I really feel like everyone who has taken the time to reply (and thank you guys for doing so) has genuinely missed the point which I was making. Why would we assume that simulated roaches would experience anything? Simulating a roach is not the same thing asbeing a roach. I don't see any evidence that simulating something in a computer would ever give the simulation the actual properties of its real life counterpart.
Well just for fun they've actually put the simulated roach brain into a little robot that had artificial antennae and other senses that emulated the roach's body, and sure enough the robot behaves like a real roach did. What's to stop someone from simulating that same sensory input to the roach brain without the robot body? How would the roach know the difference?
The argument isn't that the simulation actually takes on the properties of the simulated, but that to anything in the simulation, the simulation itself is indistinguishable from reality.