Fig 1.
A conceptual summary of the current study.
Previous studies, which correspond to the left panel, have confirmed the positive correlation between overall RTs in each vigilance task session (e.g., the numbers of long RTs in each PVT session) and long-term pupillary fluctuations amplitudes measured over several minutes immediately before each PVT session (e.g., PUIs below 0.8 Hz), which were acquired every several hours. The long-term increase in the overall RT is interpreted as a decrement in long-term vigilance. The underlying mechanism of this relationship could be ANS-related tonic activity of the LC from 0.1 to 5.0 Hz. On the other hand, the current study, which corresponds to the right panel, examined the positive correlation between the RTs per trial within one session of PVT and the short-term pupillary fluctuations amplitudes per trial within one session of PVT (i.e., M-PUIs measured with a fine temporal resolution of 10 Hz or more immediately before the target presentations during trials). In this study, the increase in RT per trial was defined as a decrease in short-term vigilance level. The underlying mechanism of this relationship might include the phasic activity of the LC from 10 to 20 Hz.
Fig 2.
The serial flow of elements in one trial of the Psychomotor Vigilance Task.
After a random interval (1,000 to 8,000 ms), a target (white dot) appeared. Participants were asked to press a key for the appearance of the target as quickly as possible. After the keypress response, the time from the appearance of the target to the keypress was displayed. The section for measuring the pupilar diameter is illustrated (from the start of the trial to the target appearance).
Fig 3.
A Conceptual example of M-PUI calculation.
The purple arrows in this figure represent the flow of M-PUI calculation. The black line in the first block represents the raw pupilar diameter’s time series. First, a blink was detected using the rule that a blink occurs when the pupilar diameter’s constant value falls below the threshold. Second, the pupilar diameter’s time series was linearly interpolated during the blink. Third, noisy time series with raw pupilar diameters were smoothed. Different smoothed data are obtained using differently sized time windows (orange, purple, and blue lines). Note that small fluctuations can be seen for smoothing with small timescales, but not for other cases. Fourth, we calculated M-PUI in the smoothed pupilar diameter’s time series by using a medium timescale in this example. Then, we calculated the absolute value for the degree of change (the sum of the magnitude of red arrows) in pupillary fluctuations. Finally, the absolute degree of change was divided by the length of a specific interval (the sum of the magnitude of green arrows).
Fig 4.
Positions of data analyses in the practice-theory trade-off.
Fig 5.
Distributions of RTs on all valid trials.
The upper panel shows a histogram of the RT distributions of all participants. The lower panel shows RT distributions of individual participants, starting with the individual with the shortest mean RT. The different colors represent histograms of different individuals, which are common to the upper and lower figures. Note that two trials (4,988 ms and 1,662 ms, respectively) were not shown for visual purposes.
Table 1.
Summary of statistics testing differences between mean M-PUIs calculated from time-series for all smoothed pupilar diameters for long and short RTs.
Fig 6.
Scatterplot and distributions of each participants’ and correlation coefficients across all participants between normalized RT and normalized M-PUI for the 50 ms Hanning window.
The top left panel represents the scatterplot between normalized RT and M-PUI for the 50 ms Hanning window for each participant and all participants. The top right panel represents the distributions of correlation coefficients between normalized RT and normalized M-PUI for the 50 ms Hanning window for each participant and all participants. The bottom panel represents all scatterplots for each participant. The different colors represent scatterplots of different individuals that are common to Fig 5. The solid red line in these panels shows the mean magnitude of the correlation coefficient for all participants, and the dotted black lines show the magnitudes of the correlation coefficient for each participant.
Fig 7.
Mean correlation coefficients between normalized M-PUI and normalized RT for each Hanning window size and numbers of participants whose correlation coefficient reached significance.
Red dots indicate the points where detailed scatterplots are shown in Fig 6. Error bars represent standard errors.
Fig 8.
Mean correlation coefficients between normalized M-PUI and normalized RT for all down-sampled time-series of pupilar diameters and numbers of participants whose correlation coefficients reached significance.
The blue line represents the mean correlation coefficients between normalized RT and normalized M-PUI calculated from the pupilar diameters down-sampled to 100 Hz. The blue bars represent the numbers of participants that reached significance in the case of 100 Hz. The green line represents the mean correlation coefficients between normalized RT and normalized M-PUI calculated from the pupilar diameters down-sampled to 50 Hz, and the green bars represent the numbers of participants that reached significance in the case of 50 Hz. The red line represents the mean correlation coefficients between normalized RT and normalized M-PUI calculated from the pupilar diameters down-sampled to 25 Hz, and the red bars represent the numbers of participants that reached significance in the case of 25 Hz.
Fig 9.
Mean correlation coefficients between normalized M-PUI and normalized RTs for different time points and numbers of participants whose correlation coefficients reached significance.