There isn't one, really.
What a camera does is 'think' everything is a uniform gray color that reflects 18% of the white light hitting it and exposes to that level. In other words, if you aim a camera at an 18% gray card and print the resultant photograph, it would (should) look identical to the card you photographed. Why?
In an 'average' scene, this is the sort of gray you would get if you mixed everything visible into a big soup. With a bit of playing around, you can actually do this on your own photographs and get remarkably similar and stable results. I've done it.
Where the 18% gray exposure metering falls flat, however is when you're imaging considerably darker or lighter images. The old 'black cat in a coal shed' would be way, way OVER-exposed by the camera's metering system as it tries to replicate the 18% gray reflectance (when the image itself would only be 'reflecting' about 8% of the light hitting it. In this case, you would have to under-expose by about 2 stops from the metered value to get the 'correct' exposure. Similarly, snow scenes tend to reflect 60~70% of the light and the camera would only be expecting 18% so it would under-expose - causing dirty gray snow. To get the correct exposure, you need to add a couple of stops over and above the camera's metered value.
And that's not even going into tricky lighting where it's important to correctly expose only a small part of a scene - and I daren't venture into 'artistic' license...where the exposures - regardless of what we think, could be construed as 'correctly exposed' to give the artist's desired results...