Fodor's primary argument for concept nativism can be summed up as follows. No theory of concept acquisition can explain why it is that we acquire the concept DOORKNOB from exposure to doorknobs rather than anything else, unless it is a theory whereby acquisition occurs via some rational process. However, reasoning requires antecedently existing representations on which to operate. So any theory of concept acquisition presupposes antecedently existing representations, i.e. concept nativism.
I agree that concept acquisition must occur via a rational process, but I disagree that rational processes in general require antecedently existing representations. If a rational process can be mechanized, it might be able to operate on non-representational inputs. I give an example of a rational process of concept acquisition, "acquisition by abduction", that can indeed be mechanized so that it operates on non-representational inputs. Thus it can operate even if the starting point for concept acquisition is a blank slate. Since there is reason to believe that the mechanistic version of acquisition by abduction is the basic functional unit of the cerebral cortex, it is entirely possible that we have no innate concepts.