Explanation-Aware Computing: Papers from the 2007 AAAI Workshop, Association for the Advancement of Artificial Intelligence, Technical Report WS-07-06, Menlo Park California, AAAI Press, 2007, 1-9.

In this article, I consider an important challenge to the popular theory of scientific inference commonly known as ‘inference to the best explanation’, one that has received scant attention.1 1 The problem is that there exists a wide array of rival models of explanation, thus leaving IBE objectionably indeterminate. First, I briefly introduce IBE. Then, I motivate the problem and offer three potential solutions, the most plausible of which is to adopt a kind of pluralism about the (...) rival models of explanation. However, I argue that how ranking explanations on this pluralistic account of IBE remains obscure and pluralism leads to contradictory results. In light of these objections, I attempt to dissolve the problem by showing why IBE does not require a ‘model’ of explanation and by giving an account of what explanation consists in within the context of IBE. (shrink)

Explanation-Aware Computing: Papers from the 2007 AAAI Workshop, Association for the Advancement of Artificial Intelligence, Technical Report WS-07-06, Menlo Park California, AAAI Press, 2007, 1-9.

A key theme in the historiographical work of Machamer has been the ways that motion is made intelligible through explanatory means of natural motion and models of the simple machines such as the lever and pendulum. One way of spelling out the explanatory value of these strategies is through the concept of equilibrium. Natural motion and simple machines allow the simplification of complex problems in terms of self-evident, intelligible equilibrium conditions. This chapter connects the theme of equilibrium and natural (...) motion across Machamer’s work on mechanisms and mechanical explanation, on Aristotle, Galileo, Descartes and Newton, and on the pendulum and mental models. Just as equilibrium can be found within science, it also becomes a model of intelligibility for doing history and philosophy of science: a normative but objective representation of the important properties of science. (shrink)

This paper is an assessment of an attempt, by James Greeno, to measure the explanatory power of statistical theories by means of the notion of transmitted information (It). It is argued that It has certain features that are inappropriate in a measure of explanatory power. In particular, given a statistical theory T with explanans variables St and explanandum variables Mj, it is argued that no plausible measure of explanatory power should depend on the probability P(Si) of occurrence of initial conditions (...) in the systems to which T applies or the magnitudes of the conditional probabilities P(Mj/Si), in the manner in which Ir does. (shrink)

In this article, I explore the compatibility of inference to the best explanation (IBE) with several influential models and accounts of scientific explanation. First, I explore the different conceptions of IBE and limit my discussion to two: the heuristic conception and the objective Bayesian conception. Next, I discuss five models of scientific explanation with regard to each model’s compatibility with IBE. I argue that Kitcher’s unificationist account supports IBE; Railton’s deductive–nomological–probabilistic model, Salmon’s statistical-relevance model, and (...) van Fraassen’s erotetic account are incompatible with IBE; and Salmon’s causal–mechanical model is merely consistent with IBE. In short, many influential models of scientific explanation do not support IBE. I end by outlining three possible conclusions to draw: (1) either philosophers of science or defenders of IBE have seriously misconstrued the concept of explanation, (2) philosophers of science and defenders of IBE do not use the term ‘explanation’ univocally, and (3) the ampliative conception of IBE, which is compatible with any model of scientific explanation, deserves a closer look. (shrink)

It has been the dominant view that probabilistic explanations of particular facts must be inductive in character. I argue here that this view is mistaken, and that the aim of probabilistic explanation is not to demonstrate that the explanandum fact was nomically expectable, but to give an account of the chance mechanism(s) responsible for it. To this end, a deductive-nomological model of probabilistic explanation is developed and defended. Such a model has application only when the probabilities occurring in (...) covering laws can be interpreted as measures of objective chance, expressing the strength of physical propensities. Unlike inductive models of probabilistic explanation, this deductive model stands in no need of troublesome requirements of maximal specificity or epistemic relativization. (shrink)

The controversy about intentional explanation of action concerns how these explanations work. What kind of model allows us to capture the dependency or relevance relation between the explanans, i.e. the beliefs and desires of the agent, and the explanandum, i.e. the action? In this paper, I argue that the causal mechanical model can do the job. Causal mechanical intentional explanations consist in a reference to the mechanisms of practical reasoning of the agent that motivated the agent to act, i.e. (...) to a causally relevant set of beliefs and desires. Moreover, the causal mechanical model can provide in efficient and unproblematic applications, unlike action explanations using ceteris paribus laws or counterfactuals. The drawback of the latter models of explanation is their modal requirement: the explanans must mention or implies sufficient and/or necessary conditions for the explanandum. Such a requirement is too strong when it comes to intentional explanation of action. (shrink)

The legitimacy of going back to the classical view of God’s action in the world based on the list of causes and understanding of chance in the works of Aristotle and Aquinas – in the context of contemporary science – seems to depend on whether there is a space for causal analysis within the current models of scientific explanation. This article offers a brief account of the path leading to negation and rediscovery of the importance of causality in (...) scientific explanation and reintroduces the semicausal position of the prominent philosopher of science, Mario Bunge, who treats causation as one of several categories of determination. The diversity of the categories he lists finds analogy in the commonly accepted pluralist approach to the search of the model which adequately describes the practice of scientific research. What is more, the same diversity of the categories of determination opens the way back to the classical Aristotle’s fourfold account of causation and his understanding of chance. This fact allows us, in turn, to defend the contemporary version of the classical notion of divine action against the accusation of methodical error in the form of imposing the notion of the ancient categories of causality on the results of contemporary scientific research, which notion, as some maintain, has little in common with the models of explanation currently accepted in natural sciences. (shrink)

This paper proposes a formal top-level model of explaining the outputs of machine-learning-based decision-making applications and evaluates it experimentally with three data sets. The model draws on AI & law research on argumentation with cases, which models how lawyers draw analogies to past cases and discuss their relevant similarities and differences in terms of relevant factors and dimensions in the problem domain. A case-based approach is natural since the input data of machine-learning applications can be seen as cases. While (...) the approach is motivated by legal decision making, it also applies to other kinds of decision making, such as commercial decisions about loan applications or employee hiring, as long as the outcome is binary and the input conforms to this paper’s factor- or dimension format. The model is top-level in that it can be extended with more refined accounts of similarities and differences between cases. It is shown to overcome several limitations of similar argumentation-based explanation models, which only have binary features and do not represent the tendency of features towards particular outcomes. The results of the experimental evaluation studies indicate that the model may be feasible in practice, but that further development and experimentation is needed to confirm its usefulness as an explanation model. Main challenges here are selecting from a large number of possible explanations, reducing the number of features in the explanations and adding more meaningful information to them. It also remains to be investigated how suitable our approach is for explaining non-linear models. (shrink)

Peter Railton (1978) has introduced the influential deductive-nomological-probabilistic (DNP) model of explanation which is the culmination of a tradition of formal, non-pragmatic accounts of scientific explanation. The other models in this tradition have been shown to be susceptible to a class of counterexamples involving intervening causes which speak against their sufficiency. This treatment has never been extended to the DNP model; we contend that the usual form of these counterexamples is ineffective in this case. However, we develop (...) below a new version which overcomes these difficulties. Thus we claim that all of the models in this tradition, DNP included, have an equal status with respect to sufficiency. (shrink)

Analytic and continental philosophies of mind are too long divided. In both traditions there is extensive discussion of consciousness, the mind-body problem, intentionality, subjectivity, perception (especially visual) and so on. Between these two discussions there are substantive disagreements, overlapping points of insight, meaningful differences in emphasis, and points of comparison which seems to offer nothing but confusion. In other words, there are the ideal circumstances for doing philosophy. Yet, there has been little discourse. This paper invites expanding discourse between these (...) two philosophical traditions. The first part briefly describes the existing literature which works across the analytic- phenomenology divide, situating my work within it as a focus on analytic physicalism and phenomenal explanation. In the longer second part, I sketch a model for explanation embedded simultaneously in both traditions. Hopefully, a theoretical framework emerges that the unlikely combination of Maurice Merleau- Ponty and Patricia Churchland could accept. In the third part, I apply the three-tiered model to a discussion of plasticity and suggest that the model both reflects existing research across three levels of analysis and can be a fruitful way to approach future research. My suggestion for a three-tiered model is quite tentative. Much less tentative is my claim that constructive dialogue between phenomeno- logical and physicalist study of consciousness is long-overdue, illuminating, and practical. (shrink)

We apply three state‐of‐the‐art models for syllogistic reasoning to data from experiments where participants received feedback for their conclusions in order to demonstrate the use of model parameters to derive new hypotheses and present possible explanations for the feedback effect.

Among the many functions of models, explanation is central to the functioning and aims of science. However, the discussions surrounding modeling and explanation in philosophy have largely remained separate from each other. This chapter seeks to bridge the gap by focusing on the puzzle of model-based explanation, asking how different philosophical accounts answer the following question: if idealizations and fictions introduce falsehoods into models, how can idealized and fictional models provide true explanations? The chapter (...) provides a selective and critical overview of the available strategies for solving this puzzle, mainly focusing on idealized models and how they explain. (shrink)

It is argued that the constraints placed on the non-law premisses of a D–N explanation are irrelevant to their function and will not salvage the deductive requirement from triviality.

Detailed examinations of scientific practice have revealed that the use of idealized models in the sciences is pervasive. These models play a central role in not only the investigation and prediction of phenomena, but in their received scientific explanations as well. This has led philosophers of science to begin revising the traditional philosophical accounts of scientific explanation in order to make sense of this practice. These new model-based accounts of scientific explanation, however, raise a number of (...) key questions: Can the fictions and falsehoods inherent in the modeling practice do real explanatory work? Do some highly abstract and mathematical models exhibit a noncausal form of scientific explanation? How can one distinguish an exploratory "how-possibly" model explanation from a genuine "how-actually" model explanation? Do modelers face tradeoffs such that a model that is optimized for yielding explanatory insight, for example, might fail to be the most predictively accurate, and vice versa? This chapter explores the various answers that have been given to these questions. (shrink)

During the 1930s and 1940s, American physical organic chemists employed electronic theories of reaction mechanisms to construct models offering explanations of organic reactions. But two molecular rearrangements presented enormous challenges to model construction. The Claisen and Cope rearrangements were predominantly inaccessible to experimental investigation and they confounded explanation in theoretical terms. Drawing on the idea that models can be autonomous agents in the production of scientific knowledge, I argue that one group of models in particular were (...) functionally autonomous from the Hughes–Ingold theory. Cope and Hardy’s models of the Claisen and Cope rearrangements were resources for the exploration of the Hughes–Ingold theory that otherwise lacked explanatory power. By generating ‘how-possibly’ explanations, these models explained how these rearrangements could happen rather than why they did happen. Furthermore, although these models were apparently closely connected to theory in terms of their construction, I argue that partial autonomy issued in extra-logical factors concerning the attitudes of American chemists to the Hughes–Ingold theory. And in the absence of a complete theoretical hegemony, a degree of consensus was reached concerning modelling the Claisen rearrangement mechanism.Keywords: Models; Explanation; Physical organic chemistry; Rearrangements. (shrink)

This paper discusses the D-N model of scientific explanation. It is suggested that explanation is a part of assertive discourse where certain principles must be observed. Then use is made of the relation between the informative content and logical content of a sentence (as shown, for instance, by Popper) to draw some of the conditions necessary for a sound model. It is claimed that the conditions of the model proposed in the present paper exhaust the insights of the (...) papers in the literature, solve the difficulties encountered by other authors, but have some damaging consequences on the D-N model of scientific explanation. (shrink)

Carl Craver proposes a mechanistic model of explanation in science motivated by a desire to intervene, as exemplified by explanations in neuroscience which in his opinion are motivated by the desire to bring the central nervous system under control. In his discussion of causal relevancy conditions of mechanistic components Craver asserts that a cause need not make its effect probable in order to explain it . Although this is supported by some interpretations, Craver’s own is highlighted by his appeal (...) to an example in neuroscience of the apparently stochastic nature of neurotransmitter release events. I propose the view that this interpretation of such a causal relevancy condition is contradictory to Craver’s own proposed first norm of explanation : that a good explanation must fully account for the explanandum phenomenon including varied manifestations of the phenomenon. By defending Craver’s explanatory motivation of intervention I demonstrate that the attributed cause of a phenomenon must necessarily make the phenomenon probable in order to effectively explain it. (shrink)

Our present models and theories of African history and prehistory are profoundly influenced by the physical anthropologists’ perceptions of human reality in present-day Africa. Professor P. V. Tobias has suggested that the present-day people of Africa, excluding the recent arrivals from Europe and Asia, descended from a common proto-Negriform stock which gave birth first to the so-called “Khoisan” (I am using here the terminology of my source, not the historically justified Khoe and San, meaning the Hottentots and the Bushmen) (...) and later to the Negro genetic type. Of greater importance to archaeologists, who deal mainly with the so-called pre-history, is the fact that, genetically speaking, the difference between the “Khoisan” and the Negroids is smaller than the difference between these two African racial types and all other non-African types. This realisation tends to have one consequence: the pre-historic and even historic developments are generally perceived as explainable from within Africa, without any need for external clues or explanations. Archaeological finds are thus taken for evidence of a basically internal evolution and the finds of foreign glass beads, though very frequent and spread all over Africa, are seen as belonging to the same category as Persian, Chinese, or Arab ceramics and are explained as indication of a coastal trade. (shrink)

This article defends the principle of sufficient reason (PSR) from a simple and direct valid argument according to which PSR implies that there is a truth that explains every truth, namely an omni-explainer. Many proponents of PSR may be willing to bite the bullet and maintain that, if PSR is true, then there is an omni-explainer. I object to this strategy by defending the principle that explanation is irreflexive. Then I argue that proponents of PSR can resist the conclusion (...) that there is an omni-explainer by denying that explanation distributes over conjunction. In the final section, I consider a plausible revision of this assumption. I argue there that, given the revised assumption, PSR seems to yield a striking picture of the explanatory structure of the universe. I explain why the resulting model does not appear utterly implausible to me. (shrink)

The belief-desire model of action explanation is deeply ingrained in multiple disciplines. There is reason to think that it is a cultural artifact. But is there an alternative? In this discussion, I will consider the radical critique of this action explanation model by Rüdiger Bittner, which argues that the model appeals to dubious mental entities, and argues for a model of reasons as responses to states or events. Instead, for Bittner, agents are reason-selectors—selecting the states or events to (...) respond to and selecting the ones the agent is disposed to respond to. By getting rid of the explanatory role of beliefs, this model runs into difficulties over errors usually attributed to false beliefs. These can be resolved by expanding the notion of dispositions to cover the case of false belief. But this suggests that the belief-reason model serves to divide the category of dispositions in an arbitrary or culturally specific way. (shrink)

Words are the essence of communication: They are the building blocks of any language. Learning the meaning of words is thus one of the most important aspects of language acquisition: Children must first learn words before they can combine them into complex utterances. Many theories have been developed to explain the impressive efficiency of young children in acquiring the vocabulary of their language, as well as the developmental patterns observed in the course of lexical acquisition. A major source of disagreement (...) among the different theories is whether children are equipped with special mechanisms and biases for word learning, or their general cognitive abilities are adequate for the task. We present a novel computational model of early word learning to shed light on the mechanisms that might be at work in this process. The model learns word meanings as probabilistic associations between words and semantic elements, using an incremental and probabilistic learning mechanism, and drawing only on general cognitive abilities. The results presented here demonstrate that much about word meanings can be learned from naturally occurring child-directed utterances (paired with meaning representations), without using any special biases or constraints, and without any explicit developmental changes in the underlying learning mechanism. Furthermore, our model provides explanations for the occasionally contradictory child experimental data, and offers predictions for the behavior of young word learners in novel situations. (shrink)

Agent-based modelling has become a well-established method in social epistemology and philosophy of science but the question of what kind of explanations these models provide remains largely open. This paper is dedicated to this issue. It starts by distinguishing between real-world phenomena, real-world possibilities, and logical possibilities as different kinds of targets which agent-based models can represent. I argue that models representing the former two kinds provide how-actually explanations or causal how-possibly explanations. In contrast, models that (...) represent logical possibilities provide epistemically opaque how-possibly explanations. While highly idealised ABMs in the form in which they are initially proposed typically fall into the last category, the epistemic opaqueness of explanations they provide can be reduced by validation procedures. To this purpose, an examination of results of simulations in terms of classes of models can be particularly helpful. I illustrate this point by discussing a class of ABMs of scientific interaction and the claim that a high degree of interaction can impede scientific inquiry. (shrink)

This paper presents a new formal model for D–N explanation that gives intuitive criteria of acceptability, avoids the known trivializations, and links explanation with confirmation theory. Although set in the twenty-five year tradition of attempts to formalize D–N explanation, it proposes a new direction for the model that is to be distinguished from the syntactical and informational approaches by its introduction of restrictions which derive from the use which the D–N model can have in hypothesis testing. This (...) model, illustrating the verificational approach, revises the classic H–O requirements and amends the notion of partial self-explanation to meet a criticism to which the H–O notion is vulnerable. (shrink)

A procedure of historical explanation is proposed which integrates two approaches used by contemporary historians. The motivational model, focusing on the various kinds of motives encountered in historical narratives, and the deductive- nomological model, which focuses on the importance of external events, can be linked together to yield a better integrated explanatory system. The two approaches can be bridged by establishing even more general laws underlying ones already applied, or by searching for substantiations of causes and laws in the (...) origin of the entire historical process. On the basis of analysis of actual historiographical problems it becomes clear that historians mingle explanations in terms of human motives and external occurrences, without making the relations between them sufficiently clear. One needs to recognize that both motives and external events can be found in the historical process. Sequences of causes and effects in narratives have as an essential feature the intertwining of causes interpreted as motives and causes interpreted as external events. The objective conditions explained by the deductive-nomological model should find their reflection in the motivational model in the way that agents take cognizance of these conditions. (shrink)