Problems of Empiricism: First Essay Wylie Breckenridge Explain the difference between reductionism and instrumentalism about theoretical

entities. What do these views agree about? What do they disagree about? Why is the theory/observation distinction important for anti-realists about scientific theories? Does it make a difference what kind of anti-realist you are; i.e., does the severity of the problem depend upon whether you are a reductionist or an instrumentalist? Scientific theories are a problem for empiricists. They are arguably our most successful and reliable sources of knowledge, and yet they don't seem to satisfy the empiricist requirement that all of our knowledge be about, and justified through, experience. Antirealism is an attempt to solve this problem. Its proponents claim that science, despite how it seems, really is grounded in experience and therefore is an empirically valid source of knowledge. For the empiricist, the only things that we have direct access to are our sensual experiences. They are all that we can talk meaningfully about, and they are all that we can know. The empiricist thinks of this experience as being presented on an 'internal screen', and if there is a world external to us then it can only be known as we 'observe' it on this screen. Empiricists differ in opinion as to what the basic components of this experience are. Phenomenalists claim that they are individual sense-data - patches of colour, snippets of sound, etc. Medium-sized object empiricists claim that they are bundles of sense-data - the ones we call chairs, tennis balls, etc. Each of these empiricists has his own language of experience - the language from which all meaningful and knowable statements are composed. For the phenomenalist this language is the language of sense-data. For the medium-sized objectist it is the language of medium-sized objects. Many scientific statements are not expressed in either of these languages. The statement "all rocks sink in water" is expressed in the language of medium-sized objects, but not in the language of individual sense-data. But the statement "electrons have a mass of 9.11 x 10-31 kg" is expressed in neither. For the empiricist, this statement is meaningless - it is not about anything that we can observe on our internal screens - or if it does have any meaning it is not a claim whose truth or falsity we can verify because, again, we cannot observe what it claims. Thus the empiricist is obliged to reject this as an empty claim, and much of science along with it. In response to this, the anti-realist argues that scientific statements should not be taken at face value, and that when correctly viewed they do indeed meet empirical requirements. There are two principal versions of antirealism: reductionism and instrumentalism. I will explain each position in turn. In what follows I will talk about two distinct subsets of any scientific language. The first, called the observation language, contains all those statements that are expressed in the language of experience as agreed upon by empiricists. For the phenomenalist, this will be the language of individual sense data. For the medium-sized object empiricist, it will be the language of medium-sized objects. The rest of the scientific language I will call the theoretical language. It is theoretical in the sense that it contains words and phrases that are only meaningful in the context of the particular scientific theory. The


It is important for the empiricist to make a distinction between these two languages. consider the following theory: All rocks contain sinkitons Anything that contains sinkitons sinks in water Even for a medium-sized objectist these statements are theoretical . because on their view it is only the statements that are expressed in the observation language that make meaningful and verifiable claims. is that a scientific theory can be completely expressed in the observation language. if so desired. then W sinks in water is as follows: 1.they are just rules for deducing observation statements from observation statements. on the other hand. taken together. I will refer to statements in the observation language as observation statements. It is the theoretical statements that pose the problem.they are not expressed in the language of experience. they are factually equivalent to this single observation statement: All rocks sink in water (In this case they are not really shorthand at all . The reductionist claims that they are just a shorthand way of expressing a factually equivalent class of observation statements and that they can. or else she will not be clear about what the theory is actually claiming. independently of the theory in which it appears. is supposed to be meaningful as it stands. It is through this class of observation statements that the theoretical statement gets its meaning.the equivalent class of observation statements is smaller. without losing any of its content. For example. According to the empiricist they are meaningless and/or unverifiable. We can see that. W is a rock 2. All rocks contain sinkitons 3. and it is through these observation statements that its truth or falsity can be verified. and to statements in the theoretical language as theoretical statements. but many predictions about what observational consequences follow from what observational premises seem to rely essentially upon them. Consider the above theory again. in effect. He claims that they are not the kinds of things that can be true or false . Anything that contains sinkitons sinks in water premise premise premise -2- .) The instrumentalist claims that theoretical statements are not really statements at all.not only is much of the theory expressed in terms of them. One way that I can use this theory to predict that if W is a rock. But they seem to be essential to it . What the reductionist is claiming. The reductionist claims that that these two statements are just a shorthand way of writing an equivalent class of observation statements. But this would probably not have been true for a more realistic example. then she must give us an account of their use of theoretical statements. and so add no empirical content to the theory. be translated into the latter. and so any requirement of meaning or factual content is misguided. Thus it is important for the empiricist to specify exactly what this language is.observation language. because of the reference to (as yet) unobservable 'sinkitons'. So if the empiricist wants to keep scientific theories as a genuine source of knowledge.

The conclusion will be true as long as the theory is true. that in keeping with empiricist requirements all the factual claims of a theory are really about matters of experience and nothing beyond. Both can say. "W contains sinkitons". nor is it a rule of deduction. and the instrumentalist medium-sized objectist. modus ponens Here I have used the theory as premises and derived the conclusion by two applications of modus ponens. W contains sinkitons 3. with the former corresponding more to a set of instructions for using the hammer. if there is indeed no such language. but is a kind of place-holder that aids the deductive process. then the thesis that all theoretical -3- . W sinks in water premise 1. the "language" of sense data is not an autonomous language. The instrumentalist claims that this is the wrong way to use the theory. However. Hammers are merely a tool for turning raw materials into finished products. The reductionist phenomenalist claims that every statement of a theory can be translated into equivalent statements about sense-data. It is not a pair of factual claims that can be used as premises in a deduction . not materials or products themselves and hence only of instrumental value to the carpenter. all rocks contain sinkitons 2.the former by claiming that they are meaningful and verifiable once they have been translated into equivalent observation statements. W is a rock 2. modus ponens 3. even in principle. And no one has yet succeeded in constructing such a language. the reductionist medium-sized objectist. He would present the argument like this: 1. W sinks in water 1.4. Nagel discusses this problem and concludes as follows: In short. one only has to attempt this translation for a statement like "there is an apple on the table beside me". anything that contains sinkitons sinks Note that this deduction contains an intermediate theoretical statement. and even then it is a difficult process. The instrumentalist claims that this is not a factual claim. I think it's more accurate to divide theoretical statements into those that the instrumentalist thinks of as deductive rules ('All rocks contain sinkitons') and those that he thinks of as place-holders ('W contains sinkitons'). therefore. This is how reductionists and instrumentalists account for the use of theoretical statements .the reductionist phenomenalist. the instrumentalist phenomenalist. An analogy sometimes used to explain instrumentalism is that theoretical statements are to scientific theories as hammers are to carpentry.4. Then in the carpentry analogy it is the latter that correspond to the hammer. I will consider one problem that is faced by each of the four types of empiricist that I have mentioned . But it is not obvious that this is possible. W contains sinkitons 5.2. To see how difficult it is. To see why. But they run into trouble elsewhere. It takes a great deal of practice to be able to isolate individual patches of light and colour from the overall scene. the latter by claiming that they are not supposed to have meanings or to be factual claims at all.they are rules of deduction in themselves.

Any survey of the statements of a scientific theory will make it clear that very few. not only is the class of observation statements vast but it is also impossible to specify completely at any point of time. it says that we can only remove the theoretical statements from the theory if we know all of their observational consequences in advance. within the theory of electromagnetism.1 The reductionist medium-sized objectist does not face this problem. Worse still. Thus the set of equivalent observation statements will be infinitely numerous. because we haven't yet observed them and we 1 Nagel. Sinkitons are not. the class of observation statements must contain all of the observational consequences of the original theoretical statement. But consider a statement like "there is a net electric charge on this plastic ruler". In effect it is saying that the reductionist program is achievable in the case of a formal language. and then all of these will reappear as axioms. p. he does not require that any translation of statements into the language of sense-data should take place . then. -4- . but he faces another (and one that also faces the phenomenalist). If we apply this to a scientific theory. are partitioned into two types (think of them as the observation statements and the theoretical statements) then we can replace it with a new formal system whose axioms are made up just of observation statements and such that both systems have the same set of observation statements as deductive consequences of the axioms. Apples are a part of our medium-sized object language because they are a familiar feature of our daily experience. To ensure that no content is lost. and that it contains no factual premises to deduce any factual consequences from (remember. are expressed in the language of sense-data. 122. without loss of content. On the other hand. The instrumentalist phenomenalist. theories do contain statements that are expressed in the language of medium-sized objects. and so the instrumentalist medium-sized objectist is not faced with an account of the theory that renders it empirically empty. does not meet the requirements of a formal language. into a class of observation statements (whatever the observation language may be).he just takes the statements as they are). including its axioms. It is claimed that a given theoretical statement can be translated. then.statements are in principle translatable into the language of pure sense contents is questionable from the outset. A similar problem is suggested by Craig's theorem. In a nutshell it says that if the sentences of a deductive system.that the theory is just a set of rules and tools of deduction. then. Moreover. and for every position in space around the ruler. This is a result from formal logic that initially appears to lend support to the reductionist thesis. we may discover in the future that a type of object previously thought to be unaffected by electric charge is actually affected by it. include things like "if such-and-such an object is placed in suchand-such a place then so-and-so will happen". the procedure involved in forming the new deductive system only works if all of the deductive consequences of the old system are known in advance. This is clearly not a desirable position. is forced to claim that the theory has no factual content at all . Consider the theory about sinkitons presented above. if any. and these consequences are all presented as axioms in the new system. however. That is. But he faces another potential dilemma. There will be a statement like this for every object that is effected by an electric charge. and so we must include in our translation statements about the behaviour of these objects as well. The consequences of this. Scientific language. and so the relevance of Craig's theorem to the reductionist thesis is questionable.

What I am experiencing is not an external world. Neither option is very appealing.either fix the observation language once and for all. But in making this claim I am not committed to believing that it is will it agrees with what I observe. I don't think of myself as really just talking about sense-data or other bundles of experience.the criterion by which we are supposed to judge which statements are factual claims and which are not . and continually reassess the factual content of the theory. then these theories are empirically inadequate and therefore questionable. nor do I think of myself as really just stipulating a rule that can be used to deduce matters of observation. And that might be enough to give empirical justification to my use of the theory. in the course reader. In short. The change comes about because the observation language . When I make a statement like "All rocks contain sinkitons". van Fraassen. but only the final products of an elaborate construction involving both the sense-data available to me and my whole system of beliefs. but a good deal of it comes from my other beliefs.that there really are things called sinkitons with the properties described.experience is not something that precedes knowledge. Maxwell. just as well as we can observe apples. Then sinkitons would become a part of the medium-sized object language.has changed. According to the empiricist. I am supposed to take this 'given' data and from it construct all of my beliefs and knowledge. and ignore any changes in what we can observe. "Arguments Concerning Scientific Realism". But it seems to me that often I construct a theory first and then check it against the evidence available to me. and the instrumentalist is obliged to say that "All rocks contain sinkitons" is now a factual statement and not a rule of deduction. These particular speculations may be unreasonable. "The Ontological Status of Theoretical Entities". I am not even committed to being interested in its truth or falsity at all. but what is happening 'in here'. I suspect that the empiricist has got things around the wrong way . or allow the observation language to evolve. The instrumentalist is thus obliged to say that the statement "All rocks contain sinkitons" is not a statement at all. The instrumentalist has a choice .haven't any idea what they might look like if we did. I suspect that reflecting on the way that I actually think when I'm using a scientific theory will not only show the inadequacy of the anti-realist account of science. but I think the suggestion that we question the empiricist starting point is not. G. It seems to me that both the reductionists and instrumentalists are doomed to face problems because the account they give of scientific language does not agree with the way that we typically use it. That assumption is that 'observation' is the passive process of 'viewing' things as they are presented on my 'internal screen'. I think of myself as making a claim about the way things are . In fact. Suppose something changes so that we can observe sinkitons. If antirealist theories about scientific language do not agree with what we actually observe in the behaviour of practising scientists. What we are viewing on our internal screens is not what is happening 'out there'. Bibliography B. Typically I am more interested in its empirical adequacy . but my internal world projected outwards. but also the inadequacy of the basic assumption of empiricism. but a rule of deduction within the theory. I suspect that I never observe my sensual experience at all. In fact. in the course reader. Part of this evidence comes from my senses. -5- . but the way that we present to ourselves the consequences of our beliefs.

Chapter 6.E. The Structure of Science. Nagel. -6- .