The Grand Design - Stephen Hawking, Leonard Mlodinow (2010)
Chapter 4. Alternative Histories
N 1999 A TEAM OF PHYSICISTS in Austria fired a series of soccer-ball-shaped molecules toward a barrier. Those molecules, each made of sixty carbon atoms, are sometimes called buckyballs because the architect Buckminster Fuller built buildings of that shape. Fuller’s geodesic domes were probably the largest soccer-ball-shaped objects in existence. The buckyballs were the smallest. The barrier toward which the scientists took their aim had, in effect, two slits through which the buckyballs could pass. Beyond the wall, the physicists situated the equivalent of a screen to detect and count the emergent molecules.
If we were to set up an analogous experiment with real soccer balls, we would need a player with somewhat shaky aim but with the ability to launch the balls consistently at a speed of our choosing. We would position this player before a wall in which there are two gaps. On the far side of the wall, and parallel to it, we would place a very long net. Most of the player’s shots would hit the wall and bounce back, but some would go through one gap or the other, and into the net. If the gaps were only slightly larger than the balls, two highly collimated streams would emerge on the other side. If the gaps were a bit wider than that, each stream would fan out a little, as shown in the figure below.
Notice that if we closed off one of the gaps, the corresponding stream of balls would no longer get through, but this would have no effect on the other stream. If we reopened the second gap, that would only increase the number of balls that land at any given point on the other side, for we would then get all the balls that passed through the gap that had remained open, plus other balls coming from the newly opened gap. What we observe with both gaps open, in other words, is the sum of what we observe with each gap in the wall separately opened. That is the reality we are accustomed to in everyday life. But that’s not what the Austrian researchers found when they fired their molecules.
In the Austrian experiment, opening the second gap did indeed increase the number of molecules arriving at some points on the screen—but it decreased the number at others, as in the figure below. In fact, there were spots where no buckyballs landed when both slits were open but where balls did land when only one or the other gap was open. That seems very odd. How can opening a second gap cause fewer molecules to arrive at certain points?
We can get a clue to the answer by examining the details. In the experiment, many of the molecular soccer balls landed at a spot centered halfway between where you would expect them to land if the balls went through either one gap or the other. A little farther out from that central position very few molecules arrived, but a bit farther away from the center than that, molecules were again observed to arrive. This pattern is not the sum of the patterns formed when each gap is opened separately, but you may recognize it from Chapter 3 as the pattern characteristic of interfering waves. The areas where no molecules arrive correspond to regions in which waves emitted from the two gaps arrive out of phase, and create destructive interference; the areas where many molecules arrive correspond to regions where the waves arrive in phase, and create constructive interference.
In the first two thousand or so years of scientific thought, ordinary experience and intuition were the basis for theoretical explanation. As we improved our technology and expanded the range of phenomena that we could observe, we began to find nature behaving in ways that were less and less in line with our everyday experience and hence with our intuition, as evidenced by the experiment with buckyballs. That experiment is typical of the type of phenomena that cannot be encompassed by classical science but are described by what is called quantum physics. In fact, Richard Feynman wrote that the double-slit experiment like the one we described above “contains all the mystery of quantum mechanics.”
The principles of quantum physics were developed in the first few decades of the twentieth century after Newtonian theory was found to be inadequate for the description of nature on the atomic—or subatomic—level. The fundamental theories of physics describe the forces of nature and how objects react to them. Classical theories such as Newton’s are built upon a framework reflecting everyday experience, in which material objects have an individual existence, can be located at definite locations, follow definite paths, and so on. Quantum physics provides a framework for understanding how nature operates on atomic and subatomic scales, but as we’ll see in more detail later, it dictates a completely different conceptual schema, one in which an object’s position, path, and even its past and future are not precisely determined. Quantum theories of forces such as gravity or the electromagnetic force are built within that framework.
Can theories built upon a framework so foreign to everyday experience also explain the events of ordinary experience that were modeled so accurately by classical physics? They can, for we and our surroundings are composite structures, made of an unimaginably large number of atoms, more atoms than there are stars in the observable universe. And though the component atoms obey the principles of quantum physics, one can show that the large assemblages that form soccer balls, turnips, and jumbo jets—and us—will indeed manage to avoid diffracting through slits. So though the components of everyday objects obey quantum physics, Newton’s laws form an effective theory that describes very accurately how the composite structures that form our everyday world behave.
That might sound strange, but there are many instances in science in which a large assemblage appears to behave in a manner that is different from the behavior of its individual components. The responses of a single neuron hardly portend those of the human brain, nor does knowing about a water molecule tell you much about the behavior of a lake. In the case of quantum physics, physicists are still working to figure out the details of how Newton’s laws emerge from the quantum domain. What we do know is that the components of all objects obey the laws of quantum physics, and the Newtonian laws are a good approximation for describing the way macroscopic objects made of those quantum components behave.
The predictions of Newtonian theory therefore match the view of reality we all develop as we experience the world around us. But individual atoms and molecules operate in a manner profoundly different from that of our everyday experience. Quantum physics is a new model of reality that gives us a picture of the universe. It is a picture in which many concepts fundamental to our intuitive understanding of reality no longer have meaning.
The double-slit experiment was first carried out in 1927 by Clinton Davisson and Lester Germer, experimental physicists at Bell Labs who were studying how a beam of electrons—objects much simpler than buckyballs—interacts with a crystal made of nickel. The fact that matter particles such as electrons behave like water waves was the type of startling experiment that inspired quantum physics. Since this behavior is not observed on a macroscopic scale, scientists have long wondered just how large and complex something could be and still exhibit such wavelike properties. It would cause quite a stir if the effect could be demonstrated using people or a hippopotamus, but as we’ve said, in general, the larger the object the less apparent and robust are the quantum effects. So it is unlikely that any zoo animals will be passing wavelike through the bars of their cages. Still, experimental physicists have observed the wave phenomenon with particles of ever-increasing size. Scientists hope to replicate the buckyball experiment someday using a virus, which is not only far bigger but also considered by some to be a living thing.
There are only a few aspects of quantum physics needed to understand the arguments we will make in later chapters. One of the key features is wave/particle duality. That matter particles behave like a wave surprised everyone. That light behaves like a wave no longer surprises anyone. The wavelike behavior of light seems natural to us and has been considered an accepted fact for almost two centuries. If you shine a beam of light on the two slits in the above experiment, two waves will emerge and meet on the screen. At some points their crests or troughs will coincide and form a bright spot; at others the crests of one beam will meet the troughs of the other, canceling them, and leaving a dark area. The English physicist Thomas Young performed this experiment in the early nineteenth century, convincing people that light was a wave and not, as Newton had believed, composed of particles.
Though one might conclude that Newton was wrong to say that light was not a wave, he was right when he said that light can act as if it is composed of particles. Today we call them photons. Just as we are composed of a large number of atoms, the light we see in everyday life is composite in the sense that it is made of a great many photons—even a 1-watt night-light emits a billion billion each second. Single photons are not usually evident, but in the laboratory we can produce a beam of light so faint that it consists of a stream of single photons, which we can detect as individuals just as we can detect individual electrons or buckyballs. And we can repeat Young’s experiment employing a beam sufficiently sparse that the photons reach the barrier one at a time, with a few seconds between each arrival. If we do that, and then add up all the individual impacts recorded by the screen on the far side of the barrier, we find that together they build up the same interference pattern that would be built up if we performed the Davisson-Germer experiment but fired the electrons (or buckyballs) at the screen one at a time. To physicists, that was a startling revelation: If individual particles interfere with themselves, then the wave nature of light is the property not just of a beam or of a large collection of photons but of the individual particles.
Another of the main tenets of quantum physics is the uncertainty principle, formulated by Werner Heisenberg in 1926. The uncertainty principle tells us that there are limits to our ability to simultaneously measure certain data, such as the position and velocity of a particle. According to the uncertainty principle, for example, if you multiply the uncertainty in the position of a particle by the uncertainty in its momentum (its mass times its velocity) the result can never be smaller than a certain fixed quantity, called Planck’s constant. That’s a tongue-twister, but its gist can be stated simply: The more precisely you measure speed, the less precisely you can measure position, and vice versa. For instance, if you halve the uncertainty in position, you have to double the uncertainty in velocity. It is also important to note that, compared with everyday units of measurement such as meters, kilograms, and seconds, Planck’s constant is very small. In fact, if reported in those units, it has the value of about 6/10,000,000,000,000,000,000,000,000,000,000,000. As a result, if you pinpoint a macroscopic object such as a soccer ball, with a mass of one-third of a kilogram, to within 1 millimeter in any direction, we can still measure its velocity with a precision far greater than even a billionth of a billionth of a billionth of a kilometer per hour. That’s because, measured in these units, the soccer ball has a mass of 1/3, and the uncertainty in position is 1/1,000. Neither is enough to account for all those zeroes in Planck’s constant, and so that role falls to the uncertainty in velocity. But in the same units an electron has a mass of .000000000000000000000000000001, so for electrons the situation is quite different. If we measure the position of an electron to a precision corresponding to roughly the size of an atom, the uncertainty principle dictates that we cannot know the electron’s speed more precisely than about plus or minus 1,000 kilometers per second, which is not very precise at all.
According to quantum physics, no matter how much information we obtain or how powerful our computing abilities, the outcomes of physical processes cannot be predicted with certainty because they are not determined with certainty. Instead, given the initial state of a system, nature determines its future state through a process that is fundamentally uncertain. In other words, nature does not dictate the outcome of any process or experiment, even in the simplest of situations. Rather, it allows a number of different eventualities, each with a certain likelihood of being realized. It is, to paraphrase Einstein, as if God throws the dice before deciding the result of every physical process. That idea bothered Einstein, and so even though he was one of the fathers of quantum physics, he later became critical of it.
Quantum physics might seem to undermine the idea that nature is governed by laws, but that is not the case. Instead it leads us to accept a new form of determinism: Given the state of a system at some time, the laws of nature determine the probabilities of various futures and pasts rather than determining the future and past with certainty. Though that is distasteful to some, scientists must accept theories that agree with experiment, not their own preconceived notions.
What science does demand of a theory is that it be testable. If the probabilistic nature of the predictions of quantum physics meant it was impossible to confirm those predictions, then quantum theories would not qualify as valid theories. But despite the probabilistic nature of their predictions, we can still test quantum theories. For instance, we can repeat an experiment many times and confirm that the frequency of various outcomes conforms to the probabilities predicted. Consider the buckyball experiment. Quantum physics tells us that nothing is ever located at a definite point because if it were, the uncertainty in momentum would have to be infinite. In fact, according to quantum physics, each particle has some probability of being found anywhere in the universe. So even if the chances of finding a given electron within the double-slit apparatus are very high, there will always be some chance that it could be found instead on the far side of the star Alpha Centauri, or in the shepherd’s pie at your office cafeteria. As a result, if you kick a quantum buckyball and let it fly, no amount of skill or knowledge will allow you to say in advance exactly where it will land. But if you repeat that experiment many times, the data you obtain will reflect the probability of finding the ball at various locations, and experimenters have confirmed that the results of such experiments agree with the theory’s predictions.
It is important to realize that probabilities in quantum physics are not like probabilities in Newtonian physics, or in everyday life. We can understand this by comparing the patterns built up by the steady stream of buckyballs fired at a screen to the pattern of holes built up by players aiming for the bull’s-eye on a dartboard. Unless the players have consumed too much beer, the chances of a dart landing near the center are greatest, and diminish as you go farther out. As with the buckyballs, any given dart can land anywhere, and over time a pattern of holes that reflects the underlying probabilities will emerge. In everyday life we might reflect that situation by saying that a dart has a certain probability of landing in various spots; but if we say that, unlike the case of the buckyballs, it is only because our knowledge of the conditions of its launch is incomplete. We could improve our description if we knew exactly the manner in which the player released the dart, its angle, spin, velocity, and so forth. In principle, then, we could predict where the dart will land with a precision as great as we desire. Our use of probabilistic terms to describe the outcome of events in everyday life is therefore a reflection not of the intrinsic nature of the process but only of our ignorance of certain aspects of it.
Probabilities in quantum theories are different. They reflect a fundamental randomness in nature. The quantum model of nature encompasses principles that contradict not only our everyday experience but our intuitive concept of reality. Those who find those principles weird or difficult to believe are in good company, the company of great physicists such as Einstein and even Feynman, whose description of quantum theory we will soon present. In fact, Feynman once wrote, “I think I can safely say that nobody understands quantum mechanics.” But quantum physics agrees with observation. It has never failed a test, and it has been tested more than any other theory in science.
In the 1940s Richard Feynman had a startling insight regarding the difference between the quantum and Newtonian worlds. Feynman was intrigued by the question of how the interference pattern in the double-slit experiment arises. Recall that the pattern we find when we fire molecules with both slits open is not the sum of the patterns we find when we run the experiment twice, once with just one slit open, and once with only the other open. Instead, when both slits are open we find a series of light and dark bands, the latter being regions in which no particles land. That means that particles that would have landed in the area of the dark band if, say, only slit one was open, do not land there when slit two is also open. It seems as if, somewhere on their journey from source to screen, the particles acquire information about both slits. That kind of behavior is drastically different from the way things seem to behave in everyday life, in which a ball would follow a path through one of the slits and be unaffected by the situation at the other.
According to Newtonian physics—and to the way the experiment would work if we did it with soccer balls instead of molecules—each particle follows a single well-defined route from its source to the screen. There is no room in this picture for a detour in which the particle visits the neighborhood of each slit along the way. According to the quantum model, however, the particle is said to have no definite position during the time it is between the starting point and the endpoint. Feynman realized one does not have to interpret that to mean that particles take no path as they travel between source and screen. It could mean instead that particles take every possible path connecting those points. This, Feynman asserted, is what makes quantum physics different from Newtonian physics. The situation at both slits matters because, rather than following a single definite path, particles take every path, and they take them all simultaneously! That sounds like science fiction, but it isn’t. Feynman formulated a mathematical expression—the Feynman sum over histories—that reflects this idea and reproduces all the laws of quantum physics. In Feynman’s theory the mathematics and physical picture are different from that of the original formulation of quantum physics, but the predictions are the same.
In the double-slit experiment Feynman’s ideas mean the particles take paths that go through only one slit or only the other; paths that thread through the first slit, back out through the second slit, and then through the first again; paths that visit the restaurant that serves that great curried shrimp, and then circle Jupiter a few times before heading home; even paths that go across the universe and back. This, in Feynman’s view, explains how the particle acquires the information about which slits are open—if a slit is open, the particle takes paths through it. When both slits are open, the paths in which the particle travels through one slit can interfere with the paths in which it travels through the other, causing the interference. It might sound nutty, but for the purposes of most fundamental physics done today—and for the purposes of this book—Feynman’s formulation has proved more useful than the original one.
Feynman’s view of quantum reality is crucial in understanding the theories we will soon present, so it is worth taking some time to get a feeling for how it works. Imagine a simple process in which a particle begins at some location A and moves freely. In the Newtonian model that particle will follow a straight line. After a certain precise time passes, we will find the particle at some precise location B along that line. In Feynman’s model a quantum particle samples every path connecting A and B, collecting a number called a phase for each path. That phase represents the position in the cycle of a wave, that is, whether the wave is at a crest or trough or some precise position in between. Feynman’s mathematical prescription for calculating that phase showed that when you add together the waves from all the paths you get the “probability amplitude” that the particle, starting at A, will reach B. The square of that probability amplitude then gives the correct probability that the particle will reach B.
The phase that each individual path contributes to the Feynman sum (and hence to the probability of going from A to B) can be visualized as an arrow that is of fixed length but can point in any direction. To add two phases, you place the arrow representing one phase at the end of the arrow representing the other, to get a new arrow representing the sum. To add more phases, you simply continue the process. Note that when the phases line up, the arrow representing the total can be quite long. But if they point in different directions, they tend to cancel when you add them, leaving you with not much of an arrow at all. The idea is illustrated in the figures below.
To carry out Feynman’s prescription for calculating the probability amplitude that a particle beginning at a location A will end up at a location B, you add the phases, or arrows, associated with every path connecting A and B. There are an infinite number of paths, which makes the mathematics a bit complicated, but it works. Some of the paths are pictured below.
Feynman’s theory gives an especially clear picture of how a Newtonian world picture can arise from quantum physics, which seems very different. According to Feynman’s theory, the phases associated with each path depend upon Planck’s constant. The theory dictates that because Planck’s constant is so small, when you add the contribution from paths that are close to each other the phases normally vary wildly, and so, as in the figure above, they tend to add to zero. But the theory also shows that there are certain paths for which the phases have a tendency to line up, and so those paths are favored; that is, they make a larger contribution to the observed behavior of the particle. It turns out that for large objects, paths very similar to the path predicted by Newton’s will have similar phases and add up to give by far the largest contribution to the sum, and so the only destination that has a probability effectively greater than zero is the destination predicted by Newtonian theory, and that destination has a probability that is very nearly one. Hence large objects move just as Newton’s theory predicts they will.
So far we have discussed Feynman’s ideas in the context of the double-slit experiment. In that experiment particles are fired toward a wall with slits, and we measure the location, on a screen placed beyond the wall, at which the particles end up. More generally, instead of just a single particle Feynman’s theory allows us to predict the probable outcomes of a “system,” which could be a particle, a set of particles, or even the entire universe. Between the initial state of the system and our later measurement of its properties, those properties evolve in some way, which physicists call the system’s history. In the double-slit experiment, for example, the history of the particle is simply its path. Just as for the double-slit experiment the chance of observing the particle to land at any given point depends upon all the paths that could have gotten it there, Feynman showed that, for a general system, the probability of any observation is constructed from all the possible histories that could have led to that observation. Because of that his method is called the “sum over histories” or “alternative histories” formulation of quantum physics.
Now that we have a feeling for Feynman’s approach to quantum physics, it is time to examine another key quantum principle that we will use later—the principle that observing a system must alter its course. Can’t we, as we do when our supervisor has a spot of mustard on her chin, discreetly watch but not interfere? No. According to quantum physics, you cannot “just” observe something. That is, quantum physics recognizes that to make an observation, you must interact with the object you are observing. For instance, to see an object in the traditional sense, we shine a light on it. Shining a light on a pumpkin will of course have little effect on it. But shining even a dim light on a tiny quantum particle—that is, shooting photons at it—does have an appreciable effect, and experiments show that it changes the results of an experiment in just the way that quantum physics describes.
Suppose that, as before, we send a stream of particles toward the barrier in the double-slit experiment and collect data on the first million particles to get through. When we plot the number of particles landing at various detection points the data will form the interference pattern pictured, and when we add the phases associated with all the possible paths from a particle’s starting point A to its detection point B, we will find that the probability we calculate of landing at various points agrees with that data.
Now suppose we repeat the experiment, this time shining lights on the slits so that we know an intermediate point, C, through which the particle passed. (C is the position of either one of the slits or the other.) This is called “which-path” information because it tells us whether each particle went from A to slit 1 to B, or from A to slit 2 to B. Since we now know through which slit each particle passed, the paths in our sum for that particle will now include only paths that travel through slit 1, or only paths that travel through slit 2. It will never include both the paths that go through slit 1 and the paths that pass through slit 2. Because Feynman explained the interference pattern by saying that paths that go through one slit interfere with paths that go through the other, if you turn on a light to determine which slit the particles pass through, thereby eliminating the other option, you will make the interference pattern disappear. And indeed, when the experiment is performed, turning on a light changes the results from the interference pattern, to a pattern like that! Moreover, we can vary the experiment by employing very faint light so that not all of the particles interact with the light. In that case we are able to obtain which-path information for only some subset of the particles. If we then divide the data on particle arrivals according to whether or not we obtained which-path information, we find that data pertaining to the subset for which we have no which-path information will form an interference pattern, and the subset of data pertaining to the particles for which we do have which-path information will not show interference.
This idea has important implications for our concept of “the past.” In Newtonian theory, the past is assumed to exist as a definite series of events. If you see that vase you bought in Italy last year lying smashed on the floor and your toddler standing over it looking sheepish, you can trace backward the events that led to the mishap: the little fingers letting go, the vase falling and exploding into a thousand pieces as it hits. In fact, given complete data about the present, Newton’s laws allow one to calculate a complete picture of the past. This is consistent with our intuitive understanding that, whether painful or joyful, the world has a definite past. There may have been no one watching, but the past exists as surely as if you had taken a series of snapshots of it. But a quantum buckyball cannot be said to have taken a definite path from source to screen. We might pin down a buckyball’s location by observing it, but in between our observations, it takes all paths. Quantum physics tells us that no matter how thorough our observation of the present, the (unobserved) past, like the future, is indefinite and exists only as a spectrum of possibilities. The universe, according to quantum physics, has no single past, or history.
The fact that the past takes no definite form means that observations you make on a system in the present affect its past. That is underlined rather dramatically by a type of experiment thought up by physicist John Wheeler, called a delayed-choice experiment. Schematically, a delayed-choice experiment is like the double-slit experiment we just described, in which you have the option of observing the path that the particle takes, except in the delayed-choice experiment you postpone your decision about whether or not to observe the path until just before the particle hits the detection screen.
Delayed-choice experiments result in data identical to those we get when we choose to observe (or not observe) the which-path information by watching the slits themselves. But in this case the path each particle takes—that is, its past—is determined long after it passed through the slits and presumably had to “decide” whether to travel through just one slit, which does not produce interference, or both slits, which does.
Wheeler even considered a cosmic version of the experiment, in which the particles involved are photons emitted by powerful quasars billions of light-years away. Such light could be split into two paths and refocused toward earth by the gravitational lensing of an intervening galaxy. Though the experiment is beyond the reach of current technology, if we could collect enough photons from this light, they ought to form an interference pattern. Yet if we place a device to measure which-path information shortly before detection, that pattern should disappear. The choice whether to take one or both paths in this case would have been made billions of years ago, before the earth or perhaps even our sun was formed, and yet with our observation in the laboratory we will be affecting that choice.
In this chapter we have illustrated quantum physics employing the double-slit experiment. In what follows we will apply Feynman’s formulation of quantum mechanics to the universe as a whole. We will see that, like a particle, the universe doesn’t have just a single history, but every possible history, each with its own probability; and our observations of its current state affect its past and determine the different histories of the universe, just as the observations of the particles in the double-slit experiment affect the particles’ past. That analysis will show how the laws of nature in our universe arose from the big bang. But before we examine how the laws arose, we’ll talk a little bit about what those laws are, and some of the mysteries that they provoke.