Sveinbjorn Thordarson's Website

Van Fraassen walks a strange line between realism and anti-realism. His position, which he calls ‘contructive empiricism’, has been described as the “most cogent and well-developed alternative to scientific realism” (Ladyman 2000). It is broadly realist concerning theories without unobservable entities but anti-realist about theories with unobservable entities. On this account, scientific theories are to be taken literally – they are statements about the world capable of being either true or false, depending on whether the descriptions they provide correspond to a mind-independent reality. However, acceptance of a given scientific theory does not commit us to holding that it is literally true, or entail that we believe in the existence of the unobservable entities it postulates. Indeed, according to Van Fraassen, the aim of scientific theories is not to give us a ‘true’ account of the world, as such, but rather to be ‘empirically adequate.’ A theory is empirically adequate if and only if what it says about that which is observable is true, or in Van Fraassen’s words, if “it saves the phenomena” (Van Fraassen 1980).

It should be clear from this brief summary of constructive empiricism that the notion of observability plays a crucial role. If we accept constructive empiricism, our level of commitment to theories hinges on this concept. A theory which only contains observables and is empirically adequate may be said to be true, in the sense that it provides a correct description of the world, whereas theories with unobservables are underdetermined; a variety of different and conflicting theories containing unobservables may be equally adequate empirically. It is therefore worth examining whether the notion of observability can bear the weight Van Fraassen assigns to it.

In response to the observable-unobservable dichotomy postulated by the logical empiricists, Grover Maxwell argues that the distinction between unobservable and observable entities is a tenuous and difficult one. The notion of observability is a vague predicate, with all the problems that this entails. We have a continuous series of unaided to aided observations running through a spectrum, with something like direct observation under ideal circumstances on one end, and complete unobservability on the other, with varying degrees of non-ideal or technically assisted observations in between. Where should the line be drawn? Maxwell’s position is that we have no clear criterion by which to draw such a line, and therefore concludes that “our drawing of the observational-theoretical line at any given point is accidental” and, furthermore, that “it has no ontological significance” (Maxwell 1998).

Maxwell is right, I believe, in his claim that any attempt to draw a clear line between that which is observable and that which is not is bound to be arbitrary and devoid of ontological significance. But, as Van Fraassen is quick to point out, the fact alone that observability is a vague predicate need not be problematic. There are many vague predicates in natural language which we find no difficulty in employing. Thus, while it is true that observability is a vague concept, it need not necessarily be a useless one. Although we may not be able to draw a clear and distinct line between observable and unobservable entities, and despite the fact that there are difficult cases in the gray area of the observability spectrum, we do have clear and unambiguous cases of observables and unobservables, which should be sufficient to make the distinction usable. Furthermore, while there is no necessary relationship between existence and observability – the former clearly does not entail the latter – the question of whether observability is epistemologically relevant is another matter, for "even if observability has nothing to do with existence […] it may still have much to do with the proper epistemic attitude to science" (Van Fraassen 1980).

Van Fraassen does not argue that the myriad of unobservable entities postulated by current scientific theories do not exist. His constructive empiricism is agnostic about unobservables, rather than directly anti-realist: they may exist, or they may not. Therefore, Maxwell’s criticisms, or any other criticisms based on the vagueness of the notion of observability, supposedly do not apply, since Van Fraassen’s reservations about unobservables are epistemological, not ontological. These reservations are closely tied to his claim that empirical adequacy should be the aim of science, as opposed to truth. In the case of any scientific theory which includes unobservable entities, the theory’s truth cannot be determined by its empirical adequacy (and indeed, this follows logically from Van Frassen’s definition of empirical adequacy). Judgment concerning the existence of unobservables should therefore be suspended because they are a needless risk.

Does Van Fraassen’s claim that unobservable entities have a special epistemological status hold? To answer that question, we must elaborate: To Van Fraassen, unobservables are unobservable-to-us – we cannot “step out of ourselves” and think of observability in objective terms. In other words, there is no eternal Platonic idea of observability. Consequently, that which counts as observable will necessarily be relative to the general biological sensory equipment of humans. The human organism is a measuring apparatus – the only one ‘directly’ available to us. From this he proceeds to argue that

it is […] not irrational to commit oneself to a search for theories that are empirically adequate, […] while recognizing that what counts as an observable phenomenon is a function of what the epistemic community is (Ibid)

Our rational level of theory commitment is therefore only to theories that account for that which is observable. We are safer if, based on the status of the epistemic community, we commit ourselves only to the existence of that which is perceivable by the cognitive apparatuses of the community of enquirers. All this shifts the entire burden of Van Fraassen’s position to his epistemology – which involves limiting what we know to what we can squarely place under the label “observable”. This, in turn, means that his position can only be defended if we accept this particular epistemology – and it obviously has alternatives.

Let me outline, as an example, the following epistemological scheme: Let us suppose that we never observe the things-in-themselves, but merely sensory experiences differing only in how far back we can trace the causal chain of their origin. If I visually observe a common, medium-sized object such as table, I can provide a certain causal story that explains how I come to see the table – rays of light fall on the table and are reflected by it through the air and to my eyes, which in turn transmit signals to the brain. When I look through a high-power microscope, I provide a qualitatively identical causal story, albeit a longer one. With purportedly unobservable entities, then, the only difference is the length of the causal story. Under this scheme, if we are constructive empiricists and adhere to this epistemology, we must base our attitude concerning the existence of entities on the length of the causal story – but there is no qualitative epistemological difference. Observability is thus reduced from a vague concept to a useless one, since there can obviously be no particular step in the causal chain which is fundamentally different from any other – in other words, there will be no clear cases, and thus the constructive empiricist position crumbles. Now, it is not my intent to defend this sort of Kantian ding an sich epistemology, but rather to demonstrate that Van Fraassen’s constructive empiricism is dependent on a certain epistemological view which one is by no means compelled to subscribe to.

Since the foundations of constructive empiricism rest on Van Frassen’s epistemology, we must ask whether it has any substantial benefits to offer. It is quite clear that Van Fraassen is trying to minimize his commitments. He is willing to be adventurous about scientific theories – they are capable of being literally true or false – but unwilling to take the step of affirming the existence of unobservables. His position thus falls somewhere between the realist and the skeptic. Is it any less problematic than the alternatives, say outright anti-realism about theories or full-fledged realism? This is not at all clear. As I see it, constructive empiricism tries to provide good grounds for why observability should be the criterion by which we determine whether to maintain that something exists. But if we are good empiricists, we are committed to basing our ideas about the world only on what has been observed, and this is distinct from what is observable. Observability is based on a great variety of contingent and incidental factors, and seems questionable as the basis on which to draw the line between what we choose to maintain actually exists and what merely serves as an instrumentally useful fiction. It is thus open to criticism via some deeply troubling and counter-intuitive thought-experiments (e.g. Churchland 1982). The rationality that Van Fraassen attributes to his position ends up being radically dependent on whether we accept an epistemology under which we can point to clear cases of observables and unobservables. For my part, I am not sure that we should wish to subscribe to such a position.

London 2007

Sveinbjorn Thordarson

References

Churchland, P. (1982): “The Ontological Status of Observables: In Praise of the Superempirical Virtues”, Pacific Philosophical Quarterly 63, pp. 226-236.

Ladyman, J. (2000): “What’s Really Wrong with Constructive Empiricism? Van Fraassen and the Metaphysics of Modality”, British Journal of the Philosophy of Science 51, pp. 837-856.

Maxwell, G. (1998): “The Ontological Status of Theoretical Entitites” in Philosophy of Science: The Central Issues, ed. Curd, Martin & Cover, J.A. pp. 1052-1063

Van Fraassen, B. C. (1980): The Scientific Image, United Sates; Oxford University Press.