In a 1913 experiment, subjects detected whether someone outside their vision was watching them with only 50.2% accuracy—essentially random chance. Our belief we can sense stares is confirmation bias: we remember the hits and forget the misses.