Wednesday, March 14, 2007

Knowing Your Own Thoughts

Readers familiar with my research will know that I'm pretty pessimistic about the reliability of introspection, even of currently ongoing conscious experience -- e.g., ongoing visual experience, imagery, emotion. Regarding our own thoughts, though, you might think it's plausible to suppose that our introspection is excellent.

I'm not so sure.

It's not clear that, even if we reach accurate judgments about our on-going thoughts, the accuracy of those judgments is due to a genuinely introspective process -- if we take introspection essentially to involve something like detection. Rather, the accuracy may be more a matter of self-attributions of thoughts being self-fulfilling, as statements like "I'm saying 'blu-bob'" are self-fulfilling. Or, similarly, our self-knowledge of our thoughts might be like the driver's knowledge of whether he'll be going left or right at the coming intersection (to borrow an example from Tori McGeer) -- knowledge we have as a result of our capacity to shape our behavior, or thoughts, to accord with our judgments about what our behavior or thoughts will be. Or maybe (with Dorit Bar-On and others) the accuracy of our self-attribution of thoughts is due to the fact that our self-attributions are simply expressions of our thoughts (like "that hurts" is an expression of pain, no more introspective than "ow!"). What all these accounts have in common is that our accuracy in self-attribution is not due to accurate detection in an introspective sense.

But there's something left out here, too. For it seems that sometimes we have accurate knowledge of recently past thoughts. Because the purported thoughts are in the past, our accuracy can't be due to self-fulfillment or self-shaping or self-expression. Of course, it's not clear that it's due to introspection exactly, either -- since ordinarily we think of introspection as a means of detecting what's currently ongoing in our minds, not what happened in the past. But the fact remains, whether we call it introspective or not, we do seem to have some accurate knowledge of recently past thoughts.

But how accurate, I wonder? I'm feeling pretty confident that about a minute ago I was thinking about getting some tea. But should I trust this confidence? People are also pretty confident in their judgments about imagery, dreams, visual experience, etc., even when they are quite plausibly mistaken. We are never proven wrong in our self-attribution of past thoughts, and maybe that underwrites our confidence -- but we have no test for the accuracy of such self-attributions, so of course we won't be proven wrong, no matter how wrong we actually are! Is there really some basis for thinking that our memories of our recent thoughts do generally accord with the thoughts themselves?


Tanasije Gjorgoski said...

Hi Eric,

Some might say that we can't be sure of our out thoughts two minutes ago, because we can't be sure that we existed two minutes ago. (e.g. the memories of the swampman about two minutes ago)

Intuitively it seems to me that there is something wrong with such possibility. I'm not sure why though. One idea that comes to mind is that maybe I'm aware of the whole 2minutes time span somehow.

Joachim Horvath said...

Hi Eric,

to suggest that we may not even know (in most cases) what we thought two minutes ago seems to be a pretty drastic skeptic hypothesis. For, if it really were true, then any more complex reasoning or planning would be impossible (at least if carried out in thought). But such complex reasoning and planning is possible and even quite common. Therefore, we do typically know what we thought two minutes ago.

But if one really had a serious concern about the reliability of introspective memory, one could test it in the following way: Write down your present thought, hide it, wait two minutes, write down what you think you thought two minutes ago and compare it with your first note. [If one worries about the interference of visual memory here, one can also do this by recording an utterance of one's thought in such a way that oneself doesn't hear what one utters while uttering it - and then go on as before.]

Eric Schwitzgebel said...

Thanks for the comments, Tanasije and Joachim!

I do suppose one could doubt that the world existed two minutes ago, but that doubt doesn't usually tempt me. I'm not much of an external world skeptic. I think it's not only possible, but maybe even reasonable, to suppose we have only poor knowledge even of our recent thoughts. Or at least I'm playing with that idea.

There is an idea of a "specious present" in some of the literature on time consciousness -- the idea of consciousness of "now" being spread out over time somehow -- but the literature on this is pretty disorganized and hard to evaluate.

Tyler Burge makes an argument similar to yours, Joachim. But I'm not sure this kind of self-knowledge is really necessary for reasoning. For one thing, it seems we can imagine a young child or an artifical life form reasoning without making judgments about its stream of thought. For another, there's a threat of a regress: If I need to know that I believe P and P->Q to conclude Q, then might it also be the case that I need to know that I believe that I believe P and P->Q? Our thought processes, it seems, can be in some way attuned or responsive to earlier thoughts without our being able to reach accurate judgments about those thoughts.

That last idea rests on a distinction I'd like to draw between "attunement" and "judgment", which will hopefully be the topic of an upcoming post!

michael metzler said...

Eric writes: “Our thought processes, it seems, can be in some way attuned or responsive to earlier thoughts without our being able to reach accurate judgments about those thoughts.”

This seems to be a part of just what it is like to think a thought, part of its proprietary phenomenology perhaps. Looking forward to the next post on “attunement”.

Anonymous said...

Hi Eric,

First time viewer of your blog... Absolutely wonderful reading - thank you for that.

I believe you're very right to question our own thoughts. Our thoughts are nothing more than tools used by our sub-conscious mind, within the conscious "workspace". I think we would all do really well to question the validity of our own thoughts. Not so much if they really happend or not, but rather if they are true.

I don't believe our introspection is excellent at all. In fact I think it's more liable to confuse ourselves and muddy up the original intention behind the thought than anything else.

Is there really some basis for thinking that our memories of our recent thoughts do generally accord with the thoughts themselves?

I think so, personally. We attatch meaning/significance to our thoughts much like we would anything else. If I was thinking "Green Tea, mmmmm smells nice" five minutes ago, and I take note of myself thinking that (either by inhereant significance of my observing my own thougths, or my associative significance with Green Tea itself (perhaps because I haven't had green tea in five years, the thought (or expression) is a matter of significance to my mind).

Of course you can replace Green Tea in the above example with anything. What matters is that when we attatch meaning to a thought (either by observing the thoughts implicitly, or by the thought having some sort of relevant significance, usually due to compounded significance via related thought imprints from the past) we give it a history. Through it's relative significance to you, it passes through time as an imprint in our memories. We can be as sure about that imprint than we can about anything else we experience. Because whether it's introspection or external speculation it's all just imprints on the memory anyway. Some are more significant to you and those will be the ones to be trusted the most, I feel.


Eric Schwitzgebel said...

Thanks for the kind words and comments, Michael and anonymous!

I do agree with you, anonymous, that if we particularly reflect on or "observe" a thought as it is ongoing, that probably increases the likelihood of remembering it correctly.

One type of data that seems to support that conclusion is the research of Ericsson and Simon: When people are asked to report on their methods of solving a problem, their reports of their methods often accord with the methods one can plausibly infer they actually used (based on reaction time, type of answer, etc.). This suggests that when we're primed to notice how we think about something, we often do characterize our thoughts correctly.

I wonder if Ericsson and Simon have some cases where they ask people in retrospect to reflect on how they solved a problem, without the people being aware in advance that they'd be asked that? Hm! I should go check!

Anonymous said...

The thread reminds me of a bumper sticker seen recently--
"You don't have to believe everything you think."