A recent profile piece in the Financial Times highlighted the breadth of Amia Srinivasan’s thought. It is not surprising, then, that the new Chichele Professor of Social and Political Theory at Oxford has written one of the most interesting recent papers in epistemology. It’s also a sign that philosophy in the analytic tradition is beginning to recognise the central importance of political theory for epistemology; something that has long been known in, for instance, the Black feminist tradition. Srinivasan’s ‘Radical Externalism’ is published in The Philosophical Review. It gives an argument for what epistemologists’s call an externalist account of justified belief; and it tries to explain why such an account might be what epistemologists call normative. But we’ll come to all of that. Let me begin by setting the scene.
The first philosophical idea I ever encountered was a sceptical challenge. It was on a television show when I was a teenager; Blackadder, I think, though I’ve never been able to track it down again. Here it is. You think the world has existed for many years. You think your memories of the past are accurate, and the traces of the past you see in the world around you are genuine. But how can you tell that the world was not created only a few seconds ago, along with you and your apparent memories and all the other misleading evidence of its great age? How could you ever rule out that possibility? And how could you justify your belief if your couldn’t rule it out?
That was my first experience of philosophy. Gradually, I came to realise that this was just one sceptical challenge among many. Here’s another: You think that your friends and family and other people you meet have conscious minds just like your own. You believe that they think like you, experience pain and sadness and love just like you. But how can you tell that they aren’t sophisticated robots with no conscious experience at all, trained to simulate human behaviour flawlessly? What evidence could you ever collect that would allow you to dismiss that possibility?
In both cases, the sceptical challenge has the same structure. You have a particular belief (the world’s old; my friends have minds). The sceptic describes a hypothetical scenario in which everything would seem to you exactly as it actually seems, and yet the world beyond your mind would be dramatically different, and the belief you have about it would be false (the world’s very young, but it was created with apparent evidence of the past and your memories; your friends have no minds, but can simulate human behaviour perfectly). You are then invited to justify believing that the world is the way you think it is, rather than the way it is in the sceptical scenario. And you find that you have no resources to draw on; everything you might use would be exactly the same in the sceptical scenario.
That’s the sceptical challenge, and it puzzled me on and off during my teenage years. What could break the symmetry between my belief and the sceptical hypothesis? Answer: nothing to which I have access. After all, in each case, the sceptical hypothesis is designed specifically to ensure that, were it true, my evidence, my memories, and anything else accessible to me would be exactly as it actually is. Perhaps, then, it is some intrinsic feature of my belief or the sceptical hypothesis that could tell between them. Perhaps I could argue that my belief is simply more plausible? But what makes it so? Is it that it’s a simpler or more elegant explanation of the things I experience? For one thing, it’s not clear that it is either of those things; and for another, it’s not clear why it would be relevant if it were. Sceptical challenges are hard to answer.
When I studied philosophy as an undergraduate, however, I learned that there was a very straightforward answer to these challenges. It’s called externalism. According to the externalist, the sceptical challenge looks unsolvable only because we’ve been in thrall to internalism, a faulty account of when a belief is justified. According to internalism, or at least the version known as access internalism, whether or not your belief is justified depends only on what’s going on in your mind, and in particular the part of your mind to which you have conscious access. The idea, roughly speaking, is that a belief is justified if the person holding it can justify that belief using what’s available to them. For the internalist, then, the sceptical challenge is difficult to answer because everything that is available to you would be the same in the sceptical scenario; so it seems impossible to adduce any of it as evidence against the sceptical hypothesis.
According to externalism, on the other hand, whether or not your belief is justified typically depends not only on what’s going on in your mind, but also what’s going on in the world beyond your mind, and the relationship between your mind and that external world. Thus, for instance, some externalists will say that a belief is justified if the process that led you to form it is highly reliable; that is, it produces a high proportion of true beliefs and a low proportion of false beliefs. For instance, for a sighted person, a belief about a nearby, reasonably large object formed on the basis of their visual experience will count as justified because that process is highly reliable: while a typical visual system can lead us astray, particularly in the case of visual illusions, it delivers true beliefs very often and false ones relatively rarely.
One of the central arguments in favour of externalism has always been that it provides a compelling response to the sceptical challenge. According to the externalist, what breaks the symmetry between your belief that the world is old and the evidence of its age veridical, on the one hand, and the sceptical hypothesis that the world is very young and the evidence of its age misleading, on the other, is that, in the world we actually inhabit, which is old and contains veridical evidence, the process by which you form your belief about its age is highly reliable. You see the evidence of the age of the universe using your visual system, or you draw it from your memory; you collect together that evidence; and you infer from it all that the universe has existed for longer than the last couple of seconds. Each step in this process is reliable: your visual system reliably forms true beliefs; so does your memory; and, in this world that we actually inhabit, extrapolating from evidence in the way you do is also a reliable process. According to the externalist, then, your belief is justified and the sceptical challenge is answered.
Against this triumph for the externalist, the internalist often pits a series of hypothetical cases where the externalist seems to get things wrong. Consider Jaya. Entirely unbeknownst to her, Jaya is blessed with a remarkably reliable ability to tell a fake Leonardo painting from a real one. Almost however good the fake is, she will be able to spot it. But she’s never used this ability before and she has no idea that she has it. One day in a museum, standing in front of a painting that the gallery attributes to Leonardo, this ability leads Jaya to form a strong belief that the painting is a fake. She doesn’t know where that belief came from, nor what features of the painting led her to conclude that. Five years later, experts verify that the painting is indeed a 19th century forgery. According to externalism, Jaya’s belief was not only true, it was justified, because it was formed by a reliable mechanism. For many people, this is the wrong verdict. Jaya was right, but she could give no justification for her belief, and indeed she wasn’t even aware of the basis for it, let alone why the basis provided support to the belief. So it wasn’t justified. So externalism gives false positives: there are unjustified beliefs it considers justified.
Another riposte to the externalist points to its false negatives. Think about a world where one of the sceptical hypotheses is true. The world really did come into existence two minutes ago, complete with me, my memories, and all the signs that it’s older than it really is. Surely then I would still be justified in believing that it was old. After all, I’d have all the same evidence! But the externalist will typically have to deny this. So there are justified beliefs it considers unjustified.
So much for the scene-setting. We’ve met internalism and externalism now. We’ve seen why internalism finds it hard to answer the sceptical challenge, why the externalist finds it easy, and why that latter victory comes at the high cost of false positives and false negatives. Let’s turn now to Srinivasan’s article. In it, she notes a different sort of sceptical challenge that externalism can meet. This challenge is even more powerful than the traditional one, so externalism’s achievement is all the greater and the argument in its favour all the stronger.
In the standard sceptical cases, there are two hypotheses, the one you believe and the sceptical one that your interlocutor uses to argue that your belief isn’t justified. Your evidence seems not to tell in favour of either, but we want to say that your belief is in fact justified. In the sorts of case Srinivasan has in mind, there are again two hypotheses, the one you believe and the sceptical one. But in this case your evidence seems to tell positively in favour of the sceptical one. Yet we want to say that your belief is justified. This is clearly a harder challenge, but, as Srinivasan notes, externalism is up to the task.
Srinivasan is thinking of cases in which an individual comes to recognise an aspect of their oppression or identifies an instance of prejudice, group-based hatred or mistreatment, or discrimination. A working class academic at an Oxford college who hears comments from their colleagues, rightly concludes that the institution is classist, and retains their belief even when the Principal of the college, also from a working class background, assures them that they have misinterpreted the comments. A woman working on a production line who rightly interprets the behaviour of a male colleague as sexual harassment and retains that judgment in the face of pressure from a misogynist society that consistently downplays such claims and dismisses them as neurotic overreactions or fabrications. A person of colour who interprets the media coverage of a Black celebrity as racial hatred, and who retains her confidence in that reading even as influential media figures, many of whom she usually trusts, bellow that it’s all been blown out of proportion.
Very often, Srinivasan contends, we wish to say that an individual is justified in ignoring the vast body of countervailing testimonial evidence that threatens to gaslight them in these cases. She claims that they are justified in retaining their belief in the face of that powerful evidence. But of course the internalist will often have to say that they are not justified, for the countervailing evidence against their belief seems much stronger to them from the inside than the original evidence for their belief. As we’ll discuss more below, that is precisely why gaslighting is so effective and so insidious. The externalist, on the other hand, can say that they are justified, since we can see from the outside that the way they produced that belief is in fact reliable; the belief stood in the right sort of relationship to the part of the world that it concerned. In the classist, racist, misogynist society such an individual inhabits, such countervailing gaslighting evidence is typically misleading, while their own ability to detect classism, misogyny, and racism is reliable. So, just as in the original sceptical cases, the externalist’s verdicts about justification square with our favoured ones.
As with the externalist solution to the traditional sceptical problems at the beginning, there are of course internalist objections. Consider, for instance, Charlie. They stand in an art gallery, looking at a painting that appears to them entirely red. They are told by the gallery curator that it is in fact a white canvas, but lit by red light. Nevertheless, Charlie persists in their belief, and in fact they turn out to be correct. Though typically an honest individual, the gallery curator was lying. Is Charlie’s belief justified? The externalist will typically have to say that it is. Yet many people feel this is the wrong answer. It is another false positive for the externalist. Perhaps they have only managed to defeat the sceptic by making it too easy for a belief to count as justified.
At this point, we seem to have reached a stalemate. There are cases that the externalist seems to get right: they save us from traditional forms of scepticism; and they render justified individuals who reliably detect instances of oppression and prejudice. And there are cases that the internalist seems to get right: Jaya and her uncanny eye for a Leonardo fake; and Charlie’s use of standard visual perception to detect the colour of a surface. How can we tell between them?
Philosophers differ on how we should adjudicate these disputes. For some, we should cleave as closely as possible to the data we collect. That is, we should present these cases to a large and diverse sample of people, collect their responses, and choose whichever account best fits the evidence. We might call this the empiricist approach. For others, our concepts are not some fixed feature of the world that we must discover by polling users. They are malleable, and it is part of our role as philosophers to recommend how they should be used. We might make these recommendations based on the general practical benefits that we anticipate accruing from using the concept in one way or the other. We might call this the pragmatist approach. Or we might follow Sally Haslanger and make these recommendations based on the political usefulness of the concept. We might call this the ameliorative approach.
The empiricist approach is likely to end either in stalemate, or in an extraordinarily complex concept, boasting myriad epicycles, designed to fit the contours of the data. And while we’ve seen Srinivasan argue that there are practical and indeed political reasons to use the externalist version of the concept, we have not yet considered whether there might be equally strong or stronger practical or political reasons to use the internalist notion.
However, all of this assumes that there are just two options: adopt the internalist concept and eschew the externalist one; or use the externalist concept and reject the internalist one. But perhaps there is a third option: accept both. For the empiricist, this would mean concluding that the term is ambiguous between two concepts. For the pragmatist and the ameliorative approach, it would mean that there’s practical or political value in dividing the concept in two and retaining both halves.
For the empiricist, it seems plausible to draw this conclusion from the stalemate. After all, this is what philosophers have done in other cases. Suppose I take a coin out of my pocket and tell you it is a trick coin, biased either towards landing heads or biased towards landing tails. You say the coin is just as likely to land heads as to land tails. Are you right? Many will say yes, and many will say no. Another stalemate. In this case, we resolve the standoff by saying that there are two concepts of probability in play. The first is epistemic, the other ontic. The first measures something like your degree of confidence, and on that concept, it is right to say that the coin is equally likely to land heads or tails, for my confidence in each outcome is the same. The second measures something about the world independent of our knowledge of it, and on that concept, it is not just as likely to land as tails. Indeed, it is either biased towards heads in which case it is more likely to land heads than tails, or it is biased towards tails in which case it is more likely to land that way than the other. These two versions of the concept of probability have been distinguished and even given their own names: credence for the epistemic concept and chance for its ontic cousin.
And indeed there are other examples: when a thirsty person holds a glass of poison that they mistakenly believe to be harmless water, we say that they have internal reasons for drinking it but external reasons to refrain. We allow the concepts of internal and external reasons to co-exist, because we recognise that both concepts are useful, and a conceptual scheme that welcomes both might be more cluttered than one that accommodates only one of them, but it more than compensates for that by yielding us greater expressive power.
Indeed, it’s worth noting that philosophers have already divided the very concept of justification, though in a different way from the one that is proposed here. If I have evidence for a belief, but I’ve formed the belief not on the basis of that evidence but on the basis of wishful thinking, for instance, we say that my belief is propositionally justified but not doxastically justified. So we already countenance different versions of justification. Why not also countenance internal and external versions?
In sum: there are certainly precedents for splitting a concept in two when the pursuit of a precise account of the single concept end in a stalemate. From Srinivasan, we have seen the practical and political import of retaining the externalist version of the concept of justification. But what of the internalist version?
As is often noted, one practical use of the concept is in legal settings. In many cases, whether a person should be found guilty of a crime depends on whether they were justified in believing that they were doing no harm. Think again of the person holding a glass of poison that they believe to be water. They offer their friend a drink, which kills them. Are they guilty of any crime? The question often turns on whether their belief that the glass contained only water was justified, where that concept is used in the internalist sense.
Such an argument might not convince Srinivasan, though, who agrees with Marx that excessive concern with assigning blame is a bourgeois preoccupation. I tend to agree. But I don’t think it requires excessive concern to think that it would be useful to have an epistemic concept that we can use to make such ascriptions. If we must choose between only using the internalist concept or only using the externalist concept, then it does perhaps show excessive concern with blameworthiness to insist on the concept that supports such judgments. But if we can choose both, and we choose to include the internalist concept for this reason, it doesn’t.
What’s more, the appeal of the internalist concept is not restricted to its role in assigning blame. We also use it to signal that a person’s beliefs or actions are internally coherent or rationally comprehensible, and that is useful because it allows us to predict their future behaviour. If I see someone behave in a way that is incomprehensible to me, but learn that they are internally justified in doing so, I learn that they must have beliefs that support or rationalise that behaviour, and that allows me to predict how they will behave in other situations. One weakness of the externalist concept is that learning that someone’s belief is justified tells us little about the rest of their cognitive states. It tells us only about the relationship between that belief and the part of the world that it concerns.
So there are pragmatic benefits to including the internalist concept of justification that go beyond its use in assigning blame. But these are not specifically political benefits to weigh against those catalogued by Srinivasan for the externalist concept. To see that there are those sorts of benefits too, we need look no further than the sort of case that Srinivasan herself considers.
In each of the cases we described above—the classist Oxford college, the sexual harassment of a woman in the workplace, and the racist coverage of a Black celebrity—the individual who has identified these phenomena is gaslighted by society as a whole. That is, the collective beliefs and attitudes of the society they inhabit bombard them with powerful but misleading evidence that threatens to undermine their belief in their own ability to discover certain sorts of truths about the world, and thereby threatens to undermine their beliefs in those truths. They are told that what seems to them good evidence of oppression or prejudice is in fact not good evidence; they are told that their reaction to it is out of proportion to the evidence they have; and they are told this often and by many people whose judgments they trust in other spheres. As Srinivasan points out, we want to say that those who resist such gaslighting are justified in doing so. And the externalist is best equipped to deliver that judgment. However, in order to understand just what is so insidious about gaslighting, we need to appeal to the internalist notion of justification as well. Let’s see why.
Note that, while Srinivasan is right to say that there’s a sense in which the person who resists gaslighting is justified in their belief, there is also a sense in which the person who does not resist is also justified, for they believe in line with the evidence that they possess. And it is important to have a concept that recognises this fact not only so that we don’t blame victims of gaslighting for their failure to resist, but also because it allows us to see how gaslighting differs from other sorts of harmful epistemic practices. Some politically repressive epistemic practices appeal to our rationality and others to our irrationality. Gaslighting is in the former camp. The goal of gaslighting is to make you doubt something you formerly believed: that a particular interaction counted as sexual harassment or rape; that a comment was racist or classist or homophobic. It achieves that goal by bombarding you with evidence to which the rational response is to drop your belief. After all, according to most, when you find out that nearly everyone disagrees with you on a topic about which you take yourself to be no more o an expert than them, the rational move—or perhaps just a rational move—is to drop your belief on that topic, even if you don’t wholeheartedly take up the opposing majority belief. This is very different from a harmful epistemic practice that appeals to our cognitive biases, for instance. Suppose I try to make you drop your belief that Nadiya is a competent algebraic topologist by reminding you of certain features of her that your implicit biases associate with a lack of mathematical competence. I’ve deployed a very different sort of epistemic weapon against your beliefs here. And, from a political point of view, this is important because the way we must defend against these weapons will be rather different. In the latter sort of case, we seek to remedy a problem with the way you think; in the former, we seek to remedy the sort of evidence you’re receiving; and it’s not hard to see that different policies might serve those different ends.
That’s one political reason for retaining the internalist notion of justification along with its externalist cousin. To see another, it’s useful to note how positively valenced normative concepts—like rationality, justification, and knowledge—work in epistemology. Most often, they apply to whatever doxastic item it is that they concern—beliefs, or the mechanisms by which they’re formed, or social institutions for the dissemination of information, etc.—whenever that item has a particular constellation of properties that we consider valuable. For instance: if someone knows something, that implies that it’s true and truth is something we value in a belief; what’s more, it implies that the fact that they believe this truth is somehow non-accidentally related to the truth itself, and again this connection between our beliefs and the world is valuable; and so on. And justification is no different. The debate between internalists and externalists is a debate about which valuable properties a belief must have in order to count as justified. Both sides, I think, agree that the properties the other side requires are valuable—indeed, some internalists will appeal to the value of having true beliefs when they argue that a justified belief must be part of a coherent set of beliefs, pointing out that only if the set of beliefs is coherent could all of them be true. They just disagree about which count for justification. As Srinivasan says herself, it would surely be better for each of the individuals from above who reliably detected the classism, misogyny, and racism that they experienced to be able to say, moreover, what it was that made their experience classist or misogynist or racist; to be able to pick out the features of it on which they based their belief, and to know why those features support the conclusion they came to. Once we see this, we can also see why the internalist’s notion of justification, and the valuable features that it requires of a belief, are also crucial for the radical project to which Srinivasan wishes to recruit externalism.
Return again to the individual experiencing classist behaviour at an Oxford college. We suppose that he has a reliable mechanism for detecting classism and forms his belief on this basis. We also suppose that he isn’t aware of this mechanism: he isn’t aware of how it works or what features of a situation it responds to; he isn’t aware of its reliability. While it certainly speaks in favour of his belief that it is formed in this way, we would surely not feel that our radical project were complete if all judgments about classism, racism, or ableism, misogyny or sexism, homophobia or transphobia were of this sort. When we raise consciousness about these oppressions and prejudices, we do not wish only that people form beliefs about them that are justified by the externalist’s lights. We want people to understand what is racist about what they experience; what it is that the reliable mechanism is picking up. If all we wanted was externally justified belief, testimony from a reliable source would suffice. It would be enough for a racism guru to whisper in a person’s ear each time they experience it or witness it. But, as the tradition of radical pedagogy has surely convinced us, understanding for ourselves the way our social world is structured is more valuable than simply knowing it on the basis of expert testimony. What we want over and above reliability are the things that the internalist’s account of justification demands: awareness of the evidence; awareness of why it is evidence for the belief; understanding of the phenomenon that our belief concerns.
Of course, you might reply that this tells in favour not of retaining both the externalist and the internalist concepts of justification but instead of retaining the externalist one, while adding also a concept that is essentially the conjunction of the externalist and internalist one. After all, what our argument in the previous paragraph shows, if anything, is the political usefulness of a concept that holds of a belief if both the internalist and externalist accounts are satisfied: the belief is formed reliably, and you’re aware of that. That is, what the internalist demands is what we want only after what the externalist demands is already in place. If this were the only consideration in favour of the internalist concept, that might be true. But when we combine this argument with the arguments above in favour of the internalist concept, it is clear that what is needed is the internalist concept and the externalist one; and with both of those in hand, we can conjoin them when we need to speak of the goal of raising consciousness.
What goes for justification here may well go for other concepts, such as knowledge and rationality. In general, when philosophers apply the method of conceptual analysis to positively valenced normative concepts in epistemology or other fields—justification, knowledge, art, rational, reasonable, moral, prudential, etc.—their analyses are usually valuable not because they trace the precise contours of the single concept to which we’ve all been referring with these words, but because in attempting to reveal those contours, they’ve enumerated valuable features of whatever the sort of thing is to which the concept applies. As in the case of justification, then, it might well be that when we do this, we reveal that there are a number of ways of collecting those features together to define concepts that are practically or politically useful. Just like the concept of justification, other normative concepts may well end up splitting into many versions.