When watching videos, our sense of reality is continuously challenged. How much can a fundamental dimension of experience such as visual flow be modified before breaking the perception of real time? Here we found a remarkable indifference to speed manipulations applied to a popular video content, a soccer match. In a condition that mimicked real-life TV watching, none of 100 naïve observers spontaneously noticed speed alterations up/down to 12%, even when asked to report motion anomalies, and showed very low sensitivity to video speed changes (Just Noticeable Difference, JND = 18%). When tested with a constant-stimuli speed discrimination task, JND was still high, though much reduced (9%). The presence of the original voice-over with compensation for pitch did not affect perceptual performance. Thus, our results document a rather broad tolerance to speed manipulations in video viewing, even under attentive scrutiny. This finding may have important implications. For example, it can validate video compression strategies based on sub-threshold temporal squeezing. This way, a soccer match can last only 80 min and still be perceived as natural. More generally, knowing the boundaries of natural speed perception may help to optimize the flow of artificial visual stimuli which increasingly surround us.We are almost literally immersed in an artificial visual world, made in large part by motion pictures. How robust is our sense of reality with such pervasive technologies? For example, how well can we detect speed alterations in a video clip? The availability of videos, videogames and virtual reality systems, through which a variety of arbitrarily altered worlds can be experienced, calls for an appraisal of the boundaries within which video speed can be manipulated without conflicting with perception. However, although much effort has been devoted to studying motion perception 1,2 , how we perceive the speed of real-life scenes is still largely unknown, due especially to stimulus complexity and the presence of cognitive, higher-order factors 3-5 . Here we have addressed natural speed perception by measuring the detectability of speed alterations applied to a soccer match video clip.
ResultsRetrospective noticing. In the first experiment we used a rather ecological setting, with observers watching a 10-minute video clip of a soccer match without any specific task. They viewed it only once and were told that at the end they would have to answer questions concerning the match, aimed to assess attention and memory capabilities. Observers were divided into five groups (N = 25 each), and each group watched the video clip at a different speed (0.88x, 0.96x, 1.00x, 1.04x and 1.12x, where 1.00x is the original video speed -30 fps; see Methods). The original audio track with voice commentary was pitch-compensated to avoid distortions when the video was played at a different speed. In general, observers were quite compliant, as suggested by self-reported attention measures (mental concentration: 60%, event-following: 81%). By contras...