The most interesting thing about the graphs isn't the slope, but that "perceived score" is far more tightly coupled to the individual's perception of their overall ability than their actual score. But that could be an artefact of how and when questions about ability were asked (conjecture: I haven't read the original paper)