That's an excellent question! Apparently it's not as I've been interpreting it. I've been thinking of the circle as a limit, but some googling reveals:
Circumplex: a circular depiction of the similarities among multiple variables. (APA Dictionary of Psychology)
The circumplex model focuses on determining how traits and emotions are structurally similar, and its underlying assumption is that a relatively seamless circular odering, or circumplex, is an economical description of the relations among traits and emotions. (from an introduction to an out of print book on circumplexes)
So I think that anything being mapped would occur on the circle and not within it. This is actually even more interesting because the intersection of the axes could be considered a neutral state, but if all states in this model must occur on the circle itself then there is no neutral state.
Perhaps someone else is more familiar with this idea?