In “Externalism Resisted,” Jonathan Vogel addresses some important, foundational questions about the nature of justification.Footnote 1 Vogel's focus is on Ernest Sosa's case against internalism about justification in Epistemic Justification.Footnote 2 Externalism about justification, according to Sosa, is the only game in town.
Vogel discusses two examples from Sosa. The examples are structurally similar, and for simplicity's sake, we will focus upon the following one:
Breakfast Case. You remember having oatmeal for breakfast, because you did experience having it, and have retained that bit of information through your excellent memory. Your counterpart self-attributes having had oatmeal for breakfast, and may self-attribute remembering that he did so (as presumably you do), but his beliefs are radically wide of the mark, as are an army of affiliated beliefs, since your counterpart was created just a moment ago [by a demon], complete with all of those beliefs and relevant experiences. Are you two on a part in respect of epistemic justification? (730)
Sosa's answer to the concluding rhetorical question is “No.” We have, according to Sosa,
(a) For you, J(B)
(b) For your counterpart, ~J(B)
where ‘B’ stands for ‘I had oatmeal for breakfast’ and ‘J’ stands for ‘The subject in question is (objectively) justified in believing that …’ Sosa regards the case as a counterexample to internalism. Following Vogel, we will focus upon “Chisholmian internalism”:Footnote 3
The concept of epistemic justification … is internal … in that one can find out directly, by reflection, what one is justified in believing at any time. (729)
Our question here is as follows: Does Vogel's critique of Sosa on the Breakfast Case (hereafter BC) constitute a successful defense of Chisholmian internalism, thereby allowing us to resist externalism about justification? Sosa characterizes the externalism which he accepts and which Vogel resists:
When a belief is epistemically justified, that is because it comes from an epistemically, truth-conducively reliable process or faculty or intellectual virtue. (729)
So let us turn to the heart of Vogel's analysis of BC.Footnote 4 Vogel maintains that a “puzzle” for Sosa's anti-internalist view of BC arises in the following way. Where “M” stands for “The subject in question remembers that …, ” we have (according to what we shall call “Sosa's Intuition”)
(2) For your counterpart, ~M(B) and ~J(B).Footnote 5
According to Sosa's Intuition, the second conjunct holds in BC in virtue of the holding of the first conjunct.Footnote 6 Now consider the following Connection Thesis (hereafter CT):
If you're justified in believing that there is a reliable information source according to which X, then you're justified in believing X. (734)
For example, if you're justified in believing that your thermometer is reliable in its reading of the current temperature as being 80 degrees, then you're justified in believing that the current temperature is 80 degrees. Let us symbolize CT as
(6) If J(R(X)), then J(X).
According to Vogel, (6) is highly plausible for any theorist of justification. Applying (6) to BC, we have
(8) For your counterpart, if J(R(B)), then J(B).
Since “Sosa allows that your counterpart is justified in believing that his belief that B arises from a reliable information source, namely memory” (735), we have
(9) For your counterpart, J(R(B)).
(9) is true even though R(B) is false (since the source of your counterpart's belief is demonic). This is just an example of fallibilism about justification. (8) and (9) yield
(10) For your counterpart, J(B).
But that contradicts the (b)-component of Sosa's intuition (2):
(b) For your counterpart, ~J(B).
This is not so much a “puzzle” for Sosa's view of BC, as Vogel says, as it is a refutation of that view on the basis of the instance (8) of CT. We need to take a step back and consider Vogel's “puzzle” in light of the distinction between doxastic and propositional justification. As BC is set up by Sosa, it appears to concern doxastic justification: “Your counterpart self-attributes having had oatmeal for breakfast” (730). That is, your counterpart believes B, and the issue of the doxastic justification of that belief accordingly arises.Footnote 7 According to Sosa's Intuition. the pertinent belief of your counterpart is not doxastically justified. Thus the ‘J’ in
(b) For your counterpart, ~J(B)
is to be interpreted as concerning doxastic justification. Thus the “puzzle” reasoning from (8) to (10) must be regarded as concerning doxastic justification if a contradiction is to be derived. In this case, this means that diachronic considerations come into play, as we proceed to explain.Footnote 8 So (b) should be interpreted as concerning time-indexed doxastic justification:
(b') For your counterpart, ~DJt(B)
where “DJt” stands for “The subject in question is doxastically justified in believing … at t,” and where t is the time at which your counterpart comes to believe B, simply accepting it as a delivery of memory without raising higher-order questions about the reliability of memory. Let us suppose that (i) at a later time t', your counterpart comes to have a higher-order belief of R(B), and (ii) this belief at t' is doxastically justified. So we have
(9’) For your counterpart, DJt'(R(B)).
At t', your counterpart's evidential basis for believing B has expanded from what it was at t, since that basis now includes his belief of R(B). Now let us return to (8), interpreted as concerning doxastic justification and properly time-indexed:
(8’) For your counterpart, if DJt'(R(B)), then DJt'(B).Footnote 9
(8’) and (9’) yield
(10’) For your counterpart, DJt'(B).
But there is no contradiction between (b') and (10’). It is a familiar point that given an expansion of one's evidential base, a belief that starts out doxastically unjustified can later become doxastically justified (the reverse point holds as well).
What if the “puzzle” is understood as concerning propositional justification? The “puzzle” would reach the following contradiction: your counterpart both has and lacks propositional justification for believing B at t. Here is how this would seemingly work.
Consider first
(b'') For your counterpart, ~PJt(B)
Here “PJt” stands for “The subject in question has propositional justification for believing that … at t.” According to Sosa's Intuition, your counterpart's belief of B at t lacks doxastic justification, and so it would seem that he also lacks propositional justification for believing B at t (more on this later; it is crucial to the question of propositional justification regarding B for your counterpart). We also have
(9’’) For your counterpart, PJt(R(B)).
Should he have reflected at t upon the apparently memorial source of his belief of B, he would have been doxastically justified in believing R(B) at t. We assume that he does not undertake this reflection as he first forms his belief of B at t; maybe he never comes to form a reflectively based belief of R(B). What of the pertinent instance of CT? It says that if your counterpart has propositional justification for believing R(B) at t, then he has propositional justification for believing B at t:
(8’’) For your counterpart, if PJt(R(B)), then PJt(B).
Why would this be true? Presumably the story is as follows. Suppose that your counterpart has propositional justification for believing R(B) at t. Suppose he had had a doxastically justified belief of R(B) at t. Then if he had inferred from R(B) to B at t', then he would have had a doxastically justified belief of B at t'. Hence, given that he has propositional justification for believing R(B) at t, he also has propositional justification for believing B at t. So from (8’’) and (9’’) we have
(10’’) For your counterpart, PJt(B).
This contradicts (b'').
We should reconsider the earlier inference from (I) Sosa's Intuition that your counterpart's belief of B at t lacks doxastic justification, to (II) your counterpart lacks propositional justification for believing B at t. It is a commonplace that this sort of inference is invalid. For example, I might believe X at t on a flimsy evidential basis (I lack doxastic justification for believing X at t) and yet possess evidence E for X which is such that if I had appealed to E in support of my belief of X at t, then that belief would have been doxastically justified at t (I possess propositional justification for believing X at t). The situation regarding your counterpart is similar. Though according to Sosa's Intuition your counterpart lacks doxastic justification for believing B at t, he may well possess an unused propositional justication for believing B at t (as described above). So the contradiction would be blocked.
So we have tried to allow Sosa to accept some version of Vogel's use of CT in BC and yet (0) avoid contradiction, while (00) holding to Sosa's Intuition about your counterpart's lack of doxastic justification for believing B at the time t when he accepts it as a deliverance of memory.Footnote 10
Let us end by considering CT's plausibility. Given
(A) If S is a bachelor, then S is male
it is plausible to hold that
(B) If J(S is a bachelor), then J(S is a male).
But this move cannot be mimicked as follows:
(A') If R(X), then X
(B') If J(R(X)), then J(X).
This is because (A') is plainly false. How about:
(A'') If R(X), then Pr(X)
(B'') If J(R(X)), then J(Pr(X))
where “Pr(X)” stand for “It is highly probable that X”. (A'') is prima facie plausible, and so is the inference to (B''). But in order to get the contradiction involved in Vogel's “puzzle”, we need to be able to get some properly interpreted version of
(10) For your counterpart, J(B)
and not just
(10*) For your counterpart, J(Pr(B)).
However, maybe we could reason from (10*) to (10). Things are tricky here, for lottery reasons. One might well think that I am justified in believing Pr(L) (where L = My ticket in the lottery is a loser) even though I am not justified in believing L. Certainly many share the intuition that I lack knowledge of L even if L is true.
Even though our response on behalf of Sosa above did not call into question the truth of CT, there is something peculiar about the defense offered since, as Vogel (737–8) himself notes, Sosa and many other externalists are committed to the falsity of the thesis. This is easily shown to be the case for the pure reliabilist who holds that reliability is necessary and sufficient for justification. For the pure reliabilist, CT is equivalent to the claim that R(R(X)) entails R(X). However, this claim is surely suspect since reliably believing a source to be reliable is no guarantee that the source is reliable.
Matters are more complex when we consider Sosa and CT. Sosa denies that reliability is sufficient for justification, but does think it to be necessary. Instead, it is reliability plus a further ‘internal’ condition that guarantees justification. Following Vogel, let us indicate that this condition obtains for a particular belief X with “I(X).” Thus, presuming we now have arrived at an attempted analysis of the concept of justification, Sosa endorses the “hybrid” (SOS)
(SOS) I(X) and R(X) if and only if J(X).
If CT is correct, then Sosa is also committed to
(20) I(R(X)) and R(R(X)) entail R(X).
The ‘antecedent’ of (20), together with the left to right direction of (SOS) yields J(R(X)). If CT is true, it follows that J(X). And now we can employ the right to left component of (SOS) to derive R(X). Though perhaps less counterintuitive than the commitment of the pure reliabilist identified above, (20) is also suspect. Suppose that (i) I have good reasons to believe that my speedometer is reliable, (ii) my beliefs about speedometer reliability are often correct, and (iii) it informs me that I am traveling at 35 miles per hour. If (20) is correct, then my speedometer must be reliable; but the preceding seems perfectly consistent with my having a bum speedometer.
We thus see that the most obvious response one could give on behalf of Sosa to Vogel's attempted reductio is to deny CT. How successful this move is may in part depend upon the independent plausibility of the thesis itself. Vogel (734) does introduce the thesis as flowing from a “conceptual connection between reliability … and justification.” However, if this is the case, then the reductio is superfluous for Sosa's theory is incompatible with a conceptual truth linking reliability and justification.
Matters are made more complicated when we consider the relationship between internalism and the connection thesis. As with externalism, there are numerous formulations of internalism. Let us focus here on Vogel's preferred version, Cartesian Internalism (CI).
Justification requires only really proper thought on the part of the subject: if a believer has obtained and sustains his belief through wholly appropriate thought, then the believer is justified in so believing – where the appropriateness of the thought is a matter purely internal to the mind of the subject, and not dependent on the environment. (729)
Vogel argues that CI is consistent with the connection thesis since if one believes that X on the basis of believing properly that one's source is reliable, it is hard to deny that the belief that X is justified. However, Vogel (fn. 20) is aware that not all beliefs will be formed in this fashion for there are presumably non-inferential beliefs such as those derived from perception. It is here that CI and CT collide. To see this, we need only turn to Vogel's own understanding of the breakfast scenario.
By contrast, your counterpart believes, but doesn't really remember, that he was in M [having breakfast experiences]. His belief results from the action of a capricious demon. Such a process of belief formation hardly seems to count as proper thought. Since your counterpart's belief isn't the outcome of proper thought, the Cartesian Internalist can agree with Sosa that your counterpart's belief is unjustified. (731)
The problem is that built into the case is the supposition (used in generating the contradiction) that our counterpart has a justified belief that his belief-forming process is reliable. If Vogel's understanding of CI is correct, we have a case of J(R(X)) yet ~J(X). Thus, CT is inconsistent with at least one version of Internalism.
The gnarly dialectical situation is as follows. We first argued, contra Vogel, that Sosa can without contradiction grant CT and still adhere to his Intuition that your counterpart lacks doxastic justification for believing B at t, when he first forms the belief via taking a seeming memory at face value. Then we considered CT's plausibility when viewed independently of the assumption of any specific theory of justification, finding CT to be far from obviously plausible. Finally, following Vogel, we noted that CT is incompatible with both pure reliabilism and hybrid theories of justification like Sosa's own. We further noted that CT is incompatible with CI, given Vogel's own analysis of BC. Why stick with CT? Vogel says, “It's difficult to give up the thought that justification is somehow connected to truth, although the exact relation between the two isn't easy to make out” (739). The usual unhelpful formulation is “Justification is truth-conducive.” But CT runs in the opposite direction: from a justified belief about reliability (truth-conduciveness) to (ground-level) justification. Vogel also says, “I tend to think that the Connection Thesis isn't easily dispensed with – in part, it reflects the elusive but essential way in which justification has to do with the ‘perspective’ of the subject” (740). We cannot readily discern the connection between the Connection Thesis and the perspective of the subject.
So in the end we resist Vogel's resistance of externalism via the use of CT.
