Animals receive rewards and punishments in different patterns. Sometimes stimuli or behaviors can become predictors of future good or bad events. Through learning, experienced animals can then avoid new but similar bad situations, or actively seek those conditions that give rise to good results. Not all good or bad events, however, can be accurately predicted. Interestingly, unpredicted exposure to presumed rewards or punishments can inhibit or enhance later learning, thus linking the two types of experiences. In Drosophila, place memories can be readily formed; indeed, memory was enhanced by exposing flies to high temperatures that are unpaired from place or behavioral contingencies. Whether it is the exposure to high temperatures per se or the lack of prediction about the exposure that is crucial for memory enhancement is unknown. Through yoking experiments, we show that the uncertainty about exposure to high temperatures positively biases later place memory. However, the unpredicted exposures to high temperature do not alter thermosensitivity. Thus, the uncertainty bias does not alter thermosensory processes. An unidentified system is proposed to buffer the high-temperature reinforcement information to influence place learning when accurate predictions can be identified.