ALL IN YOUR HEAD:

Appeals to evolution to explain the function of our subjective experience of pain should be able to describe how feelings, like anxiety or fear, can aid in evolutionary outcomes over and above systems that detect damage through purely informational means. 

In artificially intelligent (AI) systems like autonomous self-driving cars, a reinforcement learning reward function is used to make the system value danger avoidance. A reward for steering clear of something that may cause harm is enough for certain AI systems to adequately protect themselves from damage, leaving no room for affect (feelings that could be negatively or positively valenced) in motivating their behavior. It has been recently argued that functionally effective AI systems may not need to incorporate suffering into their learning, and that we ethically shouldn't. 

So what does it mean to feel pain? If a danger avoidance system can function without pain, why are we burdened with suffering when it may not be necessary for self-preservation in a potentially dangerous world? The current medical conception about the function or pain (the alarm mechanism) doesn't adequately answer this question.

Tweet   @brothersjudd