Recent developments in cosmology and particle physics suggest there could be many other universes, with different physical constants and possibly even different laws. This proposal could explain the origin of our universe and why it is fine-tuned for the development of life. But are speculations about other universes that can never been seen, based on theories that may never be testable, philosophy or science?
Although the word “universe” literally means all that exists, the longer we have studied the world, the larger it appears to have become. It is not surprising therefore that the usage of this term has changed as we have progressed from the geocentric to heliocentric to galactocentric to cosmocentric view.
Nowadays most cosmologists accept the Big Bang theory, in which the universe started in a state of great compression some 14 billion years ago. In this case, one can never see further than the distance light has travelled since the Big Bang (roughly 40 billion light-years — three times the naïve value because of expansion of space) and this might be taken to define the horizon of the observable universe. However, it would be perverse to claim that nothing exists beyond this distance. One would expect there to be other unobservable expanding domains that are still part of our Big Bang.
Recent developments in cosmology and particle physics have led to the much more radical proposal that there could also be other Big Bangs that might be completely disconnected from ours. The ensemble of universes is then sometimes referred to as the “multiverse”. As we will see, there are many motivations for invoking a multiverse. For some, it is claimed as the inevitable outcome of the physical process that generated our own universe. For others, it is proposed as an explanation for why our universe appears to be fine-tuned for life and consciousness. For others, it is seen as the result of an underlying philosophical stance that “everything that can happen in physics does happen”. The multiverse therefore arises in many different contexts and one needs to distinguish between these in assessing the idea.
It should be stressed at the outset that physicists are polarized about the notion of a multiverse. The title of this article is taken from a recent book (Carr 2007), which is based on three recent conferences on the topic, with contributions from many eminent researchers in the field. The question mark in the title indicates their broad range of attitudes to the multiverse proposal — from strong support through open-minded agnosticism to strong opposition. Nevertheless, there is no doubt that the idea has become increasingly popular in recent years. In his contribution to the book, Frank Wilczek (2007) describes the change in attitude between the first meeting in 2001 and the last one in 2005:
“The previous gathering had a defensive air. It prominently featured a number of physicists who subsisted on the fringes, voices in the wilderness who had for many years promoted strange arguments about conspiracies among fundamental constants and alternative universes. Their concerns and approaches seemed totally alien to the vanguard of theoretical physics, which was busy successfully constructing a unique and mathematically perfect universe. Now the vanguard has marched off to join the prophets in the wilderness.”
Indeed perhaps the most remarkable aspect of the book is that it testifies to the large number of eminent physicists who now find the subject interesting enough to be worth writing about.
Despite this, there is no doubt that the concept of the multiverse raises deep conceptual issues. The problem is that scientific progress has not only changed our view of the universe, it has also changed our view of the nature of science itself, and physicists are divided in their reactions to this. Indeed the authors of this article are also divided. We both accept that the multiverse has explanatory value but we differ on whether it should be regarded as legitimate science. We have written this first part of the article together, because this merely describes the various multiverse scenarios and there is no essential disagreement here. However, we have written separate sections where our views diverge, focusing on seven specific bones of contention. Readers will need to draw their own conclusions but we hope to convey the nature of the controversy.
Different multiverse proposals
Max Tegmark (2003) classifies multiverse scenarios into four different types and we start by describing these. We have mentioned that in the Big Bang theory there should be many expanding domains beyond the horizon distance. Tegmark describes this as the “Level I” multiverse, and it is relatively uncontroversial. If pursued to its logical conclusion, it leads to some bizarre possibilities (like our having identical clones at great distance if space is infinite) and these entail some philosophical problems; but it would be hard to deny its existence if not taken to extremes.
The suggestion that there could be other Big Bangs that are completely disconnected from ours is much more challenging and leads to deeper philosophical difficulties. This sort of multiverse proposal — which Tegmark labels “Level II”— usually arises from attempts to understand how our universe originated. Advocates of the Big Bang theory used to assume that known physics would break down at the BBig Bang itself because it would correspond to a “singularity” of infinite density, so one could never hope to understand what happened there (let alone before it). However, in the last few decades cosmologists have begun to address this question and with remarkable success. So if one has a model for generating our own Big Bang, it is not surprising that it can also produce other Big Bangs. The problem is that physicists have widely different views on how the different universes might arise, so there are competing models for the multiverse. Some of these come from cosmologists and others from particle physicists. Let us first examine the cosmological proposals.
Some invoke “oscillatory” models in which a single universe undergoes cycles of expansion and recollapse (Tolman 1934), though without necessarily understanding what causes the bounce. In this case, the different universes are strung out in time.
Others invoke the “inflationary” scenario, in which our observable domain is a tiny part of a single bubble that underwent an extra-fast accelerated expansion phase at some early time as a result of the effect of a scalar field (Guth 1981). This explains why the universe is so smooth and why it has almost exactly the critical density that separates ever-expanding from recollapsing models. Inflation not only implies that the observable domain is a tiny patch of a much larger universe — some versions also predict that there could be many other bubbles, corresponding to other universes with different properties spread out in space (figure 1). A variant of this idea is “eternal” inflation, in which the universe is continually self-reproducing, so that there are an infinite number of bubbles extending in both space and time (Vilenkin 1983, Linde 1986).
A more radical proposal is to invoke quantum cosmology effects at the Planck time. These occur at around 10−43 s after the Big Bang, when the classical space-time description of general relativity breaks down. In this approach one has a superposition of different histories for the universe and uses what is termed the “path integral” approach to calculate the probability of each of these. This replaces the Big Bang singularity with a bounce — time becoming imaginary there according to Hartle and Hawking (1983)— and leads to a form of the cyclic model. Quantum cosmology is most naturally interpreted in the context of the “many worlds” interpretation of quantum mechanics (Everett 1957), in which the universe branches every time an observation is made (rather than the alternative view in which the wave-function collapses). Tegmark describes this quantum multiverse as “Level III” and it is the oldest scientific form of the idea.
We now turn to multiverse proposals inspired by particle physics. The holy grail of particle physics is to find a “Theory of Everything” that unifies all the known forces. Models that unify the weak, strong and electromagnetic interactions are commonly described as “grand unified theories” and — although still unverified experimentally — have been around for nearly 30 years. Incorporating gravity into this unification has proved more difficult but there have been exciting strides in recent years, with superstring theory being the currently favoured model. There are various versions of superstring theory but they are amalgamated in what is termed “M-theory”. This supposes that the universe has more than the three dimensions of space which we actually observe, with four-dimensional physics emerging from the way in which the extra dimensions are compactified; this is described by what is called a Calabi—Yau manifold.
In one version of M-theory our universe could correspond to a four-dimensional “brane” imbedded in a higher dimensional “bulk” (Randall and Sundrum 1999). In this case, there might be many other branes and collisions between the branes might even generate Big Bangs of the kind that initiated the expansion of our own universe. This might take place repeatedly to give a form of the cyclic model (Steinhardt and Turok 2006).
It was originally hoped that M-theory would predict all the constants of Nature uniquely. However, recent developments suggest that this is not the case and that the number of compactifications could be enormous (e.g. 10500), each one corresponding to a different vacuum state and a different set of constants (Bousso and Polchinksi 2000). This is sometimes described as the “string landscape” scenario. Each solution is associated with a different minimum of the vacuum energy and corresponds to a different universe, so the values of the physical constants would be contingent on which one we happen to occupy (Susskind 2005). A crucial feature of the string landscape proposal is that the vacuum energy would be manifested as what is termed a cosmological constant. This is an extra term in the field equations of general relativity, originally introduced by Einstein to make the universe static. One of the most exciting recent developments in cosmology has been the discovery from observations of distant supernovae that the expansion of the universe is accelerating. This suggests that the density of the universe is dominated by some form of “dark energy” and this is most naturally interpreted as a cosmological constant. It is this discovery that has attracted so many string theorists to the subject.
Finally, what Tegmark describes as the “Level IV” multiverse contains completely disconnected universes, governed by different laws or mathematical structures. The assumption here is that any mathematically possible universe must exist somewhere.
We thus see how a confluence of developments in cosmology and particle physics has led to the popularity of the multiverse proposal. Indeed, the idea might be regarded as the culmination of scientific attempts to understand the largest and smallest scales. This is encapsulated in the image of the Cosmic Uroborus (figure 2), which shows the link between the macrophysical and microphysical domains of structure provided by the various forces. The significance of the head meeting the tail is that distances close to the horizon correspond to very early times, when today's observable universe was compressed to a tiny size. This is why early universe studies have led to an exciting collaboration between particle physicists and cosmologists. As one approaches the intersect point, one encounters the multiverse on the macroscopic side and M-theory on the microscopic side.
The anthropic principle
One of the remarkable features of our universe is that some of the constants of physics seem to be fine-tuned for the emergence of observers (Carter 1974, Carr and Rees 1979, Barrow and Tipler 1986, Hogan 2000, Rees 2001). These fine-tunings — dubbed “anthropic” by Brandon Carter — have been studied for some 30 years and involve both the physical constants and various cosmological parameters. Some of them are summarized in table 1. As far as we know, these anthropic relationships are not predicted by any unified theory and, even if they were, it would be remarkable that the theory should yield exactly the coincidences required. Although anthropos is the Greek for “man”, this is a misnomer because the fine-tunings have nothing to do with Homo sapiens in particular. They just seem necessary if an increasing degree of complexity is to develop as the universe expands and cools. This suggests that the anthropic principle should really be interpreted as a complexity principle.
Anthropic arguments used to be regarded with disdain by many physicists — and in some quarters still are — because they seem to exclude the more usual type of physical explanation for the values of the constants. The fact that people of a theological disposition interpreted the fine-tunings as evidence for a creator perhaps enhanced that reaction. Three very different views of the anthropic principle are illustrated by the quotations on page 2.32 from Freeman Dyson (1979), Heinz Pagels (1985) and Brandon Carter (1974). However, the multiverse proposal has led to a shift in the status of anthropic arguments because the constants may be different in the other universes. We have seen that this arises explicitly in the string landscape scenario and the constants may also vary in the different bubbles of the inflationary scenario. So although multiverse models have not generally been motivated by an attempt to explain the anthropic fine-tuning, it now seems clear that the two concepts are interlinked. For if there are many universes, the question arises as to why we inhabit this particular one and (at the very least) one would have to concede that our own existence is a relevant selection effect. Many physicists therefore regard the multiverse as providing the most natural explanation of the anthropic fine-tunings. If one wins the lottery, it is natural to infer that one is not the only person to have bought a ticket.
A multiverse with varied physical properties is certainly one possible explanation for fine-tunings: an infinite set of universes allows all possibilities and combinations to occur, so somewhere — just by chance — things will work out right for life. In assessing this view, a key issue is whether some of the physical constants are contingent on accidental features of symmetry breaking and the initial conditions of our universe or whether some fundamental theory will determine all of them uniquely. The two cases essentially correspond to the multiverse and single universe options (figure 3). This relates to a famous question posed by Einstein: “Did God have any choice when he created the universe?” If the answer is no, there would be no room for the anthropic principle. Most physicists would prefer the physical constants to be determined uniquely, but we have seen that this now appears unlikely. What we call “laws of Nature” may be local by-laws, in which case trying to predict the values of the constants may be as forlorn as Kepler's attempts to predict the spacing of the planets in our solar system based on the properties of Platonic solids.
A particularly interesting anthropic argument is associated with the cosmological constant (denoted by Λ). In the string landscape picture one might expect the value of Λ across the different universes to have a uniform distribution ranging from minus to plus the Planck value (which is 120 orders of magnitude larger than observed). The actual value therefore seems implausibly small. There is also the puzzling feature that the observed vacuum density is currently very similar to the mean matter density, a coincidence that would only apply at a particular cosmological epoch. However, as pointed out by Steven Weinberg (1987), the value of Λ is constrained anthropically because galaxies could not form (and hence life could not arise) if it were much larger than observed. So anthropic considerations in a multiverse with a wide spread of values of Λ in different domains mean that the value we observe will be much smaller than in almost any other domain. This is not the only explanation for the smallness of Λ but there is a reluctant acceptance that it may be the most plausible one.
One important question is whether our universe is typical or atypical within the ensemble. Advocates of the anthropic principle usually assume that life forms similar to our own will be possible in only a tiny subset of universes. More general life forms may be possible in a somewhat larger subset but life will not be possible everywhere. On the other hand, by invoking a Copernican perspective, Lee Smolin (1997) has argued that most of the universes should have properties like our own, so that we are typical. His own model proposes that the physical constants have evolved to their present values through a process akin to mutation and natural selection. The assumption is that whenever matter gets sufficiently compressed to undergo gravitational collapse into a black hole, it gives birth to another expanding universe in which the fundamental constants are slightly mutated. Our own universe may itself have been generated in this way (i.e. via gravitational collapse in some parent universe). Cosmological models with constants permitting the formation of black holes will therefore produce progeny (which may each produce further black holes since the constants are nearly the same), whereas those with the wrong constants will be infertile. A Darwinian process can take place, leading preferentially to universes that produce many black holes; in this case, life may be incidental.
But is the multiverse science?
Despite the growing popularity of the multiverse proposal, many physicists remain deeply uncomfortable with it. One should note that the proposal being made is that there is a really existing multiverse. Nobody has any problem imagining a hypothetical or potential ensemble of universes — cosmologists do that all the time. The question is whether such an ensemble exists in physical reality. The idea is highly speculative and, from both a cosmological and particle physics perspective, the reality of a multiverse is currently untestable — and it may always remain so. That is to say, astronomers may never be able to observe the other universes with their telescopes and particle physicists may never be able to detect the extra dimensions with their accelerators. So although physicists such as Leonard Susskind favour the multiverse because it does away with the need for a creator, other physicists regard the idea as just as metaphysical.
Martin Rees (2001) defends the notion that the multiverse is part of science by invoking what he calls the “slippery slope” argument (figure 4). Not everyone is convinced — indeed it is one of the bones of contention we discuss later — but it highlights the difficulty of delineating a clear boundary between scientific and non-scientific speculations. For defences of the multiverse idea, see Deutsch (1997), Lewis (2000), Rees (2001), Tegmark (2003), Susskind (2006) and Vilenkin (2006). For criticisms, see Gardner (2003), Ellis (2004) and Smolin (2007).
Three views on the anthropic principle
“I do not feel like an alien in this universe. The more I examine the universe and examine the details of its architecture, the more evidence I find that the universe in some sense must have known we were coming.”
“The influence of the anthropic principle on contemporary cosmological models has been sterile. It has explained nothing and it has even had a negative influence. I would opt for rejecting the anthropic principle as needless clutter in the conceptual repertoire of science.”
“The anthropic principle is a middle ground between the primitive anthropocentrism of the pre-Copernican age and the equally unjustifiable antithesis that no place or time in the universe can be privileged in any way.”