Plantinga’s reply that selection acts on behavior, not belief content, assumes beliefs and behavior can come apart. If my belief that the cliff is dangerous causes me to avoid it, selection for the behavior is selection for the belief.

This is the strongest version of the reply, associated with Fodor and with Fitelson and Sober, and it is the one to quote. Plantinga’s response is that beliefs cause behavior by their neural properties, and it is the neural properties that are selected, so semantic content comes along only if it happens to be correlated; naturalists reply that the correlation is exactly what evolution would produce. This is where the argument stands or falls, and the debate is unresolved. You should see both moves and decide for yourself.