No description available.

No description available.

No description available.

Although there is no consensus among biologists as to whether human behaviour can beexplained by evolution, there are a number of theories and models in different fields ofresearch that aim to do just that. Philosophy is no exception. In metaethics, evolutionarybiology is used to formulate an evolutionary debunking explanation. This skepticalepistemological tool is used to show that if evolution has, in some way, affected humanmorality, then we cannot have true justified belief in moral matters.An ongoing debate about the evolutionary debunking explanation is about howmuch empirical detail the evolutionary debunking explanation can demand. With this paper, Iwant to examine how philosophers writing about the evolutionary debunking explanation useevolutionary biology, as well as how much evolutionary biology is required for theevolutionary debunking explanation to be valid. I will argue that it is possible to identifythree difficulties in using evolutionary biology to formulate a philosophical tool.

The problem of historical explanation consists in how historical facts are put together. No mere collection of facts constitutes an explanation: there must be some underlying explanation for why those facts occurred in the way they did. Many competing theories of historical explanation have thus been offered, from the highly technical D-N or covering law model, to narrative-based explanations. This paper exposes the flaws in the covering law model proposed by Carl Hempel, and offers a justification for narrative-based explanations by appealing to the notion of language games as used by Ludwig Wittgenstein, as well as the narrative and paradigm models of Arthur Danto and Thomas Kuhn for explaining historical events. / Master of Arts / The problem of historical explanation consists in how historical facts are put together. No mere collection of facts constitutes an explanation: there must be some underlying explanation for why those facts occurred in the way they did. Many competing theories of historical explanation have thus been offered, from the highly technical D-N or covering law model, which imitates the methods of explanation in “hard” scientific inquiry through a careful description of initial conditions and relevant laws and formulas, to narrative-based explanations, or explanations that use a story with a beginning, a middle, and an end. This paper exposes the flaws in the covering law model proposed by Carl Hempel, and offers a justification for narrative-based explanations by appealing to the notion of language games as used by Ludwig Wittgenstein, as well as the narrative and paradigm models of Arthur Danto and Thomas Kuhn for explaining historical events. The aim of this project is to prevent scientific analysis being incorrectly applied to non-scientific entities, such as persons (e.g. Napoleon Bonaparte) and places (e.g. Russia) which are referenced in ordinary language, and which are in principle irreducible to the primary entities of the so-called “hard” sciences, such as subatomic particles and fundamental forces.

In this essay I defend the notion of deductive explanation mainly against two types of putative counterexamples: those found in genuinely indeterministic systems and those found in complex dynamic systems. Using Railton's notions of explanatory information and ideal explanatory text, deductivism is defended in an indeterministic setting. Furthermore, an argument against non-deductivism that hinges on peculiarities of probabilistic causality is presented. The use of the notion of an ideal explanatory text gives rise to problems in accounting for explanations in complex dynamic systems, regardless of whether they are deterministic or not. These problems are considered in the essay and a solution is suggested. This solution forces the deductivist to abandon the requirement that an explanation consists of a deductive argument, but it is argued that the core of deductivism is saved in so far as we, for full explanations, can still adhere to the fundamental requirement: If A explains B, then A is inconsistent with anything inconsistent with B.

Ce travail concerne principalement la notion de modularité hiérarchique et son utilisation pour expliquer la structure et le comportement dynamique des systèmes complexes au moyen de modèles modulaires hiérarchiques, ainsi qu'un concept de ma proposition, l’antimodularité, relié à la possibilité de la détection algorithmique de la modularité hiérarchique. Plus précisément, je mets en évidence la portée pragmatique de la modularité hiérarchique sur la possibilité de l’explication scientifique des systèmes complexes, c’est-à-dire, systèmes qui, selon une description de base choisie par l’observateur, peuvent être considérés comme composés de parties élémentaires discrètes interdépendantes. Je souligne que la modularité hiérarchique est essentielle même au cours de l’expérimentation visée à découvrir la structure de ces systèmes. Mais la détection algorithmique de la modularité hiérarchique se révèle être une tâche affectée par la démontrée intraitabilité computationnelle de la recherche de la meilleure description modulaire hiérarchique, et par l’excessive cherté computationnelle même des méthodes de détection approximatives de la modularité. L’antimodularité consiste en le manque d’une description modulaire appropriée aux exigences de l’observateur, manque dû ou à l’absence de modularité dans la description basique choisie du système, ou à l’impossibilité de produire algorithmiquement une description hiérarchique valide, en raison des dimensions excessives du système à évaluer en relation à la cherté computationnelle des méthodes algorithmiques. Je souligne, de plus, que la modularité et l’antimodularité dépendent du choix pragmatique d’une spécifique description de base du système, choix fait par l’observateur sur la base de ses objectifs explicatifs. Je montre comment l’antimodularité entrave la possibilité d’appliquer au moins trois types bien connus d’explication: mécanique, déductive-nomologique et computationnelle. Un quatrième type, l’explication topologique, reste par contre indemne. Ensuite j’évalue la présence de modularité dans les systèmes biologiques, avec ses possibles conséquences, et l’éventualité d’encourir dans l’antimodularité en biologie et en autres sciences: éventualité assez probable, au moins dans la biologie des systèmes. Je me permet enfin quelques spéculations métaphysiques et historiques plutôt libres. D’un point de vue métaphysique, l’antimodularité semble suggérer une position possible, selon laquelle les espèces naturelles sont modules qui ont été détectés et, en raison de l’intraitabilité computationnelle de la détection de la meilleure description modulaire hiérarchique, il est improbable qu’ils constituent la meilleure façon possible de décrire le monde, parce que la modularité des espèces naturelles assez probablement ne reflète pas la meilleure modularité possible du monde. D’un point de vue historique, l’utilisation croissante des méthodes computationnels pour la détection de la modularité ou pour la simulation de systèmes complexes, en particulier dans certains domaines de la recherche scientifique, suggère la possibilité d’imaginer une multiplicité de disciplines scientifiques émergentes, guidées par une production croissante et auto-alimentante d’explications potentiellement inintelligibles pour les capacités cognitives humaines. Cela, à mon avis, constituerait un changement historique dans la science, qui, s’il n’a pas déjà eu lieu, pourrait bien être sur le point de se produire. / This work is mainly concerned with the notion of hierarchical modularity and its use in explaining structure and dynamical behavior of complex systems by means of hierarchical modular models, as well as with a concept of my proposal, antimodularity, tied to the possibility of the algorithmic detection of hierarchical modularity. Specifically, I highlight the pragmatic bearing of hierarchical modularity on the possibility of scientific explanation of complex systems, that is, systems which, according to a chosen basic description, can be considered as composed of elementary, discrete, interrelated parts. I stress that hierarchical modularity is also required by the experimentation aimed to discover the structure of such systems. Algorithmic detection of hierarchical modularity turns out to be a task plagued by the demonstrated computational intractability of the search for the best hierarchical modular description, and by the high computational expensiveness of even approximated detection methods. Antimodularity consists in the lack of a modular description fitting the needs of the observer, a lack due either to absence of modularity in the system’s chosen basic description, or to the impossibility, due to the excessive size of the system under assessment in relation to the computational cost of algorithmic methods, to algorithmically produce a valid hierarchical description. I stress that modularity and antimodularity depend on the pragmatic choice of a given basic description of the system, a choice made by the observer based on explanatory goals. I show how antimodularity hinders the possibility of applying at least three well-known types of explanation: mechanistic, deductive-nomological and computational. A fourth type, topological explanation, remains unaffected. I then assess the presence of modularity in biological systems, and evaluate the possible consequences, and the likelihood, of incurring in antimodularity in biology and other sciences, concluding that this eventuality is quite likely, at least in systems biology. I finally indulge in some metaphysical and historical speculations: metaphysically, antimodularity seems to suggest a possible position according to which natural kinds are detected modules, and as such, due to the computational hardness of the detection of the best hierarchical modular description, they are unlikely to be the best possible way to describe the world, because the modularity of natural kinds quite probably does not reflect the best possible modularity of the world. From an historical point of view, the growing use of computational methods for modularity detection or simulation of complex systems, especially in certain areas of scientific research, hints at the envisioning of a multiplicity of emerging scientific disciplines guided by a self- sustained, growing production of possibly human-unintelligible explanations. This, I suggest, would constitute an historical change in science, which, if has not already occurred, could well be on the verge of happening. / Questo lavoro riguarda principalmente il concetto di modularità gerarchica e il suo impiego nello spiegare la struttura e il comportamento dinamico di sistemi complessi mediante modelli modulari gerarchici, nonché un concetto di mia proposta, l’antimodularità, legato alla possibilità del rilevamento algoritmico di modularità gerarchica. Nello specifico, evidenzio la portata pragmatica della modularità gerarchica sulla possibilità di spiegazione scientifica dei sistemi complessi, cioè sistemi che, secondo una descrizione di base scelta dall’osservatore, possono essere considerati come composti da parti elementari discrete interrelate. Sottolineo che la modularità gerarchica è essenziale anche nel corso della sperimentazione volta a scoprire la struttura di tali sistemi. Il rilevamento algoritmico della modularità gerarchica si rivela essere un compito affetto dalla dimostrata intrattabilità computazionale della ricerca della migliore descrizione modulare gerarchica, e affetto dal comunque elevato costo computazionale anche dei metodi di rilevamento approssimati della modularità. L’antimodularità consiste nella mancanza di una descrizione modulare adatta alle esigenze dell’osservatore, mancanza dovuta o all’assenza di modularità nella descrizione di base del sistema scelta dall’osservatore, o all’impossibilità di produrre algoritmicamente una sua descrizione gerarchica valida, per le dimensioni eccessive del sistema da valutare in rapporto al costo computazionale dei metodi algoritmici. Sottolineo che modularità e antimodularità dipendono dalla scelta pragmatica di una certa descrizione di base del sistema, scelta fatta dall’osservatore sulla base di obiettivi esplicativi. Mostro poi come l’antimodularità ostacoli la possibilità di applicare almeno tre tipi noti di spiegazione: meccanicistica, deduttivo- nomologica e computazionale. Un quarto tipo di spiegazione, la spiegazione topologica, rimane sostanzialmente immune dalle conseguenze dell’antimodularità. Valuto quindi la presenza di modularità nei sistemi biologici, e le sue possibili conseguenze, nonché l’eventualità di incorrere nell’antimodularità in biologia e in altre scienze, concludendo che questa eventualità è abbastanza probabile, almeno in biologia dei sistemi. Infine, mi permetto alcune speculazioni metafisiche e storiche piuttosto libere. Dal punto di vista metafisico, l’antimodularità sembra suggerire una posizione possibile secondo cui i generi naturali sono moduli che sono stati rilevati, e in quanto tali, a causa dell’intrattabilità computazionale del rilevamento della migliore descrizione modulare gerarchica, è improbabile che essi siano il miglior modo possibile per descrivere il mondo, perché la modularità dei generi naturali molto probabilmente non rispecchia la migliore modularità possibile del mondo. Da un punto di vista storico, il crescente utilizzo di metodi computazionali per il rilevamento della modularità o per la simulazione di sistemi complessi, in particolare in alcuni settori della ricerca scientifica, suggerisce la possibilità di immaginare una molteplicità di discipline scientifiche emergenti, guidate dalla produzione di spiegazioni potenzialmente inintelligibili dal punto di vista cognitivo umano, produzione che potrebbe iniziare ad autoalimentarsi, portando potenzialmente ad una crescita inarrestabile. Suggerisco che questo scenario cos- tituirebbe un cambiamento epocale nel campo della scienza, che, se non è già avvenuto, potrebbe benissimo essere sul punto di realizzarsi.

Reinforcement Learning (RL) algorithms are highly popular in the robotics field to solve complex problems, learn from dynamic environments and generate optimal outcomes. However, one of the main limitations of RL is the lack of model transparency. This includes the inability to provide explanations of why the output was generated. The explainability becomes even more crucial when RL outputs influence human decisions, such as in Human-Robot Collaboration (HRC) scenarios, where safety requirements should be met. This work focuses on the application of two explainability techniques, “Reward Decomposition” and “Autonomous Policy Explanation”, on a RL algorithm which is the core of a risk mitigation module for robots’ operation in a collaborative automated warehouse scenario. The “Reward Decomposition” gives an insight into the factors that impacted the robot’s choice by decomposing the reward function into sub-functions. It also allows creating Minimal Sufficient Explanation (MSX), sets of relevant reasons for each decision taken during the robot’s operation. The second applied technique, “Autonomous Policy Explanation”, provides a global overview of the robot’s behavior by answering queries asked by human users. It also provides insights into the decision guidelines embedded in the robot’s policy. Since the synthesis of the policy descriptions and the queries’ answers are in natural language, this tool facilitates algorithm diagnosis even by non-expert users. The results proved that there is an improvement in the RL algorithm which now chooses more evenly distributed actions and a full policy to the robot’s decisions is produced which is for the most part aligned with the expectations. The work provides an analysis of the results of the application of both techniques which both led to increased transparency of the robot’s decision process. These explainability methods not only built trust in the robot’s choices, which proved to be among the optimal ones in most of the cases but also made it possible to find weaknesses in the robot’s policy, making them a tool helpful for debugging purposes. / Algoritmer för förstärkningsinlärning (RL-algoritmer) är mycket populära inom robotikområdet för att lösa komplexa problem, att lära sig av dynamiska miljöer och att generera optimala resultat. En av de viktigaste begränsningarna för RL är dock bristen på modellens transparens. Detta inkluderar den oförmåga att förklara bakomliggande process (algoritm eller modell) som genererade ett visst returvärde. Förklarbarheten blir ännu viktigare när resultatet från en RL-algoritm påverkar mänskliga beslut, till exempel i HRC-scenarier där säkerhetskrav bör uppfyllas. Detta arbete fokuserar på användningen av två förklarbarhetstekniker, “Reward Decomposition” och “Autonomous policy Explanation”, tillämpat på en RL-algoritm som är kärnan i en riskreduceringsmodul för drift av samarbetande robotars på ett automatiserat lager. “Reward Decomposition” ger en inblick i vilka faktorer som påverkade robotens val genom att bryta ner belöningsfunktionen i mindre funktioner. Det gör det också möjligt att formulera en MSX (minimal sufficient explanation), uppsättning av relevanta skäl för varje beslut som har fattas under robotens drift. Den andra tillämpade tekniken, “Autonomous Policy Explanation”, ger en generellt prespektiv över robotens beteende genom att mänskliga användare får ställa frågor till roboten. Detta ger även insikt i de beslutsriktlinjer som är inbäddade i robotens policy. Ty syntesen av policybeskrivningarna och frågornas svar är naturligt språk underlättar detta en algoritmdiagnos även för icke-expertanvändare. Resultaten visade att det finns en förbättring av RL-algoritmen som nu väljer mer jämnt fördelade åtgärder. Dessutom produceras en fullständig policy för robotens beslut som för det mesta är anpassad till förväntningarna. Rapporten ger en analys av resultaten av tillämpningen av båda teknikerna, som visade att båda ledde till ökad transparens i robotens beslutsprocess. Förklaringsmetoderna gav inte bara förtroende för robotens val, vilket visade sig vara bland de optimala i de flesta fall, utan gjorde det också möjligt att hitta svagheter i robotens policy, vilket gjorde dem till ett verktyg som är användbart för felsökningsändamål.

It is natural to suppose that mental and physical properties are importantly distinct. Yet whatever this difference is, it has to be compatible with interaction between the mind and the body. Satisfaction of these desiderata leads to a paradox. If you make the mind strongly separate from the body, then there is the problem of bringing them together. If you unite them, then there is the problem of preserving their distinctiveness. It is the aim of this thesis to resolve the paradox. From the outset, it is assumed that the nature of interaction is most satisfactorily explained by an account of mental properties in monistic terms. For reasons for space, the arguments of Materialism are concentrated upon at the exposure of Idealism. Three strategies are examined, and found wanting. First, an instance of a non-reductive account provided by Davidson's 'Anomalous Monism'. Here, mental properties seem to be left with no role in influencing behaviour. Second, a review of reductionist accounts, ranging from Identity Theories to Representationalism. Criticism focuses upon the failure of reductionism to explain the connection between the function of a conscious state and its particular character. A Materialist treats mental states as if they were part of the physical universe. This implies that the nature of these states may be understood through scientific investigation, in the same manner as all other phenomena. The third strategy is to deny the above implication: that is, deny the assertion that, by existing, all aspects of an object are thereby knowable. The ideas of Colin McGinn are discussed as an example of this position. Since his arguments are equally suitable for non-Materialist purposes, they do not constitute an exclusively Materialist solution to the above paradox. This thesis offers an alternative way of pursuing the above strategy. It argues that the relation between mental states and our ways of understanding phenomena, is such that we should not expect our theories about the nature of 'mind' and the 'physical world' to employ the same terms. These properties appear distinct, not because they are different substances, but because they occupy different sides of the ‘process of understanding’ - ‘thing understood’ relationship. For convenience, this position is referred to as ‘Agnostic Materialism’. As interaction between the mind and the body is compatible with the mind having no influence upon our behaviour, it is incumbent upon the thesis to defend Materialism against the claim that mental properties are epiphenomenal. This is achieved by teasing out two ways in which such properties are considered inert: either because the workings of the mind are independent of the body; or because the mind’s processes are irrelevant to those of the body. The first claim is seen arise from the difficulty of seeing the mind as part of the physical world - a difficulty removed by the arguments in the previous paragraph. The second claim gains plausibility through a mistaken adherence to certain models of scientific explanation.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal.