According to Silins (here), Evidential Internalism follows from a plausible access principle, the principle of armchair access:
(AA) If S’s evidence includes the proposition that p, it is possible for S to know that p is included in her evidence from the armchair.
Here is the argument. Let’s assume that it is possible for S to know that she has hands. Let’s call the proposition that S has hands, ‘HANDS’. According to E = K, if S knows HANDS, S’s evidence includes HANDS. According to E = K, if S’s evidence includes HANDS, it must be the case that HANDS is true. According to AA, if S’s evidence includes HANDS, it is possible for S to know from the armchair that her evidence includes HANDS. Now, it seems that if E = K is true, S ought to be able to know that it is true from the armchair. Let’s assume that she does. We can show that Evidential Externalism is false because it has false consequences about how we can have knowledge about the external world.
It seems that:
(1) S has armchair knowledge that her evidence includes HANDS.
(2) S has armchair knowledge that if her evidence includes HANDS, it really is that HANDS is the case.
(C) Thus, S is in a position to know that HANDS is the case from the armchair.
But, (C) is supposed to be absurd. It’s not entirely clear why it is supposed to be absurd. In order for (1) to be true, we have to assume that S has introspective knowledge about her experiences. If she forms the beliefs in response to her experience we would expect her to (i.e., she believes HANDS because she undergoes an experience indistinguishable from the veridical experience of hands) and her belief proved to be correct, would that belief not turn out to be knowledge? Silins does argue (convincingly, I think) that problems arise for E = K. But, the denial of E = K does not amount to acceptance of Evidential Internalism.
Note that you can reject E = K and still derive the (allegedly) dread consequence that S is in a position to know that HANDS is the case from the armchair if she knows that she has hands. You can derive that conclusion is we assume two further claims:
NIK --> E: If S knows p non-inferentially, p is included in S's evidence.
E --> T: If p is included in S's evidence, p is the case.
Assuming that these are known from the armchair, we can still derive (C) even if we reject E = K. If you're a fan of armchair access and bothered by (C), which do you give up?
Response 1: (AA), (~C), and (~(NIK --> E))
(C) really is a dread consequence. We ought to retain (AA) and reject (NIK --> E). Non-inferential knowledge alone is not sufficient for including some proposition in your evidence, you need armchair knowledge of a proposition for that proposition to be included in your evidence.
Note that by 'evidence', Silins means a reason that provides a justification for holding a belief. I have to say that I find it exceptionally odd to think that propositions known non-inferentially could not constitute reasons that could justify further beliefs. There might be some further argument that convinces us that we should say this, but I suspect that that argument would make the armchair access argument otiose.
Response 2: (AA), (~C), and (~(E --> T))
(C) really is a dread consequence. We ought to retain (AA) and reject (E --> T).
I'm bothered by the suggestion that there can be false propositions that constitute evidence. For me, the worry is brought out by this exchange:
Scarlet: Does the prosecution have solid evidence against Mustard?
Green: Yes, they have all sorts of evidence against him: namely, that he was the last one to see the victim alive, that his alibi did not check out, that his fingerprints were on the murder weapon, and that he had written a letter containing details the police think only the killer could have known.
Now, consider a second:
Plum: How good is the prosecution’s evidence against Mustard?
Peacock: It seems that the evidence is pretty strong. However, Mustard’s prints are not on the murder weapon, his alibi checks out, and he was not the last one seen with the victim. This is all perfectly consistent with the evidence that the prosecution does have.
It seems as if Peacock’s assertion flatly contradicts Green’s assertion. But all that Peacock has done is assert that the falsity of certain propositions is consistent with other propositions about what the prosecution’s evidence consists of. So, unless we say that claims about what someone’s evidence consists of entail that those claims are true, it’s hard to see how Peacock’s assertion could contradict Green’s assertion. Peacock’s assertion speaks to the veracity of the prosecution’s claims rather than speaking directly about the evidence that they have.
Response 3: (~C), (~AA), (E --> T), and (NIK --> E)
Suppose (C) really is a dread consequence to be avoided at all costs. It seems that I know that:
(1) It's not wrong for me to believe I have hands.
This is a mildly anti-sceptical assumption. If this is true, I should also be in a position to know:
(2) It's not wrong for me to believe that I have hands given just the evidence I actually have.
Because (C) is a dread consequence, it follows from the denial of (C), (2), and (AA) that I should be able to work out that:
(3) It's not wrong for me to believe that I have hands given just the evidence that I have from the armchair even though this is not evidence that puts me in a position to know that I have hands.
But (3) clashes with the principle that asserts that if you know your evidence does not put you in a position to know, you oughtn't believe that proposition until you gather additional evidence. So, either (C) is not the dread consequence it's taken by internalists to be or it is but (AA) isn't the innocent principle it's taken to be. (Obviously, I don't think that (AA) is all that innocent because if (C) really is damning, it would force us to say that there are decisive epistemic reasons to refrain from treating some propositions known non-inferentially to you as evidence or would allow us to say that false propositions constitute evidence.)