Out of Nowhere: Introduction: The emergence of spacetime
CChapter 1: Introduction: The emergence of spacetime
Nick Huggett and Christian W¨uthrich ∗
18 January 2021
Contents “Big Bang Machine Could Destroy Earth” . . . ran an attention grabbing headline in
The Sunday Times (Leake 1999), regarding the newRelativistic Heavy Ion Collider (RHIC) at Brookhaven National Laboratory. To be fair, the mainapocalyptic concern of the paper was that the RHIC would create a form of matter in which strangequarks eat the up and down quarks found in ordinary matter. But it also discussed the possibilitythat experiments involving high-energy collision of gold ions could create microscopic blackholes,which would pull all the matter in the world into them. Such scenarios were taken seriously enoughthat they were evaluated by a panel of elders, who concluded that the chances of any such eventswere utterly minuscule (Busza et al. 1999)—happily, up to the time of writing, they have not beencontradicted by events at Brookhaven! ∗ This is a chapter of the planned monograph
Out of Nowhere: The Emergence of Spacetime in Quantum Theoriesof Gravity , co-authored by Nick Huggett and Christian W¨uthrich and under contract with Oxford University Press.More information at . This work was supported financially by the ACLS and the JohnTempleton Foundation (the views expressed are those of the authors not necessarily those of the sponsors. a r X i v : . [ phy s i c s . h i s t - ph ] J a n et’s look into the business of black hole formation more carefully to explain why physics needsan account of quantum gravity. First, the RHIC was built to probe how matter behaves underintense temperatures and pressures—in effect recreating in a tiny region the state of the universewithin the first second of its existence, when quarks and gluons flowed in a plasma rather thanbinding to form particles. The predictions tested here are largely those of quantum chromodynam-ics, the quantum theory of the strong force binding nucleons and their constituents. That is, thecollisions between heavy nuclei such as gold in the RHIC are governed by the laws of quantummechanics (QM).The concern over black holes, however, arises when one asks how general relativity (GR)—theclassical, non-quantum, theory of gravity, gets into the picture. According to GR, the spacetimemetric outside a sphere of mass M takes the form:d s = c d t (1 − GMrc ) − d r / (1 − GMrc ) , (1)where G is Newton’s gravitational constant, c is the speed of light, and r the distance from the center.(d s is the infinitesimal spacetime ‘distance’ squared—the ‘interval’—between two radial pointsseparated in time by d t and space by d r , in suitable co-ordinates; though an exact understandingis not crucial here.) What is important is that this quantity blows up when 2 GM = rc , or r = 2 GM/c . Understanding this occurrence was an important issue in the early development ofGR, but what it actually signifies is the presence of an ‘event horizon’ around the mass, from whichneither matter nor light can escape—the boundary of a black hole. Of course this story only makessense if the mass is all located within a radius of 2 GM/c , since the metric formula only holdsoutside the mass. So one can equally well say that if one has a mass M it will only form a blackhole if it is all located within a radius less than 2 GM/c .So finally, the question posed by the panel at Brookhaven was how small a region would theamount of energy to be created in collisions be located (see page 7 of the report)? Acting cautiously,they assumed the best conditions for black hole formation, supposing that all the energy producedby the collision contributes to the mass: about 50 times that of a gold atom. For a black hole ofthis mass the event horizon has a radius of 10 − m. On the other hand, a gold atom has a radiusof around 10 − m, so even supposing that all the energy is concentrated in a region the size of asuitably Lorentz-contracted nucleus, general relativity predicts that collisions will be many, manyorders of magnitude from creating a black hole.Phew. What we have then is an argument that the physics of Brookhaven lies within the domain ofrelativistic QM, but that the gravitational effects of the collisions are utterly negligible. Perhapsthe world is just ‘dappled’ in this way: in some domains, such as the motions of the planets (GRexplains the perihelion of Mercury, for instance) GR holds and QM is irrelevant; in others, asin RHIC, it is QM that holds sway, with GR entering only to provide a background geometrydetermined by ambient bodies, not by the system under consideration. But the very argumenthere shows that it is possible to bring considerations from both theories to bear on a single system, Don’t be confused if you have read of black holes being created at RHIC. In fact what has (perhaps) beenobserved is the Unruh effect, which is formally equivalent to Hawking radiation, but does not involve black holes, butacceleration, which is in a sense indistinguishable from a gravitational force according to GR. See Nastase (2005). − to 10 − s quantum fields provide the energy whichdrives the expansion of the universe, according to the laws of GR.) In the second, Hawking radiationis predicted to occur around black holes as a QM effect resulting from the geometry of spacetimegiven by GR (Hawking 1974). Indeed, assuming that the local equivalence of acceleration andgravity—the ‘equivalence principle’—holds in the quantum domain, then RHIC does provide testsfor this physics, since the radiation produced by decelerating ions can be measured (see footnote1): under the assumption, it provides indirect tests of the overlap of gravity and QM.Thus, we say, the ultimate need for a theory that in some way unifies QM and GR—a theoryof quantum gravity (QG)— arises from the existence of phenomena in our universe in which thedomains of the two overlap. There are other, more theoretical, arguments for such a theory andconcerning the form it should take. For a critical evaluation of these, arguing that overlap is bestreason to seek quantum gravity, and that empirical considerations best dictate its form see Callenderand Huggett (2001); for further discussion W¨uthrich (2005). There are then good reasons for physicists to investigate QG. This book is predicated on theview that it also has an important call on the efforts of philosophers. Indeed, we would like thiswork to encourage, by example, our colleagues to be more adventurous in their choice of topics ofenquiry. Philosophy of physics, we suggest, has a tendency to look too much to the past, and to themetaphysics of well-established physics (and of course to internecine disputes): classical statisticalmechanics, classical spacetime theory and non-relativistic quantum mechanics are ‘so twentieth (oreven nineteenth) century’, and yet have a virtual lock on the discipline. While quantum field theory(QFT) is becoming a significant topic, that is still at least half a century behind the physics!We’re overstating things somewhat for effect here: of course, even old theories do face importantfoundational problems, and their consequences for our broader understanding of the world takeconsiderable elucidation. And of course it is unfair to suggest that no philosophers of physics showan interest in contemporary physics. Indeed, since we started writing this book, there has beenan explosion of interest in QG, especially amongst a younger generation of scholars, which we findvery exciting. Still, we do say that collectively the discipline pays insufficient attention to cuttingedge physics, and hope this book in some way serves as an impetus to greater engagement.Our point is not that novelty is good for its own sake, nor that philosophy is a ‘hand-maiden’,who should dutifully follow the fashions of physics. Rather, we are inspired by recent work in thehistory and philosophy of science to believe that it is central to the business of philosophy to engagewith developing physical theories—both because the search for philosophical knowledge must beresponsive to empirical discoveries, and because philosophy has important contributions to make tothe development of physics (and other sciences). We will discuss this point at greater length below( § § §
3) of the challenges to the very idea that something a seemingly fundamental as This situation may change if the technology improves enough over the next few years to carry out the experimentproposed by Bose et al. (2017). § § § §
8) we will give an overview of the different strategies one might take towardsquantizing gravity, to relate the different proposals that we will consider in the book.
All theories of QG are, to a large extent, speculative; some of the examples that follow are morespeculative than others. However, as we shall explain below, there are good reasons to think thatthey may still teach important lessons in the search for QG. The first three examples are the focusof the following chapters; the remaining two are also illuminating, but we have discussed themelsewhere. • Causal Set Theory (CST):
As we will see in chapters 2 and 3, CST makes liberal useof GR as a vantage point for its research programme. In fact, it takes its most importantmotivation from theorems stating that given the causal structure of a spacetime, its metricis determined up to a conformal factor. In other words, the causal structure determines thegeometry of a spacetime—but not its ‘size’. Taking this cue, CST posits that the fundamentalstructure is a set of elementary events which are locally finite, partially ordered by a basiccausal relation. In other words, the fundamental structure is a causal set . The assumption oflocal finitarity is nothing but the formal demand that the fundamental structure—whateverelse it is—is discrete. Together with the demand of Lorenz invariance at the derived level, thediscreteness of causal sets forces a rather odd locality structure onto the elementary events ofthe causal set ( § § • Loop Quantum Gravity (LQG):
LQG starts out from a Hamiltonian formulation of GRand attempts to use a recipe for cooking up a quantum from a classical theory that has beenutilized with great success in other areas of physics. This recipe is the so-called canonicalquantization . The goal of applying the canonical quantization procedure is to find the phys-ical Hilbert space, i.e. the space of admissible physical states, and the operators defined onit that correspond to genuinely physical quantities. As will be seen in chapters 4-5, followingthis recipe leads rather straightforwardly into a morass of deep conceptual, interpretative,and technical issues concerning the dynamics of the theory as well as on time quite generally.We find that the states in the ‘kinematic’ Hilbert space afford a natural geometric interpreta-tion: its elements are states that give rise to physical space, yet are discrete structures with a4isordered locality structure. At least in one basis of this Hilbert space, the states appear tobe states of a granular structure, welding together tiny ‘atoms’ of space(time). It is crucial tothis picture that these atoms of space(time) are atoms in the original meaning of the word:they are the truly indivisible smallest pieces of space(time). The smooth space(time) of theclassical can thus be seen to be supplanted by a discrete quantum structure. Moreover, sincegenerically a state of this structure will be a superposition of basis states with a determinategeometry, generic states will not possess determinate geometric properties. If continuity, lo-cality, or determinate geometry was an essential property of spacetime, then whatever thefundamental structure is, it is not spacetime. In this sense, spacetime is eliminated from thefundamental theory. • String Theory:
According to string theory (chapter 6) tiny one-dimensional objects movearound space, wiggling as they go—the different kinds of vibration correspond to differentmasses (charges and spins) and hence to different subatomic particles. So it sounds as ifspacetime is built into the theory in a pretty straight-forward way; however, we will see thatthings are not so simple. First, in chapter 7 we will see that various versions which are intu-itively very different in fact correspond to the same physics—are ‘ dual ’. For instance, supposethat at least one of the dimensions of space is ‘compactified’, or circular. Then it turns outthat a theory in which the circumference of the dimension is C has the same collection ofvalues for physical quantities as a theory in which the circumference is 1 /C : i.e., that theoriesin which compactified dimensions are small are physically indistinguishable from—or ‘dual’to— those in which they are large. Other dualities relate spaces of different topologies. Thesefacts raise important questions about whether the space in which the string lives is the one weobserve, since that is definitely large though the string space could be small, or even conven-tional. Second, in chapters 8-9, we will see how the geometrical structure of spacetime—andindeed GR—arises from the behavior of large collections of strings in a ‘graviton’ state, thequantum particle that mediates gravitational forces. • Group Field Theory:
Consider rotations, by and angle θ , in the plane: a different one foreach value of 0 ≤ θ < π . These form a group under composition: a rotation by α followedby an angle β is just a rotation by α + β (and for example, a rotation by 2 π − α undoes arotation by α ). Similarly for rotations in three dimensions, and indeed similarly for Lorentztransformations (which are in fact nothing rotations in Minkowski spacetime), and so on. Foreach, composition yields a different function from any pair to third. We can thus characterizethe abstract group structure simply by the action of this function of an entirely arbitrary setof elements—forget that we started with rotations, and let the elements be anything, with acomposition rule isomorphic to that of the rotations. One could take then a set of such blankelements, which compose like rotations in the plane, but which should not be thought of asliteral rotations: no plane at all is postulated, the only manifold is that formed by the groupelements themselves—the circle of points with labels 0 − π , not a 2-dimensional plane. Andfinally one can introduce a field on this group manifold, a real number for each group element,with a dynamical law for its evolution; and indeed quantize this field. As we have emphasized,while physical space was used to guide the construction of this theory, it is not an explicit More accurately, the space that is assumed is not the space of relativistic physics, or that of planar rotations,but four copies of the group of Minkowski rotations. • Non-Commutative Geometry:
Picture a Euclidean rectangle whose sides lie along twocoordinate axes, so that the lengths of its sides are x and y —its area is x · y = y · x . Butwhat if such products fail to commute, xy (cid:54) = yx , so that area is no longer a sensible quan-tity? How can we understand such a thing—a non-commutative geometry ? By abandoningordinary images of geometry in terms of a literal space (such as the plane) and presenting itin an alternative, algebraic way. In fact, our example already starts to do so: even thinkingabout areas as products of co-ordinates uses Descartes’ algebraic approach to Euclidean ge-ometry. Once we have entered the realm of algebra, all kinds of possible modifications arise.Especially, an abstract algebra A requires an operation of ‘multiplication’, (cid:63) , but this can bea quite general map from pairs of elements, (cid:63) : A × A → A , which need not be commutative !For instance, one could define ‘multiplication’ to satisfy x (cid:63) y − y (cid:63) x = θ (a small number),and use it to generate polynomials in x and y ; these carry geometric information. Such athing is perfectly comprehensible from the abstract point of view of algebra, but it cannot begiven a familiar Euclidean interpretation via Cartesian geometry. So, such a theory seems todescribe a world that is fundamentally algebraic, not spatial (in the ordinary sense)—thereis x (cid:63) y and y (cid:63) x but no literal rectangle. If there is thus fundamentally nothing ‘in’ space,is the ultimate ontology ‘structural’, based on algebraic relations only? And how could anappearance of familiar (commutative!) space arise; especially, what significance could point-valued quantities have? We will return to this example in § inexplicability of X from Y: as some have claimed life or mind emerges from matter.On the contrary, we argue that classical spacetime structures can be explained in more fundamentalterms: indeed, it was largely to explicate how physicists do so that we wrote the book. Some mightthen say that spacetime ‘reduces’ to non-spatiotemporal QG, but we prefer to stick with the notionsof ‘explanation’ or ‘derivation’, because there are many notions of ‘reduction’, some of which aretoo strict. But we are also happy to speak of (weak) ‘emergence’ even when spacetime is derived,because the gulf between a theory that does not assume spacetime, and one that does is so great.Having spacetime or not makes a huge formal and conceptual difference, in particular because inalmost all theories prior to QG classical spacetime has apparently been one of the most basic posits.Indeed, this very gulf makes one wonder what it could mean to derive spacetime, and whether it ispossible at all.Before we proceed, we need to introduce some terminology to keep the discussion straight. Theissue is that the theories of QG often contain some object referred to as ‘space’, even when they donot assume ‘space’ in the ordinary sense. For instance, there may be a ‘Hilbert space’, or a ‘dualspace’, or ‘Weyl space’, or ‘group space’. So we will refer to spacetime in the ordinary sense as6classical’, or ‘relativistic’, or sometimes just ‘space’ or ‘spacetime’ when the context makes mattersclear. (We eschew the phrase ‘physical space’, since the other ‘spaces’ may well be part of thefundamental physical furniture. We have previously used ‘phenomenal space’, to indicate that clas-sical space is that of observable phenomena, according to the physicist’s use of ‘phenomenological’.However, this leads to confusion with the philosophical doctrine of ‘phenomenalism’, so we havedropped it.)By classical or relativistic spacetime, we mean that theorized in QM (especially QFT) andrelativity, approximated in non-relativistic mechanics, and ultimately implicated in our observationsof the physical world. As stated, that is not an entirely homogeneous concept, so we will say morelater ( §
6) about exactly what features of classical spacetime are emergent from our theories of QG.First, we turn to some challenges to the project of deriving spacetime.
Space and time are so basic to both our manifest and scientific images of the world that at first themind boggles at the thought that they might be mere ‘appearances’ or ‘phenomena’, of some deeper,more fundamental, non-spatiotemporal reality. Is a physics without spacetime even intelligible?And if it is, is spacetime the kind of thing whose existence could be explained? At its core, thisbook seeks to address these questions: on the one hand explicating the worlds described by theoriesof QG, while on the other showing how spacetime can be derived from them. But to understandthe nature and methodology of that project, it is important here to unpack the mind boggling,vertiginous panic about the very idea. Larry Sklar (1983) gave expression to this all too commonsentiment among philosophers (and physicists) when he wrote What could possibly constitute a more essential, a more ineliminable, component of ourconceptual framework than that ordering of phenomena which places them in space andtime? The spatiality and temporality of things is, we feel, the very condition of theirexisting at all and having other, less primordial, features. A world devoid of color, smellor taste we could, perhaps, imagine. Similarly a world stripped of what we take to beessential theoretical properties also seems conceivable to us. We could imagine a worldwithout electrical charge, without the atomic constitution of matter, perhaps withoutmatter at all. But a world not in time? A world not spatial? Except to some Platonists,I suppose, such a world seems devoid of real being altogether. (45)According to Sklar, a non-spatiotemporal world is inconceivable, and thus presumably not evenmetaphysically possible, let alone physically. This monograph is concerned with establishing thepossibility of a fundamentally non-spatiotemporal world, articulating the consequences of such apossibility, and defending the idea that spacetime may be merely emergent in a perfectly acceptablescientific explanation of the manifest world. So in this section we will discuss various more preciseways that one might doubt the possibility of deriving spacetime. A note on terminology: we take ‘Platonists’ to be committed to the existence of abstract entities, such aspropositions, sets, love, and justice, but also to the existence of the concrete, physical, and spatiotemporal world.Those who maintain that our world is fundamentally mathematical in nature and thus entirely consists in ultimatelyabstract entities or structures are often labelled as ‘Pythagoreans’. Since we are interested not in whether there existabstracta, but in the possibility that all physical existence is grounded in non-spatiotemporal structures, we will referto those who maintain that a fundamentally non-spatiotemporal physical world is not devoid of “real being”—nodoubt historically inaccurately—as
Pythagoreans . We take this Pythagoreanism to be Sklar’s target—and the oneof this monograph. N particles is not a function in ordinary space, but of thepositions of all the particles: Ψ( x , y , z ; x , y , z ; . . . ; x N , y N , z N ). Thus Ψ lives in ‘configuration’space, in which there are three dimensions for each particle. Albert (1996) has argued that we shouldtake the wavefunction ‘seriously’ as the ontology of the theory, and conclude that configurationspace is more fundamental than regular space—that the three dimensions of experience are mereappearances of the 3 N dimensions of reality. Whatever the merits of that view, the general ideahas been attacked by Tim Maudlin (2007). In particular he argues as follows: one mightderive a physical structure with the form of local beables from a basic ontology thatdoes not postulate them. This would allow the theory to make contact with evidencestill at the level of local beables, but would also insist that, at a fundamental level,the local structure is not itself primitive. ... This approach turns critically on whatsuch a derivation of something isomorphic to local structure would look like, wherethe derived structure deserves to be regarded as physically salient (rather than merelymathematically definable). Until we know how to identify physically serious derivativestructure, it is not clear how to implement this strategy. (3161)We have italicized the key phrase here. Suppose that one managed to show formally thatcertain derivative quantities in a non-spatiotemporal theory took on values corresponding to thevalues of classical spatiotemporal quantities; one would then be in a position to make predictionsabout derived space. However, according to the passage quoted, such a derivation (even if thepredictions were correct) would not show that spacetime had been explained. In addition, we haveto be assured that the formally derived structure is ‘ physically salient ’. We agree with Maudlinthat physical salience is required of proper—one can say ‘explanatory’—derivations: otherwise onesimply has a formal, instrumental book-keeping of the phenomena. Indeed, we agree with him thatthe issue is particularly pressing in theories of emergent spacetime. But we think that it can beaddressed in QG: one of the goals of this book is to investigate the (novel) principles of physicalsalience for theories of QG, the principles whose satisfaction makes the derivations of spacetimephysically salient. In the following chapters we will look in detail at the derivations, to make clearthe assumptions and forms of reasoning that lie behind them. In the concluding chapter 10 wewill analyze what have learned, to start to explicate what makes a derivation of spacetime in QGphysically salient. 8ut to explain that project—and its relevance to philosophy—we need to unpack the very notionof physical salience, as we understand it. There is a subtlety about the way that Maudlin makes the point, however (which we did notclearly address in Huggett and W¨uthrich (2013)). For the target derived structure in itself isprima facie physically salient: it is the physical datum to which the more fundamental theory isanswerable. (Perhaps a more fundamental theory will show that some less fundamental theory isprofoundly confused; but more generally one expects that existing, well-confirmed theories havelatched onto some genuine physical structures, and that new, better theories will simply explainhow, by subsuming the old in some broad sense.) So in that sense there is really no question of thephysical salience of the ‘derived structure’—in the sense of the structure to be derived.Rather, Maudlin is talking about a formal derivation within a proposed new theory, and thequestion of whether what is at present simply a mathematical structure, in numerical agreementwith the target structure, in fact explains it, and isn’t merely an instrument for generating pre-dictions. We would break this question down into two interconnected parts (which will also helpilluminate what is involved in explanation here). First, the question of whether and how the basalobjects or structures of the more fundamental theory accurately represent physically salient ob-jects and structures. As we shall see shortly, that question becomes far more pressing when noneof the putative objects or structures are supposed to be in spacetime. Second, does the formalderivation of the phenomenal from the more fundamental make physical sense? That the derivationexists shows that it makes sense at the level of the formalism, and especially that the derivation iscompatible with the mathematical laws. But, as Maudlin suggests, there is more to the questionof physical salience than that. And the question is especially pointed when one wonders how thespatiotemporal could ever be ‘made’ of the non-spatiotemporal. We will illustrate these ideas witha homely (and idealized in many ways) example. The ideal gas law tells that for a gas (in a box of fixed volume) pressure ∝ temperature. Ideal gastheory says nothing about the microscopic composition of gases, so these are (among) the primitivequantities of the theory, operationalized via pressure gauges (relying on forces measured via Hooke’slaw for springs), and thermometers (so relying on the linear expansion with temperature of somesubstance). This is the phenomenon to be explained by the more fundamental theory, the kineticgas model, according to which the gas is composed of atoms with mass m , whose degrees of freedomare their positions and velocities. The latter can be expressed by a vector (cid:126)V , with 3 n components:for each of the n atoms that make up the gas, three components, to describe the speed with respectto each of the three dimensions of space. Each atom has a kinetic energy associated with its velocity(1 / m(cid:126)v ); the average kinetic energy is simply their sum, divided by n : denote this quantity We are grateful to Maudlin for conversations on this topic. We believe that we capture the essence of hisidea, even if we might differ in details; and especially regarding the depth of the problem in the case of spacetimeemergence. We do, however, want to point out an important difference between the cases of emergence from QGand from configuration space: in the latter, but not the former, there is a way to formulate the theory in 3-space(as single particle wavefunctions with a tensor product). Thus in QM (but not QG) Maudlin can argue that thederivation isn’t physically salient, because the formulation from which it is derived is unnecessary in the first place.That the crucial difference between QG on the one hand and QM (and GR) on the other lies in there being noalternatives translates into a different status for spacetime functionalism in the two cases has been argued by Lamand W¨uthrich (forthcoming). The following has also been discussed in Huggett (2018). ( (cid:126)V ) ≡ m(cid:126)V . (2)Now one computes the atoms’ momentum change (per second per unit area) resulting from theircollisions with the sides of the box: assuming that the collisions are elastic, and that the atomsare distributed evenly throughout the box and with respect to their velocities, one formally derivesthat P ( (cid:126)V ) ≡ n v m(cid:126)V . (3)Clearly the two quantities are proportional: P ( (cid:126)V ) ∝ T ( (cid:126)V ) , (4)which has the form of the ideal gas law. However (and despite the suggestive names, P and T ) wehave so far said nothing to justify identifying the quantities with the pressure and temperature ofthe ideal gas law; we have only noted a formal proportionality.From this example we can abstract the following schema:If (a) fundamental quantities X can be ‘aggregated’ into α ( X ) and β ( X ), such that (b) f ( α ( X )) = g ( β ( X )) follows from fundamental laws, then the law f ( A ) = g ( B ) relatingless fundamental quantities A and B is formally derived .The term ‘aggregated’ is supposed to be vague, in order to accommodate the many ways a deriva-tion might proceed. But the underlying idea is that the more fundamental theory has (many)more degrees of freedom than the less fundamental, and somehow the more fundamental must be‘summarized’ by the less, for example by averaging, or by coarse-graining.Maudlin’s claim is that formal derivability does not suffice to properly derive phenomena: inparticular, 3-dimensional space can be formally derived from the full 3 N -dimensional configurationspace, but for Maudlin that does not make it a plausible, more fundamental alternative to ordinaryspace. And more generally, one should worry that a merely formal condition does not distinguishinstrumental calculi from serious physical accounts. And indeed, further analysis of the derivationof the ideal gas law shows that considerations of physical salience are at play.In particular, P ( (cid:126)V ) is derived by assuming that the atoms are striking the sides of the box, andexerting a force there: so acting exactly at the place and in the way that would produce a readingon a pressure gauge. And T ( (cid:126)V ) is (according to the randomness assumption) the amount of energyin any macroscopic region of the box, say the location of the bulb of a thermometer: and collisionswith the bulb will transfer kinetic energy to the molecules of the thermometer, causing thermalexpansion. Imagine if instead that P ( (cid:126)V ) only referred to the center of the box; or if T ( (cid:126)V ) referredto a single atom in the box. Then the formal derivation would not be convincing. Or suppose thatinstead of the atomic gas model we imagined that a gas was a continuous object, whose degrees offreedom were somehow described by (cid:126)V , but not as the velocities of anything (certainly not atoms).Then the formal consequences of kinetic gas theory could still be taken to hold, but they would nolonger have the interpretation that they do in the kinetic gas model; the whole derivation wouldgo through, but its physical meaning would be obscure. In short, the reason, in addition to theirproportionality, that we find P ( (cid:126)V ) and T ( (cid:126)V ) convincing as pressure and temperature, and not justquantities following a similar law, is that they are spatiotemporally coincident with those quantities,10nd involve processes capable of producing the phenomena associated with those quantities. Thederivation is not merely formal, but also physically salient.Continuing our schema:A (non-instrumental) derivation of phenomena requires, in addition to a formal deriva-tion, that (c) the derivation have physical salience .Of course, this schema does not tell us what it is to be physically salient, but the example ofthe ideal gas above illustrates two very important aspects, spatiotemporal coincidence, and theaction of a physically accepted mechanism. And this observation immediately reveals the problemfor the emergence of spacetime, because such criteria simply cannot be satisfied by derivationsfrom non-spatiotemporal theories, because they are explicitly spatiotemporal criteria. For instance,it makes no fundamental sense in such a theory to even ask where a structure is. So if suchcriteria are a priori constraints on science, then the QG program, to the extent that it involvesnon-spatiotemporal theories, is in some serious trouble. However, a second example indicates thecontextuality of physical salience, and thereby the way in which QG can hope to achieve physicalsalience in its derivations.Figure 1: Descartes’ and Newton’s competing images of gravity. On the left is pictured Descartes’vortex model: each cell represents a ball of rotating matter, with lines to indicate the directionof rotation (e.g., those surrounding f, L, Y rotate about axes in the plane shown, while thosesurrounding D, F, S rotate about axes perpendicular to the plane). The bodies at the center of acell represent suns: S is ours. On the right is the diagram from Newton’s Proposition I.1 proofof Kepler’s equal areas law for a central force (essentially, conservation of angular momentum).All that matters is the direction of the force (towards the point S ), not any ‘hypothesis’ about itsnature. Ultimately Newton will apply the proposition to the case in which S is our Sun. (Publicdomain, via Wikimedia Commons and Google Books.)11onsider the competing Cartesian and Newtonian accounts of gravity, exemplified by illus-trations from the Principles of Philosophy (Descartes 1644) and the
Mathematical Principles ofNatural Philosophy (Newton 1726), respectively: see figure 1. On the one hand we have the vor-tices of Descartes, which aimed to provide a mechanical account of gravity, in terms of the motionsand collisions of particles. On the other there is Newtonian action at a distance, which allowed him(as in Proposition I.1) to formulate and use his mathematical principles. We pass over Newton’sown ambiguous attitude towards the causes of gravitation (his refusal to ‘feign hypotheses’ on theone hand, but his speculations in the
Optiks (Newton 1730) on the other). The point to whichwe draw attention is the controversy between the Newtonians and Cartesians regarding the needfor mechanical explanation. For the latter, Newton might have captured the effects of gravity ina formally accurate way, but offered no scientific explanation for the phenomena. For example,consider Leibniz’s clear statement to Clarke:If God would cause a body to move [round a] fixed centre, without any [created thing]acting upon it . . . it cannot be explained by the nature of bodies. For, a free bodydoes naturally recede from a curve in the tangent. And therefore . . . the attraction ofbodies . . . is a miraculous thing, since it cannot be explained by the nature of bodies.(
Leibniz-Clarke Correspondence in Clarke et al. (1956))We take Leibniz’s complaint to be exactly that Newton’s derivation of the phenomena lacks physicalsalience, because only mechanical causes are physically salient explanations of unnatural motions.Of course, the Newtonians were ultimately victorious, and this Cartesian condition of physicalsalience was replaced by one that allows action at a distance, because of the success of universalgravity, and the failure of mechanical alternatives, such as Leibniz’s. But that was not the end ofthe story: through the development and empirical success of electromagnetic theory, culminatingin the development of special relativity, action at a distance was again rejected, with contact actionreplaced by the demand for local field interactions—and hence the replacement of Newtonian gravitywith general relativity. Again, we understand this demand as a criterion of physical salience,required for more than merely formal accounts. But even that is not the end of the story, forquantum mechanics experimentally conflicts with that concept of locality, and so quantum non-locality must be accommodated in some way. (Hesse 1961 is a classic telling of this tale.)By now, three points are indicated by this story: first, questions of physical salience, here in theform of the principles of locality, are genuine, controversial components of scientific enquiry. Second,such principles are historically contingent, changing in step with major advances in physics. Third,such changes are ultimately settled by, and epistemically justified by, empirical success: one of thethings that we learn in a scientific revolution is a set of criteria of physical salience for explanationappropriate to the new domain of enquiry. Put this way, we see principles of physical salienceas part of what Kuhn called the ‘disciplinary matrix’ in the
Postscript to the second edition of(1962), or what Friedman (2001) refers to as the ‘relative, constitutive a priori’. Though changes inthe principles change wholesale what theories are even candidate explanations, we don’t infer anycatastrophic incommensurability here: as we said, innovations in physical salience are grounded inempirical success, like all other scientific knowledge.So we have a general answer to the problem raised earlier.
How can a derivation of spacetimefrom a non-spatiotemporal theory ever be physically salient?
Well, it cannot satisfy the standards Note especially that we strictly distort the logic of Newton’s
Principia here: as far as Proposition I.1 is concerned,the forces could be impulses directed towards the point S . However, though Newton’s reader may not at that stageknow the nature of the force, for Newton the figure represents the action of universal gravitation.
12f physical salience that apply to theories with classical spacetime, but we should expect a non-spatiotemporal theory to require new standards. And so the real question is what are those newprinciples? Like Friedman, we see that question, and the development of such new principles as afoundational, interpretational, conceptual—hence philosophical —endeavor. We shall elaborate onhow such a project is to be conducted in §
7. We will see throughout the book how this endeavoris ineliminably philosophical in the different approaches to QG. For now we want to illustrate theproblem with an example.
Of necessity, this section is somewhat more technical than the others, and could be skipped bythose not requiring a concrete illustration of how interpretational considerations come into playin elevating a formal derivation into one (potentially) having physical salience. It elaborates anexample of a non-spatiotemporal theory already given, to show how one might come to view it asa theory from which spacetime emerges.We start with familiar, commutative geometry, for which xy = yx , in a smooth manifold ofpoints; let it be 2-dimensional for simplicity. Consider polynomials P ( x, y ) of x and y . These are‘fields’, meaning that they return a numerical value at each point ( x, y ). They form an algebrawith respect to multiplication: this just means that when you multiply two polynomials together,the result is another polynomial. Moreover, because xy = yx we have that P ( x, y ) Q ( x, y ) = Q ( x, y ) P ( x, y ), so that the algebra is commutative. (Check with P ( x, y ) = xy and Q ( x, y ) = x + y if you like.)It may seem like a rather uninteresting structure, but in fact such algebraic relations alone con-tain geometric information about the space: in this case, that it is smooth, that it is 2-dimensional,and whether it is open or closed. This fact is shown by the important Gelfand-Naimark theorem(1943), which is the foundation of ‘algebraic geometry’. Indeed, the whole structure of differentialgeometry can be recast in algebraic terms. (An interesting application is Geroch’s (1972) formula-tion of general relativity as an ‘Einstein algebra’; discussed by Earman (1989, § xy − yx = iθ and sees what happens. (And similarlyin spaces of any dimensions.) Surprisingly, one finds that the structure necessary to cast geometryin algebraic terms remains (at bottom, one can still define a derivative on the algebra, in termsof which the other structure is defined). Moreover, the Euler-Langrange equation and Noether’stheorem do not require commutativity, and so the structure of modern physics is preserved, evenin such a ‘non-commutative spacetime’—in a purely algebraic formulation.But suppose such a physics were correct: how could it explain spacetime as it appears to us?Specifically, how are we to understand events localized in space in terms of an abstract algebra?When the algebra is commutative, the Gelfand-Naimark theorem lets us interpret the elements asfields, P ( x, y ) related to regions of space; but what about the non-commutative case? The questionis just that which has concerned us in this chapter (and indeed the whole book): how can we derivethe appearance of classical spacetime from a non-spatiotemporal theory, in a physically salient way?The obvious thing to try is to (a) interpret x and y not as elements of an abstract algebra, butas fields in an ordinary plane: taking the value of the x and y coordinates at any point ( x, y ). Then This section is based on Huggett et al. (forthcoming). See also Lizzi (2009) for a more mathematical survey. (cid:63) , such that x (cid:63) y − y (cid:63) x = iθ . Then (c) construct the algebra ofpolynomial fields, but with (cid:63) -multiplication instead of regular (point-wise) multiplication. Indeed,this is exactly how one typically proceeds in non-commutative geometry: in one formulation, theoperation is ‘Moyal- (cid:63) ’ multiplication , and the fields form the ‘Weyl representation’ of the algebra.The algebra of the fields with respect to (cid:63) will be that of the abstract non-commutative algebra,and now we have referred that algebra to objects in an ordinary manifold. In particular, one couldtalk about the local region in which such-and-such a field has values less than 1, say. Indeed, onemight now wonder whether we should throw away the abstract algebra, and just treat physics in‘non-commutative geometry’ as really physics in commutative geometry, but with an unfamiliarmultiplication operation. In other words, wonder whether classical spacetime needs to be recoveredat all?Huggett, Lizzi, and Menon (forthcoming) argue that indeed it must be, for the Weyl repre-sentation has formal representational structure that exceeds its meaningful, physical content. Inparticular, the concept of a region with an area smaller than θ —a forteriori that of a point—is unde-finable in the theory. This can be seen in a couple of ways, but for instance the attempt to measurepositions more accurately leads to unphysical results. The conclusion is that, although the Weylrepresentation contains points and arbitrarily small regions, they are purely formal, and do notrepresent anything real: non-commutative geometry—even the Weyl representation—is physically‘pointless’.As a result, we cannot understand a point value of the Weyl fields as having any physicalmeaning. Rather we need to understand the fields as complete configurations: the unit of physicalmeaning for a field in non-commutative space is the function from each point to a value, P : ( x, y ) → R ; not its value , P (x , y) at any particular point (x , y). But the full configuration is equivalent tothe place of the field in the abstract algebra, and so we are back to the question of deriving locality.Here is one way to proceed, using an ansatz proposed by Chaichian, Demichev, and Presnajder2000 (discussed further in Huggett et al. forthcoming). They propose that an ordinary, commutingfield—the kind observed in classical spacetime—be related to a Weyl field W ( x, y ) by an operationof ‘smearing’. One multiplies W ( x, y ) by a θ -sized ‘bell function’ about ( X, Y ), and integrates overthe Weyl space coordinates x and y . The result is a new field Ω(
X, Y ). Extrapolating from this‘CDP ansatz’, the result of smearing is to introduce classical space into the theory. W lives inWeyl space, whose status, we argue, is only that of a formal representation of the fundamentalalgebra, while Ω should be interpreted as living in the physical space that we observe. We thusinterpret smearing as relating a function on one space to a value on another space : it relates thenon-commuting field W , represented as a function over Weyl space points ( x, y ), to the value of anobserved, commuting field Ω at physical space point ( X, Y ). That it takes a function to a value isjust mathematics; that it relates Weyl and physical spaces is a substantive physical postulate.However, it still makes no sense to consider Ω in regions smaller than θ : we have in facterased the unphysical information at such scales by smearing W . So strictly a single coordinatepair ( X, Y ) does not label a physical point. Rather, the proposal is that these smeared fieldsare approximated by observable fields over regions greater than θ ; thereby formally deriving thelatter, spatially localized objects from the former, purely algebraic objects. (In this case, we havesmearing as ‘aggregating’, in a very loose sense.) Of course, in our existing theories, fields live ina full commuting spacetime, but that is an extrapolation from our actual observations of fields, φ (cid:63) ψ ≡ φ · ψ + (cid:80) ∞ n =1 ( i ) n n ! θ i j . . . θ i n j n ∂ i . . . ∂ i n φ · ∂ j . . . ∂ j n ψ . Ω( X, Y ) ∝ (cid:82) (cid:16) e − (( X − x ) +( Y − y ) ) /θ · W ( x, y ) (cid:17) d x d y . θ . On the proposed interpretation, then,any information contained in Ω about regions less than θ is not only unobserved, but unphysical,surplus representational ‘fluff’.Now, nothing in the theory forces this picture as a physical story—it is merely an interpretationalpostulate. (Though we claim that it is conceptually coherent.) However, it has empirical conse-quences: the dynamics magnifies the θ -scale non-commutativity to observable scales (e.g., Carrollet al. 2001). If those predictions are successful, then we have evidence that the underlying non-commutative field theory and the interpretational postulate are correct. Imagining that situationthen, we claim that the situation is exactly analogous to that of the Newtonians regarding actionat a distance. That is, we would be justified in accepting the CBP ansatz and our interpretationalpostulate as novel principles of physical salience: they regulate what constitutes a physically salientderivation in the theory. In both cases, the final ground is the empirical success of the theory.So it should be clear how the example illustrates our points about physical salience and ourscheme proposed above. In the first place we have argued that non-commutative geometry isnon-spatial, in the sense that it is ‘pointless’, and so must be understood as a purely algebraictheory. Then we have explicated a possible formal derivation of localizable fields from this morefundamental theory. And finally, we have sketched a scenario in which such a derivation leads tosuccessful predictions, and hence to the conclusion that the formal derivation is physically salient,in fact explaining the appearance of localized fields, and ultimately classical spacetime. Of course,we highlight that this discovery constitutes a change in what derivations ‘deserve to be regarded asphysically salient (rather than merely mathematically definable)’, to paraphrase Maudlin.This is the pattern that we will see in more detail in the examples of the following chapters, andto which we will return in the conclusion. In the following two sections we will investigate furtherwhat is achieved by such a derivation ( § § The schema that we presented for a physically salient derivation relates closely (but not exactly)to Lewis’ (1972) account of functional identification. Since ‘spacetime functionalism’, of variousvarieties, has been recently discussed it is worth drawing the comparison, to better understand thenature of the proposed emergence. Suppose, in idealization, that a theory T is formulated as apostulate ‘ T [ t ]’. t represents what was traditionally called the ‘theoretical’ terms, though we prefer‘troublesome’ (Walsh and Button 2018, § T [ · ]’ also involves(traditionally) ‘observational’, or (according to Lewis) ‘old’, or (with Walsh and Button) ‘okay’terms, and Lewis proposes that the t are defined in terms of them by t = ιx T [ x ] . (5)That is, ‘the t are (if anything) the extant, unique things that satisfy the theory postulate’. The t are thus defined in terms of their nomic relations to one another and to the okay terms—i.e., interms of their ‘functional’ relations—and so (5) is a functional definition. As a result, the terms Lewis proposes in passing that the actual definition be modified to allow for approximate satisfaction of T [ · ]. Inour opinion that is always going to be the case in actual theories, so this modification is not optional but required, andhis discussion is a significant idealization. The harder question of how exactly the modification is to be implemented are rendered semantically okay (though they may remain metaphysically problematic).Suppose that we also hold a postulate R [ r ], where r and t do not overlap, so that our acceptance R does not depend on our views on the troublesome terms of T . Further suppose that we cometo believe T [ r ]. Now, T [ r ] , t = ιx T [ x ] (cid:15) r = t, (6)so, by definition of the t , T [ r ] deductively entails the identity of the objects of R and T . Lewis’ pointis the epistemic one that such functional identifications are thus not inductive: given a functionaldefinition, once one accepts that the objects of R play the same roles as those of T , logic andmeaning alone commit one to accepting their identity .Lewis offered electromagnetic waves and light as an example of the scheme, but of course hispoint was that ‘when’ neuroscience showed that neural states played the functional roles of mentalstates, then they would—as a matter of logic and definition—be identified. The subject of thisbook also broadly fits Lewis’ scheme: theory T is our spacetime theory assumed by QFT andrelativity theory, while R is a theory of QG. Denote them ST and QG , respectively, and use ST to functionally define any troublesome spacetime terms. Then, according to the schema of §
4, aphysically salient derivation of ST from QG shows that ‘aggregates’ of QG , described using its terms q , satisfy ST [ · ]—that they indeed play the functional roles of the objects of ST . So the identityfollows. However, there are differences to Lewis’ functionalism, which we will explain presently.Now, how to turn Lewis’ scheme into a concrete plan for the functional reduction of spacetime?The spacetime functionalism recently introduced by Lam and W¨uthrich (2018, forthcoming) is basedon the general scheme in the spirit of Kim (2005, 101f) according to which a functional reductionof higher-level properties or entities to lower-level properties or entities consists in two necessaryand jointly sufficient steps: (FR1) The higher-level entities/properties/states to be reduced are ‘functionalized’; i.e., one specifiesthe causal roles that identify them, effectively making (5) explicit.(FR2) An explanation is given of how the lower-level entities/properties/states fill this functionalrole, so that we come to accept T [ r ].If these two steps are fulfilled, then it follows that the higher-level entities/properties/states arerealized by the lower level ones.Applying the template of functional reduction to the case of the emergence of spacetime in QG,the two steps above become:(SF1) Spacetime entities/properties/states, s , are functionalized by specifying their identifying roles,such as spacetime localization, dimensionality, interval, etc. Effectively, one makes explicit s = ιx ST [ x ].(SF2) An explanation is given of how the fundamental entities/properties/states, q , postulated bythe theory of quantum gravity fill these roles, so that we come to accept ST [ q ]. is not carefully addressed by Lewis. Acceptance of a theory requires that it be meaningful, hence acceptance of R [ r ] requires that the r be referential,and thus that any troublesome r can be functionally defined in terms of the okay r as in (5), mutatis mutandis. Weaccept this assumption for Lewis’ cases, but we will see that things are more complex in the case of QG. Kim’s model involves three steps, where the second is to identify the entities in the reduction base that performthe role at stake, and the third is to construct a theory explaining how these fundamental entities perform that role.We subsume these two steps in our second stage. are the spacetime entities/properties/states. In the following chapters, after explainingeach theory of QG and its conceptual foundations, we will follow this scheme in our discussions:on the one hand describing the functional roles of spacetime entities/properties/states, and on theother showing how the theory of QG proposes that those roles are played. Of course, given theevidential state of QG, we do not claim that that these proposals are correct: we only describe howthe theories may functionally reduce spacetime, not—as far as we currently know–how they do .Several remarks are in order regarding the functionalist approach to spacetime. First, it shouldbe made clear that we take emergence and reduction to be compatible with one another, and hencefunctional reduction may serve as a template to explain the emergence of a higher-level feature, i.e.,the fact that higher-level entities exhibit novel and robust behaviour not encountered or anticipatedat the more fundamental level. Second, there is a sense in which a functionalism about spacetime must start from a broaderconception of functional reduction than is usual in the familiar functionalisms in the philosophyof mind or the philosophy of the special sciences. There, a mental or biological or other higher-level property is understood to be determined by—indeed, usually identified with —its causal role within the relevant network such as the network of mental or biological activities. If in spacetimefunctionalism the roles are still supposed to be causal , then a much broader notion of ‘causal’ mustbe at work, one that does not in any way depend on the prior existence of spacetime. As it isnot clear what that would be, it is preferable to formulate a notion of functionalism devoid of anyinsistence that the functional roles be causal.Third, the central claim of spacetime functionalism is that it is sufficient to establish only thefunctionally relevant aspects of spacetime. In particular, it is therefore not necessary to somehowderive relativistic spacetime in its full glory and in its every aspect in order to discharge the task.Naturally, this raises the question of what these functionally relevant aspects of spacetime are—thetask of (SF1). As we will see in the following chapters, different approaches to QG take differentstances on what functions are to be recovered, though broadly speaking, all aim to recover functionssufficient for the empirical significance of basic metrical and topological properties. Our stance willbe that the list of functions cannot be determined a priori from conceptual analysis of classicalspacetime theories, but by the twin demands of the empirical, and of the resources of the proposedreducing theory. In short, part of the work of each chapter will be to identify the spacetime functionsrecovered in the different approaches, and indicate how they relate to observation.Fourth, the scheme permits a form of ‘multiple realizability’, as is typical of functionalism alsoin the philosophy of mind or the philosophy of the special sciences: (SF2) allows that different(kinds of) fundamental entities might play one and the same functional role, i.e., that the ‘realizer’of spacetime might have been by something other than what it in fact is. This liberal stancespurs a concern that functionalism is too weak a condition to secure the emergence of spacetime,that the true nature of spacetime is not exhausted by its functional roles, so that none of the merefunctional realizers could ever truly be spacetime. In particular, the worry continues, a rash relianceon functionalism misses the qualitative nature of spacetime—some kind of spacetime ‘qualia’, asit were—and it is precisely such qualitative features that make spacetime what it is, and whichcannot be recovered by mere functional realization. However, the case of spacetime is disanalogousto that of mind: we agree with Knox (2014) who states that where “the fan of qualia [in thephilosophy of mind] has introspection, the fan of the [spacetime] container has only metaphor” As restated many times in our earlier publications, and in agreement with what we take to be the consensus inphilosophy of physics as stated, e.g., in Butterfield (2011a,b) and Crowther (2016, § Borrowing a distinction from Le Bihan (2018) between a “hard” and an “easy problem” ofspacetime emergence, spacetime functionalism amounts to the denial that there is a hard problemof an unbridgeable explanatory gap between the fundamental, non-spatiotemporal and the emer-gent, spatiotemporal realm. For the functionalist, what is to be shown—by a physically salientderivation—is how the fundamental degrees of freedom can collectively behave in ways such thatthey play the required spacetime roles. And nothing more. No special character, or essence, ormetaphysical nature need be accounted for. Functional identification requires no ‘luminosity’ oflight beyond the behavior of electromagnetic waves, or ‘consciousness’ beyond the functioning ofneurons. Or in our case, no special ‘spatiotemporality’ that the non-spatiotemporal could never ob-tain. Once one has shown that the non-spatiotemporal plays the roles of the spatiotemporal—andso is the spatiotemporal—no more need be said: one has a full scientific account of the emergenceof spacetime, and no ‘explanatory gap’ remains.Fifth, functionalism shows how the goal of reduction can be the scientific explanation of thefunctional roles of higher level entities/structures/states by lower level entities/structures/states(and to nothing more). But, it is debatable to what exactly spacetime functionalism is ontologicallycommitted: substances, relations, entities, structures, states, or something else. We will not furtherpursue this debate as we believe it to be orthogonal to the concerns of this book. Thus we hopethat the reader will forgive our switching between speaking of the spatiotemporal as if it were anentity, or a structure, or a state, or something different yet again. We simply aim to avoid torturingEnglish more than necessary, and no deep philosophical commitment should, for instance, be readinto our using ‘spacetime’ as a noun.Sixth, we are far from the first to suggest functionally defining space or spacetime. DiSalle (2006,chapter 2) reads Newton’s Scholium to the definition in much this way (though Huggett 2012 dis-agrees). Functionalist strategies have also become very visible in the philosophy of non-relativisticquantum mechanics, where Wallace (2012) deploys it in his defence of an Everettian interpretationand Albert (2015, Ch. 6) in support of wave function monism. Those latter applications differfrom ours because they are concerned with recovering three-dimensional physical space. In con-trast, spacetime functionalism in QG is commissioned with functionally recovering 4-dimensionalspacetime, and so relates to work by Knox (2013, 2014, 2019) in the context of classical spacetimephysics. For her, something ‘plays spacetime’s role’ and thus is spacetime “just in case it describesthe structure of inertial frames, and the coordinate systems associated with these” (2014, 15). InGR, the metric field performs spacetime’s role in this sense and thus is identified with spacetime byher. As the metric may itself not be fundamental but instead emerge from the collective behaviorof more fundamental degrees of freedom, she explicitly leaves open the possibility that the realizersof spacetime’s functions may themselves not be fundamental (Knox 2013, 18). As the relationshipbetween the fundamental degrees of freedom and the emergent spacetime realizer is left untouched We also concur with Lam and W¨uthrich (2018) in their rejection of the version of this concern articulated in Ney(2015), who worries that if the fundamental entities are not already appropriately (spatio)temporal in their nature,they cannot ‘build up’ or constitute spacetime as they are not the right kind of stuff (see also Hagar and Hemmo2013). As diagnosed by Lam and W¨uthrich, advocates of this worry seem to rely on an unreasonably narrow conceptof constitution. We might also object that if we surrendered to this worry, there would be no principled reason tothink that it would not also annihilate all other cases of presumed emergence and amount to an unyielding dualism.
18y Knox’s inertial frame functionalism, the latter does also not shed any light on it. Seventh and finally, there is an important but subtle difference in the application of Lewis’sscheme to QG from that in the cases he has in mind. Suppose we accept a spacetime theory ST [ s ],where whatever it is that performs the spacetime functions is denoted by the troublesome terms, s . These, following Lewis, we take to be defined by s = ιx ST [ x ] . (7)The okay terms appearing in ST [ · ] would refer to matter of various kinds, its relative motions andpoint-coincidences: so, for instance, the metric in GR might be defined locally in terms of its rolein determining motions under gravity or scattering amplitudes. We think that this part of Lewis’picture—which corresponds to (SF1)—fits our cases well. But what about the second part of hisscheme, involving R [ r ]? Although the result is still a functional identification, its significance hasshifted somewhat, as we shall now explain.Butterfield and Gomes (2020a; 2020b) analyze recent proposals for spacetime functionalism inexplicitly Lewisian terms. They emphasize, as we have, that in Lewis’ scheme theoretical iden-tification follows by definition alone (once the r s are known to play the role of the t s), and thatfunctional identification is a species of reduction. But they also show how various spacetime andtemporal functionalisms follow the ‘Canberra plan’, according to which the troublesome t s are notonly defined by T , but are also ‘vindicated’ by their functional identification as r s. For instance, asmental states, perhaps, turn out to be neural states so, in their examples, a temporal metric mightbe identified with purely spatial structure; then, if neural states or spatial structures are on a firm(or firmer) ontological footing than mental states or time, the identifications show that the latterare equally well grounded. They are, that is, vindicated against any metaphysical suspicions raisedagainst them. That vindication is not by itself achieved by the functional definition (5) of the t s;that merely makes the terms referential, so that they can be meaningfully employed. Put anotherway, (FR1) alone does not vindicate the mental, for instance; (FR2) is also needed, to show how themental is part of the physical. Regarding these cases, we are in agreement with Butterfield andGomes emphasis of this important distinction, and its applicability to the cases that they discuss.However, in our cases, for which R is some QG , the second step, while still involving a functionalidentification, does not follow the Canberra plan, because the troublesome terms, q , of QG are non-spatiotemporal, and so on a weaker , not firmer, footing—the ontological and semantic correlate ofempirical incoherence. Ontologically, as we have discussed, our physical and metaphysical cate-gories assume spatiotemporality, and so the natures of the q are mysterious. Semantically, we canexpect an attempt to functionally define the q s as q = ιx QG [ x ] to fail. Lewis’ scheme for functionaldefinition requires that a theory have sufficient okay terms to uniquely define the troublesome ones:if many collections of terms satisfy the putative definition, then it fails to establish reference. Butthat is what one expects in a theory that breaks from established categories as radically as a non-spatiotemporal one; the terms that we take to be okay are systematically spatiotemporal in someway, and so are expected not to appear in QG . And indeed, we contend that the theoretical con-cepts of the theories we consider in this book cannot be defined without appeal to spatiotemporal Cf. Lam and W¨uthrich (2018, 40) and Lam and W¨uthrich (forthcoming, §
3) for a more detailed discussion ofinertial frame functionalism and how it relates to our project. Or put yet another way, the t are often troublesome both semantically and ontologically: the functional definitiontakes care of the first problem, while the functional identification takes care of the second. When we use ‘troublesome’we always mean semantically. Butterfield and Gomes do not claim otherwise, and indeed acknowledge that QG will look different (2020b, 3). reversed : the non-spatiotemporal objectsof QG are vindicated via their identifications with spatiotemporal objects. Clearly this approachonly works to the extent that the spatiotemporal is itself on a firm ontological footing, which ofcourse is a topic of endless debate. To skirt such debates in this book we will remain as neutral aspossible, and not take any stand on the metaphysical nature of spacetime features such as topologyor metricity, so that our conclusions remain valid for anyone who accepts them under whateverinterpretation.Within Lewis’ framework, the vindication of the q works as follows. Suppose that non-spatiotemporal QG [ q ] has been proposed. As explained, the q are semantically troublesome and ontologically sus-pect. Moreover, until we accept that a derivation of (at least a fragment of) ST is physically salient,we have no empirical grounds for accepting QG . Such a derivation will provide, not only groundsfor QG , but also define and vindicate the q . Introducing the ‘aggregate operator’ α ( · ), according toour schema, when we have a physically salient derivation of spacetime properties, then we accept α ( q ) = ιx ST [ x ] . (8)In conjunction with ST [ s ] this entails that α ( q ) = s (9)more-or-less as for Lewis. However, the reversal of the Canberra plan makes several things different.First, semantics. As noted, the q were not antecedently defined, but now can be throughtheir—or rather the α ( q )’s—role as spacetime entities/structures/states. In other words, (8) is inpart definitional of the q : the troublesome non-spatiotemporal terms of QG can only be definedwith reference to spatiotemporal terms not native in QG . Moreover, (8) only succeeds in definingthe q if in physical fact they play the ascribed roles, and do merely mimic them formally; somethingthat the physical salience of the derivation will secure. Second, ontology. The q s are placed on afirm ontological footing—are vindicated—when we accept that the α ( q ) are in physical fact thoseentities/structures/states that play the spacetime role. Once again, acceptance of the physicalsalience of the derivation secures just that.Finally, epistemology. In Lewis’ scheme, we have independently accepted theories of, say, neu-ronal and mental states, and later discover that they play the same functional roles, entailing thatthey are identical. In our case, the acceptance that QG ’s objects (or rather their aggregates) playthe same roles as ST ’s objects, and hence are identical with them, is simultaneous with our accep-tance of QG . In general terms, the evidence for R [ r ] is no longer antecedent (or independent) of theevidence for T [ r ], but rather the very same evidence. As such the epistemic calculus is different. Inone case, observations of neuronal states can be made independently of mental states, and we onlyhave to show that they perform the relevant functions: producing suitable behaviors, for instance.In the other, observations are not independent of spacetime states, and have to support both thetruth of a theory of QG, and that its objects perform the right functions. To give evidence, that is, (8) is not purely definitional, since it also also involves an existential commitment that the q s exist. And it neednot fully define q ; we also still have that q = ιx QG [ q ] by definition. In Huggett and W¨uthrich (2013, 284), we described this approach to vindication as physical salience flowingdown to the q ‘from above’. As we noted, the theories that we plan to investigate are all speculative at present, faced withconsiderable formal and empirical uncertainties. So what can we hope to learn from a philosophicalenquiry into something that is at worst likely false, or at best a work in progress? We see thesituation as characteristic of emerging fundamental physics (and perhaps other sciences). Theprocess of discovery takes place along various fronts: obviously, new empirical work constrainstheory and requires explanation; also obviously, new mathematical formalisms are tried out andexplored; less obviously, but just as importantly, conceptual analysis of the emerging theory isundertaken. In particular, we want to stress that this last kind of work is carried out concurrentlywith the empirical and theoretical. One should not view interpretation as something that merelyhappens after an uninterpreted formal structure is presented, but as an inextricable aspect of theprocess of discovery. As such, it is something that has to be carried out on inchoate theories, inorder to help their development into a finished product.We claim that this view is supported by the historical record: we have in fact already seenthis for Newtonian gravity. But one can equally well point to the absolute-relative debate in thedevelopment of the concept of motion, or 19th century efforts to come to grips with the physicalsignificance of non-Euclidean geometry. These debates did not wait until after a theory was devel-oped to clarify its concepts; rather they had to be carried out simultaneously, as an integral partof the development of the theory (see DiSalle (2006)). Of course we are hardly the first to realizethat such philosophical issues have to be addressed together with the empirical and theoreticalones. Many of Kuhn’s (1962) arguments illustrate this point, and more recently it is a major themeof Friedman (2001). But while we agree with their focus on philosophical, conceptual analysis asan essential part of theory construction, we don’t intend to get involved in issues involving the apriori or incommensurability, instead we want to emphasize the practical role for analysis in thedevelopment of QG.In the search for a new fundamental theory, the goal is—as it was for Einstein and for Newton—anew formalism plus an interpretation that connects parts of the formalism to antecedently under-stood aspects of the physical world, especially to the empirical realm. That is, an interpretationof how the more fundamental plays the functional roles of the less fundamental. And of coursethat means undertaking the project that we have been talking about in this section, of derivingspatiotemporal predictions from theories of QG. But one never simply co-opts or invents formal-ism without some eye on the question of how it represents existing physics of interest; and as theformalism is developed it becomes possible to see more clearly how and what the new formalismrepresents. Addressing this question is of on-going importance for finding the right formalism forthe area under study. Moreover, constructing such a formalism does not typically proceed in amonolithic fashion; instead different fragments of theory are proposed, investigated, developed orabandoned. For example, think of the development of the standard model of QFT from the early21ays of quantum mechanics. So the analysis of concepts of the new theory in terms of existingphysics is often faced with a range of half-baked theories and models. All the same, lessons abouthow a more developed, less fragmented theory can be found depend on asking how the fragmentsrepresent known physics—the answers are potential clues to how the finished product could do so.We believe that contemporary QG should be thought of in just this way—certainly the frag-mentation is real! Our primary goal is to look at a range of the half-baked fragments and ask howthey connect to spatiotemporal phenomena. Since they do not do so in a familiar way, in terms of acontinuous manifold of points, the question becomes ‘how does spacetime emerge from the underly-ing physics?’. We hope, therefore, that by concentrating on the question of emergence, aside fromall the other issues involved in the search for a theory of QG, we will be performing a service tophysicists working in QG, by focussing their attention on what is already known—and remindingthem that success depends on making it part of the search. Naturally, we do not expect to findsolutions of the order of Newton or Einstein! Indeed, a lot of what we shall do is draw out answersalready given by physicists; we believe that careful philosophical analysis of these answers can helpclarify them to reveal strengths and weaknesses, and hence aid progress. (Moreover, because we arefocussed on this quite narrow issue, we can survey a wider range of approaches than most physicistsactively study, and so provide a helpful overview of the topic.) And hence we believe that in theexamples we will consider there are important clues for the development of QG which philosophicalanalysis can reveal.
Thus, our primary aim is to see how spacetime disappears and re-emerges in several approaches toQG, and to show how this is not just a technical issue for physicists to solve, but instead elicitsnumerous foundational and philosophical problems. As we work through three such approaches—causal set theory (CST), loop quantum gravity (LQG), and string theory, which were all brieflyintroduced in § String theory works within this approach, but with one important tweak: instead of quantizedpoint like particles, it deals in quantized 1-dimensional, string-like objects. This, it appears, makesall the difference to the finiteness of the theory. Chapters 6-9, address the emergence of spacetimein string theory. Chapter 6 is a fairly technical introduction of the theory, aimed at philosophersof physics: it aims to be more intuitive, and more explicit about the conceptual and physicalframework than physics textbooks usually are. For those who have some familiarity with classicaland quantum field theory, it will tell you what you need to know about strings. Chapter 7 dealswith string ‘dualities’: some fascinating and powerful symmetries that arise when space has aninteresting topology (a cylinder, say). We argue that they are the kind of symmetries are notmerely observational, but ‘go all the way down’, showing that string theory does not possess, in itsbasic objects, familiar spacetime properties, such as definite size or topology; it is for largely thatreason that spacetime ‘emerges’. Chapter 8 is again fairly technical, explaining and analyzing insome detail the derivation of the Einstein field equation for gravity, from string theory. This is acentral part of emergence, for it derives the spacetime metric, giving empirical content to spacetimegeometry, and gives rise to GR. Finally, chapter 9 draws on the material of the previous chaptersto argue that indeed spacetime emerges in string theory, how this happens, and what ‘principles ofphysical salience’ are required.The final, concluding, chapter draws on the results of the previous ones to return to the questionof this introduction. How can we see that the derivations of spacetime that we have investigated arethemselves physically salient, and what principles can we extract from them that might be helpfulin the search for QG? See Kiefer (2004, chapter 2) for a very nice survey of QFT of the gravitational field. eferences David Albert.
After Physics . Harvard University Press, Cambridge, MA, 2015.David Z Albert. Elementary quantum metaphysics. In
Bohmian mechanics and quantum theory:An appraisal , pages 277–284. Springer, 1996.Jeffrey A Barrett. Empirical adequancy and the availability of reliable records in quantum mechan-ics.
Philosophy of Science , 63:49–64, 1996.Sougato Bose, Anupam Mazumdar, Gavin W Morley, Hendrik Ulbricht, Marko Toroˇs, Mauro Pa-ternostro, Andrew A Geraci, Peter F Barker, MS Kim, and Gerard Milburn. Spin entanglementwitness for quantum gravity.
Physical Review Letters
Foundations ofPhysics , 41:920–959, 2011a.Jeremy Butterfield. Less is different: emergence and reduction reconciled.
Foundations of Physics ,41:1065–1135, 2011b.Jeremy Butterfield and Henrique Gomes. Functionalism as a species of reduction, July 2020a. URL http://philsci-archive.pitt.edu/18043/ . Submitted to: ?Current Debates in Philosophy ofScience: In honor of Roberto Torretti?, edited by Cristian Soto; (to be published in the SyntheseLibrary).Jeremy Butterfield and Henrique Gomes. Geometrodynamics as functionalism about time, October2020b. URL http://philsci-archive.pitt.edu/18339/ . Submitted to: From Quantum toClassical: Essays in memory of Dieter Zeh; edited by Claus Kiefer: Springer, Cham, 2021.Craig Callender and Nick Huggett. Why quantize gravity (or any other field for that matter)?
Philosophy of Science , 68:S382–94, 2001.Sean M. Carroll, Jeffrey A. Harvey, V. Alan Kostelecky, Charles D. Lane, and Takemi Okamoto.Noncommutative field theory and lorentz violation. 2001. URL http://arxiv.org/abs/hep-th/0105082v1 .M. Chaichian, A. Demichev, and P. Presnajder. Quantum field theory on noncommutative space-times and the persistence of ultraviolet divergences.
Nucl.Phys. , B567:360–390, 2000. URL http://arxiv.org/abs/hep-th/9812180 .Samuel Clarke, Gottfried Wilhelm Leibniz, and Robert Gavin Alexander.
The Leibniz-Clarke Corre-spondence: Together Wiith Extracts from Newton’s Principia and Opticks . Manchester UniversityPress, 1956.Karen Crowther.
Effective Spacetime: Understanding Emergence in Effective Field Theory andQuantum Gravity . Springer, Cham, 2016. 24ene Descartes.
Principia Philosophiae . Apud Ludovicum Elezvirium, 1644.Robert DiSalle.
Understanding Spacetime: The Philosophical Development of Physics from Newtonto Einstein . Cambridge University Press, Cambridge, 2006.John Earman.
World Enough and Space-Time: Absolute versus Relational Theories of Space andTime . MIT Press, Cambridge, MA, 1989.Michael Friedman.
Dynamics of Reason . CSLI Publications, 2001.I.M Gelfand and M.A Naimark. On the embedding of normed rings into the ring of operators inhilbert space.
Mat. Sbornik , 12:197–213, 1943.Robert Geroch. Einstein algebras.
Communications in Mathematical Physics , 26:271–275, 1972.Amit Hagar and Meir Hemmo. The primacy of geometry.
Studies in History and Philosophy ofScience Part B: Studies in History and Philosophy of Modern Physics , 44(3):357–364, 2013.S. W. Hawking. Black hole explosions?
Nature , 248(5443):30–31, 1974.Mary B Hesse.
Forces and fields . T. Nelson, 1961.Nick Huggett. What did newton mean by ‘absolute motion’ ? In Andrew Janiak and Eric Schliesser,editors,
Interpreting Newton: Critical Essays , pages 196–218. Cambridge University Press, 2012.Nick Huggett. Spacetime ’emergence’, December 2018. URL http://philsci-archive.pitt.edu/15440/ . To appear in Routledge Companion to Philosophy of Physics, edited by Eleanor Knoxand Alastair Wilson.Nick Huggett and Christian W¨uthrich. Emergent spacetime and empirical (in) coherence.
Studies inHistory and Philosophy of Science Part B: Studies in History and Philosophy of Modern Physics ,44(3):276–285, 2013.Nick Huggett, Fedele Lizzi, and Tushar Menon. Missing the point in noncommutative geometry.
Synthese , forthcoming.Claus Kiefer.
Quantum Gravity . Oxford University Press, 2004.Jaegwon Kim.
Physicalism, or Something Near Enough . Princeton University Press, Princeton,2005.Eleanor Knox. Effective spacetime geometry.
Studies in History and Philosophy of Modern Physics ,44:346–356, 2013.Eleanor Knox. Spacetime structuralism or spacetime functionalism? Manuscript, 2014.Eleanor Knox. Physical relativity from a functionalist perspective.
Studies in History and Philos-ophy of Modern Physics , 67:118–124, 2019.Thomas S Kuhn.
The Structure of Scientific Revolutions . University of Chicago Press, Chicago,1962. 25incent Lam and Christian W¨uthrich. Spacetime is as spacetime does.
Studies in History andPhilosophy of Modern Physics , 64:39–51, 2018.Vincent Lam and Christian W¨uthrich. Spacetime functionalism from a realist perspective.
Synthese ,forthcoming.Baptiste Le Bihan. Priority monism beyond spacetime.
Metaphysica , 19:95–111, 2018.Jonathan Leake. Big bang machine could destroy earth.
The Sunday Times , July 18 1999.David Lewis. Psychophysical and theoretical identifications.
Australasian Journal of Philosophy ,50(3):249–258, 1972.Fedele Lizzi. Noncommutative spaces.
Lecture Notes in Physics , 774:89–109, 2009.Tim Maudlin. Completeness, supervenience, and ontology.
Journal of Physics A: Mathematicaland Theoretical , 40:3151–3171, 2007.Horatiu Nastase. The rhic fireball as a dual black hole. arXiv preprint hep-th/0501068 , 2005.Isaac Newton.
Philosophiae naturalis principia mathematica , volume 3. Apud Guil. & Joh. Innys,Regiæ Societatis typographos, 1726.Isaac Newton.
Opticks . Prabhat Prakashan, 1730.Alyssa Ney. Fundamental physical ontologies and the constraint of empirical coherence.
Synthese ,192:3105–3124, 2015.Daniele Oriti. Disappearance and emergence of space and time in quantum gravity.
Studies inHistory and Philosophy of Science Part B: Studies in History and Philosophy of Modern Physics ,46:186–199, 2014.Lawrence Sklar. Prospects for a causal theory of space-time. In Richard Swinburne, editor,
Space,Time and Causality , pages 45–62. D. Reidel Publishing Company, Dordrecht, 1983.David Wallace.
The Emergent Multiverse: Quantum Theory According to the Everett Interpretation .Oxford University Press, Oxford, 2012.Sean Walsh and Tim Button.
Philosophy and Model Theory . Oxford, UK: Oxford University Press,2018.Christian W¨uthrich. To quantize or not to quantize: fact and folklore in quantum gravity.