Artificial realities fool us: Our poor sense of video speed.

We live almost literally immersed in an artificial visual reality, especially videos. Here we show that our tuning to video clip speed is quite poor, and changes during childhood. Firstly, we found a remarkable lack of sensitivity to speed manipulations when viewing a soccer match. None of 100 naïve observers spontaneously noticed speed alterations up/down to 12% (Just Noticeable Difference, JND=18%). When measured with a constant-stimuli discrimination task, speed sensitivity was still low (JND=9%). Secondly, using short clips of human motion, mixed human-physical motion, physical motion and ego-motion, speed underestimation was typical, although largely dictated by clip content (Point of Subjective Equality, PSE, from 2% to 32%). Speed biases were robust: manipulating display size or adding arbitrary soundtracks was irrelevant. PSE was uncorrelated with estimated duration of the video clips, thus pointing to a distinct sense of event speed. Thirdly, in 142 children aged 6 to 10, the “perceptually natural” video clip speed (PSE) decreased steadily until reaching adult levels, and sensitivity (JND) improved, as if maturational processes speeded up visual perception. Children were also assessed for response control, and scored for impulsivity/inattention and visuo-motor habits. PSE and JND correlated with response times, which also decreased with age, but not with response accuracy, which remained high. No correlations were found with psychological and habits scores, except videogame playing. Besides the relevance in clinical and developmental contexts, measuring the subjective sense of speed may help to optimize video reproduction speed and validate “natural” video compression techniques based on sub-threshold temporal squeezing.