Don't you think those 'experience correlations' 
     are based on proxy parameters?

Oh absolutely, I'm sure that's a huge portion of it. Nonetheless, I'm much more attuned to it now.

Situations with high CO2 are usually situations with lots of warm, stagnant, exhaled air in rooms filled with people.

Would I be able to detect high CO2 levels in an environment with cool, well-circulated air? I'm not sure. Certainly not as well, and perhaps not at all.

In practice, it probably doesn't matter. I don't think there are many of those pathologically counterintuitive CO2 situations. If the air seems stuffy or stagnant then you can be fairly certain it's full of CO2 and/or airborne pathogens.

Right. Especially if you own a meter as well, it's probably all good. I also know people that don't own meters who think they're in stale air as soon as it gets a warm in a room and need to see an open window constantly and then turn up the heating some more. But if you have a meter to tell you whether you were right, that argument doesn't work anymore, so I've changed my mind on whether using proxy metrics is necessarily bad!

    I also know people that don't own meters who think 
    they're in stale air as soon as it gets a warm in 
    a room and need to see an open window constantly 
    and then turn up the heating some more

Oh man, that seems really obnoxious. I don't even use an actual meter that way!

Our brains are incredibly good at sensing CO2, because it’s what drives respiration.

The “my lungs are burning and I must breath” signal is actually caused by high co2. We can’t perceive blood oxygen levels at all.

This is actually something hospitals have to be aware of. A tiny sliver of the population (1-10000, maybe) have greatly reduced sensation of co2, so can literally stop breathing (while awake!) if placed on high flow 100% oxygen.