Saving The Unobservable Phenomena (2017)

Walter Tay Ann Lee

Associate Professor Kuldip Singh

UNL2210 – Mathematics and Reality

22 September 2017


In this paper, I will consider the arguments that have been offered for the ‘common conception’ of constructive empiricism. I will claim that this ‘type’ of constructive empiricism is problematic because the argument from underdetermination supporting it contradicts the scientist’s belief in the existence of causality. However, I will further argue that van Fraassen’s constructive empiricism is more nuanced and does not have the same contradiction.





What is the ‘common conception’ of constructive empiricism and how does it differ from van Fraassen’s constructive empiricism? Dyck (2007) has made a powerful commentary on the apparent misreading of van Fraassen’s The Scientific Image (1980) and the supposedly imaginary arguments for constructive empiricism that scientific realists attribute to him. Some realists such as André Kukla (1998) even note, “Yet it’s curiously difficult to locate the exact place in van Fraassen’s writings where this argument is presented in fully general form.”

Thus, I shall, for brevity, introduce an imaginary single author of the ‘common conception’ of constructive empiricism and call him CE (in the style of Psillos 1997). CE begins his argument by first defining constructive empiricism, which is perhaps best defined as the denial of scientific realism. To be a constructive empiricist, one must at least be sceptical about the two main tenets of scientific realism:

[1]    That science aims at truth about the world

[2]    That if we accept a theory, we must also believe that whatever objects the language of the theory “picks out” from nature must be true, or approximately true

CE denies [1] because for him, the aim of science is not to produce true theories, but rather to produce empirically adequate theories. To clarify, CE does not deny the realist that truth exists. It is clear from van Fraassen’s (1980) writings that he himself adopts a correspondence theory of truth: “I would still identify truth of a theory with the condition that there is an exact correspondence between reality and one of its models.” CE believes that the claims our theories make may sometimes accurately describe the actual state of affairs in the external world, however, CE denies that such truth is possible from empirical data alone. This is explained in CE’s denial of [2].

CE’s denial of [2] is supported by his ‘argument from underdetermination’. To understand his argument, we must first consider three key distinctions that CE (as well as van Fraassen) makes:

[3]      Observable/Unobservable Distinction

[4]    Truth/Empirical Adequacy Distinction

[5]    Acceptance/Belief Distinction

What is the observable/unobservable distinction, [3]? For van Fraassen (1980) and CE, observability is always relative to an epistemic community and depends upon what the members of that community are able to detect with their unaided senses. For example, Mars, Singapore, and Obama are all observable; electrons, genes, and gravitational forces are not. We can see the red-ness of an apple (thus it being an observable feature), however we do not see the atoms that the apple is being composed of (thus it being an unobservable feature). For CE, observable facts are just facts that involve the observable features of observable objects.

Distinction [3] allows CE to draw a further distinction [4], which leads us to our argument from underdetermination. CE believes that all evidence is observational because if we cannot observe the unobservables, then we can never ‘know’ it to be true. CE invites us to imagine an infinite amount of empirically equivalent rivals of a theory (Psillos 1996). Since empirically equivalent theories all save the same phenomena and are equally supported by all possible evidence, all of them will always be equally believable (Dyck 2007). Thus, this implies that the belief in any specific theory among empirically equivalent rivals must be arbitrary and unfounded.

What about the belief is unfounded? For CE, it seems that scientific theories make claims about both the observables and the unobservables. Thus, CE believes that a theory is empirically adequate if everything the theory says about the observables is true (thus, saving the phenomena). Hence, from [4], a theory may be both true and empirically adequate; the theory perfectly depicts the actual state of affairs in the external world, and consequently is able to save the phenomena it purports to save. However, a theory may also be false and empirically adequate; the theory wrongly depicts the actual state of affairs in the external world, but is still able to save the phenomena. For CE, we are making a deductive leap from empirical adequacy to truth (realists call this IBE or Inference to Best Explanation) when we ‘believe’ [2] since the only thing we have is empirical data.

When discussing what the aim of science is, it seems that we are also concerned with the criteria to which a theory is accepted. And this is the main distinction [5] that leads CE to a constructive empiricist view on the aim of science. CE asks, “Do we need the belief that the theory is true to accept a theory?” For CE, there is no such need. When we accept a theory, we are merely saying that what the theory says about the observables is true. However, to believe a theory, we are also saying that what the theory says about the unobservables is true. Thus, begins CE’s scepticism about ‘general theories and explanations’ that intend to give an account of the observable world in terms of unobservable entities and processes (van Fraassen 1989). On CE’s view, we should surely accept a theory that is empirically adequate, however, belief in the truth of a theory (that is, believing what the theory says about the unobservables) can never be warranted.

Hence, it is clear why CE believes that the aim of science is to generate empirically adequate theories. However, this does not sit well with many scientific realists.





In this section, I will be expounding on a contradiction briefly mentioned in two sentences by Psillos (1997), “If IBE is generally abandoned, then we are left with a poor epistemology that admits only judgments about observed things. Cartesian scepticism might well be evaded, but Humean scepticism is in the offing.”

What is the contradiction between CE’s argument from underdetermination and our belief in the existence of causality? To begin, we must first understand the two sceptic hurdles that a scientist must overcome:

[6]    Cartesian Scepticism

[7]    Humean Scepticism

Very briefly, Cartesian scepticism [6] is an extreme form of scepticism whereby the truth of any anything cannot be known if one is able to doubt it. For example, because I am able to doubt whether my hand is actually my hand (because I could be in a dream and possibly not have hands in the real world), I can never know for sure that my hand is my hand. It is clear why Cartesian scepticism is an important hurdle for scientists. If we cannot know about anything in the world (or even if the external world exists), what are our scientific theories actually saying? However, in this paper we are not concerned with Cartesian scepticism since both scientific realists and CE seem to assume (or possibly even have an argument for) the existence of an external world, and that we can ‘know’ about the external world (or its representations) through observation.

Humean scepticism [7] is the one we are mainly concerned with. Hume is largely sceptical about the existence of causality. Hume (2007) begins his argument by dividing “all objects of human reason or enquiry” into Relations of Ideas, and Matters of Fact.

Hume (2007) explains that Relations of Ideas are propositions that are discoverable  by the mere operation of thought. That is, they do not depend on our observations of what exists in the external world. For example, he says “that the square of the hypotenuse is equal to the sum of the square of the two sides” is a proposition that can be validated with merely logical thinking. The idea here is that Relations of Ideas are not validated by observations of the external world.

Matters of Fact, on the other hand, must be validated by observations of the external world. For example, how do we know that the sun will rise tomorrow? We can only ‘know’ that the sun will rise tomorrow by observing it.

Now, it is important to note a very crucial distinction made by Hume between Relations of Ideas and Matters of Fact. The principal distinction is that the denial of a true relation of ideas implies a contradiction. What does this mean? Take, for example, the Pythagoras’ Theorem. If I say, “a2 + b2 < c2”, this denial of the true relation of ideas (that a2 + b2 = c2) implies a contradiction. We know that this proposition is false by means of logic alone. However, for Matters of Fact, its denial cannot be known independently of experience. For example, it is not inconceivable to imagine the sun rising from the west to the east rather than from the east to the west. The negation of the “true matter of fact” does not imply a contradiction. What this means is that for Matters of Fact we cannot use logic alone to assess the claim, we must validate it by means of experience and observation.

Why does this distinction matter? Hume is particularly sceptical about the nature of causation. He gives the example of a billiard ball colliding into another billiard ball. By relation of Cause and Effect, we can determine that the act of the first billiard ball colliding into the second billiard ball has caused the latter to move. However, it is not inconceivable for us to imagine that the movement of the second billiard ball was not due to the collision with the first ball, but that the two events of the balls moving just occurred in conjunction with each other. That is, Hume invites us to doubt, “Do we actually see causation? Have we seen one ball cause the other to move? Or do we just see one ball moving after the other?”

Now, because it is not inconceivable for us to imagine such a denial of causation, this implies that causation must be a matter of fact! Why? Because if the denial of causation cannot be proven to be a contradiction by means of logic alone, then it must be a matter of fact. If causation is a matter of fact, that means that the truth of causality must be in some way, dependent upon experience. For Hume (2007), the existence of causality cannot be assumed without validating it with experience, since it is a matter of fact. But what is causality and have we observed it before? There lies the crux of the problem.

Now, how does this tie in to CE’s argument from underdetermination? The key to Hume’s scepticism about causality is that because we cannot observe it, thus we cannot believe that it exists. Notice how similar this sounds to the case that CE makes for the unobservables. That is, because we cannot observe the unobservables, thus, we cannot claim that it exists. Putting it together, because causality cannot be observed (and thus, is an unobservable for the CE), we cannot ‘know’ it to be true. However, this presents a contradiction.

If we accept CE’s observable/unobservable distinction [3] and the criteria to which we believe things to be true [5], we must then reject the existence of causality. Recall that for CE, we must reject belief in the unobservable entities that are “picked out” by a scientific theory. However, if we reject causality, what then is the mechanism by which our scientific theories explain reality? Constant conjunction? Since the assertion that causality exists and the distinctions made in [3] and [5] cannot be held simultaneously, then by contradiction, CE’s argument from underdetermination must be wrong.




In this section, I will argue that van Fraassen’s constructive empiricism is more nuanced and does not have this same contradiction. A closer reading of The Scientific Image (van Fraassen 1980) causes one to realize that van Fraassen makes a much weaker claim than CE. As Cartwright (2007) explains, “This kind of argument is not available to van Fraassen. For he is not trying to talk the realist into disbelief. He is positively endorsing a specific set of beliefs.”

Recall that the goal here for van Fraassen is not an epistemology, but a view on what the aim of science is. Van Fraassen (1980) is mainly concerned with the criterion of success for science, not the epistemic reasons for believing in the unobservables. Nowhere in his book does van Fraassen claim that it is irrational to be a scientific realist (Dyck 2007). In fact, the argument from underdetermination commonly attributed to van Fraassen is in conflict with his own epistemological beliefs. To quote van Fraassen (1980):

“A complete epistemology must carefully investigate the conditions of rationality for acceptance of conclusions that go beyond one’s evidence. What it cannot provide, I think (and to that extent I am a sceptic), is rationally compelling forces upon these epistemic decisions.”

Here Dyck (2007) finds that van Fraassen is explicitly denying that there can be epistemic rules that force (dis)belief on us. That is, van Fraassen himself does not seem to believe that there can be rules that will compel us to full belief in a theory, or even belief in empirical adequacy. Van Fraassen is not trying to call all scientific realists wrong. All that he is saying is that constructive empiricism is a possible position, and we should not discount it so easily.

How does this remove the contradiction? Remember that van Fraassen (1980) has been stressing that constructive empiricism should be seen as a view of science, not as an epistemological position. That is, it doesn’t tell us what we should or should not believe, but it gives us an answer to the question ‘what is science’ (Dyck 2007)? I claim that in creating the observable/unobservable distinction, he is not compelling realists to become constructive empiricists, but rather to highlight a problem in science: that we seem to naively take as truth many of the unobservable entities that we do not observe. That is, he does not say that we are irrational in believing that unobservable entities exist, rather, we should take note of the fact that such a belief is unwarranted before believing in such entities. I believe that van Fraassen himself would accept the contradiction made in section II, however, he would take it as further proof for why we should have a distinction between what we accept and what we believe. He might even use Hume’s scepticism towards causality as a much stronger example of something that we take for granted to be true, but is ultimately unwarranted.




We cannot deny the realists their beliefs (due to the contradiction), and likewise, we cannot deny the constructive empiricists their disbeliefs (due to the unobservables). Thus, in being a constructive empiricist, all that van Fraassen (2008) is saying is that, since we have reached an impasse on the truth of the unobservables, the main aim of science should be on the observables, that is, empirical adequacy.


[2529 words]


Works Cited

Cartwright, Nancy. “Why Be Hanged for Even a Lamb?” Images of Empiricism, Jan. 2007, pp. 32–45., doi:10.1093/acprof:oso/9780199218844.003.0003.

Dyck, Maarten Van. “Constructive Empiricism and the Argument from Underdetermination.” Images of Empiricism, Jan. 2007, pp. 11–31., doi:10.1093/acprof:oso/9780199218844.003.0002.

Forrest, Peter. “Why Most of Us Should Be Scientific Realists.” Monist, vol. 77, no. 1, 1994, pp. 47–70., doi:10.5840/monist19947712.

Fraassen, Bas C. van. Empirical Stance. Yale University Press, 2008.

Fraassen, Bas C. Van. “Laws and Symmetry.” Feb. 1989, doi:10.1093/0198248601.001.0001.

Fraassen, Bas. C. Van. “The Scientific Image.” Nov. 1980, doi:10.1093/0198244274.001.0001.

Hume, David, and Stephen Buckle. An Enquiry Concerning Human Understanding and Other Writings. Cambridge University Press, 2007.

Hume, David, et al. A Treatise of Human Nature. Clarendon Press, 2009.

Kukla André. Studies in Scientific Realism. Oxford University Press, 1998.

Ladyman, James, et al. “A Defence of Van Fraassen’s Critique of Abductive Inference: Reply to Psillos.” The Philosophical Quarterly, vol. 47, no. 188, 1997, pp. 305–321., doi:10.1111/1467-9213.00061.

Monton, Bradley John. Images of Empiricism: Essays on Science and Stances, with a Reply from Bas Van Fraassen. Oxford University Press, 2008.

Psillos, Stathis. “How Not to Defend Constructive Empiricism: a Rejoinder.” The Philosophical Quarterly, vol. 47, no. 188, 1997, pp. 369–372., doi:10.1111/1467-9213.00066.

Psillos, Stathis. “On Van Fraassen’s Critique of Abductive Reasoning.” The Philosophical Quarterly, vol. 46, no. 182, 1996, p. 31., doi:10.2307/2956303.

Leave a Reply

Your email address will not be published. Required fields are marked *