If you put a fish in a raincoat, it remains a raincoat. It's a raincoat worn by a special kind of animal, but it's not a special kind of raincoat. If you find it particularly amusing to see a fish wearing a raincoat, you might intentionally put the raincoat on the fish because it's the special kind of animal it is. But it is still not thereby a special kind of raincoat. Finally, you might put the raincoat on the fish precisely to draw attention to the distinctive – water-dwelling – features of the animal. Even so, it's just a raincoat.
I want to suggest that a certain kind of argument that adult human knowledge is a distinctive kind of knowledge commits an error similar to the error in thinking that raincoats worn by fish are distinctive kinds of raincoats. Knowledge that adult human knowers have is knowledge had by a distinctive kind of animal. It is not, thereby, a distinctive kind of knowledge.
In Knowledge and Its Place in Nature, Hilary Kornblith argues for what Michael Williams refers to as Kornblith's “austere reliabilism” – that knowledge is a matter of reliability only and does does not require “justification, at least when ‘justification’ is understood in the sense of epistemic responsibility'” (Reference Kornblith2002: 90). Williams' “What's So Special About Human Knowledge?” is a plea to retain the responsibility requirement on knowledge. Williams' responsibility requirement admits, he says, of three senses, but all three of the senses characterize the kind of epistemic authority required for knowledge. Two senses will not concern us here: (1) “doing one's due diligence” – that is, being responsive to epistemic norms and (2) being subject to epistemic sanction. The sense that will concern us here, and the sense in which the responsibility requirement captures what is distinctive about adult, rational, human knowledge – is accountability. Accountability for what one believes is what characterizes the kind of knowledge that is reserved to humans.
The disagreement between Williams and Kornblith on the responsibility requirement is reflected in their view of the relation between animal and human knowledge. According to Kornblith, animal and human knowledge are continuous. According to Williams, because animals are not responsible for or accountable for their beliefs, human knowledge is distinct in kind from animal knowledge. One way to resist Williams on this might be to claim that animals are accountable in just the way humans are, and so accountability can't be what's distinctive about adult human knowledge. But if one is not tempted to go this way, and one still wants to claim that human knowledge is continuous with animal knowledge, it looks like one will want to resist Williams' claim that human knowledge requires accountability, which is just what Kornblith does.
What is Williams' argument that knowledge requires responsibility/accountability? I take it that the argument is linguistic. He says, “the best way to get a grip on the concept of knowledge is to examine our practices of claiming and attributing knowledge.” The feature of our practice that he has in mind, I think, is our tendency to, when we attribute knowledge, attribute also accountability. That is, he is taking as a premise that attributions of knowledge have the function of conferring epistemic authority on the subject of the attribution. When I say that I know when the next bus is coming, I claim epistemic authority for myself. When I say that I have no idea when the next bus is coming, but that you know, I'm claiming authority for you (and all that this involves – including accountability).
I don't mean to be interpreting Williams as offering a standard Ordinary Language argument. Such an argument might go like this:
(1) We attribute “knowledge” that P to S when and only when we believe S is accountable for their belief that P.
(2) Therefore, S knows that P only if S is accountable for their belief that P.
As with all such arguments, the success of the argument is contingent on the disposition of various warranted assertability maneuvers. That done, the argument, suitably spelled out, might provide strong evidence for its conclusion, if the premise were true. But I don't think that's the argument Williams wants to make. Nor should it be Williams' argument, because the premise is false. We don't attribute knowledge that P to S only when we believe S is accountable for their belief that P. That is, we don't use sentences that predicate “knowledge” of some subject only when that subject is accountable. We, after all, attribute “knowledge” that P to animals and to humans who aren't accountable.
Start with animals. I can say that the bomb-sniffing dog “knows” which suitcase to avoid. I do defer to the dog in this case, though I don't attribute accountability to the dog. One might think, when we attribute knowledge to humans, we do more. We also attribute accountability. But not always. Sometimes we attribute knowledge to humans without thereby attributing accountability. It's not that humans lose the ability to have animal knowledge when they acquire the ability to be accountable for their beliefs. We can attribute knowledge to adult humans of the sort we attribute to the bomb-sniffing dog. So, the claim has got to be that we often or sometimes attribute accountability to adult humans by attributing knowledge to them.
When do we do this? Presumably it will depend on the context in which we attribute knowledge, and on what is presupposed between the knowledge-attributor and their interlocutor. If it's presupposed that the subject of the attribution is an adult, rational, reflective subject, capable of accountability for the very belief knowledge of which is being attributed to them, and it's presupposed that the subject is being normally attentive to the issue about which we're attributing them knowledge, then when we attribute knowledge to them, it's plausible that we thereby attribute accountability.
But how do we get from this fact to the further claim that what's distinctive about human knowledge is accountability? The proper conclusion to draw from the linguistic facts seems to be, rather, at most this: that when an adult, rational, human believer is capable of accountability for some belief, then they know that the belief is true only if they're accountable for it. But this doesn't show that the kind of knowledge they have is special. After all, perhaps rational beings capable of accountability for their beliefs tend to be accountable for those beliefs particularly when they have animal knowledge that the belief is true. At most, then, we only have license to conclude that there is something special about the adult rational human believer – the one who is attributed knowledge.
We – and Kornblith – can certainly grant this: there is a kind of being who is accountable for its beliefs in circumstances in a way that animals are not. When we attribute knowledge to beings mutually assumed to be of this kind and in those contexts, we thereby attribute accountability. But this is not to say that we attribute a distinctive kind of knowledge to that being. You've attributed knowledge to an epistemically special kind of being. And if it's a shared presupposition between you and your interlocutor that the being is special in this way, then you'll agree that when the being has knowledge, the being will have special status vis-à-vis the proposition knowledge of which has been attributed to the being. But this won't show that the kind of knowledge attributed is especially different from the kind of knowledge attributed to animals. The premise we should grant to Williams, then, is not 1, above, but rather this one:
(1*) By attributing knowledge that P to S, we thereby attribute accountability to S, when there is a presumption that S is a rational human capable of accountability for the belief that P.
What follows from that premise is not 2, above, but rather at most this:
(2*) If S knows that P and S is an adult human capable of accountability for the belief that P, then S is accountable for P.
This conclusion could be true as much because such subjects are accountable for what beliefs they have as because there is a distinctive kind of knowledge that entails accountability. It may very well be that what makes humans distinct from animals (or, one thing that does) is that humans are accountable for what they know while animals are not. But this claim is consistent with both the view that animal knowledge is continuous with human knowledge and the view that human knowledge is different in kind from animal knowledge in requiring accountability. Only the weaker thesis – that human knowers are accountable in ways that animal knowers are not – is motivated by the assumed premise about linguistic “knows”-behavior.
A similar point can be made taking off from Linda Zagzebski's espresso machine example (Reference Zagzebski2003: 13): if you proudly show me your brand new, fabulously reliable espresso machine, I might say, after tasting a cup, “that is a fabulous cup of espresso.” By saying this, I'll be saying something about the cup of espresso and be indirectly praising the machine: I know you want praise for your new machine, and I know that one way to provide that praise is to praise the output. I might also truly say it about some fluky output of what we mutually agree and know to be a terribly unreliable and awful machine. I don't thereby commit myself to saying that there are two kinds of good cups of espresso, one produced by bad machines, and one produced by good. Nonetheless, when I attribute good cups to what are mutually agreed to be good machines, still, I can attribute worth to the machine. Likewise, when I say of an adult human subject who is accountable for their beliefs in a context in which they are so accountable that they know that P, I indirectly say that they are accountable for the belief. But that's just because, we both know, such subjects are accountable for the beliefs they have in such contexts – not because the kind of knowledge I'm attributing to the subject requires accountability.
The worry generalizes. Take any linguistic argument for a necessary condition on distinctively human knowledge that proceeds from a premise about when we ascribe “knowledge” distinctively to humans.
(A) When we attribute “knowledge” distinctively to humans we thereby characterize them as having feature F.
(B) Therefore, S has distinctively human knowledge only if S has feature F.
It may be that the conclusion follows from the premise (implicit premises suitably defended and WAMs suitably disposed-of). But, if so, we can't assume that the conclusion tells us something about what distinctive human knowledge is, or even whether there is a kind of knowledge only humans can have. To answer Williams' title question: what's so special about human knowledge? Answer: it's had by human knowers.Footnote 1
