The Aims of Reliable Knowledge: A Reply to Seungbae Park, Richard A. Healey

As a scientific realist, Seungbae Park in his (2019) paper “The Disastrous Implications of the ‘English’ View of Rationality in a Social World” seeks to repel an anti-realist attack from Bas Van Fraassen by turning his opponent’s new epistemology against his own constructive empiricism and theory of explanation …[please read below the rest of the article].

Image credit: Clayton Shonkwiler via Flickr / Creative Commons

Article Citation:

Healey, Richard A. 2019. “The Aims of Reliable Knowledge: A Reply to Seungbae Park.” Social Epistemology Review and Reply Collective 8 (9): 25-30.

The PDF of the article gives specific page numbers.

This article replies to:

As a scientific realist, Seungbae Park in his (2019) paper “The Disastrous Implications of the ‘English’ View of Rationality in a Social World” seeks to repel an anti-realist attack from Bas Van Fraassen by turning his opponent’s new epistemology against his own constructive empiricism and theory of explanation. He motivates this interesting strategy by appealing to a reciprocity principle of social epistemology which may be stated as follows: In a dispute between philosophers advocating rival epistemologies, each is entitled to apply his opponent’s own epistemological principles in arguing against him. If the argument succeeds, then the opponent’s views are thereby show to be self-undermining.

On Constructive Empiricism

Seungbae correctly notes an important distinction that others have missed. Van Fraassen’s constructive empiricism is not itself an epistemological doctrine. In particular, a constructive empiricist may consistently adopt any of a variety of epistemic attitudes toward a successful scientific theory. He may believe that it is true in what it says about observable matters while holding no beliefs about what it says about unobservables: he may believe that it is true in what it says about observable matters while disbelieving what it says about unobservables: or he may believe that it is true in what it says about everything, unobservable as well as observable.

A constructive empiricist may also reserve epistemic judgment on a successful scientific theory because he considers it premature to accept it, and so believes neither that it is true nor that it is empirically adequate. Reflection on the fate of historically successful theories may encourage adoption of this conservative epistemic attitude.

If constructive empiricism (CE) is not an epistemological doctrine, then what is it? It is a theory designed to explain scientific activity as a goal-directed enterprise. This theory takes the goal of science to be construction of acceptable theories, where what counts as acceptance involves, as belief, only that the accepted theory is empirically adequate. It is to be contrasted with scientific realism (SR), which Van Fraassen takes to offer a rival theory, also designed to explain scientific activity as a goal-directed enterprise. (SR) agrees with (CE) that the goal of science is the construction of acceptable theories, but disagrees about what counts as acceptance: for the scientific realist what it is to accept a scientific theory is simply to believe that it is true.

In the course of his argument for (CE), Van Fraassen advances a contextual theory of explanation in science (CT). He uses this to rebut a typical realist who argues that we have reason to believe that a scientific theory is not merely empirically adequate but true if it successfully explains phenomena (perhaps, phenomena that some empirically equivalent theory merely predicts or accommodates). (CT) is actually a theory not just of scientific explanation, but of a (supposedly) broader category of why-questions and their answers: for Van Fraassen views explanation as merely an application of science, not one of its constitutive goals.

Alongside (CE) and (CT), Van Fraassen has also advocated what he calls a new epistemology—not just of science. This new epistemology (NE) replaces the traditional goal of justifying belief with the more limited (and so more attainable?) goal of characterizing rational changes of belief. Though set in a Bayesian framework, (NE) does not identify rational belief-change with Bayesian conditionalization or any alternative rule. (NE) is voluntarist insofar as it allows that rational belief change is constrained by no such rule, and that each of many contrary belief changes may be rational. Van Fraassen’s rejection of a supposed rule of inference to the best explanation (IBE) is therefore just one instance of this voluntarist epistemology.

Van Fraassen certainly does not take (CE) to imply or motivate the replacement of (IBE) by a potentially weaker rule of inference to the empirical adequacy of the best explanation. He offers the contrast between what Oliver Wendell Holmes called the English and the Prussian views of law as an appropriate analogy for the contrast between his voluntarist epistemology and a more traditional rule-constrained epistemology. His English view of rationality is no mere philosophical toy but a serious proposal in general epistemology with applications far outside the scientific realism/anti-realism debate.[1]

Countering Van Fraassen

After this reconnaissance of Bas’s position, I now assess the prospects of Seungbae’s attacks. His first target is (CT), Van Fraassen’s contextual theory of explanation. In The Scientific Image Van Fraassen (1980, Chapter 5) proposed a theory of explanation that takes an explanation to be an answer to a why-question, as this arises in a certain context. In accord with his semantic conception, he thinks of this theory as providing models of the use of language in such contexts. These models are abstract objects, whose elements are themselves abstract—propositions and relations among them. If the theory is sufficiently successful, one should accept it. But does that involve believing that it is true?  Not for a constructive empiricist like Van Fraassen, for that would involve believing in the existence of all the unobservable abstracta that feature in the relevant models.

While Van Fraassen never says that he believes his own theory is true, he does say that he believes it is “basically correct” (146). Even an unsympathetic reader who does not distinguish between ‘correct’ and ‘true’ here must acknowledge a gap between ‘correct’ and ‘basically correct’. A more sympathetic reader would rather identify ‘correct’ with ‘acceptable’ and take Van Fraassen here to be stating his belief that, while not acceptable as it stands, some development of (CT) will be acceptable.  But as an empiricist, his acceptance of (CT) or a development of it would certainly not involve belief that this is true, for that would commit him to belief in all those abstracta that feature in the relevant models. His voluntarist epistemology allows that such commitment is not in itself irrational—but also that he is equally rational not to make it.

Park thinks that Van Fraassen’s voluntarist epistemology undercuts his own theory (CT) because neither he nor anyone else is irrational to disbelieve (CT). I have shown that Van Fraassen can happily admit that he does not believe that (CT) is true, while consistently maintaining that it is basically correct. Does this leave Van Fraassen in a disastrous position? I don’t see that it does. If I were Van Fraassen I would consider Seungbae’s attack on my contextual theory of explanation successfully repulsed. Turning his own voluntarist epistemology against him has not dislodged him from that position.

Regarding the “Empiricist Position”

Park’s next section (3.2) is called Empirical Adequacy. Though clearly intended as an attack on Van Fraassen’s position, it locates as target “the empiricist position that T is merely empirically adequate.”  This is puzzling. We are told nothing in this section about theory T so it is not clear how to begin assessing this “empiricist position.” Since Park began his paper by supposing that T best explains some available data I will take it that this supposition is still in effect. In that case another puzzle arises: Who occupies this “empiricist position”? Since Park’s paper is directed against Van Fraassen he apparently takes this to be Van Fraassen’s position. But it is not. (CE) does not imply that a theory that best explains some data is (merely) empirically adequate. Since it is not an epistemological position, neither does (CE) imply that a scientist should believe that such a theory is (merely) empirically adequate. To repeat, (CE) is compatible with a range of epistemic attitudes toward a successful scientific theory. Moreover, we saw that Van Fraassen’s voluntarist epistemology disavows any rule of inference to the empirical adequacy of the best explanation while allowing that a belief change describable as inference to the mere empirical adequacy of the best explanation may be rational when also would be a belief change describable as inference to its truth.

Park apparently attributes this “empiricist position” to Van Fraassen because of a passage (1985, 294) in which Van Fraassen is concerned to refute a scientific realist claim, not to state his own position. A realist may claim that of two rival theories passing the same tests a scientist will (and should) accept the more explanatory because its additional explanatory power means that it is more likely to be true. Against this claim Van Fraassen objects that the more explanatory theory cannot be more belief-worthy because it must be more informative and so, to put it crudely, must have more ways of being false. Van Fraassen acknowledges that in such a case a scientist may have reason to accept the more explanatory theory—not as true, but as empirically adequate—but she has no additional reason to believe that it is true. His appeal to economy of belief in refuting the scientific realist claim in no way commits Van Fraassen to the empiricist position that the best explanatory theory T is merely empirically adequate.

Where Van Fraassen does appeal to economy is in defending (CE) as a better theory of scientific activity than (SR) because it saddles science with none of the metaphysical commitments he associates with (SR). (CE) and (SR) offer rival accounts not of belief but of acceptance. While he thinks it is obvious that the empirical adequacy of a theory requires less by way of belief than its truth, it takes argument to show that (CE) can account for the goal-directed activity of scientists and for what they take to be required of an acceptable theory without saddling them with what he believes to be the excess cost of gratuitous metaphysics. So Park’s real target in section 3.2 should be (CE) itself, not the empiricist position that the theory T that best explains the available data is merely empirically adequate. For it is (CE) whose superiority over (SR) Van Fraassen argues for on grounds of metaphysical economy.

As a foil for his so-called empiricist position, Park here introduces a view he calls Skepticism: Some observational consequences of T are true.

Assuming T is acceptable, in part because of its explanatory power, a proponent of (CE) will believe Skepticism, as also would anyone adopting the empiricist position that the best explanatory theory T is merely empirically adequate. Park argues that Van Fraassen’s voluntarist epistemology commits him to the rationality of one who believes Skepticism but disbelieves that empiricist position. He goes on to consider an imaginary scientific episode in which such disbelief seems unwarranted and argues that this could have adverse consequences for scientific practice.

As an argument against Van Fraassen’s scientific anti-realism this misses its target, since Van Fraassen does not hold Park’s so-called empiricist position. But perhaps Park’s argument can be reformulated and redirected against (CE) itself? This would involve transforming Skepticism into a skeptical rival to (CE) and employing the similar strategy of using Van Fraassen’s voluntarist epistemology to allow as rational a negative epistemic attitude toward an explanatorily successful theory—an attitude that would fly in the face of good scientific practice.

Destructive Empiricism

I suggest the following foil for (CE):

Destructive Empiricism (DE): Science aims to give us theories some of whose observational consequences are true: and acceptance of a theory involves as belief only that some of its observational consequences are true.

(DE) retrenches the goals and epistemic commitments involved in (CE) just as (CE) retrenches those involved in (SR). So (DE) further deflates the ambitions of science while lowering the standards for realizing them. As a proponent of (CE), Van Fraassen would clearly wish to reject (DE) as a good explanation of actual scientific practice. But does his voluntarist epistemology commit him to the rationality of a destructive empiricist—i.e. one who believes (DE) but disbelieves (CE)?

A practice best explained by (DE) would not be recognizable as science. Any theory deliberately constructed to accommodate any existing observational data would be immediately accepted just because it accommodated them. The burden of testing a theory would immediately be lifted and epistemic doubts about it reduced to doubts about the data it was constructed to accommodate. Such an activity fails the most basic condition that scientific activity involve ampliative inference. The epistemic economy of (DE) is illusory: while it minimizes the epistemic cost of accepting a theory, this confers no associated benefit. Destructive empiricism deserves its name.

As a constructive empiricist offering a theory of scientific practice, Van Fraassen should accept a sufficiently successful theory as empirically adequate. Since (DE) is a wildly unsuccessful theory it would be irrational for him (or anyone else) to accept it. If (DE) is unacceptable then (CE) implies that it would be irrational for a voluntarist to believe that it is true. So Van Fraassen may—indeed should—reject (DE) despite the illusory economy promised by its condition on acceptance as compared to that of (CE).  While loosening the bonds of rational belief, voluntarism does not imply that anything goes.

What could Van Fraassen say in response to someone (whether scientist or philosopher) who comes to believe that a theory T best explains some available data while believing Skepticism but disbelieving (what Park calls the empiricist position) that T is empirically adequate? That depends on whether this skeptic deems T acceptable, given the available data. If she does, then she would be irrational not to abandon her skepticism by coming to believe that T is also empirically adequate. If she does not (yet?) deem T acceptable, then she should not believe that T is empirically adequate: it is up to her whether to persist in her disbelief in T’s empirical adequacy―voluntarism considers suspension of epistemic judgement and continued disbelief equally rational options here.

Up to this point I have played devil’s advocate by suggesting how Van Fraassen might respond to Park’s attacks on his position. Lest I be taken for a constructive empiricist, I now conclude by speaking briefly in my own voice.

Theory Acceptance and Reliable Knowledge

As an explanation of scientific practice, constructive empiricism strikes me as both too strong and too weak. Constructing empirically adequate theories is neither the sole nor the overriding aim of science, and accepting a theory involves both more and less epistemic commitment than believing that it is empirically adequate. I’ll use this famous quote from Feynman’s lectures on physics to illustrate both points.

If, in some cataclysm, all of scientific knowledge were to be destroyed, and only one sentence passed on to the next generation of creatures, what statement would contain the most information in the fewest words? I believe it is the atomic hypothesis (or the atomic fact, or whatever you wish to call it) that all things are made of atoms—little particles that move around in perpetual motion. Attracting each other when they are a little distance apart, but repelling upon being squeezed into one another (1-2).

Feynman says that this sentence contains an enormous amount of information about the world, and goes on to call it both an idea and a hypothesis. He does not call it a theory, and he says neither that it is true nor that it is empirically adequate. In fact it is neither, as Feynman well knew despite taking it to be a core piece of scientific knowledge. Electrons, light, black holes―none of these are made of atomic matter. Dalton had an atomic theory, as did Bohr, though neither is today accepted as true or empirically adequate.

Yet Feynman’s sentence encapsulates just the kind of “reliable knowledge” at which science aims. Though literally false, it is true of a large number of things that we encounter all the time, and provides the key assumption required to explain why these behave the way they do. But atoms are not observable, though they can now be imaged by atomic force and scanning tunneling electron microscopes. Against the voluntarist constructive empiricist I would argue that a scientist today would be irrational to accept our best atomic theories but not to believe that there are atoms. Our best atomic theories are certainly improvements on Dalton’s and Bohr’s, but we know the idealizations they involve render them at best approximately true: accepting such a theory does not involve even the belief that it is empirically adequate.

Both constructive empiricism and (what Van Fraassen calls) scientific realism exaggerate the importance of theories in science and set the bar on acceptance of a theory too high. A more defensible formulation of scientific realism would offer a more nuanced account of the role of theory and theory acceptance. But it would abandon empirical adequacy in favor of truth as a goal when it comes to claims about the world. We now know that there are atoms, electrons and black holes, and we know a lot about them, even though even our best theories of these things are neither true nor empirically adequate. Scientists consider even the Standard Model of high energy physics merely a low energy effective theory, and the general theory of relativity cannot be considered even empirically adequate in its treatment of certain phenomena involving black holes where quantum effects become important.

While sympathizing with Van Fraassen’s aversion to a metaphysics of causation, laws of nature and propositions, I see no reason to extend this to skepticism about features of the physical world that our unaided senses happen to be too weak to detect without instrumental aids. Feynman’s sentence encapsulates one profound example of the kind of knowledge at which science aims and it would be irrational for a scientist (or anyone else) to dismiss this as mere speculation.

Contact details: Richard A. Healey, University of Arizona,


Feynman, Richard P. 1963. The Feynman Lectures on Physics, Volume 1. Redwood City, Ca.: Addison-Wesley.

Park, Seungbae. 2019. “The Disastrous Implications of the ‘English’ View of Rationality in a Social World.” Social Epistemology 33 (1): 88-99.

Van Fraassen, Bas C. 1980. The Scientific Image.  Oxford: Clarendon Press.

Van Fraassen, Bas C. 1984. “Belief and the Will.” The Journal of Philosophy 81 (5): 235-256.

Van Fraassen, Bas C. 1985 “Empiricism in the Philosophy of Science.” In Images of Science edited by Paul M. Churchland and Clifford A. Hooker, 245-308. Chicago: University of Chicago Press.

[1] See, in particular, his (1984) “Belief and the Will.”

Categories: Critical Replies

Tags: , , , , , , ,

3 replies


  1. Constructive Empiricism in a Social World: Reply to Richard Healey, Seungbae Park – Social Epistemology Review and Reply Collective
  2. Four Points in Response to Seungbae Park, Richard A. Healey – Social Epistemology Review and Reply Collective
  3. Replies to Healey’s Comments Regarding van Fraassen’s Positions, Seungbae Park – Social Epistemology Review and Reply Collective

Leave a Reply