This version of naturalism is one that I, and many others, oppose. This opposition will probably come as no surprise; Quine seems to be claiming that philosophers interested in questions about belief and knowledge should close up shop and go home. Just as scientists warm to the heroic description of scientific work given by Popper, many philosophers turn a cold shoulder to Quine's claim that there is nothing important for us to do (unless we get psychology degrees and move to the psychology department). But aside from wanting to keep the paychecks coming, there is a deeper issue here.
In a different version of naturalism, there is such a thing as a philosophical question, distinct from the kinds of questions asked by scientists. A naturalist can think that science can contribute to the answers to philosophical questions, without thinking that science should replace philosophical questions with scientific questions. That is the version of naturalism that I defend. This contrasts with the kind of naturalism described by Quine in his 1969 paper; there we think of science as the only proper source of questions as well as the source of answers.
If we think that philosophical questions are important and also tend to differ from those asked by scientists, there is no reason to expect a replacement of epistemology by psychology and other sciences. Science is used as a resource for philosophy, not as a replacement.
What might be examples of these questions that remain relevant in naturalistic philosophy but which are not directly addressed within science itself? Many naturalists have argued that normative questions are important examples here-questions that involve a value judgment. If epistemology was absorbed by psychology, we might get a good description of how beliefs are actually formed, but apparently we would not be told which beliefforming mechanisms are good and which are bad. We would not be able to address the epistemological questions that have to do with how we should handle evidence, and how we can tell a good argument from a bad one. Those questions are central to philosophy. For the naturalist, the answers to these questions will often depend on facts about psychological mechanisms and the connections that exist between our minds and the world. But the naturalist expects that it will remain the task of philosophy to actually try to answer these questions. The sciences tend to concern themselves with different issues.
The term "normative naturalism" is often used for naturalistic views that want to retain the normative side of epistemology. (The term was coined by Larry Laudan [19871; see also Kircher 199z). I should also note here that although Quine's original discussions seemed to leave no place for normative questions in epistemology, toward the end of his career he modified his view, bringing it closer to normative naturalism (Quine 199o).
Normative naturalism accepts many (though not all) of the normative questions that have been passed down to us from traditional epistemology. But what is the basis for making these value judgments? What is the basis for a distinction between good and bad policies for forming beliefs? At this point normative naturalism confronts one aspect of the old and difficult problem of locating values in the world of facts. In the face of this problem, normative naturalists have often chosen a simple reply. The value judgments relevant to epistemology are made in an instrumental way.
In philosophical discussions of decisionmaking, an action is said to be instrumentally rational if it is a good way of achieving the goal that the agent is pursuing, whatever that goal might be. When assessing actions according to their instrumental rationality, we do not worry about where the goals come from or whether they are appropriate goals. We just ask whether the action is likely to achieve the outcome that the agent desires. And if some action A is being used as a means to B, it is a factual matter whether or not A is likely to lead the agent to B. So it is a factual matter whether or not action A is instrumentally rational for that agent.
It is uncontroversial to say that one kind of rationality is instrumental rationality. It is much more controversial to say that this is the only kind of rationality. Some normative naturalists think that instrumental rationality is the only kind of rationality that is relevant to epistemology. The problem of assessing which of a person's ultimate goals are justifiable is either rejected (because there can be no such assessment) or not addressed.
Earlier I said that John Dewey's work in the 19zos and 193os describes and applies a good version of naturalism. Dewey's handling of the question of epistemological norms and values is one strength of the view. In his 193 8 book Logic, Dewey develops a version of what would now be called normative naturalism. He says that in his epistemology, claims of "good" and "bad" reasoning are intended in the same way that we would understand claims about "good" and "bad" farming (1938,103-4). Everyone is aware, he says, that some farming techniques are better than others at achieving the usual sorts of goals that farmers have. The likely consequences of different farming decisions are a factual matter, and we learn about these consequences from experience. The methods of farming that we presently regard as good ones are not perfect, and they might be improved further in the future. But there is no philosophical problem with making value judgments of this kind. Dewey says that the same is true of value judgments in epistemology.
I have focused on the role of normative questions here, but these are not the only kinds of questions that belong within epistemology, as opposed to the sciences that "feed into" epistemology. Another set of questions asked by philosophers have to do with the relationships between our commonsense or everyday view of the world, on one hand, and the scientific picture of the world, on the other. What kind of match (or mismatch) is there between the two pictures? We find questions of this kind in epistemology: what relationship is there between the commonsense or everyday picture of human knowledge, and a scientific description of our real contact with the world?
Answering this kind of question requires that we summarize both the everyday and the scientific pictures in a concise way and then compare them. One of the fastest-moving and most interesting parts of naturalistic philosophy in recent decades has been the naturalistic philosophy of mind. How does the everyday picture of the human mind and its contents (thoughts, beliefs, desires, memories) compare with the picture of the mind that is emerging from psychology and neurobiology?
Another set of philosophical questions that remain pressing for the naturalist have to do with the relations between different sciences. The various sciences each give us fragments, based on empirical work, of what the world is like and how it runs. But do the fragments tend to fit together neatly, or are there mismatches and tensions between them? The philosopher patrols the relationships between adjacent sciences, occasionally climbing into a helicopter to get a synoptic view of how all the pieces fit together. This can result in philosophical criticism of particular scientific ideas, but the criticism is made from the point of view of our overall scientific picture.
So I can now summarize the version of naturalism that I accept. Naturalism in philosophy requires that we begin our philosophical investigations from the standpoint provided by our best current scientific picture of human beings and their place in the universe. We begin from this picture, and we do not try to give a general justification, from outside of science, for our entitlement to use it. The science we rely on is not completely certain, of course, and may eventually change. The questions we try to answer, however, need not be derived from the sciences; our questions will often be rather traditional philosophical questions about the nature of belief, justification, and knowledge. Science is a resource for settling philosophical questions, rather than a replacement for philosophy or the source of philosophy's agenda.
I should note that I am not arguing that all work done by philosophers should be naturalistic in this sense. In particular, philosophy has long served another unusual and useful role in intellectual culture; it has acted as an "incubator" for novel, speculative ideas, giving them room to develop to a point where they may become scientifically useful. There are other roles for philosophy as a discipline as well. Philosophy often benefits, in fac
t, from its somewhat loose organization and open-ended agenda. We never know what new ideas, issues, or approaches might appear. But to the extent that we can expect to solve the big problems in fields like epistemology, naturalism is probably the right approach.
So what questions should we address in naturalistic philosophy of science? Back in chapter i, I distinguished two sets of questions that would shape this book. We should try to achieve (z) a general understanding of how humans gain knowledge of the world around them, and (z) an understanding of what makes the work descended from the Scientific Revolution different from other kinds of investigation of the world. Those summaries are a start, but I can now fill in a bit more detail.
Does a naturalistic investigation of the role of observation in science support the familiar idea that observation and experiment make science responsive to the real structure of the world? One way to understand the work of some sociologists of science, including Latour, is to see it as proposing a theory of scientific change that gives no role to this notion of "responsiveness" to the world. Where exactly does a view like that go wrong? There could, in principle, be an institution that looked like what we call "science" but in which there was no genuine responsiveness to the world. Experiments would be no more than expensive "PR" exercises, and theories would change via a process of negotiation between factions. How do we know that our own science is not like this? In order to resolve this issue, we need to distinguish some "in principle" questions and some "in practice" questions. Does the nature of human thought and perception allow that, in principle, scientific belief can be made responsive to the real structure of the world? Even if this is possible in principle, do actual scientific communities operate in a way that makes this responsiveness occur in practice?
Suppose we are indeed able to vindicate the idea that science is responsive to the world; then what sort of contact with the world do our successful theories achieve? It is familiar to think of truth as the goal that we set for our theories; a good theory is one that represents the world truly. But is the traditional concept of truth a coherent and useful one here? Does it help us understand scientific progress at all? (Kuhn believed it does not.) What sense, if any, can a naturalist make of the idea of an "inductive logic," or a general theory of evidence? If we are able to isolate features of scientific thinking or scientific community structure that seem powerful and valuable, how can these be safeguarded and strengthened? Are there features of science that are self-defeating or harmful, which we might try to resist or change?
This version of naturalism will guide the remaining chapters of the book. But as I emphasized earlier, there are many different views that people like to label as naturalistic. And even philosophical discussions of science that are not carried out under the banner of naturalism have become more responsive to a variety of sources of ideas. This broadened perspective on the kinds of information that might be helpful has been a notable feature of recent philosophy. Some philosophers think the result has been chaotic, a profusion of ambiguous fragments and half-finished forays in too many directions. Discipline has been lost. But others, including me, think that the result has generally been progress.
10.3 The TheoryLadenness of Observation
In this section I will focus on a debate that developed in the z96os and continues to the present. The debate concerns the role of observation in science, and it is often called the debate about the "theoryladenness of observation." Put most simply, the debate has to do with whether observational evidence can be considered an unbiased or neutral source of information when choosing between theories, or whether observations tend to be "contaminated" by theoretical assumptions in a way that prevents them from having this role. The problem is especially important for people who want to develop empiricist views. Advocates of radical theories of science, of the kind discussed in the last few chapters, have often seen the theoryladenness of observation as a powerful argument against mainstream empiricism.
So the debate is important. The reason it is discussed in this chapter is that this debate becomes much easier to settle if it is approached from a naturalistic point of view. This issue gives us a good illustration of how naturalistic philosophy works in practice.
Our topic is observation. But "observation" is being understood in a very broad way here, to include all kinds of sensory contact with the world, all kinds of perception. Empiricists have agreed that observation is our source of knowledge about the world. Despite a good deal of disagreement within the empiricist movement, observation has generally been seen as theory-neutral. This neutrality, or absence of bias, is often the basis for the claim that observation is an "objective" way to settle disagreements.
It was against this background that arguments for the "theoryladenness" of observation developed, especially in the work of N. R. Hanson, Kuhn, and Feyerabend. These arguments are a mixture, but their intended upshot is clear: observation cannot function as an unbiased way of testing theories (or larger units like paradigms), because observational judgments are affected by the theoretical beliefs of the observer. Therefore, traditional empiricist views about the role of observation in science are false.
As I said, these arguments are a mixture. Sometimes they are about the language of observation reports, sometimes about observation as a psychological phenomenon, sometimes about the beliefs resulting from observation, sometimes all of these. And while some of the phenomena discussed in the arguments are important and challenging, others are not. Some arguments only trouble logical positivism, while others trouble all possible views about science other than radical skepticism or extreme relativism.
Let us start with the more innocuous arguments. Sometimes it is claimed that observation is guided by theory, because theories tell scientists where to look and what to look for. This is true, but no sensible empiricist has ever denied it. This fact does not affect the capacity of observation to act as a test of theory, unless scientists are refusing to look where unfriendly observations might be found. All empiricists would regard that as a breakdown of fundamental scientific procedures.
At other times it is claimed that scientists must use theoretical assumptions to decide which observations to take seriously. Some apparent observations might involve malfunctions or mistakes of various kinds and can be disregarded. The observations that affect theory choice are "filtered" through a process in which some data are discarded. Because theoretical beliefs affect this filtering, there is the possibility of bias here.
Those problems are real. What they involve, however, is the problem of holism about testing, which was introduced back in chapter z. Philosophers are still trying to unravel this problem, as they try to develop new theories of testing and confirmation. In the absence of a general solution to these problems, some pieces of an answer can be suggested. The theoretical assumptions that affect the relevance of an observation to a piece of theory can themselves be tested separately. We might also venture some low-level recommendations: perhaps in crucial tests, scientists should be more reluctant to discard observations. But in this area it is hard to know which pieces of common sense are helpful, which are trivial, and which are flat wrong.
Another set of arguments about observation concern language. When a scientist has an experience, he or she can only make this experience relevant to science by putting it into words. The vocabulary used, and the meanings of even innocent-looking terms, will be influenced by the scientist's theoretical framework. Given the interconnections between the meanings of words in a language, there is no part of language whose application to phenomena is totally "theory-free."
Some versions of this argument are not of enduring importance because they cause trouble only for the logical positivist ideal of a purely observational language, sharply distinct from the parts of language that use or assume theoretical ideas. Sometimes critics of empiricism write as if once it has been shown that the language of observation is in some sense "theoretical," that is the end of the argument and empiricism is dead. This is a m
istake. In working out the relevance of this issue to more modern forms of empiricism, everything depends on which kinds of theories affect the language of observation and on the nature of this effect. For example, maybe observational reports assume "theories" that are so low-level that the testing of real scientific theories will never be affected. We can think of the assumption that objects generally retain their shape when we are not looking at them as "theoretical" in a sense, but the effect of this assumption on observation reports does not usually matter to testing in science.
But suppose it can be shown that observation reports are affected by the kinds of theories that are themselves being tested. For example, Feyerabend tried to show that innocent-looking descriptions of motion in the seventeenth century were affected by theoretical background assumptions in this way. This looks like trouble. But even this kind of effect may or may not be philosophically important. A theory might contribute the concepts used to express an observation, without this affecting the capacity of an observation report to test the theory in question. Not every result described in terms of the concepts preferred by theory T will be an observation report that is favorable to theory T. Back in my discussion of Popper, I mentioned that an observation of rabbit fossils in Precambrian rocks would be a massive shock to evolutionary theory (section 4.6). Suppose we regard "I saw rabbit fossils in Precambrian rocks" as an observation report that is very much "laden" with biological and geological theory. Some might want to say it is so laden with theory that it is not an observation report at all. But regardless of this, the report would still be a massive shock to evolutionary theory.
So imagine that we had a simple falsificationist view of testing in science. It is clear that the fact that observation reports are expressed using concepts derived from a theory has no effect on the capacity of nature to say NO to a conjecture. Simple falsificationism is not an adequate view of testing in science, but that does not matter to the present point. The point is that an influence of theory on observational vocabulary does not, on its own, prevent observation from acting as an unbiased test of theory.
Theory and Reality Page 22