Much of our modern culture revolves around something called "science." Governments want "scientific" analysis of various problems to guide policymaking. News reports detail the latest "scientific" studies about human health. People worry about whether their religion conflicts with "science." But what is science? This turns out to be a complicated and controversial question, and whenever we try to come up with a really precise definition, we end up calling some activities "science" that we would rather exclude, or excluding some activities we would like to include (Laudan, 1996). For example, some people might distinguish science from other activities by noting that scientists perform experiments. However, some sciences are not particularly experimental, e.g., it is hard to imagine astronomers performing experiments on celestial bodies. On the other hand, astronomers do collect and record observations, even if these cannot properly be called "experiments." Is the collection of observations of the natural world the defining feature of science? Apparently it is not, since astrologers have been observing and recording the motions of heavenly bodies for millennia, and most people would not classify astrology as science. Scientists typically go on to explain their observations by creating theories that might be used to predict or control future events. However, astrologers also explain their observations by creating theories, and they certainly try to predict things by their theories (Okasha, 2002, pp. 1-2)! Furthermore, there is a certain breed of physicists, called "string theorists" who have not yet come up with a single prediction, but that does not keep them from being classed with the other scientists in the university physics departments where they work.

Even if a precise definition of "science" is not forthcoming, however, most people would agree that, in general, science does involve collecting observations about the natural world and coming up with explanations for them that might help us predict or control the future. Therefore, we could propose a loose definition of science like the following.

Another point that may trouble you about our definition of science is that we have not yet gotten rid of the astrologers. A prominent philosopher of science put it this way: "The difference between science and other endeavors that seek explanations of why things are the way they are can be found in the sorts of standards that science sets itself for what will count as an explanation, a good explanation, and a better explanation" (Rosenberg, 2000, p. 21). In order to help you understand why things like astrology (or history, or any number of other fields of study that could fit our loose definition) are not considered "science," we must explain the kind of standards scientists set for themselves when developing their stories.

Rules for Scientific Storytelling

Just like any literary genre, scientific storytelling follows certain rules that set it apart from other genres. However, it is important to realize that rules are chosen, not because no others are possible, but for convenience in attempting to accomplish certain goals. Remember that science is the art of creating explanations for natural phenomena that could be useful for controlling or predicting nature. What kinds of rules could be made to make science more useful in this way?

Rule #1: Predictive Power

Our first rule has to with the idea that scientific stories might be useful for predicting the future course of events.

On the other hand, in the 19th century the great British scientist, Lord Kelvin, suggested that the sun might be a glowing ball of liquid, formed as meteorites coalesced by gravitational attraction and generated heat from friction, etc. If this were true, Kelvin reasoned, it ought to be possible to calculate the sun's age, based on estimates of its annual heat loss. He estimated that the sun had been losing heat for a maximum of 100 million years (Thomson, 1862). Further research into the frequencies of light waves emitted by molten meteorites might also have served as a test of the predictive power of Kelvin's story. Now, it turns out that scientists since Kelvin have come up with much better ideas about what the sun is, and how its heat is generated, and these new explanations can account for many more observations than Kelvin's. For example, the light waves emitted by the sun are not characteristic of molten meteorites, and radiometric dating techniques seem to support the idea that life has existed on Earth for much longer than 100 million years. In other words, Kelvin's explanation is now considered to be flatly wrong because its predictions failed. However, it is still considered a "scientific" explanation, because it generated predictions that were not originally used in the creation of the explanation. This kind of prediction allows science to go forward, rather than getting stuck in a rut.¹.

Although some "scientific" explanations do not immediately produce predictions that we can test (remember the "string theorists,) it is easy to see why scientists prefer testable stories. That is, if we allow too many explanations that cannot be tested in any way, then it becomes harder to decide whether to prefer one story over another.

Rule#2: Prospects for Improvement

In order to fully understand why scientists prefer testable predictions, one must first come to the realization that science is not about establishing "the facts," once for all, but about a process of weeding out bad explanations and replacing them with better ones.

1. Scientists make observations about the natural world.
2. Scientists come up with explanations that can explain these observations, or at least the ones that we are most sure about, or seem most important.
3. Other consequences of these explanations are evaluated, and scientists come up with ways to observe whether those predictions are true.
4. Scientists then make these other observations to test their predictions.
5. If the predictions work out, then the original explanation may be kept. If the predictions do not work out, then scientists do one of three things.
   1. They throw out their initial explanation, and try to come up with another one that explains all (or at least most) of their relevant observations.
   2. They slightly modify their original explanation to account for the new observations.
   3. They ignore the new observations that do not fit with their explanation, assuming there must be something wrong with the observations. Then they either go on as if nothing had happened or try to improve the observations.
6. Whether they keep the original explanation, or go with another one, scientists always return at this point to Step #3, and keep repeating steps 3-6 over and over again.

Indeed, we claimed above that scientists are perfectly capable of ignoring some observations that conflict with their established explanations. Why would they do such a thing? The fact is that sometimes observations go wrong-instruments do not work correctly, experiments are contaminated, and people can be deceived in what they think they see. Furthermore, the world is a complicated place, and even if a few observations seem to conflict with an explanation, it may still be mainly correct. And if it is not immediately apparent how to fix the theory, that is no reason to throw out an otherwise perfectly good explanation. However, if observations that do not fit a scientific story keep piling up, rather than being successfully explained away, scientists begin wondering whether they ought to look harder for a better story (Kuhn, 1996).

Consider the example of Galileo. In his time, the geocentric (Earth at the center of the Universe) astronomy that was in fashion at the time was in trouble-a number of observations seemed to conflict with the theory. To overcome some of these problems, Copernicus had proposed that the Sun is at the center of the Universe, and everything else revolves around it in circular orbits. Galileo used a telescope to produce observations that he then advertised as supporting the Copernican theory. For example, he could show that the brightness of the planets changed through the year, which was predicted from the Copernican idea that the Earth should be at different distances from the planets at different times of year. However, the magnitudes of some of these variations were not nearly large enough. Also, many people who looked through Galileo's telescope distrusted it, because although it seemed to work well when pointed at objects on the Earth, optical illusions (such as double-vision) were noted when it was pointed toward the heavens (Feyerabend, 1993, pp. 86-105). Clearly, the Copernican theory had problems of its own, and many of them were not solved for decades, or even centuries, as the Copernican theory was adjusted to accommodate things like elliptical, rather than circular, orbits and better theories of optics were developed. So why did it quickly become the dominant explanation of the motion of heavenly bodies, even in the face of contradictory evidence? Perhaps the answer is that even if it had problems, the adherents of the Copernican theory saw it as a more promising avenue of research than the competition, and so they were willing to try to work out those problems.

The idea that we hope to get across here is that science employs reasonable methods for coming up with good explanations for natural phenomena. But in the end, it is a human enterprise, prone to human limitations.

Rule #3: Naturalism

The kind of human limitations just discussed are not the end of the story, however. It turns out that scientists also deliberately impose certain limitations on their craft for practical reasons.

Looking back to some of the examples already discussed, it is clear that the explanation of the sun that included the Greek god Apollo is ruled out from the start, whereas Kelvin's explanation is not. Whereas the Apollo story involves a supernatural being, Kelvin only appealed to natural causes, such as the gravitational attraction between meteors and heat generation by friction. He said he favored his explanation of the sun's heat because "No other natural explanation... can be conceived" (Thomson, 1862).

This brings us to a rather odd problem. That is, many scientists believe in Judaeo-Christian, Muslim, and other concepts of God and spirituality along with most of the rest of the world. Many of them even believe that "supernatural" events have occurred. And yet, when they practice science, they never introduce the supernatural into their explanations. For example, Lord Kelvin not only believed in a Christian concept of God, but he even used his estimate of the age of the sun to show that there could not possibly have been enough time for life on Earth to have evolved from lower forms, as Charles Darwin suggested. He went on to propose that a relatively young solar system ruled out organic evolution and this, in turn, implied an intelligent Creator. If Kelvin believed that God supernaturally generated life on earth, then why would he feel compelled to stick to "natural" explanations when offering a scientific account of the origin of the sun?

There are three practical reasons for sticking to naturalistic explanations. First, we have already seen that supernatural explanations (like Apollo's chariot) tend not to generate new predictions. Not only does this stop the scientific enterprise in its tracks, it is not very useful. That is, supposing the sun is Apollo's chariot, what can we then do with that information? Science operates by observing regularities in nature, but supernatural beings like Apollo might decide to change the natural order at any moment, and how could we predict when or why that would happen? Second, it is very difficult to place limits on which supernatural explanations are acceptable. For example, if it is acceptable to say that the sun is Apollo's chariot, then why not Odin's shiny helmet? Both of these points can be overstated, however. It might well be possible for supernatural explanations to generate new predictions-even some that could easily be tested - but in order for this to be so we usually must know something in advance about the supernatural agent in question. For example, if we say that God created the world, we can generate predictions about what the world is like only if we know something about what God could have and would have done during the Creation. And this brings us to our third reason for sticking to naturalistic explanations. Since different groups ascribe different attributes to God and other supernatural agents, if we allowed supernatural explanations in science we would end up with various versions of Christian, Muslim, Hindu, Buddhist, and Jewish science, to name but a few. In pluralistic society, and in an age when science is a big-money, publicly funded enterprise, most scientists would prefer that we all just try to come to some sort of compromise, for the moment, and that compromise entails keeping God out of the explanatory picture.

Another example of the usefulness of a naturalistic approach to science is the story of the ancient Greek physician, Hippocrates. In Hippocrates' day, illness was often attributed to the anger of the gods, and that sort of thing. In that case, a physician's job was to invoke the aid of the gods (usually Asclepius, Apollo's son) to heal the sick person. Hippocrates challenged this practice, not because he did not believe in the gods, but because he thought that the physicians of his day were often using the gods as an excuse for their own ignorance of the causes of disease. If, on the other hand, diseases were mostly the result of natural causes, one might often find natural cures (Rubenstein, 2003). This sort of attitude is very prevalent today, even among deeply religious people. That is, when people are seriously ill, they usually check into the hospital, even though they might pray for divine help, as well.

On the other hand, even if the supernatural is not allowed in scientific explanations, individual scientists may still use their religious views or other inner experiences in the creative process. For instance, Albert Einstein frequently used to muse about how "the Old Man" (referring to his impersonal concept of God) would have done things. However, when it came to his published scientific explanations, "the Old Man" never made an appearance. The Belgian scientist Friedrich Kekulé hit upon the idea that the benzene molecule has a ring structure after having a dream where a snake tried to swallow its own tail, but went on to test this idea using scientific methods (Okasha, 2002, p. 79). In the creative process, anything goes, so long as a naturalistic account can be given later.

It should always be remembered that scientists do not allow God and other supernatural agents into their stories only because there are practical reasons for it, rather than because we have to.² Furthermore, just because we can come up with a naturalistic explanation for something, it does not follow that the explanation is true. As discussed above, we can never be sure that we have hit upon the one and only possible explanation for our observations, and we can never be sure that more observations will not contradict our stories.

Once this point is clear, it should be apparent that once in a while, there will be conflicts between science and various religious viewpoints. If we do not allow the supernatural to play any part in scientific explanations, how can we expect them to always be in harmony with religious philosophies that specifically claim there are supernatural influences on the natural order? Occasional conflicts would seem to be inevitable, and therefore such conflicts should not come as a shock to anyone.

Rule #4: Uniformitarianism

Most people will agree that most of the time the world operates in a regular manner, according to some natural laws. Therefore, they have little problem with most science as it is now practiced. On the other hand, some people believe that this has not always been the case in the past. For example, some people believe that God created the world out of nothing in the not-too-distant past, and that other "miracles" have occurred in the past. This poses a problem for the "historical sciences"-those that interpret the present state of things in terms of past events. For example, consider the popular TV series, CSI. In this show, crime scene investigators (forensic scientists) examine the details of a crime scene (blood spatter patterns, angles of bullet holes, articles that seem out of place, injuries evident on a dead body, etc.) and make up stories about how the present situation might have come about. In order to test their stories, they might shoot bullets into Jell-O, try to mimic the production of blood spatters, use trigonometry to determine from where a bullet might have come, and that sort of thing. The assumption implicit in all of these activities is that the crime scene reached its present state via processes that can be mimicked in the laboratory. They do not even consider the possibility that some supernatural entity might have been involved. Why? Because if they admitted such a possibility, all their normal methods for evaluating evidence would go out the window. Furthermore, when the case reaches the courtroom, even juries packed with deeply religious people tend not to listen to pleas by defense attorneys that supernatural entities adjusted crime scenes to make the defendants look guilty. This brings us to our next rule.