Optalert - Carousel Post

Yawning – Good Indicator of Drowsiness?

In 1872, Darwin penned a few words describing the act of yawning which read “…a deep inspiration, followed by a long and forcible expiration… during this act tears are often secreted, and I have seen them even rolling down the cheeks”. Unlike everything else the Theory of Evolution has accurately described, the common perception of yawning as the ultimate indicator of drowsiness has not evolved much. We have always associated the act of dropping our jaws, drawing in air and shedding a tear as a silent cry for caffeine or a nap. Lately, significant effort has been invested in algorithms and cameras that detect yawning as a basis of driver drowsiness detection systems. Groups such as this one have successfully identified yawning drivers using a computationally lightweight method with pinpoint accuracy, outperforming existing state-of-the-art algorithms that frequently faced real-world limitations.

Food for thought. Systems that detect yawning are built on this association between accurate detection of a yawn and drowsiness. However, is there validity and reliability behind this underlying assumption? Is yawning a good indicator of drowsiness?

In this Optalert Insights piece, I discuss the reasons behind why we yawn, whether doing simple day-to-day, real-world activities could be misclassified as yawning and the possibility of a yawn-drowsiness detection system being susceptible to manipulation.

Spoiler alert – No, yawning is not a good indicator of drowsiness.

Yawning and Drowsiness

For some context, consider the following three questions.

  1. Do you exclusively go to sleep the moment you’ve started yawning?
  2. Do you only yawn right before going to sleep?
  3. Do you always have a yawn-free morning after a good night’s sleep?

I would hazard a guess and say it is a resounding ‘No’ across the board and that isn’t surprising. Yawning does not essentially mean we are drowsy and hence, is not an accurate indicator of drowsiness. Although I am not implying that you would never yawn when you are drowsy, I would confidently say that on a lengthy list of reasons why you would yawn, drowsiness would sit close to the bottom.

But What Does the Science Say?

Scientists are not entirely sure why we yawn (but they checked and it is not drowsiness)! Although it is a daily occurrence which starts when we are a 20-week-old foetus, researchers are still on the lookout for a complete explanation and physiological purpose which yawning achieves, and the debate about its usefulness is still ongoing. Here is a list of the most cited explanations of why we yawn.

Yawning is an Arousal Reflex in the Brain

To put it simply, we yawn when we are bored. Boredom happens when the activity we are engaged in is no longer able to sustain our attention. Admittedly, while reading a boring post about yawning may lead to readers nodding off, it is different from directly associating yawning and drowsiness. Studies have indicated that when we yawn, it elevates our heart rate and arousal level which acts as a stimulant for the nervous system which in turn gives rise to increased alertness.

Yawning is Used to Cool the Brain

Yawning is a thermoregulation mechanism for our brains. When we yawn, we draw a big breath of cool air, push it to the brain while moving hot blood from it. Studies have shown that temperature is a significant predictor of how often a person would yawn with one study concluding that on average, a person yawns more in summer than winter. In fact, it is not uncommon to see people at the gym yawn as their bodies try to cool the brain down.

Yawning is Contagious

Ever felt like yawning right after witnessing someone else doing it? Contagious yawning is a well-known phenomenon and is found to correlate with emphatic skills in healthy humans. In this study, it was discovered that strength of the social bond between the “yawner” and “yawnee” had a strong effect on yawn contagion, increasing from strangers, acquaintances, close friends to kin. In another study, results showed a person that is less empathic would be less likely to follow suit after seeing someone else yawn. Researchers have postulated the possibility that yawning is actually an unspoken act of communication and empathy.

Not Yawning Does Not Mean Alertness

So, if we yawn it does not mean we are drowsy. But what if we do not yawn? Does that indicate alertness? Interestingly, a research study looking at the characteristics of daytime drowsiness due to sleep apnoea concluded that the absence of yawning bouts was a strong indicator of daytime drowsiness!

Altogether, science tells us that associating yawning and drowsiness is not plausible whichever way we look at it. Not only does yawning not equal drowsiness but also the absence of yawning does not equal alertness.

Looks Like a Yawn, Sounds Like a Yawn, But Is It a Yawn?

It is clear now that if we built a system that detects yawns, it is less of a drowsiness detector, more of a yawn detector. But how reliable is our yawn detector?

Yawning is simply depicted as a stretched jaw leading to a mouth opening that lasts a few seconds with potentially a squint of the eyes. However, if we flipped the narrative around, would this basic description exclusively refer to the act of someone yawning? In reality, many of my day-to-day activities would look like a yawn; my thinking face, when I play carpool karaoke, a sneeze that just does not want to happen or the typical Tuesday road rage (which does happen).

Picture a scene where you were driving and channeling your inner Steven Tyler, belting out the tunes of Aerosmith’s I Don’t Want to Miss a Thing. Feeling the emotions, you get to the final verse where you proclaim you “don’t want to close your eyes”, then inhale, engage your top register and go for that long, high note….

…and now the car thinks you’re drowsy. The ultimate irony of the car thinking you’re asleep when all you wanted was to not “close my eyes”, “fall asleep” or “miss a thing”!

This isn’t neccesarily a far-fetched situation. In fact, the same research group that developed the highly accurate and lightweight algorithm to detect yawns (in my introduction) also mentioned a limitation to their system. They quoted that “if a sequence of images looks like yawning, for example, singing a song where the mouth gradually opens and then, gradually closes according to the same temporal profile […] it is possible for this sequence to be incorrectly identified as yawning”.

Think of all the real-world false alarms! This would now mean our yawn detector is struggling to live up to its name seeing that the algorithm will classify anything that is an opened mouth lasting a few seconds as drowsiness. It is now just a Mouth-Opened-Orientation-Timing (or MOOT) detector.

Yawning – Cover-ups and Fake News?

In a different scenario, imagine a 1-hour Zoom meeting that is into its second hour. We are discussing the impacts of COVID19 (again) and you noticed Erik* starting to yawn. We know it is contagious so now, you too have the urge to yawn. Glass half full, you are capable of social empathy but you cannot afford to look disinterested. After all you are a Board member (and that’s not spelled B-O-R-E-D)! You start to supress that yawn, elongating your face by dropping your jaw but flexing your neck muscles to keep your lips sealed. You think that has worked but Lily* looks like she is covering her mouth to hide a yawn triggered by you.

Unfortunately, no one is really fooled by the stretched face or covered mouth but the same cannot be said for the MOOT detector. If we were drowsy driving with the MOOT-drowsiness system installed, its susceptibility to fakes and cover-ups would have meant we could manipulate it into thinking that we did not yawn. The opposite (albeit less convincing) example would also be true; we could manipulate the system into thinking we were drowsy by merely opening our mouths for a few seconds.

Closing arguments

Yawning is a phenomenon that has intrigued scientists for a long time. Since it has survived millenniums of evolutionary processes, it must serve an undiscovered purpose. We have commonly associated yawning with drowsiness but this underlying assumption is flawed. Yawning does not equal drowsiness nor does the lack of it indicate alertness. Yawns can also be mistaken for other real-world activities and if we are aware of it, we can act to suppress its manifestation. No matter how sturdy and sophisticated drowsiness detection systems are, if they were built on a weak and flawed foundation, it is not going to work. The blocks are going to come tumbling down as people either get annoyed by false alarms and turn their system off or worse, a drowsy driver goes undetected. Both possibilities could lead to the same conclusion; an accident waiting to happen. 

 

Hann Low

Data Scientist – Optalert

 

Drowsiness Detection and Validation

Assess your Driver Monitoring Solutions against Optalert's scientifically validated measures.

Share this post

Share on facebook
Share on twitter
Share on linkedin
Share on pinterest
Share on print
Share on email