Here’s something that Leonard Peikoff said:

The “stolen concept” fallacy, first identified by Ayn Rand, is the fallacy of using a concept while denying the validity of its genetic roots, i.e., of an earlier concept(s) on which it logically depends.

Ayn Rand rarely explained what she meant with her more obscure phrasings; she seems to have taken for granted that she was perfectly clear. The same is true of her followers. So it’s with some caution that one should try to say what they mean by anything that they say; the most sincere effort at interpretation is routinely met with accusations of intellectual dishonesty. In this case, one might want to know what it is to “use” a concept or what it is for one concept to logically depend on another or what it is for the concepts on which a concept logically depends to be “valid”. We can only guess. Allowing C1…Cx to be words that express concepts:

One uses C1 by saying, of some x, that x is C1.
C1 is valid if, and only if, for some x, x is C1.
C1 logically depends on C2 if, and only if, C1 is valid only if C2 is valid.

Assume that that is correct. Peikoff says that Descartes commits the fallacy of “stolen concept”:

Men have been wrong, and therefore, [Descartes] implies, they can never know what is right. But if they cannot, how did they ever discover that they were wrong?

(As a matter of Descartes exegesis, that is incorrect; Descartes argues that, since error is possible, our means of cognition must be basically reliable.) It seems that one steals the concept of C1 if, and only if:

1. One uses C1.
2. One says that, possibly, for no x, x is C2.
3. C1 logically depends on C2.

For example:

Descartes uses ‘wrong’.
Descartes says that, possibly, for no x, x is right.
‘Wrong’ logically depends on ‘right’.

I don’t know why “contradicting yourself” needs a special new name, unless it’s for the parallelism between theft of material goods and “theft” of cognitive accomplishments, but at any rate, certainly one should not “steal concepts” as (I hope) Rand and her followers would define that phrase.

How would one discover the theft of a concept? Condition 3 is never apparent to observation. And one might satisfy conditions 1 and 2 without realizing it, by satisfying them at very different times. If, for instance, on one day I say that some acts are just, and on the next day I say that nothing is virtuous, then we might not notice that I’ve stolen the concept of justice (nothing is just if nothing is virtuous) because of the time gap. How can one shrink that gap?

Here’s an approach. When someone uses C1 — says, of some x, that x is C1 — you might ask him why, or how he knows, that x is C1. If his attempt to state criteria for the application of C1 involves some concept C2 that he says is invalid, then you have shown that he has stolen a concept. Thus, asking “Why?” questions is a tool for the discovery of conceptual dependencies and other relationships; one way of putting this is that answering “Why?” questions legitimates one’s ownership of a concept.

To ask “Why is that the case?” is not necessarily to ask, “Is that the case?” And neither of those questions is necessarily a way of saying, “That is not the case.” American English emphasizes hesitance and deference, so we are often uncomfortable telling someone that he is wrong or asking him to prove that he is right. Thus we will sometimes fail to say “You are wrong” but instead say, “Is that so?” And, further, instead of challenging a belief directly, we will sometimes ask “How do you know that that is so?” when we are trying to deny that that is so. (This is the same sort of usage that leads us to ask someone whether he can pass the salt, instead of saying “Pass the salt” or even “Please pass the salt.”) Nevertheless, philosophers (and scientists) routinely ask “Why?” questions because we want to know about conceptual dependencies and hierarchies when we have no doubt at all about the truth of the claim about which we are inquiring.

What that all implies is that refraining from answering a “Why?” question in a philosophical context routinely conceals the stealing of a concept. Why would one refrain from answering a “Why?” question?