Monday, September 1, 2014

Understanding human behavior

A question that came up prominently during the seminar in Mexico has been discussed recently also by Jon Cogburn at NewAPPS. The question is about what is involved or required for understanding human behavior.

Jon Cogburn says that:
One could say that given a set of discourse relevant norms held fixed, understanding in general just is the ability to make novel predictions. For Davidson/Dennett, we assume that human systems are largely rational according to belief/desire psychology and then this puts us in a position to make predictions about them. We make different normative assumptions about functional organization of organs, and different ones again about atoms. But once those are in place, understanding is just a matter or being better able to predict.
I'm writing this as a post of my own rather than as a comment at NewAPPS partly because I suspect it will be too long for a comment and partly because I don't know what all of this means and don't want to appear snarky or embarrassingly ignorant. I genuinely (non-snarkily) don't know what it means to hold fixed a set of discourse relevant norms, nor what it means to put normative assumptions in place. But what I take Cogburn to be saying is, in effect, that "understanding in general just is the ability to make novel predictions." 

Winch says that being able to predict what people are going to do does not mean that we really understand them or their activity. He cites Wittgenstein's wood-sellers, who buy and sell wood according to the area covered without regard to the height of each pile. We can describe their activity and perhaps predict their behavior but we don't, according to Winch, really understand it. He surely has a point. Here's a longish quote (from pp. 114-115 of the linked edition, pp. 107-108 of my copy):
       Some of Wittgenstein’s procedures in his philosophical elucidations reinforce this point. He is prone to draw our attention to certain features of our own concepts by comparing them with those of an imaginary society, in which our own familiar ways of thinking are subtly distorted. For instance, he asks us to suppose that such a society sold wood in the following way: They ‘piled the timber in heaps of arbitrary, varying height and then sold it at a price proportionate to the area covered by the piles. And what if they even justified this with the words: “Of course, if you buy more timber, you must pay more”?’ (38: Chapter I, p. 142–151.) The important question for us is: in what circumstances could one say that one had understood this sort of behaviour? As I have indicated, Weber often speaks as if the ultimate test were our ability to formulate statistical laws which would enable us to predict with fair accuracy what people would be likely to do in given circumstances. In line with this is his attempt to define a ‘social role’ in terms of the probability (Chance) of actions of a certain sort being performed in given circumstances. But with Wittgenstein’s example we might well be able to make predictions of great accuracy in this way and still not be able to claim any real understanding of what those people were doing. The difference is precisely analogous to that between being able to formulate statistical laws about the likely occurrences of words in a language and being able to understand what was being said by someone who spoke the language. The latter can never be reduced to the former; a man who understands Chinese is not a man who has a firm grasp of the statistical probabilities for the occurrence of the various words in the Chinese language. Indeed, he could have that without knowing that he was dealing with a language at all; and anyway, the knowledge that he was dealing with a language is not itself something that could be formulated statistically. ‘Understanding’, in situations like this, is grasping the point or meaning of what is being done or said. This is a notion far removed from the world of statistics and causal laws: it is closer to the realm of discourse and to the internal relations that link the parts of a realm of discourse.
Winch talks elsewhere (see p. 89, e.g.) about what it takes for understanding to count as genuine understanding: reflective understanding of human activity must presuppose the participants' unreflective understanding. So the concepts that belong to the activity must be understood. Without such understanding all we will generate is (p. 88) "a rather puzzling external account of certain motions which certain people have been perceived to go through."
But does Winch get to say what counts as genuine understanding? This was a point we discussed when I was at the University of Veracruz. Several students there seemed to want less than what Winch would accept as true understanding of human behavior. They did not want to empathize. They wanted to identify patterns, and if they were able to do so well enough to be able to make accurate predictions then they would be very satisfied. A "rather puzzling external account of certain motions" is basically what they were hoping to produce, as long as it allowed them to make accurate predictions.

It looks to me as though Winch would resist or even reject such a desire, but can a desire be mistaken? And I don't know how to settle the apparent disagreement between Winch and others about what is and is not real understanding. Can we just say that as long as you see the facts you may say what you like? Or is that too easy?


  1. It seems like your question relates to what we would and would not be willing to call "understanding."

    Let's start with an example. There are now computer chess programs that can beat the best players in the world. Do they understand chess? The answer seems to me to be no. But the program meets the behavioral performance of understanding as well as any human chess master does; therefore the behavioral performance of understanding can't be the same thing as understanding itself.

    This is a fairly simple example, surely open to objections, but if I were arguing with these students I would want to know if they thought the computer's performance counted as understanding chess. If they did, maybe there wouldn't be anything else to say; we're just counting different things as 'understanding.' If they didn't and they still held on to the behavioral performance as sufficient for understanding, then we'd have something interesting to chew on.

    1. Yes, this seems exactly right to me. Except that the students in question weren't really talking about understanding, or at least not understanding people. They wanted to 'understand' (if that's the right word) social structures.

    2. raises the question of whether there are such "structures" to be understood or not/

    3. Yes. I never got a very clear, concrete idea of what they wanted, and there were several different people who perhaps wanted different things. But there was a consensus (I think) that they wanted something that Winch would not count as real understanding.

    4. When you talk about social structures I think e.g. of Levi-Strauss, recurring patterns of social organization that occur across different cultures and groups.

      I would tend to see such structures (in the mind of the social scientist) as tools for recognizing patterns in the behavior and organization of cultural groups. So - understanding a group of people would sometimes make it easier to see such structures, and spotting such structures would sometimes make it easier to understand a group of people. But no necessary connection either way, except maybe at some very minimal level.

      Although I suppose at some level I do want to say that as a practical matter the normal case is that behavioral pattern recognition and understanding people tend to track each other, at least somewhat.

      A poker player would only care (qua poker player) about understanding to the degree that it enabled her to make further behavioral predictions and thus extract more money from the people she was playing with. Which would in turn lead to the suspicion that 'understanding' was 'just' deeper-level pattern recognition. And from a certain point of view, which our poker player shares with some social scientists, that's all it can be.

      That would be a philosophico-cynical collapsing of understanding into behavioral pattern recognition that one could argue with. "How do we know it's not just..."

      But then when I think about the computer chess program this seems like an error. Or consider this: I have a missal and I watch a group of people celebrate the mass. I find that I can predict what they say quite well! But I grew up in China in a remote atheist household, learned English at university, and have never read the Bible or knowingly been exposed to any tenets of any of the Abrahamic faiths. Do I understand what they're doing?

    5. I guess it might also be useful to say that when I was studying with him Winch would sometimes point out that understanding was often used as a degree term, perhaps more often than it was used as a binary term. His paper "Can we understand ourselves?" I think connects to this issue peripherally.

    6. In other words, I don't think that Peter would have insisted that understanding was just one sort of thing - even though he would have opposed (and did oppose) the idea that "flat" behavioral-external pattern recognition was sufficient for all the sorts of cognition of other human beings we would want to call understanding.

      So verstehen as opposed to positivism, but not necessarily verstehen as this special state of mind we need susceptible to unitary definition.

    7. Thanks! This is very interesting.

      Yes, I think that understanding comes in degrees. In a sense that's obvious, but when it comes to understanding people they can be more or less alien to us, including cases where they are completely alien.

      I think I agree with you, but I don't know how to argue with someone who agrees that all they have is behavioral pattern recognition, and who insists that this is enough for them. It's like Dennett saying that thermostats think. He doesn't seem to be missing any facts, just choosing to use words in a different way. It is possible to argue with someone like that, but there is no guarantee that it will get you anywhere.

    8. I don't think Dennett would say that thermostats think, except to the extent that he is broadening the word "think" to say something about the underlying nature and status of thinking itself (which is what you seem to be suggesting, Duncan). His point would be that, in terms of the science, thinking is grounded in much more basic physical processes, a point that is sometimes argued by some philosophers.

      His point doesn't mean thermostats experience thinking about things as we do, however, but only that our thinking about things is grounded in the same sorts of basic physical processes as you find in more rudimentary mechanisms like thermostats. It's a scientific point which ought to be determinable by testing at some stage of inquiry. And it has some philosophical implications.

      If the point he's making turns out to be right, then such a use would make sense, if not for thermostats (whose "thinking" is so far removed from what we ordinarily mean by that term that it really would be out of bounds) then for other, more sophisticated entities which operate on a developmental hierarchy closer to ourselves. Dennett assumes it will turn out to be right because, he argues, it's the best explanation we have for how brains do thinking.

      If so, it has linguistic implications, too.

    9. I think his point is that it can be useful to talk about thermostats having beliefs and desires (e.g. when it's broken a thermostat might "think" or "believe" that a room is too cold when it really isn't), and that when it is useful we are perfectly entitled to do so. The question of whether thermostats really have beliefs and desires is, in Dennett's view, a mistake. I sympathize with this view, as long as pragmatism doesn't blind us to important differences between people and thermostats. As the saying (from Sidney Morgenbesser?) goes, pragmatism's OK in theory but it doesn't work in practice. Or rather, what I'm saying is, pragmatism should be rejected if it doesn't work in practice.

    10. Yes, I think that's right. Pragmatism works only when it does. I don't think we have a different take on Dennett on this issue either though perhaps we're expressing it differently.

    11. Agreed. And you have brought up some points that I think Dennett makes in more recent works that I am not familiar with, so your take on him is richer than mine. I'd have to read that work to decide how much sympathy I have for it.

  2. A question: What if someone said they were comfortable with requiring only statistical patterns and call that ‘understanding’ because we don’t really come across strange wood sellers, or not very often anyway?

    Would a Winchean answer then be that not paying attention to those strange cases runs the risk of failing to notice of a level or a kind of understanding—understanding from “within”—that is present also in non-strange cases?

    That is, is there a way of arguing that we are in need of a more robust notion of understanding? Is there a way to persuade people that they need something they don’t take themselves to need? (I think this connects with your question about whether desires can be mistaken.)

    1. Is there a way to persuade people that they need something they don’t take themselves to need?

      I think so. But the way that occurs to me to do this is to let them have what they think they want and then see (perhaps with some helpful nudging) that it isn't satisfying to them. I didn't have time to do that in this case, and of course I can't say with much confidence in advance that they would not have been satisfied with the results after all.

    2. And what do you say: do we need that more robust sense or not? - I mean, are there occasions that we don't need it at all? And if so, in what situations do we need it and in what situations we don't? - Is there a case to be made for the idea that we need it in the wood-seller's situation, but we don't need it in other cases?

    3. Also, I'm not sure how to characterize the difference b/w the two kinds of understanding. I see that the shallower one involves prediction, but does it involve anything else? And does the more robust notion not involve prediction? And if both involve prediction, what is the difference?

      Do you think that saying that the first sense of understanding is "external" and the second "internal" is the right way to describe the difference? If so, since (if I'm right) 'internal' and 'external' are metaphors, what are they metaphor for?

    4. I think we need the more robust sense, and am not sure that the less robust sense is any use at all. But I want to keep an open mind about that until I have a better sense of what those who want the less robust sense want to do with it. Actually there are probably multiple less robust notions of understanding. Searle's Chinese room understands Chinese in a non-robust way, and someone who can predict what the wood-sellers are going to do understands them or their behavior in a different non-robust way. That is, the room does not only predict, but the student of the wood-sellers only predicts. Or is this really just the same external kind of understanding? The predictor could probably interact successfully with the wood-sellers, after all.

      I'm not sure what the more robust kind of understanding involves besides, or beyond, prediction. Feeling at home in a practice, perhaps, but that's surely metaphorical too. And 'feeling at home' is putting it too strongly, because I can (robustly, or at least somewhat robustly) understand behavior that is still alien to me, albeit less alien than the wood-sellers' behavior. Engaging in the practices of some religion not my own is less strange to me than the wood-sellers' behavior, but still not something in which I feel at home.

      I can't answer your last question, but I think one's sense of self is involved. Does that help at all?

    5. Searle, though, argues that the Chinese Room doesn't understand Chinese in any way, robust or otherwise. He rejects the non-robust proposal as inconsistent with what we mean by "understanding" in every case -- even though I think Reshef is right, that all we really mean by "understanding" at times, is the less robust form of it he alludes to because we're generally really only interested in behaviors.

      But then the question is whether, when challenged to say if we really believed an entity that's no more than a Searlean "Chinese Room" (the kind that behaves as if it understands Chinese) really understands, we would continue to affirm that it does. If Searle is right (and I think he is in this case), then, if pressed on the matter, we would not want to say that there's understanding present. Searle's point is that by "understanding," all we really mean is what we apply the term to in our own case.

      I don't think that's quite right (there's certainly room for linguistic variance and we certainly do that when thinking about machine "intelligence" for instance) but it's worth considering. It just may be the case that we always do presume there's a "mind" -- or some equivalent -- that understands if we think the term "understands" is applicable to the case at hand.

      And that does say something about the way we operate in the world. It says that, embedded in certain linguistic practices we employ, are very deep assumptions we would find difficult to shake off.

    6. two questions:

      i know you say you want to keep an open mind about that, but i'm wondering why you say we might not need the less robust notion at all. is the idea that we never just want to predict, and that there is always some other things we want to do?

      (i don't remember the wording, but i think Cavell says somewhere that having a concept is a matter of being able to "run" with it. perhaps this is also what an internal understanding will be--the more internal understanding you have, the more you can run with it?)

      i'm also intrigued by what you say at the end when you bring the idea of a sense of self in connection with the notions of internal and external. But I'm not sure i see what you have in mind. Can you say a bit more?

    7. Searle, though, argues that the Chinese Room doesn't understand Chinese in any way

      Right, but not everyone accepts his argument. People disagree about what counts as understanding, with some preferring a more robust view than others. Searle's argument is a kind of intuition pump, but it doesn't bring out the same intuitions with the same force in everybody.

      is the idea that we never just want to predict, and that there is always some other things we want to do?

      No, I don't mean that. But I don't know whether we need to call the ability to predict understanding. It might be useful or natural to call it that. Or it might be confusing to call it that. I don't care very much either way, but if I have to commit myself to one view or another I'd like to consider some examples and/or hear arguments for and against each position.

      the more internal understanding you have, the more you can run with it?

      That sounds right.

      The self seems to come into the question of internal and external in the following kind of way. Winch talks about needing to have an understanding of art in order to understand the history of art as something other than just some odd things that people have done (or something along those lines). So it's not that one has to be an artist, but one does need, he thinks, to understand art, to have grasped certain concepts that artists use, in order to have an internal understanding of art history. This seems to me to involve something like being able to imagine oneself in the artists' shoes. If art moved from a very elaborate style to a very simple one, say, then the historian should be able to see why this was, which involves (I think) seeing the very elaborate stuff as excessive, or at least as something that could be regarded as excessive. As in the case where you say, "I can see why you might think it's excessive, but I like it." So it's personal in a way that external understanding is not. It involves reactions that either I have or else that I can imagine myself having.

    8. side note: great expression, "intuition pump." i like the mechanical ring it has. much philosophy, i think, seems to aspire to no more than that.

    9. It's Dennett's expression, so I can't take credit for it.

    10. "Searle's argument is a kind of intuition pump, but it doesn't bring out the same intuitions with the same force in everybody."

      Yes, I think that's why the argument is so contentious in some quarters. It just hits some people one way and others another. In the end it hinges on conflicting intuitions. Searle's view rests on the idea that "understanding" can only be understood as the kind of private experiencing each of us has when we hear or read a word or phrase and get its meaning. The problem is, as you suggest, that it may very well be the case that we all get quite different experiences (different associated mental pictures, say). If I read the word "horse" maybe I think of a racehorse while someone else conjures up a wild mustang or an old nag. Still we somehow manage to understand each other, much of the time, even if may have different pictures: a frontal view or a profile of a horse in motion even or maybe even some storybook picture of a horse we recall from childhood. Searle takes no account of that but relies on the notion that something mysterious is taking place at the instant of understanding and so, because he makes no effort to understand what that might be, he leaves us with the supposition that there is a mysterious mental occurrence that counts as understanding.

      Dennett counters with a proposal that it's all about complexity of associations which, I think, leads to a more satisfying view, i.e., that understanding the meaning of terms, say, isn't about a mysterious condition of getting it nor is it about sharing the same picture (Wittgenstein put the kibosh on that notion rather definitively I think) but rather that it's about a preponderance of associative similarities so that, to the extent there is a broad match of associations on a kind of macro level there can be agreement on meanings and hence we can understand one another's words and symbols. But it isn't perfect by any means which shouldn't surprise any of us, given that we are so prone to misunderstandings and confusions. The point is that we can work through many of them much of the time by a sort of trial and error in discourse. Dennett's account, I think, is much more reasonable than Searle's disguised mysterianism.

    11. I'm not sure that Searle would accept that he's as Cartesian as this, but he may be guilty as charged. One could object to Dennett's position, too, no doubt, but I'm not in a position to do so here.

    12. Searle would certainly deny being Cartesian as he does being dualist but I think the fundamental flaw in his argument is that he is the latter and, possibly the former, in a suppressed sort of way. But as you indicated, this really hinges on competing intuitions. If one cannot see how something like understanding (interpreted, as Searle uses it, as being what happens when we understand words or symbols) can be analyzed into more basic features, if one presumes, as Searle does, that consciousness is somehow basic (ontologically irreducible), then one cannot see how Dennett's approach could work. But if one doesn't have that Searlean type intuition, then Dennett's approach, whether empirically correct or not, becomes intelligible and at least a viable way of explaining what we mean by "understanding" of the type we ascribe to ourselves and those like us when we read or hear words, etc.

    13. Yes, although we don't necessarily have to choose between Dennett and Searle. As you suggest, they could both be wrong. (Not that I'm claiming to know what to think about consciousness.)

  3. It seems to me that Cogburn's take on this is perfectly harmless and, I gues he would say, perfectly self-evident. I take it by:

    "a set of discourse relevant norms held fixed" and

    "different normative assumptions about functional organization of organs, and different ones again about atoms ... once those are in place"

    he is talking about definitions of terms within a field of (scientific) enquiry and agreement on models and methods, or the grammar of certain language games, if you will allow that terminology. That once those are in place, then the ability to predict is a mere function of the grasp of the game and its application. He's saying that a good deal of what counts as physics or biology or chemistry is a grasp of the grammar involved. That seems fairly innocuous.

    So what understanding means in these games of language is clearly defined and can be shown e.g. in an examination of a pupil by asking for, say, the answer to "Given any particular configuration of P Q R ... then?"

    That doesn't seem too far off the other use of understanding discussed here which on the surface seem to clash with this one. It is e.g. fairly common to say of someone one knows: "NN would never do such a thing". Of course there are cases in which one is mistaken and NN in fact did do what one predicted they would not, but that is again (in full accordance with the model) explained as being the result of inssufficient data. This works on the assumption that given sufficient data understanding (correct prediction) follows. The entire apparatus of total surveillance being erected around us and this point in time is a concrete expression of exaclty this mode of thought or worldview. The correlate of prediction is control.

    So Cogburn, like any good political officer, is testifying to the party line.

    The question seems to be: Is there a different account of understanding? That, it seems to me, either means is there an account of understanding that fits this worldview yet leaves room for the unpredictable, or that any alternative account of understanding would entail an alternative world view, or view of what it is to be human. These look to me to be questions dealt with in e.g. the Tractatus.

    And that is not a scientific question so much as a question of what science is or might be.

    Re the students at Veracruz: They seem to be asking for a clear set of criteria for judging when they know/understand their subject; when they can confidently say they'll pass the final and get that vital piece of paper. Which seems, fair, as that is what they're being sold/paying for.

    It's all in the game.

    1. He's saying that a good deal of what counts as physics or biology or chemistry is a grasp of the grammar involved. That seems fairly innocuous.

      Thanks. This helps. It seems both innocuous and yet somehow wrong though. To talk about grasping the grammar of the language-game makes the process sound very intellectual. I don't mean that what is meant is wrong, but this way of putting it seems potentially misleading to me. And I don't blame you for that. I think the (perhaps merely apparent) intellectualization is there already in talk of holding fixed a set of relevant norms.

      I wonder whether I should stop talking about the students at Veracruz because I fear I'm misrepresenting them. The focus of our discussion was Wittgenstein and Winch, so what the students believed about other matters only came up in passing. I did not have time to question any one person at length about their research or about their own view of social science. My sense, though, was that their interests were unlike anything Winch had really imagined that anyone would be interested in. To a very large extent I think it comes down to what you are prepared to count as understanding. Some people are satisfied with a thinner conception than others.

    2. Re the first section: I find it misleading as well, and I do agree that it is wrong, but I take it that's what he means when he says he agrees with Wittgenstein and Monk. So I figured that was how he might have phrased it if he was okay with it not sounding fancy. As it is in your quote above it does sound like you need an expensive education to be able to do what he does. Apart from that he is representative of a very common view of what Wittgenstein amounts to. In that there are all these language games that hang together internally and some of them (science) are of more use than others.In that latter he merely dovetails with mainstream consensus. (Dawkins/Dennett et al.) At a guess I might say he is earnestly trying to make Wittgenstein "relevant".

      As for the students at Veracruz, I was just thinking that they, for practical reasons, were looking for some sort of clear idea as to what they were doing and what success in their endeavours might look like. It's one way of dealing with their chosen subject. A bit like saying, right, now that all the complications are in view, and I'm not doing philosophy, what do I have to hold on to?

      I did not mean to be disprespecting of them at all. If that is what came across, it is entirely to do with me and nothing to do with them.



    3. No, I didn't think you were being disrespectful of the students. I just started to feel that I was speaking too broadly to be accurate, given that I'm generalizing about several people who might not all think the same thing and whose ideas I just don't know very well.

      I think Jon Cogburn is trying to speak in a way that brings various different people and schools of thought into conversation with each other. That seems healthy. But I don't always speak the language, or want to speak it. I don't think he wants to make Wittgenstein relevant so much as he wants to see Wittgenstein as relevant. That is, he's not a Wittgensteinian out to sell his idol to others but a non-Wittgensteinian who hopes to find something good in Wittgenstein (as he hopes to find something good in all philosophers). Again, I think that's a good thing.

      He's one of my heroes, but that doesn't mean we're bound to agree on everything (of course).

    4. Re Cogburn, thanks. That ought to serve me right and learn me to hold my tongue. Carry on.

    5. I didn't mean that nothing he says should ever be criticized! Don't hold your tongue too long, please.

  4. How do we know if so and so understands our words or if they get the theory we've carefully explained or the directions we've provided for how to get to from here to there?

    We answer questions like that by observing the behaviors that follow and reporting on them. So and so understands how to read a map by demonstrating the capacity to interpret it through words or behaviors. They understand my directions for how to ride a bike by jumping on it and riding without falling off. He/she understands how to build a bike when he/she has done so, or can give us information about how to do so with which a knowledgeable bike builder will agree. They demonstrate knowledge by passing a test (formal or informal) or by putting a theory to work (though you can do the latter without doing the former, too). We don't have to know what's going on in anyone's head to say they have understood in such cases. But we suppose something is.

    A machine that's been built to ride a bike, or to build one (or just to follow certain verbal instructions), doesn't have to have a mental life though various programming functionalities, explainable as code or in terms of electronic interactions inside the mechanism, are assumed to be happening (since mechanical devices don't to these sorts of things of their own accord). Such descriptions can be invoked to explain how the machine comes to do what it has done. But with entities like ourselves there seems to be a further question: what's going on with them that counts as understanding what I've said, shown, and so forth?

    Confronted by creatures with comprehensible behaviors of a certain type, e.g., language use (though not that exclusively because we can speak of a chimp or a dog as understanding us, too), we suppose there's also a a knower at some level or other. And in this case to ask about "understanding" implies a different vantage point, as in what's going on with them that counts as "understanding"?

    Someone tells me this is how you build a bike. If they're speaking in Vietnamese or another tongue with which I'm unfamiliar, I hear sounds and perhaps see gestures but am at sea. If they switch to English, and I'm an English speaker, and they're using vocabulary with which I'm familiar (some English sentences may be too technical for me!) I get their words (though I may still lack the capacity to do what they're directing me to do -- mechanical aptitude being a different sort of thing than understanding languages). Now the sounds my instructor is making have meaning for me. But what counts as that meaning? What's going on in me that allows me to say "now I understand"?

    This strikes me as a different sort of question and, if it's not the usual one we have recourse to when considering who understands what, it's also not an unimportant one -- especially if one's project is to discover how to improve understanding in some creatures or to build understanding in machines.

    1. How do we know if so and so understands our words

      Of course, usually this question does not come up. And I think that's important. It's not that the answer is usually obvious. Usually there is no question of this at all. (If we forget this then we are likely to come up with a misguided way of thinking about the whole issue, an over-theorized or -intellectualized way of thinking.)

      or if they get the theory we've carefully explained or the directions we've provided for how to get to from here to there?

      Usually we find out by asking them: Do you see what I mean? Is that clear? etc. Only if we don't trust them will we make any further tests or observations (although we might accidentally observe, of course).

      We answer questions like that by observing the behaviors that follow and reporting on them.

      Sometimes, but only in the relatively rare cases where we a) needed to ask (or were unsure), and b) did not trust the person to answer honestly. E.g. when dealing with children whom we are trying to teach.

      But with entities like ourselves there seems to be a further question: what's going on with them that counts as understanding what I've said, shown, and so forth?

      Isn't this just the philosophical question: what is understanding? And in what sense "is there" this question? It does not simply come up, except in very unusual circumstances. We can bring it up, but it doesn't generally arise in the course of normal life.

      This strikes me as a different sort of question and ... it's also not an unimportant one -- especially if one's project is to discover how to improve understanding in some creatures or to build understanding in machines.

      I agree. But if you want to build a machine that understands you're probably best off trying to build one that behaves as if it understands rather than trying to do the philosophy first.

  5. But with entities like ourselves there seems to be a further question: what's going on with them that counts as understanding what I've said, shown, and so forth?

    "Isn't this just the philosophical question: what is understanding? And in what sense 'is there' this question? It does not simply come up, except in very unusual circumstances. We can bring it up, but it doesn't generally arise in the course of normal life."

    Unusual circumstances perhaps, but not, I would say, "very unusual."

    I'm not sure the above is a uniquely philosophical question rather than, say, a scientific one which raises some philosophical issues, e.g., what sorts of things are we and what does that say about our place in the world?

    Well, are those faux questions then?

    It's true we don't think about such stuff most of the time as we go through our daily routines but sometimes we do and it's not illegitimate -- or what's religion, spiritual pursuits and passion for the arts all about? These are all part of life, too, and philosophy has something to say about them as well as about the sciences. Perhaps philosophy doesn't have any unique questions? Perhaps what it has is just a way of approaching questions that belong in other bailiwicks.

    The AI researcher seeks to discover what is it that brains do that constitutes the experience of being the kinds of creatures we are while the religious seeker wants to know what is his or her place in the world. All of us, from time to time, wonder about what's the right thing to do in difficult circumstances. The lover of the arts wants to immerse him or herself in created experiences offered by others -- or to create his or her own for others, the more subtle and deeply moving, the more powerful the experience. All these are part of human life only because they involve having the kind of mental life we have. So is it really "very unusual" to wonder what understanding amounts to in the different cases -- or to try to relate the cases to one another in a way that, perhaps, enhances our understanding of "understanding" itself?

    1. what's going on with them that counts as understanding what I've said, shown, and so forth?

      What this question is asking depends on the circumstances. That is, is there some practical task (such as building an understanding machine) at hand, or is this purely idle speculation? (Or something else?) If it's pure speculation then the question is philosophical, I would say, and it's not clear how we could tell a correct answer from an incorrect one. Is someone like Dennett simply wrong, or is he entitled to his opinion? Is Winch simply right, or is he defending a plausible answer to a question that has no absolutely correct answer (in the sense that others are not bad but rather mistaken about a matter of fact).

      are those faux questions then?

      I wouldn't call philosophical questions faux, but they do belong to a particular type. It's not an illegitimate type either. But it is quite different from the scientific type of question. For one thing, typically we have methods for answering scientific questions whereas for philosophical ones we don't. So it's important to know what kind of question one is dealing with.

      is it really "very unusual" to wonder what understanding amounts to in the different cases?

      I think so, yes. Not in a pejorative sense, but in the sense that it does not happen most of the time. I wonder what time it is far more often than I wonder what time is. And I think I am completely typical in this regard. And because the question "what is time?" is not forced on us by any of our normal activities it is not clear what would count as a good answer. What am I trying to do in asking this question? Or am I just expressing a mental itch? And how then can I tell a true answer from something that only seems right (because the itch goes away in its presence)? I don't mean that all philosophical questions are nonsensical. I mean that questions, like other things we say, typically get their meaning from the way they are used. And the use to which a philosophical question is (to be) put is not clear.


    1. Thanks.

      Therefore, argues Turner, "practices"—in the sense that the term is widely used in the social sciences and humanities—is a myth, and so are the "cultures" that are central to anthropological and sociological thought.

      Sounds interesting.

    2. and well argued but sadly not widely read...

  7. I agree with your point that in everyday life we don't often have occasion to ask ourselves questions like "If he says he understands my point what does that mean for him?"

    I may ask if he really understands it or does he only think so -- or did I really make myself clear or did I understand his point, to which I was responding (or am I way off the mark)? And here all I mean by "understand" seems to consist of my getting certain kinds of responses or my producing responses of my own which elicit kinds of responses I expect. But unless I want to build something that can do that, I don't much care about what's going on in my interlocutor's head beyond what it produces in the way of remarks in the appropriate set of exchanges.

    But if I wonder how he comes to think as he does, perhaps I do want to know more. An interesting case: My wife can't read maps or at least she couldn't until quite recently. Whenever we'd drive anywhere and we needed directions from a map she invariably became confused and frustrated (the advent of Google Maps and Mapquest have changed that dynamic though.) Recently however she had a Eureka moment. Driving back from Canada she insisted on reading the map without my pulling over to the roadside to take it from her. And she did it. Slowly, painstakingly, but she got the next town right and the one after that right. She was ecstatic or having achieved something that had always been dark to her.

    I've always wondered why she had such a hard time with maps though. There was the very practical question of my learning not to depend on her for directions, of course (it got so bad that we had a standing joke in my family that if she said go left, you went right, and so forth). Still, why map reading seemed to be second nature for me yet such a challenge for her intrigued me. It wasn't a practical question, I suppose but an interesting, if theoretical, one.

    I wouldn't say it was a philosophical question but it wasn't really a scientific one either. I just wanted to know what was going on in her mind, when she looked at a map, compared to what was going on in my mind. Eventually I concluded that I had a picture of the landscape, in which we were, in my mind and related the map I was looking at to that picture. The names of places and the symbols on the map had a meaning for me because, when I saw them, I connected to the places we were actually in or would soon be.

    I also saw the directions on the map in terms of where we were. If I looked at something to the north on the map I could then look at the landscape and determine the north so the markings on the map to our north became significant to me. She seemed to be missing all that. I used to kid her that her problem was that she didn't have a map in her head like I did. It wasn't that I had a real map, of course, a complete, detailed picture of the landscape like the map in her hands. It was just that I had a kind of partial template in mind, vague but roughly matching the map in her hands, when I looked at the map.

    Not much of practical consequence came from this though, perhaps, if I were teaching children, it would tell me something about how different intellects work (I'm pretty bad at doing calculations in my head, for all my map reading skills, and my wife's got me beat on that score). So maybe knowing something about how one comes to understand the markings on a map might have some consequences in the real world. It's not entirely irrelevant to the things we do in life when you think about it. And it's not merely a philosophical speculation either.

    1. Yes, I agree that these questions are not entirely irrelevant. I just think that a sentence or question might appear to have one clear or real meaning when in fact the meaning is different in different contexts. If I want to know why this person cannot read maps then I must, it seems, investigate this person. What I find out may or may not be completely irrelevant to why other people cannot read maps, or to why I can read maps. "What goes on in people's heads when they read maps?" might not have a general answer. Perhaps we all do it differently. An investigation of one person or one group will be different from an investigation of human beings as such. And a philosophical investigation will be different from a scientific one. And the results of any of these investigations may or may not contribute to a general understanding of understanding. Although actually I think what I'm suspicious of is the idea of such a general inquiry.

  8. Agreed that it's not an inquiry into the general case because, for that, you'd need to do a bit of science. Maybe set up some scenarios and test to see how well some do in those situations and get reports of what they think is going on in their own cases and then look for some generalities from which to hazard an hypothesis and then test it further on broader and broader groups of subjects and increasingly variant scenarios.

    Eventually a scientific picture would emerge and this might be relevant to understanding brains and building their synthetic equivalents. It might also have pedagogical implications. That said, a line of thought re: why one individual may be able to read maps and another can't can hardly be scientific, even if it is based on a kind of ad hoc empiricism. But that doesn't mean it has no broader value than just the immediate case. The hypothesis that my wife lacks certain kinds of visualization capacities or just has them to a less significant degree than, say, I do, does say something about what goes into how we think ,and also what certain kinds of understanding consist of.

    Importantly, I think, it says something about how we use "understanding" in different contexts while reminding us that there is a mental domain which, though hard to talk about in any definitive way, can't really be dismissed out of hand the way behaviorists of a certain sort might be inclined to do. That is, I think it's important to take account of the mental aspect of our lives in order to usefully explain things like understanding (it can't just be about behaviors) and more complicated activities we engage in such as making value judgments including and especially the ethical kind. But that's grist for another mill.

    1. The hypothesis that my wife lacks certain kinds of visualization capacities or just has them to a less significant degree than, say, I do, does say something about what goes into how we think ,and also what certain kinds of understanding consist of.

      Qua hypothesis it doesn't say anything, does it? But if it were proved correct then, yes, that would tell us something.

      What I'd be wary of is thinking that we have the word 'understanding' therefore there must be something called understanding that we can investigate. It's possible that the one word has a variety of uses (meanings). And I wouldn't want to posit a mysterious mental domain unless I really had to. But, of course, it depends what is meant by terms like 'mental domain', on what the empirical facts are, and on what various theories can and cannot explain.

    2. I don't think "mysterious" except in the sense that there are still things we don't know (which is just to say science isn't complete) or, in a different sense, that the mental part of our lives isn't amenable to referential discourse in the same way elements in the public domain are. Still it's hard to argue we don't have subjectivity even if there's no thing to be singled out for public inspection. We do have thoughts, memories , feelings, beliefs, perceptions, desires, awareness. I don't see how a full account of human activity could ever be given if we disregard this part of our lives. The problem, of course, is how NOT to disregard it without embracing confusion of the sort Wittgenstein pointed out.

    3. The problem, of course, is how NOT to disregard it without embracing confusion of the sort Wittgenstein pointed out.

      Yes, exactly.

  9. Sorry for jumping in a bit late here. A couple of thoughts. Basically, I agree with w5800: what counts as "understanding" depends on the context in which the question of understanding arises. The context also dictates what counts as "real understanding" as opposed to "superficial understanding" or "no understanding at all". There is no such thing as a "real understanding" which, as it were, stands outside of any particular circumstance.

    This point also relates to the issue of predictability. The ascription of understanding and the notion of predictability are internally linked - it's a grammatical connection. So just as the context dictates what counts as "understanding" it also dictates what counts as "predicting behaviour" - for that latter term can cover a huge range of activities.

    If I know X always gets annoyed when someone mentions religion than that's one aspect of my understanding of her. It may even allow me to predict certain gestures or exclamations she'll make if the topic crops up.

    But suppose I know that x has a deep understanding of (or empathy for) other people - far greater than mine. This allows me to predict that she will find some way of comforting someone who's deeply upset - but I cannot predict what she'll actually do. She finds words and gestures that I'd never have thought of. So, you might say, for me her responses are predictably unpredictable.

    These are both cases of "prediction" but what falls under that term differs hugely in the two cases. So to say that understanding is bound up with prediction is true, but actually says next to nothing about specific cases.

    1. Yes, thanks, I think this is right. Although I'm not sure that the context dictates what counts as real understanding. Questions like this seem like moral questions to me: what counts as real and what counts as merely superficial? Surely this is, in part, a value judgment. That doesn't mean there's no right answer, but it does mean that disagreement is intelligible.

    2. Well, maybe "dictates" is a poor choice of words. But "real understanding" is sensitive to context at least insofar as it only seems applicable in some situations. Can someone have a real understanding of, say, the 6 times table? How would that differ from simply knowing it? If you're much quicker than I am at answering "what's 7x6?" does that mean your understanding is real and mine is superficial? The concept doesn't seem to have much traction in that type of situation.

      Of course, if we're talking about (eg) understanding human nature or music or even football then the concept seems far more obviously applicable. Yet it is also more open to question. You may say that the way Jones plays a piano piece shows a real (or deep) understanding - but I might say he's too flashy or passionless, etc.

      Does this mean there's no "right" answer? I'd say: it depends on what you mean by "right answer". If you're comparing it with "what's 7x6?" then yes, there's no right answer (but really that's just pointing out that here the question of right and wrong doesn't function like that). Does it mean you shouldn't claim to be right about Jones's musical understanding? Certainly not! As you say, it's a value judgement,and so it's unlike both "7x6=42" and "chocolate tastes nicer than vanilla".

      W says in §240 that people don't come to blows over the answer to mathematical questions, and that this is grounded in agreement of judgements. And that (it seems to me) points to the fact that where we do come to blows over questions it's often in part because we lack sufficiently similar judgements.

    3. Can someone have a real understanding of, say, the 6 times table?

      I think you either know it or you don't. But perhaps Pythagoras or Plato or Frege would claim that most people lack a real understanding of it. A claim like that might be confused, but I think that's debatable (obviously it will depend on what, if anything, is meant). And it might be religious or quasi-religious. Basically I agree with you though. I'm just quibbling.

    4. Actually, that type of example did occur to me, but I left it out as I think it's more a caveat than an objection to my point. But it's still interesting. There could be (admittedly unusual or esoteric) circumstances in which it made sense to talk about a deep understanding of multiplication. But (it seems to me) that someone could exhibit such an understanding and yet be rather poor at multiplying!

    5. Yes. Actually multiplying is trivial, after all, in comparison with the glorious wonders of....