This is an essay about reasons for action. It begins with two rather ordinary observations. The first is that these two uses of the term ‘reason’ roughly correspond with the two main roles that a reason can play: the role of favoring a prospective course of action, and the role of explaining action. Each of these roles seems crucial to a philosophical account of reasons, and it is not obvious that either has claim to priority. The second observation is that accommodating each of these roles seems to place restrictions on what we can say about reasons for action, and those who lean toward giving priority to one role rather than the other tend also to give priority to the corresponding set of restrictions. They take that set as given, and then focus their efforts on finding a way to meet the other set if they can. Accommodating the explanatory role has seemed to many to require that a reason bear some relation to the motivations of the agent for whom it is reason. One might wonder: what sense could there be in calling something a reason for me to act if it were not in any way capable of explaining my being moved to act? I argue, however, that accepting this sort of internalist condition on something’s being a reason to act precludes accepting a condition of objectivity that is imposed on us if we wish to accommodate the favoring role: sometimes, at least, when we have a reason to act, we could not cease to have that reason simply by having a (perhaps radically) different set of attitudes. I then consider whether the reverse might be true of externalist theories. Does taking the favoring role as one’s starting point preclude a full account of the explanatory role of reasons? I argue that it does not. I show that an Anscombean conception of intentional action allows for a fairly clean solution to a pair of puzzles that motivate this worry. This approach relieves much of the pressure to think of reasons as being tied to motivational attitudes. / text

No description available.

Ce travail a pour objectif d'éclaircir, par la voie d'une enquête sur le langage de l'action intentionnelle et la grammaire des concepts psychologiques, la nature de notre capacité ordinaire à assigner un sens aux actions humaines et les conditions de son exercice. À l'heure actuelle, les débats sont dominés par une conception causaliste de l'action et de son explication qui innerve des problèmes tels que celui de la causalité mentale, de la réalité des états mentaux ou du statut de la « psychologie populaire ». Le causalisme pose que la différence entre une action et un simple mouvement corporel tient au fait que la première est causée par certains antécédents mentaux, distincts du mouvement, qui rendent compte de son caractère intentionnel. Expliquer une action consisterait dès lors à donner des raisons référant à ces états. Renouant avec l'approche logico-grammaticale issue de Wittgenstein et développée par Anscombe, nous entendons montrer que cette conception est erronée. Son principal défaut est de succomber à une image mentaliste selon laquelle les prédicats psychologiques utilisés dans les explications ordinaires ont toujours pour fonction de dénoter des états, des processus ou des événements internes à l'agent, distincts de l'action et pouvant entrer dans des chaînes causales. Contre cela, nous voulons établir que la différence entre action et mouvement corporel est une différence entre deux modes de description. Le mode intentionnel dans lequel se disent les actions se caractérise de façon essentielle par l'asymétrie de la première et de la troisième personnes. Expliquer une action consiste à en donner une description qui l'identifie dans sa dimension téléologique. Notre démarche est la suivante. Le premier mouvement de notre étude est consacré à la clarification des liens conceptuels entre intention et action. Après avoir rappelé les origines modernes de la conception causaliste, nous critiquons les arguments proposés par Davidson en sa faveur. Récusant leur mentalisme sous-jacent, nous montrons que le caractère intentionnel d'une action ne dépend pas de la présence d'un « ingrédient » mental. Puis nous rendons compte de la force explicative des descriptions intentionnelles en les mettant en rapport avec la spécificité du raisonnement pratique et de la connaissance d'agent. Dans un second mouvement, nous revenons sur la nature de la capacité à formuler des jugements psychologiques au sujet d'autrui. Nous examinons successivement la thèse selon laquelle les agents possèdent une théorie de l'esprit et celle selon laquelle ils recourent à un principe de charité interprétative. Contre ces approches intellectualistes, nous soutenons que l'usage sensé des termes psychologiques est un savoir-faire qui doit être mis en relation avec l'expressivité naturelle du corps humain, la dimension intrinsèquement sociale et institutionnelle de la vie humaine et l'existence de formes d'action et d'expression partagées faisant autorité. Reconnaître les conditions actionnelles de l'intelligibilité d'autrui exige de se défaire de l'illusion moderne d'un « sujet désengagé ». / My objective is to elucidate, by means of a grammatical inquiry into the language of intentional action and the use of psychological concepts, the nature of our ordinary capacity to make sense of human actions, as well as the conditions of its exercise. In contemporary debates prevails a causalist view of action and its explanation, which underlies the discussions about mental causation, the reality of mental states or the epistemic status of so-called folk psychology. Causalism holds that an action differs from a mere bodily movement by its being caused by some mental antecedents, distinct from movement, which account for its intentional character. Explaining an action would thus consist in giving reasons ultimately referring to those states. In line with the wittgensteinian logico-grammatical approach developed by Anscombe, I intend to show that this view is misconceived. Its main defect lies in its mentalism, i.e. the tendency to see psychological predicates as always functioning to denote internal episodes or events, distinct from action and liable to figure in causal chains as links bringing about the agent's behaviour. Against this, I want to establish that the distinction between action and mere movement is to be seen as a difference between two modes of description. The intentional mode in which are couched action descriptions is essentially characterized by the first/third-person asymmetry. Explaining an action consists in identifying it by a description exhibiting its teleological features. My plan is the following. The first main move is to clarify the conceptual relations between intention and action. Having brought to light the modern roots of causalism, I consider the arguments Davidson put forward in its favour, so as to refute their underlying mentalism. I show that for an action, being intentional does not depend on any occurrence of a separate mental component. Then I proceed to account for the explanatory force of intentional descriptions by relating them to the specificity of practical reasoning and agent knowledge. The second main move tackles the question to know what the capacity of making sense of others consists in. I successively consider the idea that agents possess a theory of mind and the thesis as to which they must rely on a principle of interpretative charity. Against those intellectualist views, I claim that sensible use of psychological concepts is a practical skill which is to be related to the sensibility to the natural expressivity of human body, the intrinsically social and institutional features of human life and the existence of shared standards of expression and action. Acknowledging the action-based conditions of others' intelligibility requires to dispose of the modern illusion of a « disengaged subject ».

The starting point of the dissertation is a special kind of intentional action -- Aristotelian praxis, or, in a more metaphysical register, energeia -- a kind whose agent's intention in acting must be expressible as the deliverance of one's prohairesis (``deliberate choice''), action that is the embodiment of one's conception of eupraxia (``acting well''), and, equivalently, of eudaimonia (``happiness''). It is special, since not all that we intentionally do can be intelligibly expressed as the deliverance of our conceptions of acting well. Recognition of the gaps between action in general and intentional action more specifically, and between intentional action and prohairetic action, sets the stage for a reinterpretation, not only of core aspects of Aristotle's Ethics, but also of central features of Aristotle's political recommendations. The interpretation defended here centers on the claim that, for Aristotle, defective political communities are often marked, not so much by an erroneous conception of human virtue, but by defective forms of action, forms in which agents fail to apply certain concepts to what they do. Importantly, such failures do not hang on the different failure to apply concepts correctly: the failure to act prohairetically need not come to the failure to grasp the correct conception of human virtue or of human happiness. / Government

Cette étude porte sur la notion de raisonnement. L'hypothèse principale que je défends est que ce que l'on appelle communément un raisonnement comporte un aspect processuel et un aspect inférentiel. L'aspect processuel consiste dans le fait qu'un raisonnement est un enchaînement causal d'actes ou d'états mentaux, et l'aspect inférentiel consiste dans le fait que les états mentaux en question ont des contenus propositionnels qui peuvent entretenir entre eux des relations logiques. Lorsqu'on accepte cette distinction, il apparaît qu'une théorie adéquate du raisonnement doit expliquer comment ces deux aspects sont reliés. L'objectif de cette étude est d'examiner de quelle manière différentes approches du raisonnement traitent de la relation entre ces deux aspects. Cette thèse est divisée en quatre chapitres. Le premier chapitre, à caractère introductif, présente le contexte philosophique général dans lequel s'est posée la réflexion sur le raisonnement (pratique) au cours des dernières décennies. Le deuxième chapitre s'amorce par une analyse critique de la thèse d'Anscombe (1957) selon laquelle l'action est la conclusion de l'inférence pratique. J'y montre comment cette suggestion l'oblige à rendre compte du caractère processuel du raisonnement en termes inférentiels et qu'une telle solution conduit à une compréhension inadéquate de la relation entre les aspects processuel et inférentiel de ce type de raisonnement. Dans la section qui suit, je présente l'hypothèse qui est au coeur de cette étude: j'y explique ce qui distingue les aspects processuel et inférentiel du raisonnement en indiquant quelques conséquences qu'entraîne cette distinction dans l'étude du raisonnement. J'examine ensuite comment il serait plausible, à la lumière de la distinction introduite, de rendre justice à l'intuition d'Anscombe. Le troisième chapitre porte sur la conception du raisonnement de Harman (1986). Après avoir présenté sommairement l'esprit de sa théorie et les principes de révision qu'il identifie comme étant ceux que suivent ou doivent suivre les agents lorsqu'ils raisonnent (section 1), je tente ensuite de déterminer comment Harman conçoit cet aspect processuel, et si sa théorie permet d'expliquer de manière adéquate la relation entre les aspects processuel et inférentiel du raisonnement. J'examine d'abord quelle fonction il reconnaît aux notions d'implication et d'inconsistance, et montre ensuite que si le processus dans lequel s'engage celui qui raisonne peut être conçu à la manière de Harman, c'est-à-dire à la 3ème personne, il doit aussi être conçu du point de vue de l'agent qui raisonne, c'est-à-dire à la 1ère personne. Après avoir discuté des limites inhérentes au fait de concevoir l'aspect processuel du raisonnement d'un point de vue essentiellement externe, à la 3ème personne; j'identifie quelques conséquences que fait apparaître la perspective de la 1ère personne, dans le but de montrer pourquoi une théorie adéquate du raisonnement ne peut en faire l'économie. En dernière section, j'explore quelques conséquences de la distinction proposée, notamment quant à la question de savoir quelles notions pourraient être mises en parallèle à celles de prémisse, conclusion et validité lorsqu'il est question du raisonnement. Le quatrième chapitre porte sur la conception du raisonnement de John Broome. Après avoir exposé sommairement les thèses que défend Broome (1999, 2002) sur le terrain de la normativité puisqu'elles sont déterminantes dans sa théorie du raisonnement, je montre que la distinction qu'il établit entre un raisonnement de croyance et un raisonnement d'intention s'accorde (au moins en partie) avec la distinction raisonnement/argument suggérée (section 1). Je propose ensuite une analyse critique de la position qu'il défend dans Broome (2006b) en m'opposant à l'exigence de double expression qui est nécessaire, selon lui, pour pouvoir raisonner avec des attitudes autres que des croyances. Je montre qu'une telle conception du raisonnement est beaucoup trop exigeante, et qu'elle a le désavantage principal de faire en sorte que pour pouvoir raisonner avec des attitudes autres que des croyances, vous devez vous mettre en position de 3ème personne (section 2). Dans le cadre de la dernière section, je montre que la divergence entre nos conceptions respectives du raisonnement a son origine dans les thèses qu'il défend concernant la normativité de la rationalité et la normativité des raisons. J'identifie quelques-unes des conséquences de ces thèses pour une théorie du raisonnement. ______________________________________________________________________________ MOTS-CLÉS DE L’AUTEUR : Raisonnement, Argument, Inférence, Raisons, Rationalité, Normativité.

Mémoire numérisé par la Direction des bibliothèques de l'Université de Montréal.

Este trabalho tem como objetivo o desenvolvimento de um novo método para a simulação de redução da dose de radiação em imagens mamográficas clínicas. Assim, estudos sobre a influência da redução da dose de radiação no diagnóstico do câncer de mama podem ser realizados sem que o paciente se exponha a doses extras de radiação. Uma análise preliminar foi realizada para a caracterização do ruído produzido pelo equipamento mamográfico no processo de aquisição da imagem. Essa análise evidenciou a importância de um método local de simulação, uma vez que o ruído depende da posição ao longo do campo. O novo método proposto consiste em ajustar os níveis de cinza e adicionar uma máscara de ruído Poisson, dependente do sinal, nas imagens clínicas adquiridas com a dose de radiação padrão, simulando sua aquisição com doses de radiação reduzidas. A dependência entre ruído e sinal foi criada com o uso da transformada de Anscombe. O desempenho do método proposto foi avaliado utilizando-se imagens mamográficas de um phantom antropomórfico obtidas com diferentes doses de radiação. As imagens simuladas pelo método proposto foram comparadas com as imagens reais. A similaridade entre os espectros de ruído permitiu a comparação de métricas locais da imagem. O erro percentual entre os níveis de cinza das imagens reais e simuladas se manteve inferior a 1%. O ruído adicionado manteve um erro percentual inferior a 1%. Testes de t-Student mostraram que não existe diferença estatística significante (p < 0,05) entre as imagens reais e simuladas pelo método proposto. / This work aims to develop a new method for simulating reduction of the radiation dose in clinical digital mammography. Using such method, studies regarding the influence of dose reduction in cancer diagnosis can be performed without unnecessary exposure of patients to X-ray radiation. A preliminary study characterized the noise produced by the digital mammography equipment during the acquisition process. This analysis emphasized the importance of simulating noise locally, since noise is dependent on the spatial position of the pixel. Therefore, the proposed method consists of adjusting the gray levels and adding signal-dependent Poisson noise to images acquired at the standard radiation dose. Dependency between noise and signal was created using the Anscombe transformation. The performance of the proposed method was evaluated using mammographic images of an anthropomorphic phantom acquired at different radiation doses. Images simulated using the proposed method were compared to real images acquired using the clinical equipment. Similarity between noise power spectra and local metrics validated the similarity between images. The gray level of the simulated and real images were compared using local mean and reported averaged errors smaller than 1%. The added noise was also compared and the averaged error was smaller than 1%. Statistical Student\'s t-test tests showed no statistical difference (p < 0.05) between real images and the ones simulated using the proposed method.

Este trabalho tem por objetivo o desenvolvimento de uma nova técnica de pré-processamento de imagens mamográficas digitais para melhorar o desempenho dos esquemas computacionais de auxílio ao diagnóstico (CAD) e para auxiliar na detecção precoce do câncer de mama. O método proposto efetua uma restauração nas imagens mamográficas utilizando, em uma primeira etapa, a transformada de Anscombe e o filtro de Wiener para redução do ruído quântico. Posteriormente, é utilizado o filtro inverso da função de transferência de modulação (MTF) do sistema de imagem para realce das estruturas de interesse na mamografia, como as microcalcificações, que podem ser um indicativo de câncer de mama em seu estágio inicial. Imagens mamográficas restauradas pelo método proposto foram utilizadas na avaliação de um esquema CAD para detecção automática de microcalcificações. Os resultados mostraram que o desempenho do esquema CAD apresentou uma melhora significativa quando imagens restauradas foram utilizadas, mesmo para imagens de mamas densas, que resultam normalmente em baixa taxa de detecção devido ao baixo contraste. / This work aims to developing a new technique for pre-processing digital mammographic images in order to improve the performance of computer aided-diagnosis schemes (CAD) and to assist in early detection of breast cancer. The proposed method performs a restoration in mammographic images using in a first step, the Anscombe transform and Wiener filtering to reduce image quantum noise. Subsequently, it was used the inverse modulation transfer function filtering (MTF) considering the imaging system to enhance structures of interest in mammography, such as microcalcifications, which may be an indicative of breast cancer in its early stage. Mammographic images restored by the proposed method were used in the evaluation of a CAD scheme for automatic detection of microcalcifications. The results showed that the performance of the CAD scheme had a significant improvement when restored images were used, even for images of dense breasts, which often results in low detection rate due to low contrast.

Este trabalho tem como objetivo o desenvolvimento de um novo método para a simulação de redução da dose de radiação em imagens mamográficas clínicas. Assim, estudos sobre a influência da redução da dose de radiação no diagnóstico do câncer de mama podem ser realizados sem que o paciente se exponha a doses extras de radiação. Uma análise preliminar foi realizada para a caracterização do ruído produzido pelo equipamento mamográfico no processo de aquisição da imagem. Essa análise evidenciou a importância de um método local de simulação, uma vez que o ruído depende da posição ao longo do campo. O novo método proposto consiste em ajustar os níveis de cinza e adicionar uma máscara de ruído Poisson, dependente do sinal, nas imagens clínicas adquiridas com a dose de radiação padrão, simulando sua aquisição com doses de radiação reduzidas. A dependência entre ruído e sinal foi criada com o uso da transformada de Anscombe. O desempenho do método proposto foi avaliado utilizando-se imagens mamográficas de um phantom antropomórfico obtidas com diferentes doses de radiação. As imagens simuladas pelo método proposto foram comparadas com as imagens reais. A similaridade entre os espectros de ruído permitiu a comparação de métricas locais da imagem. O erro percentual entre os níveis de cinza das imagens reais e simuladas se manteve inferior a 1%. O ruído adicionado manteve um erro percentual inferior a 1%. Testes de t-Student mostraram que não existe diferença estatística significante (p < 0,05) entre as imagens reais e simuladas pelo método proposto. / This work aims to develop a new method for simulating reduction of the radiation dose in clinical digital mammography. Using such method, studies regarding the influence of dose reduction in cancer diagnosis can be performed without unnecessary exposure of patients to X-ray radiation. A preliminary study characterized the noise produced by the digital mammography equipment during the acquisition process. This analysis emphasized the importance of simulating noise locally, since noise is dependent on the spatial position of the pixel. Therefore, the proposed method consists of adjusting the gray levels and adding signal-dependent Poisson noise to images acquired at the standard radiation dose. Dependency between noise and signal was created using the Anscombe transformation. The performance of the proposed method was evaluated using mammographic images of an anthropomorphic phantom acquired at different radiation doses. Images simulated using the proposed method were compared to real images acquired using the clinical equipment. Similarity between noise power spectra and local metrics validated the similarity between images. The gray level of the simulated and real images were compared using local mean and reported averaged errors smaller than 1%. The added noise was also compared and the averaged error was smaller than 1%. Statistical Student\'s t-test tests showed no statistical difference (p < 0.05) between real images and the ones simulated using the proposed method.

Este trabalho tem por objetivo o desenvolvimento de uma nova técnica de pré-processamento de imagens mamográficas digitais para melhorar o desempenho dos esquemas computacionais de auxílio ao diagnóstico (CAD) e para auxiliar na detecção precoce do câncer de mama. O método proposto efetua uma restauração nas imagens mamográficas utilizando, em uma primeira etapa, a transformada de Anscombe e o filtro de Wiener para redução do ruído quântico. Posteriormente, é utilizado o filtro inverso da função de transferência de modulação (MTF) do sistema de imagem para realce das estruturas de interesse na mamografia, como as microcalcificações, que podem ser um indicativo de câncer de mama em seu estágio inicial. Imagens mamográficas restauradas pelo método proposto foram utilizadas na avaliação de um esquema CAD para detecção automática de microcalcificações. Os resultados mostraram que o desempenho do esquema CAD apresentou uma melhora significativa quando imagens restauradas foram utilizadas, mesmo para imagens de mamas densas, que resultam normalmente em baixa taxa de detecção devido ao baixo contraste. / This work aims to developing a new technique for pre-processing digital mammographic images in order to improve the performance of computer aided-diagnosis schemes (CAD) and to assist in early detection of breast cancer. The proposed method performs a restoration in mammographic images using in a first step, the Anscombe transform and Wiener filtering to reduce image quantum noise. Subsequently, it was used the inverse modulation transfer function filtering (MTF) considering the imaging system to enhance structures of interest in mammography, such as microcalcifications, which may be an indicative of breast cancer in its early stage. Mammographic images restored by the proposed method were used in the evaluation of a CAD scheme for automatic detection of microcalcifications. The results showed that the performance of the CAD scheme had a significant improvement when restored images were used, even for images of dense breasts, which often results in low detection rate due to low contrast.