What it means to “want” something: A model-based theory

• by Sunny Khemlani
• in News
• posted June 10, 2022

Hillary Harner spent much of her time in my group at NRL exploring how people comprehend the word “want”. In a new paper now out in Cognitive Science, we outline how people interpret and reason about want: they build models of future possibilities, and use desire-language to contrast desires against facts. Here’s the abstract of the paper:

No present theory explains the inferences people draw about the real world when reasoning about “bouletic” relations, that is, predicates that express desires, such as want in “Lee wants to be in love”. Linguistic accounts of want define it in terms of a relation to a desirer’s beliefs, and how its complement is deemed desirable. In contrast, we describe a new model-based theory that posits that by default, desire predicates such as want contrast desires against facts. In particular, A wants P implies by default that P is not the case, because you cannot want what is already true. On further deliberation, reasoners may infer that A believes, but does not know for certain, that P is not the case. The theory makes several empirical predictions about how people interpret, assess the consistency of, and draw conclusions from desire predicates like want. Seven experiments tested and validated the theory’s central predictions. We assess the theory in light of recent proposals of desire predicates.

And here’s a link to the PDF as well as the corresponding OSF page.