Does Time Really Slow Down during a Frightening Event?
(a) Participants' estimates of the duration of the free-fall were expanded by 36%. The actual duration of the fall was 2.49 sec. (b) If a duration expansion of 36% caused a corresponding increase in temporal resolution, a 79% accuracy in digit identification during the fall would be predicted (left bar, see text). However, participants' accuracy in-flight was significantly less than expected based on this theory (middle bar, p<2×10−6). In-flight performance was no better than ground-based controls (right bar, p = 0.86), in which the experimental sequence was identical except that the participants did not perform the free fall. The performance scores are averaged over participants, each of whom performed the experiment only once and had a potential performance of 100% (correctly reported both digits), 50%, or 0%. Note that participants did show better-than-chance performance on both the in-flight experiment and ground-based control (chance = 10% accuracy) even though the alternation period had been set to 6 ms below their threshold. This performance gain might be attributable to perceptual learning; it may also be because movement of the chronometer makes it slightly easier to read due to separation of successive frames, and participants sometimes moved the device involuntarily as they hit the net. To ensure parity between the comparisons, we applied a small jerk to control participants' wrists to mimic how the device moved when free-fall participants hit the net. Asterisks represent p<0.05.