analyticity and holism

In my last post I wrote a bit about the Analytic Imperative, which I think implicitly structures scientific reasoning. In that post, I took analyticity to be a property of arguments or lines of reasoning. That’s the shiny new formulation, the one I think will pass muster in Academia. But the original formulation, the one that got this project going almost 10 years ago, was much more psychological. I’m not sure how much of this more cognitive dimension will make it into the final project. The last few years I’ve tried to hedge my bets by finding ways of talking about the same issue without needing to bring in the cognitive psychology of scientists. But in the interest of full disclosure, and because I think it may be helpful for people to get an intuitive picture of what I’m up to, I’ll try to summarize the analytic/holistic distinction as it exists in cognitive psychology.

I’ll be drawing almost entirely on an excellent book by Richard E. Nisbett called The Geography of Thought. Nisbett is a psychologist who works on things like social cognition, the culture of honour and violence in the American south and people’s use of probabilistic reasoning. But it was a 2001 meta-study called Culture and Systems of Thought: Analytic and Holistic Cognition (pdf here) that set my mind on fire. The book is an extension of that work, and I highly recommend either if you’re interested at all by what follows. He argues that it has hitherto been assumed by psychologists that human cognition is everywhere the same. By studying white, middle-class undergraduate psychology majors, researchers believed that they could make generalizations about all human beings. This assumption, Nisbett claims, is demonstrably false. Using fairly standard psychological methodology, he tries to show systematic differences in the way ‘Asians’ and ‘Westerners’ think and perceive the world.

Just as an aside, you can probably already see why I’d want to excise this from a project in analytic philosophy. Issues about race and culture are inevitably difficult, in that sticky way that only hot political questions can be. I grew up in a town so ethnically homogenous that for most of my young life, I believed that acknowledging or mentioning a persons race counted as being racist. I consider myself ill-equipped to deal with the political complexities. Nevertheless, let’s dive in anyway. I can really only give the flavour of the work that has been done.

Which two of the following things are most closely related?

  • Monkey
  • Panda
  • Banana
My natural inclination is to say that Monkey and Panda belong together, because they’re both animals and mammals. But in a 2002 study, Li-jun, Zhiyong Zhang and Nisbett found that students from mainland China and Taiwan grouped Monkey and Banana together, because they have a thematic relationship: monkeys eat bananas. American students, on the other hand, had the same inclination as I do. A similar study by Liang-hwang Chiu [1] found that using pictures of a cow, a chicken and some grass, American children would group the cow and chicken, while Chinese children would group the cow and the grass.
So what goes on here? The argument Nisbett makes is that Americans are inclined towards categorical reasoning – to take an object in isolation from its context, as an instance of some abstract category and reason on that basis. Asians, in contrast, tend to reason dialectically, seeing relationships and connections as the primary stuff of reality.

The interpersonal version of this difference is familiar enough. American children are, according to Nisbett, “protagonists of their autobiographical novels” (p.87). He describes the work of developmental psychologists Jessica Han, Michelle Leichtman and Qi Wang [2] which suggests that this is the case:

[They] asked four- and six-year-old American and Chinese children to report on daily events, such as the things they did at bedtime the night before or how they spent their last birthday. They found three remarkable things. First, although the children made more references to themselves than to others, the proportion of self-reference was more than three times higher for American children than for Chinese children. Second, the Chinese children provided many small details about events and described them in a matter-of-fact fashion [ed: they’re like little historians!]. American children talked in a more leisurely way about many fewer events that were of personal interest to them. Third, American children made twice as many references to their own internal states, such as preferences and emotions, as did the Chinese children.


Americans see themselves as relatively autonomous, self-defining entities. Context and circumstance are regarded as relatively unimportant. Asians, on the other hand, regard themselves as embedded in a set of relationships. Consider the moral philosophies of Aristotle and Confucius to see this writ large. Aristotle taught virtue ethics: goodness is a property that inheres in individuals, and to be a good individual is to satisfy a set of criteria (to fall under a set of categories). For Confucius, on the other hand, goodness was the satisfaction of a complex set of relationships of duty and obligation. The Analects describe in detail the proper relationship between fathers and sons, elder and younger brothers, mothers and children, husband and wife.


But I don’t want to give the impression that these differences are merely features of the respective culture’s self-narratives. There are demonstrable perceptual differences as well. Nisbett and his colleagues found differences in the performance of Asians
and Westerners on the ‘rod and frame’ test. The idea is you put the subject’s face in a long box, at the end of which is a rod. The angle of the rod can be manipulated independently of the box. The subject is then asked to judge when the rod is perfectly vertical. What they found was that Asians were more ‘field dependent’ – that is, they were more biased by the orientation of the frame than were Americans. The vision of Asians appeared to be literally more context dependent than that of Americans.


There are compensations, of course, for being holistic. Asians performed better on tasks where they were asked to detect correlations between events (images flashed on a screen).


So in sum, Westerners seem to rely more heavily on categories and treat objects as isolated from their environment, whereas Asians tend to see things as embedded in a context. Each strategy has its own domain. The analytic strategy is really nice if you can make it work – the messiness of particularity falls away, and you can make definitive claims about things. On the other hand, holism is a safer bet if you believe the world is a goddamn mess. If things are always changing, relying on categories is a bad mistake. If context always matters, it is never safe to use abstract categorical reasoning.

As I wrote last time, I think science is basically, characteristically analytic. That’s what makes it go. Well-meaning people like Richard Lewontin who argue for ‘dialectical’ science are just making a mistake, if by ‘dialectical’ they mean anything like holistic. For all its flaws, the analytic way of seeing is probably the best thing the Western world ever produced. From it flows things like the scientific method, and the idea of individual rights. But we need to understand its limits in order to use it properly.


[1] Chiu, L.-H. (1972) “A cross-cultural comparison of cognitive styles in Chinese and American children” International Journal of Psychology 7, 235-242
[2] Han, J.J., Leichtman, M.D., and Wang, Q. (1998) “Autobiographical memory in Korean, Chinese and American children.” Developmental Psychology 30, 326-350
Advertisements

7 thoughts on “analyticity and holism

  1. I’ve always balked a bit at the idea that there can be some fundamentally different style of thought common to one part of humanity but alien to another, but I have to admit this is probably due to an overestimation of my own power of imagination, reasoning and empathy. On the other hand having read a little Lao Tzu and Confucius, they are fascinating to me for their relative lack of explicit arguments (ie connected series of statements intended to establish a proposition) or rather unifying arguments and those works are as a result wonderfully strange to me (of course so is Hegel’s Phenomenology of Geist), compared to much of Western philosophy. Interestingly in my again very limited exposure to the Buddhist tradition in Asian there we find fairly straightforward arguments, which of course were imported originally from India (with some help from various monks and possibly a magical monkey, pig, water sprite and dragon). I’m not sure whether this sort of thing is consonant or dissonant with this sort of theory because on the one hand it suggests how thought in different cultures can be very different, but at the same time illustrates how alien thought can become deeply interwoven in a society, suggesting universal (or at least trans-cultural) appeal, capabilities of thoughts and norms. So are there different ways of thinking or the same ways but different emphasis?

    One confounding variable I would note in an English-Chinese comparison is learning the Chinese written system, probably requires a significantly different mental investment with its several thousand characters over the 26 of the Roman alphabet that could easily effect visual reasoning say. I’m not sure that is actually a problem for the theory or an example, but it is a more mechanical difference than the thesis seems to want imply.

    One other confounding thing, if you asked me to guess what words people said by doing free association one the word “monkey” I would guess “banana” before I would ever guess “panda” (I would guess “barrel” before I would guess “panda”). Its somewhat speculative but I suspect small change in understanding the question could lead to a huge change in the answers given even among English speakers.

    A quick note while duty to others, you can easily play up that aspect in Aristotle (there is an appropriate level of generosity required of the virtuous man for example and think about how much he talks about the importance of friendship). Also for all his emphasis on carefully prescribed duties Confucius said that it is the person who broadens the way, not the way that broadens the person (Ana. 15-29). It is outside my expertise but Confucius often seems to talk as if there are moral virtues that inhere in individuals that ritual and proper deference to social relations cultivate rather than define.

    • I would reject any claim that either of these cognitive styles are alien to anyone – they occur in degrees all over, and there is huge variability within cultures. For my purposes, the important point is simply that these two styles of thinking and perceiving exist. Since races make demonstrably poor scientific categories, we should also not be mislead into thinking these studies show us something essential about ‘westernness’ or ‘asianness’. But I do find it plausible that, using the crude heuristic of separating people into their country of origin, significant cognitive differences could be found.

      The differences between English and Chinese are actually one of Nisbett’s favourite examples. He argues that some of the cognitive differences come from (and presumably are also the cause of) the differences in the structure of those languages. So far from being a confounding factor, he takes it as a part of his argument.

      • Funnily as far as the language (ie the spoken language) is concerned Chinese and English are similar in their analytic character (as opposed to synthetic, ie lack of conjugation, declension etc. and use of word order and context to distinguish parts of speech) even if Chinese is more extreme. The way he describes holism, English is way more holistic than Latin or Greek (ie depends on the relations between words for meaning not characters of the words themselves) because of its lack of complex conjugation and declension even if it is less holistic than Chinese. Reading the paper you linked I note that they claim the actual usage of of English stylistically avoids contextual dependency, this is an interesting claim.

        The thing that struck me about the written script is that while English and Chinese are very different here, it hardly falls neatly into a holistic-analytic divide, it seems almost completely orthogonal. It is basically a question of memorizing a large number of symbols with phonetic content versus a small number. This could alter the visual reasoning but not necessarily along an analytic-holistic axis. If you are merely concerned with establishing different dispositions to styles of thoughts in different societies this does not matter. If you want to claim that such differences can be analyzed along a analytic-holistic axis then it most definitely does matter and it may be premature to make such a claim.

        I notice the paper claims “It can be maintained that the Western alphabet is more atomistic and analytic by nature” than Asian writing systems (presumably meaning Chinese characters), maybe but it can also “be maintained” that the Chinese characters are more atomistic, since in English for example the phonetic value of a character (letter) depends on its relations to other characters and even other words (“read” versus “read” in “I read the book last week” versus “I read science fiction novels”) . Whereas each Chinese characters stand for a definite syllable (although to be fair tonality and probably other aspects of phonetics depends on placement) and can even sometimes suggest its semantic content (although this should not be overstated since many Chinese words are polysyllabic and so the individual characters have no meaning beyond phonetics, also since spoken Chinese is ambiguous written Chinese in spoken style has the same ambiguities). Finally, it claims Chinese characters are essentially pictographic which is a hugely debatable and I think dubious claim (admittedly based on a cursory study of the language and having read a book by one expert writing a polemic against precisely such claims but it was a really good polemic). To reiterate my previous point, the pictographic makes the phonetic reading of the whole phrases analytic even if the pictograms need to each be read holistically, kind of a unity of opposites going on here.

        • Ach… Okay having skimmed through more of the paper I found an interesting note on how on one field dependence test (finding a figure inside a more complex figure) East Asian showed no more field dependence than Westerners (and possibly less). However someone else did the test comparing Malaysians (who use alphabetic characters) with Westerners and found Malaysians scored as highly field dependent. (page 297 footnote 5) So, they show at least a dim awareness of my worry that the cognitive skills acquired as part of literacy in Chinese script are orthogonal to the holistic-analytic axis they set-up.

          I find their history of science cherry picked at best…

        • Allan, you seem to be arguing that there are differences between Chinese and English that cannot be captured by the analytic/holistic distinction, but surely no one would disagree with that. You seem to have a few different meanings of ‘analytic’ in mind, and it isn’t clear to me that all of them correspond with what the authors were up to. I’d be interested to see the reference you’re drawing on though, if you can dig it up.

          I could be convinced that the differences they’re trying to establish between the written languages are specious. The differences I’m really interested in, however, are of the following form:

          “”Generic” noun phrases are more common for English speakers than for Chinese speakers, perhaps because Western languages mark in a more explicit way whether a generic interpretation of an utterance is the correct on. In fact, in Chinese there is no way to tell the difference between the sentence “the squirrels eat nuts” and “this squirrel is eating the nut”. Only context can provide this information.” ( Geography of Thought p.156)

          The questions about the cognitive value of writing systems are sort of interesting, but kind of beside the point for my project.

          • Just to be clear their is an analytic-synthetic classification of languages and in that spectrum English and Chinese are on the analytic end (or isolating) and ancient Greek and Latin are on the synthetic end ( http://en.wikipedia.org/wiki/Synthetic_language ).

            The whole question intrigues me even as I can’t help but be skeptical. I think one basic worry is actually best illustrated by the Chinese character example. There may be a holistic frame of mind that prefers Chinese style characters, also that same frame of mind may encourage frame dependent visual processing (all things being equal) and yet learning Chinese style characters can then develop the skill to identify figures embedded in a larger figure more easily, increasing frame independent analysis of some visual tasks, thus countervailing the tendency in a specific context (taken at face value the meta-study suggests all this). I expect such countervailing causes to be the rule rather than the exception. Also surely there will be multiple different tendencies interacting? And it is surely a difficult problem to characterize whatever tendencies exist (especially in the face of complex interactions). So good for Nisbett, now is the time for bold conjectures but it is also the time for humbling refutations.

            • To be clear, that is NOT the distinction they are working with. Conflating that particular analytic/synthetic distinction with the analytic/holistic distinction the study is arguing for does not appear to be helpful, or fair. It appears, on the whole, to be missing the point.

              Your speculation is well placed, I’m sure. Life is complicated, and anyone who tells you the world can be divided in two and categorized on that basis has clearly oversimplified. I rather think that isn’t what they were doing, however. As you point out with the Malaysian example, they’re willing to identify counterexamples to a statistical regularity when they find them. But unless the counterexamples outweigh the regularity, this is not a refutation, it’s honest scholarship.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s