What is knowledge?

John L.McCreery (jlm@TANUKI.TWICS.CO.JP)
Sun, 19 Dec 1993 17:19:40 JST

Two last thoughts on the broadest epistemological issue--
what we think knowledge (or *truth* or *science*) is.

In a book called _Embodiments of Mind_ Warren McCulloch (one
of the founders of automata theory and a seminal influence on
cognitive science) remarks that what he tries to do is build
machines that simulate human behavior. They are always
unsatisfactory. They are unable to do something that humans
do. At this point, McCulloch says, researchers fall into two
groups: There are those who say, "You see, it can't be done."
There is something--call it 'soul' perhaps--that machines will
never have. Then, says McCulloch, there are people like me. We
go out and build a better machine. We can always build a better

In (and here I'm not at all sure of the reference) _The Birth of
the Tragedy and the Generation of Morals_ Frierich Nietzsche
compares a scientist and a metaphysician to two men watching
Salome perform the dance of the seven veils. The scientist is
content to be tantalized as one veil after another comes off. He
will be quite happy if she never gets to the end. The
metaphysician is the bozo in the back who's always shouting
"Take it all off, NOW!"

If this is where I'm coming from, what do I make of
anthropology's "postmodern" predicament? To me the "modern"
is defined by the ascendancy of science as a model of
knowledge based on experimentation. The experiment is a way
of getting directly at what the researcher believes is the heart
of his problem, by isolating the key variables and excluding
everything else. The result, as Alfred North Whitehead noted
many years ago in _Science and the Modern World_ is a
bifurcation of human experience into stuff that is easy to
manipulate experimentally and a lot of other stuff that isn't.
Historically, the former has become the province of "tough-
minded" types who prefer "hard" problems to the soft, squishy,
can't-get-my-hands-on-this latter, which is left to the
"tende-minded," dare we say "soft-in-the-head" types who
congregate in what are called the humanities. Voila! The two
cultures, as C.P.Snow called them.

Anthropology and the other social "sciences" are born under the
banner of science. But birth and citizenship are two different
things. Because their subjects are human beings, none of this
social science tribe is allowed to perform experiments. They
can't, then, be "real scientists." Some, like economics and
certain forms of sociology, try to be as much like real
scientists as they can. In mathematical models and statistical
research methods they have something that can be manipulated
in a quasi-experimental fashion. I say "quasi-experimental"
because what gets manipulated are numbers abstracted from
realities, not, as in real experiments, physical realities
themselves. (Note, too, how statistics rests at base on the
idea that significant stuff--represented by "hard" data can be
isolated from other stuff that is written off as due to random
chance and thus, inherently, insignificant.)

What, then, of anthropology? In some parts (physical
anthropology, archeology), experiments don't bear directly on
the main questions (the evolutionary history of homo, state-
formation, that sort of thing), but they do provide a wealth of
useful circumstantial evidence (carbon-dating, DNA matching,
for example). In principle, the same could be said of
sociocultural anthropology. Here, however, there are practical

Fieldwork is the heart of our discipline, and fieldwork
historically has meant going to some exotic location, settling
in and spending a year (sometimes two, three or more) living
with people who are, at first, perfect strangers to us. Learning
the language and enough of whatever the local equivalent of
street-smarts is to survive and get on with our work consumes
most of our time. We fill our notebooks with impressions that
are more or less well-founded; add census data, photographs
(nowadays maybe videos). We get home and find ourselves in a
situation that is closest perhaps to that of the historian who
confronts a mass of heterogeneous, more or less fragmentary
records and must make something coherent of them.

To me the force of the "postmodern" claim that anthropology
is, first of all, a kind of writing is rooted in this predicament.
Whether culture itself is a text is moot. What, in fact, we
produce is mostly text of one form or another.

The nut of the problem I want to raise here is we don't have a
consensus on how to evaluate the results. Clearly we're not
doing experiments. Only rarely do we have the kind of data that
make statistical tests valid. How then can we judge what is
better and worse? The hermeneutic tradition to which Geertz
appeals is one possibility. Are there others?

John McCreery (JLM@twics.co.jp)