Why does a flickering light seem to last longer?
A flickering disk seemed to last longer than a steady one of the same duration, and people also became worse at judging its duration, consistent with the visual system using longer time windows to analyse flicker.
Source
Perception of time in articulated visual events
Study at a glance
- Design
- Human experiment — Within-subject two-interval duration discrimination: a flickering (luminance-modulated) standard disk compared with a steady comparison disk; adaptive staircase in Experiment 1, constant stimuli with confidence ratings and ROC analysis in Experiment 2
- N
- Experiment 1: 9 observers; Experiment 2: 6 observers; many trials per observer
- Population
- Young adult observers (mean age 22) at a Japanese research institute
- Outcome
- Time dilation (shift in point of subjective equality), response bias, and sensitivity to duration (d_a)
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
Observers saw two grey disks one after the other and judged which lasted longer. One disk (the standard, lasting 500 to 1100 ms) flickered in brightness at 3 or 7 cycles per second, while the other was steady and varied in duration. Experiment 1 (9 observers) used an adaptive procedure to find the point at which the two seemed equal; Experiment 2 (6 observers) used fixed durations and confidence ratings so that sensitivity could be separated from response bias. The authors then built a simple simulation in which flicker lengthens the time window used to detect the stimulus's offset.
What they found
Flickering disks were judged longer than they were (time dilation), more so for 7 Hz than 3 Hz flicker, for longer standards, and when the flickering disk came second. In Experiment 2 the dilation averaged about 146 ms for the 500 ms standard and about 249 ms for the 1100 ms standard. Flicker also biased responses toward 'longer' and reduced sensitivity to duration, and the second stimulus was judged less precisely than the first; these two losses of sensitivity added together independently. The simulation reproduced these patterns.
The limits
What it doesn't show
Samples were very small (nine and six observers) with heavy testing of each person, so the results describe a robust within-person effect but say little about individual differences. The integration-window model was only simulated and fitted qualitatively; the study did not measure neural integration windows or rule out attentional explanations directly. Only simple disks with two flicker rates were tested, so it is unclear how far the account extends to complex real-world events.
Key terms
- Time dilation
- Overestimation of how long a stimulus lasted relative to its real duration.
- Point of subjective equality (PSE)
- The comparison value at which an observer is equally likely to judge it longer or shorter than the standard; shifts in the PSE reveal perceptual biases.
- Time-order effect
- The tendency for the order of two stimuli to bias their comparison, here making the second stimulus seem longer.
- ROC analysis
- A signal-detection method using confidence ratings to separate an observer's true sensitivity from their bias toward one response.
- Uncertainty tradeoff
- The measurement principle that a signal's timing and its frequency content cannot both be estimated precisely at once; longer windows help frequency but blur timing.
Flashcards
0 of 10 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
Which condition produced the greatest time dilation in Experiment 1?
Common questions
Is time dilation just people guessing 'longer' for flickering things?
Partly, but not only. Experiment 2 separated bias from sensitivity and found both a bias toward 'longer' and a genuine loss of precision in judging the flickering stimulus's duration.
Why would the visual system trade off timing precision?
Estimating how fast something flickers works best over longer stretches of input, and using longer integration windows delays and blurs detection of when a stimulus ends, which inflates and muddies duration estimates.
Are six or nine observers enough?
For psychophysics, where each observer performs hundreds of trials and effects are measured within each person, small samples are common, but they limit claims about the general population.
More on Perception