mercredi 18 janvier 2017

Modal empiricism as pragmatic realism

So far I explained why modal empiricism is the best position on epistemological grounds:

All this is fine, but epistemology is only one side of the questions over scientific realism. The other side, which is too often neglected in my opinion, is the semantic side, in the broad sense: what exactly is scientific representation about? How does it relate to reality? Shall we interpret the content of scientific theories as descriptions of the world, or as mere tools to make accurate predictions?

Three accounts of meaning

F21. Venous enlargement in hepatic cirrhosis. Alfred Kast Wellcome L0074357
This was the central question in philosophy of science during most of the 20th century. Logical empiricists viewed scientific theories as linguistic statements and attempted to provide an analysis of their meanings.

It is common nowadays, following the semantic view of theories, to think of scientific theories as a collection of models rather than a set of statements, but although models played an important role in the argument of the previous posts, I think the statement view is roughly correct, and roughly equivalent to the semantic view.

The semantic view was a sane reaction to the idealistic conceptions of logical empiricism. Its main import is that it emphasised the role models play in empirical confrontation, as mediators between the theory and experience. But as an attempt to bypass semantic questions, I think it is misguided, because models need to be expressed in a language to have a domain of application, and axioms unify various models in a coherent scheme, so the best way to express the content of a theory is with axioms expressed in a theoretical language.

Now comes the main question of the semantic side of scientific realism: how shall we interpret this theoretical language? There are roughly three options:

structuralism:
theoretical terms are mere placeholders in the conceptual structure of the theory
reductionism:
theoretical terms should be analysed by their intensional or extensional relations to our experience, or their function in experience
essential realism:
theoretical terms directly refer to natural properties and relations, or essences

Note that the question is not about what theories achieve to do, but about what they purports to do: about their meaning, or about what would make them true if they were true. The emphasis on linguistic aspects the logical empiricist entertained is not misguided. After all, meaning is generally analysed in terms of truth conditions: the meaning of a statement is what would make this statement true. There are three traditional theories of truth, which roughly correspond to our three options: coherentism, according to which truth is coherence with a conceptual scheme, pragmatism, according to which truth has to do with the ideal success of a statement, or our capacities to assess this statement, and correspondence, according to which truth is correspondence to reality. In sum, the locus of truth-conditions, and meaning, is either in our representations only, in experience, or in an independent reality. (One could add deflationist theories of truth but I forget them here because they do not really shed light on meaning.)

To many authors, only a correspondence theory can sustain genuine realism: the others would rather lead us to various forms of empiricism, relativism or idealism. Kripke's defence of essential realism has been influential, which is a reason why the focus in philosophy of science moved away from semantic aspects, toward epistemic aspects: given that our theories purports to refer to natural properties, or to correspond to reality, how would we know that they succeed? A correspondence theory of truth introduces a gap between the content of our representations and our capacities to know that they are true, and this is where realists resort to abductive reasoning to fill the gap. The fact that this is problematic is a good reason not to forget about other options, and to keep an eye on philosophy of language. Interestingly, pragmatics became a big subject there, with an emphasis on the role of intentional aspects and context on meaning, and direct reference is not uncontroversial anyway.

This account seems unfair to structuralists, because structural realists would typically accept a correspondence theory of truth, but apply it to the structure directly. In a sense, this is a way to accommodate coherentist aspects (for the vocabulary) with a correspondence truth (for the structure). Now I'm not sure it works: Putnam's model-theoretic argument is a good argument against this kind of view. Structural realism is unstable. It must tell us what the relata of the structure are. If they're elements of experience, this is just empiricism. If they're natural properties, this is just standard realism. If they're identified by the structure only, this is a mathematical platonism, not so far from idealism, and this is vacuous as a realism (because any mathematical structure exists, abstractly: no big deal). So, basically, we fall back on one of our three options.

What about modal empiricism?

F19. Carcinoma scirrhosum diffusum ventriculi. Alfred Kast Wellcome L0074358
So what all this has to do with modal empiricism? Aren't semantic and epistemic questions independent?

Well, a first remark is that I defended that modal empiricism is not realism, because it rests on an inductive, not abductive epistemology. But the contrast between empiricism and realism is not initially a contrast in which inferences are valid or not, but in whether one holds our theories to be true, or merely empirically adequate. Could an induction on possible situations be enough to know that our theories are true after all? Couldn't modal empirical adequacy collapse to truth?

Another remark is that modal empiricism does not entertain a distinction between what is observable or not. Empirical adequacy is expressed in terms of application and predictions, which concern the objectivable aspects of situations. I emphasised how active intervention was necessary to test theories when they posit unobservable entities, such as proteins. On the surface, the arguments are similar to the arguments in favour of entity realism. Again, couldn't modal empiricism be enough to claim that these entities exist? If these entities are identified by their causal role (functionally that is) and if causal relations can be empirically assessed, they certainly exist for the modal empiricist.

Finally, note that essences and intensions are often analysed in modal terms. If interpreting a theory amounts to describe the essential properties it refers to, or to give it an intension, and if, as I have argued, modal statements are not underdetermined by experience, then perhaps the interpretation of a theory is not underdetermined either?

Take Quine's example of renates (creatures with kidneys) and cordates (creatures with a heart). Both have the same extension (all animals with a kidneys have a heart) but a different intension. We could take this intension as a description of their essence. Imagine a theory that says that renates are hairy and another that says that cordates are hairy. For a standard empiricist, both have the same empirical consequences, although they're different theories, but not so for a modal empiricist: we could intervene in the world to differentiate them (through genetic manipulations, say: create a cordate that is not are renate). But then, the interpretation of the theory matters for possible predictions, and we might as well be realists. Ok, but what if we discover that the manipulations are impossible? What if the genes that code for kidneys also code for hearts? Well, we'd have discover that renates and cordates are identical, as a matter of natural necessity. But we can still be realists, and claim to have discover the essence of cordates and renates. Again, the interpretation matters.

All this works so long as our theoretical terms somehow keep in touch with empirical observations. But what linguistic resources do we have, beyond experience and modalities, to interpret our theories? So it seems that modal empiricism is just standard realism.

Pragmatic truth and internal realism

F8. Serosa intestinorum et mesentarium. Cholera. Alfred Kast Wellcome L0074374
What is puzzling is that we defended that modal empiricism is better than realism because it does not fall prey to a pessimistic meta-induction. So what has gone wrong?

The answer, I think, lies in the theory of truth one adopts, and incidentally, on the modalities involved. Correspondence truth introduces a gap between our epistemic abilities and the truth of our theory. But for the empiricist, no correspondence truth is involved. Theoretical terms are always interpreted in how they relate to experience. They matter only insofar as they change the conditions of application and prediction of theoretical models. We are talking about possible conditions here, and modal relations are involved, but no metaphysical necessity or possible worlds: only physical necessity, possible situations in the actual world, i.e. situations that we could implement by intervention.

This is some kind of reductionist account of meaning associated with a pragmatist theory of truth: following pragmatist truth, saying that our theories are modally adequate, or saying that they're true makes no practical difference.

There is an important difference with essential realism, which is that the meaning of theoretical terms will change from theory to theory, contra Kripke. Here, a theoretical term could be ideally analysed in terms of its modal relations to conditions of application and good prediction for the theory, i.e. in terms of its functional role in experimentation. This account seems similar to logical empiricists verificationism, or to operationalism, but the modal aspect and the pragmatic aspect of this construal (in particular, the fact that it does not rest on a distinction between observable and unobservable, and that the notion of application can incorporate pragmatic or contextual aspects) can help overcome the difficulties of these standard positions (such as the reduction of dispositional terms). But the problem of theoretical change remains: observations and interventions are theory-laden, so can't a new theory change the conditions of application of our theoretical terms? Then Kripke's arguments would apply: generally, we are willing to say that we could discover that gold is not yellow (because this is an illusion for example). Gold is not identical with its manifestations: it is what causes these manifestations. Similarly, we would be willing to accept that the way we apply terms like "electron" in experiments is misguided, in light of a new theory. The term "electron" cannot be analytically synonymous with some conditions of application, even extending to possible conditions of application.

I don't think this is a problem if we reject a strict notion of analyticity: indeed, the meaning of theoretical terms will change from theory to theory. However, experimental practices generally survive theory change. We can measure temperature with a thermometer, whether we endorse classical thermodynamics or statistical physics. The reason for this is a continuity in empirical adequacy. So, precisely, modal empiricism has the resources to answer Kripke's arguments: we keep using the same terms because they approximately play the same causal role, and we can be confident that they will continue to do so because our theories are modally adequate. The meaning of "mass" changes from Newtonian gravitation to general relativity, and the rest mass we attribute to the sun changes slightly, but not drastically. If it were to change drastically, i.e. if we discovered that our theories are not modally adequate, well, we would stop using the term "mass" because it would not apply any more. So modal empiricism is not analytic, but still, it guarantee a pragmatic continuity in theoretical term use.

By the way, this is a nice way to associate intensional and essentialist aspects: if experimentation is theory laden, then the intension of theoretical terms will tend to fixate on projectible predicates, i.e. to "group" objectivable aspects in such a way that necessary relations can be attributed to these groups. They mimic essential properties, because the "analytical" necessity associated with their intension (and the structure of the theory) will tend to match physical necessity. Or in a Quinean vein, there is no strict distinction between analytic and synthetic necessity, and meaning itself can be discovered empirically.

The conclusion of this is that modal empiricism, equipped with a pragmatic theory of truth, is just realism: not genuine realism for the aficionados of correspondence truth, but, at least, pragmatic realism (or perhaps internal realism).

Conclusion

F15 Typhus abdominalis, necrosis superficialis Alfred Kast Wellcome L0074365
Of course, one can choose the semantic theory that one wishes in order to make any epistemological position a realist position that says that our theories are true. The logical empiricists did just that with their verificationist theory of meaning. The advantage of modal empiricism is that its commitment to modalities allows for a richer semantic theory that can mimic semantic realism, and thus can probably retain its benefits. The modal empiricist can use theoretical language just as a realist would do. But if pressed on metaphysical or semantic questions, the modal empiricist will not claim that theoretical terms refer to natural properties or that theories correspond to reality: they're only applicable in bounded domains of experience, and they're interpreted relatively to our epistemic position, not in terms of absolute existence. Their meaning is just the functional role they play in experience, and it can be adjusted in front of new theories. That's how the modal empiricist can use a realist language, and at the same time maintain its advantages over realism in front of anti-realist arguments.

jeudi 3 novembre 2016

Why Empiricists should Endorse Modalities (2) Scientific Rationality

In the previous post, I explained why, assuming that there is necessity in the world, relations of necessity are perfectly knowable on the basis of experience, without any recourse to abductive reasonning. That certainly works in favour of an endorsement of natural modalities for the empiricist: at least, epistemic arguments against it are not conclusive. But still, an empiricist could resist this kind of commitment by assuming that there is no necessity in the world from the start. She could interpret modal discourse in a pragmatic way, rather than assume that modal statements have truth values. In this post, I want to explain why she shouldn't.

As I said in the previous post, I don't think that the existence of necessity in the world itself (as opposed to relations of necessity if there is necessity in the world) can be confirmed or disconfirmed by experience. However, no position really comes without some general metaphysical framework, and this includes all empiricist positions: the idea that all knowledge comes from experience is not itself confirmed or disconfirmed by experience. The idea that our theories will continue to be empirically adequate in the future is not confirmed or disconfirmed by experience. The empiricist is only willing to assume the minimum necessary, and to refrain from speculations. One could frame this as some kind of transcendental argument: we need some basic assumptions to make sense of the world, but we shouldn't assume more.

This is exemplified by van Fraassen's defense of constructive empiricism. He does not say that his position is itself confirmed by experience, but he rephrases empiricism as a position about the aim of science, as a collective endeavour: its aim, according to van Fraassen, is to produce empirically adequate theories (not true theories). A scientist can well be a realist, but needs not be to be a good scientist. However, a scientist must at least assume that our theories are empirically adequate, and will continue to be in the future. Now if one thinks that scientific practice is a rational activity, one should also believe that our best theories are empirically adequate. This is a minimum.

I think this is the correct way of thinking about epistemological positions. But I think that it supports modal empiricism rather than constructive empiricism.

mardi 25 octobre 2016

Why empiricists should endorse modalities (1): modal knowledge

Let us take stock. In the posts so far (linked below in order),

  • I presented modal empiricism: the view that our best scientific theories are empirically adequate for all possible situations to which they would apply.
  • I detailed the conception of empirical adequacy on which the position rests. It is not cast in terms of a model of the universe, as usually, but in terms of situations to which different models apply, and, I think this conception is more connected to scientific practice than the usual ones.
  • This conception of empirical adequacy does not commit us to modal empiricism, but I explained how modal empiricism is able to answer the no-miracle argument, while retaining the advantages of empiricism when it comes to theory change.
  • Finally, I explained why, according to me, scientific realism is misguided: it rests on (meta-)abduction for its justification, but abduction, however central it is in scientific practice, is not a principle of justification, but a strategic device to select good hypotheses: hypotheses that we should test first, that is. It doesn't exempt us from further empirical tests if we want to justify these hypotheses. But scientific realism cannot itself be tested empirically.

What makes modal empiricism a version of empiricism, not of scientific realism, is that in contrast with realism, the modalities to which it is committed are arrived at by induction on possible situations, not by abduction. Relations of necessity are no explanations to regularities, but regularities extended to the possible.

samedi 24 septembre 2016

Against abduction

Abduction (or "inference to the best explanation") is the cornerstone of scientific realism. There are always many different theories, or hypothesis that could account for some given phenomena, but scientists make their choice on the basis of non-empirical criteria, such as simplicity: they choose the best explanation. According to a realist, this means one likely to be true. This, in essence, is abductive reasoning: an inference from non-empirical, explanatory virtues to truth, or to likelihood. Furthermore, the realist claims that her position, that our best scientific theories are approximately true, is itself the best explanation to their predictive success. That's what we could call a meta-abduction: a justification of abduction (best explanations are true) by means of abduction (that best explains their success, so it's true). So it's clear that abduction is essential to realism, perhaps its definite characteristic. But is abduction a valid form of inference?

samedi 3 septembre 2016

Is empirical success a miracle?

In the previous post, I detailed my conception of empirical adequacy: a theory is empirically adequate if for every model of the theory, for all situations to which the model would apply, the model would make correct predictions. Depending on the range of situation we consider (situations actually experimented, actual situations we could experiment in principle...) on can derive different versions of empiricism. Modal empiricism is the view that our theories are empirically adequate for all possible situations.

In this post, I would like to explain why modal empiricism can respond to the no-miracle argument for scientific realism, and why it is not threatened by a meta-induction argument. But before that, we must examine the different kinds of induction that are involved in our definition.

lundi 22 août 2016

Empirical Adequacy: a Proposal

In the last post, I criticised van Fraassen's definition of empirical adequacy. According to van Fraassen, a theory is empirically adequate if it has at least one model such that all observable phenomena fit inside (they correspond to the empirical substructures of the model). My criticisms were the following: it rests on a problematic distinction between observable and unobservable, it does not take into account interventions and manipulations, which are central in scientific experimentation, and it refers to an hypothetical model of the universe, which is unnecessary and disconnected from scientific practice.

Can we do better? I think we can if we directly refer to scientific experimentation instead of coming up with an abstract reconstruction of empirical adequacy. Empirical adequacy should simply be framed in terms of the good predictions of models when they apply to various situations. Thus I suggest the following definition:

A theory is empirically adequate exactly if, for all its models, and for all concrete situations in the world, if the model applies to the situation, then its predictions are correct.

Here it is: that's a pretty simple definition. Now, of course, I need to expand a bit what all this means. This is the aim of the present post. But let me begin with an illustration.

Take as a concrete situation the evolution of the solar system during a certain period of time. A Newtonian model of the solar system applies to this situation if it correctly describes the planets and the sun, with their respective initial positions and masses. It makes good predictions if the evolution of the position of planets in the model correspond to the positions that we could observe in this situation. If this is so, then our model of the solar system is empirically adequate for this situation. If all models of the theory that we could apply in the world are empirically adequate for all situations to which they apply in the world, then our theory is empirically adequate.

I will now explain in more details what I mean by situation, application and prediction.

samedi 20 août 2016

What is Empirical Adequacy? Against Observability and the Model of the Universe.

In the last post, I explained how I conceive of physical theories: roughly, a vocabulary and axioms, from which we build models applicable to particular types of situations. Models can be mapped to concrete situations in the world, and we can compare their predictions to data models extracted from these situations. From this picture, we can ask: what is it, for a theory, to be empirically adequate?

Van Fraassen, who is one of the main contemporary defenders of empiricism, proposes a definition of empirical adequacy which is the following:

"A theory is empirically adequate exactly if what it says about the observable things and events in this world, is true—exactly if it ‘saves the phenomena’. A little more precisely: such a theory has at least one model that all the actual phenomena fit inside. I must emphasize that this refers to all the phenomena; these are not exhausted by those actually observed, nor even by those observed at some time, whether past, present, or future."

This definition dates back to his 1980 book "the scientific image" (You can find a similar one in a footnote of his 1989 "laws and symmetry"). What he means by "fit inside" is explicated later in the book in terms of isomorphism between data models and the "empirical substructures" of theoretical models. I think it's not the right way to understand empirical adequacy for several reasons, and although his notion of "fit inside" was challenged by some authors, who proposed more sophisticated accounts, this is hardly my main point of contention.