Over at Victor Reppert's blog, I've been confronting a form of the argument from reason (AfR). The AfR takes many forms, but the basic scheme is to say that there's something about the process of inference that makes it impossible to explain with any naturalistic model.
In this particular case, the AfR argues that inference is invariant across all possible universes, and that means that inference is physics independent. The argument then says that a thing that is physics independent is non-physical, and, therefore, no naturalistic theory can explain that thing.
I find it easy to see that this argument fails because I can cite counterexamples. The best one I have found so far is the concept of a plasma. A plasma is a high-density, high-temperature state wherein bound states in a medium lose their identity, and the medium becomes a soup of bound-state components and force-carriers. An ionic plasma is an ionized gas, of the type you find in a neon light. In an ionic plasma, atoms become unbound, and you get a soup of electrons, ions, and photons.
However, long after we learned of ionic plasmas, we discovered that there might be quark-gluon plasmas (QGP). In a QGP, we have a soup of quarks and gluons instead of bound nuclei. The physics of QGP, known as quantum chromodynamics, is completely different from the physics of ionic plasmas, which are electromagnetic.
This means that plasma is an abstraction that is physics independent. Indeed, we can imagine very different universes that would support plasmas. Yet, if the ability of an abstraction to apply across universes with different physics were a sign that the abstraction were non-physical, we would have to conclude that plasmas were non-physical. Clearly, plasmas are physical phenomena, and so the premise about the physical portability of abstractions is false.
Here is a systematic dismemberment of the argument. The argument goes like this:
1) We can abstract the possibility of other lawful universes.
2) In every abstraction of a lawful universe, there is a different set of laws.
3) However, every such universe contains common set of laws that establish non-contradiction. (Without these, the laws would predict an outcome and its negation, and not be laws at all.)
4) The conditions under which one classifies an inference as valid are common across every universe. In other words, inference is a specification that is independent of all the laws in the universe except the common logical laws.
5) Premise: Assume that minds in our universe are the result of physical processes of our universe.
6) Any physical system that meets a non-physical criterion must be portable across universes without changes.
7) Minds in our universe would cease to function if physical laws were changed. That is, changing physical laws of our universe without changing anything else in our universe (e.g., the structure of brains), would result in systems that no longer meet the specification for a mind that makes correct inferences.
8) Physical minds are not portable across universes without breaking the criterion or requiring redesign.
9) Therefore, due to (6), assumption (5) is incorrect.
This argument fails because (6) is false.
(6) is false because we can imagine more counterexamples like the following. A "hammer" is an example of an abstract, non-physical criterion that can apply to matter. A hammer is something that amplifies hardness and pressure with kinetic energy. Hammers can exist in a great many universes. Consider a universe slightly different from our own, one in which Iron is a liquid or a gas. An Iron hammer in this alternate universe would no longer meet the abstract criterion for being classified as a hammer. However, in the alternate universe, we could make a hammer with a brass head that meets the abstract criteria. Clearly, physical specifics do not have to be portable across universes for there to be physical implementations in other universes. There will certainly be universes that cannot support hammers, but this is not important.
There's another confusion buried in the argument. What does it mean when we say that correct inference is fixed in all lawful universes? Presumably, these other universes do not actually exist, so we are not referring to physical universes. Rather, we are referring to abstractions. We are saying that if we create an abstraction for another universe, that abstraction is subject to rules of inference. Hence, the very portability of inference is not across physical systems, but across abstractions of physical systems. That is, a mind needs to be able to port inferences across its own abstractions. Here's the experiment. If I create a physical implementation of a mind, and that mind creates abstractions for other universes, then that mind must be able to apply rules of inference to its abstractions of other universes. That's the burden a physical mind has to meet.