Wolfgang Schwarz

Srinivasan's "radical externalism"

Internalism about justification is often supported by intuitions about cases. Srinivasan (2020) argues that these intuitions can't be trusted, because there are analogous cases in which they go in the opposite direction. I'll explain why I'm not convinced.

I should say that I'm not sure what this debate is about. Are we talking about some pre-theoretic folk concept of justification? Or about a concept that plays some important theoretical role? Srinivasan acknowledges (in footnote 10) that there might not be a single, precise folk concept of justification. I agree. To clarify her topic, she says that she is interested in the kind of justification that is a precondition for knowledge. This doesn't really help me. I think that 'knowledge' is context-dependent, and that it sometimes means no more than 'true belief'. There is no interesting justification condition that is present in every case of knowledge.

I'll proceed on the assumption that we're doing old-fashioned conceptual analysis. We'll look at how we are inclined to apply the concept of justification in various hypothetical scenarios, and see if we can find a pattern.

Here, then, are three classical scenarios that are thought to support internalism about justification.

Clairvoyant. Norman's brain has been connected to a device that picks up satellite information about the current location of the US President. Whenever it picks up such information, it causes Norman to believe that the President is at that location. Norman knows nothing about this. Today, the device makes him believe that the President is in New York.

Externalist theories of justification tend to say that Norman's belief is justified. Intuitively, however, it is not.

Dogmatist. Mary looks at a red statue in a gallery and believes that it is red. A gallery assistant tells her that the statue is actually white but illuminated by red light. Mary ignores this evidence and keeps believing that the statue is red.

Some externalist theories of justification say that Mary's belief is (still) justified. Intuitively, it is not.

Vat. Jane is a brain in a vat, under the illusion that she is a normal embodied person. She believes that she has hands.

Externalist theories tend to say that Jane's belief isn't justified. Intuitively, it is.

For what it's worth, I agree with these intuitive judgements. Or rather, I can easily hear a sense of 'justified' that agrees with the judgements. I can also hear a sense of 'justified' on which every true belief (or perhaps every true belief for which there is evidence – whether available to the subject or not) is justified. But let's focus on the first sense.

Srinivasan asks us to consider the following cases.

Racism. Nour is having dinner at a friend's house. The host is racist, and manifests his racism in subtle behavioural cues that Nour subconsciously picks up. As a result, she comes to believe that the host is racist.

Intuitively, Srinivasan says, Nour's belief is justified.

Classism. Charles has experienced classist discrimination at his college. The college Master assures him that he is misinterpreting the relevant events. Charles dismisses this evidence and keeps believing that he has experienced discrimination.

Intuitively, Srinivasan says, Charles's belief is justified.

Sexism. Radha lives in an environment where everyone believes that misbehaving women deserve to be beaten by their husbands. She adopts this belief and maintains it after thorough reflection.

Intuitively, Srinivasan says, Radha's belief isn't justified.

I agree with these intuitive judgements.

The judgements are in line with what some externalist accounts would predict. This alone is obviously not an argument in favour of the externalist accounts. Nobody ever thought that externalist accounts go wrong in every possible case. The objection to externalism is that it goes wrong in some cases – like Clairvoyant, Dogmatist, and Vat. The rules of conceptual analysis say that a correct analysis should get all clear cases right. A single (clear) counterexample is enough to bring down an analysis.

So why does Srinivasan bring up the new cases? I'm not entirely sure what her argument is.

One line of thought, to which she appears sympathetic, is that the new cases are exactly analogous to the old cases in every epistemic respect: Clairvoyant is just like Racism, Dogmatist is just like Classism, Vat is just like Sexism. Since justification is an epistemic matter, it follows that no credible theory of justification can give different verdicts about these pairs. Our intuitive judgements are incoherent and can't be trusted.

The problem with this line of thought is that it is hard to show that the relevant pairs are exactly alike in every epistemically relevant respect. In fact, there is a strong prima facie reason to think that this is false. The reason is precisely that the cases seem to differ with respect to whether the relevant beliefs are justified.

Another line of argument to which Srinivasan appears sympathetic is that some standard internalist accounts of justification give the wrong verdict in the new cases.

The problem with this argument is that it lends little support to externalism. That there are counterexamples to a purported analysis X hardly shows that a certain alternative Y is correct – especially if there are also counterexamples to Y. For example, that there are counterexamples to most alternatives to the JTB analysis of knowledge hardly shows that the JTB analysis is correct.

So I'm not sure how we're supposed to get an argument for externalism out of the new cases.

But let's have a look at the supposedly analogous cases, and see if we can find a pattern in our intuitive judgements.

Take Clairvoyant and Racism. Srinivasan's idea is that Norman and Nour both come to have the relevant beliefs through a reliable causal process, without having any insight into how the process works.

Srinivasan suggests that our different judgements about justification have to do with the fact that Nour is picking up on an oppressive environment of "bad ideology". There might be something to this, but I don't think it's the full explanation.

Imagine a variant of Racism in which the host isn't racist. Instead, he has Asperger's. The Asperger's only shows up in subtle behavioural cues to which Nour is subconsciously sensitive. (Her brother also has Asperger's. That's how she became sensitive to the cues.) As a result, she forms the belief that the host has Asperger's, but she can't explain on what basis she has the belief. Like in Racism, I'd say that her belief is justified. But it has nothing to do with bad ideology.

I suspect that the relevant difference between Clairvoyant and Racism is the strangeness of the process by which Norman forms his beliefs. There's nothing terribly strange about beliefs that are formed on the basis of subtle cues that the subject can't verbally identify. Nour has no strong reason to distrust her hunch that the host is racist (or that he has Asperger's.) Things are different for Norman. It is highly unusual to form accurate beliefs about the President's location through an implanted brain device. Norman has strong reason to distrust his hunch that the President is in New York.

To be fair, Srinivasan stipulates that Norman has "no evidence or reasons of any kind for or against" the hypothesis that he has the relevant clairvoyant power. This blocks my diagnosis. But it makes the case much harder to imagine. How could Norman have no evidence or reasons of any kind against the hypothesis that he has been implanted with a President-tracking device? We all have ample evidence about the kind of processes through which we come to form our beliefs about the present state of the world around us. We have strong reason to think that we generally come to form such beliefs through our sense organs. Without substantive further evidence, all of us should give low credence to the hypothesis that we have extra-sensory clairvoyant powers. How could this be different for Norman? Does he not know that he forms most of his beliefs about his environment with the help of his sense organs? Does he have special evidence suggesting that he has extra-sensory clairvoyant powers? If not, then his evidence isn't neutral on whether he has clairvoyant powers. It strongly favours the hypothesis that he does not. If we expand the scenario so that Norman's evidence is really balanced on whether he has the clairvoyant powers, then my intuition that his belief isn't justified goes away.

Next, Dogmatist and Classism. Srinivasan's idea is that Mary and Charles both have non-misleading evidence for a certain proposition (that the statue is red, that Charles has been subject to discrimination), but then receive misleading evidence against that proposition (an authority tells them that the proposition is false).

OK, but this is a very coarse-grained and superficial similarity. We can readily see important differences if we consider to what degree the relevant beliefs are supported by the information available to Mary and Charles.

We can assume that Mary knows that things sometimes appear red even though they are actually white, because they are illuminated by red light. Let's say she rationally thinks that this is true of around 0.1% of all things that appear red. She has no special information about the statue she is looking at. Initially, she can therefore be 99.9% confident that this statue is red. Now a gallery assistant says that the statue is actually white. We can use Bayes' Theorem to see how this should impact Mary's confidence. We have to ask two questions. First, how likely is it (given Mary's evidence) that the assistant would say that the statue is white if it's actually red? Answer: very unlikely. Second, how likely is it that the assistant would say that the statue is white if it's actually white? Answer: very likely. By Bayes' Theorem, it follows that Mary's degree of belief that the statue is red should significantly decrease.

Now compare Charles. What does he think about the prevalence of classist discrimination in schools before coming to the college? In Srinivasan's story, he thinks it is common. His prior ("base rate") in the hypothesis that he will experience discrimination might be, say, 0.6. However, the Master has assured him before his arrival that there's no discrimination in his college. We assume that Charles trusts the Master's assurance, so that upon arrival he gives low credence to being subject to discrimination. Now he experiences discrimination. For the cases to be parallel, we must assume that Charles no longer trusts the Master's assurance. His evidence of discrimination is strong. He becomes (say) 99.9% convinced that he is experiencing discrimination. Next, the Master reassures him again that there's no discrimination: that he is misinterpreting the relevant events. As before, we have to ask two questions. First, how likely is it (by Charles's lights) that the Master would say this if there were discrimination? Second, how likely is it that the Master would say this if there were no discrimination? The answer to the second question is obviously 'very high'. But what about the first? Before reporting his concerns to the Master, Charles was 99.9% confident that the Master was wrong when he initially told him that there is no discrimination in his college. Based on this experience, Charles can hardly trust the Master's second assurance. It wouldn't be irrational if he didn't put much trust in it at all. By Bayes' Theorem, Charles's degree of belief that he has been subject to discrimination should not significantly decrease.

(Radical externalists might disagree with this analysis. But I find it intuitive. More importantly, it's a very natural thing for internalists to say. And the question, remember, is whether internalists can find a disanalogy between the two cases.)

Finally, Vat and Sexism. Srinivasan's idea is that the internal evidence available to Radha strongly supports that she deserves to be beaten, just as the internal evidence available to Jane supports that she has hands.

But why should be accept this?

Why might one think that Jane's internal evidence supports that she has hands? Because it perceptually appears to Jane as if she has hands. A perceptual appearance that p is pro tanto evidence that p. Jane has no internal evidence suggesting that things are not as they appear. So her internal evidence supports that she has hands.

Radha's case is obviously different. It doesn't perceptually appear to her as if she deserves to be beaten. (She doesn't suffer a perceptual illusion or hallucination.) Rather, she has heard people defend a certain view which entails (together with other things she has reason to believe) that she deserves to be beaten. Does her internal evidence support the truth of these views? I would certainly reject a general principle by which hearing people defend a view is pro tanto evidence that the view is correct. I see no good reason to think that the judgement for Jane should carry over to Radha. True, some particular internalist proposal might give the same verdict about the two cases. But it's trivial to come up with an account that tells them apart.

Now, as I said at the outset, I may have misunderstood Srinivasan's project. As a Williamson fan, she would no doubt reject my suggestion that she is engaged in old-fashioned conceptual analysis. But whatever her project might be, I don't think she has undermined the view that intuitions about cases support internalism over externalism.