A received picture in metaethics distinguishes between two kinds of evaluative terms.
The first - thin terms, such as 'good' - express merely evaluative content. The second -
thick terms, such as 'kind' - have both evaluative content and descriptive entailments.
This apparent saturation of content has made thick predicates a central part of
various metaethical debates, including over non-cognitivism and over the nature
of ethical knowledge. Here I argue that the received picture is based on a mistake.
Thick predicates are as undiluted by description as thin predicates. Both express
evaluative states of mind. Evaluative states of mind, I will argue, come in different
flavors, flavors I will call evaluative dimensions. Once evaluative dimensions are on
the table, we can see more clearly the difference between thick and thin predicates:
thick predicates supply their own evaluative dimensions, whereas thin predicates
are sensitive to dimensions that come from elsewhere.
This view can explain why uses of thick terms communicate descriptive content
even if they don’t entail it: they communicate that the user has the kind of reason
that informs evaluation in the dimension of the thick term. It can also explain why
we avoid using some thick terms: we are not willing or able to evaluate in every
dimension. If this view is right, then thick predicates are of philosophical interest
for reasons other than the ones typically attributed to them: they shed light on an
undertheorized aspect of our cognitive economy, dimension-relative evaluation.