The topic of this thesis is cognitive control: how the brain organises itself to perform the many tasks it is capable of and how it switches flexibly among them. Task-switching experiments reveal a substantial cost in reaction time and accuracy after a switch in tasks. This "switch cost" is reduced by preparation (suggesting anticipatory task-set reconfiguration), but not eliminated. The thesis focuses on the sources of the "residual" cost. Most accounts attribute it to response selection being prolonged on a task-switch trial by task conflict, e.g. by 'task-set inertia' — persisting activation/inhibition of the previous task's S-R rules — or their associative reactivation by the stimulus. Four experiments used event-related potentials (ERPs) to determine which stages of task processing are influenced by a change in tasks, looking for delays in process-specific markers in the ERP. Experiments 1 and 2 showed that a prepared switch to a reading task from a perceptual judgement delayed early ERP markers of lexical access by a large fraction of the RT switch cost, suggesting that a substantial part of the residual cost arises in processes earlier than response selection, possibly due to task-related attentional inertia. Markers of lexical access observed in the non-lexical task were larger on switch than repeat trials, providing the first electrophysiological evidence of task-set inertia. Experiment 3 examined the effects of an unprepared switch in the same way. ERP waveforms were modulated by a switch before markers of lexical access were evident, suggesting additional processing demands compete for resources with lexical access. A simple delay, however, was not found; post-stimulus task-set reconfiguration does not just insert an extra processing stage. Experiment 4 looked for a delay in the onset of an early ERP marker of emotional processing when the task switched between categorising facial expression and classifying a superimposed letter. No such delay was found in this case, and ERP markers of emotion processing were present to the same extent in the letter task. This suggests that, given appropriate spatial attention, processing facial emotion unfolds automatically, independent of attention allocation to the facial features. Experiments 5-7 further explored the link between conflict due to processing the irrelevant stimulus dimension and the ERP post-stimulus negativity that accompanies the residual cost. The negativity could be elicited even on trials of non-switching blocks by prior training on classifying the irrelevant attribute of the stimulus using the same responses. But this effect did not seem to result from the trained class of irrelevant attribute attracting more attention. Finally, Experiment 8 followed up an incidental observation in Experiment 1 to establish the novel observation that a task-switching context abolishes the usual ERP correlate of withholding a response in a go/no-go paradigm, suggesting an interesting interaction between task-set control and response inhibition.

This study served to further investigate learning and memory, and to offer a potential tool to support educational interventions. More specifically, this was accomplished by an investigation of the physiologic changes in the brain that occurred while students learned medical anatomy. A group of 29 students taking the Gross Anatomy course at Boston University School of Medicine participated in the study. Testing occurred in two sessions: prior to the course and at the completion of the course. For each session, scalp EEG was recorded while participants were shown 176 anatomical terms (132 relevant to the course and 44 obscure) and asked to respond with "Can Define", "Familiar", or "Don't Know". Behavioral results indicated a positive correlation between participants' course grades and performance on the experimental tasks. EEG results were analyzed for event-related potential (ERP) components related to two memory components: familiarity and recollection. Results had a number of indications. For Don't Know responses, a stronger early frontal, late parietal, and late frontal effect occurred more so for terms of Session 1 compared to Session 2. For an analysis of just Session 2 data, results indicated increased activity of the early frontal, late parietal, and late frontal effects for Can Define responses only. Session 2 Can Define responses elicited a stronger early frontal ERP, occurring between 300 and 500 milliseconds yet, the most post-retrieval processing and monitoring appeared for Can Define terms of Session 2. Ultimately, we focused on investigating two points: 1) the effect of classroom learning on memory, and 2) the examination of ERPs as a tool to guide education interventions. Specifically, ERPs would potentially indicate markers to predict whether students would retain materials long before behavioral measures indicate these results. This has potential to determine whether long-lasting or transient learning will occur; as well as the potential to support early intervention strategies for not just students, but also individuals with learning disabilities or memory impairments.

What are the effects of non-balanced bilingualism on cognitive performance? And how do proficient, non-native speakers acquire and use lexical, syntactic and semantic information during sentence processing? Whilst there is growing research on these topics, there is no firm consensus on how to answer these questions. In the literature on cognitive effects of bilingualism, this lack of consensus has even resulted in radically opposing views and a heated debate. In this thesis, I seek to provide a balanced treatment of the literature and to address the above-mentioned questions by employing behavioral and neurophysiological paradigms. First, using a structural priming paradigm, I examine how proficient, non-native speakers of different native language backgrounds (Romance and Germanic) acquire lexically-specific syntactic restrictions of non-alternating verbs in English. Results from these experiments suggest that, although non-native speakers partially acquire lexically-specific syntactic restrictions, their knowledge is not native-like. Moreover, transfer from the first language does not seem to play a role in the acquisition of the relevant restrictions. Second, using Event-Related Potentials (ERPs) I examine whether proficient non-native Spanish-English speakers draw on different forms of semantic information such as relatedness and animacy incrementally during sentence comprehension. Results of these experiments suggest that, while relatedness facilitates processing (indexed by N400s) in both native and non-native speakers, effects of animacy are smaller in non-native speakers, relative to native speakers. Third, I employ a series of auditory attentional tasks and measures of lexical access and verbal fluency to assess cognitive functions in non-balanced bilinguals with different levels of language proficiency. Results show a bilingual advantage in inhibitory control and a non-significant trend towards bilingual better performance in attentional switching, and the groups exhibit similar performance on verbal fluency. Results of all the studies are discussed in the context of the existing literature on cognitive performance in bilinguals and accounts of language processing in native and non-native speakers and suggestions for future research are provided.

This project analyzed gender and color-specific differences in event-related potentials (ERPs). Previous studies have shown that males process color differently than females. In a recent study, sex differences were found in ERPs during a visual object recognition task. There were higher EEG amplitudes in females (especially P300) than males. Significant sex and color-specific differences have been found in diseases involving altered dopamine (DA) machinery. Thus, we analyzed differences between ERPs in males vs females during a color task. We also compared the color-specific differences in ERPs between males and females. Males and females participated in EEG recording sessions for 2 color studies during a color-go-no-go task, where two studies examined the gender and color-specific differences in ERPs, respectively. Data from 32 males and 24 females and 21 females and 31 males, respectively, in two color studies demonstrated significant sex-specific differences in ERPs during a color-go-no-go task. Males consistently showed higher EEG amplitudes (particularly P300) than females, which is contradictory to what we demonstrated previously in the object recognition task, indicating different color processing systems in males and females. Regarding color-specific differences, no significant differences were found in P300s between the three colors red, green and blue in males and females when each color was the relevant stimulus, suggesting that color is not a marker for inducing ERPs in normal subjects. These studies will provide the impetus to compare patients having altered DA mechanisms such as in attention-deficit hyperactivity disorder (ADHD), Parkinson's, or chemical addiction.

The aim of the current series of experiments was to further explore the boundary conditions of the recognition memory old/new effect in the context of the recognition/associative recall task (Rugg, Schloerscheidt, Doyle, Cox, & Patching, 1996). The study by Rugg et al. was replicated and extended by manipulating both the semantic relatedness between study items and the timing of recall. Eventrelated potentials (ERPs) were recorded from 17 scalp electrode sites during performance of a recognition/associative recall task. Forty participants were visually presented with four blocks of 50 word pairs which were either unrelated (Experiments 1 and 2) or weakly semantically related (Experiments 3 and 4). Participants were instructed to form an association between the members of each word pair. At test, the first members of each pair were visually presented intermixed with a similar number of unstudied items. Participants were required to discriminate (i.e., recognise) previously studied items (old) from new items. Participants were also required to recall the study associate for words judged old, and to provide confidence levels for each recognition decision on a 3-point scale. Recall was either immediate (Experiments 1 and 3) or delayed (Experiments 2 and 4). Relative to ERPs to new items, the ERPs elicited by words correctly recognised and for which the associate was correctly recalled exhibited a positive-going shift between 500-800 ms poststimulus onset. The effect was maximal at posterior temporal-parietal electrode sites (the parietal old/new effect). Although the effect was not lateralised to the left hemisphere, this result may be due to the variability in encoding strategies employed by the participants. Behavioural data consistently indicated that response confidence is confounded with response category. The ERP results also revealed that the old/new effect is not evident following the experimental control of response confidence, and that immediate recall is associated with a negative-going shift at posterior electrode sites between 800-1100 ms poststimulus onset. Manipulating the semantic relatedness between the word pairs did not influence the distribution of the old/new effect. The results are discussed in terms of the view that the parietal old/new effect reflects neural activity associated with the recollection of specific previous experiences, and may reflect retrieval processes supported by the medial temporal lobe memory system (Moscovitch, 1992, 1994; Squire, 1992; Squire, Knowlton, & Musen, 1993). It is suggested that future research extend the current findings by examining the influence of response confidence in alternative recognition memory paradigms.

Since its first use in human in 1929, the electroencephalogram (EEG) has become one of the most important diagnostic tool in clinical neurophysiology. However, their use in clinical studies is limited because the huge quantity of collected information is complicated to treat. Indeed, it is very difficult to have an overall picture of this multivariate problem. In addition to the impressive quantity of data to be treated, an intrinsic problem with electroencephalograms is that the signals are "contaminated" by body signals not directly related to cerebral activity. However, these signals do not interest us directly to evaluate treatment effect on the brain. Removing these signals known as "parasitic noise" from electroencephalograms is a difficult task. We use clinical data kindly made available by the pharmaceutical company Eli Lilly (Lilly Clinical Operations S.A., Louvain-la-Neuve, Belgium). Particular types of analyses were already carried out on these data, most based on frequency bands. They mainly confirmed the enormous potential of EEG in clinical studies without much insight in the understanding of treatment effect on the brain. The aim of this thesis is to propose and evaluate a panel of statistical techniques to clean and to analyze electroencephalograms. The first presented tool enables to align curves such as selected parts of EEGs before any further statistical treatment. Indeed, when monitoring some continuous process on similar units (like patients in a clinical study), one often notices a typical pattern common to all curves but with variation both in amplitude and dynamics across curves. In particular, typical peaks could be shifted from unit to unit. This complicates the statistical analysis of sample of curves. For example, the cross-sectional average usually does not reflect a typical curve pattern: due to shifts, the signal structure is smeared or might even disappear. Another of the presented tools is based on the preliminary linear decomposition of EEGs into statistically independent signals. This decomposition provides on the one hand an effective cleaning method and on the other hand a considerable reduction of the quantity of data to be analyzed. The technique of decomposition of our signals in statistically independent signals is a well-known technique in physics primarily used to unmix sound signals. This technique is named Independent Component Analysis or ICA. The last studied tool is functional ANOVA. The analysis of longitudinal curve data is a methodological and computational challenge for statisticians. Such data are often generated in biomedical studies. Most of the time, the statistical analysis focuses on simple summary measures, thereby discarding potentially important information. We propose to model these curves using non parametric regression techniques based on splines.

Since its creation in 1998, the Implicit Association Test (IAT) has become a commonly used measure in social psychology and related fields of research. Studies of the cognitive processes involved in the IAT are necessary to establish the validity of this measure and to suggest further refinements to its use and interpretation. The current thesis used ERPs to study cognitive processes associated with the IAT. The first experiment found significant differences in P300 amplitude in the Congruent and Incongruent conditions, which were interpreted as a reflection of greater equivocation in the Incongruent condition. The second experiment tested the task-set switching account of the IAT in much greater detail by analyzing each trial type separately. In the Congruent condition, all trial types elicited the same amplitude P300. Local probability, and the consequent checking and updating of working memory, was thought to be responsible for differences between trials of the Incongruent condition that required or did not require a task switch. The final experiment examined the role of working memory in the IAT by introducing obtrusive and irrelevant auditory stimuli. The results of Experiment 3 indicated that the introduction of an obtrusive and irrelevant auditory increment deviant has little overall effect on the IAT, and a similar effect on switch and no-switch trials within the Incongruent condition. This could have been because both the Congruent and Incongruent conditions of the IAT make such extensive demands on central processing resources that few are available to allow for the switching of attention, or it is possible that the IAT does not require significant updating of working memory. The usefulness of ERPs in the study of the IAT effect is demonstrated by the current research. In particular, the finding that behavioural results were not always consistent with the ERP results demonstrates that electrophysiological measures can complement traditional behavioural measures.

Since its creation in 1998, the Implicit Association Test (IAT) has become a commonly used measure in social psychology and related fields of research. Studies of the cognitive processes involved in the IAT are necessary to establish the validity of this measure and to suggest further refinements to its use and interpretation. The current thesis used ERPs to study cognitive processes associated with the IAT. The first experiment found significant differences in P300 amplitude in the Congruent and Incongruent conditions, which were interpreted as a reflection of greater equivocation in the Incongruent condition. The second experiment tested the task-set switching account of the IAT in much greater detail by analyzing each trial type separately. In the Congruent condition, all trial types elicited the same amplitude P300. Local probability, and the consequent checking and updating of working memory, was thought to be responsible for differences between trials of the Incongruent condition that required or did not require a task switch. The final experiment examined the role of working memory in the IAT by introducing obtrusive and irrelevant auditory stimuli. The results of Experiment 3 indicated that the introduction of an obtrusive and irrelevant auditory increment deviant has little overall effect on the IAT, and a similar effect on switch and no-switch trials within the Incongruent condition. This could have been because both the Congruent and Incongruent conditions of the IAT make such extensive demands on central processing resources that few are available to allow for the switching of attention, or it is possible that the IAT does not require significant updating of working memory. The usefulness of ERPs in the study of the IAT effect is demonstrated by the current research. In particular, the finding that behavioural results were not always consistent with the ERP results demonstrates that electrophysiological measures can complement traditional behavioural measures.

In this dissertation, the effect of mild sleep deprivation on attentional allocation and performance monitoring was investigated using a variety of event-related potential (ERP) paradigms with ecologically realistic periods of sleep deprivation. Seventeen female young adults completed several tasks under alert and sleepy conditions, after 3 and 20 hours of wakefulness, respectively. Objective behavioural measures of response times and error rates indicated virtually no decrements that could be attributed exclusively to sleepiness; however, there were consistent alterations in the ERPs indicative of subtly reduced attentional resources and performance monitoring. The first study (Chapter 2) examined the effect of distraction on the P300, an ERP component related to attention and stimulus processing. Participants performed an auditory oddball task with and without a secondary visual working memory task. Response times (RTs) and P300 amplitudes were affected by the addition of the secondary working memory task. However, an interaction showed that the P300 latency was significantly increased by the secondary task only in the sleepy condition, indicating that processing speed is impaired by a secondary task only when the participant is sleepy. The next study (Chapter 3) used a Go/NoGo contingent negative variation (CNV) task. The CNV is reflective of sustained attention, and is known to be associated with frontal lobe functioning. This task was performed twice, with and without a financial incentive for fast responses, to assess the effect of motivation. The P300 amplitude to the first stimulus and CNV prior to the second were clearly larger to Go stimuli for both levels of alertness when the participant was motivated by the financial incentive. However, with no incentive in the sleepy condition, there was reduced differentiation of the two types of stimuli, indicating a reduced ability to discriminate between important and less important information. In chapters 4 and 5, performance monitoring was examined using two tasks, the Eriksen Flanker task and the Anti-Saccade task, producing an ERP related to errors with two basic components: the error-negativity (Ne/ERN) and error-positivity (Pe), thought to be related to error recognition and error evaluation, respectively. In both data sets, the amplitude of the Ne/ERN was not significantly reduced by sleep deprivation, but the amplitude of the Pe was. In addition, smaller anti-saccade errors produced reduced Ne/ERN amplitudes compared to larger anti-saccade errors. Another marker of performance monitoring is post-error slowing, which was present in the flanker task only during the alert condition. These results indicate that error detection or recognition (Ne/ERN) appears to be relatively preserved during sleep deprivation, but further error evaluation (Pe) and compensation (post-error slowing) are impaired. Taken together, the findings demonstrate that even mild sleep deprivation has a subtle but reliable effect on electrocortical activity associated with attention and performance monitoring despite an absence of behavioural changes, indicating deleterious effects before behavioural changes are observed. Therefore, relying on behavioural tests to determine at what point an individual becomes unsafe to operate machinery or perform various tasks may be misleading.

In this dissertation, the effect of mild sleep deprivation on attentional allocation and performance monitoring was investigated using a variety of event-related potential (ERP) paradigms with ecologically realistic periods of sleep deprivation. Seventeen female young adults completed several tasks under alert and sleepy conditions, after 3 and 20 hours of wakefulness, respectively. Objective behavioural measures of response times and error rates indicated virtually no decrements that could be attributed exclusively to sleepiness; however, there were consistent alterations in the ERPs indicative of subtly reduced attentional resources and performance monitoring. The first study (Chapter 2) examined the effect of distraction on the P300, an ERP component related to attention and stimulus processing. Participants performed an auditory oddball task with and without a secondary visual working memory task. Response times (RTs) and P300 amplitudes were affected by the addition of the secondary working memory task. However, an interaction showed that the P300 latency was significantly increased by the secondary task only in the sleepy condition, indicating that processing speed is impaired by a secondary task only when the participant is sleepy. The next study (Chapter 3) used a Go/NoGo contingent negative variation (CNV) task. The CNV is reflective of sustained attention, and is known to be associated with frontal lobe functioning. This task was performed twice, with and without a financial incentive for fast responses, to assess the effect of motivation. The P300 amplitude to the first stimulus and CNV prior to the second were clearly larger to Go stimuli for both levels of alertness when the participant was motivated by the financial incentive. However, with no incentive in the sleepy condition, there was reduced differentiation of the two types of stimuli, indicating a reduced ability to discriminate between important and less important information. In chapters 4 and 5, performance monitoring was examined using two tasks, the Eriksen Flanker task and the Anti-Saccade task, producing an ERP related to errors with two basic components: the error-negativity (Ne/ERN) and error-positivity (Pe), thought to be related to error recognition and error evaluation, respectively. In both data sets, the amplitude of the Ne/ERN was not significantly reduced by sleep deprivation, but the amplitude of the Pe was. In addition, smaller anti-saccade errors produced reduced Ne/ERN amplitudes compared to larger anti-saccade errors. Another marker of performance monitoring is post-error slowing, which was present in the flanker task only during the alert condition. These results indicate that error detection or recognition (Ne/ERN) appears to be relatively preserved during sleep deprivation, but further error evaluation (Pe) and compensation (post-error slowing) are impaired. Taken together, the findings demonstrate that even mild sleep deprivation has a subtle but reliable effect on electrocortical activity associated with attention and performance monitoring despite an absence of behavioural changes, indicating deleterious effects before behavioural changes are observed. Therefore, relying on behavioural tests to determine at what point an individual becomes unsafe to operate machinery or perform various tasks may be misleading.