<p>Deep learning has been increasingly adopted in a multitude of settings. Yet, its strong performance relies on processing data during inference that is in-distribution with its training data. Deep learning input data during deployment, however, is not guaranteed to be in-distribution with the model's training data and can often times be distorted, either intentionally (e.g., by an adversary) or unintentionally (e.g., by a sensor defect), leading to significant performance degradations. In this dissertation, we develop algorithms for a variety of applications to improve the performance of deep learning models in the presence of distorted data. We begin by first designing feature engineering methodologies to increase classification performance in noisy environments. Here, we demonstrate the efficacy of our proposed algorithms on two target detection tasks and show that our framework outperforms a variety of state-of-the-art baselines. Next, we develop mitigation algorithms to improve the performance of deep learning in the presence of adversarial attacks and nonlinear signal distortions. In this context, we demonstrate the effectiveness of our methods on a variety of wireless communications tasks including automatic modulation classification, power allocation in massive MIMO networks, and signal detection. Finally, we develop an uncertainty quantification framework, which produces distributive estimates, as opposed to point predictions, from deep learning models in order to characterize samples with uncertain predictions as well as samples that are out-of-distribution from the model's training data. Our uncertainty quantification framework is carried out on a hyperspectral image target detection task as well as on counter unmanned aircraft systems (cUAS) model. Ultimately, our proposed algorithms improve the performance of deep learning in several environments in which the data during inference has been distorted to be out-of-distribution from the training data. </p>

Understanding what constitutes typical adult processing of infant cues and the potential individual variations associated with it is an important early step in comprehending how caregiving practices arise. By adopting a multi-method approach, the present work aimed to enrich the knowledge on adults’ cognitive and electrophysiological responses to infant cues, by focusing on the role of the adults’ sex, the perceived quality of early care from caregivers, and the actual involvement in childcare. Four studies were presented to cover different perspectives on adults’ responses to infant cues, from cognition to electrophysiology and parental behaviors. Methodologies encompassed cognitive experimental tasks, EEG, standardized observations, and self-reported measures. Samples included non-parents, different-sex parents, and same-sex parents. Overall, the present work responded to the recent interest in understanding adults’ responses to infant cues going beyond a heteronormative perspective. Our findings reinforced the argument that differences between males and females, if present in the response to infant stimuli, might not always be biologically determined, but reinforced by societal norms. In addition, we supported the relevance of accounting for both past and current experiences of care when investigating the adults’ cognitive and electrophysiological responses to infant cues. Eventually, we started to gain some knowledge on the neurophysiological correlates of sensitive caregiving in different family contexts. On the whole, considering the complexity of plural family models, we hopefully contributed to socializing the idea that there are different ways of conceiving and understanding parenting nowadays.

Heightened impulsivity and compulsivity are often found in association with both dysfunctional everyday behavior and with psychopathology. Impulsivity and compulsivity are also linked to alterations in behavioral response inhibition and its electrophysiological correlates. However, they are rarely examined jointly and their effect outside of clinical samples is still disputed. This study assesses the influence and interaction of impulsivity and compulsivity as measured by questionnaires (Barratt Impulsiveness Scale, UPPS Impulsive Behavior Scale, and Obsessive–Compulsive Inventory-Revised) on behavioral performance and event-related potentials (N2, P3a, and P3b) in a visual Go/Nogo task. Data from 250 participants from the general population (49% female; age M = 25.16, SD = 5.07) were collected. We used robust linear regression as well as regression tree analyses, a type of machine learning algorithm, to uncover potential non-linear effects. We did not find any significant relationship between the self-report measures and behavioral or neural inhibition effects in either type of analysis, with the exception of a linear effect of the lack of premeditation subscale of the UPPS Impulsive Behavior Scale on behavioral performance. The current sample size was large enough to uncover even small effects. One possibility is that inhibitory performance was unimpaired in a non-clinical sample, suggesting that the effect of these personality traits on inhibition and cognitive control may require a clinical sample or a more difficult task version. Further studies are needed to uncover possible associations and interactions to delineate when impulsivity and compulsivity lead to dysfunctional everyday behavior and psychopathology.

L’attention visuo-spatiale peut être déployée à différentes localisations dans l’espace indépendamment de la direction du regard, et des études ont montré que les composantes des potentiels reliés aux évènements (PRE) peuvent être un index fiable pour déterminer si celle-ci est déployée dans le champ visuel droit ou gauche. Cependant, la littérature ne permet pas d’affirmer qu’il soit possible d’obtenir une localisation spatiale plus précise du faisceau attentionnel en se basant sur le signal EEG lors d’une fixation centrale. Dans cette étude, nous avons utilisé une tâche d’indiçage de Posner modifiée pour déterminer la précision avec laquelle l’information contenue dans le signal EEG peut nous permettre de suivre l’attention visuelle spatiale endogène lors de séquences de stimulation d’une durée de 200 ms. Nous avons utilisé une machine à vecteur de support (MVS) et une validation croisée pour évaluer la précision du décodage, soit le pourcentage de prédictions correctes sur la localisation spatiale connue de l’attention. Nous verrons que les attributs basés sur les PREs montrent une précision de décodage de la localisation du focus attentionnel significative (57%, p<0.001, niveau de chance à 25%). Les réponses PREs ont également prédit avec succès si l’attention était présente ou non à une localisation particulière, avec une précision de décodage de 79% (p<0.001). Ces résultats seront discutés en termes de leurs implications pour le décodage de l’attention visuelle spatiale, et des directions futures pour la recherche seront proposées. / Visuospatial attention can be deployed to different locations in space independently of ocular fixation, and studies have shown that event-related potential (ERP) components can effectively index whether such covert visuospatial attention is deployed to the left or right visual field. However, it is not clear whether we may obtain a more precise spatial localization of the focus of attention based on the EEG signals during central fixation. In this study, we used a modified Posner cueing task with an endogenous cue to determine the degree to which information in the EEG signal can be used to track visual spatial attention in presentation sequences lasting 200 ms. We used a machine learning classification method to evaluate how well EEG signals discriminate between four different locations of the focus of attention. We then used a multi-class support vector machine (SVM) and a leave-one-out cross-validation framework to evaluate the decoding accuracy (DA). We found that ERP-based features from occipital and parietal regions showed a statistically significant valid prediction of the location of the focus of visuospatial attention (DA = 57%, p < .001, chance-level 25%). The mean distance between the predicted and the true focus of attention was 0.62 letter positions, which represented a mean error of 0.55 degrees of visual angle. In addition, ERP responses also successfully predicted whether spatial attention was allocated or not to a given location with an accuracy of 79% (p < .001). These findings are discussed in terms of their implications for visuospatial attention decoding and future paths for research are proposed.

Zusammenhänge zwischen neurokognitiven Indikatoren und Verhaltensindikatoren der Gesichterkognition können Gehirnsysteme und neuronale Subprozesse identifizieren, die individuellen Unterschieden im Verhalten zugrunde liegen. Diese Dissertation zeigt, dass Ereigniskorrelierte Potentiale (EKPs) als neurokognitive Indikatoren für die Erforschung individueller Unterschiede eingesetzt werden können, denn sie weisen die gleichen hohen psychometrischen Qualitäten wie andere Fähigkeitsindikatoren auf und messen daher individuelle Unterschiede in der neuronalen Verarbeitung zuverlässig und stabil über die Zeit. Auf der Verhaltensebene wurden drei Teilfähigkeiten der Gesichterkognition etabliert: Gesichterwahrnehmung, Gesichtergedächtnis und Gesichtergeschwindigkeit. EKPs wurden in Strukturgleichungsmodellen verwendet, um den Beitrag neurokognitiver Indikatoren an individuellen Unterschieden dieser Gesichterkognitionsfähigkeiten zu schätzen. Für 85 Probanden wurden Beziehungen zwischen den Gesichterkognitionsfähigkeiten und der P100, N170, der sogenannten Differenz aufgrund des Gedächtnisses (Dm) und dem frühen sowie späten Wiederholungseffekt (ERE und LRE) etabliert. Spezifische Anteile individueller Unterschiede in der Gesichterkognition auf der Verhaltensebene wurden durch individuelle Unterschiede im Zeitverlauf der strukturellen Gesichteranalyse (N170 Latenz) sowie in der Reaktivierung von Repräsentationen gespeicherter Gesichtsstrukturen (ERE) als auch personen-spezifischen Wissens (LRE) erklärt. Keinen Anteil an individuellen Unterschieden erklärten hingegen frühe Wahrnehmungsprozesse (P100), die neuronale Aktivierung während der strukturellen Gesichteranalyse (N170 Amplitude) und Prozesse der Gedächtnisenkodierung von Gesichtern (Dm). Diese Ergebnisse zeigen, dass individuelle Unterschiede in der Gesichterkognition von der strukturellen Gesichteranalyse sowie von der Effizienz und Geschwindigkeit des Zugriffs auf Gedächtnisinhalte zu Gesichtern und Personen abhängt. / Individual differences in perceiving, learning, and recognizing faces were shown on the behavioral and neural level but were rarely related to one another. By determining relationships between behavioral and neurocognitive indicators of face cognition, brain systems and neural sub-processes can be identified that underlie individual variations on the behavioral level. The present dissertation laid the foundation for using event-related potentials (ERPs) as neurocognitive indicators in individual differences research. ERP components were shown to possess the same high psychometric qualities as behavioral ability measures and thus to measure individual differences of neural processing reliably and stably across time. On the behavioral level, three component abilities of face cognition were established: face perception, face memory, and the speed of face cognition. ERP components were used in structural equation models that estimated contributions of neurocognitive indicators to the individual differences in these face cognition abilities. Regression analysis was used to determine the contributions of P100, N170, the so called difference due to memory (Dm), as well as early and late repetition effects (ERE and LRE) to face cognition abilities in 85 participants. Certain amounts of variance in face cognition as seen on the behavioral level were accounted for by individual differences in the temporal dimension of structural encoding of a face (N170 latency) and in the re-activation of both stored facial structures (ERE) and person-identity information (LRE). In contrast, processes of early vision (P100), the neural activation of structural face encoding (N170 amplitude), and memory encoding of new faces (Dm) did not show any contribution to individual differences in face cognition. These findings show that individual differences in face cognition depend on the speed of structurally encoding faces and on the efficiency and speed of accessing face and person memory.

L'étude par neuro-imagerie de l'activité de plusieurs cerveaux en interaction (hyperscanning) permet d'étendre notre compréhension des neurosciences sociales. Nous proposons un cadre pour l'hyperscanning utilisant les interfaces cerveau-ordinateur multi-utilisateur qui inclut différents paradigmes sociaux tels que la coopération ou la compétition. Les travaux de cette thèse comportent trois contributions interdépendantes. Notre première contribution est le développement d'une plateforme expérimentale sous la forme d'un jeu vidéo multijoueur, nommé Brain Invaders 2, contrôlé par la classification de potentiels évoqués visuels enregistrés par électroencéphalographie (EEG). Cette plateforme est validée par deux protocoles expérimentaux comprenant dix-neuf et vingt-deux paires de sujets et utilise différentes approches de classification adaptative par géométrie riemannienne. Ces approches sont théoriquement et expérimentalement comparées et nous montrons la supériorité de la fusion des classifieurs indépendants sur la classification d'un hypercerveau durant la seconde contribution. L'analyse de coïncidence des signaux entre les individus est une approche classique pour l'hyperscanning, elle est pourtant difficile quand les signaux EEG concernés sont transitoires avec une grande variabilité (intra- et inter-sujet) spatio-temporelle et avec un faible rapport signal-à-bruit. En troisième contribution, nous proposons un nouveau modèle composite de séparation aveugle de sources physiologiquement plausibles permettant de compenser cette variabilité. Une solution par diagonalisation conjointe approchée est proposée avec une implémentation d'un algorithme de type Jacobi. A partir des données de Brain Invaders 2, nous montrons que cette solution permet d'extraire simultanément des sources d'artéfacts, des sources d'EEG évoquées et des sources d'EEG continues avec plus de robustesse et de précision que les modèles existants. / The study of several brains interacting (hyperscanning) with neuroimagery allows to extend our understanding of social neurosciences. We propose a framework for hyperscanning using multi-user Brain-Computer Interfaces (BCI) that includes several social paradigms such as cooperation or competition. This dissertation includes three interdependent contribution. The first contribution is the development of an experimental platform consisting of a multi-player video game, namely Brain Invaders 2, controlled by classification of visual event related potentials (ERP) recorded by electroencephalography (EEG). The plateform is validated through two experimental protocols including nineteen and twenty two pairs of subjects while using different adaptive classification approaches using Riemannian geometry. Those approaches are theoretically and experimentally compared during the second contribution ; we demonstrates the superiority in term of accuracy of merging independent classifications over the classification of the hyperbrain during the second contribution. Analysis of inter-brain synchronizations is a common approach for hyperscanning, however it is challenging for transient EEG waves with an great spatio-temporal variability (intra- and inter-subject) and with low signal-to-noise ratio such as ERP. Therefore, as third contribution, we propose a new blind source separation model, namely composite model, to extract simultaneously evoked EEG sources and ongoing EEG sources that allows to compensate this variability. A solution using approximate joint diagonalization is given and implemented with a fast Jacobi-like algorithm. We demonstrate on Brain Invaders 2 data that our solution extracts simultaneously evoked and ongoing EEG sources and performs better in term of accuracy and robustness compared to the existing models.

In this dissertation the effects of variations in stimulus properties and CTOA, in auditory attention tasks were explored using recently developed approaches to EEG analysis including LIMO. The last experiment was structured using information theory, designing an effective experiment. Four studies were carried out using a number parity decision task, that employed different combinations of cueing Tone (T), Novel (N) and the Goal (G) stimuli. In the first EEG study, contrary to previous findings (Polich 2002, 2007) in control participants, no correlation between the time of a novel condition to the next novel condition and P300 amplitude was found. Therefore single trial across-subject averaging of participants’ data revealed significant correlations (r > .3) of stimulus properties (such as probability, frequency, amplitude and duration) on P300, and even r > .5 was found when N was an environmental sound in schizophrenic patients. In the second EEG study, simultaneously with fMRI recordings, the participants that showed significant behavioural distraction evoked brain activations and differences in both hemispheres (similar to Corbetta, 2002, 2008) while the participants, as a whole, produced significant activations mainly in left cortical and subcortical regions. A context analysis was run in distracted participants contrasting the trials immediately prior to the G trials, resulting in different prefrontal activations, which was consistent with studies of prefrontal control of visual attention (Koechlin 2003, 2007). In the third EEG study, the distractor noise type was manipulated (white vs environmental sounds) as well as presence or absence of scanner background noise in a blocked design. Results showed consistent P300, MMN and RON due to environmental noise. In addition, using time constants found in MEG results (Lu, Williamson & Kaufman, 1992) and adding the CTOA to the analysis, an information theory framework was calculated. After the simulation of the information of the experiment, a saddle indentation in the curve of the information measure based on the states of the incoming signal at around 300 ms CTOA was found. This saddle indentation was evident in more than 60 novel trials. In the fourth study, the CTOA and stimulus properties were manipulated in a parametric experiment. Based on the three studies, reducing complexity if the task (first study), using more than 60 stimuli in the novel conditions (third study). The CTOA randomly varying between 250 ms or 500 ms. Thirty-eight ANCOVA with 2 categorical and 1 continuous regressors were conducted and determined which time and channels elicited reliably signatures (p <.05) in the whole participants at short CTOA. Results revealed differences for the waveforms of current condition by depending on which condition appeared previously as well in terms of frequency and duration in scalp frontal electrodes (such as the second study). These results were interpreted as a consequence of switching between modes of attention and alerting states which resulted in the activation of frontal areas. Moreover, contextual analyses showed that systematic manipulation of stimulus properties allowed the visualization of the relationships between CTOA, executive function and orienting of attention.

The purpose of this research was to explore the effects of small monetary or economic gains and/or losses on choice behavior through the use of a computerized game and to determine gain/loss ratio differences using both behavioral and electrophysiological measures. Participants (N=53) played the game in several 36 minute sessions. These sessions operated with concurrent variable-interval schedules for both rewards and penalties. Previously, asymmetrical effects of gains and losses have been identified through cognitive studies, primarily due to the work of nobel laureates Daniel Kahneman and Amos Tversky (1979). They found that the effect of a loss is twice (i.e., 2:1) that of a gain. Similar results have been observed in the behavioral laboratory as exemplified by the research of Rasmussen and Newland (2008), who found a 3:1 ratio for the effect of losses versus gains. The asymmetry of gains and losses was estimated behaviorally and through event-related brain potentials (ERPs) and the cognitive (Kahneman and Tversky) and behavioral (Rasmussen and Newland) discrepancy elucidated. In the game, the player moves an animated submarine around sea rocks to collect yellow coins and other treasures on the sea floor. Upon collecting a coin, one of three things can happen: The player triggers a penalty (loss), the player triggers a payoff (gain), or there is no change. The behavioral measures consisted in counting the number of clicks, reinforces, and punishers and then determining ratio differences between punished (loss) and no punished condition (gain) conditions. The obtained gain/loss ratio corresponded to an asymmetry of 2:1. Similarly ratio differences were found between male and female, virtual money and cash, risk averse versus risk seeking, and generosity versus profit behavior. Also, no ratio difference was found when players receive information about other player's performances in the game (players with information versus players without information). In electroencephalographic (EEG) studies, visual evoked potentials (VEPs) and ERPs components (e.g., P300) were examined. I found increased ERP amplitudes for the losses in relation to the gains that corresponded to the calculated behavioral asymmetry of 2:1. A correlational strategy was adopted that sought to identify neural correlates of choice consistent with cognitive and behavioral approaches. In addition, electro cortical ratio differences were observed between different sets of electrodes that corresponded to the front, middle, and back sections of the brain; differences between sessions, risk averse and risk seeking behavior and sessions with concurrent visual and auditory stimuli and only visual were also estimated.

人們對於錯誤訊息處理經由自我覺察或外在回饋之管道，可藉由事件關聯電位分別測得ERN (error-related negativity) 及FRN (feedback-related negativity)。過去研究曾指出雙側作業(Flanker task)中錯誤所引發的ERN會受到以圖片呈現的短期情緒所調節，然而對於回饋誘發的FRN與個體情緒調節的關係則未曾被探討過。過去FRN的研究認為唯有當受試者所進行的作業為增強學習作業時，受試者對於回饋結果的預期狀態才能反映於FRN的反應強度。本研究利用兩個實驗分別採用非增強學習作業及增強學習作業，其中並以IAPS情緒圖片進行短期情緒的引發，在受試者於實驗中對其反應結果的不同預期狀態，探測受試者FRN受短期情緒調節的效果。 實驗一利用非增強學習作業，結果顯示FRN的強度可以反映受試者對於回饋結果的預期狀態，其中以非預期時FRN的強度為最大，預期時FRN的強度為最小；另外，正向情緒圖片對於FRN具有調節效果，正向情緒下FRN反應強度小於中性以及負向情緒下FRN反應強度。實驗二利用增強學習作業，前述的FRN強度反映受試者對回饋結果的預期效果，只有在實驗前半段的嘗試次中被觀測到，此效果未見於全部嘗試次納入分析；另外，實驗二中沒有觀察到情緒對於FRN的調節效果。 綜合而言，本研究發現受試者唯有持續處於學習的情形下，FRN才能反映受試者對於回饋結果的預期狀態，情緒對FRN的調節效果也僅於此情況下才能被觀測到。 / Error-related information in human can be processed via self-awareness and/or feedback given externally, which are measurable by the use of event-related potential (ERP) and termed error-related negativity (ERN) and feedback-related negativity (FRN) respectively. Previous studies showed that short-term affective stimuli would modulate the magnitude of ERN elicited by Flanker task. However, such modulation effect has not been tested on FRN. Furthermore, the magnitude of FRN is indicated to be related to the expectancy states toward feedback when the subject is undergoing a reinforcement learning task. Present study, thus, was designed to test the affective modulation effect on FRN in two separate tasks. In which, emotional pictures adopted from IAPS were used as the short-term affective stimuli, and different expectancy states in both non-reinforcement learning task (Experiment1) and reinforcement learning task (Experiment 2) were manipulated. In the results of Experiment 1, the magnitude of FRN was larger under the unexpected condition in comparing to the expected one. Modulation effect of short-term affective stimuli on FRN was obtained when positive emotion pictures were presented in non-reinforcement learning task, which FRN amplitude was significantly smaller in comparing to those measured after the presentation of neutral and negative pictures. In the results of Experiment 2, FRN elicited in the unexpected condition was only obtained from analyzing the dada collected in the first half of trails. Such effect was not confirmed when the data from all trials were analyzed. A lack of modulation effect of short-term affective stimuli on FRN appeared in Experiment 2. In conclusion, it is indicated that the expectancy depended FRN is most apparent when the subject is undergoing a continuous learning-demanded process. Meanwhile, short-term affective stimuli can modulate such FRN.

研究目的 失眠的過度激發假說是目前最廣為接受的失眠病因之一，失眠者通常可觀察到具有身心過度激發的狀態，其中認知神經過度激發假說更由中樞神經的角度探討失眠者的過度激發現象，且已有許多證據支持此觀點。然而過度激發因素在失眠病程上的影響仍不明確，不論在前置因子或是持續因子的層面，激發因素皆經常被提及與討論。本研究以未失眠但具有高睡眠脆弱特質之族群作為研究對象，企圖模擬失眠者在失眠初發之前的狀態，另加入睡前及早晨的自主神經系統與主觀激發狀態之測量，比較其與低睡眠脆弱特質組在有無壓力誘發狀態下，兩組的睡眠中訊息處理程度、睡前及早晨的自主神經系統激發程度與主觀激發狀態，藉以了解（1）壓力是否會引發個體進入較高的主、客觀激發狀態；(2)睡眠脆弱特質在不同的激發程度下，是否會對於個體的睡前、睡眠中及早晨的激發程度造成影響，以初步推論激發狀態因素在失眠病程中所扮演之角色。 方法 以壓力下失眠反應量表（Ford Insomnia Response to Stress Test, FIRST）區分出高睡眠脆弱特質組（HV）13位（6男、7女）、低睡眠脆弱特質組（LV）14位（6男、8女），年齡為19-32歲之健康成人。每位受試者均需在睡眠實驗室進行3個夜晚之夜間睡眠記錄，包含第一晚的PSG測量以排除個案有其他睡眠疾患，並讓受試者習慣實驗環境；第二晚及第三晚則交互平衡操弄基礎情境或壓力情境，在壓力情境下，受試者被告知將進行即席演講，需在睡前抽出演講題目，並於隔天早晨喚醒後10分鐘內發表一篇內容充實、呼應主題之演說，隨後測量心跳變異率（heart rate variability，HRV）、平均心跳間距（r-r interval，RRI）、主觀身心激發狀態（Pre sleep arousal scale，PSAS）等主客觀激發程度，入睡後進行整夜的ERP（事件相關電位[Event-related potential，ERP]）測量，誘發ERPs之實驗流程以oddball典範為主，並於早晨起床後再次進行HRV、RRI及PSAS之測量。 結果 以2（組別）x 2（情境）之混和設計二因子變異數分析比較各組在不同情境下的主客觀激發程度指標是否有所不同。結果顯示壓力確實能引發受試者進入較高的主客觀激發狀態，包含睡前及早晨有較快的平均心跳速率、較高的主觀身心激發程度，而在NREM睡眠中可發現HV在有壓力的情境下，對外界的訊息處理程度顯著較LV組增加，包含較快的N1與較慢的N350出現，但HV在壓力下亦有較快的P2與較強的P900，顯示對外界訊息的抑制功能上升，且隨著睡眠更深，進入慢波睡眠時，P2及P900亦顯著較LV強，表現出更多的抑制外界訊息能力。 結論 本研究探討睡眠脆弱特質在不同激發狀態下對於睡眠的影響，結果顯示壓力能引發個體較高的激發狀態，然而在未失眠族群中，高睡眠脆弱特質對於認知激發狀態下的睡眠中訊息處理程度雖有提高，但可發現高睡眠脆弱特質組亦有較高的抑制外界訊息能力，而不致於使得激發狀態過度影響其睡眠，此結果部分支持失眠的認知神經過度激發假說，過度激發因素可能在持續因子的部分影響性較大，造成個體產生對睡眠的抱怨。另一方面，本研究結果亦彰顯失眠的高危險群的及早介入預防之重要性。 / Introduction: Conditioned hyperarousal has been theorized to be a major etiological factor in chronic insomnia patients. It has been suggested that insomniacs showed more arousal than non-insomniacs in many psychological and physiological measures, including enhanced information processing during sleep as measured by NREM ERPs. However, the role of hyperarousal on the course of development of insomnia remained unclear, it has been considered as a predisposing trait and/or perpetuating factor in the cause of insomnia. The present study compared arousal levels around sleep onset, during sleep and in the morning between non-insomniac individuals with low and high vulnerability to stress-related sleep disturbance, as high vulnerable individuals were proposed to be predisposed to chronic insomnia. The objectives of the study are therefore to investigate: 1) the effect of an acute stressor on arousal level prior, after, and during sleep, and 2) the differences in stress reactivity between individual with high and low vulnerability to stress-related sleep disturbances. Method: Fourteen healthy individuals scoring low (LV) and thirteen healthy individuals scoring high (HV) on the Ford Insomnia Response to Stress Test (FIRST) were studied. All subjects had to sleep in the lab for three nights: a screening/ adaptation night, then a baseline night and a stress night. The sequences of the later two nights were counterbalanced across subjects. During the stress night, subjects were informed that they would be required to give a speech immediately after waking up. Heart rate variability (HRV), r-r interval (RRI) of EKG and subjective arousal level (Pre sleep arousal scale, [PSAS]) was accessed before lights off and in the morning. Then sleep ERP was recorded during the night. An odd-ball paradigm was conducted to evoke ERPs throughout the night. Result: Mixed two-way ANOVAs were used to compare the differences between vulnerabilities and conditions. Both HV ans LV showed higher subjective arousal level and smaller RRI under the stress condition than the baseline condition. During NREM sleep, HV showed increased information processing under stress condition than LV as shown by faster N1, slower N350. Also, HV showed increased inhibition under stress condition than LV, with faster P2 and higher P900. Stronger P2 and P900 were also observed in HV when reached slow-wave sleep stage, comparing to LV. Conclusion: Both groups showed more arousal under stress condition than baseline condition, indicating an acute stress can lead to a higher arousal level. Moreover, high vulnerable individuals showed increased information processing during NREM sleep than low vulnerable individuals, but also showed more inhibition to prevent sleep from being disturbed by external stimulus. The results suggest that increased information processing in reacting to stress may be a predisposing trait for sleep disturbances. However, sleep may be protected by a compensatory inhibitory process in those who have not developed a long-term insomnia.