ABSTRACT
We can begin to approach what might be called the peculiarity of literature as a
form of cognitive practice by comparing how literary works end with how other
pieces of intellectual work end. A proof in mathematics ends by reaching its final
line, where each line that is not an axiom is generated in explicit accord with a
rule of inference that in principle anyone might follow. Reports of experimental
results generated in a lab specify procedures that were followed in setting up
equipment and carrying out tests. While they often also offer conjectural inter-
pretations of results and suggestions for further work, they describe minimally a
procedure that anyone might follow in order to achieve a like-enough result.
Hence we can speak readily of objective evidence that a certain state of affairs
can be produced so-and-so. In statistical social science, one finds reports of
results from questionnaires or other data about populations expressed in
numerical terms. Under the assumption that a larger population will not be too
different from a sample, one can draw conclusions about distributions of traits
and tendencies of development. History undertakes to tell us what happened,
and the claims of professional historians are supported with reference to primary
sources, indicated in footnotes. In economics, one often finds abstract mathe-
matical models that describe processes of income distribution or GNP growth,
for example, that are imagined to occur underneath a confusing surface of extra
variables that induce deviations from the model. Among these cognitive prac-
tices, literature is perhaps most like economics in giving a model of certain
processes in the world. This is scant comfort, however, since whether the pro-
cesses described by economic models really do occur, on the one hand, or are
rather fairy tales invented by clever calculators, on the other, is itself a subject of
more than a little dispute. Literary models, moreover, if that is what literary
texts offer us, are in even worse shape, since they focus only on very small
numbers of mostly made-up cases, and they lack even the potential of refine-
ment through the incorporation of further data.