In IRS, the content of a concept is taken in terms of its inferential relations rather than associative relations where the meaning of the concept is identified by the role the concept and its inferences play in language and thought. Alternatively, in other words, IRS claims, in the simplest form that the content of a concept is in the role that the concept plays in the mind
The roots of IRS lie somewhere between Wittgenstein suggestion about meaning is use and Stellar’s observation that the content of logical constant can be specified by the roles in inference.
IRS is able to cope up with two things, according to F&P. At first, when you consider Concepts, they do not have any extension or real existence in the world to have an inference like a square circle or Santa Claus. Although more or less, the people who have such concepts are likely to provide a definition for it, trying to prove that their concept has content. As to the relationship between Empty Concepts and IRS, in the work of F&P is not clear. What I gather by this point is, plausibly, Isolated concepts cannot have content. Technically, a Concept C can have content only when it has inferential relations to another concept D. Thus, if C does not have any inferential relation with any other concept, it cannot have content.
In the work of Stellar, F&P talk about how inferential rules are constituted by ‘interaction rule’ and ‘elimination rule’ and by this tried to explain how the concept AND, and the concept TREE are semantically different.