Because, in a way, propositional logic only represents facts about finitely many objects, not knowledge. In other words, it doesn't allow you to deduce facts about objects you _haven't_ seen before. If you want to represent knowledge, you'll want to be able to represent information such as: Objects always either have property $A$ or $B$, i.e. a sentence such as $$ \forall x\: (A(x) \lor B(x)) $$
In propositional logic, the best you can do is add all statements of the form $$ A_x \lor B_x $$ for all _known_ objects $x$. But that doesn't allow you to deduce anything about a previously unknown $x$.