"Mary's room" tells us nothing

I have a low tolerance for philosophical thought experiments that posit strange situations that nobody has actually encountered -- "experience machines," violinists hooked up to innocent people's kidneys, planets where water's chemical formula is XYZ, etc. I think they make a fundamental mistake about what intuition is. Intuition is a practical skill. Therefore it is rooted in actual experience of encountering decisions to be made in real life. Intuition has some weight in those situations. So I have no in-principle objection to philosophical arguments that demonstrate that one's beliefs about one situation must be wrong because they imply a counterintuitive result in some other familiar situation. But there seems to me to be no reason to think that an intuition about an unfamiliar case should carry any weight at all.

Consider the classic "Mary's room" thought experiment. The point of this scenario is supposed to be to demonstrate that subjective experiences ("qualia") exist in addition to the physical facts about the universe. Here's how Ian Pollock summarizes it:

Mary is a genius neuroscientist. She knows everything there is to know empirically about how the mind and brain work, and in particular she exhaustively knows every single scientific fact that bears on color vision: optics, how the retina works, how the retina’s signals get processed in the brain, et cetera.

However, Mary has had an unusual upbringing. Raised by philosophers (a sketchy proposition in the best of times), she has been kept in a monochrome room for her entire life and has never seen primary colors — never seen a red rose, or a green leaf, or a blue sky. (Nitpicky readers will notice that this would be actually pretty hard to accomplish — for one thing, she would have to have her skin painted monochrome, never look at a bright light through her closed eyelids and see the red color that results, never dream in color... but never mind. This! Is!! PHILOSOPHY!!!)

Eventually, her philosophical zookeepers take a day off from heroically pushing obese people off bridges, and they take her out of her monochrome cage. She comes out into the world and sees a dandelion for the first time, saying “Wow, so that’s what yellow looks like! Cool!” You can imagine all her other epiphanies for yourself.

The gotcha being that she appears to have learned something new — what it’s like to see yellow — on being released. But we stated that she already knew all the scientific facts! Therefore, there must be facts out there that are beyond the reach of science, even in principle — mental facts, or subjective facts, perhaps. Facts about what it is like to have certain experiences. That’s the idea, anyway.

Here's the problem: To decide whether Mary learned anything new upon first seeing the color yellow, we have to try to imagine knowing everything about color without having seen it. But that's not something any of us have any experience with. So our intuitions are going to be dominated by imagining what it's like to know quite a lot about something. But in the latter case everyone would agree that direct experience teaches you something new. Intuitions about "Mary's room" are useless in demonstrating whether qualia are non-physical facts. At best, this thought experiment can just illustrate the opposing positions. And perhaps some enterprising scientist with no respect for the IRB could use it to design an empirical test of the issue. But probing the thought experiment conceptually does nothing because we haven't had the right kind of experiences to callibrate our intuitions with.

(As a side note, my heavily Pragmatist outlook leads me to doubt the conceptual coherence of the idea of complete scientific knowledge of some topic, of the sort Mary is hypothesized to have. Scientific knowledge is a matter of approximate yet useful models. The idea of a complete cataloging of every fact strikes me as a sort of naive "mirror theory" of truth.)

"Partisan bias" is quite rational

A new poll suggests that voters' views on immigration shift when party labels are attached or removed. Specifically, when read excerpts from the Democratic and Republican party platforms on immigration, rural swing state voters preferred the GOP's position by a margin of 50%-39%. But when read the platforms without being told which party's is which, these voters preferred the Democratic position, 40%-49%. This result is being billed as an example of partisan bias -- voters really like Democratic policies but they have an irrational loyalty to the Republicans anyway.

The first thing to note is that the size of the shift isn't all that huge. Ten percentage points is politically significant, especially since in a closely divided electorate that shift switches which position has greater support. But it still means that the vast majority of voters would not change their position as a result of hearing party labels.

I also don't think reacting to party labels is necessarily irrational. The platforms are both made up of vague boilerplate. Those of us who follow the immigration debate closely can recognize the shibboleths and infer which specific policy measures they connect to -- but the average poll respondent doesn't have that depth of knowledge. Voters don't formulate preferred policy positions on every issue and then pick the party that will get closest to those positions. Instead, they figure out which party, or which specific leaders, they can trust. A trustworthy party or leader is one who 1) appears knowledgeable and competent with respect to the issue, and 2) appears to care about people like you. If you've come to trust the Republican party to have your interests at heart -- because they share your broader values -- then hearing that a bit of vague boilerplate is Republican vague boilerplate, that's a signal that you can expect that policy to work out in the "right" way when it comes to the specifics of implementation. That's a much easier and more reliable strategy than trying to infer what "comprehensive immigration reform" might actually be.

Libertarians should give their property to Native Americans

Many libertarians subscribe to a theory of property ownership (derived from John Locke and Robert Nozick) that says you have a right to any property that you acquired either a) by making use of something that was previously unowned, or b) by getting it fairly from someone who had a legitimate right to it. This is a useful theory in rejecting redistributive taxation. But it raises a troublesome question for any libertarian living in the US, since it seems to imply that we must immediately return all land stolen from the Native Americans, which is pretty much the entire country. Here's a passage from noted libertarian intellectual Murray Rothbard, which seems to very straightforwardly imply the need to turn over Native land:

Suppose, for example, that Jones possesses a watch, and that we can clearly show that Jones's title is originally criminal, either because (1) his ancestor stole it, or (2) because he or his ancestor purchased it from a thief (whether wittingly or unwittingly is immaterial here). Now, if we can identify and find the victim or his heir, then it is clear that Jones's title to the watch is totally invalid, and that it must promptly revert to its true and legitimate owner. Thus, if Jones inherited or purchased the watch from a man who stole it from Smith, and if Smith or the heir to his estate can be found, then the title to the watch properly reverts immediately back to Smith or his descendants, without compensation to the existing possessor of the criminally derived "title." Thus, if a current title to property is criminal in origin, and the victim or his heir can be found, then the title should immediately revert to the latter.

I found this passage interesting because it is quoted by contemporary libertarian Bryan Caplan as part of an argument for why libertarianism does not require returning Native land. Caplan reasons:

In the extremely unlikely event that any particular Indian can show that he personally is the rightful heir of a particular Indian who was wrongfully dispossessed of a particular piece of property, the current occupants should hand him the keys to his birthright and vacate the premises. Otherwise the current occupants have the morally strongest claim to their property, and the status quo should continue.

Here, Caplan is introducing a new requirement: individual ownership. It is indeed unlikely that we'll often be able to track down the individual heir of the individual Native from whom a piece of land was stolen, because historically most tribes did not practice individual ownership of land. Any land stolen prior to the Dawes Act (and some acquired afterward, since not all tribes recognized the legitimacy of the forced division of tribal land into individually-owned parcels) was the collective property of the tribe. That means its legitimate heirs are all present-day members of the tribe who owned it at the time of its theft.

I don't know exactly where Caplan's house is, but since he works at George Mason University it's plausible he lives on land stolen from the Patawomeck tribe, which is alive and well. Or perhaps his house is on Doeg land. As far as I can tell from a little Google and Wikipedia work, there are no surviving Doegs. But that genocide doesn't necessarily mean Caplan can breathe easy -- the Doegs were allied with several neighboring related tribes, so there is a plausible argument that their allies, some of whom survive today, are their proper heirs.

Caplan could reply that collective ownership is illegitimate, and only individual ownership is recognized by libertarian philosophy. Certainly this sort of terra nullius doctrine has been popular in the past. But it has no basis in the Locke/Nozick theory of property. The root of that theory is the freedom of individuals to acquire, use, and dispose of property as they see fit, as long as it doesn't violate someone else's rights. Thus, people have the freedom to create collectivities such as tribes and grant those collectivities property-ownership rights (including the right to refuse to transfer collective property to any individual).

I am not a libertarian, and thus I am not a proponent of a strict Locke/Nozick theory of property. I think Nozickian property systems are justified only insofar as they contribute to social well-being, and that property can legitimately be taken for a variety of socially beneficial uses (including redistribution, and reparations for past injustices). But it seems clear that a consistent Locke/Nozick libertarian ought to demand the return of all land to the descendants of its pre-European owners except in the small number of cases where a) the relevant tribe, and all affiliated tribes that might make a claim to heirship, have been completely destroyed, or b) the Native owners made a legitimate and fair transfer of property to a non-Native (as many tribes did have traditions of welcoming strangers and immigrants).