). Fractionation and Localization of Distinct Frontal Lobe Processes: Evidence from Focal Lesions in Humans. In Donald T. Stuss & R. T. Knight (Eds.), Principles of Frontal Lobe Function (pp. 392 407). New York: Oxfo rd University Press. Retrieved from http://dx.doi.org/10.1093/acprof:oso/9780195134971.001.0001 Wierenga, C. E., Benjamin, M., Gopinath, K., Perlstein, W. M., Leonard, C. M., Rothi, L. J., Conway, T., et al. (2008). Age related changes in word retrieval: r ole of bilateral frontal and subcortical networks. Neurobiol Aging 29 (3), 436 51. Wilson, M. D. (1988). The MRC Psycholinguistic Database: Machine Readable Dictionary, Version 2. Behavioural Research Methods, Instruments and Computers 20 (1), 6 11. Zlatar Z. Z., McGregor, K. M., Dzierzewski, Joe M, Bauer, Andrew, Phan, S., Towler, Stephen, Cohen, Matthew, et al. (unpublished). Functional language networks in sedentary and physically active older adults.
75 BIOGRAPHICAL SKETCH Ilana Fawn Levy received her Bachelor of Arts degree in cognitive science from the University of Virginia (Charlottesville) in May 2003. Following graduation Ilana worked at the National Institutes of Mental Health in the Laboratory of Brain and Cognition from 2003 to 2005 after recei ving a post baccalaureate intramural research training award from the NIH. She received her Ph.D. from the Clinical and Health Psychology doctoral program at the University of Florida in 201 2 with a concentration in clinical neuropsychology. While at UF, s he worked in the Brain Imaging Rehabilitation and Cognition lab under the mentorship of Bruce Crosson, Ph.D studying the neural correlates of language processing Ilana completed her pre doctoral Internship at the Boston Consortium through the VA healthca re system in Boston.