In his "Thinking it Through" textbook, Appiah writes

It is a big step from saying that some of our mental states are things that other people can know about, to saying, with the behaviorists, that all of them must be in this way public. Yet one of the most influential philosophical arguments of recent years has just this conclusion. The argument was made by the Austrian-born philosopher Ludwig Wittgenstein, whose work we will discuss again in the chapter on language. Wittgenstein began by supposing that anyone who believed in the essentially private thoughts of Descartes’ philosophy of mind would find it quite acceptable to suppose that someone could name a private experience—one, that is, that nobody else could know about. And indeed, as we shall see in Chapter 3, Thomas Hobbes, who was an English philosopher who reacted against some of Descartes’ ideas, thought that we used words as names of our private thoughts in order to remember them. He called them “marks” of our thoughts. To use marks in this way, someone would have to have a rule that they should use the name just on the occasions where that private experience occurred. Wittgenstein argued that obeying such a rule required more than that there should be both circumstances when it was and circumstances when it wasn’t appropriate to use the name. He thought that it also required that it should be possible to check whether you were using the name in accordance with the rule. And he offered a very ingenious argument that was supposed to show that such checking was impossible. If Wittgenstein was right, there could be no such “private languages.” And his argument is called, for that reason, the private-language argument.

I am trying to follow the line of argumentation. Is this meant to be a reductio argument? Is Appiah saying "suppose we accept that such private thoughts exist. Then this implies (according to Hobbes) that we have private 'marks'. Wittgenstein then says that these 'marks' can only exist if we have a way of checking(i.e. the existence of 'marks' implies the ability to check), and yet such checking is impossible -- therefore private mental events are impossible"?

  • You can interpret it as a reductio at this surface level, but Appiah does not really give a substantive idea of what the argument is in this passage. For that, see SEP, Private Language Argument.
    – Conifold
    Aug 31 at 19:57
  • Oh yes, I agree that Appiah does not state the private language argument here (I think that’s the very next chapter). What I suppose I’m trying to ascertain is the point of the private language argument. Am I correct in interpreting its conclusion as a reductio against the notion of having completely private mental events at all? @Conifold
    – EE18
    Aug 31 at 20:25
  • The point, yes, but the argument is more subtle than that. Wittgenstein argues not so much that assuming private language leads to a contradiction as that the very notion of it is nonsensical in the first place, and falls apart when thought through.
    – Conifold
    Aug 31 at 23:01
  • Got it, thank you! And that incoherence is precisely the “contradiction” arrived at in the reductio? I’d be happy to accept an answer to that effect if you’d like! Thanks so much for your help. @Conifold
    – EE18
    Sep 1 at 0:34
  • 1
    Private mental events are certainly possible if not necessary. There're basically two kinds of private mental events, access-consciousness (a-type) and phenomenal-consciousness (p-type), by the naming alone one can understand the former can be accessed via the felicitous mastery expression of public languages or other mime mediums (such as holding a white flower) or even medium free (Rumi's soundless sound), while the latter cannot be expressed and accessed as formulated in the famous contemporary hard problem of consciousness which seems relevant to your book's topic... Sep 1 at 23:38

1 Answer 1


See this answer to a closely related question: How serious are believers in the private language argument?

Your question is focused on what appear to be the flaws in W's "error correction" argument. As the SEP article admits, the "error correction" part of W's argument is -- unconvincing -- and if it WERE convincing, it would also prohibit public language.

The SEP article and Conifold argue that there is instead some "logic" problem with private language, although neither SEP nor Conifold specify clearly what this problem is. W's focus tended to lean him toward behaviorism and the Verifiability Principle, which basically cannot be satisfied for anything that is intrinsically private. An alternate "logic" problem is to assume that language exists to communicate, and communication requires others, so must therefore be external.

In general, what exists in our world is an empirical question, not a logic one, and we appear to use private language (for instance, I do a visual/structural set of space transforms in my head to figure out how to pack a set o items, which I need to translate internally to share externally, and the internal knowledge and transforms are completely private), which should trump any such logic argument. Digging in a bit more on the logic claim, the potential for one to use language to communicate with oneself refutes this second "logic" claim, and the rejection of both behaviorism and the Verification Principle in philosophy and psychology invalidates the first, so there do not actually appear to be any such "logic" constraints prohibiting private language.

  • Thank you for this very interesting answer!
    – EE18
    Sep 25 at 17:21
  • 1
    The standard model of communication is something like the transmission of information from one point to another. Doesn't that make the idea that talking to oneself is a bit problematic? (I am NOT suggesting that we do not talk to ourselves.)
    – Ludwig V
    Sep 25 at 17:31
  • @LudwigV -- Hume argued a "bundle theory" model of self, that we are different selves instant to instant. Under bundle theory, we NEED language to communicate with our future selves! Most philosophical ideas contain kernels of truth, and in many cases they then overgeneralize, and I think this is true for Hume and bundle theory. But the utility of bundle theory to show how self-communication is not only not a contradiction, but actually ESSENTIAL is -- still valid.
    – Dcleve
    Sep 25 at 18:11
  • 1
    So talking to oneself is just one self talking to another? Still, if both selves share some memories, one self saying to another "Your train is due at 9:20" is a bit puzzling. Or does each self have its own memories?
    – Ludwig V
    Sep 25 at 20:41
  • @LudwigV I do not endorse Hume's bundle theory, but it does capture an aspect of the diverse/random non-contiguous nature of at least some of our thinking. Talking TO oneself helps to focus and channel our sometimes scattered thoughts in a particular direction, and reinforce a conclusion we want to carry out. Is this communicating with oneself? I would think it has to be.
    – Dcleve
    Sep 25 at 20:56

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .