This post in the eleventh in a series on Science, Technology, and Society. The previous post in the series is here, and the following post is here. All posts in the series have previously appeared on the Partially Examined Life group page on Facebook.
“What we represent to ourselves behind experiences exists only in our understanding.”
Bas van Fraassen (1941 – ) is a Dutch-American philosopher of science at Princeton University, famous for a unique form of scientific agnosticism called constructive empiricism, meant to steer a middle course between constructionism and realism.
The constructionist account of science emphasizes context at the expense of theory, while the realist account takes the opposite approach. For constructionists, a theory is a useful story about experience—that we have one theory and not another is the product of a historically contingent process. Just as three different people who witness the same event can tell three very different stories about it, three different societies can tell three different stories about experience. These stories can be very different, yet all offer an adequate account of the evidence. Therefore theories describe the context in which they were created at least as much as they do anything real about the universe. Because every theory emerged from some context and some dry of values, one cannot be separated from the other. For constructionists, they are invented, not discovered.
Realists, on the other hand, regard scientific theories as actual descriptions of actual reality, exclusive and contradictory to alternative models. Belief in their literal truth is the only possible outcome of a rational inquiry into their nature. Context does not create theories – theories are successful precisely when they transcend their context of creation in order to describe the universe as it really is. If the scientific revolution had occurred in Ming China instead of Bourbon Europe, we would still have the same laws of motion that Newton discovered, only they would not be called Newton’s laws. Context is trivial, explanatory success profound. Theories are discovered, not invented.
In The Scientific Image (1980) and Laws and Symmetry
(1989), Bas van Fraassen offered a different approach. Perhaps surprisingly, he did so by reviving aspects of the positivist program. Philosophy of science should, he held, be regarded as the clarification of relationships between theory and observation, language and logic. A commitment to the utility of a theory does not, on van Fraassen’s account, imply a commitment to its status as truth—only to its correspondence to observation. In other words, a successful theory should be regarded “empirically adequate,” a more practical standard than “literal truth.”
Knowledge stops with the unaided human senses. Objects or processes that cannot be directly observed, but only “detected” with instruments, cannot be empirically known, because the senses, as such, do not and cannot observe them. Rather, the instrument does, and the senses observe what the instrument observes. Inference to the best explanation is valid in the former case but not the latter. For instance, if you hear scratching in the walls, squealing and scurrying at night, and you notice your cheese has gone missing, you might reasonably infer the presence of a mouse. You can know it because, even if you haven’t seen the mouse, you could in principle, and if you wait long enough you probably will. On the other hand, if you run a large hadron collider and detect subatomic debris consistent with the theory that calls for the existence of the Higgs boson, you cannot know that the boson actually exists, because you have not and cannot see it. What you can know is whether or not your theory is “empirically adequate.”
Because many, if not most, of the things with which science has to deal cannot be directly observed, the central question of science is not “What is the truth about nature?” but “what counts as an empirically adequate explanation?” This in turn is largely a question of context. Rather than ask, like the constructionists, “Who created this theory, and for what purpose?” we should ask “Who is going to use this theory, and for what purpose?” In this respect, an explanation is like a tool: A good hammer is not much use for washing clothes, nor a washing machine for driving nails. Similarly, quantum mechanics is not a good explanation for time dilation, nor is relativity for the weak nuclear force.
Philosophically, the payoff for this view is that it allows the philosopher to handle scientific theories without committing himself to their literal truth, which can be undesirable for reasons we’ve already seen. Similarly, it makes room for constructionism and realism within the same framework. The cost is that it treats as uncertain the existence of some things, like the Higgs boson, about which scientists are reasonable certain, and for which we have good evidence. Science and philosophy of science are not entirely reconciled, but neither is van Fraassen engaged in a radical and far-reaching critique like the constructionists.
Daniel Halverson is a graduate student studying the history of Science, Technology, and Society in Nineteenth century Germany. He is also a regular contributor to the PEL Facebook page.
“Knowledge stops with the unaided human senses”
That’s a nice piece of dogma, is there a reason for believing this?
To be precise, the quote should say “Knowledge of the external world stops with the unaided human senses,” as I’m sure all empiricists will agree that we have a priori knowledge of logical and mathematical truths.
Here’s one possible argument for this claim.
1. With respect to the external world, beliefs produced by unaided sense perception enjoy a relative consensus that far outstrips other possible belief forming mechanisms (e.g. ESP, religious experience, Cartesian intuitions, etc.)
2. There is some roughly defined threshold with respect to consensus regarding beliefs about the external. We should doubt whether belief forming processes that don’t meet this threshold actually produce knowledge.
3. Unaided sense perception is the only process that passes this threshold, the others don’t even come close.
4. Therefore, knowledge of the external stops with the unaided senses.
“Unaided sense perception is the only process that passes this threshold”
Why don’t aided sense perception pass this threshold? (I’m thinking of microscopes and the like not ESP)
I think your question points to an inadequacy of describing sense perception as “unaided.” Perhaps a better distinction would be between direct and indirect perception, although even this distinction is not quite accurate.
I think van Fraassen’s skepticism is directed towards theoretical entities, like electrons and black holes. The existence of these entities is inferred through what we take to be observation of their effects. This is what I consider to be “indirect” perception. I have almost zero knowledge of experimental physics, so I have no idea what kind of tools experimental physicists use. Let’s just suppose that they have some tool and it in certain situations, it gives you a number. What are you directly perceiving? You’re just perceiving a number. However, physicists might take these numbers and then conclude the presence of an electron.
This is certainly different from what you might perceive using a microscope, telescope, or a pair of binoculars. These sorts of tools aid in direct perception. I don’t think any empiricist would want to deny that beliefs coming from this sort of aided perception can count as knowledge.
So, perhaps a better way to word the principle is this, “Knowledge of the external world stops with direct sense perception.”
“I think van Fraassen’s skepticism is directed towards theoretical entities”
No, he dislikes microscopes specifically, however you are free to develop Fraassens thoughts into something more sensible.
“This is certainly different”
Sure it’s different, the question is how this difference motivates the claim that knowledge can come from one kind of perception but not from the other.
Take a simple example, the speedometer of a car. It shows the speed of the car. I know the speed of the car by looking at the speedometer.
Now some philosopher of science comes along and say that I don’t really know the speed of the car because I can’t see it directly. The natural reply is that this is just nonsense, I know it as well as if I could see it directly. If all men were born deaf I’m sure the same philosopher would say there were no such thing as sound since it could then only be detected and not actually heard.
I don’t know much about van Fraassen. My only familiarity is through secondary literature. So the position that I am defending might be some idealized via that is not accurately attributable to BVF. I am actually not very sympathetic to this kind of empiricism, so am just playing a sort of devil’s advocate.
Your case regarding the car speedometer doesn’t quite fit the mold. Here’s why. There are other ways of empirically judging the speed of a car. The reading on a car speedometer, is therefore verifiable using these other means. We’re okay with gaining knowledge via the reading on a speedometer roughly for the same reason that we’re okay with gaining knowledge via testimony.
Suppose, instead, that you had a vehicle that is in motion. You only have one instrument that purports to measure the speed of the vehicle, but all that it gives you is a number. You have no way of checking via other empirical means how fast the vehicle is going. In this case, are we confident in claiming that we know the vehicle’s speed?
You give the analogy about being born deaf, Here’s something similar. We (apparently) don’t have a way of sensing immaterial entities, like the God of theism. What should we say about claims that we know that God exists?
You also make an interesting distinction between sense and detection. What exactly is the difference between the two?
Here’s another popular argument for BVF style empiricism. It’s an argument from underdetermination. Underdetermination is what occurs when two or more logically inconsistent theories fit with the available evidence.
1. A belief p that arises via indirect sense perception can be underdetermined.
2. Underdetermination defeats epistemic justification for p.
3. Therefore, justification for belief in p can be defeated.
“We’re okay with gaining knowledge via the reading on a speedometer roughly for the same reason that we’re okay with gaining knowledge via testimony.”
If you are okay with knowledge via testimony, you are no longer an empiricist. (Or atleast the term has become bent to allow for almost any kind of evidence). Then if you wish let’s examine my point of view.
“In this case, are we confident in claiming that we know the vehicle’s speed?”
Sure why not, there are cases analogous to this. You might think that the chance of being in error about the thing measured means there can be no real knowledge, but then there couldn’t be any real knowledge at all since there’s always the possibility of being in error.
“What should we say about claims that we know that God exists?”
I have no idea what to say about it.
“You also make an interesting distinction between sense and detection. What exactly is the difference between the two?”
I contend there is no distinction between them in the case of knowledge, I just used the words to distinguish (for the sake of argument) between what a person does when he sees something, and what a machine does when it sees something.
“A belief p that arises via indirect sense perception can be underdetermined”
Sure it can be, but it need not be. So too can a belief that arises via direct sense perception or via any other means.
I’ll try to weave my points into a more cohesive whole so as to give a sense of perspective to this discussion.
First, a definition of empiricism. This is what I take to be a fundamental principle of any form of empiricism:
Suppose that p is a claim about the external world.
Some epistemic agent x is justified in believing p if and only if it is possible that x’s belief in p is formed via sense perception.
You claim that “If you are okay with knowledge via testimony, you are no longer an empiricist. (Or at least the term has become bent to allow for almost any kind of evidence).” Your reading of empiricism seems unnecessarily restrictive. It is not a claim that any serious empiricist will agree to. You can, of course, insist on this principle, but then it seems to me that you are no longer engaging in productive discussion with most self-identifying empiricists.
Why think that the above principle adequately captures the empiricist position? Based on my limited experience with the literature, the principle does a good job of including what empiricists tend to agree is knowledge and exclude what empiricists tend to take as non-knowledge. So, for instance, just about every empiricists will agree that we can have justified beliefs via memory and testimony. However, the sorts of beliefs that an empiricist will permit are only the ones that could have also been a product of sense perception. So, if someone tells me that it is raining outside, I can justifiably believe this claim (holding constant conditions regarding the reliability of the individual giving the testimony), since I could check for myself whether or not it is raining. However, if someone told me that my dead uncle is present in my bedroom, then I wouldn’t be justified in believing this claim.
The principle excludes the sorts of claims you would expect empiricists to reject. Assuming that supernatural entities are not detectable via sense perception, the empiricist principle gives us the result that belief in such entities is not epistemically justified.
The strength of the principle will depend partly on how you interpret possibility. A wider sense of possibility, such as metaphysical possibility, will make the principle more permissive, whereas a narrower sense, like nomological possibility, will make it more restrictive.
Okay, if you are with me in taking this principle to constitute empiricism, then the next step is to argue for its truth.
I’ve given several arguments above. I won’t rehash them here, but instead show how they are related and also give clarification in response to your comments.
One argument for empiricism is the argument from underdetermination. You reply by stating that sense perception could be underdetermined. Furthermore, you state that indirect perception might, but need not be underdetermined. I’m not exactly sure what to make of your response, so I will try to clarify my point.
The argument depends on the truth of the following claim:
If p is strongly underdetermined, then belief in p is not justified.
There is a distinction between weak and strong underdetermination. Weak underdetermination occurs when there are several logically inconsistent theories that are supported by the evidence, but the inclusion more evidence will exclude all but one theory. Strong underdetermination occurs when no amount of available evidence is sufficient to narrow the field down to one theory.
The argument holds that non-empirical beliefs about the external world are subject to strong underdetermination. This is where my example about God and religion was supposed to provide some illumination. Supposing that religion makes claims about the external world, the empiricist will hold that religious claims face strong underdetermination. Consider the persistent disagreement between major religions as inductive evidence for this kind of underdetermination.
Is strong underdetermination sufficient for epistemic defeat? This is contentious. It depends on how high your standards are for knowledge and epistemic justification.
You stated that sense perception could be subject to underdetermination. Perhaps so. This is where my argument from consensus comes in. It may be the case that claims arising from sense perception are underdetermined, but let’s compare this with claims about the external arising from faculties outside of sense perception. This difference seems quite stark.
Finally, we’ll need to consider how all this applies to the debates regarding theoretical entities in science. We may come to form a belief on the basis of what a machine tells us. Is this belief justified? By putting the above principle to use, the answer seems clearer. Reports from a machine or instrument are basically a form of testimony. We are justified in believing those reports if and only if we could somehow arrive at the same belief via our own sense perception. Without being able to believe such claims via sense perception, such reports are subject to strong underdetermination, which if we accept the above claim, defeats justification.
Empiricists like van Fraassen basically express skepticism regarding certain theoretical entities. They hold that with respect to certain claims in physics, it is not possible for sense perception to verify those claims.