This article is a draft version put up for comments and criticism. Please do not distribute!

A game of dice: physics, randomness, and naturalism

According to Bob Semes, executive director of the Jefferson Center, I was supposed be a “skeptical scientist” in this Summer Institute. But when I first met Alan Wallace, a fellow presenter, I did not have a straightforward answer when he asked me what perspective I represented. I replied “confusion”—which sounded a lot like “Confucian,” so I immediately had some embarrassed explanation to do.

Now, I do in fact argue for a skeptical view: I defend naturalism, or physicalism, or scientific materialism, or whatever you want to call it. I do not think there are any spiritual realities over and above what is realized in the physical world. And my variety of naturalism emphasizes the fundamentally random elements of the universe. We can explain our world as a combination of chance and necessity. Einstein insisted that God does not play dice with the world; I see the world as a game of dice.

I have written a couple of books arguing for such a view, and fortunately Bob Semes found them interesting and invited me to give a presentation. But the Summer Institute is not only interested in science-based arguments concerning religious views; the cultural and ethical aspects of questions about science and religion are just as important. In that case, some degree of confusion might be appropriate. After all, much of religious life does not directly involve supernatural claims; instead, it organizes a personal and communal need for morality and meaning. These are not matters that typically fall in the domain of natural science. Still, it is interesting that belief in a spiritual reality transcending merely natural goings-on is usually central to human views of meaning and morality. When a group of Christians join together in a prayer circle, they may also be engaging a form of therapy or social exchange. But their conviction in a very supernatural Jesus is integral to their prayers. So naturalism is relevant to our wider concerns, if only because it undercuts a very common device for perceiving humanly relevant meaning in the universe. Naturalism does this without presenting any straightforward substitute, hence the confusion.

Let me also clarify my view by contrasting it with liberal religious options. I have considerable sympathy for liberal religion; in a world of resurgent fundamentalisms, religious liberals stand out as being respectable, sophisticated, and humane. A more liberal public religious climate is good for science. Nevertheless, liberal religion also depends on gods, souls, transcendent forces. I find liberal theologians to be frustratingly evasive about science and religion. Almost always, they remain content with a superficial compatibilism. Few supernatural beliefs are subject to direct testing, especially when made suitably vague, but this does not help make them plausible.

And yet, I should add that I do not urge any kind of rationalist triumphalism. Our present scientific understanding also suggests that it will be very difficult for naturalism to gain widespread acceptance. Moreover, in matters involving identity and morality, naturalism puts us in an ambiguous position. I come from the Enlightenment rationalist tradition, but I think Enlightenment rationalists have put too much emphasis on a traditional style of philosophy. We have proclaimed the lawful order of nature, and opposed monotheism with our own variety of moral universalism. I want to draw attention to the more disorderly aspects of nature, and suggest that moral pluralism might make more sense in an accidental world.

Let me put questions about morality and meaning aside for now, and concentrate on why I think naturalism is correct. In a minimal sense, naturalists deny that there are any gods, souls, or ghosts.  There are no disembodied spirits—no supernatural agents of any kind. Yet not only our religions but the stories we tell are full of ghosts, vampires, gods, ancestral spirits, and so on. Supernatural agents are the stuff of horror movies and folktales as much as religious faith. Indeed, cognitive scientists have lately come up with some very interesting ideas about why belief in such beings are so ubiquitous in human cultures. It appears that we are intuitive mind-matter dualists, and that we are naturally drawn to concepts such as bodiless spirits. These concepts catch our attention by violating a basic expectation about the “person” category while still allowing most inferences appropriate to persons. Although such beliefs come to us naturally, naturalists insist that they are mistaken.

Now, some sort of spirit realm could be real—many paranormal claims, for example, could be evidence for such a reality. Parapsychologists have often sought support for dualism, and theologians impressed with the paranormal welcome the opportunity to talk about “agent causation” that is not reducible to physical interactions, about spirits acting on matter. In many ways, parapsychology has been an antimaterialist research program. It has not, however, succeeded. Indeed, it seems any sort of good evidence for the existence of supernatural agents is hard to come by.

If rejecting supernatural agents was all there was to naturalism, it would mainly be an aggressive affirmation of common sense, a kind of finger-wagging skepticism about agents said to violate basic expectations about persons. But there is more to the supernatural than glorified superstition, and in the end, naturalism offends common sense much more than any religion. This becomes clear, for example, when we discuss the question of design in the universe. As creationists know very well, our common sense screams that any order or functional complexity we find looks very much like it is due to intelligent design. And world religions picture our universe hierarchically, as a place shaped from the top down by supernatural agents. Complex order, theologians say, is ultimately due to a rational God who brings order out of chaos. A spoon bent by a psychic, paranormalists tell us, is a sign of spirit or consciousness acting on mere matter to bend it to its will. The complexities of cellular biochemistry, intelligent design creationists declare, must come from information injected into the world by something beyond mere physics. Such a perception of the world has a strong commonsense appeal.

Naturalists go against this common sense, claiming that a bottom-up picture is more accurate. We say that complexity, including life and mind, is assembled out of the lifeless and mindless substrate of mere physics. The molecules that chemists work with, for example, are made out of the particles and forces described by physicists. There are no “chemical souls.” Life, the domain of biology, is in turn made up of processes that the physical sciences study.  There is no “life force” or “animal soul.” Note that this is not at all the same as saying that the most interesting, most fundamental explanations belong to the physicists, and everyone else works out the consequences of a theory of elementary particles. Some of the most exciting work in modern science has to do with explaining complexity, with working out the connections between life and mind and physics. At the very least, biology is a test of physics, since if our understanding of physics cannot set the stage for life, it is clearly incomplete. Moreover, concepts fundamental to understanding complexity, such as thermodynamics, computation, Darwinian evolution and so on, do not depend on the details of microscopic physics.

Evolution, in fact, is a very good illustration. Darwinian variation-and-selection explains complex adaptations by combining chance and necessity—random and orderly behavior—just like all physical explanations. Darwinian evolution is more than the observation that a cockroach and a human have a common ancestor. Evolution is a non-directed, non-progressive process that does not reveal any purpose behind the history of life. Darwinian evolution closely ties in with the physics of complexity, being a wonderful solution to the problem of how information arises in the fundamentally mindless, impersonal world of physics.

In the natural sciences, the bottom-up picture of naturalism is very well supported. Still, many people wonder if theories like evolution can still fit into an ultimately top-down picture of the world. Liberal theists believe that evolution is God’s way of creating, and it is not hard to find theologians endorsing a “lite” version of intelligent design, according to which God is the ultimate source of information. They are, I think, wrong. But to see why, we need to flesh out the bottom-up picture of the word with a new emphasis. Randomness has become very important in our current understanding of the natural world, and this makes it particularly implausible that there is still a God directing nature from behind the scenes.

Physical explanations rely on chance and necessity. We are more used to the notion of the laws of nature—for example, we can calculate the structure of a water molecule by using quantum mechanics, determining how H₂O is a stable structure that has particular bond angles and so forth. We can write down an equation for gravity. Less known is how quantum mechanics is full of random events, such as decays of radioactive nuclei. In general, all physical explanations combine chance and necessity: rules and dice. A good way to phrase my version of physicalism is this: I translate “everything (life, minds, everything) is physically realized” into “everything is explained by combinations of chance and necessity.” Furthermore, chance and necessity are inseparable, in that the most basic laws of nature—the rules—tell us what sort of dice we roll to generate our world.

All this needs some explaining. Let me start by contrasting it to a generic notion of divine design. Intelligent design proponents, and supernaturalists in general, do not think that chance and necessity suffice to account for our world. They often say that intelligence is a separate principle, one that cannot be reduced to physics. Our concern, then, is not just with basic physics but also with purpose, intelligence and persons. When I claim that chance and necessity suffice to describe our world, I also mean that nothing is irreducibly personal—that all agents we know are entirely physical. I argue that artificial intelligence is possible. And I insist that cognitive neuroscience—like biology, a field of study that is continuous with physics—will lead the way to understand minds within a physical world.

Let us step back to physics before building up to brains. After all, we need to know more about what randomness is and how it shows up in physics. One route to randomness is to talk about disorder. Disorder is a major theme in macroscopic, many-particle physics, whether it comes in boundary conditions, thermal physics, or the behavior of objects like a flipped coin. Disorder, however, need not be due to randomness; it might arise from sources such as dynamical chaos. Indeed, in classical (nineteenth century) physics, we can always obtain more information to improve our predictive power.

For true randomness, which is an extreme form of disorder, we have to bring in quantum physics. Radioactive decay events, for example, happen completely at random. The results of quantum measurements cannot be predicted. Moreover, there is no way to obtain further information to improve predictability. The randomness in a quantum event is fundamental—not like the macroscopic, classical disorder in the flip of a coin. And since quantum mechanics is our most fundamental theory of physics, its randomness appears everywhere. In quantum cosmology, for example, it becomes natural to think of multiple universes, with differing physical laws, randomly branching off from one another.

A particularly interesting way accidents come into play in physics is in determining the everyday laws of nature we live by. Physicists expect the most basic laws of physics to be very simple expressions of fundamental symmetries. As the story goes, physicists would like to find a “theory of everything” that would fit on a T-shirt. But clearly, an equation on a T-shirt would contain practically no information. It would tell us very little about the world we live in. The more complex low-temperature world we inhabit comes about through symmetry breaking. Consider a crowd of people milling around in the reception after a talk. They all face different directions; it is a disorderly scene. But in an audience, people are far less energetic. They move around less, are quieter, and they tend to sit facing in the same direction. The reason is that they are interested in hearing a speaker. With symmetry breaking, we get the same sort of order, but without any externally imposed direction. The reason is the mutual interactions of the components of a system. Say we have a group of people whom, for whatever reason, like to sit facing the same direction. Unless they are very energetic, continually walking around a room, they will tend to line up with each other, even without the presence of a speaker. Inn fact, they will face a completely arbitrary direction. Similarly, our world is realized through a process of symmetry breaking that randomly picks out a low temperature environment. The most basic laws of physics only tell us what sort of dice were rolled to generate our history.

In other words, randomness is a fundamental aspect of modern physics. Still, supernaturalists and Enlightenment rationalists both have a distaste for randomness. Many of us suspect randomness is only a label for ignorance. Maybe we do not know the deeper laws underlying the apparent chaos, or maybe a God directs the seemingly random events in cosmology and biological evolution. If physics does not supply us with the hidden causes, perhaps a kind of metaphysical cause beyond nature is exactly what we need. After all, wouldn’t a world that was a game of dice be a formless chaos?

This is another occasion where naturalism must confront common sense. We have a hard time believing our universe can be an accident. For a naturalist, at some level this must be so: explanations have to come to an end, and at some point we can only say the world is as it is and there is no further reason we can discern for that fact. Nevertheless, we keep thinking that everything must have a cause, and that calling something an accident must just cover up an ignorance of the real causes. What, then, of the randomness in modern physics?

We now need a better understanding of randomness. Mathematically, a random sequence is one that lacks any pattern. A series of coin flips that alternates tails and heads, “THTHTHTHTHTHTHTH…,” does not look very random, while one that goes “HTHTTTTHHTHHHTTH…” looks like a better candidate. Why, since any sequence of heads and tails is equally probable with a fair coin? The haphazard one, lacking pattern, looks much more random. Indeed, in the limit of an infinite sequence, that lack of pattern is precisely what defines randomness. Unlike an alternating sequence, we cannot predict the next coin toss in a random sequence. There is no “theory” that reveals a pattern. In both everyday life and in science we depend on our pattern recognition skills to figure out what is going on and make sense of it in terms of causes. A series of random events, being patternless, means we cannot recognize a pattern and place it in a network of causes. A random event is, at some level, an uncaused event. And in physics, we call something random when we see no pattern and no good prospect of finding one.  When we have to say it’s a brute fact.

What, then, about the worry that randomness signifies a formless chaos? This is a misconception, similar to the misconception that randomness stands for ignorance. Consider a demonstration I often do with my students. I give each in a class of about 35 students a coin, and ask them to pretend to be a radioactive nucleus. In each time step, they have an equal chance to decay or to survive. So I tell them to flip their coin: if they get heads they die, if the get tails they live to flip again. After the first round, about 15 to 20 survive, then about 8 or 10, until everyone decays away after about 7 or 8 rounds. Now, the time when each individual in the class decays is unpredictable. But the population of “live nuclei” in the class will tend to halve with each step. Each experiment starting with 35 students is different, but even with such a small number, we see what looks like an exponential decay with fluctuations. If we have the more than 10²⁰ nuclei involved in macroscopic samples, the fluctuations will be miniscule and measurements will be guaranteed to produce a smooth exponential decay curve. Randomness means complete unpredictability of individual events, but also strict statistical predictability in large populations. Randomness is not complete ignorance, and can imply strong predictability—only for populations, not individuals.

So, again: the fundamental laws of physics are full of randomness. They give us a framework for understanding the accidents that produce our history. And this in turn means that everyday notions of cause and effect are not fundamental features of  our universe. Cause and effect emerge from a microscopic substrate of random events.

What, then, about persons? I have been presenting a very bottom-up picture, suggesting that specially spiritual realities are out of place in our universe. But then, it is easy to suspect that physical science should be all about mindless stuff doable through chance and necessity. Why should a physical style of explanation work across the board? Maybe it fails when dealing with persons, with minds—maybe the generic version of intelligent design is correct and intelligence is a separate principle, not reducible to any combination of chance and necessity.

I do not think so. Part of the reason is my judgment that cognitive neuroscience is making slow but real progress in explaining how minds work. Old-fashioned dualism is out of fashion, even if many philosophers continue to favor weaker, more obscurantist cousins of dualism. But another reason comes from some of my own work, which addresses the question of whether chance and necessity can produce real intelligence––genuine creativity.

This is an interesting question, especially when we think about machines that may be able to realize any proposed physical model of a mind. But then, there is an immediate problem: computers, for example, are not creative. Their output is entirely determined by their programming, their past history, and present input. Imagine a piece of paper that said the following:

Günaydın! Bugün hava iyi, ancak yarın daha kötü olacak gibi. Bulut çok, ama ne yapar, belli değil.

For anyone who cannot read Turkish, this does not signify anything. But we still recognize it as very likely to be a meaningful message in a human language. (It says “Good morning! Today the weather is fine, but tomorrow it looks like it will be worse. There are many clouds, but it’s not clear what will happen.”) We also know that even if a machine printed the text, the real source of the message is human. After all, a computer is not capable of creating new information beyond its programming and input.

Indeed, it seems clear that we are not the same as computers. Humans are creative––we are flexible, not bound by pre-programmed rules.  We always might figure out a new way to do things. There is a similar Gödelian critique of machine intelligence, based on the observation that any system of rules is rigid, so it must have blind spots. Humans are supposed to be nonalgorithmic, beyond computer programs and their rigid predetermined rules. Intelligent design proponents adapt this argument to claim that no mere mechanism (including Darwin’s) can be creative.

Humans are, most probably, nonalgorithmic. But it turns out that the missing ingredient for machine creativity and flexibility is randomness. For example, in games where an opponent can adapt to a set strategy, occasional random behavior can be a very good idea. This is because randomness is the best possible source for novelty and unpredictability. As in Darwinian evolution, we need to combine rules and randomness to create new functionality.

This is important, so I will risk some technical language. The most famous nonalgorithmic functions, such as “Turing’s halting function,” are called oracles.  Not only computers, but we also cannot compute them. So we need flexibility without oracles. And a machine can use randomness (which is maximally nonalgorithmic) as a source of novelty, to break out of ruts. Additionally, there is a completeness theorem showing that all infinite sequences are partly random, so that the only tasks beyond rules and randomness are oracles, for which infinite information must be known. In that case, there is very good reason to believe that any human output can be produced by mechanisms that include a random element.

Such a view fits in very well with current trends in artificial intelligence research, and in cognitive and brain science. Indeed, Darwinian variation-and selection is taking on an important role in explaining minds. It appears that intelligence relies on broadly Darwinian processes combining chance and necessity.

If I am right, and even personal properties such as intelligence can be explained through an ultimately physical account, the bottom-up, naturalistic, accidental picture of the world is most likely correct. There are no gods, no souls, no spirits beyond nature. Even so, the implications of my views for religion, or even for the rationality of supernatural belief, are less clear.

That may be an odd thing to say, since I have just tried to make a case that supernatural beliefs are mistaken. Nevertheless, naturalism has significant disadvantages. The naturalistic picture is counterintuitive––it goes against ingrained and socially reinforced habits of thinking. Such habits work well enough, most of the time, at little cost. They will not, and perhaps should not, be discarded easily. After all, naturalism is much more costly––it requires specialized knowledge and training to acquire new habits that go against the grain of human nature. It hard to expect reflective naturalism to become widespread, even in secular Europe.

Another area where naturalism has to prompt an ambiguous response is morality. If we need a clear-cut, practical guide to moral behavior, naturalism does not offer anything immediate. Indeed, my version of naturalism rejects hard moral objectivism, tending toward moral pluralism or “moral ecology.” In other words, I claim that there are many viable, successfully reproducing patterns of interests and ways of life. For most people, most of the time, these will include supernatural beliefs.

So this is why I do not want to end on a note of triumphant rationalism. In fact, let me give liberal religion some backhanded praise. For science-minded nonbelievers such as myself, liberal religion looks too evasive, too concerned to protect “God” from scientific criticism. I like to say that theological conservatives are wrong, but liberals are not even wrong. Nevertheless, practically speaking, liberal religion is a good social compromise. Moreover, the strength of liberal religion in the wider society is vital for science and other enterprises I care deeply about.

This, I admit, puts me in a difficult spot. For example, in the April 24, 2006 issue of The Nation, Michael Lerner accuses scientific naturalists of all sorts of intellectual sins:

The left has been captivated by a belief that has been called scientism…––the belief that the only things are real or can be known are those that can be empirically observed and measured. As a religious person, I don’t rely on science to tell me what is right or wrong or what love means or why my life is important… Claims about God, ethics, beauty and any other face of human experience that is not subject to empirical verification––all these spiritual dimensions of life––are dismissed by the scientistic worldview as inherently unknowable and hence meaningless… The view that what is real and knowable is that which can be empirically verified or measured is a view that itself cannot be empirically measured or verified and thus by its own criterion is unreal or unknowable. It is a religious belief system with powerful adherents.

Now, I think Lerner is profoundly mistaken. “Spiritual dimensions” do not get a free pass to avoid science-based criticism. And science is a much broader enterprise than Lerner conceives! But I find this sort of criticism troubling, since I usually side with Lerner in political matters. He thinks some attenuated sense of the supernatural is vital, and my views are a variety of “scientism.”

So here again is the confusion I face, with no good resolution that I know of. I argue that the question of a specially spiritual reality involves all our sciences, philosophy, history––the best of all our knowledge. And according to the best of our knowledge, we inhabit a universe that is a game of dice. The most accurate view of the world is physical, even random in the end. Naturalism explains the world much better than its rivals. Yet thoughtful naturalists perhaps also have to be ambivalent about naturalism.

Taner Edis is associate professor of physics at Truman State University, Kirksville, MO. His latest book is An Illusion of Harmony: Science and Religion in Islam, to be published by Prometheus Books in 2007.