Announcer
The following program features simulated voices generated for educational and philosophical exploration.
Leonard Jones
Good afternoon. I'm Leonard Jones.
Jessica Moss
And I'm Jessica Moss. Welcome to Simulectics Radio.
Leonard Jones
Yesterday we examined quantum mechanics and the measurement problem. Today we turn to a foundational question about scientific reasoning itself. How do we justify inferring universal laws from particular observations? Is scientific induction rationally defensible, or does it rest on assumptions we cannot justify?
Jessica Moss
This is Hume's problem of induction—the question of why we should expect the future to resemble the past, or unobserved cases to resemble observed ones. It strikes at the heart of scientific methodology.
Leonard Jones
Our guest is Dr. Bas van Fraassen, Professor Emeritus of Philosophy at Princeton University and author of The Scientific Image and Laws and Symmetry. He's developed constructive empiricism as an alternative to scientific realism. Welcome, Dr. van Fraassen.
Dr. Bas van Fraassen
Thank you both. I should say at the outset that I think Hume was fundamentally correct—induction cannot be rationally justified in the way many philosophers hope. But this doesn't make science irrational.
Jessica Moss
Let's begin with Hume's argument. Can you reconstruct it for us?
Dr. Bas van Fraassen
Hume asks: what justifies our belief that unobserved cases will resemble observed cases? Any attempt to justify this principle either begs the question by assuming what we're trying to prove, or attempts to derive it from experience—which itself presupposes induction. It's a vicious circle or infinite regress. No deductive argument can bridge the gap from observed to unobserved.
Leonard Jones
Let me be precise about this. You're saying that any inductive inference requires the assumption that nature is uniform—that the future will resemble the past. But we can't justify this uniformity principle without already relying on induction?
Dr. Bas van Fraassen
Exactly. If you try to argue that nature has been uniform in the past, therefore it will be uniform in the future, you're already making an inductive inference. You're assuming the very principle you're trying to establish. The circularity is unavoidable.
Jessica Moss
What about probabilistic approaches? Can't we say that while we can't be certain about inductive inferences, they're probably correct?
Dr. Bas van Fraassen
This faces the same problem. To say an inductive inference is probably correct requires assuming something about the probability that nature is uniform. But how do we know that probability? Any attempt to estimate it from past cases again presupposes induction.
Leonard Jones
Many philosophers have attempted to solve the problem of induction. What about the pragmatic response—that induction works, and that's justification enough?
Dr. Bas van Fraassen
This is Reichenbach's pragmatic justification. He argued that if any method can succeed in predicting the future, induction will. But this argument has problems. First, it assumes we can identify what 'working' means independently of inductive assumptions. Second, it doesn't tell us which inductive method to use—there are infinitely many.
Jessica Moss
Can you explain that last point? I thought induction was a single method.
Dr. Bas van Fraassen
Consider Goodman's new riddle of induction. Define 'grue' as a property things have if they're green before some future time T and blue afterwards. All emeralds we've observed have been green, but they've also all been grue. So why infer that future emeralds will be green rather than grue? Both inferences seem equally justified by past observations.
Leonard Jones
But surely green is a more natural property than grue. The latter is gerrymandered, defined in terms of an arbitrary time.
Dr. Bas van Fraassen
This is the standard response—we should project natural properties, not gerrymandered ones. But what makes a property natural? If we say it's the properties that appear in fundamental laws of nature, we're being circular again. We're trying to justify induction by appealing to laws that we only know through induction.
Jessica Moss
This is deeply unsettling. If induction can't be justified, and all scientific reasoning depends on induction, doesn't this undermine science entirely?
Dr. Bas van Fraassen
I don't think so, and this is where constructive empiricism comes in. I argue that science doesn't require justified induction to be rational. Science aims not at true theories about unobservable reality, but at empirically adequate theories—theories that correctly predict observable phenomena.
Leonard Jones
You're drawing a distinction between accepting a theory as true and accepting it as empirically adequate. Can you elaborate?
Dr. Bas van Fraassen
When we accept a theory, we believe it's empirically adequate—that what it says about observable things and events is true. But we need not believe its claims about unobservable entities and processes. We can remain agnostic about electrons and quantum fields while accepting that theories involving them make accurate predictions.
Jessica Moss
But this seems to avoid rather than solve the problem. Don't we still need induction to infer that a theory empirically adequate so far will remain empirically adequate in the future?
Dr. Bas van Fraassen
Yes, and I don't claim to solve Hume's problem. I accept it. My point is that we can be rational in science without solving it. We can accept theories tentatively, recognizing that acceptance is a pragmatic stance rather than a claim to have justified true belief about nature's uniformity.
Leonard Jones
Let me probe this distinction between rationality and justification. You seem to be saying that induction can be rational even though it can't be justified. How can this be coherent?
Dr. Bas van Fraassen
I think we need to distinguish epistemic rationality—which concerns what we're justified in believing—from practical rationality—which concerns what we're reasonable to do or accept given our goals. Science can be practically rational without providing epistemic justification for its inductive inferences.
Jessica Moss
This sounds like you're giving up on the idea that science gives us knowledge about the world. Isn't that too high a price to pay?
Dr. Bas van Fraassen
I think science gives us knowledge about observable phenomena. What I'm skeptical about is whether we have knowledge of unobservable entities and processes, or knowledge that nature's patterns will continue. But this skepticism doesn't prevent us from using scientific theories effectively or from making progress in prediction and technological application.
Leonard Jones
But many scientists would say their theories do give us knowledge about unobservables. The electron is as real as anything we can see. How do you respond to scientific realists who make this argument?
Dr. Bas van Fraassen
Realists often appeal to inference to the best explanation—the idea that we should believe theories that best explain observed phenomena. But this just pushes the problem back. Why think the best explanation is likely to be true? Any argument for this principle seems to presuppose induction or something equally problematic.
Jessica Moss
What about the success of science? Doesn't the fact that scientific theories work so well give us reason to think they're approximately true?
Dr. Bas van Fraassen
This is the no-miracles argument—it would be miraculous if theories made accurate predictions unless they were approximately true. But I think this underestimates how many failed theories there have been, and how empirical adequacy can be achieved without truth. Evolution provides an analogy—organisms are adapted to their environment not because they have true beliefs about it, but through selection.
Leonard Jones
You're suggesting that scientific theories are selected for empirical adequacy rather than truth? That's an interesting analogy, but doesn't it presuppose that there are stable patterns in nature to which theories must conform?
Dr. Bas van Fraassen
Yes, and I don't deny there are regularities in our observations. What I deny is that we can justify the belief that these regularities reflect universal laws that will hold for unobserved cases. We can observe patterns without being justified in projecting them indefinitely into the future.
Jessica Moss
This seems to lead to a kind of skepticism. If we can't justify induction, how do we make any decisions about the future? Don't we rely on inductive reasoning constantly in practical life?
Dr. Bas van Fraassen
We do rely on it, and I'm not saying we shouldn't. I'm making a philosophical point about justification, not a practical recommendation to abandon induction. We can recognize that our inductive practices lack ultimate justification while continuing to use them because we have no alternative and they've proven useful.
Leonard Jones
But this seems unsatisfying. You're saying we should continue doing something we acknowledge we can't rationally justify. Isn't there something incoherent about that?
Dr. Bas van Fraassen
Only if you think rationality requires justification all the way down. I think that's an unrealistic standard. At some point, our practices rest on assumptions we can't further justify. This doesn't make them irrational—it makes them basic to how we navigate the world.
Jessica Moss
Let me ask about the implications for scientific practice. If constructive empiricism is correct, should scientists change how they work? Should they stop making claims about unobservable entities?
Dr. Bas van Fraassen
No, I think scientists should continue as they are. Constructive empiricism is a philosophical interpretation of science, not a proposal to reform scientific practice. Scientists can legitimately talk about electrons and fields as tools for organizing and predicting observations, even if we remain agnostic about their ultimate reality.
Leonard Jones
But doesn't this create a tension? If scientists believe their theories describe reality, and you say we should only accept them as empirically adequate, aren't you disagreeing with scientific practice?
Dr. Bas van Fraassen
I think many scientists are actually constructive empiricists in practice, even if they use realist language. When push comes to shove, what matters is whether theories make accurate predictions, not whether they reveal underlying reality. The instrumentalist attitude is widespread in working science.
Jessica Moss
What about cases where scientific theories make surprising predictions that turn out to be correct—like the discovery of Neptune from perturbations in Uranus's orbit? Doesn't that suggest the theories are revealing real underlying causes?
Dr. Bas van Fraassen
These cases are impressive, but they can be explained by empirical adequacy without requiring truth about unobservables. The theory was adequate for predicting where a massive object should be to explain the perturbations. That it turned out to be a planet is consistent with empirical adequacy—we observed what the theory predicted we would observe.
Leonard Jones
I want to return to the problem of induction itself. You say it can't be solved, but haven't there been attempts to provide probabilistic solutions through Bayesian confirmation theory?
Dr. Bas van Fraassen
Bayesian approaches face their own problems. They require assigning prior probabilities to hypotheses, but these priors can't be justified without begging the question. Different prior probability assignments can lead to radically different conclusions from the same evidence. The choice of priors is ultimately subjective.
Jessica Moss
So you're skeptical of formal confirmation theory as a solution to induction?
Dr. Bas van Fraassen
I think it's useful as a framework for thinking about evidence and belief revision, but it doesn't solve Hume's problem. It just formalizes our inductive practices—it doesn't justify them. The probability calculus tells us how to update beliefs given evidence, but not why we should expect evidence-based beliefs to track truth.
Leonard Jones
Let me ask a final question about the relationship between your constructive empiricism and realism. Is there any evidence that could persuade you toward realism?
Dr. Bas van Fraassen
I think the debate is less about evidence and more about philosophical commitments regarding what science aims to achieve and what counts as rational belief. Realists and empiricists can agree on all the empirical facts while disagreeing about what attitude to take toward theoretical claims. It's a philosophical difference, not an empirical one.
Jessica Moss
That's fascinating—you're saying the debate about scientific realism is itself not a scientific question but a philosophical one about the aims and norms of inquiry.
Dr. Bas van Fraassen
Exactly. And I think recognizing this helps us see that both positions can be coherent. We can have productive science whether we adopt realist or empiricist interpretations. The interpretation doesn't change the predictions or the evidence.
Leonard Jones
Though it might change what we think science achieves and how we understand its relationship to truth.
Dr. Bas van Fraassen
Indeed. And those are questions worth continued philosophical investigation, even if they don't admit of empirical resolution.
Jessica Moss
Dr. van Fraassen, thank you for this rigorous examination of induction and scientific inference.
Dr. Bas van Fraassen
Thank you. These foundational questions deserve ongoing attention from both philosophers and scientists.
Leonard Jones
We'll return tomorrow with more philosophical inquiry.
Jessica Moss
Good afternoon.