I have seen the opposite behaviour in a cheap security camera system. It can be set so it records only in the event of movement in the field of view of the camera, when there is no movement, the frame rate drops to save storage. It seems to do so by comparing two succesive frames and if it detects difference among the frames it records all the frames, otherwise, it only records one frame every two seconds.
This particular camera does not have anything tha could "detect" movement by itself, it only has the lens and a few IR emitting diodes. The movement detection feature can be defeated, for example by walking straight towards the camera (since it produces very little variation from one frame to the next). Walking across the field of view, in the other hand, triggers the recording mode almost inmediately.
Is my guess that a camera with liveview, or a camera with face recognition can be programmed on a similar principle. Such a feature could be practical for the kind of user who preffers to shoot in automatic all the time. The logic is fairly simple, the camera "sees" many frames, and when it detects enough of them that do not change, picks any of them to record it as a picture.
Another application of such logic could be for example to take pictures of landscape or cityscape automatically removing any moving objects, like passing cars or people. It would work as a very long exposure comprised of many "frames". The "frame difference" algorithm would be used to drop any frame that "does't fit" and after enough time, the camera would produce a picture of "everything that was still".
Indeed such simple algorithms would have numerous drawbacks, and possible would need fine tuning of parameters to suit them for particular situations, thus rendering themselves useless for the "automatic user". Besides, even a user of exclusively the auto modes can produce good results by making small adjustments, so everything is reduced to a market decission: there may not be enough buyers of the feature to make it a profitable selling feature.