I think of surprise as the effect of data on the update of your prior probabilities. If you previously considered an hypothesis unlikely, but the data changes t

I think of surprise as the effect of data on the update of your prior probabilities. If you previously considered an hypothesis unlikely, but the data changes that probability to be much higher it is a surprise (same with voting a prior high probability down).