<%BANNER%>

A Tale of Two Stroops in Traumatic Brain Injury


PAGE 1

A TALE OF TWO STROOPS IN TRAUMATIC BRAIN INJURY By PAUL SEIGNOUREL A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2003

PAGE 2

Copyright 2003 by Paul J. Seignourel

PAGE 3

ACKNOWLEDGMENTS I thank my advisor, William M. Perlstein; and my collaborators on this project, Michael A. Cole, Jason A. Demery, and Diana L. Robins. iii

PAGE 4

TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iii LIST OF TABLES.............................................................................................................vi LIST OF FIGURES..........................................................................................................vii ABSTRACT..................................................................................................................... viii CHAPTER 1 INTRODUCTION........................................................................................................1 Performance of Brain-Injured Patients on the Stroop Task..........................................6 Predictions..................................................................................................................10 2 METHODS.................................................................................................................14 Participants.................................................................................................................14 Materials.....................................................................................................................15 Card Stroop..........................................................................................................15 Single-Trial Stroop..............................................................................................16 Neurobehavioral Rating Scale (NRS).................................................................17 WAIS-III subtests................................................................................................19 North American Adult Reading Test (NAART).................................................19 Procedures...................................................................................................................19 Statistical Analyses.....................................................................................................19 3 RESULTS...................................................................................................................22 Neuropsychological Tests...........................................................................................22 Card Stroop.................................................................................................................22 Single-Trial Stroop: Speed-Accuracy Tradeoff Analyses..........................................23 Single-Trial Stroop: RTs.............................................................................................24 Single-Trial Stroop: Error Rates.................................................................................25 Neurobehavioral Rating Scales...................................................................................27 4 DISCUSSION.............................................................................................................32 iv

PAGE 5

LIST OF REFERENCES...................................................................................................44 BIOGRAPHICAL SKETCH.............................................................................................51 v

PAGE 6

vi LIST OF TABLES Table page 1-1 Studies of performance of TBI patients on the Stroop task.......................................8 2-1 Demographics and injury variables by group..........................................................16 3-1 Performance on neuropsyc hological tests by group................................................22 3-2 Means and standard deviations of medi an RT (ms) in the single-trial Stroop........24 3-3 Means and standard deviations of e rror rates (%) in the single-trial Stroop...........26 3-4 ANOVA for error rates in the single-trial Stroop....................................................26 3-5 Internal consistency (alpha-cronbach) estimates for the NRS for self and significant others ratings by group.........................................................................28 3-6 Means and standard deviations of NRS surrogate factor scores and total score by group...................................................................................................................28 3-7 Correlations between self and si gnificant-others ratings on the NRS....................29 3-8 Correlations between ratings on the NRS and Stroop performance (TBI group)....30 3-9 Multiple regressions with self and si gnificant-others ratings as independent variables and error ra tes (color naming) as the de pendent variables (TBI group)..30 3-10 Correlations between self -ratings on the NRS and color-n aming error rates at the long and short delays separately (TBI group)..........................................................31 3-11 Multiple regressions with color-naming erro r rates at the long and short delays as independent variables and NRS self-ratings as dependent variables (TBI group)..31

PAGE 7

LIST OF FIGURES Figure page 2-1 Schema of the single-trial Stroop. Participants are first presented with an instructional cue (Color or Word), followed by a delay (1 or 5 seconds) and an imperative stimulus (congruent, neutral or incongruent).....................................17 3-1 Number of items completed on the card Stroop by group and condition. Error bars represent standard errors...........................................................................................23 3-2 Error rates in the incongruent color-naming condition of the single-trial Stroop by group and delay. Error bars represent standard errors..............................................27 vii

PAGE 8

Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science A TALE OF TWO STROOPS IN TRAUMATIC BRAIN INJURY By Paul J. Seignourel May 2003 Chair: William M. Perlstein Major Department: Clinical and Health Psychology The Stroop Color-Word task has frequently been used to examine attentional deficits in Traumatic Brain Injury (TBI). However, critical review of the literature suggests that TBI patients do not consistently show disproportionate reaction time (RT) interference on the Stroop task, inconsistent with a frequent assertion of increased interference. We hypothesized that combining inhibition and working memory (WM) requirements in a new, computerized version of the Stroop would increase sensitivity to TBI. We examined the performance of healthy subjects and TBI patients on card and single-trial versions of the Stroop paradigm. The TBI patients did not show disproportionately increased RT interference compared to controls on either version, but instead greater error-rate interference on the single-trial version, especially at the long delay. Error rates, but not reaction times, were related to reports of symptomatology in TBI patients. Results suggest that errors (rather than RTs) are more sensitive to cognitive deficits in TBI; and that such deficits, and everyday life difficulties experienced by TBI viii

PAGE 9

survivors, may be related to a difficulty in the preparation to override prepotent response tendencies. ix

PAGE 10

CHAPTER 1 INTRODUCTION Traumatic brain injury (TBI), with an incidence of 500,000 to 1.9 million cases each year (Lezak, 1995), is a significant problem in the United States. It is also a unique problem, in that it affects large numbers of young individuals: the highest incidence of brain injury occurs in the 15 to 24 year range, with high incidence rates also in the 0 to 5 year range and for the elderly. Among individuals who sustain TBI, some experience only transient symptoms, such as brief loss of consciousness (LOC) and post-traumatic amnesia (PTA). Others, however, continue to experience significant problems long after injury. These symptoms include frequent headaches, irritability and restlessness, sleep disturbances, and various cognitive difficulties (such as poor concentration and attention, reduced processing speed, and impaired memory) (Levin, Eisenberg, & Benton, 1991; Levine, 1988; van Zomeren & van den Burg, 1985): a constellation of symptoms sometimes referred to as postconcussional syndrome (Lishman, 1988). In a study of 103 patients with moderate-to-severe brain injuries, Olver, Ponsford and Curran (1996) found that at a 2-year follow-up, 61% of the patients reported word-finding problems, 60% had difficulties concentrating and 56% reported feeling more depressed than before the injury. Moreover, only 40% were able to drive and only 41% were employed full-time. Percentages were similar at 5-year follow-up, with the exception of ability to drive, which increased to 48%; and full-time employment, which decreased to 34%. Such findings suggest that TBI, at least in the moderate-to-severe range, results in disabling and long-lasting difficulties for a great number of individuals. 1

PAGE 11

2 Closed head injuries, in which the skull is intact and the brain is not exposed, account for about 90% of all brain injuries (Lezak, 1995). Anatomically, brain damage after closed head injury is generally diffuse, resulting from several distinct damaging mechanical forces to many different areas of the brain. Some evidence exists, however, that because of the shape of bones protuberances in the brain, bruising due to coup (the blow at the point of impact) and contrecoup (bruising in the area opposite to the blow) is often more pronounced in the frontal and temporal lobes (Levin & Kraus, 1994). Axonal shearing or diffuse axonal injury (DAI), which is an excessive stretching of nerve fibers and blood vessels due to violent acceleration and deceleration (Adams, Graham, Murray, & Scott, 1982), might be more concentrated in these areas as well (Groswasser, Reider-Groswasser, Soroker, & Machtey, 1987). Consistent with evidence of frontal damage, studies of individuals with TBI often show difficulties on complex tasks, despite normal or quasi-normal functioning in basic cognitive domains such as arousal, language and perception. For example McDowell, Whyte and DEsposito (1997) compared brain-injured patients with control participants on a simple visual reaction time (RT) task and on a dual task, consistent with patient reports of difficulties negotiating multiple simultaneous real-world tasks (van Zomeren & van den Burg, 1985). They found that, in addition to generalized slowing, TBI patients showed greater decrements in performance compared to control participants during dual-task conditions. Similarly, Levine, Dawson, Boutet, Schwartz and Stuss (2000) found that TBI patients were impaired on the Revised Strategy Application Test (R-SAT), a complex and unstructured task in which participants, in order to perform satisfactorily, must select items on the basis of their length while refraining from the tendency to complete all items sequentially.

PAGE 12

3 Despite major differences, the two previously mentioned tasks share at least one essential characteristic: in both cases, participants tend to perform poorly if they complete all items automatically (that is, by simply responding to each item presented to them and without any overall strategy toward optimal performancebottom-up processing). Conversely, good performance depends on participants strategy being guided by the overall goal and structure of the task (top-down processing). In other words, both tasks require that participants actively maintain goal representations to guide their behavior and the order in which they perform the task. Using a more general framework, Braver and Barch (2002page 1) defined context representations as any task-relevant information that is generally represented in such a form that it can bias processing in the pathways responsible for task performance. Through extensive behavioral studies and connectionist computational models, Cohen, Barch, Carter and Servan-Schreiber (1999) showed that schizophrenic patients cognitive difficulties on a range of seemingly dissimilar tasks could be explained by a single deficit in context maintenance. According to their model, context representations are maintained on-line by the prefrontal cortex (PFC), through a mechanism where dopamine (DA) plays a prominent role; the putative disruptions in the dopaminergic circuitry in schizophrenic patients are thought to be primarily responsible for their context maintenance deficits. More recently, Braver et al. (2001) showed that performance decrements and, counterintuitively, performance improvements in older adults could also be explained by a deficit in context maintenance. Importantly, the term context maintenance, in this framework, refers to a concept more general than goal representations: goals represent only one type of representation, among others, that can bias processing of subsequent information. In the AX-Continuous

PAGE 13

4 Performance Task (AX-CPT), for instance, the cue provides the context essential for proper processing of the probe, without representing per se the goal of the task (Cohen et al., 1999). Context maintenance, on the other hand, is a much more specific concept than executive functioning, which has been used to describe a number of abilities including planning, problem-solving, shifting mental set, and inhibition of prepotent responses. In fact, Cohen, Braver and OReilly (1996) noted that the term dysexecutive syndrome (often used to describe the set of symptoms exhibited by patients with frontal lobe lesions) is essentially descriptive, and does not specify which mechanisms are responsible for executive control. 1 They added that their framework, by specifying the mechanisms responsible for cognitive control at a computational and neurobiological level, leads to specific predictions regarding the performance of individuals with context maintenance deficits. For the purpose of our study, two of these predictions are especially relevant. First, representations of context are especially important when the response required by the task at hand is competing with a more frequent and automatic response. This characteristic of context maintenance is similar to what has been called inhibition of the prepotent response. It is required in the Stroop task, used in this study, but also in the antisaccade task (Crevits, Hanse, Tummers, & Van Maele, 2000) and in the R-SAT (Levine et al., 2000), previously mentioned. Second, deficits in context maintenance should be more 1 Context maintenance may not account for all mechanisms involved in executive control. In the framework described by Cohen and Servan-Schreiber (1992), the maintenance of context representations is carried out by the prefrontal cortex (PFC) and mediated by DA. Carter et al. (2000), however, showed that the role of the anterior cingulate (AC) in cognitive control is evaluative rather than strategic. In other words, the AC might be another area of the brain responsible for cognitive control, complementary to the maintenance of context carried out by the PFC.

PAGE 14

5 pronounced when the task requires participants to hold context representations in mind for a prolonged period of time. This second characteristic of context maintenance shows its similarity with working memory; in fact, Braver et al. (2001) recently described the maintenance of context representations as a subset of representations within working memory. In support of this hypothesis, Cohen et al. (1996) compared the performance of schizophrenics and control participants on the AX-CPT; they found that the sensitivity of schizophrenics to the cue decreased after a long delay compared to a short delay between the cue and the probe; whereas the performance of control participants was not affected by delay. Moreover, the same pattern of results was observed when comparing two neural network simulations designed to model normal performance and performance associated with a reduced gain in the context layer. Cohen et al. (1999), on the other hand, failed to find significant delay effects for both schizophrenics and healthy participants on a single-trial version of the Stroop paradigm we used. The primary goal of our study was to examine the hypothesis that patients with moderate-to-severe TBI exhibit difficulties maintaining context representations, as evidenced by poor performance when they must maintain representations over a delay in order to inhibit prepotent response tendencies. This hypothesis stems from the relation among context maintenance, PFC, and dopaminergic neuromodulation, on the one hand; and from the aforementioned finding that closed head injuries often result in damage to the frontal lobes, on the other hand. It is also consistent with the various symptoms experienced by such patients (some of which, including lack of behavioral control, disorganization, attention difficulties, and memory deficits, suggest dysfunction of the frontal lobes) (Kraus & Maki, 1997; Levin & Kraus, 1994; Rieger & Gauggel, 2002).

PAGE 15

6 More specifically, our study used a single-trial version of the Stroop task, which places greater demands on working memory than the card Stroop often used in clinical settings. Before we describe the tasks in detail and state our specific predictions, we review the evidence that TBI critically affects performance on the Stroop task. Performance of Brain-Injured Patients on the Stroop Task The Stroop task has been considered by several researchers to be a prototypical instrument for measuring prepotent response inhibition 2 (Miyake, Emerson, & Friedman, 2000; Miyake, Friedman et al., 2000), a function often attributed, in part, to the frontal lobes (Jahanshahi et al., 1998; Kiefer, Marzinzik, Weisbrod, Scherg, & Spitzer, 1998). Although there are many different versions of the original task (MacLeod, 1991), the basic paradigm consists of naming the color of the ink of an incongruent color word. For example, one would be presented with the word BLUE printed in red ink. In order to provide the correct answer, red, the participant must overcome the natural, more automatic or prepotent tendency to simply read the word and respond blue. A version of the Stroop task most widely used in clinical settings (Golden, 1978) comprises three conditions: Word reading, in which participants must read aloud columns of color words. Color naming, in which participants are required to name the color of the ink of rows of Xs. 2 Although some researchers refer to the Stroop task as a measure of selective or focused attention (see e.g., Batchelor, Harvey, & Bryant, 1995; Kingma, La Heij, Fasotti, & Eling, 1996; Salo, Avishai, & Lynn, 2001), we find the term inhibition of a prepotent response more specific. Selective attention is the ability to focus ones attention on a specific stimulus or a specific dimension of a stimulus among an array of distractors. Not included in this definition is a key component of the Stroop, which is the fact that in this task the distractor (i.e., the color word) tends to elicit an automatic (and in this case incorrect) response; whereas the correct response (naming the color of the ink) is much less natural and automatic.

PAGE 16

7 Color-word naming (or interference), in which participants must name the color of the ink of incongruent color words. Studies examining TBI patients performance on the Stroop paradigm have relied on a variety of task versions and scoring methods (Table 1-1). In some studies, researchers have used a single measure (generally performance on the color-word-naming condition) to compare brain-injured patients and control participants. These researchers have generally found that TBI patients were slower than control participants (Guskiewicz, Riemann, Perrin, & Nashner, 1997; Rojas & Bennett, 1995; Stuss et al., 1985; Trennery, Crosson, DeBoe, & Leber, 1989), although not always significantly so (Guskiewicz et al., 1997; Potter, Jory, Bassett, Barrett, & Mychalkiw, 2002). Interpretation of these results, however, is uncertain. Although researchers often consider slower performance in the interference condition of the Stroop task as a sign of deficits in inhibiting a prepotent response (or selective attention), it may simply result from generalized slowing, a very common cognitive sequela of TBI (Lezak, 1995). Researchers using the difference score between the color-word naming and the color-naming conditions have obtained mixed results. In three studies, TBI patients had a significantly greater mean interference score than control participants (Bohnen, Twijnstra, & Jolles, 1992; McDowell et al., 1997). In three other studies, however, there was no significant difference between the two groups (Batchelor et al., 1995; Bate, Mathias, & Crawford, 2001; Bohnen, Jolles, & Twijnstra, 1992), and in one study,

PAGE 17

8 Table 1-1. Studies of performance of TBI patients on the Stroop task Authors Participants Tasks a Score Results Asikainen et al. (1999) 92 severe TBI, 22 moderate TBI CWN CWN No significant diff. Batchelor et al. (1995) 35 mild TBI, 35 controls WR, CN, CWN, mod. CWN CWN CN No significant diff. Bate et al. (2001) 35 severe TBI, 35 controls WR, CN, CWN, mod. CWN CWN CN No significant diff. Bohnen et al. (1992) 44 mild TBI, 44 controls WR, CN, CWN, mod. CWN CWN CN No significant diff. Bohnen et al. (1992) 10 mild TBI with PCS, 10 mild TBI without PCS WR, CN, CWN, mod. CWN CWN CN Greater interference in TBI patients Guskiewicz et al. (1997) 11 athlete mild TBI, 11 athlete controls WR, CN, CWN CWN No significant diff. McDowell et al. (1997) 25 moderate-to-severe TBI, 24 controls CN, CWN CWN CN Greater interference in TBI Ponsford & Kinsella (1992) 47 severe TBI, 30 orthopoedic patients WR, CN, CWN CWN CN Errors Greater interference in controls No significant diff. Potter et al. (2002) 24 mild TBI, 24 controls CWN, PC cong., PC incong. CWN PC incong. Errors No significant diff. No significant diff. No significant diff. Rojas & Bennett (1995) 25 mild TBI, 25 controls CWN, WCN CWN TBI slower than controls Stuss et al. (1985) 20 CHI, 20 controls WR, CN, CWN CN CWN TBI slower than controls No significant diff. Trennery et al. (1989) 65 TBI, 106 controls CN, CWN, WCN CWN TBI slower than controls Vakil et al. (1995) 25 TBI (all severity levels) 27 controls CN, CWN CWN-CN Greater interference in TBI a WR = word reading; CN = color naming (naming the color of a row of Xs); CWN = color-word naming (naming the color of incongruent color words); mod. CWN = modified color-word naming, in which a few word-reading items are interspersed among incongruent color-word naming items; WCN = word-color reading (reading color words printed in incongruent colors); PC cong. and PC incong. = computer version where congruency varies with each trial but task remains constant.

PAGE 18

9 control participants actually had a greater interference score than brain-injured patients (Ponsford & Kinsella, 1992). 3 Finally, researchers examined error rates in only two studies. Ponsford and Kinsella (1992) failed to find any difference between control participants and TBI patients on a card version of the Stroop. Potter et al. (2002) found no significant difference between the two groups on either a card version or a computer version of the Stroop. Three main possible explanations may account for these mixed findings. First, it may be that TBI patients do not actually have deficits in inhibiting prepotent response tendencies, but rather difficulties on many tasks due to nonspecific, generalized slowing. Although this explanation may account for the results obtained with the Stroop task in brain-injured patients, the findings by McDowell, Whyte and DEsposito (1997) and Levine et al. (2000) suggest that these patients do have more difficulties with tasks requiring some form of cognitive control, of which inhibition of prepotent response tendencies is a prominent example. Secondly, even though TBI patients may have difficulties with inhibitory processes, the Stroop paradigm may not be sensitive to these deficits. Finally, the Stroop paradigm may have the potential to detect deficits in TBI patients, but the versions of the task used in previous studies may not be sensitive to these deficits. In support of this explanation, all the tasks used in previous research on TBI were card versions of the Stroop paradigm. As pointed out by Perlstein, Carter, Barch and Baird (1998), these versions require participants to correct mistakes during task performance. As a result, errors and speed are confounded in the RT data, leading to a possible decrease in sensitivity. Possibly even more crucial is the fact that conditions are blocked in the card version of the Stroop. In other words, the task requires participants to read words on a first page, and name colors on a different page. As a result, it may become easier, over time, to overcome the influence of the words of incongruent stimuli and to focus instead 3 Bohnen and his collaborators (Bohnen, Jolles et al., 1992; Bohnen, Twijnstra et al., 1992) have also developed a new condition, in which a few word-reading items are interspersed among incongruent color-word naming items. For this new condition, they found an increased interference score in brain-injured patients in at least two independent studies. Interestingly, their modification is not without commonalities with our computer version, as in both cases participants have to switch rapidly between reading words and naming colors, thereby increasing dramatically the context maintenance requirements of the task.

PAGE 19

10 on the color of the ink. As noted previously, deficits in context maintenance are more likely to become apparent in tasks with greater working memory demands. In a task with mixed conditions, such as the computer version used in our study, color naming and word reading vary randomly with each new trial, and participants must constantly update, maintain and manipulate task instructions to perform appropriately. If the difficulties of brain-injured patients are indeed due to impairments in maintaining context representations, then such a task, combining working memory and inhibition of prepotent response tendency requirements, may be better suited than card versions of the Stroop paradigm to detect deficits in context maintenance due to TBI. Predictions In our study, we examined the performance of TBI patients on two versions of the Stroop paradigm: the classic three-card version of Golden (1978), often used in the clinical context, and a single-trial computerized version developed by Cohen et al. (1999). In the single-trial version, the conditions varied with each trial, and participants were first given an instructional cue (reading the word or naming the color), and then shown an imperative stimulus varying in congruency. Previous single-trial computer versions of the Stroop typically consisted of three color-naming conditions of the present design (Perlstein et al., 1998; Salo et al., 2001). Our addition of the three word-reading conditions requires participants to maintain the instruction on-line and use it to bias the processing of the imperative stimulus. In other words, this modification introduces a working-memory component to the basic Stroop paradigm, which we believe will increase its sensitivity to context maintenance deficits induced by TBI. To examine even more closely the influence of working memory on task performance, we also manipulated the delay between the instruction and the imperative stimulus, with a short, 1-second delay and a long, 5-second delay (Figure 2.1 shows the variables in the single-trial Stroop task).

PAGE 20

11 Our main hypothesis stems from our review of the literature and the characteristics of these two tasks: we predicted that the computer-based, single-trial Stroop (but not the card Stroop) would be sensitive to context maintenance deficits in TBI patients. More specifically, we predicted that TBI patients would be slower than control participants in each of the three conditions of the card Stroop (i.e., generalized slowing), but that they would not be disproportionately slower in the incongruent condition (no increased difference score). Conversely, for the single-trial Stroop, we predicted that, in addition to generalized slowing, TBI patients would show disproportionately longer RTs than control participants in the incongruent, color-naming condition, resulting in an increased difference score between incongruent and neutral conditions. In terms of error rates, we expected TBI patients to make more errors than control participants only in the incongruent color-naming condition. We also predicted that, due to a degradation of context representations, brain-injured patients would make more errors at the long than short delay, whereas the performance of control participants would not be influenced by delay (resulting in an interaction between group and delay in the incongruent, color-naming condition). 4 An additional goal of our study was to examine the relation between performance of TBI patients on the Stroop task and subjective reports of symptomatology. As imaging techniques proliferate and become a standard tool for evaluating brain trauma, the role of neuropsychologists in clinical practice tends to move away from lesion localization to a 4 Unlike for RT data, we did not compute an interference score (incongruent neutral) for error rates, because we expected very low error rates in the neutral condition. Instead, our prediction regarding a differential deficit of TBI patients at the long delay represented our attempt to disentangle overall performance impairment from specific deficits in maintenance of context representations.

PAGE 21

12 more refined description of patients functional impairments and potential for rehabilitation (Groth-Marnat, 2000). In the future, patients needs and pressure from insurance companies will most likely result in a demand for neuropsychological tests to become better predictors of symptomatology in brain-injured patients. To our knowledge, until now, only one study (Levine et al., 2000) has examined the relationship between Stroop performance and reports of impairment after TBI. Levine et al. have found a small correlation (R 2 = 0.14) between Stroop RT interference score and total score on the Sickness Impact Profile, a multi-dimensional measure of health status for individuals with general medical conditions (Bergner, Bobbitt, Pollard, Martin, & Gilson, 1976; Pollard, Bobbitt, Bergner, Martin, & Gilson, 1976). In our study, we used self-reports and significant-other reports of symptomatology obtained on a modified version of the Neurobehavioral Rating Scale (NRS), a 27-item measure designed specifically to assess impairments after brain injury. We predicted that performance on the single-trial Stroop, being more sensitive to impairments in TBI, would be more strongly correlated than performance on the card Stroop with reports of symptomatology. In addition, we were interested in the following questions: In the single-trial Stroop, would RTs and errors contribute independently to predicting reports of impairments, or would one of them account for most of the predicted variance in the NRS? Would errors be a better predictor of NRS scores at the long than at the short delay, suggesting that performance after a delay (i.e., with increased working memory requirements) is more strongly related to reports of impairment in TBI? Would self-reports or significant-other reports be most strongly correlated with Stroop performance? In relation to this last question, Sunderland, Harris and Baddeley (1983) found that significant-others ratings of TBI patients memory difficulties correlated more strongly than self-reports with their performance on standard memory tests. If such results were replicated in our study, it would suggest that significant others, when present, may be better informants of impairment after

PAGE 22

13 brain injury than the patients themselves, a finding that may have important implications for the assessment and management of TBI.

PAGE 23

CHAPTER 2 METHODS Participants Participants were 23 individuals with moderate-to-severe closed-head injury and 21 demographically-matched, healthy participants recruited through advertising in the local community. Consistent with the definition of Alexander (1995), we defined injuries of the moderate-to-severe range as any injury resulting in loss of consciousness (LOC) longer than 30 minutes, post-traumatic amnesia (PTA) longer than 24 hours or presence of lasting focal signs and positive neuroimaging findings. 1 We did not distinguish between moderate and severe brain injury. Data were collected from interviews with TBI patients at the time of the experiment. Because most patients did not remember their score on the Glasgow Coma Scale (GCS) at the time of injury, this variable was not used in our classification of severity. Excluded from the study were all individuals below 18 or above 55 years of age, and all participants reporting a history of schizophrenia or bipolar disorder, a history of formally diagnosed attention-deficit/hyperactivity disorder, a history of learning disability, a history of chronic alcohol or drug abuse extending within a 6-months period before testing, a history of other acquired brain damage (e.g., stroke, epilepsy), a history of inpatient psychiatric hospitalization predating brain injury, or a history of significant depression or anxiety predating brain injury and extending within a 1 In our sample, all of the participants classified as moderate-to-severe TBI, but one reported an LOC greater than 30 minutes or a PTA greater than 24 hours. This participant reported an LOC and PTA of 1 minute, but presented with linear skull fracture, subdural hematoma and significantly impaired senses of smell and taste. 14

PAGE 24

15 2-year period before testing. In addition, brain-injured partic ipants had to be at least 1 month after injury. Exclusion based on cu rrent medication use was determined on a case-by-case basis, depending on whether the medication had documented effects on concentration and attention. Demographic variables of brain-injured pati ents and control participants are shown in Table 2-1. The two groups did not di ffer significantly in terms of gender, 2 (1, 42), p > 0.5, age, t (42) = -1.8, p > 0.08, education, t (42) = 0.2, p > 0.8, mothers education, t (42) = 1.5, p > 0.1, and fathers education, t (42) = -0.6, p > 0.5. Median scores (ranges) for time since injury, LOC, and PTA in the TBI group were 65 months (1.5 to 444 months), 72 hours (1 minute to 2160 hours) and 692 hours (1 minute to 4320 hours), respectively. Materials Card Stroop We used the three-card version of the Str oop test most frequently used in clinical settings (Golden, 1978). This version includes three cards, administered in fixed order, that consist of five columns of 20 items each. The first card contains lists of color words (BLUE, GREEN and RED) printed in black ink, which participants were instructed to read aloud as quickly and as accurately as possible (word-reading condition). The second card contains rows of four colored Xs and part icipants were instructed to name the color of the printed Xs (color-naming condition). Th e third card contains lists of words printed in incongruent colors (e.g., BL UE printed in red ink), and pa rticipants were required to name the printed color of the ink (colo r-word-naming or conflic t condition). For each card, participants were given 45 seconds to co mplete as many items as possible, without skipping any. Whenever particip ants made a reading or nami ng error, they were briefly instructed to corre ct their response.

PAGE 25

16 Table 2-1. Demographics and injury variables by group Controls participants ( n = 21) TBI patients ( n = 23) Males/females 12/9 15/8 M SD M SD Age (years) 36.3 10.1 42.0 11.2 Education (years) 14.0 1.6 13.9 2.0 Mothers education (years) 13.9 3.0 12.7 2.3 Fathers education (years) 13.6 3.4 14.1 2.6 For the different subtests of the card Stroop, Golden (1975) found test-retest reliabilities ranging from 0.73 to 0.89 in a normal sample. Golden (1976) also found the card Stroop to be fairly accurate in distinguishing brain-injured individuals from normal participants and psychiatric patients. Single-Trial Stroop Participants also performed a computer-based, single-trial version of the Stroop task, originally developed by Cohen et al. (1999). Stimuli were presented on an Apple Macintosh computer using PsyScope software (Cohen, MacWhinney, Flatt, & Provost, 1993). At the beginning of each trial, they were presented with an auditory cue (Word or Color), followed by a visual stimulus. They were instructed to respond verbally to the stimulus, and their RTs were determined by a voice-activated relay connected to the computer. The examiner recorded responses manually for coding of accuracy. The same basic three colors and color words (blue, red, and green) as in the card Stroop were used. There were two task conditions (color naming and word reading), two delay conditions (corresponding to a delay of 1s and 5s, respectively, between the cue and the probe); and three congruency conditions (congruent, neutral, and incongruent); for a total of 12 trial types (Figure 2-1). In the congruent condition, stimuli were color words presented in their own color ink. Incongruent stimuli consisted of color words printed in one of the two

PAGE 26

17 remaining colors, and neutral stimuli were four colored Xs for color-naming trials and color words displayed in white for word-reading trials. 2 A total of 180 experimental trials were distributed randomly across trial types, resulting in an average number of 15 trials of each type. Because of random effects, this number was not always equal to 15, but fell between 8 and 24. Because of its novelty, reliability and validity estimates were not available for this computer-based task. Instructions Delay Congruency Incongruent Neutral Color Word Congruent Long Short Figure 2-1. Schema of the single-trial Stroop. Participants were first presented with an instructional cue (Color or Word), followed by a delay (1 or 5 seconds) and an imperative stimulus (congruent, neutral or incongruent). Neurobehavioral Rating Scale (NRS) A modified version of the NRS was used to estimate current functioning in TBI patients and control participants. This 27-item instrument was originally based on clinician ratings, and was specifically designed to estimate cognitive, behavioral, and emotional changes after brain injury (Levin et al., 1987). We used the modified version, 2 Although one could argue that white is also a color, and that neutral stimuli in the word-reading condition were not really neutral, the important point here is that White was not part of the response set in any of the trial types. A further indication that the white color did not substantially interfere with the reading of the color words is that none of our participants ever answered White to any of the stimuli.

PAGE 27

18 developed by Mathias and Coats (1999), to enable self-report. Each of the 27 items was rated using a 5-point scale (from never to very often). In addition to self-report ratings, each participant was given a copy of the scale to be completed by a significant other. Previous versions of the NRS (Levin et al., 1987) showed high inter-rater reliability (0.88 and 0.90 for two different samples of brain-injured patients) and internal consistency for self and significant-others ratings ranged from 0.92 to 0.96 (Mathias & Coats, 1999). One brain-injured patient had a missing item on self-report. Four significant others of normal participants and one significant other of a brain-injured patient failed to return their questionnaires. Finally, one significant-other report of a brain-injured patient had a missing item. NRS data for these participants were excluded from relevant analyses. In addition to the total score, previous factor-analytic studies of the NRS and the NRS-Revised (NRS-R), a revised version of the NRS developed by Vanier, Mazaux, Lambert, Dassa and Levin (2000), suggested that at least three factors (relatively stable across studies) could be extracted from this instrument: a cognition/attention factor, a hyperactivity factor, and a mood/affect factor (Levin et al., 1987; McCauley et al., 2001; Vanier et al., 2000). Based on these previous studies, we calculated three surrogate factor scores by summing ratings across selected items (Hair, Anderson, & Black, 1995), separately for each of the two rating forms (self and significant-others ratings). The cognition/attention factor included Items 3 (Disorientation), 7 (Confusion), 10 (Memory), and 23 (Planning). The hyperactivity factor included Items 8 (Disinhibition), 11 (Restlessness), 22 (Excitement), and 25 (Tenseness). The mood/affect factor included Items 6 (Emotional (Guilt), 13 (Depressed Mood), 14 (Irritability), and 15 (Motivation).

PAGE 28

19 WAIS-III subtests The Digit Span forward and backward and Digit Symbol-Coding subtests of the Weschler Adult Intelligence TestThird Edition (WAIS-III) were administered as estimates of working memory and processing speed, respectively. Reliability estimates of the Digit Span ranged from 0.84 to 0.93 and its correlation with the working memory index of the WAIS-III was estimated at 0.83 in a normative sample ( WAIS-III and WMS-III Technical Manual 1997). Test-retest reliability estimates of the Digit Symbol-Coding ranged from 0.81 to 0.87 and its correlation with the processing speed index of the WAIS-III was estimated at 0.91 ( WAIS-III and WMS-III Technical Manual 1997). North American Adult Reading Test (NAART) The NAART is often used in neuropsychological evaluations as a measure of premorbid intelligence after brain injury or the onset of a dementia (Spreen & Strauss, 1991). It has been shown to be a good estimate of WAIS-R and WAIS-III scores, especially in the average range of intellectual abilities (Johnstone, Callahan, Kapila, & Bouman, 1996). However, it may tend to overestimate low IQ scores and underestimate high IQ scores (Johnstone et al., 1996). Procedures Participants were administered all of the tasks described above as part of a larger battery. The total duration of testing fell between 4 and 5 hours, and order was counterbalanced across tasks. Statistical Analyses Analyses of the Stroop data were conducted separately for the word-reading and color-naming tasks. For the card Stroop, the primary dependent variables were the

PAGE 29

20 number of items completed on each card within the 45-second time-limit. We conducted a 2 x 2 analysis of variance (ANOVA) with group as the between-subjects factor and condition (color naming and color-word naming) as the within-subject factor. We conducted a separate t-test for the word-reading condition. For the single-trial Stroop task, RTs and error rates were analyzed separately. For each trial type and each participant, we calculated the median RT for correct responses 3 and conducted separate 2-Group x 2-Delay x 3-Congruency ANOVAs for the word-reading and color-naming tasks. Between-group planned comparisons evaluated interference effects (incongruent RT vs neutral RT) in the color-naming task. The raw data for error rates were normalized using the arcsin transformation (Neter, Wasserman, & Kutner, 1985) before all analyses. We conducted separate 2 x 2 x 3 ANOVAs for word-reading and color-naming tasks. To statistically evaluate the primary prediction of interest, we also conducted a focused Group x Delay ANOVA on the incongruent color-naming condition to test our prediction of increased error rates in TBI patients at the long delay. Given the expected restricted range of NRS scores in healthy controls, we restricted our analyses of correlations between ratings on the NRS and performance on the Stroop task to the TBI group. To limit the number of correlations computed, we used five simple performance indices from the Stroop tasks: For the card Stroop, total number of items computed in the color-word naming condition. 3 In situations where subjects variability is expected to be important, and in which the probability of outliers may be different across conditions, Median RTs are fairly robust to outliers and have good power compared to other measures of central tendency (Ratcliff, 1993).

PAGE 30

21 For the single-trial Stroop, median RT in the incongruent word-reading task, median RT in the incongruent condition of the color-naming task, error rates in the word-reading task, and error rates in the color-naming task (for the four indices of the single-trial Stroop, we used the average of scores at the short and at the long delay).

PAGE 31

CHAPTER 3 RESULTS Neuropsychological Tests Means and standard deviations of the performance of TBI patients and control participants on various neuropsychological tests are presented in Table 3-1. Table 3-1. Performance on neuropsychological tests by group Control participants ( n = 21) TBI patients ( n = 23) M SD M SD NAART (errors) 26.2 9.2 33.7 14.1 Digit span forward (raw scores) 10.4 1.9 9.7 2.5 Digit span backward (raw scores) 6.7 2.1 5.7 2.2 Digit symbol (# items completed) a 89.1 15.7 65.3 16.4 a Because the Digit Symbol was included only later in our protocol, the data for this task include only 19 control participants and 15 TBI patients out of our total sample. The two groups did not differ in terms of their performance on the Digit Span Forward, t (43) = 0.6, p > 0.3, or Backward, t (43) = 1.6, p > 0.11. Brain-injured patients, however, committed significantly more errors on the NAART than control participants, t (43) = -2.1, p = 0.046, suggesting slightly lower verbal IQ in TBI patients. TBI patients also completed significantly fewer items on the Digit Symbol task than control participants, t (32) = 4.3, p < 0.001, suggesting lower processing speed in TBI patients. Card Stroop Means and standard errors of the number of items completed on each card for each group are presented in Figure 3-1. Control participants completed significantly more items than TBI patients in the word-reading condition, t (42) = 6.2, p < 0.001. The 2 x 2 ANOVA including group as a between-factor and color-naming and color-word-naming 22

PAGE 32

23 conditions as repeated measures yielded main effects of group, F (1, 42) = 45.3, p < 0.001, and condition, F (1, 42) = 440.1, p < 0.001. The Group x Condition interaction was also significant, F (1, 42) = 8.2, p < 0.01, reflecting a smaller interference score (number of items completed in the color-naming condition minus number of items completed in the color-word-naming condition) for brain-injured patients (M = 25.4, SD = 8.5) than for control participants (M = 33.4, SD = 10.0). These results suggest that Stroop interference was less strong in TBI patients than in control participants, a finding inconsistent with the hypothesis of a deficit in inhibition of a prepotent response in brain-injured individuals. 20406080100120WRCNCWNCondition Control TBI Figure 3-1. Number of items completed on the card Stroop by group and condition. Error bars represent standard errors. Single-Trial Stroop: Speed-Accuracy Tradeoff Analyses In order to ensure that there were no speed-accuracy tradeoffs in either group, we calculated the correlations between median RT and error rates in the incongruent condition for both groups, separately for color-naming and word-reading tasks. All estimates were positive except for the control group in the color-naming task, where the correlation was negative, r (20) = -0.25, although not significantly, t (20) = -1.1, p > 0.2.

PAGE 33

24 These results suggest that speed-accuracy tradeoffs did not play a significant role in the performance of TBI patients or control participants. Single-Trial Stroop: RTs Means and standard deviations of median RT for each group and each condition are presented in Table 3-2. In the word-reading task, control participants were faster than brain-injured patients, F (1, 42) = 16.7, p < 0.001, and RTs were longer at the short than long delay, F (1, 42) = 5.6, p < 0.03. There was also a significant effect of congruency, F (2, 42) = 24.5, p < 0.001. Fishers post-hoc comparisons revealed that RTs were longer in the incongruent than in the congruent ( p < 0.001) and neutral ( p < 0.001) conditions, whereas the difference between the neutral and congruent conditions was not significant ( p > 0.5). None of the interaction effects reached significance (all p > 0.1). These findings suggest generalized slowing in TBI patients. Table 3-2. Means and standard deviations of median RT (ms) in the single-trial Stroop Control participants ( n = 21) TBI patients ( n = 23) M SD M SD Word reading Short delay Congruent 820 174 951 139 Neutral 786 174 992 158 Incongruent 894 204 1082 286 Long delay Congruent 767 149 936 152 Neutral 732 200 957 154 Incongruent 839 136 1015 172 Color naming Short delay Congruent 838 199 1027 179 Neutral 770 138 967 155 Incongruent 1035 192 1310 249 Long delay Congruent 775 159 1014 215 Neutral 726 97 940 144 Incongruent 987 193 1265 258

PAGE 34

25 In the color-naming task, control participants were faster than brain-injured patients, F (1, 42) = 24.9, p < 0.001, and RTs were longer at the short delay than long delay, F (1, 42) = 10.7, p < 0.003. There was also a significant effect of congruency, F (2, 42) = 89.8, p < 0.001. Fishers post-hoc comparisons revealed that RTs were longer in the incongruent than in the congruent ( p < 0.001) and neutral ( p < 0.001) conditions. RTs were also significantly longer in the congruent than in the neutral condition, p < 0.01. None of the interaction effects reached significance, all p > 0.1. More importantly, the two groups were compared with respect to the interference score (difference between the incongruent and neutral conditions), at the short and long delays separately. Control participants ( M = 265, SD = 132) did not differ significantly from TBI patients ( M = 343, SD = 215) at the short delay, t (42) = -1.4, p > 0.1, nor at the long delay (control participants: M = 262, SD = 130; TBI patients: M = 326, SD = 170), t (42) = -1.4, p > 0.1. These results suggest generalized slowing but no disproportionate Stroop interference in TBI patient. Single-Trial Stroop: Error Rates Means and standard deviations for the error rates are displayed in Table 3-3, and the ANOVA results are presented in Table 3-4. In order to decompose the significant interaction between group and congruency in the word-naming task, we conducted separate analyses for each of the congruency condition, with delay as a within-subjects factor and group as a between-subjects factor. In the incongruent, but not in the congruent and neutral conditions, brain-injured patients committed significantly more errors than control participants, F (1, 42) = 6.8, p < 0.03. None of the other effects were statistically significant.

PAGE 35

26 Table 3-3. Means and standard deviations of error rates (%) in the single-trial Stroop Control participants (n = 21) TBI patients (n = 23) M SD M SD Word reading Short delay Congruent 0 0 0 0 Neutral 0 0 .4 1.7 Incongruent 0 0 2.0 4.6 Long delay Congruent 0 0 0 0 Neutral 0 0 .7 2.3 Incongruent .3 1.3 3.1 6.0 Color naming Short delay Congruent 0 0 0 0 Neutral 0 0 0 0 Incongruent 7.3 8.6 14.5 18.3 Long delay Congruent .7 2.2 0 0 Neutral .4 1.8 .2 1.0 Incongruent 2.8 6.0 16.5 23.1 Table 3-4. ANOVA for error rates in the single-trial Stroop Word reading Color naming Source df F p df F P Between subjects Between subjects Group 3 6.8 0.013 3 5.0 0.03 Within subjects Within Subjects Delay 1 1.3 > 0.2 1 .1 > 0.5 Congruency 2 6.7 0.002 2 20.9 < 0.01 Group x delay 3 .6 > 0.4 3 2.2 > 0.1 Group x congruency 6 5.3 0.007 6 5.9 0.004 Delay x congruency 2 .7 > 0.5 2 .7 > 0.4 Group x delay x congruency 6 .2 > 0.5 6 3.3 0.04 We followed the same procedure to decompose the significant three-way interaction in the color-naming task. In the congruent and neutral conditions, none of the effects reached significance (all p s > 0.1). In the incongruent condition, there was a main effect of group, F (1, 42) = 7.2, p < 0.02, and, more importantly, a near-significant interaction between group and delay, F (1, 42) = 4.0, p = 0.052. Consistent with our

PAGE 36

27 primary prediction, the difference between the two groups was significant at the long delay, t (42) = 3.2, p < 0.003, but not at the short delay, t (42) = 1.5, p > 0.1 (Figure 3-2). In terms of effect sizes, the difference between the two groups was moderate at the short delay, d = 0.49, and large at the long delay, d = 0.97. In addition, complementary analyses revealed that control participants committed significantly fewer errors at the long than short delay, t (42) = 2.7, p < 0.02, whereas error rates for TBI patients did not differ as a function of delay, t (42) = -0.3, p > 0.5. Consistent with our primary prediction, these results suggest a disruption over time of context representations in TBI patients compared to healthy participants. 0510152025ShortLongDelayErrors (%) Controls TBI Figure 3-2. Error rates in the incongruent color-naming condition of the single-trial Stroop by Group and Delay. Error bars represent standard errors. Neurobehavioral Rating Scales Cronbach-alpha internal consistency estimates for self and significant-others ratings are presented in Table 3-5. Internal consistency estimates were satisfactory (above 0.80) for self-ratings in the TBI group, but tended to be lower (below 0.80) in the control group. They also tended to be low in certain subscales of significant-others ratings, for both the TBI and the control group. The lower estimates for the control group are not

PAGE 37

28 surprising, given that the NRS was developed specifically for assessing impairment after brain injury. In addition, they should not constitute a significant threat to the validity of our study, since our analyses of the relations between reports of impairment and performance on the Stroop tasks were restricted to the TBI group. The lower internal consistency estimates for the significant-others ratings of brain-injured patients, on the other hand, suggest that the relationship between these ratings and the performance of TBI patients should be interpreted with caution. Table 3-5. Internal consistency (alpha-cronbach) estimates for the NRS for self and significant-others ratings by group Control participants TBI patients Self-rating Sign. Other Self-rating Sign. Other Total 0.93 (21) a 0.93 (17) 0.93 (22) 0.88 (21) Cognition 0.69 (21) 0.60 (17) 0.82 (23) 0.85 (22) Hyperactivity 0.65 (21) 0.74 (17) 0.86 (23) 0.66 (21) Mood 0.80 (21) 0.82 (17) 0.82 (23) 0.45 (22) a The number of participants used for this particular estimate is in parentheses. Because of missing questionnaires and missing items, this number varies slightly for each subscale. Means and standard deviations for each of the surrogate factor scores and for the total NRS score are reported in Table 3-6. Table 3-6. Means and standard deviations of NRS surrogate factor scores and total score by Group Control participants (n = 17) TBI patients (n = 20) Comparison M SD M SD t p Self-ratings Cognition 0.82 0.53 2.06 1.02 4.8 < 0.001 Hyperactivity 1.02 0.58 1.79 1.04 2.8 < 0.01 Affect 0.92 0.65 1.84 1.02 3.6 < 0.001 Total 0.82 0.46 1.81 0.79 5.2 < 0.001 Significant-other ratings Cognition 0.62 0.57 2.09 1.00 5.0 < 0.001 Hyperactivity 1.16 0.70 1.55 0.68 1.4 0.17 Affect 1.11 0.71 1.50 0.64 1.8 0.08 Total 0.88 0.54 1.57 0.54 3.5 < 0.002

PAGE 38

29 Data from four control participants and three TBI patients were not included due to missing items or because the significant other failed to mail or bring back the questionnaire. NRS scores show that brain-injured patients viewed themselves as having significantly more cognitive and emotional difficulties and as being more hyperactive than control participants. Significant others also rated TBI patients as more cognitively impaired than control participants. There was, however, no significant difference between the two groups in the significant-other ratings of hyperactivity or emotional functioning. Table 3-7. Correlations between self and significant-others ratings on the NRS C/S H/S M/S T/S C/O H/O M/O T/O Cognition self 0.61 ** 0.55 ** 0.82 ** 0.19 0.16 0.25 0.20 Hyperactivity self 0.33 0.59 ** 0.82 ** 0.10 0.08 0.27 0.13 Mood self 0.52 ** 0.61 ** 0.87 ** 0.01 0.25 0.14 0.19 Total self 0.79 ** 0.72 ** 0.82 ** 0.10 0.19 0.25 0.21 Cognition other 0.27 0.18 -0.18 0.04 0.78 ** 0.79 ** 0.88 ** Hyperactivity other 0.21 0.64 ** 0.44 0.41 0.07 0.82 ** 0.93 ** Mood other 0.28 0.62 ** 0.47 0.55 ** 0.29 0.65 ** 0.93 ** Total other 0.33 0.59 ** 0.23 0.44 0.61 ** 0.72 ** 0.80 ** Note. Upper right = control participants; Lower left = TBI patients. In the upper row, C stands for cognition, H for hyperactivity, M for mood and T for total; S stands for self-ratings and O for significant-others ratings. *p < 0.05. **p < 0.01. Correlations between the surrogate factor scores of the NRS and between self and significant-other ratings are presented in Table 3-7. Correlations between the scores on the various factor scores of the NRS were generally moderate to high within raters, but low between self and significant-other ratings, with the notable exception of significant and moderate correlations between self-ratings of hyperactivity and significant-others ratings of hyperactivity, mood and total NRS score in the TBI group. Correlations between ratings on the NRS and performance indices on the Stroop tasks are presented in Table 3.8. As predicted, performance on the card Stroop was not significantly correlated with self and significant-others ratings of reports of

PAGE 39

30 symptomatology. RTs in the single-trial task were also unrelated to reports of impairment, as were error rates in the word-reading task, with the exception of a significant correlation with self-ratings of mood. Error rates in the color-naming task, however, were significantly correlated with all three surrogate factor scores of the NRS for self-ratings, and with significant-others ratings of hyperactivity. Table 3-8. Correlations between ratings on the NRS and Stroop performance (TBI group) Card Stroop a RT word RT color Errors word Errors color Cognition self 0.34 0.15 0.21 0.33 0.45 Hyperactivity self 0.06 0.25 0.38 0.35 0.48 Mood self 0.05 0.22 0.27 0.46 0.68 ** Total self 0.26 0.24 0.31 0.49 0.61 ** Cognition other 0.15 0.17 0.22 -0.09 -0.01 Hyperactivity other -0.13 0.12 0.27 0.33 0.49 Mood other 0.07 0.01 0.11 0.22 0.41 Total other 0.10 0.18 0.38 0.19 0.34 a For the card Stroop, we reversed the sign of the correlation, as a larger number of items completed means better performance, whereas larger RTs and error rates in the single-trial Stroop mean worse performance. *p < 0.05. **p < 0.01. We conducted multiple regression analyses to examine the extent to which self and significant-others ratings independently predicted color-naming error rates (Table 3-9). Table 3-9. Multiple regressions with self and significant-others ratings as independent variables and error rates (color naming) as the dependent variables (TBI group) Total R 2 Self-ratings a Significant other a Cognition 0.21 0.47 -0.13 Hyperactivity 0.28 0.25 0.33 Mood 0.47 0.62 0.12 Total 0.34 0.55 0.08 a Normalized beta weights. *p < 0.05. For the cognition and mood subscales, and for the total NRS score, self-ratings remained significant predictors of error rates whereas significant-others ratings did not reliably contribute to the predicted variance. These results suggest that self-reports of

PAGE 40

31 impairment in TBI may be more related than significant-other reports to performance on cognitive tasks. We also computed the correlations between self-ratings on the NRS and color-naming error rates at the long and short delays separately (Table 3-10). As correlations tended to be qualitatively higher with errors at the long delay, we conducted multiple regressions with error rates at the short and long delays as independent variables, and cognition, hyperactivity, mood and total score as dependent variables. As seen in Table 3-11, normalized beta weights were generally not significant, except for mood, where error rates at the long, but not at the short delay, were uniquely predictive of self-ratings. Table 3-10. Correlations between self-ratings on the NRS and color-naming error rates at the long and short delays separately (TBI group) Errors short delay Errors long delay Cognition 0.38 0.45 Hyperactivity 0.44 0.45 Mood 0.56 ** 0.70 ** Total 0.54 ** 0.59 ** *p < 0.05. **p < 0.01. Table 3-11. Multiple regressions with color-naming error rates at the long and short delays as independent variables and NRS self-ratings as dependent variables (TBI group) Total R 2 Errors short delay a Errors long delay a Cognition 0.21 0.10 0.38 Hyperactivity 0.23 0.24 0.27 Mood 0.49 0.07 0.65 Total 0.37 0.21 0.43 a Normalized beta weights. *p < 0.05.

PAGE 41

CHAPTER 4 DISCUSSION The results of our study suggest that the performance of brain-injured patients on two versions of the Stroop paradigm is consistent with the presence of a deficit in the preparation to override prepotent response tendencies. On the card Stroop, TBI patients did not show deficits in inhibition of prepotent response tendencies beyond generalized slowing: they were slower than control participants on every condition of the task, but not disproportionately slower in the color-word naming (or interference) condition. 1 This finding is consistent with previous literature showing generalized slowing of TBI patients (Guskiewicz et al., 1997; Rojas & Bennett, 1995; Stuss et al., 1985; Trennery et al., 1989) but mixed results across studies regarding disproportionate slowing in the interference condition of the Stroop task (Batchelor et al., 1995; Bate et al., 2001; Bohnen, Jolles et al., 1992; Bohnen, Twijnstra et al., 1992; McDowell et al., 1997; Ponsford & Kinsella, 1992). The same result was observed when analyzing the RT data from the single-trial, computer-based version of the task. Again, TBI patients were slower on every condition (for both the color-naming and the word-reading tasks), but not disproportionately slower 1 In fact, as shown in the results section, the difference score between the color-word naming and the color-naming conditions was larger in control participants than in TBI patients. If considered too literally, these results may suggest that individuals having sustained brain injury show better inhibition of prepotent response tendencies than healthy participants. As pointed out by Chapman, Chapman, Curran, and Miller (1994), though, RTs difference scores cannot always be assumed to be free of the effects of generalized slowing, especially when the two groups being considered differ significantly in terms of their baseline scores. This is precisely the case in our study, where brain-injured patients completed fewer items than controls in the color-naming condition. 32

PAGE 42

33 on the incongruent, color-naming condition compared to control participants. In contrast, examination of error rates revealed that TBI patients showed a pattern of performance suggestive of inhibitory deficits: they committed more errors than control participants on the incongruent, but not on the congruent and neutral conditions of the color-naming task. Even more importantly, their error rates differed from those of control participants at the long, but not at the short delay of the incongruent condition. This later result is consistent with the main prediction stemming from the hypothesis of a context maintenance deficit in a given population (Cohen et al., 1999; Cohen et al., 1996): the poorest performance should be observed in conditions in which there are both inhibitory and working memory requirements. This is precisely the case for the long delay, incongruent color-naming condition of the single-trial Stroop, where participants have to maintain the context (i.e., task instructions) over the delay in order to inhibit the prepotent tendency to read the word. Our overall findings parallel those of Perlstein et al. (1998), who found that schizophrenics showed increased error rate interference on the single-trial Stroop, but no increased RT interference on the card or single-trial Stroop. They suggest that working memory and inhibition of prepotent response tendencies are two closely related processes, a hypothesis consistent with the results of Roberts, Hager and Heron (1994), who found that requiring healthy participants to maintain and manipulate information in working memory decreased performance on the antisaccade task. In order to examine the ecological validity (i.e., the relation between performance and day-to-day difficulties experienced by TBI patients) of the two versions of the Stroop paradigm used in our study, we compared various indices of performance of TBI patients with their report and significant-others report of impairment after injury along the three

PAGE 43

34 surrogate factor scores of the NRS (Levin et al., 1987): cognitive functioning, hyperactivity and mood. Number of items completed in the card Stroop and median RTs in the single-trial Stroop were not significantly related to reports of impairment. Error rates, on the other hand, especially in the color-naming task, were significantly correlated with reports of impairment along all three surrogate factor scores. The correlations generally fell in the moderate range and were of greater magnitude than the relation between RT interference score and Sickness Impact Profile total score reported in (Levine et al., 2000). Interestingly, the relation between performance and impairment after injury tended to be stronger for self-reports than significant-others reports. This result contrasts with the finding by Sunderland et al. (1983) that significant-others ratings of TBI patients memory difficulties correlated more strongly than self-reports with their performance on standard memory tests. This discrepancy between the two studies raises the possibility that self-report of TBI patients may be more reliable for general areas of functioning (i.e., cognition or mood) than for specific deficits after injury. Our results suggest that the long-lasting emotional, behavioral and cognitive difficulties incurred by individuals having sustained brain injury (Mathias & Coats, 1999; Olver et al., 1996) may be related to and predicted by their performance on tasks requiring the maintenance of context representations. Future studies are needed to examine whether the relation between cognitive performance and self-reports of impairment generalizes to objective psychosocial outcomes after brain injury, such as social isolation, employment status or ability to drive. One pervasive issue confronted by cognitive neuropsychologists is the difficulty to show the specificity of deficits in a given population. Simply stated, the finding that a

PAGE 44

35 group of patients differs significantly from control participants on a given task is not sufficient to conclude that this group is specifically impaired regarding the ability purportedly measured by the task: indeed, several general factors, such as poor attention, concentration, or motivation, result in poor performance across a broad array of tasks. Moreover, as Chapman, Chapman, Curran, and Miller (1994) pointed out, the use of difference scores for RTs (or number of items completed), such as in the traditional card Stroop, cannot always be assumed to remove the effects of generalized slowing. Similarly, the use of difference scores for error rates does not always correct for the effects of generalized deficits, especially when the two groups compared in the study differ regarding their baseline error rates (Chapman & Chapman, 1988, 1989). In our study, for instance, TBI patients and control participants error rates differed in the incongruent, but not in the congruent or neutral conditions of the color-naming task of the single-trial Stroop. This finding, however, does not provide strong evidence for the specificity of the deficits shown by brain-injured patients, because both groups were at floor levels in the congruent and neutral conditions. The fact, however, that the two groups differed significantly at the long, but not at the short delay of the incongruent condition, suggests that the deficits experienced by TBI patients are specific rather than generalized. As we mentioned before, this finding is exactly in line with the predictions stemming from the hypothesis of a deficit in the maintenance of context representations. As suggested by our study, the concept of context representations provides a theoretical framework with the potential to explain a vast array of seemingly disparate deficits in a variety of populations, such as schizophrenics (see e.g., Cohen et al., 1999; Cohen et al., 1996; Perlstein et al., 1998) and older adults (see Braver & Barch, 2002;

PAGE 45

36 Braver et al., 2001). 2 This framework also provides a starting point for the study of the neurochemical and neurobiological bases of cognitive control: namely, Cohen and Servan-Schreiber (1992; 1993) have postulated that the maintenance of context representations is related to the DA system in the prefrontal cortex (PFC), and specifically the dorsolateral prefrontal cortex (DL-PFC). Indeed, a large body of literature now supports an important role of the DL-PFC in working memory and cognitive control (see Salmon, Heindel, & Hamilton, 2001, for a review), and several studies have shown the role of the DA system on cognitive control tasks (Braver et al., 2001). As mentioned before, neuroimaging studies (Levin & Kraus, 1994) suggest that the frontal lobes may be especially vulnerable to TBI. Functional imaging studies of TBI patients during working memory tasks resulted in a more dispersed pattern of cerebral activation than in healthy controls, with increased lateralization to the right hemisphere, especially in the parietal and frontal lobes (Christodoulou et al., 2001; McAllister et al., 1999). In addition, McDowell, Whyte, & DEsposito (1998) conducted a double-blind, placebo controlled study on the effects of a DA receptor agonist on patients having sustained TBI. They found that the performance of TBI patients who were administered the DA agonist improved specifically on dual tasks and other tasks with cognitive control demands. 2 An alternate conceptual framework, developed by Kimberg and Farah (1993), may also explain the findings of our study: these authors propose that a weakening of the associations among working memory representations is responsible for the various executive deficits observed following frontal lobe damage. They note that their account of frontal lobe dysfunction and the maintenance of context representations model developed by Cohen and Servan-Schreiber (1993) yield similar predictions for many executive task, but differ in terms of the neural mechanisms underlying normal and impaired performance. They also contend that a deficit in context maintenance, contrary to their model, fails to account for the deficits in contextual memory observed in patients with frontal lobe lesions. Although we chose to present the model of Cohen and Servan-Schreiber as a framework for developing our predictions, we note that both theories may equally explain the results of our study.

PAGE 46

37 Taken together, these findings provide some evidence for the hypothesis that the disturbances of cognitive control experienced by brain-injured patients may be mediated by a disruption of the DA system in the DL-PFC. It should be pointed out, however, that the DL-PFC is not the only area of the PFC that has been implicated in the completion of the incongruent condition of the Stroop task. In a study of patients with focal lesions localized in different areas of the frontal lobes, Stuss, Floden, Alexander, Levine, and Katz (2001) found that an increased interference score was observed for patients in the superior medial frontal areas. Lateral lesions were associated with increased error rates rather than a disproportionate interference score. In an event-related potential (ERP) study of Stroop performance in mild TBI patients, Potter, Jory, Basset, Barrett and Mychalkiw (2002) found an increased negativity, compared to control participants, in a latency range consistent with the activation of the anterior cingulate (AC) gyrus. They interpreted these findings as suggesting that brain-injured patients may have to engage in more effortful processing to achieve the same performance than control participants. Using functional magnetic resonance imaging (fMRI), MacDonald, Cohen, Stenger and Carter (2000) found a dissociation of the role of the DL-PFC and the AC on a long-delay version of the single-trial Stroop task in healthy participants: whereas the DL-PFC was more active, during task preparation, for the color-naming conditions (consistent with a role in cognitive control), the AC was more active when responding to incongruent stimuli, consistent with a role in conflict monitoring. These results, combined with the findings by Potter et al. (2002), suggest that TBI patients may compensate for their difficulties with cognitive control related to impaired functioning of the DL-PFC by an increased

PAGE 47

38 activation of the AC during performance monitoring. Future research using functional imaging and event related potentials (ERPs) to examine the relation between DL-PFC and AC activation in TBI patients and healthy participants may confirm or disconfirm this hypothesis and help uncover the mechanisms of recovery and compensation after brain injury. A number of limitations and alternate explanations to the present findings require discussion. First, any study of TBI must take into account the heterogeneity of this population in terms of time since injury, severity and injury localization (Lezak, 1995). Previous studies examining the performance of TBI patients on the Stroop task have varied greatly in terms of injury severity (see Table 1-1). In our experiment, all TBI patients had sustained moderate-to-severe injuries. Obviously, our results cannot readily generalize to mild TBI, which presents with a set of unique issues in terms of assessment and outcome (Binder, 1997; Binder, Rohling, & Larrabee, 1997; Mathias & Coats, 1999). Similarly, a direct comparison of our results with previous findings must take into account the great variability in terms of time since injury across studies. According to Rao and Lyketsos (2000), cognitive deficits after TBI can be divided into four successive phases: a first period of coma (which may occur or not), a phase of confusion, agitation and PTA (lasting from a few days to a month), a third period of recovery of cognitive functions (up to 24 months), and a fourth phase of generally stable cognitive sequelae (see Cripe, 1987; Levin et al., 1987). The performance of TBI patients on any cognitive task, including the Stroop, is likely to vary depending on when they are tested after the time of the trauma. In our study, most TBI participants (17 out of 23) were tested more than two years after injury, corresponding to the phase of stable cognitive deficits

PAGE 48

39 according to Rao and Lyketsos (2000) classification. The average time since injury in the study conducted by Stuss et al. (1985) corresponded to the same fourth phase of cognitive deficits ( M = 2.6 years). By contrast, Bohnen, Jolles, and Twijnstra (1992) conducted testing from 6 to 14 days after injury. Other researchers (Bate et al., 2001) included patients with a wide range of time since injury. Future research is needed to explore in a systematic way the relationship between time since injury and performance on the Stroop task in TBI patients. Secondly, the absence of disproportionate slowing of TBI patients in the incongruent condition of the single-trial Stroop (and in the color-word naming condition of the card Stroop) may have an alternate explanation than poor sensitivity of RTs to context maintenance deficits. Namely, we did not discuss so far the possibility than the prepotent tendency to read words may not be as strong for brain-injured patients as it is for control participants. If this were the case, brain-injured patients would have less to inhibit in order to name the color of incongruent stimuli, and the lack of disproportionate slowing could not be readily interpreted as the absence of an inhibitory (or context maintenance) deficit. Fortunately, the data from our study allow us to estimate the prepotency of the tendency to read words in brain-injured patients and control participants. First, we compared the two groups with respect to median RTs in the word-reading task, averaged across all conditions. Control participants ( M = 806, SD = 159) were indeed faster than TBI patients ( M = 989, SD = 137) on this measure, t (42) = -4.1, p < 0.001. This result, however, cannot be readily interpreted as indicative of a lesser prepotency to read words in the TBI group, because previous literature shows that brain injury consistently results in generalized slowing across a broad array of tasks

PAGE 49

40 (Lezak, 1995). In order to control for generalized slowing, we conducted the same analysis, but after entering the number of items completed in the Digit Symbol task as covariate. Using this procedure, control participants ( M = 835, SD = 42) did not differ significantly from brain-injured patients ( M = 974, SD = 51) in terms of residual scores, F (1, 30) = 0.2, p > 0.1. 3 As a different and complementary way to control for generalized slowing, we compared the two groups on word-reading and color-naming tasks: if brain-injured patients were less drawn than control participants to read the word of presented stimuli, they should show a decreased difference score between word-reading and color-naming tasks. In order to investigate this possibility, we averaged the median RTs for all word-reading conditions, on the one hand, and all color-naming conditions, on the other hand. We then subtracted these totals from one another and compared the two groups with respect to this difference score. Using this procedure, control participants actually had a smaller difference score ( M = 49, SD = 75) than TBI patients ( M = 98, SD = 84), t (42) = -2.1, p < 0.05. These results suggest that a weaker prepotency to read words in brain-injured patients is an unlikely explanation for our finding of a lack of disproportionate slowing in the incongruent condition of the Stroop task for this group. Another limitation of our study is that we did not measure error rates in the card version of the Stroop paradigm. As a result, our finding of poor sensitivity of the card Stroop to the inhibitory deficits after TBI may be due to a general poor sensitivity of RTs rather than the characteristics of the task itself (i.e., the confounding of errors and number of items completed and the fact that conditions are blocked, thereby reducing context maintenance requirements). This is an important point, as our goal in this study was to 3 The above analysis was conducted with 19 control participants and 15 TBI patients out of our total sample.

PAGE 50

41 show that the performance of TBI patients on different versions of the Stroop paradigm may be better explained by a deficit in context maintenance than by poor inhibition of prepotent response tendencies. Future research should take this possibility into account and include the measurement of error rates in the card Stroop to disentangle these two explanations. In addition, the pattern of error rates in the incongruent, color-naming condition of the single-trial Stroop was somewhat different from our predictions: although there was, as we predicted, an interaction between group and delay, this interaction was due to a reduction in error rates from the short to the long delay in control participants rather than an increase in TBI patients. In other words, the extended delay allowed healthy participants to reduce their error rates, which we did not expect, whereas the error rates of TBI patients remained stable across the two delay conditions. These results suggest that TBI patients have difficulty in the preparation to override prepotent response tendencies: contrary to healthy participants, they were not able to use the increased delay to reduce their error rates. This finding may seem surprising, as the short and long delays were 1 and 5 seconds, respectively, and 1 second may seem a sufficient time to process task instructions (which, in this study, consisted in a simple auditory word, color or word) and prepare the inhibition of the prepotent tendency to read the word. Future research should attempt to examine the processes through which control participants manage to decrease their error rates between 1 and 5 seconds, and to determine what specific deficits in brain-injured patients are responsible for their inability to accomplish a similar decrease. The use of specific strategies (such as rehearsing task instructions in ones mind

PAGE 51

42 during the delay) in brain-injured patients and other populations may be one methodology to explore these questions in future research. Finally, our measure of everyday functioning, the NRS, does not have well-established psychometric properties, especially under its self and significant-other reports versions. Although useful, because designed specifically for impairments after TBI, it should be supplemented, in future research, by more well-researched and documented measures of everyday functioning, such as the Sickness Impact Profile (Bergner et al., 1976; Pollard et al., 1976). As a positive note, though, we should point out that the internal consistency estimates in our sample of brain-injured patients tended to be satisfactory (above 0.80) for self-ratings, despite the small number of items constituting each of the subscales. However, they tended to be lower for significant others ratings, especially for the hyperactivity and mood subscales, which, combined with the finding of somewhat higher correlations of self-ratings with error rates, suggests that TBI patients themselves rather than significant others may be better informants of their symptomatology after brain injury. Future research is needed to confirm and explore these findings in greater detail. In our study, we found a significant and moderate correlation between error rates in the incongruent color-naming condition of the single-trial Stroop and the three surrogate factor scores of the NRS we extracted from the questionnaires. We failed, however, to find any meaningful pattern of correlations that would relate performance of TBI patients on the Stroop task and specific domains of impairment in everyday life. In fact, the two domains of the NRS we expected to be most related with error rates were cognitive functioning (because the Stroop task is a cognitive task in nature) and hyperactivity

PAGE 52

43 (because it has frequently been related to malfunctioning of the frontal lobes, which also play a crucial role in context maintenance). If anything, though, the strongest correlations in our study were found with the mood factor of the NRS. Future research, using well-established measures of impairment after illness or injury, should examine more closely the relation between performance on the Stroop task and domains of everyday functioning. We should point out, however, that our study is one of the first to examine the ecological validity of the Stroop task in TBI patients, and that the finding of moderate correlations with a self-report measure of impairment is an encouraging finding regarding the usefulness of the Stroop paradigm for predicting adjustment after brain injury. In conclusion, the results of our study suggest that: The performance of TBI patients on a single-trial version of the Stroop task designed to maximize inhibitory and working memory requirements is consistent with a deficit in the preparation to override prepotent response tendencies. Error rates are more sensitive than RTs to these deficits in TBI populations. Deficits in the preparation to override prepotent response tendencies in TBI patients have meaningful ecological implications, that is, they are significantly related to self-reports of impairment after brain-injury. These findings suggest the possibility of developing new tools for the assessment and prediction of cognitive deficits after TBI. They also offer a framework for the study of the neural mechanisms responsible for the impairment, recovery and compensation strategies used by individuals having sustained brain injuries.

PAGE 53

LIST OF REFERENCES Adams, J. H., Graham, D. I., Murray, L. S., & Scott, G. (1982). Diffuse axonal injury due to nonmissile head injury in humans: an analysis of 45 cases. Annals of Neurology, 12(6), 557-563. Alexander, M. P. (1995). Mild traumatic brain injury: pathophysiology, natural history, and clinical management. Neurology, 45(7), 1253-1260. Batchelor, J., Harvey, A. G., & Bryant, R. A. (1995). Stroop colour word test as a measure of attentional deficit following mild head injury. The Clinical Neuropsychologist, 9(2), 180-186. Bate, A. J., Mathias, J. L., & Crawford, J. R. (2001). Performance on the Test of Everyday Attention and standard tests of attention following severe traumatic brain injury. Clin Neuropsychol, 15(3), 405-422. Bergner, M., Bobbitt, R. A., Pollard, W. E., Martin, D. P., & Gilson, B. S. (1976). The sickness impact profile: validation of a health status measure. Medical Care, 14(1), 57-67. Binder, L. M. (1997). A review of mild head trauma. Part II: Clinical implications. Journal of Clinical and Experimental Neuropsychology, 19(3), 432-457. Binder, L. M., Rohling, M. L., & Larrabee, J. (1997). A review of mild head trauma. Part I: Meta-analytic review of neuropsychological studies. Journal of Clinical and Experimental Neuropsychology, 19(3), 421-431. Bohnen, N., Jolles, J., & Twijnstra, A. (1992). Modification of the Stroop colour word test improves differentiation between patients with mild head injury and matched controls. The Clinical Neuropsychologist, 6(2), 178-184. Bohnen, N., Twijnstra, A., & Jolles, J. (1992). Performance in the Stroop color word test in relationship to the persistence of symptoms following mild head injury. Acta Neurol Scand, 85(2), 116-121. Braver, T. S., & Barch, D. M. (2002). A theory of cognitive control, aging cognition, and neuromodulation. Neuroscience and Biobehavioral Reviews, 26(7), 809-817. 44

PAGE 54

45 Braver, T. S., Barch, D. M., Keys, B. A., Carter, C. S., Cohen, J. D., Kaye, J. A., Janowsky, J. S., Taylor, S. F., Yesavage, J. A., Mumenthaler, M. S., Jagust, W. J., & Reed, B. R. (2001). Context processing in older adults: evidence for a theory relating cognitive control to neurobiology in healthy aging. Journal of Experimental Psychology: General, 130(4), 746-763. Carter, C. S., MacDonald, A. M., Botvinick, M., Ross, L. L., Stenger, V. A., Noll, D., & Cohen, J. D. (2000). Parsing executive processes: strategic vs. evaluative functions of the anterior cingulate cortex. Proceedings of the National Academy of Sciences: U S A, 97(4), 1944-1948. Chapman, L. J., & Chapman, J. P. (1988). Artifactual and genuine relationships of lateral difference scores to overall accuracy in studies of laterality. Psychological Bulletin, 104(1), 127-136. Chapman, L. J., & Chapman, J. P. (1989). Strategies for resolving the heterogeneity of schizophrenics and their relatives using cognitive measures. Journal of Abnormal Psychology, 98(4), 357-366. Chapman, L. J., Chapman, J. P., Curran, T. E., & Miller, M. B. (1994). Do children and the elderly show heightened semantic priming? How to answer the question. Developmental Review, 14, 159-185. Christodoulou, C., DeLuca, J., Ricker, J. H., Madigan, N. K., Bly, B. M., Lange, G., Kalnin, A. J., Liu, W. C., Steffener, J., Diamond, B. J., & Ni, A. C. (2001). Functional magnetic resonance imaging of working memory impairment after traumatic brain injury. Journal of Neurology, Neurosurgery, and Psychiatry, 71(2), 161-168. Cohen, J. D., Barch, D. M., Carter, C., & Servan-Schreiber, D. (1999). Context-processing deficits in schizophrenia: converging evidence from three theoretically motivated cognitive tasks. Journal of Abnormal Psychology, 108(1), 120-133. Cohen, J. D., Braver, T. S., & O'Reilly, R. C. (1996). A computational approach to prefrontal cortex, cognitive control and schizophrenia: recent developments and current challenges. Philosophical Transactions of the Royal Society of London Series B, 351(1346), 1515-1527. Cohen, J. D., MacWhinney, B., Flatt, M. R., & Provost, J. (1993). PsyScope: An interactive graphic system for designing and controlling experiments in the psychology laboratory using Macintosh computers. Behavioral Research Methods: Instruments and Computers, 25, 257-271. Cohen, J. D., & Servan-Schreiber, D. (1992). Context, cortex, and dopamine: a connectionist approach to behavior and biology in schizophrenia. Psychological Review, 99(1), 45-77.

PAGE 55

46 Cohen, J. D., & Servan-Schreiber, D. (1993). A theory of dopamine function and its role in cognitive deficits in schizophrenia. Schizophrenia Bulletin, 19(1), 85-104. Crevits, L., Hanse, M. C., Tummers, P., & Van Maele, G. (2000). Antisaccades and remembered saccades in mild traumatic brain injury. Journal of Neurology, 247(3), 179-182. Cripe, L. I. (1987). The neuropsychological assessment and management of closed head injury: General guidelines. Cognitive Rehabilitation, 5(1), 18-22. Golden, C. J. (1975). The measurement of creativity by the Stroop Color and Word Test. Journal of Personality Assessment, 39(5), 502-506. Golden, C. J. (1976). Identification of brain disorders by the Stroop Color and Word Test. Journal of Clinical Psychology, 32(3), 654-658. Golden, C. J. (1978). Stroop Color and Word Test, Stoelting. Chicago. Groswasser, Z., Reider-Groswasser, I., Soroker, N., & Machtey, Y. (1987). Magnetic resonance imaging in head injured patients with normal late computed tomography scans. Surgical Neurology, 27(4), 331-337. Groth-Marnat, G. (2000). Neuropsychological assessment in clinical practice : a guide to test interpretation and integration. New York: Wiley. Guskiewicz, K. M., Riemann, B. L., Perrin, D. H., & Nashner, L. M. (1997). Alternative approaches to the assessment of mild head injury in athletes. Med Sci Sports Exerc, 29(7 Suppl), S213-221. Hair, J. F., Anderson, R. E., & Black, W. C. (1995). Multivariate Data Analysis (4th ed.). Englewood Cliffs, NJ: Prentice Hall. Jahanshahi, M., Profice, P., Brown, R. G., Ridding, M. C., Dirnberger, G., & Rothwell, J. C. (1998). The effects of transcranial magnetic stimulation over the dorsolateral prefrontal cortex on suppression of habitual counting during random number generation. Brain, 121 ( Pt 8), 1533-1544. Johnstone, B., Callahan, C. D., Kapila, C. J., & Bouman, D. E. (1996). The comparability of the WRAT-R reading test and NAART as estimates of premorbid intelligence in neurologically impaired patients. Archives of Clinical Neuropsychology, 11(6), 513-519. Kiefer, M., Marzinzik, F., Weisbrod, M., Scherg, M., & Spitzer, M. (1998). The time course of brain activations during response inhibition: evidence from event-related potentials in a go/no go task. Neuroreport, 9(4), 765-770.

PAGE 56

47 Kimberg, D. Y., & Farah, M. J. (1993). A unified account of cognitive impairments following frontal lobe damage: the role of working memory in complex, organized behavior. Journal of Experimental Psychology: General, 122(4), 411-428. Kingma, A., La Heij, W., Fasotti, L., & Eling, P. (1996). Stroop interference and disorders of selective attention. Neuropsychologia, 34(4), 273-281. Kraus, M. F., & Maki, P. M. (1997). Effect of amantadine hydrochloride on symptoms of frontal lobe dysfunction in brain injury: case studies and review. Journal of Neuropsychiatry and Clinical Neurosciences, 9(2), 222-230. Levin, H., & Kraus, M. F. (1994). The frontal lobes and traumatic brain injury. Journal of Neuropsychiatry and Clinical Neurosciences, 6(4), 443-454. Levin, H. S., Eisenberg, H. M., & Benton, A. L. (1991). Mild Head Injury. New York: Oxford University Press. Levin, H. S., High, W. M., Goethe, K. E., Sisson, R. A., Overall, J. E., Rhoades, H. M., Eisenberg, H. M., Kalisky, Z., & Gary, H. E. (1987). The neurobehavioural rating scale: assessment of the behavioural sequelae of head injury by the clinician. Journal of Neurology, Neurosurgery, and Psychiatry, 50(2), 183-193. Levine, B., Dawson, D., Boutet, I., Schwartz, M. L., & Stuss, D. T. (2000). Assessment of strategic self-regulation in traumatic brain injury: its relationship to injury severity and psychosocial outcome. Neuropsychology, 14(4), 491-500. Levine, M. J. (1988). Issues in neurobehavioural assessment of mild head injuries. Cognitive Rehabilitation, 6, 14-20. Lezak, M. D. (1995). Neuropsychological assessment (3rd ed.). New York: Oxford University Press. Lishman, W. A. (1988). Physiogenesis and psychogenesis in the 'post-concussional syndrome'. British Journal of Psychiatry, 153, 460-469. MacDonald, A. W., 3rd, Cohen, J. D., Stenger, V. A., & Carter, C. S. (2000). Dissociating the role of the dorsolateral prefrontal and anterior cingulate cortex in cognitive control. Science, 288(5472), 1835-1838. MacLeod, C. M. (1991). Half a century of research on the Stroop effect: an integrative review. Psychological Bulletin, 109(2), 163-203. Mathias, J. L., & Coats, J. L. (1999). Emotional and cognitive sequelae to mild traumatic brain injury. Journal of Clinical and Experimental Neuropsychology, 21(2), 200-215.

PAGE 57

48 McAllister, T. W., Saykin, A. J., Flashman, L. A., Sparling, M. B., Johnson, S. C., Guerin, S. J., Mamourian, A. C., Weaver, J. B., & Yanofsky, N. (1999). Brain activation during working memory 1 month after mild traumatic brain injury: a functional MRI study. Neurology, 53(6), 1300-1308. McCauley, S. R., Levin, H. S., Vanier, M., Mazaux, J. M., Boake, C., Goldfader, P. R., Rockers, D., Butters, M., Kareken, D. A., Lambert, J., & Clifton, G. L. (2001). The neurobehavioural rating scale-revised: sensitivity and validity in closed head injury assessment. Journal of Neurolology, Neurosurgery, and Psychiatry, 71(5), 643-651. McDowell, S., Whyte, J., & D'Esposito, M. (1997). Working memory impairments in traumatic brain injury: evidence from a dual-task paradigm. Neuropsychologia, 35(10), 1341-1353. McDowell, S., Whyte, J., & D'Esposito, M. (1998). Differential effect of a dopaminergic agonist on prefrontal function in traumatic brain injury patients. Brain, 121 ( Pt 6), 1155-1164. Miyake, A., Emerson, M. J., & Friedman, N. P. (2000). Assessment of executive functions in clinical settings: problems and recommendations. Seminars in Speech and Language, 21(2), 169-183. Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T. D. (2000). The unity and diversity of executive functions and their contributions to complex "Frontal Lobe" tasks: a latent variable analysis. Cognitive Psychology, 41(1), 49-100. Neter, J., Wasserman, W., & Kutner, M. H. (1985). Applied linear statistical models: regression, analysis of variance, and experimental designs (2nd ed.). Homewood, Ill.: R.D. Irwin. Olver, J. H., Ponsford, J. L., & Curran, C. A. (1996). Outcome following traumatic brain injury: a comparison between 2 and 5 years after injury. Brain Injury, 10(11), 841-848. Perlstein, W. M., Carter, C. S., Barch, D. M., & Baird, J. W. (1998). The Stroop task and attention deficits in schizophrenia: a critical evaluation of card and single-trial Stroop methodologies. Neuropsychology, 12(3), 414-425. Pollard, W. E., Bobbitt, R. A., Bergner, M., Martin, D. P., & Gilson, B. S. (1976). The Sickness Impact Profile: reliability of a health status measure. Medical Care, 14(2), 146-155. Ponsford, J., & Kinsella, G. (1992). Attentional deficits following closed-head injury. J Clin Exp Neuropsychol, 14(5), 822-838.

PAGE 58

49 Potter, D. D., Jory, S. H., Bassett, M. R., Barrett, K., & Mychalkiw, W. (2002). Effect of mild head injury on event-related potential correlates of Stroop task performance. Journal of the International Neuropsychological Society, 8(6), 828-837. Rao, V., & Lyketsos, C. (2000). Neuropsychiatric sequelae of traumatic brain injury. Psychosomatics, 41(2), 95-103. Ratcliff, R. (1993). Methods for dealing with reaction time outliers. Psychological Bulletin, 114(3), 510-532. Rieger, M., & Gauggel, S. (2002). Inhibition of ongoing responses in patients with traumatic brain injury. Neuropsychologia, 40(1), 76-85. Roberts, R. J., Hager, L. D., & Heron, C. (1994). Prefrontal cognitive processes: Working memory and inhibition in the Antisaccade task. Journal of Experimental Psychology: General, 123, 374-393. Rojas, D. C., & Bennett, T. L. (1995). Single versus composite score discriminative validity with the Halstead-Reitan battery and the Stroop test in mild brain injury. Archives of Clinical Neuropsychology, 10, 101-110. Salmon, D. P., Heindel, W. C., & Hamilton, J. M. (2001). Cognitive abilities mediated by frontal-subcortical circuits. In J. L. Cummings (Ed.), Frontal-subcortical circuits in psychiatric and neurological disorders (pp. 114-150). New York: The Guilford Press. Salo, R., Avishai, H., & Lynn, C. R. (2001). Interpreting Stroop interference: An analysis of differences between task versions. Neuropsychology, 15, 462-471. Spreen, O., & Strauss, E. (1991). A compendium of neuropsychological tests : administration, norms, and commentary. New York: Oxford University Press. Stuss, D. T., Ely, P., Hugenholtz, H., Richard, M. T., LaRochelle, S., Poirier, C. A., & Bell, I. (1985). Subtle neuropsychological deficits in patients with good recovery after closed head injury. Neurosurgery, 17(1), 41-47. Stuss, D. T., Floden, D., Alexander, M. P., Levine, B., & Katz, D. (2001). Stroop performance in focal lesion patients: dissociation of processes and frontal lobe lesion location. Neuropsychologia, 39(8), 771-786. Sunderland, A., Harris, J. E., & Baddeley, A. D. (1983). Do laboratory tests predict everyday memory? A neuropsychological study. Journal of Verbal Learning and Verbal Behavior, 22(3), 341-357. Trennery, M. R., Crosson, B., DeBoe, J., & Leber, W. R. (1989). Stroop Neuropsychological Screening Test Manual. Odessa, FL: Psychological Assessment Resources.

PAGE 59

50 Vanier, M., Mazaux, J. M., Lambert, J., Dassa, C., & Levin, H. S. (2000). Assessment of neuropsychologic impairments after head injury: interrater reliability and factorial and criterion validity of the Neurobehavioral Rating Scale-Revised. Archives of Physical Medecine and Rehabilitation, 81(6), 796-806. van Zomeren, A. H., & van den Burg, W. (1985). Residual complaints of patients two years after severe head injury. Journal of Neurology, Neurosurgery, and Psychiatry, 48(1), 21-28. WAIS-III and WMS-III Technical Manual. (1997). San Antonio, TX: The Psychological Corporation.

PAGE 60

BIOGRAPHICAL SKETCH Paul J. Seignourel studied mathematics in France and received a Ph.D. in probability from the University Paris VI and Ecole Polytechnique in 1999. He then moved to the United States and, after a year of post-baccalaureate studies in psychology, entered the Department of Clinical and Health Psychology at the University of Florida. 51


Permanent Link: http://ufdc.ufl.edu/UFE0000783/00001

Material Information

Title: A Tale of Two Stroops in Traumatic Brain Injury
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000783:00001

Permanent Link: http://ufdc.ufl.edu/UFE0000783/00001

Material Information

Title: A Tale of Two Stroops in Traumatic Brain Injury
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000783:00001


This item has the following downloads:


Full Text












A TALE OF TWO TROOPS IN TRAUMATIC BRAIN INJURY


By

PAUL SEIGNOUREL

















A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2003

































Copyright 2003

by

Paul J. Seignourel















ACKNOWLEDGMENTS

I thank my advisor, William M. Perlstein; and my collaborators on this project,

Michael A. Cole, Jason A. Demery, and Diana L. Robins.
















TABLE OF CONTENTS
page

A C K N O W L E D G M E N T S .................................................................... ......... .............. iii

LIST OF TABLES ......... ........................ ..............................vi

L IST O F FIG U R E S .... ...... ................................................ .. .. ..... .............. vii

A B S T R A C T .......................................... .................................................. v iii

CHAPTER

1 IN TR OD U CTION ............................................... .. ......................... ..

Performance of Brain-Injured Patients on the Stroop Task .........................................6
P reduction s ............................................................................ 10

2 M E T H O D S .......................................................................................................14

P a rtic ip a n ts ........................................................................................................... 1 4
M a te ria ls ........................................................................................................1 5
C a rd S tro o p .....................................................................................1 5
Single-Trial Stroop ............................................... .......... 16
Neurobehavioral Rating Scale (NRS) ........................................ .....17
WAIS-III subtests .................................................... .... ........19
North American Adult Reading Test (NAART) ..................................... 19
P ro c e d u re s ........................................................................................1 9
S statistical A n aly se s ............................................................................................... 19

3 R E S U L T S .............................................................................2 2

N europ sy chological T ests..................................................................................... 22
C ard Stroop ...................................... .... ....................................22
Single-Trial Stroop: Speed-Accuracy Tradeoff Analyses .......................................23
Single-T rial Stroop : R T s....................................................................................... 24
Single-Trial Stroop: Error R ates .......................................................... .. .....25
Neurobehavioral Rating Scales............................ ...............27

4 D ISC U S SIO N ............................................................................... 32










L IST O F R E F E R E N C E S ........................................................................ ..................... 44

BIOGRAPH ICAL SKETCH .............................................................. ...................51


























































v
















LIST OF TABLES


Table page

1-1 Studies of performance of TBI patients on the Stroop task................... ..............

2-1 Demographics and injury variables by group.......................................................16

3-1 Performance on neuropsychological tests by group .............................................22

3-2 Means and standard deviations of median RT (ms) in the single-trial Stroop ........24

3-3 Means and standard deviations of error rates (%) in the single-trial Stroop ..........26

3-4 ANOVA for error rates in the single-trial Stroop..................................................26

3-5 Internal consistency (alpha-cronbach) estimates for the NRS for self and
significant other's ratings by group ......................... ......... ...............28

3-6 Means and standard deviations of NRS surrogate factor scores and total score
b y g ro u p .......................................................................... 2 8

3-7 Correlations between self and significant-others' ratings on the NRS....................29

3-8 Correlations between ratings on the NRS and Stroop performance (TBI group)....30

3-9 Multiple regressions with self and significant-others' ratings as independent
variables and error rates (color naming) as the dependent variables (TBI group) ..30

3-10 Correlations between self-ratings on the NRS and color-naming error rates at the
long and short delays separately (TBI group).................................. ... ..................31

3-11 Multiple regressions with color-naming error rates at the long and short delays as
independent variables and NRS self-ratings as dependent variables (TBI group) ..31















LIST OF FIGURES


Figure page

2-1 Schema of the single-trial Stroop. Participants are first presented with an
instructional cue ("Color" or "Word"), followed by a delay (1 or 5 seconds) and
an imperative stimulus (congruent, neutral or incongruent). ...................................17

3-1 Number of items completed on the card Stroop by group and condition. Error bars
represent standard errors. ............................................... ............................... 23

3-2 Error rates in the incongruent color-naming condition of the single-trial Stroop by
group and delay. Error bars represent standard errors. ...........................................27















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

A TALE OF TWO TROOPS IN TRAUMATIC BRAIN INJURY

By

Paul J. Seignourel

May 2003

Chair: William M. Perlstein
Major Department: Clinical and Health Psychology

The Stroop Color-Word task has frequently been used to examine attentional

deficits in Traumatic Brain Injury (TBI). However, critical review of the literature

suggests that TBI patients do not consistently show disproportionate reaction time (RT)

interference on the Stroop task, inconsistent with a frequent assertion of increased

interference. We hypothesized that combining inhibition and working memory (WM)

requirements in a new, computerized version of the Stroop would increase sensitivity to

TBI. We examined the performance of healthy subjects and TBI patients on card and

single-trial versions of the Stroop paradigm. The TBI patients did not show

disproportionately increased RT interference compared to controls on either version, but

instead greater error-rate interference on the single-trial version, especially at the long

delay. Error rates, but not reaction times, were related to reports of symptomatology in

TBI patients. Results suggest that errors (rather than RTs) are more sensitive to cognitive

deficits in TBI; and that such deficits, and everyday life difficulties experienced by TBI









survivors, may be related to a difficulty in the preparation to override prepotent response

tendencies.














CHAPTER 1
INTRODUCTION

Traumatic brain injury (TBI), with an incidence of 500,000 to 1.9 million cases

each year (Lezak, 1995), is a significant problem in the United States. It is also a unique

problem, in that it affects large numbers of young individuals: the highest incidence of

brain injury occurs in the 15 to 24 year range, with high incidence rates also in the 0 to 5

year range and for the elderly. Among individuals who sustain TBI, some experience

only transient symptoms, such as brief loss of consciousness (LOC) and post-traumatic

amnesia (PTA). Others, however, continue to experience significant problems long after

injury. These symptoms include frequent headaches, irritability and restlessness, sleep

disturbances, and various cognitive difficulties (such as poor concentration and attention,

reduced processing speed, and impaired memory) (Levin, Eisenberg, & Benton, 1991;

Levine, 1988; van Zomeren & van den Burg, 1985): a constellation of symptoms

sometimes referred to as postconcussional syndrome (Lishman, 1988). In a study of 103

patients with moderate-to-severe brain injuries, Olver, Ponsford and Curran (1996) found

that at a 2-year follow-up, 61% of the patients reported word-finding problems, 60% had

difficulties concentrating and 56% reported feeling more depressed than before the

injury. Moreover, only 40% were able to drive and only 41% were employed full-time.

Percentages were similar at 5-year follow-up, with the exception of ability to drive,

which increased to 48%; and full-time employment, which decreased to 34%. Such

findings suggest that TBI, at least in the moderate-to-severe range, results in disabling

and long-lasting difficulties for a great number of individuals.









Closed head injuries, in which the skull is intact and the brain is not exposed,

account for about 90% of all brain injuries (Lezak, 1995). Anatomically, brain damage

after closed head injury is generally diffuse, resulting from several distinct damaging

mechanical forces to many different areas of the brain. Some evidence exists, however,

that because of the shape of bone's protuberances in the brain, bruising due to coup (the

blow at the point of impact) and contrecoup (bruising in the area opposite to the blow) is

often more pronounced in the frontal and temporal lobes (Levin & Kraus, 1994). Axonal

shearing or diffuse axonal injury (DAI), which is an excessive stretching of nerve fibers

and blood vessels due to violent acceleration and deceleration (Adams, Graham, Murray,

& Scott, 1982), might be more concentrated in these areas as well (Groswasser,

Reider-Groswasser, Soroker, & Machtey, 1987). Consistent with evidence of frontal

damage, studies of individuals with TBI often show difficulties on complex tasks, despite

normal or quasi-normal functioning in basic cognitive domains such as arousal, language

and perception. For example McDowell, Whyte and D'Esposito (1997) compared

brain-injured patients with control participants on a simple visual reaction time (RT) task

and on a dual task, consistent with patient reports of difficulties negotiating multiple

simultaneous real-world tasks (van Zomeren & van den Burg, 1985). They found that, in

addition to generalized slowing, TBI patients showed greater decrements in performance

compared to control participants during dual-task conditions. Similarly, Levine, Dawson,

Boutet, Schwartz and Stuss (2000) found that TBI patients were impaired on the Revised

Strategy Application Test (R-SAT), a complex and unstructured task in which

participants, in order to perform satisfactorily, must select items on the basis of their

length while refraining from the tendency to complete all items sequentially.









Despite major differences, the two previously mentioned tasks share at least one

essential characteristic: in both cases, participants tend to perform poorly if they complete

all items automatically (that is, by simply responding to each item presented to them and

without any overall strategy toward optimal performance-bottom-up processing).

Conversely, good performance depends on participants' strategy being guided by the

overall goal and structure of the task (top-down processing). In other words, both tasks

require that participants actively maintain goal representations to guide their behavior and

the order in which they perform the task. Using a more general framework, Braver and

Barch (2002-page 1) defined context representations as "any task-relevant information

that is generally represented in such a form that it can bias processing in the pathways

responsible for task performance." Through extensive behavioral studies and

connectionist computational models, Cohen, Barch, Carter and Servan-Schreiber (1999)

showed that schizophrenic patients' cognitive difficulties on a range of seemingly

dissimilar tasks could be explained by a single deficit in context maintenance. According

to their model, context representations are maintained on-line by the prefrontal cortex

(PFC), through a mechanism where dopamine (DA) plays a prominent role; the putative

disruptions in the dopaminergic circuitry in schizophrenic patients are thought to be

primarily responsible for their context maintenance deficits. More recently, Braver et al.

(2001) showed that performance decrements and, counterintuitively, performance

improvements in older adults could also be explained by a deficit in context maintenance.

Importantly, the term context maintenance, in this framework, refers to a concept more

general than goal representations: goals represent only one type of representation, among

others, that can bias processing of subsequent information. In the AX-Continuous









Performance Task (AX-CPT), for instance, the cue provides the context essential for

proper processing of the probe, without representing per se the goal of the task (Cohen et

al., 1999). Context maintenance, on the other hand, is a much more specific concept than

executive functioning, which has been used to describe a number of abilities including

planning, problem-solving, shifting mental set, and inhibition of prepotent responses. In

fact, Cohen, Braver and O'Reilly (1996) noted that the term dysexecutive syndrome

(often used to describe the set of symptoms exhibited by patients with frontal lobe

lesions) is essentially descriptive, and does not specify which mechanisms are responsible

for executive control.1 They added that their framework, by specifying the mechanisms

responsible for cognitive control at a computational and neurobiological level, leads to

specific predictions regarding the performance of individuals with context maintenance

deficits.

For the purpose of our study, two of these predictions are especially relevant. First,

representations of context are especially important when the response required by the task

at hand is competing with a more frequent and automatic response. This characteristic of

context maintenance is similar to what has been called inhibition of the prepotent

response. It is required in the Stroop task, used in this study, but also in the antisaccade

task (Crevits, Hanse, Tummers, & Van Maele, 2000) and in the R-SAT (Levine et al.,

2000), previously mentioned. Second, deficits in context maintenance should be more


1 Context maintenance may not account for all mechanisms involved in executive control.
In the framework described by Cohen and Servan-Schreiber (1992), the maintenance of
context representations is carried out by the prefrontal cortex (PFC) and mediated by DA.
Carter et al. (2000), however, showed that the role of the anterior cingulate (AC) in
cognitive control is evaluative rather than strategic. In other words, the AC might be
another area of the brain responsible for cognitive control, complementary to the
maintenance of context carried out by the PFC.









pronounced when the task requires participants to hold context representations in mind

for a prolonged period of time. This second characteristic of context maintenance shows

its similarity with working memory; in fact, Braver et al. (2001) recently described the

maintenance of context representations as a subset of representations within working

memory. In support of this hypothesis, Cohen et al. (1996) compared the performance of

schizophrenics and control participants on the AX-CPT; they found that the sensitivity of

schizophrenics to the cue decreased after a long delay compared to a short delay between

the cue and the probe; whereas the performance of control participants was not affected

by delay. Moreover, the same pattern of results was observed when comparing two neural

network simulations designed to model normal performance and performance associated

with a reduced gain in the context layer. Cohen et al. (1999), on the other hand, failed to

find significant delay effects for both schizophrenics and healthy participants on a

single-trial version of the Stroop paradigm we used.

The primary goal of our study was to examine the hypothesis that patients with

moderate-to-severe TBI exhibit difficulties maintaining context representations, as

evidenced by poor performance when they must maintain representations over a delay in

order to inhibit prepotent response tendencies. This hypothesis stems from the relation

among context maintenance, PFC, and dopaminergic neuromodulation, on the one hand;

and from the aforementioned finding that closed head injuries often result in damage to

the frontal lobes, on the other hand. It is also consistent with the various symptoms

experienced by such patients (some of which, including lack of behavioral control,

disorganization, attention difficulties, and memory deficits, suggest dysfunction of the

frontal lobes) (Kraus & Maki, 1997; Levin & Kraus, 1994; Rieger & Gauggel, 2002).









More specifically, our study used a single-trial version of the Stroop task, which places

greater demands on working memory than the card Stroop often used in clinical settings.

Before we describe the tasks in detail and state our specific predictions, we review the

evidence that TBI critically affects performance on the Stroop task.

Performance of Brain-Injured Patients on the Stroop Task

The Stroop task has been considered by several researchers to be a prototypical

instrument for measuring prepotent response inhibition2 (Miyake, Emerson, & Friedman,

2000; Miyake, Friedman et al., 2000), a function often attributed, in part, to the frontal

lobes (Jahanshahi et al., 1998; Kiefer, Marzinzik, Weisbrod, Scherg, & Spitzer, 1998).

Although there are many different versions of the original task (MacLeod, 1991), the

basic paradigm consists of naming the color of the ink of an incongruent color word. For

example, one would be presented with the word BLUE printed in red ink. In order to

provide the correct answer, red, the participant must overcome the natural, more

automatic or prepotent tendency to simply read the word and respond blue. A version of

the Stroop task most widely used in clinical settings (Golden, 1978) comprises three

conditions:

* Word reading, in which participants must read aloud columns of color words.

* Color naming, in which participants are required to name the color of the ink of
rows of Xs.

2 Although some researchers refer to the Stroop task as a measure of selective or focused
attention (see e.g., Batchelor, Harvey, & Bryant, 1995; Kingma, La Heij, Fasotti, &
Eling, 1996; Salo, Avishai, & Lynn, 2001), we find the term "inhibition of a prepotent
response" more specific. Selective attention is the ability to focus one's attention on a
specific stimulus or a specific dimension of a stimulus among an array of distractors. Not
included in this definition is a key component of the Stroop, which is the fact that in this
task the distractor (i.e., the color word) tends to elicit an automatic (and in this case
incorrect) response; whereas the correct response (naming the color of the ink) is much
less natural and automatic.









* Color-word naming (or interference), in which participants must name the color of
the ink of incongruent color words.

Studies examining TBI patients' performance on the Stroop paradigm have relied

on a variety of task versions and scoring methods (Table 1-1). In some studies,

researchers have used a single measure (generally performance on the color-word-naming

condition) to compare brain-injured patients and control participants. These researchers

have generally found that TBI patients were slower than control participants

(Guskiewicz, Riemann, Perrin, & Nashner, 1997; Rojas & Bennett, 1995; Stuss et al.,

1985; Trennery, Crosson, DeBoe, & Leber, 1989), although not always significantly so

(Guskiewicz et al., 1997; Potter, Jory, Bassett, Barrett, & Mychalkiw, 2002).

Interpretation of these results, however, is uncertain. Although researchers often consider

slower performance in the interference condition of the Stroop task as a sign of deficits in

inhibiting a prepotent response (or selective attention), it may simply result from

generalized slowing, a very common cognitive sequela of TBI (Lezak, 1995).

Researchers using the difference score between the color-word naming and the

color-naming conditions have obtained mixed results. In three studies, TBI patients had a

significantly greater mean interference score than control participants (Bohnen,

Twijnstra, & Jolles, 1992; McDowell et al., 1997). In three other studies, however, there

was no significant difference between the two groups (Batchelor et al., 1995; Bate,

Mathias, & Crawford, 2001; Bohnen, Jolles, & Twijnstra, 1992), and in one study,









Table 1-1. Studies of performance of TBI patients on the Stroop task


Authors
Asikainen et al.
(1999)

Batchelor et al.
(1995)

Bate et al.
(2001)

Bohnen et al.
(1992)

Bohnen et al.
(1992)


Guskiewicz et
al. (1997)

McDowell et
al. (1997)

Ponsford &
Kinsella
(1992)
Potter et al.
(2002)

Rojas &
Bennett
(1995)
Stuss et al.
(1985)

Trennery et al.
(1989)
Vakil et al.
(1995)


Participants
92 severe TBI,
22 moderate
TBI
35 mild TBI,
35 controls

35 severe TBI, 35
controls

44 mild TBI, 44
controls

10 mild TBI with
PCS, 10 mild
TBI without
PCS
11 athlete mild
TBI, 11 athlete
controls
25 moderate-to-
severe TBI, 24
controls
47 severe TBI, 30
orthopoedic
patients
24 mild TBI, 24
controls

25 mild TBI, 25
controls

20 CHI, 20
controls

65 TBI, 106
controls
25 TBI (all
severity levels)
27 controls


Tasksa
CWN


WR, CN,
CWN,
mod. CWN
WR, CN,
CWN,
mod. CWN
WR, CN,
CWN,
mod. CWN
WR, CN,
CWN,
mod. CWN

WR, CN,
CWN

CN, CWN


WR, CN,
CWN

CWN, PC
cong., PC
incong.
CWN, WCN


WR, CN,
CWN

CN, CWN,
WCN
CN, CWN


Score
CWN


CWN CN


CWN CN


CWN CN


CWN CN



CWN


CWN CN


CWN- CN

Errors
CWN
PC incong.
Errors
CWN


CN

CWN
CWN


Results
No significant diff.


No significant diff.


No significant diff.


No significant diff.


Greater interference
in TBI patients


No significant diff.


Greater interference
in TBI

Greater interference
in controls
No significant diff.
No significant diff.
No significant diff.
No significant diff.
TBI slower than
controls

TBI slower than
controls
No significant diff.
TBI slower than


controls
CWN-CN Greater interference
in TBI


aWR = word reading; CN = color naming (naming the color of a row of Xs); CWN = color-word
naming (naming the color of incongruent color words); mod. CWN = modified color-word
naming, in which a few word-reading items are interspersed among incongruent color-word
naming items; WCN = word-color reading (reading color words printed in incongruent colors);
PC cong. and PC incong. = computer version where congruency varies with each trial but task
remains constant.









control participants actually had a greater interference score than brain-injured patients

(Ponsford & Kinsella, 1992).3 Finally, researchers examined error rates in only two

studies. Ponsford and Kinsella (1992) failed to find any difference between control

participants and TBI patients on a card version of the Stroop. Potter et al. (2002) found no

significant difference between the two groups on either a card version or a computer

version of the Stroop.

Three main possible explanations may account for these mixed findings.

* First, it may be that TBI patients do not actually have deficits in inhibiting
prepotent response tendencies, but rather difficulties on many tasks due to
nonspecific, generalized slowing. Although this explanation may account for the
results obtained with the Stroop task in brain-injured patients, the findings by
McDowell, Whyte and D'Esposito (1997) and Levine et al. (2000) suggest that
these patients do have more difficulties with tasks requiring some form of cognitive
control, of which inhibition of prepotent response tendencies is a prominent
example.

* Secondly, even though TBI patients may have difficulties with inhibitory
processes, the Stroop paradigm may not be sensitive to these deficits.

* Finally, the Stroop paradigm may have the potential to detect deficits in TBI
patients, but the versions of the task used in previous studies may not be sensitive
to these deficits. In support of this explanation, all the tasks used in previous
research on TBI were card versions of the Stroop paradigm. As pointed out by
Perlstein, Carter, Barch and Baird (1998), these versions require participants to
correct mistakes during task performance. As a result, errors and speed are
confounded in the RT data, leading to a possible decrease in sensitivity. Possibly
even more crucial is the fact that conditions are blocked in the card version of the
Stroop. In other words, the task requires participants to read words on a first page,
and name colors on a different page. As a result, it may become easier, over time,
to overcome the influence of the words of incongruent stimuli and to focus instead

3 Bohnen and his collaborators (Bohnen, Jolles et al., 1992; Bohnen, Twijnstra et al.,
1992) have also developed a new condition, in which a few word-reading items are
interspersed among incongruent color-word naming items. For this new condition, they
found an increased interference score in brain-injured patients in at least two independent
studies. Interestingly, their modification is not without commonalities with our computer
version, as in both cases participants have to switch rapidly between reading words and
naming colors, thereby increasing dramatically the context maintenance requirements of
the task.









on the color of the ink. As noted previously, deficits in context maintenance are
more likely to become apparent in tasks with greater working memory demands. In
a task with mixed conditions, such as the computer version used in our study, color
naming and word reading vary randomly with each new trial, and participants must
constantly update, maintain and manipulate task instructions to perform
appropriately. If the difficulties of brain-injured patients are indeed due to
impairments in maintaining context representations, then such a task, combining
working memory and inhibition of prepotent response tendency requirements, may
be better suited than card versions of the Stroop paradigm to detect deficits in
context maintenance due to TBI.

Predictions

In our study, we examined the performance of TBI patients on two versions of the

Stroop paradigm: the "classic" three-card version of Golden (1978), often used in the

clinical context, and a single-trial computerized version developed by Cohen et al.

(1999). In the single-trial version, the conditions varied with each trial, and participants

were first given an instructional cue (reading the word or naming the color), and then

shown an imperative stimulus varying in congruency. Previous single-trial computer

versions of the Stroop typically consisted of three color-naming conditions of the present

design (Perlstein et al., 1998; Salo et al., 2001). Our addition of the three word-reading

conditions requires participants to maintain the instruction on-line and use it to bias the

processing of the imperative stimulus. In other words, this modification introduces a

working-memory component to the basic Stroop paradigm, which we believe will

increase its sensitivity to context maintenance deficits induced by TBI. To examine even

more closely the influence of working memory on task performance, we also manipulated

the delay between the instruction and the imperative stimulus, with a short, 1-second

delay and a long, 5-second delay (Figure 2.1 shows the variables in the single-trial Stroop

task).









Our main hypothesis stems from our review of the literature and the characteristics

of these two tasks: we predicted that the computer-based, single-trial Stroop (but not the

card Stroop) would be sensitive to context maintenance deficits in TBI patients. More

specifically, we predicted that TBI patients would be slower than control participants in

each of the three conditions of the card Stroop (i.e., generalized slowing), but that they

would not be disproportionately slower in the incongruent condition (no increased

difference score). Conversely, for the single-trial Stroop, we predicted that, in addition to

generalized slowing, TBI patients would show disproportionately longer RTs than control

participants in the incongruent, color-naming condition, resulting in an increased

difference score between incongruent and neutral conditions. In terms of error rates, we

expected TBI patients to make more errors than control participants only in the

incongruent color-naming condition. We also predicted that, due to a degradation of

context representations, brain-injured patients would make more errors at the long than

short delay, whereas the performance of control participants would not be influenced by

delay (resulting in an interaction between group and delay in the incongruent,

color-naming condition).4

An additional goal of our study was to examine the relation between performance

of TBI patients on the Stroop task and subjective reports of symptomatology. As imaging

techniques proliferate and become a standard tool for evaluating brain trauma, the role of

neuropsychologists in clinical practice tends to move away from lesion localization to a


4 Unlike for RT data, we did not compute an interference score incongruentt neutral) for
error rates, because we expected very low error rates in the neutral condition. Instead, our
prediction regarding a differential deficit of TBI patients at the long delay represented our
attempt to disentangle overall performance impairment from specific deficits in
maintenance of context representations.









more refined description of patients' functional impairments and potential for

rehabilitation (Groth-Marnat, 2000). In the future, patient's needs and pressure from

insurance companies will most likely result in a demand for neuropsychological tests to

become better predictors of symptomatology in brain-injured patients. To our knowledge,

until now, only one study (Levine et al., 2000) has examined the relationship between

Stroop performance and reports of impairment after TBI. Levine et al. have found a small

correlation (R2 = 0.14) between Stroop RT interference score and total score on the

Sickness Impact Profile, a multi-dimensional measure of health status for individuals

with general medical conditions (Bergner, Bobbitt, Pollard, Martin, & Gilson, 1976;

Pollard, Bobbitt, Bergner, Martin, & Gilson, 1976).

In our study, we used self-reports and significant-other reports of symptomatology

obtained on a modified version of the Neurobehavioral Rating Scale (NRS), a 27-item

measure designed specifically to assess impairments after brain injury. We predicted that

performance on the single-trial Stroop, being more sensitive to impairments in TBI,

would be more strongly correlated than performance on the card Stroop with reports of

symptomatology. In addition, we were interested in the following questions:

* In the single-trial Stroop, would RTs and errors contribute independently to
predicting reports of impairments, or would one of them account for most of the
predicted variance in the NRS?

* Would errors be a better predictor of NRS scores at the long than at the short delay,
suggesting that performance after a delay (i.e., with increased working memory
requirements) is more strongly related to reports of impairment in TBI?

* Would self-reports or significant-other reports be most strongly correlated with
Stroop performance? In relation to this last question, Sunderland, Harris and
Baddeley (1983) found that significant-others' ratings of TBI patients' memory
difficulties correlated more strongly than self-reports with their performance on
standard memory tests. If such results were replicated in our study, it would suggest
that significant others, when present, may be better informants of impairment after






13


brain injury than the patients themselves, a finding that may have important
implications for the assessment and management of TBI.














CHAPTER 2
METHODS

Participants

Participants were 23 individuals with moderate-to-severe closed-head injury and 21

demographically-matched, healthy participants recruited through advertising in the local

community. Consistent with the definition of Alexander (1995), we defined injuries of

the moderate-to-severe range as any injury resulting in loss of consciousness (LOC)

longer than 30 minutes, post-traumatic amnesia (PTA) longer than 24 hours or presence

of lasting focal signs and positive neuroimaging findings.1 We did not distinguish

between moderate and severe brain injury. Data were collected from interviews with TBI

patients at the time of the experiment. Because most patients did not remember their

score on the Glasgow Coma Scale (GCS) at the time of injury, this variable was not used

in our classification of severity. Excluded from the study were all individuals below 18 or

above 55 years of age, and all participants reporting a history of schizophrenia or bipolar

disorder, a history of formally diagnosed attention-deficit/hyperactivity disorder, a

history of learning disability, a history of chronic alcohol or drug abuse extending within

a 6-months period before testing, a history of other acquired brain damage (e.g., stroke,

epilepsy), a history of inpatient psychiatric hospitalization predating brain injury, or a

history of significant depression or anxiety predating brain injury and extending within a


1 In our sample, all of the participants classified as moderate-to-severe TBI, but one
reported an LOC greater than 30 minutes or a PTA greater than 24 hours. This participant
reported an LOC and PTA of 1 minute, but presented with linear skull fracture, subdural
hematoma and significantly impaired senses of smell and taste.









2-year period before testing. In addition, brain-injured participants had to be at least 1

month after injury. Exclusion based on current medication use was determined on a

case-by-case basis, depending on whether the medication had documented effects on

concentration and attention.

Demographic variables of brain-injured patients and control participants are shown

in Table 2-1. The two groups did not differ significantly in terms of gender, &(1, 42), 1 >

0.5, age, t(42) = -1.8, p > 0.08, education, t(42) = 0.2, p > 0.8, mother's education, t(42) =

1.5, 2 > 0.1, and father's education, t(42) = -0.6, 2 > 0.5. Median scores (ranges) for time

since injury, LOC, and PTA in the TBI group were 65 months (1.5 to 444 months), 72

hours (1 minute to 2160 hours) and 692 hours (1 minute to 4320 hours), respectively.

Materials

Card Stroop

We used the three-card version of the Stroop test most frequently used in clinical

settings (Golden, 1978). This version includes three cards, administered in fixed order,

that consist of five columns of 20 items each. The first card contains lists of color words

(BLUE, GREEN and RED) printed in black ink, which participants were instructed to

read aloud as quickly and as accurately as possible (word-reading condition). The second

card contains rows of four colored Xs and participants were instructed to name the color

of the printed Xs (color-naming condition). The third card contains lists of words printed

in incongruent colors (e.g., BLUE printed in red ink), and participants were required to

name the printed color of the ink (color-word-naming or conflict condition). For each

card, participants were given 45 seconds to complete as many items as possible, without

skipping any. Whenever participants made a reading or naming error, they were briefly

instructed to correct their response.









Table 2-1. Demographics and injury variables by group
Controls participants TBI patients
(n = 21) (n = 23)
Males/females 12/9 15/8
M SD M SD
Age (years) 36.3 10.1 42.0 11.2
Education (years) 14.0 1.6 13.9 2.0
Mother's education (years) 13.9 3.0 12.7 2.3
Father's education (years) 13.6 3.4 14.1 2.6

For the different subtests of the card Stroop, Golden (1975) found test-retest

reliabilities ranging from 0.73 to 0.89 in a normal sample. Golden (1976) also found the

card Stroop to be fairly accurate in distinguishing brain-injured individuals from normal

participants and psychiatric patients.

Single-Trial Stroop

Participants also performed a computer-based, single-trial version of the Stroop

task, originally developed by Cohen et al. (1999). Stimuli were presented on an Apple

Macintosh computer using PsyScope software (Cohen, MacWhinney, Flatt, & Provost,

1993). At the beginning of each trial, they were presented with an auditory cue ("Word"

or "Color"), followed by a visual stimulus. They were instructed to respond verbally to

the stimulus, and their RTs were determined by a voice-activated relay connected to the

computer. The examiner recorded responses manually for coding of accuracy. The same

basic three colors and color words (blue, red, and green) as in the card Stroop were used.

There were two task conditions (color naming and word reading), two delay conditions

(corresponding to a delay of Is and 5s, respectively, between the cue and the probe); and

three congruency conditions (congruent, neutral, and incongruent); for a total of 12 trial

types (Figure 2-1). In the congruent condition, stimuli were color words presented in their

own color ink. Incongruent stimuli consisted of color words printed in one of the two









remaining colors, and neutral stimuli were four colored Xs for color-naming trials and

color words displayed in white for word-reading trials.2 A total of 180 experimental trials

were distributed randomly across trial types, resulting in an average number of 15 trials

of each type. Because of random effects, this number was not always equal to 15, but fell

between 8 and 24. Because of its novelty, reliability and validity estimates were not

available for this computer-based task.


Figure 2-1. Schema of the single-trial Stroop. Participants were first presented with an
instructional cue ("Color" or "Word"), followed by a delay (1 or 5 seconds)
and an imperative stimulus (congruent, neutral or incongruent).

Neurobehavioral Rating Scale (NRS)

A modified version of the NRS was used to estimate current functioning in TBI

patients and control participants. This 27-item instrument was originally based on

clinician ratings, and was specifically designed to estimate cognitive, behavioral, and

emotional changes after brain injury (Levin et al., 1987). We used the modified version,


2 Although one could argue that white is also a color, and that neutral stimuli in the
word-reading condition were not really "neutral", the important point here is that "White"
was not part of the response set in any of the trial types. A further indication that the
white color did not substantially interfere with the reading of the color words is that none
of our participants ever answered "White" to any of the stimuli.


Short


PIP Congruent

Neutral

Incongruent









developed by Mathias and Coats (1999), to enable self-report. Each of the 27 items was

rated using a 5-point scale (from "never" to "very often"). In addition to self-report

ratings, each participant was given a copy of the scale to be completed by a significant

other. Previous versions of the NRS (Levin et al., 1987) showed high inter-rater

reliability (0.88 and 0.90 for two different samples of brain-injured patients) and internal

consistency for self and significant-other's ratings ranged from 0.92 to 0.96 (Mathias &

Coats, 1999). One brain-injured patient had a missing item on self-report. Four

significant others of normal participants and one significant other of a brain-injured

patient failed to return their questionnaires. Finally, one significant-other report of a

brain-injured patient had a missing item. NRS data for these participants were excluded

from relevant analyses.

In addition to the total score, previous factor-analytic studies of the NRS and the

NRS-Revised (NRS-R), a revised version of the NRS developed by Vanier, Mazaux,

Lambert, Dassa and Levin (2000), suggested that at least three factors (relatively stable

across studies) could be extracted from this instrument: a cognition/attention factor, a

hyperactivity factor, and a mood/affect factor (Levin et al., 1987; McCauley et al., 2001;

Vanier et al., 2000). Based on these previous studies, we calculated three surrogate factor

scores by summing ratings across selected items (Hair, Anderson, & Black, 1995),

separately for each of the two rating forms (self and significant-other's ratings). The

cognition/attention factor included Items 3 (Disorientation), 7 (Confusion), 10 (Memory),

and 23 (Planning). The hyperactivity factor included Items 8 (Disinhibition), 11

(Restlessness), 22 (Excitement), and 25 (Tenseness). The mood/affect factor included

Items 6 (Emotional (Guilt), 13 (Depressed Mood), 14 (Irritability), and 15 (Motivation).









WAIS-III subtests

The Digit Span forward and backward and Digit Symbol-Coding subtests of the

Weschler Adult Intelligence Test-Third Edition (WAIS-III) were administered as

estimates of working memory and processing speed, respectively. Reliability estimates of

the Digit Span ranged from 0.84 to 0.93 and its correlation with the working memory

index of the WAIS-III was estimated at 0.83 in a normative sample (WAIS-III and

WMS-III Technical Manual, 1997). Test-retest reliability estimates of the Digit

Symbol-Coding ranged from 0.81 to 0.87 and its correlation with the processing speed

index of the WAIS-III was estimated at 0.91 (WAIS-III and WMS-III Technical Manual,

1997).

North American Adult Reading Test (NAART)

The NAART is often used in neuropsychological evaluations as a measure of

premorbid intelligence after brain injury or the onset of a dementia (Spreen & Strauss,

1991). It has been shown to be a good estimate of WAIS-R and WAIS-III scores,

especially in the average range of intellectual abilities (Johnstone, Callahan, Kapila, &

Bouman, 1996). However, it may tend to overestimate low IQ scores and underestimate

high IQ scores (Johnstone et al., 1996).

Procedures

Participants were administered all of the tasks described above as part of a larger

battery. The total duration of testing fell between 4 and 5 hours, and order was

counterbalanced across tasks.

Statistical Analyses

Analyses of the Stroop data were conducted separately for the word-reading and

color-naming tasks. For the card Stroop, the primary dependent variables were the









number of items completed on each card within the 45-second time-limit. We conducted

a 2 x 2 analysis of variance (ANOVA) with group as the between-subjects factor and

condition (color naming and color-word naming) as the within-subject factor. We

conducted a separate t-test for the word-reading condition.

For the single-trial Stroop task, RTs and error rates were analyzed separately. For

each trial type and each participant, we calculated the median RT for correct responses3

and conducted separate 2-Group x 2-Delay x 3-Congruency ANOVAs for the

word-reading and color-naming tasks. Between-group planned comparisons evaluated

interference effects incongruentt RT vs neutral RT) in the color-naming task. The raw

data for error rates were normalized using the arcsin transformation (Neter, Wasserman,

& Kutner, 1985) before all analyses. We conducted separate 2 x 2 x 3 ANOVAs for

word-reading and color-naming tasks. To statistically evaluate the primary prediction of

interest, we also conducted a focused Group x Delay ANOVA on the incongruent

color-naming condition to test our prediction of increased error rates in TBI patients at

the long delay.

Given the expected restricted range of NRS scores in healthy controls, we restricted

our analyses of correlations between ratings on the NRS and performance on the Stroop

task to the TBI group. To limit the number of correlations computed, we used five simple

performance indices from the Stroop tasks:

* For the card Stroop, total number of items computed in the color-word naming
condition.


3 In situations where subjects' variability is expected to be important, and in which the
probability of outliers may be different across conditions, Median RTs are fairly robust to
outliers and have good power compared to other measures of central tendency (Ratcliff,
1993).






21


For the single-trial Stroop, median RT in the incongruent word-reading task,
median RT in the incongruent condition of the color-naming task, error rates in the
word-reading task, and error rates in the color-naming task (for the four indices of
the single-trial Stroop, we used the average of scores at the short and at the long
delay).














CHAPTER 3
RESULTS

Neuropsychological Tests

Means and standard deviations of the performance of TBI patients and control

participants on various neuropsychological tests are presented in Table 3-1.

Table 3-1. Performance on neuropsychological tests by group
Control participants TBI patients
(n = 21) (n = 23)
M SD M SD
NAART (errors) 26.2 9.2 33.7 14.1
Digit span forward (raw scores) 10.4 1.9 9.7 2.5
Digit span backward (raw scores) 6.7 2.1 5.7 2.2
Digit symbol (# items completed)a 89.1 15.7 65.3 16.4
aBecause the Digit Symbol was included only later in our protocol, the data for this task include
only 19 control participants and 15 TBI patients out of our total sample.

The two groups did not differ in terms of their performance on the Digit Span

Forward, t(43) = 0.6, p > 0.3, or Backward, t(43) = 1.6, p > 0.11. Brain-injured patients,

however, committed significantly more errors on the NAART than control participants,

t(43) = -2.1, p = 0.046, suggesting slightly lower verbal IQ in TBI patients. TBI patients

also completed significantly fewer items on the Digit Symbol task than control

participants, t(32) = 4.3, p < 0.001, suggesting lower processing speed in TBI patients.

Card Stroop

Means and standard errors of the number of items completed on each card for each

group are presented in Figure 3-1. Control participants completed significantly more

items than TBI patients in the word-reading condition, t(42) = 6.2, P < 0.001. The 2 x 2

ANOVA including group as a between-factor and color-naming and color-word-naming










conditions as repeated measures yielded main effects of group, F(1, 42) = 45.3, p < 0.001,

and condition, F(1, 42) = 440.1, p < 0.001. The Group x Condition interaction was also

significant, F(1, 42) = 8.2, p < 0.01, reflecting a smaller interference score (number of

items completed in the color-naming condition minus number of items completed in the

color-word-naming condition) for brain-injured patients (M = 25.4, SD = 8.5) than for

control participants (M = 33.4, SD = 10.0). These results suggest that Stroop interference

was less strong in TBI patients than in control participants, a finding inconsistent with the

hypothesis of a deficit in inhibition of a prepotent response in brain-injured individuals.


120



80 -

600

40 -

20
WR CN CWN
Condition

Figure 3-1. Number of items completed on the card Stroop by group and condition. Error
bars represent standard errors.

Single-Trial Stroop: Speed-Accuracy Tradeoff Analyses

In order to ensure that there were no speed-accuracy tradeoffs in either group, we

calculated the correlations between median RT and error rates in the incongruent

condition for both groups, separately for color-naming and word-reading tasks. All

estimates were positive except for the control group in the color-naming task, where the

correlation was negative, r(20) = -0.25, although not significantly, 1(20) = -1.1, p > 0.2.









These results suggest that speed-accuracy tradeoffs did not play a significant role in the

performance of TBI patients or control participants.

Single-Trial Stroop: RTs

Means and standard deviations of median RT for each group and each condition are

presented in Table 3-2. In the word-reading task, control participants were faster than

brain-injured patients, F(1, 42) = 16.7, p < 0.001, and RTs were longer at the short than

long delay, F(1, 42) = 5.6, p < 0.03. There was also a significant effect of congruency,

F(2, 42) = 24.5, p < 0.001. Fisher's post-hoc comparisons revealed that RTs were longer

in the incongruent than in the congruent (p < 0.001) and neutral (p < 0.001) conditions,

whereas the difference between the neutral and congruent conditions was not significant

(p > 0.5). None of the interaction effects reached significance (all P > 0.1). These findings

suggest generalized slowing in TBI patients.

Table 3-2. Means and standard deviations of median RT (ms) in the single-trial Stroop
Control participants TBI patients
(n = 21) (n = 23)
M SD M SD
Word reading
Short delay
Congruent 820 174 951 139
Neutral 786 174 992 158
Incongruent 894 204 1082 286
Long delay
Congruent 767 149 936 152
Neutral 732 200 957 154
Incongruent 839 136 1015 172
Color naming
Short delay
Congruent 838 199 1027 179
Neutral 770 138 967 155
Incongruent 1035 192 1310 249
Long delay
Congruent 775 159 1014 215
Neutral 726 97 940 144
Incongruent 987 193 1265 258









In the color-naming task, control participants were faster than brain-injured

patients, F(1, 42) = 24.9, p < 0.001, and RTs were longer at the short delay than long

delay, F(1, 42) = 10.7, p < 0.003. There was also a significant effect of congruency, F(2,

42) = 89.8, p < 0.001. Fisher's post-hoc comparisons revealed that RTs were longer in the

incongruent than in the congruent (p < 0.001) and neutral (1 < 0.001) conditions. RTs

were also significantly longer in the congruent than in the neutral condition, p < 0.01.

None of the interaction effects reached significance, all p > 0.1.

More importantly, the two groups were compared with respect to the interference

score (difference between the incongruent and neutral conditions), at the short and long

delays separately. Control participants (M = 265, SD = 132) did not differ significantly

from TBI patients (M = 343, SD = 215) at the short delay, t(42) = -1.4, 2 > 0.1, nor at the

long delay (control participants: M = 262, SD = 130; TBI patients: M = 326, SD = 170),

t(42) = -1.4, p > 0.1. These results suggest generalized slowing but no disproportionate

Stroop interference in TBI patient.

Single-Trial Stroop: Error Rates

Means and standard deviations for the error rates are displayed in Table 3-3, and

the ANOVA results are presented in Table 3-4. In order to decompose the significant

interaction between group and congruency in the word-naming task, we conducted

separate analyses for each of the congruency condition, with delay as a within-subjects

factor and group as a between-subjects factor. In the incongruent, but not in the congruent

and neutral conditions, brain-injured patients committed significantly more errors than

control participants, F(1, 42) = 6.8, p < 0.03. None of the other effects were statistically

significant.









Table 3-3. Means and standard deviations of error rates (%) in the single-trial Stroop
Control participants TBI patients
(n = 21) (n = 23)
M SD M SD
Word reading
Short delay
Congruent 0 0 0 0
Neutral 0 0 .4 1.7
Incongruent 0 0 2.0 4.6
Long delay
Congruent 0 0 0 0
Neutral 0 0 .7 2.3
Incongruent .3 1.3 3.1 6.0
Color naming
Short delay
Congruent 0 0 0 0
Neutral 0 0 0 0
Incongruent 7.3 8.6 14.5 18.3
Long delay
Congruent .7 2.2 0 0
Neutral .4 1.8 .2 1.0
Incongruent 2.8 6.0 16.5 23.1

Table 3-4. ANOVA for error rates in the single-trial Stroop
Word reading Color naming
Source df F df F P
Between subjects Between subjects
Group 3 6.8 0.013 3 5.0 0.03
Within subjects Within Subjects
Delay 1 1.3 > 0.2 1 .1 > 0.5
Congruency 2 6.7 0.002 2 20.9 < 0.01
Group x delay 3 .6 > 0.4 3 2.2 > 0.1
Group x congruency 6 5.3 0.007 6 5.9 0.004
Delay x congruency 2 .7 > 0.5 2 .7 > 0.4
Group x delay x congruency 6 .2 > 0.5 6 3.3 0.04

We followed the same procedure to decompose the significant three-way

interaction in the color-naming task. In the congruent and neutral conditions, none of the

effects reached significance (all es > 0.1). In the incongruent condition, there was a main

effect of group, F(1, 42) = 7.2, p < 0.02, and, more importantly, a near-significant

interaction between group and delay, F(1, 42) = 4.0, p = 0.052. Consistent with our










primary prediction, the difference between the two groups was significant at the long

delay, 1(42) = 3.2, p < 0.003, but not at the short delay, t(42) = 1.5, p > 0.1 (Figure 3-2).

In terms of effect sizes, the difference between the two groups was moderate at the short

delay, d = 0.49, and large at the long delay, d = 0.97. In addition, complementary

analyses revealed that control participants committed significantly fewer errors at the

long than short delay, t(42) = 2.7, p < 0.02, whereas error rates for TBI patients did not

differ as a function of delay, t(42) = -0.3, p > 0.5. Consistent with our primary prediction,

these results suggest a disruption over time of context representations in TBI patients

compared to healthy participants.


25

20 :n tr,:,

~.15
0
10

5

0 T
Short Long
Delay

Figure 3-2. Error rates in the incongruent color-naming condition of the single-trial
Stroop by Group and Delay. Error bars represent standard errors.

Neurobehavioral Rating Scales

Cronbach-alpha internal consistency estimates for self and significant-others'

ratings are presented in Table 3-5. Internal consistency estimates were satisfactory (above

0.80) for self-ratings in the TBI group, but tended to be lower (below 0.80) in the control

group. They also tended to be low in certain subscales of significant-others' ratings, for

both the TBI and the control group. The lower estimates for the control group are not









surprising, given that the NRS was developed specifically for assessing impairment after

brain injury. In addition, they should not constitute a significant threat to the validity of

our study, since our analyses of the relations between reports of impairment and

performance on the Stroop tasks were restricted to the TBI group. The lower internal

consistency estimates for the significant-others' ratings of brain-injured patients, on the

other hand, suggest that the relationship between these ratings and the performance of

TBI patients should be interpreted with caution.

Table 3-5. Internal consistency (alpha-cronbach) estimates for the NRS for self and
significant-other's ratings by group
Control participants TBI patients
Self-rating Sign. Other Self-rating Sign. Other
Total 0.93 (21)a 0.93 (17) 0.93 (22) 0.88 (21)
Cognition 0.69 (21) 0.60 (17) 0.82 (23) 0.85 (22)
Hyperactivity 0.65 (21) 0.74 (17) 0.86 (23) 0.66 (21)
Mood 0.80 (21) 0.82 (17) 0.82 (23) 0.45 (22)
aThe number of participants used for this particular estimate is in parentheses. Because of missing
questionnaires and missing items, this number varies slightly for each subscale.

Means and standard deviations for each of the surrogate factor scores and for the

total NRS score are reported in Table 3-6.

Table 3-6. Means and standard deviations of NRS surrogate factor scores and total score
by Group
Control participants TBI patients
(n = 17) (n = 20)
M SD M SD t
Self-ratings
Cognition 0.82 0.53 2.06 1.02 4.8 < 0.001
Hyperactivity 1.02 0.58 1.79 1.04 2.8 < 0.01
Affect 0.92 0.65 1.84 1.02 3.6 < 0.001
Total 0.82 0.46 1.81 0.79 5.2 < 0.001
Significant-other
ratings
Cognition 0.62 0.57 2.09 1.00 5.0 < 0.001
Hyperactivity 1.16 0.70 1.55 0.68 1.4 0.17
Affect 1.11 0.71 1.50 0.64 1.8 0.08
Total 0.88 0.54 1.57 0.54 3.5 < 0.002









Data from four control participants and three TBI patients were not included due to

missing items or because the significant other failed to mail or bring back the

questionnaire. NRS scores show that brain-injured patients viewed themselves as having

significantly more cognitive and emotional difficulties and as being more hyperactive

than control participants. Significant others also rated TBI patients as more cognitively

impaired than control participants. There was, however, no significant difference between

the two groups in the significant-other ratings of hyperactivity or emotional functioning.

Table 3-7. Correlations between self and significant-others' ratings on the NRS
C/S H/S M/S T/S C/O H/O M/O T/O
Cognition -self 0.61* 0.55* 0.82* 0.19 0.16 0.25 0.20
Hyperactivity self 0.33 0.59** 0.82** 0.10 0.08 0.27 0.13
Mood- self 0.52** 0.61* 0.87** 0.01 0.25 0.14 0.19
Total- self 0.79** 0.72** 0.82* 0.10 0.19 0.25 0.21
Cognition- other 0.27 0.18 -0.18 0.04 0.78** 0.79** 0.88*
Hyperactivity other 0.21 0.64** 0.44* 0.41 0.07 0.82** 0.93**
Mood- other 0.28 0.62** 0.47* 0.55** 0.29 0.65 0.93**
Total other 0.33 0.59** 0.23 0.44 0.61** 0.72** 0.80** -
Note. Upper right = control participants; Lower left = TBI patients. In the upper row, C stands for
cognition, H for hyperactivity, M for mood and T for total; S stands for self-ratings and O for
significant-other's ratings.
*p < 0.05. **p < 0.01.

Correlations between the surrogate factor scores of the NRS and between self and

significant-other ratings are presented in Table 3-7. Correlations between the scores on

the various factor scores of the NRS were generally moderate to high within raters, but

low between self and significant-other ratings, with the notable exception of significant

and moderate correlations between self-ratings of hyperactivity and significant-other's

ratings of hyperactivity, mood and total NRS score in the TBI group.

Correlations between ratings on the NRS and performance indices on the Stroop

tasks are presented in Table 3.8. As predicted, performance on the card Stroop was not

significantly correlated with self and significant-others' ratings of reports of









symptomatology. RTs in the single-trial task were also unrelated to reports of

impairment, as were error rates in the word-reading task, with the exception of a

significant correlation with self-ratings of mood. Error rates in the color-naming task,

however, were significantly correlated with all three surrogate factor scores of the NRS

for self-ratings, and with significant-others' ratings of hyperactivity.

Table 3-8. Correlations between ratings on the NRS and Stroop performance (TBI group)
Card RT word RT color Errors Errors
Stroopa word color
Cognition -self 0.34 0.15 0.21 0.33 0.45
Hyperactivity self 0.06 0.25 0.38 0.35 0.48*
Mood self 0.05 0.22 0.27 0.46* 0.68*
Total self 0.26 0.24 0.31 0.49 0.61*
Cognition other 0.15 0.17 0.22 -0.09 -0.01
Hyperactivity other -0.13 0.12 0.27 0.33 0.49*
Mood other 0.07 0.01 0.11 0.22 0.41
Total other 0.10 0.18 0.38 0.19 0.34
aFor the card Stroop, we reversed the sign of the correlation, as a larger number of items
completed means better performance, whereas larger RTs and error rates in the single-trial Stroop
mean worse performance.
*p < 0.05. **p < 0.01.

We conducted multiple regression analyses to examine the extent to which self and

significant-others' ratings independently predicted color-naming error rates (Table 3-9).

Table 3-9. Multiple regressions with self and significant-others' ratings as independent
variables and error rates (color naming) as the dependent variables (TBI
group)
Total R Self-ratingsa Significant other
Cognition 0.21 0.47 -0.13
Hyperactivity 0.28 0.25 0.33
Mood 0.47 0.62* 0.12
Total 0.34 0.55* 0.08
aNormalized beta weights.
*p < 0.05.

For the cognition and mood subscales, and for the total NRS score, self-ratings

remained significant predictors of error rates whereas significant-others' ratings did not

reliably contribute to the predicted variance. These results suggest that self-reports of









impairment in TBI may be more related than significant-other reports to performance on

cognitive tasks.

We also computed the correlations between self-ratings on the NRS and

color-naming error rates at the long and short delays separately (Table 3-10). As

correlations tended to be qualitatively higher with errors at the long delay, we conducted

multiple regressions with error rates at the short and long delays as independent variables,

and cognition, hyperactivity, mood and total score as dependent variables. As seen in

Table 3-11, normalized beta weights were generally not significant, except for mood,

where error rates at the long, but not at the short delay, were uniquely predictive of

self-ratings.

Table 3-10. Correlations between self-ratings on the NRS and color-naming error rates at
the long and short delays separately (TBI group)
Errors short delay Errors long delay
Cognition 0.38 0.45
Hyperactivity 0.44* 0.45*
Mood 0.56** 0.70**
Total 0.54** 0.59**
*p < 0.05. **p < 0.01.

Table 3-11. Multiple regressions with color-naming error rates at the long and short
delays as independent variables and NRS self-ratings as dependent variables
(TBI group)
Total R Errors short delaya Errors long delaya
Cognition 0.21 0.10 0.38
Hyperactivity 0.23 0.24 0.27
Mood 0.49 0.07 0.65*
Total 0.37 0.21 0.43
aNormalized beta weights.
*p < 0.05.














CHAPTER 4
DISCUSSION

The results of our study suggest that the performance of brain-injured patients on

two versions of the Stroop paradigm is consistent with the presence of a deficit in the

preparation to override prepotent response tendencies. On the card Stroop, TBI patients

did not show deficits in inhibition of prepotent response tendencies beyond generalized

slowing: they were slower than control participants on every condition of the task, but not

disproportionately slower in the color-word naming (or interference) condition.1 This

finding is consistent with previous literature showing generalized slowing of TBI patients

(Guskiewicz et al., 1997; Rojas & Bennett, 1995; Stuss et al., 1985; Trennery et al., 1989)

but mixed results across studies regarding disproportionate slowing in the interference

condition of the Stroop task (Batchelor et al., 1995; Bate et al., 2001; Bohnen, Jolles et

al., 1992; Bohnen, Twijnstra et al., 1992; McDowell et al., 1997; Ponsford & Kinsella,

1992). The same result was observed when analyzing the RT data from the single-trial,

computer-based version of the task. Again, TBI patients were slower on every condition

(for both the color-naming and the word-reading tasks), but not disproportionately slower



1 In fact, as shown in the results section, the difference score between the color-word
naming and the color-naming conditions was larger in control participants than in TBI
patients. If considered too literally, these results may suggest that individuals having
sustained brain injury show better inhibition of prepotent response tendencies than
healthy participants. As pointed out by Chapman, Chapman, Curran, and Miller (1994),
though, RTs' difference scores cannot always be assumed to be free of the effects of
generalized slowing, especially when the two groups being considered differ significantly
in terms of their baseline scores. This is precisely the case in our study, where
brain-injured patients completed fewer items than controls in the color-naming condition.









on the incongruent, color-naming condition compared to control participants. In contrast,

examination of error rates revealed that TBI patients showed a pattern of performance

suggestive of inhibitory deficits: they committed more errors than control participants on

the incongruent, but not on the congruent and neutral conditions of the color-naming task.

Even more importantly, their error rates differed from those of control participants at the

long, but not at the short delay of the incongruent condition. This later result is consistent

with the main prediction stemming from the hypothesis of a context maintenance deficit

in a given population (Cohen et al., 1999; Cohen et al., 1996): the poorest performance

should be observed in conditions in which there are both inhibitory and working memory

requirements. This is precisely the case for the long delay, incongruent color-naming

condition of the single-trial Stroop, where participants have to maintain the context (i.e.,

task instructions) over the delay in order to inhibit the prepotent tendency to read the

word. Our overall findings parallel those ofPerlstein et al. (1998), who found that

schizophrenics showed increased error rate interference on the single-trial Stroop, but no

increased RT interference on the card or single-trial Stroop. They suggest that working

memory and inhibition of prepotent response tendencies are two closely related

processes, a hypothesis consistent with the results of Roberts, Hager and Heron (1994),

who found that requiring healthy participants to maintain and manipulate information in

working memory decreased performance on the antisaccade task.

In order to examine the ecological validity (i.e., the relation between performance

and day-to-day difficulties experienced by TBI patients) of the two versions of the Stroop

paradigm used in our study, we compared various indices of performance of TBI patients

with their report and significant-others' report of impairment after injury along the three









surrogate factor scores of the NRS (Levin et al., 1987): cognitive functioning,

hyperactivity and mood. Number of items completed in the card Stroop and median RTs

in the single-trial Stroop were not significantly related to reports of impairment. Error

rates, on the other hand, especially in the color-naming task, were significantly correlated

with reports of impairment along all three surrogate factor scores. The correlations

generally fell in the moderate range and were of greater magnitude than the relation

between RT interference score and Sickness Impact Profile total score reported in

(Levine et al., 2000). Interestingly, the relation between performance and impairment

after injury tended to be stronger for self-reports than significant-others' reports. This

result contrasts with the finding by Sunderland et al. (1983) that significant-others'

ratings of TBI patients' memory difficulties correlated more strongly than self-reports

with their performance on standard memory tests. This discrepancy between the two

studies raises the possibility that self-report of TBI patients may be more reliable for

general areas of functioning (i.e., cognition or mood) than for specific deficits after

injury. Our results suggest that the long-lasting emotional, behavioral and cognitive

difficulties incurred by individuals having sustained brain injury (Mathias & Coats, 1999;

Olver et al., 1996) may be related to and predicted by their performance on tasks

requiring the maintenance of context representations. Future studies are needed to

examine whether the relation between cognitive performance and self-reports of

impairment generalizes to objective psychosocial outcomes after brain injury, such as

social isolation, employment status or ability to drive.

One pervasive issue confronted by cognitive neuropsychologists is the difficulty to

show the specificity of deficits in a given population. Simply stated, the finding that a









group of patients differs significantly from control participants on a given task is not

sufficient to conclude that this group is specifically impaired regarding the ability

purportedly measured by the task: indeed, several general factors, such as poor attention,

concentration, or motivation, result in poor performance across a broad array of tasks.

Moreover, as Chapman, Chapman, Curran, and Miller (1994) pointed out, the use of

difference scores for RTs (or number of items completed), such as in the traditional card

Stroop, cannot always be assumed to "remove" the effects of generalized slowing.

Similarly, the use of difference scores for error rates does not always correct for the

effects of generalized deficits, especially when the two groups compared in the study

differ regarding their baseline error rates (Chapman & Chapman, 1988, 1989). In our

study, for instance, TBI patients' and control participants' error rates differed in the

incongruent, but not in the congruent or neutral conditions of the color-naming task of the

single-trial Stroop. This finding, however, does not provide strong evidence for the

specificity of the deficits shown by brain-injured patients, because both groups were at

floor levels in the congruent and neutral conditions. The fact, however, that the two

groups differed significantly at the long, but not at the short delay of the incongruent

condition, suggests that the deficits experienced by TBI patients are specific rather than

generalized. As we mentioned before, this finding is exactly in line with the predictions

stemming from the hypothesis of a deficit in the maintenance of context representations.

As suggested by our study, the concept of context representations provides a

theoretical framework with the potential to explain a vast array of seemingly disparate

deficits in a variety of populations, such as schizophrenics (see e.g., Cohen et al., 1999;

Cohen et al., 1996; Perlstein et al., 1998) and older adults (see Braver & Barch, 2002;









Braver et al., 2001).2 This framework also provides a starting point for the study of the

neurochemical and neurobiological bases of cognitive control: namely, Cohen and

Servan-Schreiber (1992; 1993) have postulated that the maintenance of context

representations is related to the DA system in the prefrontal cortex (PFC), and

specifically the dorsolateral prefrontal cortex (DL-PFC). Indeed, a large body of literature

now supports an important role of the DL-PFC in working memory and cognitive control

(see Salmon, Heindel, & Hamilton, 2001, for a review), and several studies have shown

the role of the DA system on cognitive control tasks (Braver et al., 2001). As mentioned

before, neuroimaging studies (Levin & Kraus, 1994) suggest that the frontal lobes may be

especially vulnerable to TBI. Functional imaging studies of TBI patients during working

memory tasks resulted in a more dispersed pattern of cerebral activation than in healthy

controls, with increased lateralization to the right hemisphere, especially in the parietal

and frontal lobes (Christodoulou et al., 2001; McAllister et al., 1999). In addition,

McDowell, Whyte, & D'Esposito (1998) conducted a double-blind, placebo controlled

study on the effects of a DA receptor agonist on patients having sustained TBI. They

found that the performance of TBI patients who were administered the DA agonist

improved specifically on dual tasks and other tasks with cognitive control demands.

2 An alternate conceptual framework, developed by Kimberg and Farah (1993), may also
explain the findings of our study: these authors propose that a weakening of the
associations among working memory representations is responsible for the various
"executive" deficits observed following frontal lobe damage. They note that their account
of frontal lobe dysfunction and the maintenance of context representations model
developed by Cohen and Servan-Schreiber (1993) yield similar predictions for many
"executive" task, but differ in terms of the neural mechanisms underlying normal and
impaired performance. They also contend that a deficit in context maintenance, contrary
to their model, fails to account for the deficits in contextual memory observed in patients
with frontal lobe lesions. Although we chose to present the model of Cohen and
Servan-Schreiber as a framework for developing our predictions, we note that both
theories may equally explain the results of our study.









Taken together, these findings provide some evidence for the hypothesis that the

disturbances of cognitive control experienced by brain-injured patients may be mediated

by a disruption of the DA system in the DL-PFC.

It should be pointed out, however, that the DL-PFC is not the only area of the PFC

that has been implicated in the completion of the incongruent condition of the Stroop

task. In a study of patients with focal lesions localized in different areas of the frontal

lobes, Stuss, Floden, Alexander, Levine, and Katz (2001) found that an increased

interference score was observed for patients in the superior medial frontal areas. Lateral

lesions were associated with increased error rates rather than a disproportionate

interference score. In an event-related potential (ERP) study of Stroop performance in

mild TBI patients, Potter, Jory, Basset, Barrett and Mychalkiw (2002) found an increased

negativity, compared to control participants, in a latency range consistent with the

activation of the anterior cingulate (AC) gyms. They interpreted these findings as

suggesting that brain-injured patients may have to engage in more effortful processing to

achieve the same performance than control participants. Using functional magnetic

resonance imaging (fMRI), MacDonald, Cohen, Stenger and Carter (2000) found a

dissociation of the role of the DL-PFC and the AC on a long-delay version of the

single-trial Stroop task in healthy participants: whereas the DL-PFC was more active,

during task preparation, for the color-naming conditions (consistent with a role in

cognitive control), the AC was more active when responding to incongruent stimuli,

consistent with a role in conflict monitoring. These results, combined with the findings

by Potter et al. (2002), suggest that TBI patients may compensate for their difficulties

with cognitive control related to impaired functioning of the DL-PFC by an increased









activation of the AC during performance monitoring. Future research using functional

imaging and event related potentials (ERPs) to examine the relation between DL-PFC

and AC activation in TBI patients and healthy participants may confirm or disconfirm

this hypothesis and help uncover the mechanisms of recovery and compensation after

brain injury.

A number of limitations and alternate explanations to the present findings require

discussion. First, any study of TBI must take into account the heterogeneity of this

population in terms of time since injury, severity and injury localization (Lezak, 1995).

Previous studies examining the performance of TBI patients on the Stroop task have

varied greatly in terms of injury severity (see Table 1-1). In our experiment, all TBI

patients had sustained moderate-to-severe injuries. Obviously, our results cannot readily

generalize to mild TBI, which presents with a set of unique issues in terms of assessment

and outcome (Binder, 1997; Binder, Rohling, & Larrabee, 1997; Mathias & Coats, 1999).

Similarly, a direct comparison of our results with previous findings must take into

account the great variability in terms of time since injury across studies. According to

Rao and Lyketsos (2000), cognitive deficits after TBI can be divided into four successive

phases: a first period of coma (which may occur or not), a phase of confusion, agitation

and PTA (lasting from a few days to a month), a third period of recovery of cognitive

functions (up to 24 months), and a fourth phase of generally stable cognitive sequelae

(see Cripe, 1987; Levin et al., 1987). The performance of TBI patients on any cognitive

task, including the Stroop, is likely to vary depending on when they are tested after the

time of the trauma. In our study, most TBI participants (17 out of 23) were tested more

than two years after injury, corresponding to the phase of stable cognitive deficits









according to Rao and Lyketsos (2000) classification. The average time since injury in the

study conducted by Stuss et al. (1985) corresponded to the same fourth phase of cognitive

deficits (M = 2.6 years). By contrast, Bohnen, Jolles, and Twijnstra (1992) conducted

testing from 6 to 14 days after injury. Other researchers (Bate et al., 2001) included

patients with a wide range of time since injury. Future research is needed to explore in a

systematic way the relationship between time since injury and performance on the Stroop

task in TBI patients.

Secondly, the absence of disproportionate slowing of TBI patients in the

incongruent condition of the single-trial Stroop (and in the color-word naming condition

of the card Stroop) may have an alternate explanation than poor sensitivity of RTs to

context maintenance deficits. Namely, we did not discuss so far the possibility than the

prepotent tendency to read words may not be as strong for brain-injured patients as it is

for control participants. If this were the case, brain-injured patients would have less to

inhibit in order to name the color of incongruent stimuli, and the lack of disproportionate

slowing could not be readily interpreted as the absence of an inhibitory (or context

maintenance) deficit. Fortunately, the data from our study allow us to estimate the

prepotency of the tendency to read words in brain-injured patients and control

participants. First, we compared the two groups with respect to median RTs in the

word-reading task, averaged across all conditions. Control participants (M = 806, SD =

159) were indeed faster than TBI patients (M = 989, SD = 137) on this measure, t(42) =

-4.1, p < 0.001. This result, however, cannot be readily interpreted as indicative of a

lesser prepotency to read words in the TBI group, because previous literature shows that

brain injury consistently results in generalized slowing across a broad array of tasks









(Lezak, 1995). In order to control for generalized slowing, we conducted the same

analysis, but after entering the number of items completed in the Digit Symbol task as

covariate. Using this procedure, control participants (M = 835, SD = 42) did not differ

significantly from brain-injured patients (M = 974, SD = 51) in terms of residual scores,

F(1, 30) = 0.2, p > 0.1.3 As a different and complementary way to control for generalized

slowing, we compared the two groups on word-reading and color-naming tasks: if

brain-injured patients were less drawn than control participants to read the word of

presented stimuli, they should show a decreased difference score between word-reading

and color-naming tasks. In order to investigate this possibility, we averaged the median

RTs for all word-reading conditions, on the one hand, and all color-naming conditions, on

the other hand. We then subtracted these totals from one another and compared the two

groups with respect to this difference score. Using this procedure, control participants

actually had a smaller difference score (M = 49, SD = 75) than TBI patients (M = 98, SD

= 84), t(42) = -2.1, p < 0.05. These results suggest that a weaker prepotency to read

words in brain-injured patients is an unlikely explanation for our finding of a lack of

disproportionate slowing in the incongruent condition of the Stroop task for this group.

Another limitation of our study is that we did not measure error rates in the card

version of the Stroop paradigm. As a result, our finding of poor sensitivity of the card

Stroop to the inhibitory deficits after TBI may be due to a general poor sensitivity of RTs

rather than the characteristics of the task itself (i.e., the confounding of errors and number

of items completed and the fact that conditions are blocked, thereby reducing context

maintenance requirements). This is an important point, as our goal in this study was to

3 The above analysis was conducted with 19 control participants and 15 TBI patients out
of our total sample.









show that the performance of TBI patients on different versions of the Stroop paradigm

may be better explained by a deficit in context maintenance than by poor inhibition of

prepotent response tendencies. Future research should take this possibility into account

and include the measurement of error rates in the card Stroop to disentangle these two

explanations.

In addition, the pattern of error rates in the incongruent, color-naming condition of

the single-trial Stroop was somewhat different from our predictions: although there was,

as we predicted, an interaction between group and delay, this interaction was due to a

reduction in error rates from the short to the long delay in control participants rather than

an increase in TBI patients. In other words, the extended delay allowed healthy

participants to reduce their error rates, which we did not expect, whereas the error rates of

TBI patients remained stable across the two delay conditions. These results suggest that

TBI patients have difficulty in the preparation to override prepotent response tendencies:

contrary to healthy participants, they were not able to use the increased delay to reduce

their error rates. This finding may seem surprising, as the short and long delays were 1

and 5 seconds, respectively, and 1 second may seem a sufficient time to process task

instructions (which, in this study, consisted in a simple auditory word, "color" or "word")

and prepare the inhibition of the prepotent tendency to read the word. Future research

should attempt to examine the processes through which control participants manage to

decrease their error rates between 1 and 5 seconds, and to determine what specific

deficits in brain-injured patients are responsible for their inability to accomplish a similar

decrease. The use of specific strategies (such as rehearsing task instructions in one's mind









during the delay) in brain-injured patients and other populations may be one methodology

to explore these questions in future research.

Finally, our measure of everyday functioning, the NRS, does not have

well-established psychometric properties, especially under its self and significant-other

reports versions. Although useful, because designed specifically for impairments after

TBI, it should be supplemented, in future research, by more well-researched and

documented measures of everyday functioning, such as the Sickness Impact Profile

(Bergner et al., 1976; Pollard et al., 1976). As a positive note, though, we should point

out that the internal consistency estimates in our sample of brain-injured patients tended

to be satisfactory (above 0.80) for self-ratings, despite the small number of items

constituting each of the subscales. However, they tended to be lower for significant

others' ratings, especially for the hyperactivity and mood subscales, which, combined

with the finding of somewhat higher correlations of self-ratings with error rates, suggests

that TBI patients themselves rather than significant others may be better informants of

their symptomatology after brain injury. Future research is needed to confirm and explore

these findings in greater detail.

In our study, we found a significant and moderate correlation between error rates in

the incongruent color-naming condition of the single-trial Stroop and the three surrogate

factor scores of the NRS we extracted from the questionnaires. We failed, however, to

find any meaningful pattern of correlations that would relate performance of TBI patients

on the Stroop task and specific domains of impairment in everyday life. In fact, the two

domains of the NRS we expected to be most related with error rates were cognitive

functioning (because the Stroop task is a cognitive task in nature) and hyperactivity









(because it has frequently been related to malfunctioning of the frontal lobes, which also

play a crucial role in context maintenance). If anything, though, the strongest correlations

in our study were found with the mood factor of the NRS. Future research, using

well-established measures of impairment after illness or injury, should examine more

closely the relation between performance on the Stroop task and domains of everyday

functioning. We should point out, however, that our study is one of the first to examine

the ecological validity of the Stroop task in TBI patients, and that the finding of moderate

correlations with a self-report measure of impairment is an encouraging finding regarding

the usefulness of the Stroop paradigm for predicting adjustment after brain injury.

In conclusion, the results of our study suggest that:

* The performance of TBI patients on a single-trial version of the Stroop task
designed to maximize inhibitory and working memory requirements is consistent
with a deficit in the preparation to override prepotent response tendencies.

* Error rates are more sensitive than RTs to these deficits in TBI populations.

* Deficits in the preparation to override prepotent response tendencies in TBI patients
have meaningful ecological implications, that is, they are significantly related to
self-reports of impairment after brain-injury.

These findings suggest the possibility of developing new tools for the assessment

and prediction of cognitive deficits after TBI. They also offer a framework for the study

of the neural mechanisms responsible for the impairment, recovery and compensation

strategies used by individuals having sustained brain injuries.















LIST OF REFERENCES


Adams, J. H., Graham, D. I., Murray, L. S., & Scott, G. (1982). Diffuse axonal injury due
to nonmissile head injury in humans: an analysis of 45 cases. Annals of Neurology,
12(6), 557-563.

Alexander, M. P. (1995). Mild traumatic brain injury: pathophysiology, natural history,
and clinical management. Neurology, 45(7), 1253-1260.

Batchelor, J., Harvey, A. G., & Bryant, R. A. (1995). Stroop colour word test as a
measure of attentional deficit following mild head injury. The Clinical
Neuropsychologist, 9(2), 180-186.

Bate, A. J., Mathias, J. L., & Crawford, J. R. (2001). Performance on the Test of
Everyday Attention and standard tests of attention following severe traumatic brain
injury. Clin Neuropsychol, 15(3), 405-422.

Bergner, M., Bobbitt, R. A., Pollard, W. E., Martin, D. P., & Gilson, B. S. (1976). The
sickness impact profile: validation of a health status measure. Medical Care, 14(1),
57-67.

Binder, L. M. (1997). A review of mild head trauma. Part II: Clinical implications.
Journal of Clinical and Experimental Neuropsychology, 19(3), 432-457.

Binder, L. M., Rohling, M. L., & Larrabee, J. (1997). A review of mild head trauma. Part
I: Meta-analytic review of neuropsychological studies. Journal of Clinical and
Experimental Neuropsychology, 19(3), 421-431.

Bohnen, N., Jolles, J., & Twijnstra, A. (1992). Modification of the Stroop colour word
test improves differentiation between patients with mild head injury and matched
controls. The Clinical Neuropsychologist, 6(2), 178-184.

Bohnen, N., Twijnstra, A., & Jolles, J. (1992). Performance in the Stroop color word test
in relationship to the persistence of symptoms following mild head injury. Acta
Neurol Scand, 85(2), 116-121.

Braver, T. S., & Barch, D. M. (2002). A theory of cognitive control, aging cognition, and
neuromodulation. Neuroscience and Biobehavioral Reviews, 26(7), 809-817.









Braver, T. S., Barch, D. M., Keys, B. A., Carter, C. S., Cohen, J. D., Kaye, J. A.,
Janowsky, J. S., Taylor, S. F., Yesavage, J. A., Mumenthaler, M. S., Jagust, W. J.,
& Reed, B. R. (2001). Context processing in older adults: evidence for a theory
relating cognitive control to neurobiology in healthy aging. Journal of Experimental
Psychology: General, 130(4), 746-763.

Carter, C. S., MacDonald, A. M., Botvinick, M., Ross, L. L., Stenger, V. A., Noll, D., &
Cohen, J. D. (2000). Parsing executive processes: strategic vs. evaluative functions
of the anterior cingulate cortex. Proceedings of the National Academy of Sciences:
U S A, 97(4), 1944-1948.

Chapman, L. J., & Chapman, J. P. (1988). Artifactual and genuine relationships of lateral
difference scores to overall accuracy in studies of laterality. Psychological Bulletin,
104(1), 127-136.

Chapman, L. J., & Chapman, J. P. (1989). Strategies for resolving the heterogeneity of
schizophrenics and their relatives using cognitive measures. Journal of Abnormal
Psychology, 98(4), 357-366.

Chapman, L. J., Chapman, J. P., Curran, T. E., & Miller, M. B. (1994). Do children and
the elderly show heightened semantic priming? How to answer the question.
Developmental Review, 14, 159-185.

Christodoulou, C., DeLuca, J., Ricker, J. H., Madigan, N. K., Bly, B. M., Lange, G.,
Kalnin, A. J., Liu, W. C., Steffener, J., Diamond, B. J., & Ni, A. C. (2001).
Functional magnetic resonance imaging of working memory impairment after
traumatic brain injury. Journal of Neurology, Neurosurgery, and Psychiatry, 71(2),
161-168.

Cohen, J. D., Barch, D. M., Carter, C., & Servan-Schreiber, D. (1999).
Context-processing deficits in schizophrenia: converging evidence from three
theoretically motivated cognitive tasks. Journal of Abnormal Psychology, 108(1),
120-133.

Cohen, J. D., Braver, T. S., & O'Reilly, R. C. (1996). A computational approach to
prefrontal cortex, cognitive control and schizophrenia: recent developments and
current challenges. Philosophical Transactions of the Royal Society of London
Series B, 351(1346), 1515-1527.

Cohen, J. D., MacWhinney, B., Flatt, M. R., & Provost, J. (1993). PsyScope: An
interactive graphic system for designing and controlling experiments in the
psychology laboratory using Macintosh computers. Behavioral Research Methods:
Instruments and Computers, 25, 257-271.

Cohen, J. D., & Servan-Schreiber, D. (1992). Context, cortex, and dopamine: a
connectionist approach to behavior and biology in schizophrenia. Psychological
Review, 99(1), 45-77.









Cohen, J. D., & Servan-Schreiber, D. (1993). A theory of dopamine function and its role
in cognitive deficits in schizophrenia. Schizophrenia Bulletin, 19(1), 85-104.

Crevits, L., Hanse, M. C., Tummers, P., & Van Maele, G. (2000). Antisaccades and
remembered saccades in mild traumatic brain injury. Journal of Neurology, 247(3),
179-182.

Cripe, L. I. (1987). The neuropsychological assessment and management of closed head
injury: General guidelines. Cognitive Rehabilitation, 5(1), 18-22.

Golden, C. J. (1975). The measurement of creativity by the Stroop Color and Word Test.
Journal of Personality Assessment, 39(5), 502-506.

Golden, C. J. (1976). Identification of brain disorders by the Stroop Color and Word Test.
Journal of Clinical Psychology, 32(3), 654-658.

Golden, C. J. (1978). Stroop Color and Word Test, Stoelting. Chicago.

Groswasser, Z., Reider-Groswasser, I., Soroker, N., & Machtey, Y. (1987). Magnetic
resonance imaging in head injured patients with normal late computed tomography
scans. Surgical Neurology, 27(4), 331-337.

Groth-Marnat, G. (2000). Neuropsychological assessment in clinical practice : a guide to
test interpretation and integration. New York: Wiley.

Guskiewicz, K. M., Riemann, B. L., Perrin, D. H., & Nashner, L. M. (1997). Alternative
approaches to the assessment of mild head injury in athletes. Med Sci Sports Exerc,
29(7 Suppl), S213-221.

Hair, J. F., Anderson, R. E., & Black, W. C. (1995). Multivariate Data Analysis (4th ed.).
Englewood Cliffs, NJ: Prentice Hall.

Jahanshahi, M., Profice, P., Brown, R. G., Ridding, M. C., Dirnberger, G., & Rothwell, J.
C. (1998). The effects of transcranial magnetic stimulation over the dorsolateral
prefrontal cortex on suppression of habitual counting during random number
generation. Brain, 121 (Pt 8), 1533-1544.

Johnstone, B., Callahan, C. D., Kapila, C. J., & Bouman, D. E. (1996). The comparability
of the WRAT-R reading test and NAART as estimates of premorbid intelligence in
neurologically impaired patients. Archives of Clinical Neuropsychology, 11(6),
513-519.

Kiefer, M., Marzinzik, F., Weisbrod, M., Scherg, M., & Spitzer, M. (1998). The time
course of brain activations during response inhibition: evidence from event-related
potentials in a go/no go task. Neuroreport, 9(4), 765-770.









Kimberg, D. Y., & Farah, M. J. (1993). A unified account of cognitive impairments
following frontal lobe damage: the role of working memory in complex, organized
behavior. Journal of Experimental Psychology: General, 122(4), 411-428.

Kingma, A., La Heij, W., Fasotti, L., & Eling, P. (1996). Stroop interference and
disorders of selective attention. Neuropsychologia, 34(4), 273-281.

Kraus, M. F., & Maki, P. M. (1997). Effect of amantadine hydrochloride on symptoms of
frontal lobe dysfunction in brain injury: case studies and review. Journal of
Neuropsychiatry and Clinical Neurosciences, 9(2), 222-230.

Levin, H., & Kraus, M. F. (1994). The frontal lobes and traumatic brain injury. Journal of
Neuropsychiatry and Clinical Neurosciences, 6(4), 443-454.

Levin, H. S., Eisenberg, H. M., & Benton, A. L. (1991). Mild Head Injury. New York:
Oxford University Press.

Levin, H. S., High, W. M., Goethe, K. E., Sisson, R. A., Overall, J. E., Rhoades, H. M.,
Eisenberg, H. M., Kalisky, Z., & Gary, H. E. (1987). The neurobehavioural rating
scale: assessment of the behavioral sequelae of head injury by the clinician.
Journal of Neurology, Neurosurgery, and Psychiatry, 50(2), 183-193.

Levine, B., Dawson, D., Boutet, I., Schwartz, M. L., & Stuss, D. T. (2000). Assessment
of strategic self-regulation in traumatic brain injury: its relationship to injury
severity and psychosocial outcome. Neuropsychology, 14(4), 491-500.

Levine, M. J. (1988). Issues in neurobehavioural assessment of mild head injuries.
Cognitive Rehabilitation, 6, 14-20.

Lezak, M. D. (1995). Neuropsychological assessment (3rd ed.). New York: Oxford
University Press.

Lishman, W. A. (1988). Physiogenesis and psychogenesis in the 'post-concussional
syndrome'. British Journal of Psychiatry, 153, 460-469.

MacDonald, A. W., 3rd, Cohen, J. D., Stenger, V. A., & Carter, C. S. (2000).
Dissociating the role of the dorsolateral prefrontal and anterior cingulate cortex in
cognitive control. Science, 288(5472), 1835-1838.

MacLeod, C. M. (1991). Half a century of research on the Stroop effect: an integrative
review. Psychological Bulletin, 109(2), 163-203.

Mathias, J. L., & Coats, J. L. (1999). Emotional and cognitive sequelae to mild traumatic
brain injury. Journal of Clinical and Experimental Neuropsychology, 21(2), 200-
215.









McAllister, T. W., Saykin, A. J., Flashman, L. A., Sparling, M. B., Johnson, S. C.,
Guerin, S. J., Mamourian, A. C., Weaver, J. B., & Yanofsky, N. (1999). Brain
activation during working memory 1 month after mild traumatic brain injury: a
functional MRI study. Neurology, 53(6), 1300-1308.

McCauley, S. R., Levin, H. S., Vanier, M., Mazaux, J. M., Boake, C., Goldfader, P. R.,
Rockers, D., Butters, M., Kareken, D. A., Lambert, J., & Clifton, G. L. (2001). The
neurobehavioural rating scale-revised: sensitivity and validity in closed head injury
assessment. Journal of Neurolology, Neurosurgery, and Psychiatry, 71(5), 643-651.

McDowell, S., Whyte, J., & D'Esposito, M. (1997). Working memory impairments in
traumatic brain injury: evidence from a dual-task paradigm. Neuropsychologia,
35(10), 1341-1353.

McDowell, S., Whyte, J., & D'Esposito, M. (1998). Differential effect of a dopaminergic
agonist on prefrontal function in traumatic brain injury patients. Brain, 121 (Pt 6),
1155-1164.

Miyake, A., Emerson, M. J., & Friedman, N. P. (2000). Assessment of executive
functions in clinical settings: problems and recommendations. Seminars in Speech
and Language, 21(2), 169-183.

Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T.
D. (2000). The unity and diversity of executive functions and their contributions to
complex "Frontal Lobe" tasks: a latent variable analysis. Cognitive Psychology,
41(1), 49-100.

Neter, J., Wasserman, W., & Kutner, M. H. (1985). Applied linear statistical models:
regression, analysis of variance, and experimental designs (2nd ed.). Homewood,
Ill.: R.D. Irwin.

Olver, J. H., Ponsford, J. L., & Curran, C. A. (1996). Outcome following traumatic brain
injury: a comparison between 2 and 5 years after injury. Brain Injury, 10(11), 841-
848.

Perlstein, W. M., Carter, C. S., Barch, D. M., & Baird, J. W. (1998). The Stroop task and
attention deficits in schizophrenia: a critical evaluation of card and single-trial
Stroop methodologies. Neuropsychology, 12(3), 414-425.

Pollard, W. E., Bobbitt, R. A., Bergner, M., Martin, D. P., & Gilson, B. S. (1976). The
Sickness Impact Profile: reliability of a health status measure. Medical Care, 14(2),
146-155.

Ponsford, J., & Kinsella, G. (1992). Attentional deficits following closed-head injury. J
Clin Exp Neuropsychol, 14(5), 822-838.









Potter, D. D., Jory, S. H., Bassett, M. R., Barrett, K., & Mychalkiw, W. (2002). Effect of
mild head injury on event-related potential correlates of Stroop task performance.
Journal of the International Neuropsychological Society, 8(6), 828-837.

Rao, V., & Lyketsos, C. (2000). Neuropsychiatric sequelae of traumatic brain injury.
Psychosomatics, 41(2), 95-103.

Ratcliff, R. (1993). Methods for dealing with reaction time outliers. Psychological
Bulletin, 114(3), 510-532.

Rieger, M., & Gauggel, S. (2002). Inhibition of ongoing responses in patients with
traumatic brain injury. Neuropsychologia, 40(1), 76-85.

Roberts, R. J., Hager, L. D., & Heron, C. (1994). Prefrontal cognitive processes: Working
memory and inhibition in the Antisaccade task. Journal of Experimental
Psychology: General, 123, 374-393.

Rojas, D. C., & Bennett, T. L. (1995). Single versus composite score discriminative
validity with the Halstead-Reitan battery and the Stroop test in mild brain injury.
Archives of Clinical Neuropsychology, 10, 101-110.

Salmon, D. P., Heindel, W. C., & Hamilton, J. M. (2001). Cognitive abilities mediated by
frontal-subcortical circuits. In J. L. Cummings (Ed.), Frontal-subcortical circuits in
psychiatric and neurological disorders (pp. 114-150). New York: The Guilford
Press.

Salo, R., Avishai, H., & Lynn, C. R. (2001). Interpreting Stroop interference: An analysis
of differences between task versions. Neuropsychology, 15, 462-471.

Spreen, O., & Strauss, E. (1991). A compendium of neuropsychological tests :
administration, norms, and commentary. New York: Oxford University Press.

Stuss, D. T., Ely, P., Hugenholtz, H., Richard, M. T., LaRochelle, S., Poirier, C. A., &
Bell, I. (1985). Subtle neuropsychological deficits in patients with good recovery
after closed head injury. Neurosurgery, 17(1), 41-47.

Stuss, D. T., Floden, D., Alexander, M. P., Levine, B., & Katz, D. (2001). Stroop
performance in focal lesion patients: dissociation of processes and frontal lobe
lesion location. Neuropsychologia, 39(8), 771-786.

Sunderland, A., Harris, J. E., & Baddeley, A. D. (1983). Do laboratory tests predict
everyday memory? A neuropsychological study. Journal of Verbal Learning and
Verbal Behavior, 22(3), 341-357.

Trennery, M. R., Crosson, B., DeBoe, J., & Leber, W. R. (1989). Stroop
Neuropsychological Screening Test Manual. Odessa, FL: Psychological
Assessment Resources.






50


Vanier, M., Mazaux, J. M., Lambert, J., Dassa, C., & Levin, H. S. (2000). Assessment of
neuropsychologic impairments after head injury: interrater reliability and factorial
and criterion validity of the Neurobehavioral Rating Scale-Revised. Archives of
Physical Medecine and Rehabilitation, 81(6), 796-806.

van Zomeren, A. H., & van den Burg, W. (1985). Residual complaints of patients two
years after severe head injury. Journal of Neurology, Neurosurgery, and Psychiatry,
48(1), 21-28.

WAIS-III and WMS-III Technical Manual. (1997). San Antonio, TX: The Psychological
Corporation.















BIOGRAPHICAL SKETCH

Paul J. Seignourel studied mathematics in France and received a Ph.D. in

probability from the University Paris VI and Ecole Polytechnique in 1999. He then

moved to the United States and, after a year of post-baccalaureate studies in psychology,

entered the Department of Clinical and Health Psychology at the University of Florida.