Physics and the Direction of Time

views updated


Our experience of the temporality of things seems to be an experience of a radically asymmetric feature of the world. Although we do know some things about what the future will be like, we have an access to past events that is not given to us of events in the future. We take ourselves as having memories of the past but not of the future and as having records of the past but not of the future. In our explanatory accounts of what happens in the world, we explain present and future by reference to what happened in the past, but we typically do not explain the past by referring to the future. We take it that there is causation in the worldthat events determine one another to occur. But, intuitively, we think of causation and determination as directed from past to future. We have distinctive attitudes to past and future. Of the past we may have regrets, for example, but our concern for the future will be rather things such as anxiety or anticipation. So profound are these apparent differences between past and future that they are often promoted into the realm of deep metaphysics. Sometimes it is argued that the past is fixed, subject to some version or another of immutability, whereas the future remains merely a domain of open possibilities. In an even more extreme view it is argued that what is past has a determinate reality whereas the future remains a realm to which we cannot even attribute any kind of determinate being.

One might take these asymmetric features of time as irreducible, primitive properties of the world. And one might take our awareness of these features as somehow direct and not further explicable. Alternatively, one might argue for some basic, asymmetrical, metaphysical aspect of time as grounding all the asymmetries discussed above. For example, there are proposed branching models of the world in which a tree of possibilities is constantly pruned into a single actuality as time goes on and the present moves inexorably into the future. One problem with any such model is the need to respect the results of modern physics, especially special and general relativity, so as to reconcile the usual assumption in the metaphysical models of a unique global present with the denial of any such objective feature of the world in the relativistic accounts of spacetime. Another alternative would be to take a temporally asymmetric notion of causation as primitive and argue that all the other intuitive asymmetries follow from the fundamental asymmetry of causation.

Naturalistic Theories of Time Asymmetry

On the other hand, one could seek for some naturalistic account of the temporal asymmetries. Here, one looks at what our best available scientific theories tell us about the actual physical structure of the world in the hopes of finding some physical process characterized by fundamental physics that could serve to ground or explain the existence and nature of the fundamental temporal asymmetries. Much work has been done in this direction, but more needs to be done to make such a naturalistic account fully convincing. It is to this approach that this entry is directed.

Physics presents us with a paradox. Although most of its fundamental laws are often alleged to be time-reversal invariant and unable, it is therefore claimed, to ground any fundamental asymmetry in time of processes in the world, physics also describes a number of alleged time-asymmetric features of the world at a very general level. Measurement processes in quantum mechanics are often alleged to be asymmetric in time. We see radiation outbound in spherical waves from accelerated charged particles but not spontaneous collapsing spheres of radiation converging on a particle and accelerating it. Subtle experiments seem to show that some of the interactions of the elementary particles show asymmetries in time that may, indeed, require positing a fundamental law governing them that itself describes a lawlike asymmetry in time for the world. Most importantly, thermodynamics seems to reveal to us a world that is time asymmetric. A metal bar hot at one end and cold at the other when kept in isolation becomes uniformly warm all over. But an isolated uniformly warm bar does not spontaneously become hot at one end and cold at another.

A naturalistic account of the direction of time requires more than finding some physical process that is time asymmetric. It also requires even more than finding a fundamental process that has such time asymmetry. Suppose the weak interactions of the elementary particles obey a time-asymmetric law. How would such a fact be of any use in accounting for our intuitive sense that there are records and memories of the past and not of the future, or that causation proceeds from past to future, or that the past is determinate and fixed and the future a realm of mere possibilities? What is needed from a naturalistic theory of the direction of time is some appropriate connection between the physical facts introduced in the account as grounding the direction of time and those features that characterize our intuitive, deeply rooted sense of the asymmetry of time.

Thermodynamic Asymmetries

The thermodynamic asymmetries, being pervasive elements of our everyday experience, provide the most promising physical basis for a naturalistic account of the direction of time. Here, two fundamental questions must be explored: (1) Why does the world show these deep asymmetries in time of physical processes? (2) How could the existence of these asymmetric processes account for the intuitive asymmetries we attribute to the world in time? Neither question is trivially answered.

The contemporary explanation of the thermodynamic laws starts with the realization that macroscopic objects are composed of a vast number of microscopic constituents. Macroscopic thermodynamic properties, then, are thought of as grounded in such microscopic features of a system as the total energy of its microscopic constituents or the average energy of some one of these. The microscopic constituents are assumed to obey the standard dynamical laws, originally classical dynamics and now quantum mechanics. How the system behaves, then, will depend upon these laws and upon whatever initial conditions are possessed by the microscopic constituents, with the system also subject to such macroscopic constraints as exist (such as a gas being confined to a box).

Probabilistic methods were soon invoked to deal with the behavior of the vast number of microscopic constituents. These led to such theories as the kinetic theory of gases and the more abstract statistical mechanics. One consequence of this new viewpoint was a rethinking of the thermodynamic laws to allow for such possibilities as fluctuations away from the equilibrium state, even for an isolated system. A deep conceptual problem for this theory is the understanding of why the probabilistic posits that are made, and that work so well for prediction and explanation, hold in the world. Are they brute posits to be otherwise unexplained? To what degree can they be extracted from the dynamical laws governing the behavior of the microscopic constituents? Need the fundamental dynamical laws be modified to find an appropriate explanation for the fundamental probabilistic posits (and, perhaps, to solve other outstanding problems as well, such as the nature of the measurement process in quantum mechanics)? Another crucial question is the degree to which the probabilistic posits can be shown consistent with the underlying dynamics and the degree to which they can be weakened with the empirical results still forthcoming.

Furthermore, arguments that have existed from the early days of the theory indicate that introducing probability into the theory is not, by itself, enough to ground a theory of the direction of time. Probabilistic considerations would suggest that the world we live in should be a world where all systems are at equilibrium, not at all like the world we actually live in with its vast pool of nonequilibrium systems and its parallel movement from nonequilibrium to equilibrium of temporarily isolated systems. In addition, to obtain the desired nonequilibrium results in the theory, the theory's probabilistic posits must be applied in a temporally asymmetric way, being taken as correctly applicable to temporally initial, but not to temporally final, states of isolated systems.

Cosmological Considerations

From early days of the theory, cosmology has been invoked as providing the needed supplementary posits. Ludwig Boltzmann's assistant Scheutz suggested the possibility that the cosmos was, in general, in equilibrium, with the part of it with which we were familiar in a local (if large from our perspective) fluctuation away from equilibrium. Our local cosmos, then, was in equilibrium in the past and will be again in the future. Boltzmann added the anthropic observation that we could not find ourselves in one of the pervasive equilibrium regions of the cosmos since such a region could not support the flows of energy necessary for life. To this Boltzmann added the additional proposal that the reason we found our region heading toward equilibrium in the future time direction and not in the past time direction is that our very meaning of the future direction of time was that the future time direction was determined by that temporal direction in which our local region of the cosmos had a succession of states closer to equilibrium (of higher entropy).

Current cosmology, insofar as it is a discipline open to observation and empirical test, is doubtful of this early cosmological speculation. The current model, rather, is of a universe (at least as far as we know) that is distinctly unsymmetrical in time. In particular, it is posited that there is a singularity in which the cosmos is all at a single spatial point in the past time direction some tens of billions of years ago, the so-called Big Bang cosmology.

Even accepting this model of the universe is not enough to get the thermodynamic asymmetries out of the cosmology. Instead, it is generally agreed, one must make a specific assumption about the Big Bang state of the cosmos, that it is a low-entropy, that is, a highly nonequilibrium, state. The usual posit is that the space of the world at the Big Bang is highly smooth, this being for gravitational force the low-entropy condition. The idea is that as matter clumps from uniformly distributed into stars (and galaxies, etc.), the matter, initially in thermal equilibrium, becomes highly nonuniform and in a grossly nonequilibrium state, with hot stars radiating out to cold space. The decrease in the matter's entropy is continually being paid for by the great increase in the entropy of the spatial distribution that has gone from uniform to clumped.

The idea, then, is that the universe as a whole must be posited to have an initial highly nonequilibrium starting point. It is this posit that must be added to the standard probabilistic assumptions to get us the result of a predicted nonequilibrium condition for the world as we find it, and a predicted, temporally asymmetric, approach to equilibrium in the future and not into the past, for system temporarily isolated from their environments. Here the grand cosmic initial condition is being invoked to generate the temporal asymmetry unobtainable from the allegedly time-symmetric dynamical laws alone. Getting the result about the temporarily isolated subsystems of the universe requires a little more, in the form of a demonstration that from the temporally asymmetric behavior of the universe as a whole one can derive, with either no additional temporally asymmetric assumptions at all or with some posited additional asymmetry of dynamics, parallel increase of entropy of so-called branch systems in the same time direction in which the entropy of the cosmos as a whole is increasing.

From Asymmetries in Time to the Direction of Time

But then there is the second question noted above as well. Why should we think that this pervasive asymmetry of physical systems in time has anything to do with our intuitive sense of the asymmetry of time itself? Once again, the mere fact that there is some asymmetry in time of systems of the world, even a lawlike temporal asymmetry, is not enough to establish the naturalist's claim. What else is required?

Boltzmann hinted at an answer in his famous paper "On Zermelo's Paper 'On the Mechanical Explanation of Irreversible Processes'" where he claimed that what we took to be the future direction of time was just the direction of time in which the entropy of our local portion of the universe was on the increase. In that paper he drew a trenchant analogy between the temporal case and an intuitive spatial asymmetry accounted for by gravity. Originally we might think of space as being asymmetric, with one direction being down and its opposite up. Eventually, though, we realize that what we call the down direction is just the direction of the local gravitational force. On antipodal points of the earth, the local downward directions are directed oppositely to one another. And in a region of space in which there was no gravitational force, there would be no downward direction. Just so in a region of the universe not in equilibrium, Boltzmann maintained, the direction of time in which entropy was increasing would be the future time direction, and in a region of the universe at equilibrium, there would be no future direction of time and no past direction (although there would still be two oppositely directed directions of time).

What makes Boltzmann's remarks about gravity and down so plausible? It is the fact that we have in gravity and its consequences a complete explanation of all the facts that we take as distinguishing the downward direction of space from all the other directions. Down is the direction in which, for example, rocks fall. We even have an explanation, invoking the fluid in our semicircular canals, of how it is that we can tell without external observation which direction is the downward one.

The analogous argument in the case of the direction of time would require a sustained argument to the effect that it is the existence of the asymmetric processes of systems in the world described by thermodynamics, and explained by statistical mechanics combined with cosmology, that provides a full explanatory account of all those features of the world that we take as marking out an asymmetric nature to time. What would need explaining is why we have memories and records of the past and not of the future, why we take causation as going from past to future and not the other way around, why we think of the past as fixed and determinate and of the future as a realm of possibility, and, also, of how it is that we think we can tell, without inference, of any pair of events which is the earlier and which the later.

Sometimes it is claimed that the entropic theory of time direction is supported by the fact that we cannot tell of a film of events whether it is being run in the right or the wrong direction except by reference to entropic facts portrayed by the film. Sometimes it is argued against the entropic account that we can tell of events in the world what their order is in time without noting any entropic features of them. Both arguments are misguided. What would be required of an entropic account would be some kind of explanation of all the intuitive asymmetries that constitute our sense of the direction of time, not a demonstration that our judgments of time order are always inferences from directly observed entropic facts.

A number of tentative suggestions have been made in this direction. Hans Reichenbach suggested that records might be analyzed as causal interventions that induced a macro low-entropy change into what would otherwise be a macro high-entropy state. But many records do not fit his paradigm. There is no fully developed extant argument to the effect that the very existence of records, and presumably those mental records we call memories, can have their time asymmetry directly accounted for by the entropic asymmetry of processes in time.

One might argue that such intuitive asymmetries as the direction of causation and the difference in fixedness of past and future are derivative from the asymmetry of records, our taking as the fixed and the determining that which we can know to be the case from records. Or, alternatively, one might try to directly account for the asymmetry of causation out of entropic-like facts. An example of that can again be found in a tradition stemming from Reichenbach where it is noted that spatially separated correlated events can often have their correlation explained by some common past event casually connected to both correlated events but not by any such correlating event in the future of the correlated pair (the so-called fork asymmetry). Another approach stems from David Lewis. Here, causation is analyzed in terms of counterfactual conditionals. It is suggested that the fact that an even has a numerous extended range of effects in its future, but not in its past, grounds our intuition that there are forward looking but not back tracking counterfactual assertions that we accept, and that this underlies our intuitions about the time asymmetry of the causal relation.

Even though no fully worked-out account of these sorts exists, the naturalistic approach to the direction of time remains the only plausible alternative to metaphysical accounts. The latter remain hard to explicate, and it is hard to understand how they might provide new insights into the intuitive asymmetries of time. The former, in its usual thermodynamic version, is at least clear in its intentions and of an intrinsic plausibility. The further pursuit of this naturalistic program is well worthwhile.

See also Boltzmann, Ludwig; Lewis, David; Reichenbach, Hans; Time; Time, Being, and Becoming.


Albert, D. Time and Chance. Cambridge, MA: Harvard University Press, 2000.

Boltzmann, L. "On Zermelo's Paper 'On the Mechanical Explanation of Irreversible Processes.'" In Kinetic Theory. Vol. 2, edited by S. Brush, 239245. Oxford: Pergamon Press, 19651972.

Davies, P. Physics of Time Asymmetry. Berkeley: University of California Press, 1974.

Earman, J. "An Attempt to Add a Little Direction to 'The Problem of the Direction Of Time.'" Philosophy of Science 41 (1974): 1547.

Horwich, P. Asymmetries in Time. Cambridge, MA: MIT Press, 1987.

Lewis, D. "Counterfactual Dependence and Time's Arrow." In Philosophical Papers. Vol. 2. edited by D. Lewis, 3266. Oxford: Oxford University Press, 19831986.

McCall, S. A Model of the Universe. Oxford: Oxford University Press, 1994.

Mehlberg, H. Time, Causality and the Quantum Theory. Dordrecht, Netherlands: Reidel, 1980.

Mellor, H. Real Time, II. London: Routledge, 1998.

Price, H. Time's Arrow and Archimedes' Point: New Directions for the Physics of Time. New York: Oxford University Press, 1996.

Reichenbach, H. The Direction of Time. Berkeley: University of California Press, 1956.

Sklar, L. Physics and Chance. Cambridge, U.K.: Cambridge University Press, 1993.

Sklar, L. "Up and Down, Left and Right, Past and Future." Nous 15 (1981): 111129.

Strevens, M. Bigger Than Chaos. Cambridge MA: Harvard University Press, 2003

Tooley, M. Time, Tense, and Causation. Oxford: Clarendon Press, 1997.

Lawrence Sklar (1996, 2005)

About this article

Physics and the Direction of Time

Updated About content Print Article