Soma uses two versions of the Psychomotor Vigilance Test: the standard PVT (5 or 10 minutes) and the Brief PVT, known as the PVT-B (3 minutes). Both tests measure sustained attention and vigilance by tracking how quickly and consistently you respond to a visual stimulus.
A critical part of how we score these tests is how we handle very slow responses. This article explains exactly what we do, why we do it, and why it produces a more accurate and meaningful picture of your cognitive state than using no cutoff at all.
The Two Cutoffs
PVT (5 minute and 10 minute versions) Any response at 500ms or slower is classified as a lapse. The RT value of that response is discarded from the mean reaction time calculation, but the lapse itself is counted and tracked separately.
The PVT was developed by Dinges and Powell in 1985 at the University of Pennsylvania. Their work showed that responses at or above 500ms are not slow reactions. They are moments where attention has dropped out entirely. Below 500ms you are measuring processing speed. Above it you are measuring attentional failure. The two are fundamentally different and need to be treated differently.
PVT-B (3 minute version) Any response at 355ms or slower is classified as a lapse. The RT value of that response is discarded from the mean reaction time calculation, but the lapse itself is counted and tracked separately.
The 355ms cutoff was established by Basner, Mollicone, and Dinges in 2011. Because the PVT-B is shorter and faster paced, athletes naturally produce fewer and shorter lapses than they would on the standard PVT. Applying the 500ms cutoff to a 3 minute test would dramatically undercount attentional failures. The lower threshold corrects for this, bringing lapse frequency in line with the standard PVT and making both tests directly comparable.
Why We Remove Slow Responses From Mean RT
When you average reaction times together, very slow responses have a disproportionate effect on the result. A single 2000ms response can shift a mean by tens of milliseconds, even if the other 59 responses in that window were fast and consistent. The problem is that a 2000ms response is not a slow reaction. It is not a reaction at all in the vigilance sense. It means attention dropped out entirely for that moment. Including it in the mean reaction time is like measuring how fast a runner is by averaging their race splits with the time they stopped to tie their shoe. The resulting number accurately represents neither their speed nor how often they stopped. Discarding the RT value of any response at or slower than the cutoff threshold gives you a clean, uncontaminated measure of how fast you actually are when you are engaged. It answers the question: when you are paying attention, how sharp is your processing speed?
Why We Still Track Every Lapse
Excluding the RT values of responses at or slower than 355ms (PVT-B) and 500ms (PVT) from the mean does not mean those responses are ignored. Every one of them is captured, counted, and reported separately as a lapse. The RT value is discarded, but the attentional failure is always recorded. This matters because lapses and mean reaction time measure two fundamentally different things. Mean RT (below the cutoff) tells you about your processing speed when attention is present. Lapse count tells you about the stability and consistency of that attention over time. A person can have fast mean RT and many lapses. A person can have slow mean RT and very few lapses. These represent completely different cognitive profiles with different implications for performance. Tracking lapse count separately also makes the metric much more sensitive to fatigue, sleep deprivation, and cognitive load. Research consistently shows that lapse count increases more dramatically under fatigue than mean RT does. If you only looked at mean RT, you would miss the most important signal. If you only looked at lapses and ignored mean RT, you would miss the baseline speed picture. Together, they tell the full story.
The Result: Two Clean, Distinct Metrics
Soma reports both of the following for every PVT and PVT-B session:
Mean Reaction Time (ms) tells you about processing speed when attention is present. Lapse Count tells you about the stability and consistency of that attention over time. This two-metric approach is grounded in the original PVT methodology developed by Dinges and Powell (1985) and validated extensively across sleep deprivation, fatigue, and occupational performance research. It is the same approach used in aerospace research, military performance science, and clinical sleep medicine.
Why This Is More Accurate Than No Cutoff
Using no cutoff at all produces a single blended mean that is compromised by both extreme slowness and genuine speed in the same number. It cannot tell you whether a higher score reflects a person who was consistently slow or one who was mostly fast with occasional complete attentional failures. These are very different states requiring very different responses. By separating clean reaction time from attentional lapses, Soma gives you data that is actually actionable. You can see whether your issue on a given day is processing speed, attentional consistency, or both.
What it means when all data returns zero
If your athlete's session returns a mean reaction time of zero across all data points, this does not mean something went wrong with the test. It means every single response in that session was slower than the cutoff threshold. For the PVT that means every response was 500ms or slower. For the PVT-B that means every response was 355ms or slower. Every one of those responses has still been captured and counted as a lapse. The lapse count will reflect the full session. But because no response fell below the threshold, there is no clean reaction time data to report, so the mean returns zero. When you see this, do not look at the mean. Look at the lapse count. That is where the entire story of that session lives. A full-zero mean is itself a significant signal. It means the athlete was not having occasional attentional failures. Attention was compromised for the entire duration of the task. This is most commonly seen under significant fatigue, sleep deprivation, high cognitive load, or very early in a training block before the athlete has adapted to the task demands. It can also simply mean the athlete was not engaging with the task properly on that occasion.
References
Basner M, Mollicone D, Dinges DF. Validity and sensitivity of a brief psychomotor vigilance test (PVT-B) to total and partial sleep deprivation. Acta Astronautica. 2011;69(11):949-959.
Dinges DF, Powell JW. Microcomputer analyses of performance on a portable, simple visual RT task during sustained operations. Behavior Research Methods, Instruments, and Computers. 1985;17(6):652-655.
Basner M, Dinges DF. Maximizing sensitivity of the psychomotor vigilance test (PVT) to sleep loss. Sleep. 2011;34(5):581-591.