Doing Science: Observation & Inference

SUMMARY:

When we encounter something unknown, we try to figure out what it is, how it has come to be that way, and why. Scientists do this by observing nature, a process often called empirical analysis. In many cases (and especially in evolutionary biology), we can only infer an explanation, because the causes we are interested in cannot be observed directly.

The scientific method is just common sense, consistently applied. It begins with observations, which are used to formulate a testable hypothesis. The hypothesis is then used to formulate a prediction, which is then tested, either via further observations or via a controlled experiment. The results are then compared with the original hypothesis, which may continue to be applied, modified, or even rejected. A theory is a hypothesis that has been repeatedly tested and has not yet been shown to be invalid. All scientific theories are therefore open to revision, and none of them are “true” in an absolute sense.

EVOLUTIONARY PSYCHOLOGY 1.1.2

I began the previous chapter by observing that people are curious about people. While this is undoubtedly the case, it isn't the whole story. We are also extremely curious about the world around us. We (and our primate relatives) are notable among animals for our almost unlimited curiosity. This is especially the case with children, whose curiosity (like a cat's) can sometimes get them into serious trouble. No matter - if we survive a close encounter with something curious, the experience usually pays off. As the German philosopher Friedrich Nietzsche said, “what doesn't kill us makes us stronger.”

Once again, evolutionary psychology is the scientific study of human behavior from an evolutionary perspective. In other words, evolutionary psychologists (just like the rest of us) are curious about what we do and why we do it. As we will see, this curiosity is both useful and dangerous. Useful, because the more we learn about what we do and why we do it, the better able we are to understand our behavior and, if we are lucky, predict it as well.

And dangerous, because a scientific understanding of what we do and why we do it undermines some of the most deeply held and passionately defended preconceptions about who we are and why we do what we do. Evolutionary psychology forces us to answer, as carefully and dispassionately as we can, the most fundamental questions every thinking person has ever struggled with:

Who are we?

Where have we come from?

Why are we here?

Why do we do what we do?

and

How do we know?

The answers that evolutionary psychology gives to these questions may not be what all of us want to hear. But, they have a quality that most of the answers to these questions lack: they can be discovered by simply observing the world around us, and observing ourselves. To see why, let's begin with the last question first.

HOW DO WE KNOW?

There is a question we need to get out of the way right in the beginning - that is, are the answers that evolutionary psychology provides to those questions "true"? To answer that question, consider the following scenario:

It's a Monday morning, and you're running a little late. You rush around doing all of those last-minute things - taking a shower, brushing your teeth, getting dressed, finding your car keys, checking to see if you have everything you need for the day, etc. Then, you dash outside, jump in your car, and head for the freeway…but you don't get very far, because up ahead, the road is blocked off by a cordon of fire trucks and fire fighters, all surrounding what looks like the wreckage of a house fire. [Cue sound effects: distant sirens, people talking in hushed voices, a radio popping and growling, maybe a dog barking, and (for atmosphere) a drizzly rain pattering on the road…or maybe it's just water from the hose that guy is spraying around]

If you're like almost any human being, you check your headlong rush to elsewhere and rubberneck a little. You drive by slowly as the fire police wave you through, checking out the damage: smelling smoke, hearing the hiss of steam and the growl of the radios, and looking at all of the wreckage.

And then, if you're curious, you begin trying to figure out what happened. If you think about it, there are at least three logically consistent answers to the question, "What happened here"?

There has been a house fire, which started accidentally.

There has been a house fire, which was set on purpose.

Someone (perhaps a creative, experienced and well-funded movie director and crew) has just staged what looks like a house fire.

How can you tell the difference? Consider:

You did not witness the event (remember, you came on the scene described earlier, after the incident happened).

All that you have available to you is what you can see right now (and hear and touch and so forth).

To make this more interesting, let's imagine that relatively few of the people standing around saw the actual event either. And, even if they did, you would have to take their word for what they saw (the same would be true for a written account of what happened). Furthermore, the occupants of the house (assuming there were some) aren't around, so they can't tell you what happened either.

Given the foregoing, what can you conclude vis-à-vis the three proposed explanations of what happened? Consider the following questions:

Can you be absolutely certain that this situation happened by accident?

Can you be absolutely certain that this situation happened on purpose?

Can you be absolutely certain that this situation has been staged?

If you answer these questions with the viewpoint of a scientist, the answer to all three of these questions must be NO. All you have to go on is what you can observe, and what you observe is compatible with any one of the three explanations proposed.

So, how do you decide which to believe?

The answer, if you are a scientist, is that you provisionally accept the simplest explanation that best fits all that you have observed and experienced in this event and events like it that you have experienced in the past. If you are lucky, you may never have experienced a house fire yourself. However, you almost certainly have seen the aftermath of one, either directly or in the form of photographs, movies, videos, etc. And so:

You make your best guess, based on the information available and what you know from past experience.

LOGICAL INFERENCE

What you are doing when you make a guess like this is inferring that an event that you have not actually observed has, in fact, taken place. This is precisely what the theory of evolution does, and when you apply the theory to the natural world, you are using essentially the same reasoning that you would use to decide whether a house fire had occurred before you left your house.

Inference is the basis for all reasoning, including scientific reasoning. It's what we will use throughout this series to try to answer the "big questions" we posed at the beginning of this chapter.

So, how do evolutionary psychologists use logical inference to figure out what we do and why we do it? We use “common sense,” systematically applied: that is, we use what is often called “the scientific method”.

The Scientific Method

Evolutionary psychology, like biology, chemistry and physics, is ultimately based on observations:

Evolutionary psychology is an empirical science: it is based on observations of the natural world.

There are non-empirical sciences, such as mathematics and symbolic logic, which are formulated on the basis of abstract principles, usually without direct observation of the real world. However, all of the empirical sciences, including evolutionary psychology, depend fundamentally on observations of objects and processes in nature.

Theories in the empirical sciences, including evolutionary psychology, are developed using a logical procedure often called the
scientific method. Many people think that the scientific method is similar to magic, or is so difficult to understand and apply that only highly trained scientists can use it. Nothing could be further from the truth:

The scientific method is just "common sense" consistently applied.

In general, the scientific method consists of six or seven steps, always beginning with observations of the real, natural world:

Step One: You observe the world around you, focusing your attention on a particular object or process you find interesting:

For example, if you encounter a green apple growing on an apple tree and you have never eaten an apple before, you might be curious to see what it tastes like. So you try it, and the apple tastes sour.

What can you conclude about green apples at this point? You might be tempted to say (as many of my students often do), “green apples are sour.” However, recall that this is the very first apple you have ever tasted. Can you, on the basis of a single experience, conclude that “all green apples are sour?”

No; in fact, on the basis of this single experience:

The only warranted conclusion (that is, the only conclusion that you can rely on now and in the future) is that “this particular green apple is sour.”

Don't get me wrong; this isn't absolutely nothing in the way of evidence. At least you know something about this particular apple. However, given your experience, that's all you know. At this point, to extend your observation about this green apple to all green apples would be unwarranted.

This is what is known in the empirical sciences as anecdotal evidence:

Anecdotal evidence is based on single observations.

That is, anecdotal evidence, while it may be based on observation, is not logically connected to other, similar observations. Therefore, it cannot and should not be used as the basis for generalizations about nature, at least not in the empirical sciences, where you want to be reasonably certain that your generalizations can help you reliably predict what will happen in the future.

So, if you want to learn something about green apples as a class of objects, what do you do next?

Step Two: You repeat the experience, attempting to see if a similar event happens as a result.

For example, if after tasting your first green apple you come upon another one, you might be tempted to taste it as well. So (being curious) you do, and the next green apple tastes sour too.

Can you come to a conclusion about all green apples yet? Well, you might, but to do so would not yet be warranted. Why? Because all you know about so far is that these two similar events seem to be logically connected. Well, how are they connected? Simple: they are similar.

Transductive Reasoning

Finding similarities between separate events is sometimes referred to as transductive reasoning (or “transduction”). In other words,

Transductive reasoning is formulating a generalization based on an analogy between two apparently similar objects or events.

Another way of stating this is that:

Transductive reasoning is arguing by analogy.

We all use transductive reasoning and transductive arguments all the time. Children in particular use transductive reasoning almost exclusively, as Jean Piaget pointed out in his books on child development and learning. As young children we gradually build a "map" of the reality that we experience, using transductive reasoning to relate the objects and processes that we experience to each other.

However, from a scientific standpoint, the logical validity of arguments by analogy is extremely weak. This is because such arguments are only as valid as the analogy upon which they are based, and there is no way to establish such validity within the limits of the argument so far. In our example, we may validly state that “this green apple is sour, like that green apple,” but to extend that analogy to all green apples would not be warranted…not yet, anyway.

So, how can we strengthen the validity of a generalization based on a single analogy? By repeating it:

Step Three: You ask yourself a more general question about what you have observed:

You ask the question, "Are all green apples sour?" And then you taste another one, and it's sour too. And then you taste another one, and it's sour as well. And then you taste another and another and another, and they're all sour.

Now you're onto something. Rather than having information about a single experience, logically unconnected to anything else, or a single analogy of doubtful validity, you now have information about a logical class of experiences, all connected by similar outcomes. What do you do next?

Step Four: You formulate a hypothesis: that is, you formulate a tentative generalization (a “guess,” really) about a class of objects or events, based on the pattern of events that you have observed so far. Based on the foregoing observations, a reasonable hypothesis about green apples would be:

"Green apples are sour"

Inductive Reasoning

There is a term used to describe the kind of reasoning that scientists use to formulate hypotheses: it is called inductive reasoning (or "induction").

Inductive reasoning is formulating a generalization based on a series of individual cases.

Another way of stating this is that:

Inductive reasoning is arguing from the particular to the general.

Inductive reasoning is how virtually all human knowledge and understanding begins. Whenever we encounter a new phenomenon, we try to logically connect it with similar phenomena that we have already experienced. As we do so, we construct larger and larger connected sets of observations about reality (what some educational psychologists call concept maps).

Notice two things about any conclusions you might formulate using inductive reasoning:

The validity of any generalization is only as good as the number of similar observations that have been used to formulate that generalization.

and

Regardless of how many observations may have been made, you cannot be absolutely certain that the generalization that you have formulated is universally applicable.

This is because you can only observe a small subset of all possible cases of whatever it is you are interested in. After all, no matter how many apples you bite, you might not (yet) have tasted a Granny Smith (an apple that tastes sweet when it is still green!)

Prediction

Up to this point, your reasoning processes have not really been any different than what everyone does all the time. But remember, one of the primary goals of the empirical sciences is to be able to predict the future. So what do you do next? You make a prediction:

Step Five: You formulate a prediction: that is, you formulate a guess about what will happen if you perform another observation, given your generalization:

Your generalization (i.e. hypothesis) is that "green apples are sour." Therefore, a prediction that follows logically from this generalization is "this is a green apple; therefore, it is sour."

Deductive Reasoning

As you might expect, there is also a term used to describe the kind of reasoning that scientists use to formulate predictions based on hypotheses: it is called deductive reasoning (or "deduction").

Deductive reasoning is formulating a prediction about a specific case based on a generalization.

Another way of stating this is that:

Deductive reasoning is arguing from the general to the particular.

Deductive reasoning is how many of us make judgments about the objects and processes we see around us. Aristotle, in his work on logic collectively called The Organon, taught that all logic is based on deduction, and showed how logical deductions could be formulated and applied to many situations.

However, there are two caveats about any predictions you might formulate using deductive reasoning:

The validity of your prediction is only as valid as your generalization.

If your generalization is based on a small number of individual observations (especially only one), then it is unlikely to be very useful in making predictions that will be supported by further observation.

Regardless of how you have formulated your generalization (i.e. your hypothesis), if you do not then test it as rigorously as you can, you haven't really done any science.

In fact, you haven't really done anything useful...yet. Even the ancient Greeks used deductive reasoning; what makes the modern scientific method different is what you do next.

Step Six: You test your prediction; that is, make further observations that could either confirm or deny the validity of your hypothesis. There are two somewhat different ways to do this:

Make some more observations, similar to the ones that led you to formulate your hypothesis in the first place; this is sometimes called discovery science.

In the case of your green apple hypothesis, this would consist of simply tasting another green apple (or several more - as many as you can stomach).

Perform an experiment.

This means performing two kinds of observations: an experimental test, where you manipulate the variable that you are testing, and a control test, where you do not manipulate the same variable.

In the case of our green apples, there is no experimental or control test. However, in many scientific tests of hypotheses, control tests are used to determine if the variable being manipulated actually affects the outcome.

For example, you might have noticed that the grass along the road is often brown and stunted in the early spring. You suspect that this might be due to salt used on the roads during the winter (i.e. your hypothesis is, "Salt causes grass to become brown and stunted"). So, you design an experiment that tests this hypothesis: you grow some grass, and then apply various solutions of salt and water to see if they affect the color and growth of the grass. Your control test would therefore be water without salt, and your experimental tests would be solutions containing increasing concentrations of salt. If the result of your experiment looked like this:

you would indeed be warranted in concluding that salt does cause grass to become brown and stunted, and since salt is applied to the roads during the winter, you would also be warranted in inferring that road salt was the most likely cause of the brown and stunted grass you observed along the road in the spring.

Inductive Reasoning (Again)

Note that whichever way you test your prediction (by further observations or by experiment), you are once again using inductive reasoning. This means that all of the conditions listed above still apply:

The validity of your conclusions is only as good as the number of similar observations that you have used to test your hypothesis

and

Regardless of how many observations you may have made, you cannot be absolutely certain that the hypothesis that you have tested and confirmed is universally applicable.

Acceptance or Rejection of Hypothesis

We're almost there; just one more step (or two):

Step Seven: You compare your test results with the prediction that you made using your hypothesis.

If the results are pretty close to the ones predicted, then you have validated your hypothesis.

However,

If the results are significantly different from the ones you predicted, you have falsified your hypothesis.

In this case, you must take one more (and in many ways the most important) step.

Step Eight: You modify (or completely reformulate) your hypothesis and repeat all of the steps.

So, you try one more green apple, and this time you discover it's sweet (it's a Granny Smith)!

What do you do? You modify your hypothesis: "Most green apples are sour, except for Granny Smith apples."

And then you keep on testing...

Theories, Laws, and Truth in Science

After a hypothesis has been rigorously tested, how do scientists refer to it? The call it a theory:

A theory is a hypothesis that has been repeatedly tested and has not yet been shown to be invalid.

That is, all of it's predictions have been observed to conform to reality out so far).

Notice that many non-scientists (including many journalists and science writers, and virtually all creationists) use the word "theory" to mean what a scientist means when s/he uses the word "hypothesis;" that is, a tentative guess about the way the world works, which has not yet been thoroughly tested.

When a scientist uses the word "theory," s/he is generally referring to what a non-scientist would call a scientific law. This difference in usage flows from the tendency of scientists to consider that virtually no scientific principle is ever absolutely and completely confirmed. It's only as good as the experiments that have been done so far to test it. This means that what scientists refer to as "theories" generally have a great deal of evidence backing them up, more than all other alternative explanations. In other words:

Scientific theories (such as the theory of evolution) are what most non-scientists refer to as "scientific laws".

One of the most important implications of the foregoing is that nothing is really "true" in science, using the commonly accepted definition of "truth" That is, no scientific theory (i.e. “scientific law”) is always and absolutely "true":

All scientific theories are always open to revision.

Even a cursory look at the history of science indicates that theories that were once considered "true" are now either highly modified or have been thrown out altogether.

So, what is science?

Science is our best guess today at how the universe works based on the evidence we have observed so far, unless and until we find out otherwise.

Essential Reading:

Popper, Sir Karl (1959) The Logic of Scientific Discovery.

Supplemental Reading:

Aristotle. The Organon. Available online here.

Nietzsche, F. W. (1889) The Twilight of the Gods (Die Götzen-Dämmerung). Available online here.

Questions to Consider:

1. How much of what we know about the world around us do we know directly (i.e. via direct observation) and how much do we know via logical inference?

2. What is the scientific definition of “truth” and is there any other kind?

************************************************

As always, comments, criticisms, and suggestions are warmly welcomed!

--Allen