Fig 1.
Illustration of the experiments flow.
Task A) Binary task used in Experiments 1–2. Illustrative example for a single trial (L-FOV condition in Experiment 1 at the 55% sp, see Section 2.1). During a trial, participants indicated whether “more than half” or “less than half” of the duration had passed, at a temporal sampling point that was 55% of the total duration (16.5s). Task B) The double production task used in Experiment 3. First participants were presented with an empty scene for 30s (ground truth) while performing an auxiliary task (detecting the presence of A or L on screen with a corresponding keypress). Then they were asked to press a key after 30s had elapsed (empty production task). At this point, the empty scene was replaced with a scene in the H-VC or L-VC condition (high or low visual complexity, respectively). After the change of scene, participants had to press a key when 30 more seconds had elapsed (visual complexity production task). In the H-VC condition, the four screens of the virtual room displayed different videos at the same time. In the L-VC condition, the four screens presented the same video synchronously.
Fig 2.
A) Stimuli used in the prestudies. Left: The 2D, grayscale stimuli we adapted from Xuan et al [6]. Right: The 3D lamps used in our follow-up attempts to directly extend the procedure of Xuan et al. to interval timing. B) Illustrative examples of the stimuli used in our experiments depicting natural scenes. From left to right, in each column: high contrast (H-CON), low contrast (L-CON), 360° panoramas (H-FOV), and their associated most salient crops (L-FOV). Fig 1 (Task B) also shows an example of high and low visual complexity (H-VC and L-VC).
Fig 3.
From left to right: Aggregation of trials by luminance contrast, temporal frequency and field of view conditions. Y axis: % of answers. X axis: 45% and 55% sampling points (sp). The displayed bars of each graph correspond to pover,45 at the 45% sp (overestimation) and to punder,55 at the 55% sp (underestimation). The percentage of correct answers for each case is complementary to the displayed value in the graph (pcorr,45+pover,45 = 100% and pcorr,55+punder,55 = 100%). Note that overestimation is always more frequent in low magnitude levels and underestimation in high magnitude levels, regardless of the visual feature.
Fig 4.
Results for the follow-up replication study with conventional displays.
Left: Luminance contrast trials. Right: Temporal frequency trials. Y axis: % of answers. X axis: 45% and 55% sampling points (sp). The displayed bars of each graph correspond to pover,45 at the 45% sp (overestimation) and to punder,55 at the 55% sp (underestimation). The percentage of correct answers for each case is complementary to the displayed value in the graph (pcorr,45+pover,45 = 100% and pcorr,55+punder,55 = 100%). Like in Experiment 1, overestimation is always more frequent in low magnitude levels and underestimation in high magnitude levels, regardless of the visual feature.
Fig 5.
Y axis: % of answers. X axis: Temporal duration of the trials (30s, 1min, 3min, 5min). The displayed bars correspond to punder,55 at the 55% sp (underestimation). The percentage of correct answers for each case is complementary to the displayed value in the graph (pcorr,55+punder,55 = 100%). Like in Experiment 1, underestimation is more frequent in high magnitude levels, in this case in trials with a duration of up to three minutes.
Fig 6.
A: Experiment 3. Participants take longer to realize 30s have passed in H-VC, which suggests that time is compressed in the presence of larger perceived visual changes. B: NASA-TLX score (workload measure) for the visual complexity production tasks of each level. Right violin plot (“diff”) shows individual differences in NASA-TLX score (L-VC MINUS H-VC). Note that the negative values in the diff violin plot mean that for some users L-VC was experienced as more demanding than H-VC. Significant differences are marked with an asterisk.
Table 1.
List of acronyms used through this work.
Table 2.
Accuracy (% of correct answers) is higher for high magnitude levels of contrast, frequency and FoV at 45% sp (“less than half” answers), while at 55% sp (“more than half” answers) it is always higher for low magnitude levels of contrast and frequency. At 45% sp, overestimation (incorrect responses, “more than half” answers) occurs more frequently for low magnitude levels. At 55% sp, underestimation (incorrect responses, “less than half” answers) occurs more frequently at high magnitude levels.
Table 3.
A) Results of the follow-up replication study with conventional displays.
Accuracy (% of correct answers) is higher for high magnitude levels of contrast and frequency at 45% sp (“less than half” answers), while at 55% (“more than half” answers) it is always higher for low magnitude levels of contrast and frequency. At 45% sp, overestimation (incorrect responses, “more than half” answers) occurs more frequently for low magnitude levels. At 55% sp, underestimation (incorrect responses, “less than half” answers) occurs more frequently at high magnitude levels.