Why Software Brains can not feel pain

That simulated rain doesn't get anyone wet is an old saw. But it's a straw man, no one would claim otherwise. Simulated rain on a computer is just bits of electricity coursing through wires, where rain is droplets of H20 free falling and slamming into our physical bodies. But now imagine a virtual human sharing a simulation with that simulated rain. He would get wet, he would feel the rain as much as he can feel anything. So does he feel anything?

Consider Bostrom's simulation hypothesis[1]. If our universe is a simulation, would that suddenly mean that we aren't conscious, that we don't feel our rain when it hits us? Our own perceived consciousness does not prove our universe is real. It is not sufficient to say "I feel therefore I am real". A detailed simulated brain would be exactly as conscious as ours is, which is not proof either is actually conscious, but proof they are equivalent.

The confusion is the simulated brain would be part of it's world not our world. We can tie the two together with sensors and actuators, giving the virtual brain a physical presence in our world. Then it could feel our rain. Or the reverse, we could feel a simulated rain if we had a device that stimulated our skin in just the right way.

Imagine the extreme opposite of an embodied brains, a little black cube that contains a detailed virtual human with no input or output to the real world. Does it feel? Do the same experiment in reverse, imagine our world is a simulation but no one is watching it, no one is interacting, no one cares. If a virtual human stubs his toe and no one is watching, does it hurt?

[1] http://en.wikipedia.org/wiki/Simulation_hypothesis

"When is a simulation the same as the thing being simulated?" I wrote a detailed analysis of this question in 2005 at http://thinkinghard.com/blog/WhenIsASimulation.html.

My article is a refutation of John Searle's argument that a mechanical simulation of a person who "understands" Chinese is not itself a thing that "understands" Chinese, i.e. a simulation of "understanding" Chinese is distinct from actually understanding Chinese.

Panu is repeating John Searle's argument, but replacing "understand Chinese" with "feel pain".

And the answer I found to my question is: a simulation is the same as the thing being simulated _if_ the thing being simulated is an information processing system. For example, a simulation of a washing machine is not a washing machine. But, a simulation of a calculator _is_ a calculator. A realistic simulation of a calculator would look just like one of those calculator programs you have on your computer or your mobile phone. And those programs actually _are_ calculators - because we use them to do calculations for us just like we used to use real calculators to do calculations for us.

The question of whether simulated pain is the same as actual pain is equivalent to the question of whether pain is an aspect of an information processing system (which I think it is), or whether it is something else.

An information processing system that "feels pain", will be less and less inclined to do something that is "painful". This will be part of the decision making process of that system. If the system is somehow able to "ignore" its own painful feelings, it will end up making bad decisions - just like a person who can't feel pain ends up making bad decisions. Typically people who don't feel pain suffer accumulated injuries and generally fail to look after themselves, even when they intellectually know that their insensitivity to pain puts them at risk, and as a result they become disabled and die at a younger age than those of us who suffer pain in the normal manner.

i don't really think the article sold the point. Software brain doesn't necessarily equate to a simulated brain.