The subjective attitude verb ‘consider’ has been understudied. But its semantics can illuminate the semantics of ‘find’, the nature of subjective belief, and the phenomenon of faultless disagreement.

‘Consider’ requires belief in the prejacent, but also requires it to be subjective. However, ‘consider’ tracks a broader form of subjectivity than the subjective attitude verb ‘find’:

(1) a. I believe 2 is even.
b. #I consider 2 (to be) even.
c. #I find 2 (to be) even.

(2) a. I believe mushrooms are tasty.
b. I consider mushrooms (to be) tasty.
c. I find mushrooms (to be) tasty.

(3) a. I believe racehorses are athletes.
b. I consider racehorses (to be) athletes.
c. #I find racehorses (to be) athletes.

I argue that the notion of subjectivity that ‘consider’ tracks is idealized disagreement. Intuitively, a sentence or small clause S displays idealized disagreement arises when two agents with “the right kind” of evidence can rationally disagree over whether S is true. I call this right kind of evidence ‘canonical evidence’. For any particular sentence, I diagnose its canonical evidence using familiarity inferences, inferences that the relevant attitude holder has certain evidence when it is embedded under ‘consider’. For example, ‘Mushrooms are tasty’ has the familiarity inference in (4a). This result is evidenced by the infelicity (4b):

(4) a. I have tasted mushrooms before.
b. #I consider mushrooms tasty, but I have never tasted mushrooms before.

On my view, ‘consider’ therefore tracks an intuitive notion of subjectivity, on which the (canonical) evidence underdetermines a unique rational position on whether a sentence is true.

My formal account of the semantics of ‘consider’ is one on which ‘consider’ has the at-issue/truth-conditional meaning meaning of ‘believe’ but presupposes two things. First, it presupposes that its prejacent displays idealized disagreement. Second, it presupposes that the relevant agent, the type e argument of ‘consider’, has canonical evidence on the prejacent.