Carl Craver investigates what we are doing when we sue neuroscience to explain what's going on in the brain.

The concept of mechanism is analyzed in terms of entities and activities, organized such that they are productive of regular changes. Examples show how mechanisms work in neurobiology and molecular biology. Thinking in terms of mechanisms provides a new framework for addressing many traditional philosophical issues: causality, laws, explanation, reduction, and scientific change.

Carl F. Craver investigates what we are doing when we use neuroscience to explain what's going on in the brain. When does an explanation succeed and when does it fail? Craver offers explicit standards for successful explanation of the workings of the brain, on the basis of a systematic view about what neuroscientific explanations are.

We sketch a framework for building a unified science of cognition. This unification is achieved by showing how functional analyses of cognitive capacities can be integrated with the multilevel mechanistic explanations of neural systems. The core idea is that functional analyses are sketches of mechanisms , in which some structural aspects of a mechanistic explanation are omitted. Once the missing aspects are filled in, a functional analysis turns into a full-blown mechanistic explanation. By this process, functional analyses are seamlessly integrated (...) with multilevel mechanistic explanations. (shrink)

Completeness is an important but misunderstood norm of explanation. It has recently been argued that mechanistic accounts of scientific explanation are committed to the thesis that models are complete only if they describe everything about a mechanism and, as a corollary, that incomplete models are always improved by adding more details. If so, mechanistic accounts are at odds with the obvious and important role of abstraction in scientific modelling. We respond to this characterization of the mechanist’s views about abstraction and (...) articulate norms of completeness for mechanistic explanations that have no such unwanted implications. _1_ Introduction _2_ A Balancing Act: When Do Details Matter? _3_ The Norms of Causal Explanation _4_ The Norms of Constitutive Explanation _5_ Salmon-Completeness _6_ From More Details to More Relevant Details _7_ Non-explanatory Virtues of Abstraction _8_ From Explanatory Models to Explanatory Knowledge _9_ Mechanistic Completeness Reconsidered _10_ Conclusion. (shrink)

We argue that intelligible appeals to interlevel causes (top-down and bottom-up) can be understood, without remainder, as appeals to mechanistically mediated effects. Mechanistically mediated effects are hybrids of causal and constitutive relations, where the causal relations are exclusively intralevel. The idea of causation would have to stretch to the breaking point to accommodate interlevel causes. The notion of a mechanistically mediated effect is preferable because it can do all of the required work without appealing to mysterious interlevel causes. When interlevel (...) causes can be translated into mechanistically mediated effects, the posited relationship is intelligible and should raise no special philosophical objections. When they cannot, they are suspect. (shrink)

We argue that dynamical and mathematical models in systems and cognitive neuro- science explain (rather than redescribe) a phenomenon only if there is a plausible mapping between elements in the model and elements in the mechanism for the phe- nomenon. We demonstrate how this model-to-mechanism-mapping constraint, when satisfied, endows a model with explanatory force with respect to the phenomenon to be explained. Several paradigmatic models including the Haken-Kelso-Bunz model of bimanual coordination and the difference-of-Gaussians model of visual receptive fields are (...) explored. (shrink)

Not all models are explanatory. Some models are data summaries. Some models sketch explanations but leave crucial details unspecified or hidden behind filler terms. Some models are used to conjecture a how-possibly explanation without regard to whether it is a how-actually explanation. I use the Hodgkin and Huxley model of the action potential to illustrate these ways that models can be useful without explaining. I then use the subsequent development of the explanation of the action potential to show what is (...) required of an adequate mechanistic model. Mechanistic models are explanatory. (shrink)

According to one large family of views, scientific explanations explain a phenomenon (such as an event or a regularity) by subsuming it under a general representation, model, prototype, or schema (see Bechtel, W., & Abrahamsen, A. (2005). Explanation: A mechanist alternative. Studies in History and Philosophy of Biological and Biomedical Sciences, 36(2), 421–441; Churchland, P. M. (1989). A neurocomputational perspective: The nature of mind and the structure of science. Cambridge: MIT Press; Darden (2006); Hempel, C. G. (1965). Aspects of scientific (...) explanation. In C. G. Hempel (Ed.), Aspects of scientific explanation (pp. 331–496). New York: Free Press; Kitcher (1989); Machamer, P., Darden, L., & Craver, C. F. (2000). Thinking about mechanisms. Philosophy of Science, 67(1), 1–25). My concern is with the minimal suggestion that an adequate philosophical theory of scientific explanation can limit its attention to the format or structure with which theories are represented. The representational subsumption view is a plausible hypothesis about the psychology of understanding. It is also a plausible claim about how scientists present their knowledge to the world. However, one cannot address the central questions for a philosophical theory of scientific explanation without turning one’s attention from the structure of representations to the basic commitments about the worldly structures that plausibly count as explanatory. A philosophical theory of scientific explanation should achieve two goals. The first is explanatory demarcation. It should show how explanation relates with other scientific achievements, such as control, description, measurement, prediction, and taxonomy. The second is explanatory normativity. It should say when putative explanations succeed and fail. One cannot achieve these goals without undertaking commitments about the kinds of ontic structures that plausibly count as explanatory. Representations convey explanatory information about a phenomenon when and only when they describe the ontic explanations for those phenomena. (shrink)

Many areas of science develop by discovering mechanisms and role functions. Cummins' (1975) analysis of role functions-according to which an item's role function is a capacity of that item that appears in an analytic explanation of the capacity of some containing system-captures one important sense of "function" in the biological sciences and elsewhere. Here I synthesize Cummins' account with recent work on mechanisms and causal/mechanical explanation. The synthesis produces an analysis of specifically mechanistic role functions, one that uses the characteristic (...) active, spatial, temporal, and hierarchical organization of mechanisms to add precision and content to Cummins' original suggestion. This synthesis also shows why the discovery of role functions is a scientific achievement. Discovering a role function (i) contributes to the interlevel integration of multilevel mechanisms, and (ii) provides a unique, contextual variety of causal/mechanical explanation. (shrink)

It is common to defend the Homeostatic Property Cluster ( HPC ) view as a third way between conventionalism and essentialism about natural kinds ( Boyd , 1989, 1991, 1997, 1999; Griffiths , 1997, 1999; Keil , 2003; Kornblith , 1993; Wilson , 1999, 2005; Wilson , Barker , & Brigandt , forthcoming ). According to the HPC view, property clusters are not merely conventionally clustered together; the co-occurrence of properties in the cluster is sustained by a similarity generating ( (...) or homeostatic ) mechanism . I argue that conventional elements are involved partly but ineliminably in deciding which mechanisms define kinds , for deciding when two mechanisms are mechanisms of the same type, and for deciding where one particular mechanism ends and another begins. This intrusion of conventional perspective into the idea of a mechanism raises doubts as to whether the HPC view is sufficiently free of conventional elements to serve as an objective arbiter in scientific disputes about what the kinds of the special sciences should be. (shrink)

In “What Makes a Scientific Explanation Distinctively Mathematical?” (2013b), Lange uses several compelling examples to argue that certain explanations for natural phenomena appeal primarily to mathematical, rather than natural, facts. In such explanations, the core explanatory facts are modally stronger than facts about causation, regularity, and other natural relations. We show that Lange's account of distinctively mathematical explanation is flawed in that it fails to account for the implicit directionality in each of his examples. This inadequacy is remediable in each (...) case by appeal to ontic facts that account for why the explanation is acceptable in one direction and unacceptable in the other direction. The mathematics involved in these examples cannot play this crucial normative role. While Lange's examples fail to demonstrate the existence of distinctively mathematical explanations, they help to emphasize that many superficially natural scientific explanations rely for their explanatory force on relations of stronger-than-natural necessity. These are not opposing kinds of scientific explanations; they are different aspects of scientific explanation. (shrink)

Philosophers of neuroscience have traditionally described interfield integration using reduction models. Such models describe formal inferential relations between theories at different levels. I argue against reduction and for a mechanistic model of interfield integration. According to the mechanistic model, different fields integrate their research by adding constraints on a multilevel description of a mechanism. Mechanistic integration may occur at a given level or in the effort to build a theory that oscillates among several levels. I develop this alternative model using (...) a putative exemplar of reduction in contemporary neuroscience: the relationship between the psychological phenomena of learning and memory and the electrophysiological phenomenon known as Long-Term Potentiation. A new look at this historical episode reveals the relative virtues of the mechanistic model over reduction as an account of interfield integration. (shrink)

In the 1950s and 1960s, an interfield interaction between molecular biologists and biochemists integrated important discoveries about the mechanism of protein synthesis. This extended discovery episode reveals two general reasoning strategies for eliminating gaps in descriptions of the productive continuity of mechanisms: schema instantiation and forward chaining/backtracking. Schema instantiation involves filling roles in an overall framework for the mechanism. Forward chaining and backtracking eliminate gaps using knowledge about types of entities and their activities. Attention to mechanisms highlights salient features of (...) this historical episode while providing general reasoning strategies for mechanism discovery. (shrink)

Network analysis is increasingly used to discover and represent the organization of complex systems. Focusing on examples from neuroscience in particular, I argue that whether network models explain, how they explain, and how much they explain cannot be answered for network models generally but must be answered by specifying an explanandum, by addressing how the model is applied to the system, and by specifying which kinds of relations count as explanatory.

The dominant neuroscientific theory of spatial memory is, like many theories in neuroscience, a multilevel description of a mechanism. The theory links the activities of molecules, cells, brain regions, and whole organisms into an integrated sketch of an explanation for the ability of organisms to navigate novel environments. Here I develop a taxonomy of interlevel experimental strategies for integrating the levels in such multilevel mechanisms. These experimental strategies include activation strategies, interference strategies, and additive strategies. These strategies are mutually reinforcing, (...) providing a kind of interlevel and intratheoretic robustness that has not previously been recognized. (shrink)

We sketch the mechanistic approach to levels, contrast it with other senses of “level,” and explore some of its metaphysical implications. This perspective allows us to articulate what it means for things to be at different levels, to distinguish mechanistic levels from realization relations, and to describe the structure of multilevel explanations, the evidence by which they are evaluated, and the scientific unity that results from them. This approach is not intended to solve all metaphysical problems surrounding physicalism. Yet it (...) provides a framework for thinking about how the macroscopic phenomena of our world are or might be related to its most fundamental entities and activities. (shrink)

Hodgkin and Huxley’s model of the action potential is an apparent dream case of covering‐law explanation in biology. The model includes laws of physics and chemistry that, coupled with details about antecedent and background conditions, can be used to derive features of the action potential. Hodgkin and Huxley insist that their model is not an explanation. This suggests either that subsuming a phenomenon under physical laws is insufficient to explain it or that Hodgkin and Huxley were wrong. I defend Hodgkin (...) and Huxley against Weber’s heteronomy thesis and argue that explanations are descriptions of mechanisms. †To contact the author, please write to: Department of Philosophy, Philosophy‐Neuroscience‐Psychology Program, Washington University in St. Louis, One Brookings Drive, Wilson Hall, St. Louis, MO 63130; e‐mail: [email protected] (shrink)

In what sense are the activities and properties of components in a mechanism explanatorily relevant to the behavior of a mechanism as a whole? I articulate this problem, the problem of constitutive relevance, and I show that it must be solved if we are to understand mechanisms and mechanistic explanation. I argue against some putative solutions to the problem of constitutive relevance, and I sketch a positive account according to which relevance is analyzed in terms ofrelationships of mutual manipulability between (...) the behavior of a mechanism as a whole and the properties and activities of its components. My account is a causal-mechanical account in the sense that it is a particular expression of the idea that constitutive explanation is a matter of showing how an explanandum phenomenon is situated within the causal structure of the world. It is thus offered as a rival to epistemic (argument-centered) and psychological accounts of interlevel explanation. (shrink)

The construct “remembering” is equivocal between an epistemic sense, denoting a distinctive ground for knowledge, and empirical sense, denoting the typical behavior of a neurocognitive mechanism. Because the same kind of equivocation arises for other psychologistic terms, the effort to spot and remedy the confusion in the case of remembering might prove generally instructive. The failure to allow these two senses of remembering equal play in their respective domains leads, I argue, to unnecessary confusion about memory externalism, the possibility of (...) episodic memory in non-human species, and the thesis of memory continuism. By distinguishing these equivocal senses of remembering, we thus gain leverage on understanding how the distinctive epistemic norms that define many of our psychologic terms are more plausibly related to the capacities studied by empirical science, given that neither identity nor elimination are possible. (shrink)

The dominant neuroscientific theory of spatial memory is, like many theories in neuroscience, a multilevel description of a mechanism. The theory links the activities of molecules, cells, brain regions, and whole organisms into an integrated sketch of an explanation for the ability of organisms to navigate novel environments. Here I develop a taxonomy of interlevel experimental strategies for integrating the levels in such multilevel mechanisms. These experimental strategies include activation strategies, interference strategies, and additive strategies. These strategies are mutually reinforcing, (...) providing a kind of interlevel and intratheoretic robustness that has not previously been recognized. (shrink)

In what sense are the activities and properties of components in a mechanism explanatorily relevant to the behavior of a mechanism as a whole? I articulate this problem, the problem of constitutive relevance, and I show that it must be solved if we are to understand mechanisms and mechanistic explanation. I argue against some putative solutions to the problem of constitutive relevance, and I sketch a positive account according to which relevance is analyzed in terms ofrelationships of mutual manipulability between (...) the behavior of a mechanism as a whole and the properties and activities of its components. My account is a causal-mechanical account in the sense that it is a particular expression of the idea that constitutive explanation is a matter of showing how an explanandum phenomenon is situated within the causal structure of the world. It is thus offered as a rival to epistemic and psychological accounts of interlevel explanation. (shrink)

We argue that neuroeconomics should be a mechanistic science. We defend this view as preferable both to a revolutionary perspective, according to which classical economics is eliminated in favour of neuroeconomics, and to a classical economic perspective, according to which economics is insulated from facts about psychology and neuroscience. We argue that, like other mechanistic sciences, neuroeconomics will earn its keep to the extent that it either reconfigures how economists think about decision-making or how neuroscientists think about brain mechanisms underlying (...) behaviour. We discuss some ways that the search for mechanisms can bring about such top-down and bottom-up revision, and we consider some examples from the recent neuroeconomics literature of how varieties of progress of this sort might be achieved. (shrink)

Lange’s collection of expanded, mostly previously published essays, packed with numerous, beautiful examples of putatively non-causal explanations from biology, physics, and mathematics, challenges the increasingly ossified causal consensus about scientific explanation, and, in so doing, launches a new field of philosophic investigation. However, those who embraced causal monism about explanation have done so because appeal to causal factors sorts good from bad scientific explanations and because the explanatory force of good explanations seems to derive from revealing the relevant causal (or (...) ontic) structures. The taxonomic project of collecting examples and sorting their types is an essential starting place for a theory of non-causal explanation. But the title of Lange’s book requires something further: showing that the putative explanations are, in fact, explanatory and revealing the non-causal source of their explanatory power. This project is incomplete if there are examples of putative non-causal explanations that fit the form but that nobody would accept as explanatory (absent a radical revision of intuitions). Here we provide some reasons for thinking that there are such examples. (shrink)

For the greater part of the last 50 years, it has been common for philosophers of mind and cognitive scientists to invoke the notion of realization in discussing the relationship between the mind and the brain. In traditional philosophy of mind, mental states are said to be realized, instantiated, or implemented in brain states. Artificial intelligence is sometimes described as the attempt either to model or to actually construct systems that realize some of the same psychological abilities that we and (...) other living creatures possess. The claim that specific psychological. (shrink)

Long-Term Potentiation (LTP) is a kind of synaptic plasticity that many contemporary neuroscientists believe is a component in mechanisms of memory. This essay describes the discovery of LTP and the development of the LTP research program. The story begins in the 1950's with the discovery of synaptic plasticity in the hippocampus (a medial temporal lobe structure now associated with memory), and it ends in 1973 with the publication of three papers sketching the future course of the LTP research program. The (...) making of LTP was a protracted affair. Hippocampal synaptic plasticity was initially encountered as an experimental tool, then reported as a curiosity, and finally included in the ontic store of the neurosciences. Early researchers were not investigating the hippocampus in search of a memory mechanism; rather, they saw the hippocampus as a useful experimental model or as a structure implicated in the etiology of epilepsy. The link between hippocampal synaptic plasticity and learning or memory was a separate conceptual achievement. That link was formulated in at least three different ways at different times: reductively (claiming that plasticity is identical to learning), analogically (claiming that plasticity is an example or model of learning), and mechanistically (claiming that plasticity is a component in learning or memory mechanisms). The hypothesized link with learning or memory, coupled with developments in experimental techniques and preparations, shaped how researchers understood LTP itself. By 1973, the mechanistic formulation of the link between LTP and memory provided an abstract framework around which findings from multiple perspectives could be integrated into a multifield research program. (shrink)

It is a common assumption in contemporary cognitive neuroscience that discovering a putative realized kind to be dissociably realized (i.e., to be realized in each instance by two or more distinct realizers) mandates splitting that kind. Here I explore some limits on this inference using two deceptively similar examples: the dissociation of declarative and procedural memory and Ramachandran's argument that the self is an illusion.

We address Turkheimer’s argument that genome-wide association studies of behaviors and psychiatric traits will fail to produce coherent explanations. We distinguish two major sources of potential i...

In a recent issue of The Monist, Alisa Bokulich argues that those who embrace an ontic conception of scientific explanation are committed to rejecting an explanatory role for idealized, i.e., deliberately false, models. Her argument is based on an inaccurate characterization of the ontic view. Indeed, her positive view of idealization embraces rather than opposes the ontic conception. Because Bokulich is not alone in this misunderstanding, an effort to diagnose and correct it might prevent scholars from talking past one another (...) and, hopefully, nudge future discussions down more inviting paths. (shrink)

An adequate understanding of the ubiquitous practice of mechanistic explanation requires an account of what Craver termed “constitutive relevance.” Entities or activities are constitutively relevant to a phenomenon when they are parts of the mechanism responsible for that phenomenon. Craver’s mutual manipulability account extended Woodward’s account of manipulationist counterfactuals to analyze how interlevel experiments establish constitutive relevance. Critics of MM argue that applying Woodward’s account to this philosophical problem conflates causation and constitution, thus rendering the account incoherent. These criticisms, we (...) argue, arise from failing to distinguish the semantic, epistemic, and metaphysical aspects of the problem of constitutive relevance. In distinguishing these aspects of the problem and responding to these critics accordingly, we amend MM into a refined epistemic criterion, the “matched interlevel experiments” account. Further, we explain how this epistemological thesis is grounded in the plausible metaphysical thesis that constitutive relevance is causal betweenness. (shrink)

In Minds, Brains, and Norms , Pardo and Patterson deny that the activities of persons (knowledge, rule-following, interpretation) can be understood exclusively in terms of the brain, and thus conclude that neuroscience is irrelevant to the law, and to the conceptual and philosophical questions that arise in legal contexts. On their view, such appeals to neuroscience are an exercise in nonsense. We agree that understanding persons requires more than understanding brains, but we deny their pessimistic conclusion. Whether neuroscience can be (...) used to address legal issues is an empirical question. Recent work on locked-in syndrome, memory, and lying suggests that neuroscience has potential relevance to the law, and is far from nonsensical. Through discussion of neuroscientific methods and these recent results we show how an understanding of the subpersonal mechanisms that underlie person-level abilities could serve as a valuable and illuminating source of evidence in legal and social contexts. In so doing, we sketch the way forward for a no-nonsense approach to the intersection of law and neuroscience. (shrink)

It is common in psychiatry and other sciences to describe an individual or a type of individual in terms of its disposition to manifest specific effects in a particular range of circumstances. According to one understanding, dispositions are statistical regularities of an individual or type of individual in specific circumstances. According to another understanding, dispositions are properties of individuals in virtue of which such regularities hold. This entry considers a number of ways of making each of these senses of disposition (...) more precise while discussing a number of dangers lurking in careless use of the concept of a disposition. (shrink)

Hodgkin and Huxley’s 1952 model of the action potential is an apparent dream case of covering-law explanation. The model appeals to general laws of physics and chemistry (specifically, Ohm’s law and the Nernst equation), and the laws, coupled with details about antecedent and background conditions, entail many of the significant properties of the action potential. However, Hodgkin and Huxley insist that their model falls short of an explanation. This historical fact suggests either that there is more to explaining the action (...) potential than subsuming it under a general laws or that Hodgkin and Huxley were wrong about the explanatory import of their model. In this paper, I defend Hodgkin and Huxley’s view that their model alone does not explain the action potential (contra Weber 2005). I argue further that neuroscientists lacked crucial explanatory details about the action potential until they could describe the molecular and ionic mechanisms by virtue of which their model holds (see Bogen 2005). Mathematical generalizations are important epistemic tools for assessing mechanistic explanations, but they are neither necessary nor sufficient for adequate explanations, even at the lowest levels of organization where biological phenomena are integrated with physics and chemistry. (shrink)

The thesis of physical supervenience is widely understood and endorsed as the weakest assertion that all facts are tethered to the physical facts. As an exercise in exploring the constitutive commitments of an ontic view of mechanistic explanation, I entertain a weaker tethering relation, stochastic physical supervenience, the possibility of which is suggested by analogy with the apparent failure of causal determinism in certain areas of physical science. Considering this possibility helps to clarify the constitutive commitments of mechanistic explanation and (...) to explore the motivations for accepting the PS thesis. (shrink)