Is it possible to believe something without believing what logically follows from it? Not if belief is a relation between a believer and a classical possible worlds proposition. Perhaps belief therefore should be understood as a relation between a believer and a sentence? Arguably, no. Because that wouldn’t account for the following fact: whenever it’s true to utter the English sentence 'Eric believes that it’s raining', it’s also true to utter the Swedish sentence 'Eric tror att det regnar'. Belief has to be a relation between a believer and something else. But what, exactly? Some suggest structured propositions. These provide, as they say, more fineness of grain. But how do we understand the relation between a believer and such a proposition? In virtue of what does it obtain? Also, we have what I call the nand-problem, suggesting that structured propositions may in the end provide too much fineness of grain.​