Externalism assumes that internalism misses the very point of the act of justifying statements. Epistemic luck about justifiers should not be a significant factor in cases of true knowledge, not fortuitous ‘true opinion.’ The justifier can be independent of the beliefs of Adam, resting on the fact that the justifier is in fact a proper justifier. In other words, if The Times is an accurate predictor, it does not matter if Adam have awareness of his access to an accurate predictor.
Even if Adam cannot provide a good reason why The Times is a reliable source, as long as The Times is in fact a proper justifier of p, and if Adam forms the belief p after reading The Times, and if the content of The Times is in fact true, then Adam is in possession of knowledge. To generalize, in the externalist program Adam will have a justified true belief iff
- irrespective of belief in the justifier of p, the good reason is a proper justifier for p,
- believes that p, and
- p is in fact true.
For example, take Nozick’s three necessary conditions for knowledge:
- p is true
- s believes that p
- s’s belief that p is sensitive.
‘Sensitive’ is to be understood as follows: Adam’s true belief that p is sensitive iff Adam would not have believed p if p were false. If Adam would have believed p if p were false, his belief that p is insensitive. If Adam would not have believed p if p were false, his belief that p is sensitive. In other words, S’s belief that p is sensitive when in (roughly) all those worlds closest to the actual world, p is also true. This solution sidesteps the pit the internalist program falls down, providing a litmus test for certifying statements as true or false. However, one consequence of the externalist program is that knowledge exists quite often without a knowing subject: individuals do not know, and cannot know, that they do in fact know.
Some people might find this counter-intuitive result satisfying to their intuitions, but it does not satisfy the justificationist thesis. It is impossible to determine whether or not one has satisfied Nozick’s third condition. A touchstone exists, but it is inaccessible. For instance, the basic universal term ‘glass’ entails a long list of dispositions (‘breakable’, ‘able to melt at 2300 C°, etc.), so every statement that employs it (for instance, “Here is a glass of water”) assumes these universal terms are true in all cases.
This assumption is highly dubious, if the history of science is any indication. These disposition claims are highly conjectural. If Adam were to assume that these terms are true in all closest possible worlds or even in this very case, he is begging the question, namely that there is in fact glass and not some glass-like substance with a somewhat different disposition. The possibility that Adam is wrong cannot be ruled out.
This problem present in all versions of the externalist program is as follows. Imagine a computer that answers every yes-or-no question truthfully, yet the computer answers only with the terms ‘ja’ or ‘nein,’ alternating apparently at random between the two (the apparently random switch between the two terms makes sense only in an unfathomable context understood only by a very complex algorithm).
Adam cannot know when the computer, even though the computer only tells the truth, is indicating a statement is true or a statement is false. Another computer with the very complex algorithm is required in order to translate the answers.
Even if it is assumed that the closest possible worlds do satisfy the third condition, and assuming that naïve realism is true, Adam must assume that he exists in one of these worlds and not in a wholly different possible world where his subjective experiences are mistaken.
Adam knows that he must possess knowledge when the three criteria are satisfied, yet he does not know when the three criteria are in fact satisfied. The results of the three necessary conditions for knowledge are indecipherable to Adam. Adam could be a brain in a vat, or his experience of the sheep on the hill may not be sheep on the hill; ‘nein’ could mean ‘yes’ or ‘no’. These possibilities — realism is false; or, if realism is true, appearances do not properly correspond to the facts — cannot be ruled out by Adam.