> Pulse oximeters have some limitations. They can only employ light at two wavelengths. Thus the devices can only distinguish between hemoglobin and oxygenated hemoglobin. When carboxyhemoglobin and methemoglobin are also present, there are two additional wavelengths required for differentiation. In the presence of elevated carboxyhemoglobin levels, pulse oximetry overestimates the true saturation of oxygen as carboxyhemoglobin binds with a higher affinity than oxygen. In the case of carbon monoxide poisoning, the absorbance spectrum of carbon monoxide is very similar to hemoglobin, which results in a falsely high level of oxygen (overestimation of oxygen saturation) ...
Why only 2?
1. CO2 has effects on the human body of its own that aren't simply a lack of oxygen, and vice-versa. [0]
2. The baseline proportions involved aren't close, so even doubling CO2 isn't going to show up easily as a large swing in in oxygen%.
For example, the article references a study where the CO2 proportion going from 0.04% -to 0.25% correlates to mental problems.
Even if the watch could sample atmosphere directly, is it sensitive enough to detect a shift from 21.00% -> 20.79% oxygen?
As it's estimating oxygen in the owner's blood, it might not detect anything different at all... not if the owner's body has already compensated by breathing harder or by "underclocking" their brain to make dumber decisions.
I'm finding that pretty difficult to believe, to be quite honest with you.
And before you say "aha, carbon dioxide brain fog!" consider that I'm about a mile from the sea with a 40mph onshore breeze. This air is about as oxygenated as it gets.
I don't think it's too far-fetched for a quarter of that to cause subconscious cognitive effects, that could be measured in tests.