I go into situation A with expected reward 100. I receive reward 100. Yes - I might have gotten reward 100 by going into the situation with expected reward -1

I go into situation A with expected reward 100. I receive reward 100. Yes - I might have gotten reward 100 by going into the situation with expected reward -12 .. but that would be a dumb thing to do. That is the nature of acting intelligently under uncertainty - you assume things that might not be true (but probably are).