[T]he very possibility of belief presupposes the existence of cognitive faculties with the intrinsic purpose of seeking truth. Similarly, the very possibility of intentions or purposes presupposes the existence of cognitive faculties with the intrinsic purpose of supplying deliberation with goals. Human will cannot be the source of all purpose (function, value, etc.), since brain processes can constitute the human will only by having the appropriate intrinsic purposes (and functions).
Clayton Littlejohn is unconvinced:
It seems that Koons is saying that I couldn't have a mind unless I had parts that had purposes. Isn't it a mistake to say that I have the purposes I do because my brain or cognitive processes has those very purposes? To say that I have the purpose of discovering truth because part of my brain does seems to make a number of mistakes. What's odd is that he goes beyond this in suggesting that I, CL, have the purposes I do because CL's cognitive processes have purposes AND (worse) that they aren't the same as my purposes. Odd, no?
I think that misunderstands Koons' argument, though. It isn't a simple fallacy of division -- Koons isn't assuming that a property of the whole must be present in its parts, or anything like that. Instead, as I understand it, his starting point is the assumption that cognition is inherently normative. One commonly hears this view expressed through the slogan: "belief aims at truth" -- the idea being that this 'aim'/purpose is partly constitutive of belief. I think this is false (on any strong interpretation of the claim), but at least it's not as odd as would be an argument from division where the property transmitted from the whole to the part isn't even the same!
Anyway, returning to the central question: do homunculi have objective purposes? It depends how much we build into the concept of "purpose". There's a very weak sense of the claim that's unobjectionable. We can talk of biological "functions" which mean nothing more than "naturally selected operation", for example (as Brandon points out in comments to Clayton's post). We can describe certain functional roles -- e.g. information acquisition -- that could aid an organism's survival, and no doubt (some of) our cognitive processes evolved precisely because they fill such a role. (To clarify: the initial appearance of a trait is mere chance; what's more predictable is that a functionally apt trait is retained and promoted in the population through natural selection.) But there's nothing metaphysically deep about such talk; such "purposes" invoke nothing over and above the historical facts.
[Besides, I don't think those historical facts are essential to cognition in any case. If a freak quantum accident created an atom-for-atom replica (cf. "Swampman") of me, he would certainly have a mind just as I do, and arguably with all the same beliefs.** But any ascription of "objective purposes" to his accidentally assembled parts would presumably be our mere projection. Unless, I suppose, we individuate purposes the same way (I propose) we do mental states, by forward-looking rather than historical criteria. But that seems ad hoc, and - more importantly - still wouldn't add any metaphysical depth to the claims.]
Koons' view would appear to imply that there could be people that were just like us in every descriptive respect (incl. physics + qualia), and yet they somehow "don't have minds" due to lacking the essential normative components. They have (non-mental?) states that are exactly like beliefs in every respect, except that they don't intrinsically "aim at truth", so they're not "beliefs". Similarly, they form fake "intentions", arising from cognitive processes that happen in fact to "supply deliberation with goals", but because this is not their "intrinsic purpose", none of it really counts.
Now that does seem odd.
(Perhaps one could hold that objective purposes supervene on the descriptive facts, just like morality does, so that it's not really possible to subtract it whilst holding all descriptive facts the same? But then the descriptive facts suffice to establish mentality after all: the supervening normative properties come along for free!)
* = A fun aside: the GFP quote a delightful passage from Michael Freyn which draws on the metaphor of homunculi to motivate skepticism about free will:
I discover that I am not an absolute ruler after all. I am a mere constitutional fiction, a face on the postage stamps, a signature at the bottom of decrees written by unidentified powers behind the throne over which I have no control... even my private entertainments are devised for me by invisible courtiers working in parts of the palace that I have never entered, and could never find my way to go.
** = Bracketing issues about semantic externalism -- "narrow content" will do for present purposes -- I think it's most plausible to hold that beliefs are functionally reducible (perhaps modulo qualia, if you hold that beliefs have an essential phenomenal "feel" to them), e.g. to facts about one's present dispositions and such.