Anthropic Reasoning

See allHide authors and affiliations

Science  12 Aug 2005:
Vol. 309, Issue 5737, pp. 1022-1023
DOI: 10.1126/science.1111446

Does extraterrestrial intelligent life exist? The fact that we can even ask this question relies on an important truth: The properties of our universe have allowed complexity (of the type that characterizes humans) to emerge. Obviously, the biological details of humans and their emergence depend on contingent features of Earth and its history. However, some requirements would seem generic for any form of life: galaxies, stars, and (probably) planets had to form; nucleosynthesis in stars had to give rise to atoms such as carbon, oxygen, and iron; and these atoms had to be in a stable environment where they could combine to form the molecules of life.

We can imagine universes where the constants of physics and cosmology have different values. Many such “counterfactual” universes would not have allowed the chain of processes that could have led to any kind of advanced life. For instance, even a universe with the same physical laws and the same values of all physical constants but one—a cosmological constant Λ (the “pressure” of the physical vacuum) higher by more than an order of magnitude—would have expanded so fast that no galaxies could have formed. Other properties that appear to have been crucial for the emergence of complexity are (i) the presence of baryons (particles such as protons and neutrons); (ii) the fact that the universe is not infinitely smooth, allowing for the formation of structure (quantified as the amplitude of the fluctuations in the cosmic microwave background, Q); and (iii) a gravitational force that is weaker by a factor of nearly 1040 than the microphysical forces that act within atoms and molecules—were gravity not so weak, there would not be such a large difference between the atomic and the cosmic scales of mass, length, and time.

A key challenge confronting 21st-century physics is to decide which of these dimensionless parameters such as Q and Λ are truly fundamental—in the sense of being explicable within the framework of an ultimate, unified theory—and which are merely accidental. The possibility that some are accidental has certainly become viable in the context of the “eternal inflation” scenario (1-3), where there are an infinity of separate “big bangs” within an exponentially expanding substratum. Some versions of string theory allow a huge variety of vacua, each characterized by different values of Λ (or even different dimensionality) (4). Both these concepts entail the existence of a vast ensemble of pocket universes—a “multiverse.” If some physical constants are not fundamental, then they may take different values in different members of the ensemble. Consequently, some pocket universes may not allow complexity or intelligent life to evolve within them. Humans would clearly have to find themselves in a pocket universe that is “biophilic.” Some otherwise puzzling features of our universe may then simply be the result of the epoch in which we exist and can observe. In other words, the values of the accidental constants would have to be within the ranges that would have allowed intelligent life to develop. The process of delineating and investigating the consequences of these biophilic domains is what has become known as anthropic reasoning.

You are here.

Plot of cosmological constant Λ versus amplitude of fluctuations in cosmic microwave background Q. Shaded region shows conditions that allow existence of complexity.


Anthropic considerations are beginning to be seriously discussed, especially in relation to dark energy. During the past 7 years, it has become clear that the expansion of the universe is accelerating (5, 6), with dark energy contributing about 70% of the critical density required to sustain a geometrically flat universe (7). The question that arises is why we happen to live in the first and probably only time in the history of the universe in which the matter density and dark energy density are roughly equal.

The questions used to be: Why should empty space exert a force? Why should there be a cosmological constant Λ? Now we ask: Why is the force so small? If there was an inflationary era with a large cosmic repulsion, how could that force have switched off (or somehow have been neutralized) with such amazing precision? In our present universe, Λ is lower by a factor of about 10120 than the value that seems natural to theorists.

Following the original suggestion by Weinberg, some models have been constructed in which the cosmological constant is a random variable, whose a priori probability distribution is determined by the laws of physics (in the framework of inflationary cosmology). If we assume that we are typical (mediocre) observers, the approximate coincidence between the time when the dark energy starts to dominate the cosmic energy density and the present age of the universe does seem to find a natural explanation: If Λ were larger, then the acceleration would have overwhelmed gravity before galaxies had a chance to form (8, 9).

The situation becomes more complex when more than one physical parameter is postulated to be a random variable. For instance, Q, the amplitude of the fluctuations generated in the aftermath of the big bang, could take different values in other universes (10). However, in the anthropically allowed domain, Λ and Q could be correlated, in the sense that in a universe that has higher-amplitude density fluctuations, structure could still form even in the presence of considerable acceleration. For Q values smaller than about 10−6, only small structures of dark matter would form, and even those would do so rather late in the universe's life (see the figure). Gas within these structures would be so dilute that it would not cool radiatively, thereby precluding star formation. For values of Q larger than about 10−3, large structures would collapse gravitationally to form mostly monstrous black holes. For any given value of Q, above a certain value of Λ no galactic-mass bound systems would form before accelerating expansion commences. Clearly, our own pocket universe lies within the anthropically allowed domain. However, a more definitive answer to the question of how “typical” our location is will have to await a better understanding of the probability distributions of Λ and Q over the ensemble of universes.

We would argue that both the multiverse and anthropic reasoning are bona fide topics for (albeit speculative) scientific discourse. Improved understanding of string theory or inflation may give us firmer grounds for assessing whether our big bang was unique, or merely one of many. In the latter case, some of the physical constants and cosmic numbers that we have hitherto regarded as fundamental are in fact accidental. Yet the mere mention of “anthropic reasoning” and the “multiverse” tends to raise the blood pressure of some physicists. There are two reasons for this.

First, the potential existence of an ensemble of unobservable universes appears to be in conflict with the “scientific method” (which requires theories to be falsifiable by observations or experiments) and therefore in the realm of metaphysics. But there is actually a “fuzzy” boundary between what we define to be observable and what is not. The capabilities of present-day telescopes obviously imply a “horizon” beyond which nothing can be observed; a more fundamental limit to observations lies at the particle horizon, that is, the spherical surface around us (effectively at infinite redshift) from which photons emitted at the big bang are just reaching us.

In the simplest universe described in the textbooks, the Einstein-de Sitter model, the expansion decelerates, the redshift of any galaxy decreases, and all galaxies currently lying beyond the horizon will eventually come within it; they will thus become observable in the far future. But we are not in an Einstein-de Sitter universe: We are in a flat but accelerating one. In this universe, any galaxies now beyond the horizon will stay that way. A generic accelerating universe contains galaxies that we can never, even in principle, observe. If there are galaxies in “our” pocket universe—the aftermath of our big bang—that will never be observable, is it then much of a leap to envisage unobservable galaxies that came into being from other big bangs—part of a multiverse?

As mentioned above, there are theories that predict many big bangs: One possibility is the existence of brane worlds, that is, many universes embedded in a higher dimensional space. Another is “eternal inflation.” What we have traditionally called “the universe” could be just one patch of space-time in a vast cosmic archipelago—a “landscape,” some call it.

We do not know if these theories are correct. But they are speculative science, not metaphysics. Could there be many big bangs? If so, are they characterized by a range of values of the cosmic numbers—Q, Λ, and so forth? Would the same physics apply throughout the multiverse? What could give us confidence in unobservable universes?

The answer is clear—we will believe in them if they are predicted by a theory that gains credibility because it accounts for things that we can observe. We believe in quarks, and in what general relativity says about the inside of black holes, because our inferences are based on theories corroborated in other ways. Specifically, if a theory has testable and falsifiable predictions in the observable parts of the universe, we should seriously consider and be prepared to accept its predictions in parts of the universe (or multiverse) that are not accessible to direct observations.

A second reason why some physicists are hostile to the multiverse concept is that anthropic reasoning seems to point to a fundamental limitation of physics—even the “end of physics.” But this objection is, in our opinion, a purely psychological one. Physicists would like, above all else, to discover a uniquely self-consistent set of equations that determines all microphysical constants, and the recipe for the big bang. They therefore hope that future theories will reveal that all physical parameters are uniquely determined. But there is no reason why physical reality should be structured according to their preferences. It is good that many physicists are motivated to seek a theory that uniquely derives all fundamental numbers and constants, but they may be doomed to failure.

The quest for first-principles explanations may prove as vain as Kepler's quest for a beautiful mathematical formula that described the solar system. If future developments bear out the possibility of a multiverse, then anthropic arguments will offer the only “explanation” that we will ever have for some features of our universe. At the moment, we have no firm reason to close off any of the options. In view of our current ignorance as to what is truly fundamental and what is not, we should keep an open mind about all the options. What we have traditionally called fundamental constants and laws could be mere parochial bylaws in our cosmic patch. They might derive from some overarching theory governing the ensemble, but might not be uniquely fixed by that theory.

Finally, one may wonder whether anthropic reasoning has any predictive power. In principle it has. For instance, imagine that the cosmological constant is the only random variable. If, as some arguments suggest, it is drawn from a flat probability distribution, then in a random member of the multiverse one would expect it to take a high value. In the subset of anthropically allowed universes, however, clearly there is some upper limit above which structure and complexity would not emerge. If our universe were a mediocre member of the ensemble (as Copernican humility would require), the expectation is that the value of the cosmological constant in our universe would not be much smaller than this upper limit. Put differently, if observations showed that the cosmological constant is smaller than the anthropic threshold by a factor of 105, this would make any anthropic arguments seriously questionable. As it turns out, however, the value determined from observations of high-redshift supernovae and from the spectrum of the fluctuations in the cosmic microwave background is smaller than the threshold by only a factor of 5 to 10, not inconsistent with anthropic expectations.

The next few decades are expected to witness better constraints on the nature of dark energy, experimental tests of supersymmetry and symmetry breaking, and perhaps the detection of the gravity waves that originated from inflation (11). These and other unexpected discoveries will undoubtedly shed some light on the reality of the multiverse and on the uniqueness (or not) of the laws of physics. Our universe isn't the neatest and simplest. It has the rather arbitrary-seeming mix of ingredients in the parameter range that allows us to exist. Until we know for sure which type of universe or multiverse we live in, anthropic reasoning is certainly one option in the physicists' arsenal.


View Abstract

Navigate This Article