Personally I find any attempt to measure (or mentally pre-measure or should I say preventively observe) light below LV12 to be so tricky to be useless. When the human iris begins to open we are not aware of it and so we cannot compensate.
I would never try to guess the light inside of a church for instance.
Situations between LV15 and LV12 more than being judged by the "eye" are judged by experience, observing the scene and what can influence the lighting (such as white walls, or dark walls near the scene) and "reasoning" about them.
Because the eyes are very adaptable and are totally unreliable as a measure instrument the mental appreciation of light should be thought as more the result of a reasoning based on experience rather than on a visual impression. The height of the sun over the horizon, the amount of clouds, the presence of a bright reflected surface can be observed "objectively" and there is no margin of error in this observations. The error comes when trying to infer, or mentally calculate, which is the lighting created by these factors.
The only other situations where I think things become easy is floodlit monuments at night, where LV4 normally works decently well.
I would in any case confirm my mental hypothesis with a light meter any time I have it with me and if it's the right kind (for monuments at night, for instance, an in-camera light meter is in many situation pretty useless and LV 4 probably works better unless one has a proper instrument such as an incident light meter or a spot light meter). Mentally pre-calculating the exposure is as said in another thread important to avoid silly mistakes such as ISO setting mistakes.
A floodlit monument at night outside of the EV 3-5 range would make me pause and doubt, just as an exposure far from EV14 in sun light. It's just a way to be mentally aware and vigilant about exposure instead of relying solely on the instrument. Instruments are precise but their use can prone to error. That's true especially for light meters.
In any case for B&W the entire reasoning should be skewed toward overexposure in case of least doubt.