Learning in pain: probabilistic inference and (mal)adaptive control - World Wide