Option 1: Different concepts, same phenomenology
That is, she has the same phenomenology as Mary, but doesn’t have the same cognitive contents (it would be like a duplicate of me thinking ‘Where is mom?’, but the content of ‘mom’ is not fixed in Swamp Eric, or the same as in Eric). When she thinks ‘Apples are red’ she hasn’t the same thought as Mary, but she has the same phenomenology. I think this is a fairly common view.

Option 2: Same concepts, same phenomenology
This is the internalist option. It seems to fly in the face of Twin Earth and such that convinced me that (for some types of concepts) externalism is right. But it is interesting for me that concepts about qualia or experience may not have the same type of externalist story (though this is very complicated–perhaps qualia content fixation is via a causal chain where the ‘outside’ thing that does the fixing is not outside the brain, but outside the system doing the conceptualizing of qualia–e.g., connections to a perceptual sytem or whatever).

Option 3: Different concepts, different phenomenology
This is externalist all the way. Dretske would go for this option. Many don’t like it, but as he says why should we trust our intuitions about phenomenology when nobody has a really good theory.

Option 4: Same concepts, different phenomenology
I don’t think anyone holds this. That would suggest they buy historicist accounts of phenomenal content but not conceptual content, and that would be just odd.

Options 2 and 4 are probably least common, at least for concepts about individuals. For concepts about experiences, things get really weird. I frankly am not sure what I think…

And I hadn’t even thought of Chase’s epistemic considerations, but I think the semantic considerations are enough to keep my mind boggled at the present.

Regarding the instability of memory sort of thing, I think that’s going to be appealing to a lot of gappers. Jesse Prinz, for example, comes to mind. But I think it ultimately doesn’t fly. What the right view of phenomenal memory is is something I tried to convey in some earlier posts related to a slightly different project. See:

http://www.petemandik.com/blog/2009/03/30/raffman’s-rainbow-unraveled/

http://www.petemandik.com/blog/2009/04/01/over-the-rainbow/

and

http://www.petemandik.com/blog/2009/04/06/postscript-on-diachronic-discrimination-failure/

If I understand the point you want to press, you suspect gappy physicalists can’t cite a historical necessary condition on knowledge that pre-release Mary fails but Swamp Mary passes. The sort of causal theory of knowledge you offer as a specimen is meant to illustrate why that’s so. Of course, the theory you mention is hopelessly implausible, because it fails to rule out all manner of deviant causal chains. I wonder if a more plausible story will have the same problem.

Here’s a story a gappist might try to tell. Let’s be ahistorical about contents, so Swamp Mary has the right concepts to count as believing that it is like X to see red. If I’m following you, that will mean that prerelease Mary also has the right concepts, because she knows all the physical facts and we are presupposing physicalism.

1. I actually don’t think semantic ahistoricism + physicalism implies that prerelease Mary has the required phenomenal concepts. That’s because I don’t think knowing all the physical facts requires having all the concepts in terms of which the physical facts could be expressed. I might know all the physical facts there are to know about death and who died without having the true belief that Abe gave up the ghost on April 15, because I might not have the ‘giving up the ghost’ concept. But let’s bracket that point.

2. We’ll assume that prerelease Mary, Swamp Mary, and postrelease Mary all believe that seeing red is like X. A gappist needs to explain why only postrelease Mary’s belief qualifies as knowledge. Swamp Mary’s belief isn’t knowledge because it has no history at all, and there historical conditions on knowing. So now we need to find a historical condition on knowing that prerelease Mary fails but postrelease Mary passes.

To do that, we need to know more details about how Mary’s brain works. But here’s one way it might work out. Postrelease Mary’s beliefs about what it is like to see read are sustained partly by processes that keep them calibrated to the experiences she has when she sees red, as well as the usual inferential and memorial mechanisms. Prerelease Mary’s beliefs are sustained only by those inferential and memorial mechanisms.

Now let’s make one more assumption. Let’s assume phenomenal concepts are unstable in memory. For example, if Mary has a belief with the syntax ‘X-is-what-it’s-like-to-see-red’, the X concept might fade or gradually shift in its content. It might go from referring to what it’s like to see red to referring to what it’s like to see dark orange, and so the belief-token might stop being true. It could turn out, then, that memory alone is not a very reliable mechanism for sustaining phenomenal beliefs, unless it is supplemented with the sort of calibrating processes that prevent the degradation of phenomenal concepts. Those processes, though, take actual experiences as inputs. They can’t calibrate your X concept to red experiences unless you actually have some red experiences.

If Mary’s brain works in something like this way, it will turn out that prerelease Mary’s phenomenal beliefs are not sustained by suitably reliable processes, although postrelease Mary’s phenomenal beliefs are. The former are not knowledge, and the latter are knowledge. Voila, an epistemic gap of sorts!

All that said, it’s a contingent epistemic gap. I don’t know if gappy physicalists mean for the gap to be contingent or necessary. If they mean for it to be necessary, then this is no comfort to them. For my part, it doesn’t much matter, because I think it’s more interesting to try to explain why there seems to be an epistemic gap than trying to explain why there is on.

My discussion of historicism so far has focused on the semantic requirements on knowledge: S knows P only if has a belief or other state that represents P. Might the gappy-physicalist deny the possibility of Swamp Mary along historicist lines that are distinctively epistemic? Such line of thought would grant that Swamp Mary has a state that represents the phenomenal facts, but that the state fails some other criterion for knowledge (e.g. justification, non-Gettierizability), a criterion arguably involving factors that supervene on historical relations. For such a line to be promising, it will need to both (1) make a plausible case for denial of Swamp Mary’s phenomenal knowledge and (2) be consistent with the denial of pre-release Mary’s phenomenal knowledge.

Re: (1), one difficulty that arises is that it seems quite compelling to say that Swamp Mary’s true belief about the phenomenal facts suffices for phenomenal knowledge. The following remark by Alter (1998. A limited defense of the knowledge argument. Phil. Stud.) seems relevant to the case at hand.

[L]et us ask what exactly [pre-release Mary’s] lack of factual knowledge consists in. We color-sighted folk in the outside world are supposed to know facts that she does not, but what distinguishes her epistemic state from ours? The difference does not seem to turn on justification. That is, her problem is not that she has the same beliefs as we, but her beliefs, unlike ours, are unjustified – as though she suspects that seeing red has a certain distinctive phenomenal quality, the same one we know it to have, but she cannot confirm her suspicion. Rather, if she lacks knowledge of facts about color experiences, this would seem to be because she lacks the appropriate beliefs: certain propositions, which we grasp, are inaccessible to her. (p. 46)

Alter’s claim here is that pre-release Mary’s lack is due to a lack of access to certain propositions. While, of course, it doesn’t follow that gaining access to such propositions thereby suffices for phenomenal knowledge, such a claim does have a high degree of plausibility. It would be compelling, I should think, to the gappy physicialist to hold that post-release Mary’s gain of psychosemantic access to the phenomenal facts suffices for phenomenal knowledge.
There may, however, be gappy physicalists interested in pursuing the case that post-release Mary’s phenomenal knowledge is not grounded simply in her satisfaction of semantic criteria, and that there is an additional factor true of her that, lacking in Swamp Mary, makes Swamp Mary’s true representation of the phenomenal facts either unjustified or a Gettier case. I cannot here survey all the possible ways of developing such a line, but will need to briefly sketch one representative sample and what I think is problematic about it. Take for example a very simple form of reliabilism that holds that a true representation of P is justified or un-Gettierizable iff it is caused by it being the case that P. Clearly, such an account entails Swamp Mary’s phenomenal ignorance. So, this form of reliabilism will get (1). However, what’s to guarantee pre-release Mary’s phenomenal ignorance, and thus, (2)? Since the current dialectic between gappy and non-gappy physicalists presupposes physicalism, there’s no fact that physically omniscient pre-release Mary fails to have a representation of. And though her access to the facts has been mediated by black and white media, there’s no bar to saying that her true representations of the phenomenal facts are caused by the facts they represent. As a joint defense of (1) and (2), the simple form of reliabilism here scouted seems unpromising.

Re: nomological psychosemantics and the factivity argument, someone pursuing such a line needs to give an account of why it is that there’s a cognitive/epistemic asymmetry (stated in terms of recognitional capacities or whatever) between Swamp Mary and pre-release Mary. It begs the question against those (like me and Dennett) who argue that there is no asymmetry to say that it’s obvious that there is. We deny that there’s anything Swamp Mary is disposed to recognize that pre-release Mary isn’t also capable of recognizing.

I’m a bit surprised by what you’re saying here Pete. The flagship historicist theory of knowledge is reliabilism, and it’s plausible independently of considerations about egocentric historical knowledge.

Given reliabilism, Swamp Mary doesn’t have knowledge EVEN IF she has the required phenomenal concepts. She’s a Gettier case: her beliefs about what it’s like are true and justified, but they are not knowledge because they are not caused and sustained by suitably reliable cognitive processes. I find the notion of a phenomenal Gettier case pretty amusing.

What you probably care more about, though, is that your criticism of historicism doesn’t seem to cut much ice against reliabilism. Reliabilism has no more or less trouble accounting for knowledge of general truths than knowledge of particular ones, because reliabilism is fine with ampliative inferences. The inputs to the cognitive system might concern particular cases, but that doesn’t mean the outputs can’t be generalities.

Notice that the examples of generalities you mentioned are plausibly apriori. Reliabilism handles apriori knowledge pretty well by positing reliable mechanisms that are input-neutral or don’t require input at all. But the mechanisms responsible for generating and sustaining phenomenal knowledge aren’t supposed to be input-neutral at all, are they?

Your comments about Dretske make me suspect you are conflating two different dimensions of historicism. One dimension is semantic. If you are a historicist about content, then you have to think that there are historical conditions necessary for knowing that p because you think there are historical conditions necessary for THINKING that p. But the other dimension is epistemic. You can take an ahistorical view about content and still insist that there are historical necessary conditions for knowing that p.

-Chase

— and if it’s unknowable, this

What do you think of the suggestions that (1) be assimilated to some combination of Quotation and Actual-cause and that (2)&(3) be assimilated to Nomological?

Re: (1), the way I usually think of the Humean account is that ideas, representing impressions in virtue of being *copies* of them, are, in the usual case at least, both causal consequences of them (thus Actual-cause) and bear first-order resemblances to them (thus Quotation). If a Swamp being has Humean ideas of impressions she’s never had, she’s got quasi-copies that satisfy the resemblance criterion but not the etiological one.

Re: (2)&(3), my thought is that if P has a disposition to Q, then P has some property, R, constituting a categorical base that (in concert with various triggering conditions) is nomically related to Q. But maybe there are dispositions without laws? I’m not very up on the relevant literature.

“As with direct phenomenal concepts, the content of standing phenomenal concepts [those that don't derive from a current experience] does not conceptually supervene on the physical. A question arises as to what determines their content. I will not try to analyse that matter here, but I think it is plausible that their content is determined by some combination of (1) non-sensory phenomenal states of a cognitive sort, which bear a relevant relation to the original phenomenal quality in question — e.g. a faint Humean phenomenal “idea” that is relevantly related to the original “impression”; (2) dispositions to have such states; and (3) dispositions to recognize instances of the phenomenal quality in question.”

As far as I can tell, these three options are distinct from the four options you consider.