Machine learning algorithms may radically improve our ability to diagnose and treat disease. For moral, legal, and scientific reasons, it is essential that doctors and patients be able to understand and explain the predictions of these models. Scalable, customisable, and ethical solutions can be achieved by working together with relevant stakeholders, including patients, data scientists, and policy makers.

DCPS is a connectionist production system interpreter that uses distributed representations. As a connectionist model it consists of many simple, richly interconnected neuron‐like computing units that cooperate to solve problems in parallel. One motivation for constructing DCPS was to demonstrate that connectionist models are capable of representing and using explicit rules. A second motivation was to show how “coarse coding” or “distributed representations” can be used to construct a working memory that requires far fewer units than the number of different (...) facts that can potentially be stored. The simulation we present is intended as a detailed demonstration of the feasibility of certain ideas and should not be viewed as a full implementation of production systems. Our current model only has a few of the many interesting emergent properties that we eventually hope to demonstrate: It is damage‐resistant, it performs matching and variable binding by massively parallel constraint satisfaction, and the capacity of its working memory is dependent on the similarity of the items being stored. (shrink)

The first English-language collection devoted to Hegel’s Philosophy of Subjective Spirit.

The combination of ethnographic and cognitive neuroscience research provides considerable insight into the origin and symbolism of Native Californian rock art. Although made by different social groups for different purposes in various parts of the state, the ethnographic record demonstrates that the art depicts the mental imagery and somatic hallucinations of trance, taken to represent supernatural experiences. When this art is viewed from a cognitive neuroscience perspective, it suggests that the shamanistic state of consciousness was far from primarily "ecstatic," instead (...) often involving the generation of unpleasant emotions; and that biochemical changes during trance necessitated the making of rock art if the memory of these supernatural experiences was to be preserved. (shrink)

Fundamental questions in evolution concern deep divisions in the living world and vertical versus horizontal information transfer. Two contrasting views are: (i) three superkingdoms Archaea, Eubacteria, and Eukarya based on vertical inheritance of genes encoding ribosomes; versus (ii) a prokaryotic/eukaryotic dichotomy with unconstrained horizontal gene transfer (HGT) among prokaryotes. Vertical inheritance implies continuity of cytoplasmic and structural information whereas HGT transfers only DNA. By hypothesis, HGT of the translation machinery is constrained by interaction between new ribosomal gene products and vertically (...) inherited cytoplasmic structure made largely of preexisting ribosomes. Ribosomes differentially enhance the assembly of new ribosomes made from closely related genes and inhibit the assembly of products from more distal genes. This hypothesis suggests experiments for synthetic biology: the ability of synthetic genomes to “boot,” i.e., establish hereditary continuity, will be constrained by the phylogenetic closeness of the cell “body” into which genomes are placed. (shrink)

Physiological investigation has demonstrated that the central nervous system monitors body composition and adjusts energy intake and expenditure to stabilize total adipose tissue mass. Genetic variations in the signalling molecules involved in this regulatory system account for the heritable component of body fat content. The application of molecular techniques to rodent models of Mendelian obesity has resulted in the characterization of five loci at which mutations produce an abnormal accumulation of body fat. The genes at these loci include agouti, which (...) encodes a molecule that antagonizes the binding of alpha melanocyte‐stimulating hormone to its receptor; fat, which encodes carboxypeptidase E; tubby, which encodes a putative phosphodiesterase; obese, which encodes a circulating satiety protein; and diabetes, which encodes the receptor for the obese gene product. A more detailed understanding of the functional interrelationships of these genes should lead to important new insights into the causes and potential therapies for human obesity. (shrink)

The shape and turnover of photoreceptor membranes appears to depend on associated actin filaments. In dipterans, the photoreceptor membrane is microvillar. It is turned over by the addition of new membrane at the bases of the microvilli and by subsequent shedding, mostly from the distal ends. Each microvillus contains actin filaments as a component of its cytoskeletal core. Two myosin I‐like proteins co‐localize with the actin filaments. It is suggested that one of the myosin I‐like proteins might be linked to (...) the microvillar membrane. By interacting with the actin filaments, this motor should move the membrane of a microvillus in a distal direction, thus providing a possible mechanism for the turnover of the membrane.A vertebrate photoreceptor cell contains a small cluster of actin filaments in its connecting cilium at the site where new transductive disk membranes are formed. Disruption of the actin filaments perturbs disk morphogenesis. The most likely explanation for this perturbation is that the process of initiating a new disk is inhibited. Conventional myosin (myosin II) is found in the connecting cilium with the same distribution as actin. A simple model is proposed to illustrate how the actin–myosin system of the connecting cilium might function to initiate the morphogenesis of a disk membrane. (shrink)

I contend that though Pippin argues persuasively that the most trenchant critics of modernity ? Nietzsche and Heidegger ? failed to break free of modernity, it remains unclear whether the unending character of modernity is thereby defended as an unfinished project, or rather is a fate to which we are condemned. The thesis of my paper is that the ambiguity is traceable to Pippin's Hegelian narrative of legitimation. On his view, the demands of modern self?consciousness inaugurated by Kant and radicalized (...) by Hegel prohibit every appeal to a given, even where the given is construed as contingent historical shifts, or perspectives, or Ereignisse. But this rejection of appeals to something outside the narrative of legitimation implies that even the significance of the narrative can only be an effect of the narrative itself. But if this is true, there is on Pippin's own terms no way of establishing the significance of a narrative about modernity and hence no reason for finding it compelling. Thus the very radicalization of the Kantian critique of dogmatism that Pippin claims is the hallmark of modernity renders unavailable the legitimation of modernity. (shrink)

We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealised explanation game in which players collaborate to find the best explanation for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping causal (...) patterns of variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions. (shrink)

We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealisedexplanation gamein which players collaborate to find the best explanation(s) for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping causal patterns of (...) variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a (conditionally) optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions. (shrink)

As machine learning has gradually entered into ever more sectors of public and private life, there has been a growing demand for algorithmic explainability. How can we make the predictions of complex statistical models more intelligible to end users? A subdiscipline of computer science known as interpretable machine learning (IML) has emerged to address this urgent question. Numerous influential methods have been proposed, from local linear approximations to rule lists and counterfactuals. In this article, I highlight three conceptual challenges that (...) are largely overlooked by authors in this area. I argue that the vast majority of IML algorithms are plagued by (1) ambiguity with respect to their true target; (2) a disregard for error rates and severe testing; and (3) an emphasis on product over process. Each point is developed at length, drawing on relevant debates in epistemology and philosophy of science. Examples and counterexamples from IML are considered, demonstrating how failure to acknowledge these problems can result in counterintuitive and potentially misleading explanations. Without greater care for the conceptual foundations of IML, future work in this area is doomed to repeat the same mistakes. (shrink)

We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealised explanation game in which players collaborate to find the best explanation for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping causal (...) patterns of variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions. (shrink)

In response to recent corporate ethical and financial disasters there has been increased pressure on business schools to improve their teaching of corporate ethics. Accreditation bodies, such as the Association to Advance Collegiate Schools of Business (AACSB), now require member institutions to develop the ethical awareness of business students, either through a dedicated subject or an integrated coverage of ethics across the curriculum. This paper describes an institutional approach to the incorporation of a comprehensive multi-disciplinary ethics framework into the business (...) curriculum. We discuss important implications for the assessment of ethics within institutional assurance practices, and address critical issues related to the support of academics whenrequired to incorporate new ethics material within their subject which may be outside their field of expertise. As an example, the successful application of the framework within the marketing discipline is presented and discussed. (shrink)

Necessity and sufficiency are the building blocks of all successful explanations. Yet despite their importance, these notions have been conceptually underdeveloped and inconsistently applied in explainable artificial intelligence, a fast-growing research area that is so far lacking in firm theoretical foundations. In this article, an expanded version of a paper originally presented at the 37th Conference on Uncertainty in Artificial Intelligence, we attempt to fill this gap. Building on work in logic, probability, and causality, we establish the central role of (...) necessity and sufficiency in XAI, unifying seemingly disparate methods in a single formal framework. We propose a novel formulation of these concepts, and demonstrate its advantages over leading alternatives. We present a sound and complete algorithm for computing explanatory factors with respect to a given context and set of agentive preferences, allowing users to identify necessary and sufficient conditions for desired outcomes at minimal cost. Experiments on real and simulated data confirm our method’s competitive performance against state of the art XAI tools on a diverse array of tasks. (shrink)

Unsupervised learning algorithms are widely used for many important statistical tasks with numerous applications in science and industry. Yet despite their prevalence, they have attracted remarkably little philosophical scrutiny to date. This stands in stark contrast to supervised and reinforcement learning algorithms, which have been widely studied and critically evaluated, often with an emphasis on ethical concerns. In this article, I analyze three canonical unsupervised learning problems: clustering, abstraction, and generative modeling. I argue that these methods raise unique epistemological and (...) ontological questions, providing data-driven tools for discovering natural kinds and distinguishing essence from contingency. This analysis goes some way toward filling the lacuna in contemporary philosophical discourse on unsupervised learning, as well as bringing conceptual unity to a heterogeneous field more often described by what it isnot(i.e., supervised or reinforcement learning) than by what itis. I submit that unsupervised learning is not just a legitimate subject of philosophical inquiry but perhaps the most fundamental branch of all AI. However, an uncritical overreliance on unsupervised methods poses major epistemic and ethical risks. I conclude by advocating for a pragmatic, error-statistical approach that embraces the opportunities and mitigates the challenges posed by this powerful class of algorithms. (shrink)

From Kierkegaard’s famous polemic against Hegel’s system, and Marx’s rejection of the “mysticism” of reason, to Heidegger’s claim that Hegel completes the tradition of western metaphysics, and contemporary critics’ identification of Hegel as the authoritative spokesman — the “Master” — for the principles of unity and identity, a standard view has governed interpretations and evaluations of Hegel’s philosophy. Though familiarity with the positions just cited reveals considerable disparity, one does not need an especially discerning eye to recognize the common features (...) of the figures drawn. Hegel is, in all of them, the philosopher of the Absolute, the proponent of metaphysics without equal in the modern world. (shrink)

_The first English-language collection devoted to Hegel’s_ Philosophy of Subjective Spirit.

_Real Essentialism_ presents a comprehensive defence of neo-Aristotelian essentialism. Do objects have essences? Must they be the kinds of things they are in spite of the changes they undergo? Can we know what things are really like – can we define and classify reality? Many, if not most, philosophers doubt this, influenced by centuries of empiricism, and by the anti-essentialism of Wittgenstein, Quine, Popper, and other thinkers. _Real Essentialism_ reinvigorates the tradition of realist, essentialist metaphysics, defending the reality and knowability (...) of essence, the possibility of objective, immutable definition, and its relevance to contemporary scientific and metaphysical issues such as whether essence transcends physics and chemistry, the essence of life, the nature of biological species, and the nature of the person. (shrink)

Political advertising has long been a target for criticism regarding unethical behaviour. This study looks at the attitudes of Australian advertising agency executives and politicians towards ethical issues relating to political advertising. A sample of 101 advertising agency executives and 46 federal politicians were compared and some attitudinal differences were found, which could be areas of tension in the agency-client relationship.

This paper tests the following hypothesis: that the prime matter of classical Aristotelian-Scholastic metaphysics is numerically identical to energy. Is P=E? After outlining the classical Aristotelian concept of prime matter, I provide the master argument for it based on the phenomenon of substantial change. I then outline what we know about energy as a scientific concept, including its role and application in some key fields. Next, I consider the arguments in favour of prime matter being identical to energy, followed by (...) the arguments against. The method used is that of ontological profile comparison: does the profile of prime matter match, in key features, that of energy? An affirmative answer, that P=E, would be a momentous discovery: it would show that one of the most neglected and derided ideas of pre-modern metaphysics – a contributor to its downfall in the wake of the Scientific Revolution – was right all along. From a negative answer we would still learn much about the interaction of science and metaphysics. It turns out, however, given what we currently know, that the answer is not quite as simple as one might hope. (shrink)

Applied Ethics focuses the central concepts of traditional morality from the companion book Moral Theory - rights, justice, the good, virtue, and the fundamental value of human life - on a number of pressing contemporary problems, including abortion, euthanasia, animals, capital punishment, and war.

We examine two different descriptions of the behavioral functions of the hippocampal system. One emphasizes spatially organized behaviors, especially those using cognitive maps. The other emphasizes memory, particularly working memory, a short-term memory that requires iexible stimulus-response associations and is highly susceptible to interference. The predictive value of the spatial and memory descriptions were evaluated by testing rats with damage to the hippocampal system in a series of experiments, independently manipulating the spatial and memory characteristics of a behavioral task. No (...) dissociations were found when the spatial characteristics of the stimuli to be remembered were changed; lesions produced a similar deficit in both spatial and nonspatial test procedures, indicating that the hippocampus was similarly involved regardless of the spatial nature of the task. In contrast, a marked dissociation was found when the memory requirements were altered. Rats with lesions were able to perform accurately in tasks that could be solved exclusively on the basis of reference memory. They performed at chance levels and showed no signs of recovery even with extensive postoperative training in tasks that required working memory. In one experiment all the characteristics of the reference memory and working memory procedures were identical except the type of memory required. Consequently, the behavioral dissociation cannot be explained by differences in attention, motivation, response inhibition, or the type of stimuli to be remembered. As a result of these experiments we propose that the hippocampus is selectively involved in behaviors that require working memory, irrespective of the type of material that is to be processed by that memory. (shrink)

The question whether two things can be in the same place at the same time is an ambiguous one. At least three distinct questions could be meant: Can two things simpliciter be in the same place at the same time? Can two things of the same kind be in the same place at the same time? Can two substances of the same kind be in the same place at the same time? The answers to these questions vary. In what follows, (...) all three will be discussed in the light both of recent and of classic earlier examinations of the matter. The following theses will be defended. (shrink)

Whereas Mereological Essentialism is the thesis that the parts of an object are essential to it, Reverse Mereological Essentialism is the thesis that the whole is essential to its parts. Specifically – since RME is an Aristotelian doctrine – it is a claim not about objects in general but about substances. Here I set out and explain RME as it should be understood from the perspective of the Aristotelian-Scholastic tradition, as well as proposing a kind of master argument for believing (...) it. A number of objections (many of which have been raised by Kathrin Koslicki or Robert Koons) are then considered, the replies to which help further to clarify and motivate RME. The final section considers some important questions concerning parts and matter in light of Ross Inman’s recent defence of RME under the guise of what he calls Substantial Priority. Considering these questions further illustrates right and wrong ways of understanding RME. Overall, the case for RME is strong albeit with a number of difficulties that need to be resolved through further investigation. (shrink)

The old scholastic principle of the "convertibility" of being and goodness strikes nearly all moderns as either barely comprehensible or plain false. "Convertible" is a term of art meaning "interchangeable" in respect of predication, where the predicates can be exchanged salva veritate albeit not salva sensu: their referents are, as the maxim goes, really the same albeit conceptually different. The principle seems, at first blush, absurd. Did the scholastics literally mean that every being is good? Is that supposed to include (...) a cancer, a malaria parasite, an earthquake that kills millions? If every being is good, then no being is bad—but how can that be? To the contemporary philosophical mind, such bafflement is understandable. It derives from the systematic dismantling of the great scholastic edifice that took place over half a millennium. With the loss of the basic concepts out of which that edifice was built, the space created by those concepts faded out of existence as well. The convertibility principle, like virtually all the other scholastic principles (not all, since some do survive and thrive in analytic philosophy), could not persist in a post-scholastic space wholly alien to it. (shrink)