However, a mentalist approach to“truth-conditions”differs fro mthe philosophical/logical tradition in twoi mportant
respects.
First, as stressed in Chapter 10,“truth-conditions” are not taken to constitute absolute truths in the world or in
possible worlds. Rather, they are taken to be the conditions by which a language userjudgesthat some conceptualized
individual is a member of a category; and the form of these conditions is constrained by human psychology, not by
logical necessity. Such psychological conditions can be studied experimentally, as a rich research tradition has shown.
Second—and this is the burden of the present section—the main philosophical tradition adopts rather uncritically
Tarski's assumption (1956) that the truth-conditions for a sentence constitute a set of conditions that are individually
necessary and collectively sufficient to guarantee truth of the sentence:“Snow is whiteis true if and only if the following
conditions hold:....”Even the avowed mentalist Fodor, in arguing against definitions, assumes that necessary and
sufficientconditions are theappropriatewaytoaccomplishsemanticdecompositionof lexicalitems, e.g.“Somethingis
abachelorif and only if it is human, male, adult, and unmarried.”
By contrast, much conceptualist semantic research (e.g. Jackendoff 1983; Lakoff 1987) stresses the insufficiency of
Tarskian conditions to characterize the richness of human categories. This section presents examples of some of the
problems that have arisen.
11.6.1 Categories with graded boundaries
Consider the categoryred. This cannotbe identified with a particular perceivedhue, since a broad range of hues are all
calledred. However, as hues shade imperceptibly from red toward orange, there comes a point where observers are no
longer clear about their judgments. Observers may hesitateor disagree with one another. In addition, the judgment of
a particular hue may depend on what hues have immediatelypreceded it in presentation(if presented after focal red, it
is judged orange, butif afterfocal orange, itis judged red). Thatis, thereis a focalrange in whichjudgmentsare secure
and consistent,butitshades intoa borderlinerangeinwhichthereisconflictwitha neighboringcategory, and inwhich
judgments become less secure and more context-dependent.
This“fuzziness”in theboundary of a category is nota matter of speakers“not knowing themeaning ofred”;rather,it
isinherentinthestructureoftheconceptitself.Onecanmakeredmore“Tarskian”bystipulatingrigidboundarieswith
orange, pink, purple, and brown, but then oneis notdealing with theordinary meaning of theword. Similarly, onecan
create a newcategoryred-orangeat theboundary, but then thesame sort of fuzziness occurs at theboundary of red and