Theories of Explanation

In a recent paper, Erasmus et al. (2021) defend the idea that the ambiguity of the term “explanation” in explainable AI (XAI) can be solved by adopting any of four different extant accounts of explanation in the philosophy of science: the Deductive Nomological, Inductive Statistical, Causal Mechanical, and New Mechanist models. In this chapter, I show that the authors’ claim that these accounts can be applied to deep neural networks as they would to any natural phenomenon is mistaken. I also (...) provide a more general argument as to why the notion of explainability as it is currently used in the XAI literature bears little resemblance to the traditional concept of scientific explanation. It would be more fruitful to use the label “understandable AI” to avoid the confusion that surrounds the goal and purposes of XAI. In the second half of the chapter, I argue for a pragmatic conception of understanding that is better suited to play the central role attributed to explanation in XAI. Following De Regt (2017) and Kuorikoski and Ylikoski (2015), the conditions of satisfaction for understanding an ML system are fleshed out in terms of an agent’s success in using the system. (shrink)

Distinctively mathematical explanations (DMEs) explain natural phenomena primarily by appeal to mathematical facts. One important question is whether there can be an ontic account of DME. An ontic account of DME would treat the explananda and explanantia of DMEs as ontic structures and the explanatory relation between them as an ontic relation (e.g., Pincock 2015, Povich 2021). Here I present a conventionalist account of DME, defend it against objections, and argue that it should be considered ontic. Notably, if indeed it (...) is ontic, the conventionalist account seems to avoid a convincing objection to other ontic accounts (Kuorikoski 2021). (shrink)

Ground is all the rage in contemporary metaphysics. But what is its nature? Some metaphysicians defend what we could call, following Skiles and Trogdon (2021), the inheritance view: it is because constitutive forms of metaphysical explanation are such-and-such that we should believe that ground is so-and-so. However, many putative instances of inheritance are not primarily motivated by scientific considerations. This limitation is harmless if one thinks that ground and science are best kept apart. Contrary to this view, we believe that (...) ground is a highly serviceable tool for investigating metaphysical areas of science. In this paper, we defend a naturalistic version of the inheritance view which takes constitutive scientific explanation as a better guide to ground. After illustrating the approach and its merits, we discuss some implications of the emerging scientific conception for the theory of ground at large. (shrink)

One assumption of the new mechanistic approach is that there are two kinds of mechanistic explanations: etiological and constitutive ones. While the former explain phenomena in terms of their preceding causes, the latter are supposed to refer to mechanisms that constitute phenomena. Based on arguments by Kaiser and Krickel (Br J Philos Sci 68(3):745–779, 2017) and Krickel (The mechanical world, vol. 13, Springer International Publishing, 2018), I will show that this view is too narrow. Indeed, three different types of explanation (...) are usually subsumed under the label “constitutive explanation”. However, one of those types of explanation is not a version of constitutive explanation. Rather it is a variant of etiological explanation. As a result, I will show that there are four types of mechanistic explanation, two variants of etiological explanation—which I will call output mechanistic explanations and input-output mechanistic explanations—and two variants of constitutive explanation—which I will call filler mechanistic explanations and dimensioned mechanistic explanations. Keeping these apart is crucial as they come with different ontological implications. An evaluation of the mechanistic approach regarding its stance on reduction, levels, and interlevel causation crucially depends on which notion of mechanistic explanation one has in mind. (shrink)

There has been a push in recent years to provide better explanations for how AIs make their decisions. Most of this push has come from the ethical concerns that go hand in hand with AIs making decisions that affect humans. Outside of the strictly ethical concerns that have prompted the study of explainable AIs (XAIs), there has been research interest in the mere possibility of creating XAIs in various domains. In general, the more accurate we make our models the harder (...) they are to explain. Go playing AIs like AlphaGo and KataGo provide fantastic examples of this phenomenon. In this paper, I discuss a non-exhaustive list of the leading theories of explanation and what each of these theories would say about the explainability of AI-played moves of Go. Finally, I consider the possibility of ever explaining AI-played Go moves in a way that meets the four principles of XAI. I conclude, somewhat pessimistically, that Go is not as imminently explainable as other domains. As such, the probability of having an XAI for Go that meets the four principles is low. (shrink)

Recent years have seen growing interest in modifying interventionist accounts of causal explanation in order to characterise noncausal explanation. However, one surprising element of such accounts is that they have typically jettisoned the core feature of interventionism: interventions. Indeed, the prevailing opinion within the philosophy of science literature suggests that interventions exclusively demarcate causal relationships. This position is so prevalent that, until now, no one has even thought to name it. We call it “intervention puritanism” (I-puritanism, for short). In this (...) paper, we mount the first sustained defence of the idea that there are distinctively noncausal explanations which can be characterized in terms of possible interventions; a position we call “intervention liberalism” (I-liberalism, for short). While many have followed James Woodward (2003) in committing to I-puritanism, we trace support for I-liberalism back to the work of Jaegwon Kim (1974). Furthermore, we analyse two recent sources of scepticism regarding I-liberalism: debate surrounding mechanistic constitution; and attempts to provide a monistic account of explanation. We show that neither literature provides compelling reasons for adopting I-puritanism. Finally, we present a novel taxonomy of available positions upon the role of possible interventions in explanation: weak causal imperialism; strong causal imperialism; monist intervention puritanism; pluralist intervention puritanism; monist intervention liberalism; and finally, the specific position defended in this paper, pluralist intervention liberalism. (shrink)

This is the first chapter of a book that I'm writing entitled Kantsequentialism: A Morality of Ends. The chapter has six sections: (1) The Distinction between a Moral Theory and a Complete Account of Morality, (2) The Best Explanation, (3) Fitting the Data as Opposed to the Facts, (4) Epistemic Justification and Phenomenal Conservatism, (5) Neither Overfitting nor Underfitting the Data, and (6) Trusting Our Moral Intuitions. Thus, the chapter begins by providing an account of what a moral theory is (...) and how that differs from a complete account of morality. It then argues that there are only four explanatory virtues: modesty, coherence, unifying power, and predictive power. Thus, it argues that simplicity is not itself a virtue. It explains how it's important for a theory to avoid either overfitting or underfitting the data. The chapter then explains that the data for our philosophical theories are our intuitions (which are intellectual seemings), and it defends phenomenal conservatism -- the view that we are justified in believing that things are as they seem unless and until grounds for doubt arise. Lastly, it responds to worries about the trustworthiness of our moral intuitions stemming from both experimental philosophy and evolutionary debunking arguments. (shrink)

It does not seem particularly daring to say that one objective of science education is to enable students to understand different phenomena in the world in their mutual relationship. This is roughly equivalent to promoting knowledge of scientific explanations, which involve resorting to regular relationships between certain phenomena and which, certainly, is different from knowledge of this or that type of event taken in isolation. In this text, we will draw attention to two opposing tendencies that, however, tend towards the (...) common result of hindering the achievement of such an objective. On the one hand, we find characterizations of scientific explanation that assimilate it to argumentation and, therefore, lose sight of what specifically makes explanations explanatory; that is, they ask too little of a scientific explanation. On the other hand, those that consider paradigmatic cases of scientific explanation −such as those achieved by subsuming phenomena such as tides to Newton's laws− do not really constitute explanations but merely descriptions; it is about asking too much of an explanation here. We will therefore point out that a greater attention to the elaborations provided by the philosophy of science about the notion of explanation is crucial to better serve educational purposes. (shrink)

We develop and apply a multi-dimensional conception of explanatory depth towards a comparative analysis of inflationary and bouncing paradigms in primordial cosmology. Our analysis builds on earlier work due to Azhar and Loeb (2021) that establishes initial condition fine-tuning as a dimension of explanatory depth relevant to debates in contemporary cosmology. We propose dynamical fine-tuning and autonomy as two further dimensions of depth in the context of problems with instability and trans-Planckian modes that afflict bouncing and inflationary approaches respectively. In (...) the context of the latter issue, we argue that the recently formulated trans-Planckian censorship conjecture leads to a trade-off for inflationary models between dynamical fine-tuning and autonomy. We conclude with the suggestion that explanatory preference with regard to the different dimensions of depth is best understood in terms of differing attitudes towards heuristics for future model building. (shrink)

Park (2017, 2018, 2019) argues that Bas van Fraassen uses inference to the best explanation to defend his contextual theory of explanation. If Park is right, then van Fraassen is in trouble because he rejects IBE as a rational rule of inference. In this reply, I argue that van Fraassen does not use IBE in defending the contextual theory of explanation. I distinguish between several conceptions of IBE: heuristic IBE, objective Bayesian IBE, and ampliative IBE. I argue that van Fraassen (...) holds the ampliative conception of IBE and that his rejection of IBE concerns only ampliative IBE. I also argue that van Fraassen’s defense of the contextual theory of explanation, at best, can be interpreted as an instance of heuristic IBE, but not ampliative IBE. Therefore, I argue, Park’s criticism of van Fraassen misfires. (shrink)

In this article, I will provide a critical overview of the form of non-deductive reasoning commonly known as “Inference to the Best Explanation” (IBE). Roughly speaking, according to IBE, we ought to infer the hypothesis that provides the best explanation of our evidence. In section 2, I survey some contemporary formulations of IBE and highlight some of its putative applications. In section 3, I distinguish IBE from C.S. Peirce’s notion of abduction. After underlining some of the essential elements of IBE, (...) the rest of the entry is organized around an examination of various problems that IBE confronts, along with some extant attempts to address these problems. In section 4, I consider the question of when a fact requires an explanation, since presumably IBE applies only in cases where some explanation is called for. In section 5, I consider the difficult question of how we ought to understand IBE in light of the fact that among philosophers, there is significant disagreement about what constitutes an explanation. In section 6, I consider different strategies for justifying the truth-conduciveness of the explanatory virtues, e.g., simplicity, unification, scope, etc., criteria which play an indispensable role in any given application of IBE. In section 7, I survey some of the most recent literature on IBE, much of which consists of investigations of the status of IBE from the standpoint of the Bayesian philosophy of science. (shrink)

A traditional account of coincidences has it that two facts are coincidental whenever they are not related as cause and effect and do not have a common cause. A recent contribution by Lando : 132–151, 2017) showed that this account is mistaken. In this paper, I argue against two alternative accounts of coincidences, one suggested by Lando, and another by Bhogal : 677–694, 2020), and defend a third one in their place. In short, I propose that how explanatory links relate (...) to non-coincidental facts in explanation is what drives a wedge between coincidences and non-coincidences. This proposal is not susceptible to the worries I raise, and is more general, since it is not restricted to coincidences and non-coincidences involving physical facts. (shrink)

A Correction to this paper has been published (changed to open access).

Proponents of ontic conceptions of explanation require all explanations to be backed by causal, constitutive, or similar relations. Among their justifications is that only ontic conceptions can do justice to the ‘directionality’ of explanation, i.e., the requirement that if X explains Y , then not-Y does not explain not-X . Using topological explanations as an illustration, we argue that non-ontic conceptions of explanation have ample resources for securing the directionality of explanations. The different ways in which neuroscientists rely on multiplexes (...) involving both functional and anatomical connectivity in their topological explanations vividly illustrate why ontic considerations are frequently (if not always) irrelevant to explanatory directionality. Therefore, directionality poses no problem to non-ontic conceptions of explanation. (shrink)

Lange argues that some natural phenomena can be explained by appeal to mathematical, rather than natural, facts. In these “distinctively mathematical” explanations, the core explanatory facts are either modally stronger than facts about ordinary causal law or understood to be constitutive of the physical task or arrangement at issue. Craver and Povich argue that Lange’s account of DME fails to exclude certain “reversals”. Lange has replied that his account can avoid these directionality charges. Specifically, Lange argues that in legitimate DMEs, (...) but not in their “reversals,” the empirical fact appealed to in the explanation is “understood to be constitutive of the physical task or arrangement at issue” in the explanandum. I argue that Lange’s reply is unsatisfactory because it leaves the crucial notion of being “understood to be constitutive of the physical task or arrangement” obscure in ways that fail to block “reversals” except by an apparent ad hoc stipulation or by abandoning the reliance on understanding and instead accepting a strong realism about essence. (shrink)

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)

Over the last decades, network-based approaches have become highly popular in diverse fields of biology, including neuroscience, ecology, molecular biology and genetics. While these approaches continue to grow very rapidly, some of their conceptual and methodological aspects still require a programmatic foundation. This challenge particularly concerns the question of whether a generalized account of explanatory, organisational and descriptive levels of networks can be applied universally across biological sciences. To this end, this highly interdisciplinary theme issue focuses on the definition, motivation (...) and application of key concepts in biological network science, such as explanatory power of distinctively network explanations, network levels, and network hierarchies. (shrink)

Normative explanations, which specify why things have the normative features they do, are ubiquitous in normative theory and ordinary thought. But there is much less work on normative explanation than on scientific or metaphysical explanation. Skow (2016) argues that a complete answer to the question why some fact Q occurs consists in all of the reasons why Q occurs. This paper explores this theory as a case study of a general theory that promises to offer us a grip on normative (...) explanation which is independent of particular normative theories. I first argue that the theory doesn’t give an adequate account of certain enablers of reasons which are important in normative explanation. I then formulate and reject three responses on behalf of the theory. But I suggest that since theories of this general sort have the right kind of resources to illuminate how normative explanation might be similar to and different from explanations in other domains, they nonetheless merit further exploration by normative theorists. (shrink)

Efforts to formalize qualitative accounts of inference to the best explanation (IBE) confront two obstacles: the imprecise nature of such accounts and the unusual logical properties that explanations exhibit, such as contradiction-intolerance and irreflexivity. This paper aims to surmount these challenges by utilising a new, more precise theory that treats explanations as expressions that codify defeasible inferences. To formalise this account, we provide a sequent calculus in which IBE serves as an elimination rule for a connective that exhibits many of (...) the properties associated with the behaviour of the English expression ‘That... best explains why ... ’. We first construct a calculus that encodes these properties at the level of the turnstile, i.e. as a metalinguistic expression for classes of defeasible consequence relations. We then show how this calculus can be conservatively extended over a language that contains a best-explains-why operator. (shrink)

Brute facts are facts that don't have explanations. They are instrumental in our attempts to give accounts of other facts or phenomena, and so they play a key role in many philosophers' views about the structure of the world. This volume explores neglected questions about the nature of brute facts and their explanatory role.

According to Skow (2016, 2017), correct answers to why-questions only cite causes or grounds, but not non-accidental regularities. Accounts that cite non-accidental regularities typically confuse second-level reasons with first-level reasons. Only causes and grounds are first-level reasons why. Non-accidental regularities are second-level reasons why. I first show that Skow's arguments for the accusation of confusion depend on the independent thesis that only citations of first-level reasons why are (parts of) answers to why-questions. Then, I argue that this thesis is false. (...) Consequently, the claim that correct answers to why-questions only cite causes or grounds is refuted as well. (shrink)

This volume is a collection of original essays focusing on a wide range of topics in the History and Philosophy of Science. It is a festschrift for Peter Machamer, which includes contributions from scholars who, at one time or another, were his students. The essays bring together analyses of issues and debates spanning from early modern science and philosophy through the 21st century. Machamer’s influence is reflected in the volume’s broad range of topics. These include: underdetermination, scientific practice, scientific models, (...) mechanistic explanation in contemporary and historical science, values in science, the relationship between philosophy and psychology, experimentation, supervenience and reductionism. (shrink)

Recent attempts to reconcile the ontic and epistemic approaches to explanation propose that our best explanations simply fulfill epistemic and ontic norms simultaneously. I aim to upset this armistice. Epistemic norms of attaining general and systematic explanations are, I argue, autonomous of ontic norms: they cannot be fulfilled simultaneously or in simple conjunction with ontic norms, and plausibly have priority over them. One result is that central arguments put forth by ontic theorists against epistemic theorists are revealed as not only (...) question-begging, but ultimately self-defeating. Another result is that a more nuanced reconciliation of the epistemic and ontic views is required: we should regard good explanatory practice as a dynamic process with distinct phases of epistemic and ontic success. (shrink)

Theories of explanation seek to tell us what distinctively explanatory information is. The most ambitious ones, such as the DN-account, seek to tell us what an explanation is, tout court. Less ambitious ones, such as causal theories, restrict themselves to a particular domain of inquiry. The least ambitious theories constitute outright skepticism, holding that there is no reasonably unified phenomenon to give an account of. On these views, it is impossible to give any theories of explanation at all. I argue (...) that both the less ambitious and outright skeptical varieties are committed to a certain context-sensitivity of our explanatory discourse. And though this discourse is almost certainly context-sensitive in some respects, it does not exhibit the context-sensitivity less than fully ambitious theories are committed to. Therefore, all accounts that seek to restrict themselves in scope, including causal accounts of explanation, fail. (shrink)

There are several important criticisms against the unificationist model of scientific explanation: Unification is a broad and heterogeneous notion and it is hard to see how a model of explanation based exclusively on unification can make a distinction between genuine explanatory unification from cases of ordering or classification. Unification alone cannot solve the asymmetry and irrelevance problems. Unification and explanation pull in different directions and should be decoupled, because for good scientific explanation extra ad explanandum information is often required. I (...) am presenting a possible solution to those problems, by focusing on an often overlooked but important element of how theoretic unification is achieved—the conceptual frameworks of theories. The core conceptual assumptions behind theories are decisive for discriminating between explanatory and non-explanatory unification. The conceptual framework is also flexible enough to balance the tension between informativeness and maximum systematization in constructing explanatory inferences. A short case study of orthogenetic and Darwinian explanations in paleontology is presented as an illustration of how my addition to the unificationist model is applicable to a historical debate between rival explanations. (shrink)

We reassess Woodward’s counterfactual account of explanation in relation to regularity explananda. Woodward presents an account of causal explanation. We argue, by using an explanation of Kleiber’s law to illustrate, that the account can also cover some noncausal explanations. This leads to a tension between the two key aspects of Woodward’s account: the counterfactual aspect and the causal aspect. We explore this tension and make a case for jettisoning the causal aspect as constitutive of explanatory power in connection with regularity (...) explananda. (shrink)

A fresh, daring, and genuine alternative to the traditional story of scientific progress Explaining the world around us, and the life within it, is one of the most uniquely human drives, and the most celebrated activity of science. Good explanations are what provide accurate causal accounts of the things we wonder at, but explanation's earthly origins haven't grounded it: we have used it to account for the grandest and most wondrous mysteries in the natural world. Explanations give us a sense (...) of understanding, but an explanation that feels right doesn't mean it is true. For every true explanation, there is a false one that feels just as good. A good theory's explanations, though, have a much easier path to truth. This push for good explanations elevated science from medieval alchemy to electro-chemistry, or a pre-inertial physics to the forces underlying nanoparticles. And though the attempt to explain has existed as long as we have been able to wonder, a science timeline from pre-history to the present will reveal a steep curve of theoretical discovery that explodes around 1600, primarily in the West. Ranging over neuroscience, psychology, history, and policy, Wondrous Truths answers two fundamental questions-Why did science progress in the West? And why so quickly? J.D. Trout's answers are surprising. His central idea is that Western science rose above all others because it hit upon successive theories that were approximately true through an awkward assortment of accident and luck, geography and personal idiosyncrasy. Of course, intellectual ingenuity partially accounts for this persistent drive forward. But so too does the persistence of the objects of wonder. Wondrous Truths recovers the majesty of science, and provides a startling new look at the grand sweep of its biggest ideas. (shrink)

This work contrasts the formalist approach to defining explanation in science, exemplified in the Deductive-Nomenological Model of Carl G. Hempel, with the contextualist approach of Thomas Kuhn. It is argued that both of these attempts to define the explanatory processes of science are inadequate. A connection is made between the view of rationality upon which each view is based and the way that it defines explanation. It is argued that a process of thought, which scientific explanation represents, is considered rational (...) by these thinkers only if it meets the criteria of constraint under a rule, distance from subjective considerations and instrumental connection of method and goal. Given this strong definition of rationality, explanation is defined by both Kuhn and Hempel as thought within the constraint of a theory or a theoretical paradigm. Under this construal, explanation is separated as a process from theory invention which both thinkers hold to involve primarily non-cognitive criteria of choice. It is argued in this work that such a view contradicts the actual practice of science where theory invention seems to be primarily a process of rational explaining. ;A detailed historical case of an attempt to explain organic morphogenesis is given. It centers on the theory of generation of Pierre Louis Moreau de Maupertuis, an 18th Century scientific thinker. This case is meant to illustrate the kinds of cognitive interest and strategies of invention that go into dealing with an explanatory problem in science. ;Finally some general programmatic conclusions are made relating the foregoing arguments and the case, to the need to explore the nature of scientific explanation in the domain of theory invention where the reasoning processes seem to involve analogical reasoning and critical metaphysical thought. (shrink)

The focus in the literature on scientific explanation has shifted in recent years towards modelbased approaches. The idea that there are simple and true laws of nature has met with objections from philosophers such as Nancy Cartwright (1983) and Paul Teller (2001), and this has made a strictly Hempelian D-N style explanation largely irrelevant to the explanatory practices of science (Hempel & Oppenheim, 1948). Much of science does not involve subsuming particular events under laws of nature. It is increasingly recognized (...) that science across the disciplines is to some degree a patchwork of scientific models, with different methods, strategies, and with varying degrees of successful prediction and explanation. And so accounts of scientific explanation have reflected this change of perspective and model-based approaches have flourished in the explanation literature (Batterman, 2002b; Bokulich, 2008; Craver, 2006; Woodward, 2003). (shrink)

The age-old controversy pitting explanation against understanding has amazingly mutated in our time because nowadays each of these procedures attempts to replicate and even occasionally mimetizes the features traditionally ascribed to the competing practice.

The current antireductionist consensus rests in part on the indefensibility of the deductive-nomological model of explanation, on which classical reductionism depends. I argue that the DN model is inessential to the reductionist program and that mechanism provides a better framework for thinking about reductionism. This runs counter to the contemporary mechanists’ claim that mechanism is an alternative to reductionism. I demonstrate that mechanists are committed to reductionism, as evidenced by the historical roots of the contemporary mechanist program. This view shares (...) certain core commitments with reductionism. It is these shared commitments that constitute the essential elements of the reductionist program. (shrink)

When scientist investigate why things happen, they aim at giving an explanation. But what does a scientific explanation look like? In the first chapter (Theories of Scientific Explanation) of this book, the milestones in the debate on how to characterize scientific explanations are exposed. The second chapter (How to Study Scientific Explanation?) scrutinizes the working-method of three important philosophers of explanation, Carl Hempel, Philip Kitcher and Wesley Salmon and shows what went wrong. Next, it is the responsibility of current philosophers (...) of explanation to go on where Hempel, Kitcher and Salmon failed. However, we should go on in a clever way. We call this clever way the pragmatic approach to scientific explanation and clarify briefly what this approach consists in. The third chapter (A Toolbox for Describing and Evaluating Explanatory Practices) elaborates the pragmatic approach by presenting a toolbox for analysing scientific explanation. In the last chapter (Examples of Descriptions and Evaluations of Explanatory Practices) the approach is illustrated with real-life examples of scientists aiming at explaining. (shrink)

This article originated in a “cultural futures” course I taught in Seoul in 2007.1 As part of their semester project, students interviewed friends and family to identify futures that were likely to precipitate profound cultural shifts in their lives. Not surprisingly, “Korean unification” was at the top of students’ lists. After all, then-president Roh Moo-hyun had in many ways continued the “Sunshine” policies of his predecessor, President Kim Dae-jung, culminating in a largely symbolic train journey from the South to the (...) North in May 2007 and, perhaps more meaningfully, a new wave of emotional “separated family reunions” (isan’gajok sangbong). But what cultural and societal shifts might unification bring? Here .. (shrink)

This article discusses the rationality principle, especially in Popper's version, on the occasion of a commentary of Maurice Lagueux's book, Rationality and Explanation in Economics (2010).

Evolutionary developmental biology (evo-devo) is considered a ‘mechanistic science,’ in that it causally explains morphological evolution in terms of changes in developmental mechanisms. Evo-devo is also an interdisciplinary and integrative approach, as its explanations use contributions from many fields and pertain to different levels of organismal organization. Philosophical accounts of mechanistic explanation are currently highly prominent, and have been particularly able to capture the integrative nature of multifield and multilevel explanations. However, I argue that evo-devo demonstrates the need for a (...) broadened philosophical conception of mechanisms and mechanistic explanation. Mechanistic explanation (in terms of the qualitative interactions of the structural parts of a whole) has been developed as an alternative to the traditional idea of explanation as derivation from laws or quantitative principles. Against the picture promoted by Carl Craver, that mathematical models describe but usually do not explain, my discussion of cases from the strand of evo-devo which is concerned with developmental processes points to qualitative phenomena where quantitative mathematical models are an indispensable part of the explanation. While philosophical accounts have focused on the actual organization and operation of mechanisms, properties of developmental mechanisms that are about how a mechanism reacts to modifications are of major evolutionary significance, including robustness, phenotypic plasticity, and modularity. A philosophical conception of mechanisms is needed that takes into account quantitative changes, transient entities and the generation of novel types of entities, feedback loops and complex interaction networks, emergent properties, and, in particular, functional-dynamical aspects of mechanisms, including functional (as opposed to structural) organization and distributed, system-wide phenomena. I conclude with general remarks on philosophical accounts of explanation. (shrink)