value.This iscertainlynotperceivable,butitinfluencesthewayonereasonsabouttheobject, including one's goalsand
desires concerning the object. Value can be established by all sorts of means, mostly very indirect. Thus value seems
like a prime candidate for a type of inferential feature. Another candidate is high-level taxonomic category. For
instance, the classification of a particular object as an artifact predicts only that people typically use this object for
something. But it does not predict what it looks like, how it works, or what one does with it. Rather, beyond the
ontological feature‘physicalobject’, which does have perceptual consequences,artifactis not on the whole a collection
of perceptual features; rather it bridges to all sorts of inferences that can in turn potentiallyengage perceptual features.
Now suppose there were a concept that hadonlyinferential features. This then would be an abstract concept such as
those expressed bythe value of my watchandthe meaning of this word. Fro mthe point of view of language, abstract objects
cannot be distinguished fro mconcrete objects: all that is necessary for reference is an indexical feature. The interfaces
to language cannot“see”whether a concept's descriptive features are perceptual or inferential. Thus as far as language
is concerned, reference proceeds in thesame fashion whether or notthereis a conceptualized physicalobjectout there
in the world.
If abstract concepts have indexical features and descriptivefeatures, in principle theyshould have ontological category
features and valuationsas well.An interesting issue for future research is to explore what ontological category features
might pertain to abstract entities. As for valuations, comparebeliefandopinionwithvalue, meaning, andobligation.The
former twoare understoodas“subjective”: theyare objects inpeople'sminds, henceinternalrelativized toa subject.By
contrast, thelatter threearenormallyunderstoodas“objective partsoftheworld”—eventhoughonreflectionnothing
has value without someone to value it, nothing has meaning without someone to understand it, and there are no
obligations in the absence of individuals who understand the social system. That is, talking about someone's beliefs is
liketalking abouttheir mentalimagesor dreams (Jackendoff1983: ch.11); whiletalking aboutsomeone'sobligationsis
more like talking about their car (Jackendoff 1999). In short, what causes something to be judged“real”is its having
the valuationexternal—even when it is a patently abstract object!^167
Itis a fascinatingquestionwhyvalues, meanings, and obligations should be understood as“objective,”and I a mnotin
a position to answer it here.
REFERENCE AND TRUTH 323
(^167) Thus an individual's havinga“theory ofmind”(inthesense ofPremack and Woodruff1978 and Wimmer and Perner 1983) couldfunctionallydepend ontheirbeingable
to recognizethevaluationinternal asattributed to otherindividuals. The literature suggests thatchildren do notdevelopthis abilityuntilabout 5 years of age, and thatmost
animals lack it almost entirely.