Are Computational Transitions Sensitive to Semantics?

The formal conception of computation (FCC) holds that computational processes are not sensitive to semantic properties. FCC is popular, but it faces well-known difficulties. Accordingly, authors such as Block and Peacocke pursue a ‘semantically-laden’ alternative, according to which computation can be sensitive to semantics. I argue that computation is insensitive to semantics within a wide range of computational systems, including any system with ‘derived’ rather than ‘original’ intentionality. FCC yields the correct verdict for these systems. I conclude that there is only one promising strategy for semantically-laden theorists: identify special computational systems that help generate their own semantic properties, and then show that computation within those systems is semantically-laden.