1.1 Preliminaries
Let’s call ‘Old Mechanism’ (or mechanical natural philosophy) the general conception about the nature of the world and of science that was developed in the seventeenth century by thinkers such as Descartes, Boyle, Huygens and Leibniz. The currently popular New Mechanism too constitutes a general framework for understanding science and nature, which emerged and spread in philosophy of science in the beginning of the twenty-first century. In this chapter, we look at some main aspects of the historical development of the concept of mechanism, by examining the differences and similarities between New and Old Mechanism. The motivation comes from the explicit intention of new mechanists to link the current mechanical philosophy with its older counterpart in the seventeenth century (see Machamer et al. Reference Machamer, Darden and Craver2000). However, there are several questions to be asked about the real relationship between the new and old mechanical philosophy.
A key similarity between Old and New Mechanism is that, for many old and new mechanists, Mechanism is both a metaphysical position about the structure of reality and a methodological thesis about the general form of scientific explanation and the methodology of science. The main aim of this chapter is to examine the relation between the metaphysics of mechanisms and the methodological role of mechanical explanation in the practice of science, by presenting and comparing the key tenets of Old and New Mechanism. We will use this historical examination to motivate our deflationary account of mechanism developed in the book.
1.2 Old versus New Mechanism
The mechanical world view of the seventeenth century was both a metaphysical thesis and a scientific theory. It was a metaphysical thesis insofar as it was committed to a reductionist account of all worldly phenomena to configurations of matter in motion subject to laws. In particular, it was committed to the view that all macroscopic phenomena are caused, and hence are accounted for, by the interactions of invisible microscopic material corpuscles. Margaret Wilson captured this view succinctly:
The mechanism characteristic of the new science of the seventeenth century may be briefly characterised as follows: Mechanists held that all macroscopic bodily phenomena result from the motions and impacts of submicroscopic particles, or corpuscles, each of which can be fully characterised in terms of a strictly limited range of (primary) properties: size, shape, motion and, perhaps, solidity and impenetrability.
But this metaphysical thesis did, at the same time, license a scientific theory of the world, namely, a certain conception of scientific explanation and of theory construction. To offer a scientific explanation of a worldly phenomenon X was to provide a configuration Y of matter in motion, subject to laws, such that Y could cause X. A mechanical explanation then was (a species of) causal explanation: to explain that Y causes X was tantamount to constructing a mechanical model of how Y brings about X. The model was mechanical insofar as it was based on resources licensed by the metaphysical world view, namely, action of particles by contact in virtue of their primary qualities and subject to laws of motion.Footnote 1
Nearly four centuries later, the mechanical world view has become prominent again within philosophy of science. It’s become known as the ‘New Mechanical Philosophy’ or the ‘New Mechanism’ and has similar aspirations as the old one. New Mechanism, as Stuart Glennan puts it, ‘says of nature that most or all the phenomena found in nature depend on mechanisms – collections of entities whose activities and interactions, suitably organized, are responsible for these phenomena. It says of science that its chief business is the construction of models that describe, predict, and explain these mechanism-dependent phenomena’ (Reference Glennan2017, 1).
So, New Mechanism too is a view about both science and the metaphysics of nature. And yet, in New Mechanism the primary focus has been on scientific practice, and in particular on the use of mechanisms in discovery, reasoning and representation (see Glennan Reference Glennan2017, 12). The focus on the metaphysics of mechanisms has emerged as an attempt to draw conclusions about the ontic signature of the world starting from the concept of mechanism as it is used in the sciences. According to Glennan, as the research into the use of mechanism in science developed, ‘it has been clear to many participants in the discussion that metaphysical questions are unavoidable’ (Reference Glennan2017, 12). It is fair to say that New Mechanism aims to ground the metaphysics of mechanisms on the practice of mechanical explanation in the sciences.
1.3 Old Mechanism: From Metaphysics to Practice
A rather typical example of the interplay between the metaphysical world view and the scientific conception of the world in the seventeenth century was the attempted mechanical explanation of gravity.
1.3.1 Mechanical Models of Gravity
Let us start with Descartes. The central aim of the third and fourth parts of Descartes’s Principia Philosophiae, published in 1644, was the construction of an account of natural phenomena. In Cartesian physics, the possible empirical models of the world are restricted from above by a priori principles which capture the fundamental laws of nature and from below by experience. Between these two levels there are various theoretical hypotheses, which constitute the proper empirical subject matter of science. These are mechanical hypotheses; they refer to configurations of matter in motion. As Descartes explains in Part III, 46, of the Principia, since it is a priori possible that there are countless configurations of matter in motion that can underlie the various natural phenomena, ‘unaided reason’ is not able to figure out the right configuration of matter in motion. Mechanical hypotheses are necessary but experience should be appealed to, in order to pick out the correct one: ‘[W]e are now at liberty to assume anything we please [about the mechanical configuration], provided that everything we shall deduce from it is {entirely} in conformity with experience’ (III, 46; Reference Descartes, Miller and Miller1982, 106).
These mechanical hypotheses aim to capture the putative causes of the phenomena under investigation (III, 47). Hence, they are explanatory of the phenomena. Causal explanation – that is, mechanical explanation – proceeds via decomposition. It is a commitment of the mechanical philosophy that the behaviour of observable bodies should be accounted for on the basis of the interactions among their constituent parts and particles, hence, on the basis of unobservable entities. Descartes states (IV, 201; Reference Descartes, Miller and Miller1982, 283) that sensible bodies are composed of insensible particles. But to get to know these particles and their properties a bridge principle is necessary, that is, a principle that connects the micro-constituents with the macro-bodies. According to this principle, the properties of the minute particles should be modelled on the properties of macro-bodies. Here is how Descartes put it:
Nor do I think that anyone who is using his reason will be prepared to deny that it is far better to judge of things which occur in tiny bodies (which escape our senses solely because of their smallness) on the model of those which our senses perceive occurring in large bodies, than it is to devise I know not what new things, having no similarity with those things which are observed, in order to give an account of those things [in tiny bodies]. {E.g., prime matter, substantial forms, and all that great array of qualities which many are accustomed to assuming; each of which is more difficult to know than the things men claim to explain by their means}.
In this passage Descartes does two things. On the one hand, he advances a continuity thesis: it is simpler and consonant with what our senses reveal to us to assume that the properties of micro-objects are the same as the properties of macro-objects. This continuity thesis is primarily methodological. It licenses certain kinds of explanations: those that endow matter in general, and hence the unobservable parts of matter, with the properties of the perceived bits of matter. It therefore licenses as explanatory certain kinds of unobservable configurations of matter, namely, those that resemble perceived configurations of matter. On the other hand, however, Descartes circumscribes mechanical explanation by noting what it excludes, that is, by specifying what does not count as a proper scientific explanation. He’s explicit that the Aristotelian-scholastic metaphysics of substantial forms and powerful qualities is precisely what is abandoned as explanatory by the mechanical philosophy.Footnote 2
All this was followed in the investigation of the mechanism of gravity and the (in)famous vortex hypothesis according to which the planets are carried by vortices around the sun. A vortex is a specific configuration of matter in motion – matter revolving around a centre. The underlying mechanism of the planetary system then is a system of vortices:
[T]he matter of the heaven, in which the Planets are situated, unceasingly revolves, like a vortex having the Sun as its center, and … those of its parts which are close to the Sun move more quickly than those further away; and … all the Planets (among which we {shall from now on} include the Earth) always remain suspended among the same parts of this heavenly matter.
The very idea of this kind of configuration is suggested by experience, and by means of the bridge principle it is transferred to the subtle matter of the heavens. Hence, invisibility doesn’t matter. The bridge principle transfers the explanatory mechanism from visible bodies to invisible bodies. More specifically, the specific continuity thesis used is the motion of ‘some straws {or other light bodies} … floating in the eddy of a river where the water doubles back on itself and forms a vortex as it swirls’. In this kind of motion we can see that the vortex carries the straws ‘along and makes them move in circles with it’. We also see that ‘some of these straws rotate about their own centers, and that those which are closer to the center of the vortex which contains them complete their circle more rapidly than those which are further away from it’. More importantly for the explanation of gravity, we see that ‘although these whirlpools always attempt a circular motion, they practically never describe perfect circles, but sometimes become too great in width or in length.’ Given the continuity thesis, we can transfer this mechanical model to the motion of the planets and ‘imagine that all the same things happen to the Planets; and this is all we need to explain all their remaining phenomena’ (III, 30; Reference Descartes, Miller and Miller1982, 96). Notably, the continuity thesis offers a heuristic for discovering plausible mechanical explanations.
Huygens (Reference Huygens1690/1997) came to doubt the vortex theory, ‘which formerly appeared very likely’ to him (p. 32). He didn’t thereby abandon the key tenet of mechanical philosophy. For Huygens too the causal explanation of a natural phenomenon had to be mechanical. He said, referring to Descartes: ‘Mr Descartes has recognized, better than those that preceded him, that nothing will be ever understood in physics except what can be made to depend on principles that do not exceed the reach of our spirit, such as those that depend on bodies, deprived of qualities, and their motions’ (pp. 1–2).
Huygens posited a fluid matter that consists of very small parts in rapid motion in all directions and which fills the spherical space that includes all heavenly bodies. Since there is no empty space, this fluid matter is more easily moved in circular motion around the centre, but not all parts of it move in the same direction. As Huygens put it ‘it is not difficult now to explain how gravity is produced by this motion’ (p. 16). When the parts of the fluid matter encounter some bigger bodies, like the planets, ‘these bodies [the planets] will necessarily be pushed towards the center of motion, since they do not follow the rapid motion of the aforementioned matter’. And he added: ‘This then is in all likelihood what the gravity of bodies truly consists of: we can say that this is the endeavor that causes the fluid matter, which turns circularly around the center of the Earth in all directions, to move away from the center and to push in its place bodies that do not follow this motion.’ In fact, Huygens devised an experiment with bits of beeswax to show how this movement towards the centre can take place.
Newton of course challenged all this, along the lines that the very idea of causal explanation should be mechanical. But before we take a look at his reasons and their importance for the very idea of mechanical explanation, we should not fail to see the broader metaphysical grounding of the mechanical project. For, as we noted, in the seventeenth century mechanism offered the metaphysical foundation of science.
1.3.2 Mechanical versus Non-Mechanical Explanation
The contours of this endeavour are well known. Matter and motion are the ‘ultimate constituents’ of nature, or, as Boyle (Reference Boyle and Stewart1991, 20) put it, the ‘two grand and most catholic principles of bodies’. Hence, all there is in nature (but clearly not the Cartesian minds) is determined (caused) by the mechanical affections of bodies and the mechanical laws. Here is Boyle again: ‘[T]he universe being once framed by God, and the laws of motion being settled and all upheld by his incessant concourse and general providence, the phenomena of the world thus constituted are physically produced by the mechanical affections of the parts of matter, and what they operate upon one another according to mechanical laws’ (Reference Boyle and Stewart1991, 139).
The Boylean conception, pretty much like the Cartesian, took it that the new mechanical approach acquired content by excluding the then dominant account of explanation in terms of ‘real qualities’: the scholastics ‘attribute to them a nature distinct from the modification of the matter they belong to, and in some cases separable from all matter whatsoever’ (pp. 15–16). Explanation based on real qualities, which are distinct (and separable) from matter, is not a genuine explanation. They are posited without ‘searching into the nature of particular qualities and their effects’ (p. 16). They offer sui generis explanations: why does snow dazzle the eyes? Because of ‘a quality of whiteness that is in it, which makes all very white bodies produce the same effect’ (p. 16). But what is whiteness? No further story about its nature is offered, but just that it’s a ‘real entity’ inhering in the substance. Why do white objects produce this effect rather than that? Because it is in their nature to act thus.
Descartes made this point too when, in his Le Monde, he challenged the scholastic rivals to explain how fire burns wood, if not by the incessant and rapid motion of its minute parts. In his characteristic upfrontness, Descartes contrasted two ways to explain how fire burns wood. The first is the Aristotelian way, according to which ‘the ‘form’ of fire, the ‘quality’ of heat and the ‘action’ of burning’ are ‘very different things in the wood’ (Descartes Reference Descartes and Gaukroger2004, 6). The other is his own mechanistic way: when the fire burns wood, ‘it moves the small parts of the wood, separating them from one another, thereby transforming the finer parts into fire, air, and smoke, and leaving the larger parts as ashes’ (Reference Descartes and Gaukroger2004, 6).
This causal explanation, based as it is on matter in motion, is preferable precisely because it is explanatory of the burning; in contrast, the Aristotelian is not, precisely because it does not make clear the mechanism by which the fire consumes the wood: ‘[Y]ou can posit “fire” and “heat” in the wood and make it burn as much as you please: but if you do not suppose in addition that some of its parts move or are detached from their neighbours then I cannot imagine that it would undergo any alteration or change’ (p. 6). To the then dominant account of real qualities, the new mechanical metaphysics juxtaposed a different view of qualities. For something to be a quality it should be determined by the mechanical affections of matter, that is, by ‘virtue of the motion, size, figure, and contrivance, of their own parts’ (Boyle Reference Boyle and Stewart1991, 17). Hence, there can be no change in qualities unless there is a change in mechanical affections. Though ‘catholic or universal matter’ is common to all bodies (being, as Boyle [p. 18] put it, ‘a substance extended, divisible, and impenetrable’), it is diversified by motion, which is regulated by laws.
The key point then is that the mechanical account of nature is both a metaphysical grounding of science and a (the) way to do science: offering mechanical explanations of the phenomena. It covers everything, from the very small to the very large. Here is Boyle again: ‘For both the mechanical affections of matter are to be found, and the laws of motion take place, not only in the great masses and the middle-sized lumps, but in the smallest fragments of matter; and a lesser portion of it, being as well a body as a greater, must, as necessarily as it, have its determinate bulk and figure’ (p. 143).
The metaphysical grounding of mechanical explanation renders it a distinct kind of explanation, which separates it sharply from rival accounts. Concomitantly, it becomes very clear what counts as a non-mechanical alternative. An explanation couched in terms of ‘nature, substantial forms, real qualities, and the like’ is ‘unmechanical’ (p. 142). But a sui generis chemical account of nature is unmechanical too. As Boyle put it:
[T]hough chemical explications be sometimes the most obvious and ready, yet they are not the most fundamental and satisfactory: for the chemical ingredient itself, whether sulphur or any other, must owe its nature and other qualities to the union of insensible particles in a convenient size, shape, motion or rest, and contexture, all which are but mechanical affections of convening corpuscles.
The opposition to both of these non-mechanical accounts is weaved around a certain metaphysical account of the world as fundamentally mechanical and a reductive-decompositional account of scientific explanation itself.
1.3.3 Boyle on Mechanical Explanation
Boyle’s discussion of the nature of mechanical explanation in his ‘About the Excellency and Grounds of the Mechanical Hypothesis’ deserves further analysis, as it is particularly relevant for our purposes in this book. In that essay, Boyle contrasts mechanical with other kinds of explanations and points out that only the former provide information about how exactly a result is produced. What is particularly interesting for us is that Boyle focuses on what can be broadly described as ‘medical’ examples. Consider the following passage:
They that, to solve the phenomena of nature, have recourse to agents which, though they involve no self-repugnancy in their very notions, as many of the judicious think substantial forms and real qualities to do, yet are such that we conceive not how they operate to bring effects to pass – these, I say, when they tell us of such indeterminate agents as the soul of the world, the universal spirit, the plastic power, and the like, though they may in certain cases tell us some things, yet they tell us nothing that will satisfy the curiosity of an inquisitive person, who seeks not so much to know what is the general agent that produces a phenomenon, as by what means, and after what manner, the phenomenon is produced.
Here, Boyle points out that, in giving an explanation, it is not enough to state what the cause is; what is more important is to state how exactly a cause operates to bring about the effect. Failure to do this, Boyle thinks, is the main problem with explanations that merely state a causal agent without providing further information as to the manner that this agent acts. Boyle goes on to give an example of such an unsatisfactory medical explanation:
The famous Sennertus and some other learned physicians tell us of diseases which proceed from incantation: but sure it is but a very slight account that a sober physician, that comes to visit a patient reported to be bewitched, receives of the strange symptoms he meets with and would have an account of, if he be coldly answered that it is a witch or the devil that produces them.
Similarly,
it would be but little satisfaction to one that desires to understand the causes of what occurs to observation in a watch, and how it comes to point at and strike the hours, to be told that it was such a watchmaker that so contrived it; or to him that would know the true cause of an echo to be answered that it is a man, a vault, or a wood, that makes it.
The point that Boyle makes in these passages is that, quite apart from the accusation that notions such as substantial forms and real qualities are obscure, such explanations as well as others (e.g., in terms of plastic powers, which Boyle thinks are not as bad as the ones offered by the scholastics) do not fulfil what he takes as a general adequacy condition that an explanation has to satisfy, that is, to provide information as to how exactly a cause acts. Consider, for example, the following causal claim: administering of the poison led to the death of the person. We can interpret Boyle as saying that, qua explanation of death, such an explanation is incomplete; what is missing is how exactly administering of the poison led to death. Stating that the poison possessed the power to bring about death is tantamount to saying that it in fact produced death, that is, that it was the cause of death. What we need in addition to this, however, is the way it did so. A way to satisfy this demand is to provide information about the changes that the poison produced within the organism and explain how they eventually led to death. Here is how Boyle puts it:
I consider that the chief thing that inquisitive naturalists should look after in the explicating of difficult phenomena is not so much what the agent is or does, as what changes are made in the patient to bring it to exhibit the phenomena that are proposed, and by what means, and after what manner, those changes are effected: so that, the Mechanical philosopher being satisfied that one part of matter can act upon another but by virtue of local motion or the effects and consequences of local motion, he considers that as, if the proposed agent be not intelligible and physical, it can never physically explain the phenomena, so, if it be intelligible and physical, it will be reducible to matter and some or other of those only catholic affections of matter already often mentioned.
According to Boyle, then, what an ‘inquisitive person’ should do in offering an explanation of how an outcome such as death by poison comes about is to describe the series of changes that led from the event that counts as the cause to the resulting outcome. Moreover, such an explanation should be given in mechanical terms. The reason is that only by means of local motion can we understand how a cause (i.e., a part of matter) can operate on something (i.e., another part of matter). So, non-physical causes cannot explain (since they do not act by means of local motion), and for physical causes to be explanatory, they have to be reduced to matter and to the ‘catholic affections’ of matter, including motion. The requirement that an adequate explanation has to state the means by which the cause acts is here supplemented by the further requirement that the account of how exactly it acts has to be given in mechanical terms. We can view the resulting account of explanation as a combination of a methodological thesis (i.e., that an explanation should state how exactly the cause acts) with an ontological one (i.e., that it should be given in mechanical terms). Boyle goes on:
whatever be the physical agent, whether it be inanimate or living, purely corporeal or united to an intellectual substance, the above-mentioned changes, that are wrought in the body that is made to exhibit the phenomena, may be effected by the same or the like means, or after the same or the like manner … And if an angel himself should work a real change in the nature of a body, it is scarce conceivable to us men how he could do it without the assistance of local motion, since, if nothing were displaced or otherwise moved than before (the like happening also to all external bodies to which it related), it is hardly conceivable how it should be in itself other than just what it was before.
Boyle argues here that since real change requires local notion, a mechanical explanation can always be given no matter what the exact nature of the agent is. Even if the agent is immaterial, to produce a certain outcome is to produce a series of changes in local motion. This leads, finally, to the following point:
From the foregoing discourse it may (probably at least) result that if, besides rational souls, there are any immaterial substances (such as the heavenly intelligences and the substantial forms of the Aristotelians) that regularly are to be numbered among natural agents, their way of working being unknown to us, they can but help to constitute and effect things, but will very little help us to conceive how things are effected: so that, by whatever principles natural things be constituted, it is by the Mechanical principles that their phenomena must be clearly explicated.
Boyle here points out that even if substantial forms were to be accepted, they (as well as other immaterial substances) are useless if our aim is to understand how the phenomena are brought about. The thought here is that, in order to explain phenomena, we have to explain how they come about; but we can do this only by stating how the various causal agents produce changes in the properties of parts of matter by means of local motion; we cannot conceive how substantial forms can result in local motion; so, substantial forms are useless in offering explanations of how phenomena are produced.
In sum, Boyle’s main thought is that a satisfactory causal explanation has to explain the way a cause acts in bringing about a certain phenomenon. To do this, a causal explanation has to be mechanical, where a mechanical explanation explains how the effect is brought about in terms of changes in the mechanical properties of parts of matter, including local motion. As we will explain in Section 1.4, in giving our account of mechanism and mechanistic explanation, we will be in agreement with Boyle’s insights. But whereas we will keep his methodological thesis (i.e., that in giving a mechanistic explanation in science one has to explain how exactly a cause acts by describing the sequence of causal steps leading from the cause to the effect), we will reject the ontological thesis (i.e., that there exists a privileged description of this causal sequence, either in physicochemical terms or in terms of one’s favourite metaphysics of causal processes).
1.3.4 Newton against Mechanism
When Newton offered a non-mechanical account of gravity, he primarily challenged the idea that legitimate scientific explanation ought to be mechanistic, at least in the narrow sense of taking all action to be by contact. There is a sense in which Newton prioritised explanation by unification under laws and not by mechanisms. This is seen in the Preface to the second (1713) edition of the Principia, authored by Roger Cotes under the supervision of Newton. In this preface, Cotes presents Newton’s method as a middle way (via media) between Aristotelianism and Mechanism. To be sure, the mechanical explanations offered by the Cartesians were an improvement over the Scholastic explanations because they relied on demonstrations on the basis of laws. Still, taking ‘the foundation of their speculations from hypotheses’, the mechanists are ‘merely putting together a romance [i.e., fiction], elegant perhaps and charming, but nevertheless a romance’ (Newton Reference Newton and Janiak2004, 43).
Thus put, the point sounds epistemic; it concerns the increased risk involved in hypothesising a mechanism which is supposed to underpin, and hence to causally explain, a certain phenomenon. Cotes adds:
But when they [the mechanists] take the liberty of imagining that the unknown shapes and sizes of the particles are whatever they please, and of assuming their uncertain positions and motions, and even further of feigning certain occult fluids that permeate the pores of bodies very freely, since they are endowed with an omnipotent subtlety and are acted on by occult motions: when they do this, they are drifting off into dreams, ignoring the true constitution of things, which is obviously to be sought in vain from false conjectures, when it can scarcely be found out even by the most certain observations.
Still, it’s fair to say that Newton’s via media was based on a different understanding of scientific explanation: it should certainly look for causes – hence, scientific explanation should be causal – but the sought-after causes need not act by the principles of Mechanism. Newton’s way, Cotes says, is to ‘hold that the causes of all things are to be derived from the simplest possible principles’, but unlike the mechanists’ way, it ‘assume(s) nothing as a principle that has not yet been thoroughly proved from phenomena’. The ‘explication of the system of the world most successfully deduced from the theory of gravity’ is the ‘most illustrious’ example of Newton’s way (p. 32).
Newton emphatically denied feigning any hypotheses about the cause of gravity. For him, ‘it is enough that gravity really exists and acts according to the laws that we have set forth and is sufficient to explain all the motions of the heavenly bodies and of our sea’ (p. 92). Gravity according to Newton is a non-mechanical force since it ‘acts not in proportion to the quantity of the surfaces of the particles on which it acts (as mechanical causes are wont to do) but in proportion to the quantity of solid matter, and whose action is extended everywhere to immense distances, always decreasing as the squares of the distances’ (p. 92). He added that the very motion of the comets makes it plausible to think that the regular elliptical motion of the planets (as well as of their satellites) cannot ‘have their origin in mechanical causes’ (p. 90).
In his already mentioned Discourse on the Cause of Gravity (1690), Huygens expressed his dissatisfaction with Newton’s failure to offer a mechanical explanation of the cause of gravitational attraction. Favouring his own explanation of gravity in terms of the centrifugal force of the subtle and rapidly moving matter that fills the space around the Earth and the other planets, Huygens noted that Newton’s theory supposes that gravity is ‘an inherent quality of corporeal matter’. ‘But’, he immediately added, such a hypothesis ‘would distance us a great deal from mathematical or mechanical principles’ (Reference Huygens1690/1997, 35).
Yet Huygens had no difficulty in granting that Newton’s law of gravity was essentially correct when it comes to accounting for the planetary system. As he put it:
I have nothing against Vis Centripeta, as Mr. Newton calls it, which causes the planets to weigh (or gravitate) toward the Sun, and the Moon toward the Earth, but here I remain in agreement without difficulty because not only do we know through experience that there is such a manner of attraction or impulse in nature, but also that it is explained by the laws of motion, as we have seen in what I wrote above on gravity.
Explaining the fact that gravity depends on the masses and diminishes with distance ‘in inverse proportion to the squares of the distances from the centre’ (p. 37) was, for Huygens, a clear achievement of Newton’s theory despite the fact that the mechanical cause of gravity remained unidentified.
Commitment to mechanical explanation was honoured by Gottfried Wilhelm Leibniz too. In a piece titled ‘Against Barbaric Physics: Toward a Philosophy of What There Actually Is and against the Revival of the Qualities of the Scholastics and Chimerical Intelligences’ (written between 1710 and 1716), he defended the mechanical view by arguing that corporeal forces should be grounded mechanically when it comes to their application to the natural world. Leibniz was very clear that though he allowed ‘magnetic, elastic, and other sorts of forces’, they are permissible ‘only insofar as we understand that they are not primitive or incapable of being explained, but arise from motions and shapes’ (Leibniz Reference Leibniz, Ariew and Garber1989, 313). So, forces are necessary, but a condition for their applicability to the natural world is that they are seen as ‘arising from motions and shapes’. What he took it to be ‘barbarism in physics’ was to posit sui generis, that is non-mechanically grounded, ‘attractive and repulsive’ forces that act at a distance (pp. 314–15). Newton’s gravity was supposed to be such a barbaric force!
In a letter he sent to Nicolaas Hartsoeker (Hanover, 10 February 1711), Leibniz makes it clear that the proper scientific explanation should be mechanical. It is not enough for scientific explanation to identify the law by means of which a certain force acts; what is also required is the specification of the mechanism by means of which it acts. The mechanism is, clearly, on top of the law and given independently of it. Without the mechanism the power is ‘an unreasonable occult quality’. He says:
Thus the ancients and the moderns, who own that gravity is an occult quality, are in the right, if they mean by it that there is a certain mechanism unknown to them, whereby all bodies tend towards the center of the earth. But if they mean that the thing is performed without any mechanism by a simple primitive quality, or by a law of God, who produces that effect without using any intelligible means, it is an unreasonable occult quality, and so very occult, that it is impossible it should ever be clear, though an angel, or God himself, should undertake to explain it.
Newton couldn’t disagree more. In an unsent letter written circa May 1712 to the editor of the Memoirs of Literature, Newton referred explicitly to Leibniz’s letter to Hartsoeker and stressed that it is not necessary for the introduction of a power – such as gravity – to specify anything other than the law it obeys; no extra requirements should be imposed, and in particular no requirement for a mechanical grounding:
And therefore if any man should say that bodies attract one another by a power whose cause is unknown to us, or by a power seated in the frame of nature by the will of God, or by a power seated in a substance in which bodies move and float without resistance and which has therefore no vis inertiae but acts by other laws than those that are mechanical: I know not why he should be said to introduce miracles and occult qualities and fictions into the world. For Mr. Leibniz himself will scarce say that thinking is mechanical as it must be if to explain it otherwise be to make a miracle, an occult quality, and a fiction.
Note well Newton’s point. The fact that an explanation does not conform to a certain mechanical framework does not make it fictitious, occult or miraculous. Non-mechanical explanations are legitimate insofar as they identify the law that covers or governs a certain phenomenon. Hence, Newton promotes a methodological shift: causal explanation without mechanisms but subject to laws.
Causal explanation then need not be mechanical to be legitimate and adequate. This is Newton’s key thought. In breaking with a tradition which brought under the same roof a certain metaphysical conception of the world and a certain view of scientific explanatory practice, Newton distinguished the two and laid emphasis on the explanatory practice itself, thereby freeing it from a certain metaphysical grounding.
Though this is not the end of the story of Old Mechanism (more will be said in the next chapter), Newton’s key thought, we shall argue, is of relevance in the current debates over New Mechanism, to which we shall now turn our attention.
1.4 New Mechanism: From Practice to Metaphysics
It is useful to differentiate between two ways to conceptualise mechanisms in the post-1970 literature. First, mechanism has been used as a primarily metaphysical concept, mostly aiming to illuminate the metaphysics of causation. Second, mechanism has been taken to be a concept used in science, and philosophical accounts of mechanism have aimed to elucidate this concept.
To be sure, some philosophical approaches to mechanism, most notably Glennan’s (Reference Glennan1996), blend these two conceptions (the metaphysical one and the concept-in-use). However, it’s fair to say that there are two quite distinct points of origin of the recent philosophical accounts of mechanism: the first starts from metaphysics (as was the case for Descartes and other old mechanists), the second from scientific practice. Using this distinction between mechanism as a primarily metaphysical concept and as a concept-in-use in science, we can differentiate between two kinds of approaches to the metaphysics of mechanisms.
On the first approach, the aim is to show what the connection is between mechanism qua a metaphysical category and other central metaphysical concepts, notably, causation. In the context of the metaphysics of causation, ‘mechanistic’ accounts are theories about the link between cause and effect. Such theories are meant to be anti-Humean in that they view causation as a productive relation; that is, the cause somehow brings about or produces the effect. The aim of the mechanistic view of causation is to illuminate the productive relation between the cause and the effect by positing a mechanism that connects them and by explicating ‘mechanism’ in a suitable way such that causal sequences are differentiated from non-causal ones. The central thought, then, is that A causes B if and only if there is a mechanism connecting A and B.
Two kinds of views have become prominent: those that characterise the mechanism that links cause and effect in terms of the persistence, transference or possession of a conserved quantity (Mackie Reference Mackie1974; Salmon Reference Salmon1997; Dowe Reference Dowe2000) and those that connect a mechanistic account to causal production with a power-based one (see Harré Reference Harré1970 for an early such view). Despite their differences, these views share in common the claim that mechanisms are the ontological tie that constitutes Hume’s ‘secret connexion’. We call such mechanisms mechanisms-of. Mechanisms-of are ontological items that underlie or constitute certain kind of processes, that is, those that can be deemed causal. We will deal with these accounts in more detail in Chapter 5 (see also Psillos Reference Psillos2002).
On the second approach, working out a metaphysics of mechanisms is not the starting point but rather the end point of inquiry. Starting with mechanism as a concept-in-use in science, one tries first to give a general characterisation of this concept and then to derive metaphysical conclusions, that is, conclusions about the (mechanistic) structure of the world. This kind of bottom-up inquiry has yielded several well-known general accounts of mechanisms as well as theses about the ontic signature of a mechanistic world.
1.4.1 The Metaphysics of New Mechanism
Here are three well-known general characterisations of a mechanism in recent mechanistic literature:
Mechanisms are entities and activities organized such that they are productive of regular changes from start or set-up to finish or termination conditions.
A mechanism for a behavior is a complex system that produces that behavior by the interaction of a number of parts, where the interactions between parts can be characterized by direct, invariant, change-relating generalizations.
A mechanism is a structure performing a function in virtue of its component parts, component operations, and their organization. The orchestrated functioning of the mechanism is responsible for one or more phenomena.
The focus on mechanism as a concept-in-use is common to all three accounts; none of the three accounts can be viewed as falling under the rubric of mechanistic theories of causation. And yet, all these and similar accounts yield specific metaphysical commitments about what kind of things in the world mechanisms are. All these accounts are committed to the thesis that a general characterisation of mechanism must itself be cashed out in metaphysical terms.Footnote 3 Hence, talk of mechanisms in science is taken to have quite direct consequences about the kind of ontology presupposed by such talk. In order to substantiate this point, let us look at the three accounts mentioned earlier in some more detail.
Peter Machamer, Lindley Darden and Carl Craver’s (henceforth MDC) account is perhaps the most ontologically inflated, as it is explicitly committed to both entities and activities as distinct and separate ontological categories. It is thus committed to a particular view about the metaphysics of causation: causation within mechanisms is to be characterised in terms of production, where the productive relation is captured by the various different kinds of activities identified by science.
Glennan’s case is interesting, since in his Reference Glennan2002 article he refrains from taking mechanisms to entail a productive account of causation. Instead, within-mechanism interactions are characterised in terms of invariant, change-relating generalisations. As we will see below, however, Glennan has connected his account of mechanisms with a power-based understanding of causation. Hence, he is committed to causal powers as parts of the building blocks of mechanisms.
Last, Bechtel and Abrahamsen’s account does not include a specific characterisation of what mechanistic causation amounts to at all. Here, however, as in the other two accounts, we have a series of general terms, the meaning of which needs to be unpacked. So, MDC include in their accounts ‘entities’ and ‘organisation’; Glennan in his early formulations includes ‘complex system’ and ‘parts’; and Bechtel and Abrahamsen talk about ‘structure’, ‘function’, ‘parts’ and ‘organisation’.
All these accounts suggest the further need to explain what this ‘new mechanical ontology’ of entities, activities, organisation of parts into wholes and so on amounts to; what, in general terms, the constituents of mechanisms are and what their relations are to more traditional metaphysical categories, such as things, properties, powers and processes.
Notably, there has been a tendency recently to offer a more minimal general characterisation of a mechanism. For example, according to Illari and Williamson:
A mechanism for a phenomenon consists of entities and activities organized in such a way that they are responsible for the phenomenon.
Glennan’s recent version is almost identical:
A mechanism for a phenomenon consists of entities (or parts) whose activities and interactions are organised so as to be responsible for the phenomenon.
Glennan calls this account Minimal Mechanism. The key motivation here is for a general characterisation of mechanism broad enough to capture examples of mechanisms in different fields, from physics to the social sciences. But even in this minimal mode, mechanisms, according to Glennan, ‘constitute the causal structure of the world’ (Reference Glennan2017, 18).
This minimal account of mechanism might appear to fit the bill of capturing a concept-in-use in science. On closer inspection, however, it is committed to a rather rich metaphysical account of mechanism: the minimal account is not more minimal than the metaphysically inflated accounts noted above. The reason is that both of the foregoing minimal accounts still invite questions about the ontic status of mechanisms. For example, how exactly do entities and activities differ? What is the relation between activities and interactions? How should organisation be understood? Glennan (Reference Glennan2017, 13) explicitly talks about a ‘new mechanical ontology’ as the upshot of the minimal account. The ‘minimal mechanism’, he adds, ‘is an ontological characterization of what mechanisms are as things in the world’ (p. 19).
New Mechanism, then, aims to provide a new ontology of mechanisms. We can identify three commonly accepted key theses concerning mechanical ontology:
(1) The world consists of mechanisms.
Thesis 1 is a typical view among mechanists: mechanisms are taken to be things in the world, with (more or less) objective boundaries.Footnote 4 Ours is a mechanistic world. As Glennan puts it, ‘[t]hat is just how we have found the world to be’ (Reference Glennan2017, 240).
(2) A mechanism consists of objects of diverse kinds and sizes structured in such a way that, in virtue of their properties and capacities, they engage in a variety of different kinds of activities and interactions such that a certain phenomenon P is brought about.
Thesis 2 (or something very similar) can be taken as the common core of the general characterisations of mechanism as a concept-in-use given by new mechanists. It identifies the components of a mechanism and the relations among them. As mechanisms are things in the world (thesis 1), their components are also particular things in the world. Besides, these parts engage in activities by being ‘active, at least potentially’ (Glennan Reference Glennan2017, 21). Activity is understood as a manifestation of the powers things have. Glennan is quite explicit that ‘Activities manifest the powers (capacities) of the entities involved in the Activity’ (p. 31). Positing powers is supposed to explain why ‘activities are powerful’; being powerful, activities are what ‘an entity does, not merely something that happens to an entity’ (p. 32). But activities are not enough. Interactions are needed too because ‘there is no production without interaction’ (p. 22). ‘The fundamental point of ontological agreement among the New Mechanists’, as Glennan (Reference Glennan2017, 21 n. 6) puts it, is that entities cannot exist without activities or activities without entities. It’s not hard to see that the minimal account of mechanism is taken to imply or suggest a rather substantive metaphysical conception of mechanism, which, until further notice at least, is broadly neo-Aristotelian.
(3) The main way to explain a certain phenomenon P is to offer the mechanism that produces it.
Thesis 3 connects the previous theses with a claim about explanation (and more specifically, causal explanation): since in a mechanistic world phenomena are produced by mechanisms, the main task of scientific explanation is to identify the mechanism that produces a certain phenomenon, that is (by thesis 2), to identify the organised entities and activities that produce the phenomenon.
Despite their differences, there are important similarities between Old and New Mechanism (which justify viewing both positions as mechanistic). On the one hand, as we saw, new mechanists differ from their seventeenth-century predecessors in that they do not start their analysis with a metaphysical concept of mechanism; rather, they aim at giving a general characterisation of mechanism as a central concept of scientific practice. This characterisation is non-reductive in that it is not committed to the view that mechanisms are configurations of matter in motion subject to laws (and contact action). But, on the other hand, they are committed to mechanisms being configurations of powerful entities engaged in activities and interactions. As Glennan puts it: ‘Mechanisms are particular and compound, made up of parts (entities) whose activities and interactions are located in particular regions of space and time’ (Reference Glennan2017, 57). Hence, New Mechanism is similar to seventeenth-century Mechanism, in that it is committed to a mechanical ontology. This ontology (theses 1 and 2 above), while not a global metaphysics in the sense of the seventeenth century, is still a thesis about the ontic signature of the world. Here is Glennan again: ‘New Mechanist ontology is an ontology of compound systems. It suggests that the properties and activities of things must be explained by reference to the activities and organization of their parts’ (p. 57). Instead of resulting in a ‘flat’ ontology where everything there is consists in matter in motion, this new mechanical metaphysics ends up with a hierarchy of particular things – mechanisms – which may contain a diverse set of entities and activities, rather than the limited set endorsed by the corpuscularians, and whose productivity is grounded in causal powers, rather than in a few fundamental laws of motion.
But we can ask: Are these ontological commitments really necessary in order to understand scientific practice? Are they licensed by the practice of science? Remember here that the primary aim of new mechanists is to give a general characterisation of mechanism as a concept-in-use. So, ideally, the general account of mechanism should capture as far as possible the extension of a concept-in-use in the various sciences. The minimal account of mechanism discussed so far, though broad enough to play this role, inflates the concept-in-use by making it amenable to a certain metaphysical description of its basic components.
Note that our claim is not that the metaphysical questions are not philosophically interesting questions to ask; they are, especially if we are interested in giving an account of the ontological structure of reality. Moreover, such a kind of project has to be informed by what science has to say about the world. If, however, our aim is to understand how a specific concept – mechanism – is used in scientific practice, these questions seem, at least prima facie, irrelevant, especially if a general characterisation of mechanism is possible that does not include such things.
1.4.2 Mechanism in Scientific Practice
A metaphysically deflationary view of mechanism as a concept-in-use that is broad enough to capture all examples of mechanisms that we find in science seems indeed possible. This is skinnier than those accounts of mechanism offered by Illari and Williamson and by Glennan. We nonetheless claim that this skinny account is enough to capture the concept-in-use. It will be the main aim of Chapter 4 to present this account in detail, and the subsequent chapters will further illuminate various features of the view. Here we will just introduce the basic idea behind our account. This skinny or, as we will prefer to describe it, deflationary account of mechanism is achieved by dropping the reference to activities and interactions and by understanding mechanism as the causal pathway of a certain phenomenon, described in the language of theory. According to this account that we call Causal Mechanism (CM), a mechanism in science just is a causal pathway described in theoretical language:Footnote 5
(CM) A mechanism is a theoretically described causal pathway.
The central idea behind CM stems from the Boylean insight introduced in Section 1.3.3: when scientists talk about a ‘mechanism’, what they try to capture is the way (i.e., the causal pathway) a certain result is produced. Say, for example, that a pathologist tries to find out how a certain disease state is brought about. They will look for a specific mechanism, that is, a causal pathway that involves various causal links between, for example, a virus infection and changes in properties of the organism that ultimately lead to the disease state. In pathology such causal pathways are referred to as the ‘pathogenesis‘ of a disease, and when pathologists talk about the ‘mechanisms‘ of a disease, it is such pathways that they have in mind (see Lakhani et al. Reference Lakhani, Dilly and Finlayson2009).
According to CM, then, mechanisms and causation are closely related: when two events are causally connected, there is a mechanism (i.e., a causal pathway) that connects them and accounts for the specific way that the cause brings about the effect. Scientists succeed in identifying a mechanism, if they succeed in describing the relevant causal pathway in terms of the theoretical language of the particular scientific field. An especially clear example of the identification of a new mechanism is the case of the mechanism of cell death known as apoptosis; we will examine this case in detail in the next chapter.
The view of mechanisms as causal pathways differentiates CM from accounts that explicitly view mechanisms as complex systems (Glennan Reference Glennan1996), kinds of structures (Bechtel & Abrahamsen Reference Bechtel and Abrahamsen2005) or more generally as organised entities of some sort; it doesn’t differentiate it from more processual views, such as the MDC account. CM stresses that mechanisms are not systems, but causal processes. It is therefore closer to the older Salmon-Dowe view, as well as to the Boylean conception sketched above, than some more recent accounts.
There exists, however, a very important difference between Boyle’s notion of mechanical explanation and CM. As we saw earlier, for Boyle and other mechanical philosophers of the seventeenth century, mechanical explanations had to be couched in very specific terms, that is, in terms of the changes produced by parts of matter to the ‘mechanical affections’ (including motion) of other parts of matter. Thus, the methodological claim of mechanical philosophers – that is, that to explain how the phenomena are produced one should identify the mechanisms that produce them – did incorporate a claim about the specific theoretical language that such explanations should be couched in. And the main justification of this latter claim was ontological: what really exists in the world is matter that behaves according to certain laws that govern its motion. Old Mechanism, then, combined a methodological claim about the preferred form of scientific explanation with an ontological claim, that is, a claim about how the world is constructed.
In contrast to this more restricted way to conceive of mechanistic explanation, according to CM there is no privileged theoretical language in terms of which the causal pathway that produces the phenomena has to be described. This is, in one sense, in agreement with the dominant views of what a mechanism is: as both MDC (Machamer et al. Reference Machamer, Darden and Craver2000) and Glennan (Reference Glennan1996) stress in the papers that first offered general characterisations of the notion of mechanism, the contemporary concept is more general than its seventeenth-century counterpart, as the parts of a mechanism interact in various ways (e.g., by chemical interactions) and thus are not ‘mechanisms’ in the restricted seventeenth-century sense of the term.
In another sense, however, CM is different from current accounts, as it stresses that there is no privileged ontological description of a mechanism. So, while current accounts combine the methodological claim that science should discover the causal pathways that produce the phenomena with an ontological claim about the metaphysics of mechanisms, CM deflates the metaphysics and puts the methodological claim at the centre. We shall further examine this feature of CM by revisiting Newton’s views. Before this, however, let us dispel a natural but pointed objection. Recall our main thesis that ‘A mechanism is a theoretically described causal pathway.’ Does this invite the interpretation that a mechanism does not exist before it is theoretically described? We, of course, do not believe this; our view is that mechanisms just are independently and objectively existing causal processes. The aim of the phrase ‘theoretically described’ is to highlight that the causal processes that constitute mechanisms are to be described in the theoretical terms of the relevant scientific domain and not in terms of general ontological categories.
1.5 Newton Revisited
What does Newton’s critique of the Old Mechanism have to do with our understanding (and criticism) of New Mechanism? In a letter to Leibniz dated 16 October 1693, Newton challenged him to offer a mechanical explanation of ‘gravity along with all its laws by the action of some subtle matter’ and to show that ‘the motion of planets and comets will not be disturbed by this matter’. If this were available, Newton said, he would be ‘far from objecting’. But no such explanation was forthcoming and Newton was happy to reiterate his view that ‘since all phenomena of the heavens and of the sea follow precisely, so far as I am aware, from nothing but gravity acting in accordance with the laws described by me; and since nature is very simple … all other causes are to be rejected’ (Newton Reference Newton and Janiak2004, 108–9). Newton does not simply say that causal explanation might not be mechanical. His point is that causal explanation should be liberated from the tenets of (the narrowly understood) Old Mechanism. It would not be enough to offer a mechanical account of the cause of gravity; the laws that gravity obeys should be mechanically explicable, and, as Newton repeatedly stressed, this was not forthcoming. Though causal explanation matters, it doesn’t matter if it is subject to various (old) mechanical constraints.
We noted already that the new mechanical conception of nature is far from the seventeenth-century conception that everything should be accounted for in terms of (configurations of) matter in motion. So it’s far from us to tar New Mechanism with the same brush as Old Mechanism. For instance, the key ontology of the old mechanical picture was justified by and large a priori, whereas the key ontology of New Mechanism is grounded in scientific practice; in this case, it is practice that constrains metaphysics. Be that as it may, we are now going to argue that there exists a kind of Newtonian move against New Mechanism too.
What is clear from the present discussion is that, regardless of the main difference noted above, the new idea of mechanism is no less metaphysically loaded than the old one. Where the seventeenth-century mechanists looked for stable arrangements of matter in motion subject to laws, the twenty-first-century mechanists look for stable arrangements of powerful entities engaged in various activities and interactions. These mechanisms are supposed to be the building blocks of nature, and the scientific task is to unravel them. They underpin ‘mechanistic explanations’ which, as Glennan put it, show ‘how the organized activities and interactions of some set of entities cause and constitute the phenomenon to be explained’ (Reference Glennan2017, 223). Mechanistic explanation ‘always involves characterizing the activities and interactions of a mechanism’s parts’ (p. 223). Where the seventeenth-century mechanists saw ‘action by contact’ as a requisite for a proper mechanical explanation, new mechanists see powers and ‘activities’.
Why is Newton’s key thought relevant to the modern debates about mechanisms? The key thought, to repeat, was that causal explanation should identify causes and the laws that govern their action irrespective of whether or not these causes can be taken to satisfy further (mostly metaphysically driven) constraints. In other words, Newton showed that certain causal explanations of phenomena (in terms of non-mechanical forces) are both legitimate and complete insofar as they identify the right causes and are empirically grounded.
We take it that the point CM stresses, is, mutatis mutandis, analogous to Newton’s. The point of CM is that causal explanation need not be mechanistic in the new mechanists’ ontic sense and that being couched in the way new mechanists propose, causal explanation is subjected to constraints unwarranted by scientific practice. Insofar as mechanism is a concept-in-use in science, it may well be seen referring to the causal pathway of the phenomenon to be explained, couched in the language of theories. Preserving the spirit of Newton’s key thought, we might say that causal explanation is legitimate even if we bracket the issue of ‘what mechanisms or causes are as things in the world’ (Glennan Reference Glennan2017, 12) or the issue of what activities are and how they are related to powers and the like. The issue then is not ‘an ontological characterization of what mechanisms are as things in the world’ (p. 19), but a methodological characterisation of them as causal pathways described in the language of theories.
To press the analogy a bit more, questions such as ‘If entities, activities, and the mechanisms they constitute are compounds, of what are they compounded? Where does one entity or activity or mechanism end, and when does another begin? And on what account do we decide that a collection of interacting entities is to count as a whole mechanism?’ (p. 29) are pretty much like the questions concerning the cause of the properties of gravity that Newton thought need not be asked and answered for a scientifically legitimate conception of causal explanation.
We do not want to claim that questions such as the above are not connected to scientific practice. After all, even the question of the cause of gravity that Newton refrained from answering was connected to scientific practice. The point, rather, we take from Newton is that answering these questions is not required for offering adequate causal explanations of the phenomena under study. Similarly, for CM, answering questions such as the above is not required in order to have legitimate mechanistic explanations. In other words, the properties of mechanism over and above those that are required by its methodological use need not be specified; nor is there an explanatory lacuna if they are not.
According to CM, the concept of mechanism as used in practice need not, and should not, be understood in a metaphysically inflated sense. Hence, new mechanists, in offering such metaphysically inflated accounts, need to show that such accounts are indeed indispensable for doing good mechanistic science. To conclude, as Newton remained agnostic about the underlying mechanism of gravity, so CM remains agnostic about the metaphysical ground of any particular causal pathway. As in the case of gravity, it is enough that mechanisms qua causal pathways really exist and act as they do.
