Subjective expressions pose a problem for truth-conditional models of meaning: what does truth mean when expressions such as “Avocados are delicious” can be false for one speaker and true for another? A spate of recent linguistic and philosophical literature starting with (Lasersohn 2005) has offered various solutions to the problem, but most of them did not define what subjective expressions are, identifying them ad-hoc, “We know one when we see one.” This would predict that expressions are either subjective or not, but that is not actually the case – there are contexts where seemingly-subjective expressions are objective, and vice versa.

This talk, based on my dissertation research, uses findings from psychophysics and aesthetic research to offer a predictive model for when expressions are subjective. I argue that for an expression to be subjective, the perceptions and/or the categories that form its evidential basis must vary among speakers. For example, smells and tastes vary because people with different genes have differing smell receptors, while an expert would classify the features she sees in an artwork according to different criteria than an amateur would, but similarly to another expert.

I show how this model explains contextual shifts in subjectivity and provides clarity to other models, from expressivism to Lasersohn’s judge-index relativism to Kennedy and Willer’s counterstance contingency (2017). Subjective expressions reflect differences in how language maps to reality, but also reflect differences in how speakers can perceive reality itself, and thus play an important role in establishing norms of consensus and collaboration.