The Scientific Image

"Van Fraassen’s position is subtle. He does not deny that unobservable entities exist. Rather, he says that one need not believe in their existence in order to have a reasonable view of science and its practice. His anti-realism is based on the empiricist tenet that belief should be constrained by what is observable and actual and amounts to a kind of agnosticism about the existence of unobservables. He therefore defends what he calls ‘Constructive Empiricism’ in opposition to scientific realism. His account of scientific realism is in my view somewhat idiosyncratic, since he takes scientific realism to involve two theses: one axiological and another doxastic (about belief). The axiological thesis says that theories aim at truth; the doxastic thesis says that acceptance of a theory involves belief in its truth. Whereas, he takes Constructive Empiricism to say that theories aim at empirical adequacy , and that acceptance of a theory involves belief only in its empirical adequacy (though he adds that acceptance involves more than belief, viz. , commitment to use the theory to interpret the worldly phenomena). I said that this way to view the realism debate is idiosyncratic, since one can be a scientific realist or a constructive empiricist without thinking that theories have achieved their respective aims. Be that as it may, van Fraassen’s key insight is that an empiricist should set limits to what is accepted on the basis of experience, and since he thinks that unobservable entities are beyond experience, belief in them should be “supererogatory”. His key argument is that the extra risk that realists seem to take by believing in the reality of unobservable entities is illusory since the theory can only be proved wrong by showing that it is empirically inadequate. In any case, he adds, the claim that a theory is empirically adequate ( i.e. , that it saves all phenomena) is always more (or at least as) probable than the claim that a theory is true. I think a key problem with van Fraassen’s view is that it is inherently unstable. To see this, we have to reflect a bit more on the notion of empirical adequacy. A theory is empirically adequate if and only if it saves all phenomena, past, present, and future. Now, this is a no less utopian aim than proving the theory true, since at any given moment of time, scientists have only a finite amount of data available. Hence, even if these data do not refute the theory, they are far from proving that the theory is empirically adequate. The claim that a theory is empirically adequate is already ‘inflated’ vis-à-vis the available data, which show at most that a theory is unrefuted. But why go for belief in empirical adequacy as opposed to belief in truth? If the argument is that the former is epistemically safer than the latter, then this makes Constructive Empiricism unstable: the epistemic safety principle, if sensible at all, makes safer the even weaker belief in the claim that the theory is unrefuted (as opposed to the stronger belief that the theory is empirically adequate). Empiricism could be stricter than constructive empiricism: it could claim that the aim of science is to produce unrefuted theories. Constructive Empiricism is more liberal than this, but in being so, it sets some boundaries to what can be known that does go beyond what a strict version of empiricism would allow, viz. , that only what has been experienced can be known. But then, there is no logical obstacle in setting the boundaries a little higher, as realists demand. The issue of observability has drawn considerable attention among philosophers of science. There is a famous argument, by Grover Maxwell, to the effect that all entities are observable under suitable circumstances. Maxwell’s point was that ‘observability’ should be best understood as ‘detectability through or by means of some instrument’. Now, van Fraassen takes it to be the case that an entity is observable if it could be observed by a suitably placed observer. This claim is modal (‘ could be observed’), but it is not clear how the modality is to be understood. Are dinosaurs, for instance, ‘observable’ even if their observation would require time travel? And are sun spots ‘observable’ even if, strictly speaking, no one could be close enough to the sun to observe them? To be sure, van Fraassen claims that observability concerns empirically discoverable facts about humans qua organisms in the world. But even if we were to grant that there is an empirically discoverable divide between observable entities and unobservable ones, we are still left with a question: why should the observable/unobservable distinction capture the border between what is epistemically accessible and what is not? “Are dinosaurs, for instance, ‘observable’ even if their observation would require time-travel?” What van Fraassen has failed to establish is that the boundaries of experience should include only claims about unobserved-yet-observables and that they ought to exclude all claims about unobservables. In fact, there is a venerable empiricist tradition, exemplified by Hans Reichenbach and Wesley Salmon, according to which an empiricist epistemology can lead to accepting the reality of unobservable entities, based on suitable ampliative methods, without thereby abandoning empiricism. It’s interesting that the key anti-realist claim has been based on some kind of epistemic dichotomy—that some realms of being are cognitively impenetrable by us. The dichotomy has been by and large, vertical: there is something epistemically suspicious with the unobservable per se , or some aspects of the unobservable. As we have seen, for Constructive Empiricism, the epistemic dichotomy is drawn quite sharply along the line of the observable/unobservable distinction. What’s worth noting is that subsequent forms of anti-realism were also weak realist positions, since the dichotomy is now drawn within the realm of the unobservable, therefore allowing that there is epistemic access to some unobservable parts of reality. Get the weekly Five Books newsletter The two most promising but ultimately failing anti-realist views are Kyle Stanford’s neo-instrumentalism and Derek Turner’s historical hypo-realism. In the latter, the dichotomy is between the past and the tiny. Turner claims that we can know more about the tiny than the past; hence, it is safer to be a scientific realist about the tiny unobservables, such as electrons. He bases his claim on a distinction between a unifier (an entity that plays a unifying role) and a producer (an entity that can be manipulated to produce new phenomena), and argues that past (un)observables (like dinosaurs) can at best be unifiers, whereas tiny unobservables can be producers, too. “Turner claims that we can know more about the tiny than the past; hence, it is safer to be a scientific realist about the tiny unobservables, such as electrons” In Stanford’s case, the epistemic dichotomy is between those entities to which there is an independent route of epistemic access (mediated by theories that cannot be subjected to serious doubt) and those entities to which all supposed epistemic access is mediated by high-level theories. Kyle Stanford takes it that the former are epistemically accessible, while the latter are impenetrable. High-level theories are taken to be useful conceptual tools for guiding action rather than maps of a reality unavailable to the senses. Part of Stanford’s motivation is the claim that, given the past record of science (especially when it comes to high-level theories), it is likely that the truth lies in the space of currently unconceived alternatives to extant scientific theories. But, for one, this kind of argument neglects the fact that as science grows, the space of unconceived alternatives is constrained and restricted by what we already know; that is, by well-established scientific theories. For another, the problem of the existence of unconceived alternatives is a general problem for epistemology. Given that there is no deductive link between evidence and theory, it is always possible that current theories are false, and hence that there are unconceived alternatives to them. This gives rise to the issue of under what conditions we are entitled to talk about justification and knowledge, which is a general epistemological problem to be dealt with independently of the realism debate."