I would start looking at it as an issue of thermal capacitance. The concrete changes temperature slowly, once the mass is heated it stays warm a long time. To mitigate hydrostatic pressure, the backfill around an underground garage will be porous - i.e. not in full contact with the heat sink of the earth.

The next factor is the heat load of the cars. It is not just that they produce heat, it is that a substantial portion of that heat is latent - combustion produces hot water vapor. The elevated temperature within the garage means more water vapor and hence latent heat within the microenvironment.

Finally, the quality of the good air is relative. It is going to be drawn from an urban environment at street level. And drawn at high volumes and low velocity (to reduce noise). Human comfort in an underground parking garage is not a meaningful design consideration.

There's a positive feedback loop until the point that street-level urban air begins to provide some equilibrium. Or at least that would be my guess.