The Anthropic Principle Revisited

Life, the cosmos and everything - physicsworld.com

Life, the cosmos and everything

Cosmologists who study the link between life in the universe and the values of the physical constants were once viewed with suspicion by other scientists. But a recent high-profile conference at Cambridge showed that the subject is fast becoming academically respectable.

Stars in the making

The notion that certain features of the universe, such as the values of the physical constants, may be constrained by the requirement that intelligent observers can arise was first mooted nearly 40 years ago. This "anthropic principle" has been a focus of controversy (even intense antipathy in some quarters) ever since. However, judging by a conference that took place in Cambridge at the end of August, the notion seems to be attracting the interest of an increasing number of eminent physicists. The meeting - the first in a series supported in part by the Templeton Foundation - took place at the Cambridge home of Martin Rees, one of the foremost advocates of the anthropic principle. Future meetings will address the biological and philosophical aspects of the subject.
What is the anthropic principle?

There are various versions of the anthropic principle. The "weak" version accepts the laws of nature and the values of the physical constants as given and claims that the existence of life then imposes a selection effect on where and when we observe the universe. For example, the current age of the universe cannot be less than the nuclear-burning time of a massive star - otherwise there would not have been enough time for the chemical elements that are essential for life to have been generated by stellar nucleosynthesis. On the other hand, the universe cannot be much older than this because the stars would have all burned out. This means that life can only exist when the universe has roughly its observed age. This is a logical consequence of our existence and is relatively uncontroversial.

The "strong" version of the anthropic principle suggests that the presence of observers imposes constraints on the physical constants themselves. In other words, life could only arise if the constants were close to their observed values. Some people might infer from this the existence of a creator who tailor-made the universe for our benefit. However, cosmologists have recently realized that processes in the early universe may naturally have generated an ensemble of universes, each having different values of the constants. We live in one of the universes that is conducive to life. Even though invoking multiple universes is highly speculative, this makes the strong anthropic principle much more palatable from a physical point of view since it just becomes an aspect of the weak version.

In order to argue that the universe is fine-tuned for the emergence of observers, one must specify who qualifies for this description, and not everybody agrees on this. Brandon Carter, who first coined the term "anthropic principle" in 1974, introduced the meeting by emphasizing that the concept can be refined in various ways according to whether one includes every conceivable observer (including ants and extraterrestrials) or just Homo sapiens. He proposed a "refined" anthropic principle, in which the observer is "weighted" according to the amount of information processed. It is not clear, however, that consciousness is the key feature of the anthropic constraints. Other speakers stressed that many of the fine-tunings are just associated with the development of complexity.
Evidence for the anthropic principle

As Virginia Trimble emphasized, the prerequisites for getting out of bed in the morning are many and varied! In particular, the existence of life (or at least our particular form of it) requires the formation of a hierarchy of structures - planets, stars and galaxies - and, as successive speakers pointed out, each of these seems to require rather special conditions.

Carl Murray focused on planet formation. The discovery of several dozen extra-solar planetary systems in recent years suggests that our solar system is far from unique, although he did emphasize that merely having planets is not enough for life to occur since the Earth seems to have been fortunate in various other ways. For example, it is known that the Moon has played an important role as a climate regulator. If the Moon were much smaller, the spin axis of the Earth would change chaotically - leading to catastrophic weather variations that could exclude the emergence of life. Another fortunate aspect of our solar system is that the outer planets seem to have played an important role in the formation of the inner ones.

Our presence on Earth might be regarded as an example of the weak anthropic principle. Rather more controversial are the anthropic conditions that seem to be associated with stars. I discussed in my talk how these involve constraints between the dimensionless "coupling constants" that describe the strengths of the fundamental interactions - in particular the electric fine-structure constant a = e2/h-bar c ~ 1/137, the gravitational fine-structure constant aG = Gmp2/h-bar c = 5 x 10-39, and also the weak fine-structure constant aW = gmec2/h-bar3 x 10-10, where G is the gravitational constant, g is the Fermi constant, mp is the mass of a proton, h-bar is the Planck constant divided by 2 pi, c is the speed of light and me is the mass of an electron.

It seems that aG must be roughly a20 for both "convective" and "radiative" stars to exist (prerequisites for planets and supernovae, respectively) and roughly aW4 for neutrinos to eject the envelope of a star in a supernova explosion (necessary for the dissemination of heavy elements). These "coincidences" might be regarded as examples of the strong anthropic principle.

Several contributors highlighted an even more striking example associated with stars. This involves the strong interaction and concerns the generation of carbon (another prerequisite for our form of life) in the helium-burning phase of red giant stars. This occurs via a reaction in which two alpha particles unite to form a beryllium nucleus that then combines with another alpha particle to form carbon. However, as the late Fred Hoyle (see Sir Fred Hoyle 1915 - 2001 and page 11 of this issue, print version) first pointed out, the beryllium would decay before interacting with another alpha particle were it not for the existence of a remarkably finely tuned resonance in this interaction. This fact is sometimes presented as an anthropic prediction but, as Trimble intriguingly pointed out, there may have been evidence for this resonance in the data even before Hoyle suggested that it be sought in the laboratory.

Heinz Oberhummer, who has studied this resonance in more detail, reported some beautiful work showing how the amount of oxygen and carbon produced in red giant stars varies with the strength and range of the nucleon interactions. His work indicates that the nuclear interaction must be tuned to at least 0.5% if one is to produce both these elements to the extent required for life.
Cosmological anthropic constraints

The anthropic constraints associated with the formation of galaxies involve various cosmological parameters, such as the density of the matter in the universe, the amplitude of the initial density fluctuations, the photon-to-baryon ratio and the cosmological constant (an extra term Einstein introduced into his field equations for cosmological reasons and which may cause the universe to accelerate). Some of these parameters might be determined by processes in the early universe rather than being prescribed freely as part of the initial conditions. However, as Martin Rees discussed, even small deviations from the observed values of such parameters would exclude the formation of structures like galaxies and their subsequent fragmentation into stars.

Host with the most

An interesting twist on these arguments was provided by Anthony Aguirre, who described anthropic constraints on so-called cold cosmological models, in which the initial ratio of photons to baryons (i.e. ordinary matter like protons and neutrons) is much smaller than currently observed. He pointed out that such models could provide life-supporting conditions with very different values of the cosmological parameters and coupling constants to those found in our universe. Both Rees and Aguirre stressed the importance of calculating the probability distribution for such parameters across the different universes because this is the only way of testing the multiple universe or "multiverse" proposal. For example, if the distribution for the amplitude of the density fluctuations fell off too slowly, we would be surprised to be in a universe with a value as small as is observed.
Fundamental constants

In assessing the anthropic principle, a key issue is whether some fundamental theory will eventually determine all the constants uniquely or whether some of them are contingent on initial conditions or accidental features of symmetry breaking. In the first case there is no room for the anthropic principle and the anthropic fine-tunings must just be regarded as coincidental. In the second case, there may be room for anthropic arguments.

One first has to decide which physical constants should be regarded as fundamental. Unification theories predict relationships between some of the constants, so one is certainly not free to vary all of them. Craig Hogan identified the coupling constants associated with the four interactions and some basic mass-scales (e.g. the masses of the electron and the up and down quarks) as fundamental. Although features of biology are not sensitive to the values of these constants, the existence of stable atoms and an interesting range of chemical elements certainly are. For example, even small changes in the quark and electron masses would make the proton, deuteron or hydrogen atom unstable.

The particle physicists at the meeting expressed various views on how likely such tunings are to result from some fundamental theory. As John Donoghue emphasized, "fine-tuning" arises in various different contexts in particle physics - why, for example, are the cosmological constant and strong charge-parity (CP) violation so small? - even though most of these may have no anthropic significance. However, some of them do and he particularly stressed anthropic constraints on the "vacuum expectation value" of the Higgs field, which determines the masses of all the ordinary particles.

At least some physical parameters would appear to be contingent. For example, Frank Wilczek, who first posited the existence of a light particle called the "axion" in order to explain the lack of CP violation in strong interactions, pointed out that the density of these particles would now be much larger than the baryonic density unless an angle associated with the initial conditions of the axion field were tiny. Such a large axion density would be incompatible with the formation of galaxies - and so is anthropically disallowed. The only reasonable explanation for axions and baryons having comparable densities is to invoke an early "inflationary" phase for the universe, in which it expands exponentially fast due to the effect of a cosmological constant. The axion angle would then have different values in different places and we would necessarily live in a region where this angle was very small.

Most physicists would probably prefer the constants to be determined by more conventional physics. So how likely could that be? The current favourite candidate for a fundamental theory is the string model. This posits that space-time is either 10-dimensional (superstring theory) or 11-dimensional (M-theory), with four-dimensional physics emerging from the compactification of the extra dimensions. Unlike the Standard Model of particle physics, which does not incorporate gravity and contains several dozen free parameters, M-theory may predict all the fundamental constants uniquely. This point was emphasized by Malcolm Perry. The only input would then be the string scale (related to the size of the 11th dimension). However, the situation is probably not as clear-cut as this since M-theory only predicts that the number of vacuum states should be discrete; the constants may be uniquely determined within each one but could be different across the states themselves. The crucial issue is whether the number of vacuum states is sufficiently large and their spacing sufficiently small to allow some room for anthropic constraints. This issue remains unresolved.

A new twist arises if the (so-called) constants vary in time even in our universe. This is expected in many unification theories since the constants should be related to the size of the compact internal dimensions, which would be expected to change during at least part of the universe's history. This theme was taken further by John Barrow. He is part of a team that recently claimed to have found positive evidence for a variation in a of about seven parts in a million by studying absorption lines in several hundred galaxies (see When constants are not constant by Chris Carilli). His attempts to model this effect suggest that a should remain constant during both the early "radiation-dominated" phase of the universe and the late "curvature-dominated" or "cosmological-constant-dominated" phases. However, a can vary over the intermediate matter-dominated phase, which would make it difficult to satisfy the anthropic constraints on a for an extended period if the curvature or cosmological constant were too close to zero.
Quantum cosmology

One reason why many cosmologists now take the anthropic principle seriously is that the "many worlds" interpretation of quantum mechanics seems to be the only sensible context in which to discuss "quantum cosmology" - the branch of physics that tries to describe what happened near the big bang. As emphasized by Jim Hartle, quantum theory allows many mutually incompatible histories. However, it only makes sense to consider the initial conditions that led to the classical behaviour that we observe today. (With complete ignorance of the initial conditions, the quantum fluctuations could be arbitrarily large and the emergence of a classical world would not be possible.) Within this restriction, quantum cosmology allows many different worlds or "branches", all with different values of the constants, and this validates the strong anthropic principle.

Nevertheless, the cosmologists present had widely different views on how the different worlds might arise. Andrei Linde and Alex Vilenkin invoked "eternal" inflation, in which the universe is eternally self-reproducing. This version of inflation predicts that there may be an infinite number of exponentially large domains - all with different laws of low-energy physics and different coupling constants. Indeed, Linde regarded inflation as the only plausible basis for anthropic arguments. Vilenkin argued that there is a well motivated prescription within the eternal-inflation scenario for calculating probability distributions for the various constants, showing that the distributions should be weighted by the volume of the universe in which each set of values pertains.

Brief history of time

On the other hand, Stephen Hawking objected to the eternal-inflation model on the grounds that it extends to the infinite past and thus violates his "no boundary" proposal for the origin of the universe. This proposal requires that the universe start at a finite time and it avoids the initial singularity by requiring time to become imaginary there (i.e. time is multiplied by (-1)1/2 so that the metric starts off Euclidean rather than Lorentzian). Hawking uses the path-integral approach to calculate the probability of a particular history but only sums over those histories that lead to observers.

Neil Turok elaborated on this theme, showing that there are so-called instantons that represent classical solutions of the Euclidean equations that possess a continuation to real Lorentzian space-time. Although the path integral favours inflationary periods shorter than required, anthropic selection can salvage this since one only considers histories containing observers. This permits either open or closed universes but he argued that Hawking's favoured (closed) solution is unstable.
More radical physics?

The final day of the conference focused on more radical deviations from standard physics as well as some philosophical issues. Richard Gott presented another version of the many-worlds principle, speculating how the existence of closed timelike curves in general relativity could allow the universe to create itself. Max Tegmark discussed anthropic constraints on the dimensionality of space and time: three spatial dimensions are required for the stability of planetary orbits and more than one time dimension would destroy causality. He also raised the issue of whether it is sufficient to consider universes with different values of the coupling constants, or whether one should also contemplate universes with different physical laws or even different mathematical foundations. This might be the only way to explain anthropic coincidences if the physical constants within a given set of laws turn out to be uniquely specified.

Bill Stoeger discussed the legitimacy of anthropic arguments. He argued that the weak anthropic principle is a logical necessity - but that the strong version only makes sense if variations in the initial conditions of the universe or the values of the constants or the laws of nature allow some scope for anthropic selection. The multiverse proposal may accommodate this possibility, but how legitimate is it, he argued, to invoke the existence of other universes for which there may never be any direct evidence?

Lee Smolin stressed that it is only justifiable if one has a theory that independently predicts the existence of these universes, and that such a theory, to be scientific, must be falsifiable. He argued that most of the universes should have properties like our own and that this need not be equivalent to requiring the existence of observers.

Smolin's own approach invoked a form of natural selection. He argued that the formation of black holes might generate new universes in which the constants are slightly mutated. In this way, after many generations, the parameter distribution will peak around those values for which black-hole formation is maximized. This proposal involves very speculative physics, since we have no understanding of how the baby universes are born. However, it has the virtue of being testable since one can calculate how many black holes would form if the parameters were different.

A few speakers touched on the issue of consciousness. This is a topic usually eschewed by physicists, but Don Page emphasized that physics is primarily concerned with observations and these are, at root, conscious perceptions. He argued that a particular observer's experience should be a random sample of all conscious experiences and discussed how one might derive the probability measure for this sample. Ultimately this must depend on unknown laws connecting consciousness with physics. Linde also proposed that consciousness might play a crucial role in the world, speculating that it might exist (like space-time) even without matter.

Such considerations may go beyond the domain of legitimate science. But perhaps the main message of the meeting was that developments in modern physics may require one to extend one's view of what constitutes legitimate science anyway. The anthropic principle may not yet have attained complete scientific respectability, but it can no longer be dismissed as nothing more than mere metaphysics.

"Anthropic arguments in fundamental physics and cosmology" was held in Cambridge from 30 August -1 September.
About the author

Bernard Carr is in the Astronomy Unit, School of Mathematical Sciences, Queen Mary, University of London, UK.