Introspection in logical argument

Descartes' mechanical baby in white on blue

The claim

A friend and I had a conversation recently on the use of introspective evidence in logical arguments. Here's the claim:

Introspective evidence cannot be used in sound logical argument because statements that are the result of introspection have unknown truth-values.


  1. Premises used in sound logical arguments must be definitively true.
  2. Statements derived from introspection are sometimes true and sometimes not true.
  3. Introspection alone cannot always determine whether a statement derived thusly is true or not true.
  4. Statements that could either be true or not true have unknown truth-values (are not definitively true).


5) By (2), (3), and (4), statements derived from introspection are not definitively true.

6) By (1) and (5), statements derived from introspection cannot be used in sound logical arguments.

Support for the premises

  1. In formal deductive arguments, for a conclusion to necessarily be true, the argument must be both valid and sound. If valid, the conclusion cannot be false if all the premises are true -- i.e., the premises entail the conclusion. If sound, all the premises are true. Arguments can easily be valid but not sound if one or more premises are false. To know for certain that the argument is sound, one must know for certain that all the premises are true and that the argument is valid.

  2. (a) Definition of introspection: I am defining 'introspection' as a process that gives one knowledge about what's going on in one's own mind. This process may involve self-observation, self-reflection, and self-reports of one's mental states and internal motivations, as well as explanations and reasons for one's thoughts, feelings, and behaviors. This is not a universal definition. (See, e.g., the Standford Encyclopedia of Philosophy for a decent discussion of introspection.) Not all thoughts are introspective; one can think about things other than one's mental states, motivations, feelings, and the causes of these.

    (b) Definition of truth: I am concerned primarily with the accuracy or truth of introspection. For now (until I do more reading and possibly change my mind about what makes sense), I am adhering to a correspondence theory of truth: if one says the world is flat and it is indeed flat, then it is true that the world is flat. If one's internal representation of a thing in the world is consistent with the thing in the world whenever it occurs, then there is a correspondence and the thing is true. E.g., one's introspections about the causes of one's behavior are true if they accurately reflect the causes of one's behavior.

    The problem with this is that we don't know when we've made a mistake, and we don't know for certain that a thing is true even if it is true. The pragmatic theories of truth set out by Peirce, James, and Dewey attempt, in different but related ways, to deal with this issue. You can know when something's false. You can't prove that something's true. I'm planning on reading up on their respective theories.

    Other theories of truth are less appealing (also see Moser & vander Nat, 1987).

    (c) Truth of introspection: The Stanford Encyclopedia of Philosophy summarizes many relevant studies of why introspection doesn't always give us accurate information. It includes descriptions of studies of introspection about causes of behavior, about attitudes, and about conscious experience.

    The seminal paper on why we're not very good at discerning the causes of our thoughts, behaviors, and their underlying processes was Nisbett and Wilson's (1977) work. They re-analyzed the rope-swinging experiment (see this blog for a lovely description of that experiment), which was originally done by Maier in 1931 (see this pop science discussion of Maier's work), as well as many others.

    Wilson has written a bunch of other papers on the subject since.

    There's also Gazzaniga's (1995) fairly famous studies of split-brain patients. Although some might argue that the split-brain research is not relevant to the normal human with the un-lesioned brain, it's just one piece in a larger puzzle, and there's a decent amount of other evidence drawn from pretty average populations of people.

    One might argue that some results of introspection can be trusted. That could be true. How will we know which introspections are trustworthy? I'd be wary of assuming wrongly. The safest assumption is that we don't know whether our introspections accurately reflect the causes of our behavior, our attitudes, or anything else.

  3. As seen in part (2c), we rarely if ever know whether our introspection has resulted in a true statement. Introspection alone cannot reveal the truth-value of our observation (e.g., of our motivation for performing a particular behavior). If you bring evidence from an outside source, e.g., the observation that the experimenter set the rope swinging or that Gazzaniga primed his split-brain patients with a word shown only to one hemisphere, then the truth-value has no longer been determined merely through introspection.

  4. Fairly straightforward, but the change in wording necessitated a further premise.

With all that in mind...

Let's look at my claim again:

Introspective evidence cannot be used in sound logical argument because statements that are the result of introspection have unknown truth-values.

Please note that I am not claiming that introspection fails to be useful in cognitive science, philosophy, psychology, or where have you. It can be a great tool, particularly when studying people's subjective responses to situations.

I am arguing that introspection does not necessarily accurately reflect the true causes of any mental state or process, and thus, that statements derived from introspection about mental states and processes cannot be taken at face value to be necessarily true. Secondly, we generally do not know when or if our introspection reflects true causes or whether our introspection is a confabulation, and as such, we cannot assume any given introspective statement is a true reflection. The truth of any given introspective statement is unknown. Introspective statements about different internal events may be more or less likely to be true (i.e., we may have better self-knowledge of, say, our feelings than we do of the causes of our behaviors).

Recall what I said earlier about sound logical arguments: all the premises must be true and the argument must be valid. If you cannot determine whether one or more of your premises are true, then you do not know whether your argument is sound. If you are using introspective evidence in one or more premises, you cannot know whether your argument is sound. Your argument must be sound for you to be sure that your conclusion is true.

Thus, if you are just using introspective evidence, you cannot know whether your conclusion is true.

And that's why I'd argue that introspective evidence cannot be used in sound logical arguments.


Gazzaniga, Michael S., 1995, Consciousness and the cerebral hemispheres, in The Cognitive Neurosciences, Michael S. Gazzaniga (ed.), Cambridge, MA: MIT, 1391-1400.

Maier, N. R. F. (1931). Reasoning in humans: II. The solution of a problem and its appearance in consciousness. Journal of Comparative Psychology, 12(2), 181-194.

Moser, P., & vander Nat, A. (1987). Human Knowledge. Oxford, NY: Oxford University Press.

Nisbett, Richard, & Wilson, Timothy. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84, 231-259. [PDF]

Tuesday, September 27, 2011 - tags: cognitive-science philosophy


  • said: Reply

    Interesting. I think you are right that we cannot rely on introspection alone. It is just a starting point. As to truth, I give up and go with the pragmatists. Truth is what works for now until we get a better version. I don’t think we can really know.

  • said: Reply

    I agree that this definitely holds for deductive arguments, but that doesn’t mean that introspection doesn’t provide *evidence*; it can still be a good foundation for induction. (Bayes’ rule is the key to all epistemology, etc.)

  • said: Reply

    @Jaime: I need to read/think more about truth before I go entirely with the pragmatists, but you may be right.

    @WrongBot Indeed. (I was trying to think of more to say than that, but I agree with you, so.)

  • said: Reply

    isn’t correspondence kinda circular?. something is proven true by comparing to something assumed to be true. in the end, maybe we have just best guesses of what is true – best because they have been working for us humans so far – but we might have to change them again. thanks for the topic. interesting.

Comments are closed.