The environment is abundant with visual information. Each moment, this information competes for representation in the brain. From billboards and pop-up ads to smart phones and flat screens, in modern society our attention is constantly drawn from one salient object to the next. Learning how to focus on the objects that are most important for the current task is a major developmental hurdle. Fortunately, rewards help us to learn what is important by providing feedback signals to the brain. Sometimes, in adolescence for example, reward seeking can become the pre-potent response. This can ultimately lead to risky and impulsive behaviors that have devastating consequences. Until recently, little has been known about how rewards operate to influence the focus of attention. In this document, I first demonstrate the robustness of various behavioral paradigms designed to measure reward processing in vision. I found that even mundane rewards, such as images of money, are effective enough to prime the attentional system on the basis of value. Remarkably, this effect extended to images of Monopoly money. This observation suggests that whole classes of visual stimuli, such as food, pornography, commercial logos, corporate brands, or money, each with its own reward salience value, are likely vying for representation in the brain. This work has implications for the growing digital economy as it suggests that novel value systems, such as the digital currency Bitcoin, could eventually become as psychologically relevant as physical currency provided sufficient use and exposure. Likewise, this work has implications for gamification in the industrial setting. Next, I examined the sensitivity of the system to make optimal economic decisions. When faced with an economic choice normative theories of decision-making suggest that the economic actor will choose the response that affords the greatest expected utility. Contrary to this account, I developed a new behavioral paradigm (reward contingent capture) and reveal that the attentional homunculus is a fuzzy mathematician. Specifically, I found that low-level attentional processes conform to the same probability distortions observed in prospect theory. This finding supports a unified value learning mechanism across several domains of cognition and converges with evidence from monkey models. Then, I demonstrate the influence of rewards on high-order search parameters. I found that images of money can implicitly encourage observers to preferentially adopt one of two search strategies – one that values salience versus one that values goals. Together, my results expose two distinct ways in which the very same rewards can affect attentional behavior – by tuning the salience of specific features and by shaping global search mode settings. Lastly, I draw from my empirical results to present a unified model of the manifold role of rewards on visual attention. This model makes clear predictions for clinical applications of rewarded attention paradigms because it incorporates a dimension of complexity upon which learning processes can operate on attention. Thus, future work should acknowledge how individual traits such as developmental trajectory, impulsivity, and risk-seeking factors differentially interact with low- and high-level attentional processes. In sum, this document puts forward the notion that rewards serve a compelling role in visual awareness. The key point however is not that rewards can have an effect on attention but that due to the nature of visual processing, reward signals are likely always tuning attention. In this way we can consider reward salience an attentional currency. This means then that deciding where to attend is a matter of gains and losses.

La capture contingente de l’attention est un phénomène dans lequel les mécanismes d’orientation endogène et exogène de l’attention interagissent, de sorte qu’une propriété qui est pertinente à la tâche en cours, et donc qui fait l’objet de contrôles attentionnels descendants, endogènes, capture l’attention de façon involontaire, exogène, vers sa position spatiale. Dans cette thèse, trois aspects de ce phénomène ont été étudiés. Premièrement, en explorant le décours temporel de la capture contingente de l’attention et la réponse électrophysiologique à des distracteurs capturant ainsi l’attention, il a été établi que le déficit comportemental symptomatique de cette forme de capture était lié à un déploiement de l’attention visuospatiale vers la position du distracteur, et que ce traitement spatialement sélectif pouvait être modulé par le partage d’autres propriétés entre le distracteur et la cible. Deuxièmement, l’utilisation des potentiels liés aux événements a permis de dissocier l’hypothèse de capture contingente de l’attention et l’hypothèse de capture pure de l’attention. Selon cette interprétation, un stimulus ne peut capturer l’attention aux stades préattentifs de traitement que s’il présente le plus fort signal ascendant parmi tous les stimuli présents. Les contrôles attentionnels descendants ne serviraient donc qu’à désengager l’attention d’un tel stimulus. Les résultats présentés ici vont à l’encontre d’une telle interprétation, puisqu’un déploiement de l’attention visuospatiale, indexé par la présence d’une N2pc, n’a été observé que lorsqu’un distracteur périphérique possédait une caractéristique pertinente à la tâche en cours, même lorsque ses propriétés de bas niveau n’étaient pas plus saillantes que celles des autres items présents. Finalement, en utilisant un paradigme où la cible était définie en fonction de son appartenance à une catégorie alphanumérique, il a été démontré que des contrôles attentionnels en faveur d’un attribut conceptuel pouvaient guider l’attention visuospatiale de façon involontaire, rejetant une nouvelle fois l’hypothèse de la capture pure de l’attention. / Contingent involuntary orienting is a phenomenon in which endogenous and exogenous attentional mechanisms interact, such that an item captures attention only if it shares an attribute that is relevant for the task at hand. Hence, top-down attentional control settings are established endogenously in favour of the relevant attribute, but stimuli sharing this attribute draw attention to their location involuntarily. The present thesis explores three aspects of this phenomenon. First, by studying the time course of this contingent capture effect, and by measuring event-related potentials (ERPs) to capturing distractors, it has been established that the performance deficits linked to contingent capture are in fact due to a deployment of visuospatial attention to the location of the distractor. Moreover, this spatially selective processing of the capturing distractor can be modulated if the distractor shares another target attribute, beside the target defining attribute. The ERP technique also permitted the dissociation of the contingent involuntary orienting hypothesis and the pure capture with brief attentional dwell time hypothesis. According to the latter interpretation, only salient singletons have the ability to capture attention at preattentive stages of processing. Therefore, top-down attentional control settings serve only to disengage attention from the location of such singletons when they do not share target features. The present results argue against this interpretation, because a deployment of visuospatial attention, indexed by the presence of the N2pc, was observed only in response to peripheral distractors sharing the target-defining attribute, even when all items in the stimulus displays were equated in terms of bottom-up salience. Lastly, when alphanumeric category was used to define the target, it was shown that top-down attentional control settings in favour of such conceptual attributes could be successfully implemented and used to guide visuospatial attention in an exogenous fashion, providing further evidence against the pure capture hypothesis.

La capture contingente de l’attention est un phénomène dans lequel les mécanismes d’orientation endogène et exogène de l’attention interagissent, de sorte qu’une propriété qui est pertinente à la tâche en cours, et donc qui fait l’objet de contrôles attentionnels descendants, endogènes, capture l’attention de façon involontaire, exogène, vers sa position spatiale. Dans cette thèse, trois aspects de ce phénomène ont été étudiés. Premièrement, en explorant le décours temporel de la capture contingente de l’attention et la réponse électrophysiologique à des distracteurs capturant ainsi l’attention, il a été établi que le déficit comportemental symptomatique de cette forme de capture était lié à un déploiement de l’attention visuospatiale vers la position du distracteur, et que ce traitement spatialement sélectif pouvait être modulé par le partage d’autres propriétés entre le distracteur et la cible. Deuxièmement, l’utilisation des potentiels liés aux événements a permis de dissocier l’hypothèse de capture contingente de l’attention et l’hypothèse de capture pure de l’attention. Selon cette interprétation, un stimulus ne peut capturer l’attention aux stades préattentifs de traitement que s’il présente le plus fort signal ascendant parmi tous les stimuli présents. Les contrôles attentionnels descendants ne serviraient donc qu’à désengager l’attention d’un tel stimulus. Les résultats présentés ici vont à l’encontre d’une telle interprétation, puisqu’un déploiement de l’attention visuospatiale, indexé par la présence d’une N2pc, n’a été observé que lorsqu’un distracteur périphérique possédait une caractéristique pertinente à la tâche en cours, même lorsque ses propriétés de bas niveau n’étaient pas plus saillantes que celles des autres items présents. Finalement, en utilisant un paradigme où la cible était définie en fonction de son appartenance à une catégorie alphanumérique, il a été démontré que des contrôles attentionnels en faveur d’un attribut conceptuel pouvaient guider l’attention visuospatiale de façon involontaire, rejetant une nouvelle fois l’hypothèse de la capture pure de l’attention. / Contingent involuntary orienting is a phenomenon in which endogenous and exogenous attentional mechanisms interact, such that an item captures attention only if it shares an attribute that is relevant for the task at hand. Hence, top-down attentional control settings are established endogenously in favour of the relevant attribute, but stimuli sharing this attribute draw attention to their location involuntarily. The present thesis explores three aspects of this phenomenon. First, by studying the time course of this contingent capture effect, and by measuring event-related potentials (ERPs) to capturing distractors, it has been established that the performance deficits linked to contingent capture are in fact due to a deployment of visuospatial attention to the location of the distractor. Moreover, this spatially selective processing of the capturing distractor can be modulated if the distractor shares another target attribute, beside the target defining attribute. The ERP technique also permitted the dissociation of the contingent involuntary orienting hypothesis and the pure capture with brief attentional dwell time hypothesis. According to the latter interpretation, only salient singletons have the ability to capture attention at preattentive stages of processing. Therefore, top-down attentional control settings serve only to disengage attention from the location of such singletons when they do not share target features. The present results argue against this interpretation, because a deployment of visuospatial attention, indexed by the presence of the N2pc, was observed only in response to peripheral distractors sharing the target-defining attribute, even when all items in the stimulus displays were equated in terms of bottom-up salience. Lastly, when alphanumeric category was used to define the target, it was shown that top-down attentional control settings in favour of such conceptual attributes could be successfully implemented and used to guide visuospatial attention in an exogenous fashion, providing further evidence against the pure capture hypothesis.