What does it mean that computation is content-involving?

In this article we analyze the thesis that computations are content-involving (CIC). We show that the thesis can be understood in a stronger and a weaker sense, depending on whether semantics affects both the individuation of computation and the transitions between computational states (the computational dynamics), or only the former (so that the formality condition is preserved). We argue against both versions of CIC. Then we put forward a moderate syntactic view, according to which the concept of computation as abstractly characterized is essentially syntactic, but the computational explanation of physical systems makes an unavoidable use of content-involving vocabulary. This does not mean that implemented computations get their content essentially.