Hindsight bias is the tendency to retrospectively think of outcomes as being more foreseeable than they actually were. It is a robust judgment bias and is difficult to correct (or "debias"). In the experiments reported here, we used a visual paradigm in which performers decided whether blurred photos contained humans. Evaluators, who saw the photos unblurred and thus knew whether a human was present, estimated the proportion of participants who guessed whether a human was present. The evaluators exhibited visual hindsight bias in a way that matched earlier data from judgments of historical events surprisingly closely. Using eye tracking, we showed that a higher correlation between the gaze patterns of performers and evaluators (shared attention) is associated with lower hindsight bias. This association was validated by a causal method for debiasing: Showing the gaze patterns of the performers to the evaluators as they viewed the stimuli reduced the extent of hindsight bias.