Laudan Critique of Underdetermination

Laudan uses "underdetermination" as short for the thesis that a theoretical explanation is undeterdetermined by its data (as if there clearly were a set of data to which a theoretical explanation applies) in the sense that other explanations might also "work."

He assumes that it is important goal of science to produce theories that can be used to explain their data.

An underdetermination thesis can be:

1. deductive or ampliative
2. egalitarian or nonunique
3. compatible, entailing, explanatory, or empirically supported

---

A theory could be deductively determined to be true by the data. Example: Some whole number is greater than 2 and less then 4. The theory is that the number is 3.

Of course, that never happens in scienc A theory could be ampliatively (that is a strange choice of words that just means nondeductively) determined by the data. Example: I am in room 311 Social Sciences is determined by, say, my memory, that fact that the right class is here, by the number on the door.

---

If a theory is not determinately given by the data, it could be the case that ("egalitarian") all theories are equally compatible with the data, or it could be that there is more than one possible theory given by the data ("nonunique").

---

• A theory could be compatible with the data—that is, the data don't rule it out. Example: Everything that happens is caused by sprites who make it look like there are scientific laws.
• A theory could entail the data, that is, it could be that the data are inevitable according to the theory. Example: the fact that when you decrease the volume that a gas occupies, the pressure goes up in accordance with Boyle's law is entailed by the theory that there are gas sprites who cause the gas to obey Boyle's law.
• A theory could explain the data. For example, the Newtonian theory of gasses explains Boyle's law.
• A theory could be empirically supported. That is, for example, we might have some reason to believe that the Newtonian theory of gasses is the right explanation of Boyle's law. (van der Waals)

---

Now, what kinds of underdetermination do we actually have some reason to believe?

Certainly, Humean underdetermination: deductive, nonunique, compatibilist. That is shown, for example, by the fact that for any theory, the theory that that theory applies everywhere except possibly to the data being considered is compatible with the data and there is no deductive reason to reject that theory. There are many such theories, and so we have nonuniqueness.

Quine in fact argues for at least one case of a stronger kind of underdetermination: Lots of empiricists (those who think all knowledge is based on experience) have tried for centuries to see how we could come to know that there are physical objects on the basis of our sense impressions. The consensus now is that it can't be done.

If our data is the data of the senses, that data does not ampliatively determine that the theory that there are physical objects is true. The data are compatible with the alternative theory that there are no physical objects.

ampliative, nonunique, compatibilist underdetermination.

---

A portion of what Laudan must believe to say what he does:

Now, what consequences have been take to follow from underdetermination? The bad consequence is that what theories we believe are not constrained by data, we just make them up freely with no empirical constraints.

Surely, we want extract as much as we can from our data, and so we will require that our theories be consistent with what we can find ampliatively from our data. In addition, we prefer theories that are empirically supported, then explanatory, then entailing. That means that the bad consequence would only follow from ampliative, nonunique, empirically supported underdetermination.

Laudan says that Quine has produced no argument for such underdetermination. He concludes that Quine is not entitled to draw the bad consequence.

That is all true. However, it is not relevant to what Quine actually claims.

In Two Dogmas, what Quine actually says is

No particular experiences are linked with any particular statements ... . 72

Any statement can be held true come what may, ...72

... it is misleading to speak of the empirical content of an individual statement. 72

Quine is not criticizing any of the views Laudan mentions. He is criticizing a part of the received view, its semantic component. His target is not how we choose theories, it is what statements mean.

The logical positivists, at least in the received view, held that every statement just claims correlations between or the existence of certain observations.

If that were correct, we'd be able to directly use observations to assess particular laws and claims. We can't, as Duhem noted for physics and Quine noted for life. There is no way to directly assess particular laws and claims, rather, our theories as a whole confront our experience as a whole. That does not say, and was never intended to say that our theories do not confront our experience.

So, why does Quine say that "any statement can be held true come what may?" What does Quine mean by can? He does not mean that he thinks it would be reasonable to hold a statement true come what may, which seems to be what Laudan takes him to mean. If a statement were connected by definition with certain things that might come, then it would be incoherent to hold the statment if they came. Example: If "it is 80 degrees" meant "it appears that the red line is at the 80 mark," then it would be impossible ... .

According to the received view logic is just a matter of conventions we have adopted. There could therefore, as a matter of principle, be no empirical reason to consider changing our logic. To hold that, proponents of the received view must hold that there is a distinct analytic-synthetic (conventional-empirical) distinction. Quine denies that, and hence denies that there is a fact of the matter about which statements are empirical and which are definitions. Quine argues that that is absolutely true, and hence that there can be no such thing as a purely conventional truth. Since logical truths seem to be a good example of truths that are not empirical, he needs to argue that empirical considerations could even lead us to revise logic.

---

How could Laudan be so wrong?

The work of Quine and others led to the demise of the received view. When Quine wrote Two Dogmas, the received view, even though it may have been the case that no one really believed it, was the standard view, and everyone understood without being told, that it was Quine's target. If you read what Quine says without that context, you misread him.

-- ShaughanLavine - 31 Jan 2006