It is widely accepted within contemporary philosophy of perception that the content of visual states cannot be characterized simply as a list of represented features. This is because such characterization leads to the so-called, “Many Properties problem”, i.e. it does not allow us to explain how the visual system is able to distinguish between scenes containing different arrangements of the same features. The usual solution to the Many Properties problem is to characterize some elements of content as subjects, to which (...) features are attributed by a predication-like relation. In this paper, I reconsider this solution and claim that the Many Properties problem can be solved without invoking the notion of “subject”. What is more, I argue that an alternative approach has stronger justification given the empirical data concerning human vision. (shrink)