Finally we must briefly mention the Subjective Interpretation. It identifies probability with a person's degree of conviction or belief in one proposition or another. Surprisingly, Subjective Interpretation is, in my view, a realistic one, since it aims to apply probability theory to the real world. For, when we say that ``Mr. Smith has subjective probability 0.9 for that the lovely bay mare Willow will come in first'', we are talking about a realperson's degree of belief in a realevent, a real winning of a real mare, shown by the photo-finish.
Beyond the obvious problem that subjective probability could be applied only for a very restricted part of reality, Subjective Interpretation suffers from the same difficulty as the propensity approach: It claims the existence of a separate quantity, the degree of belief of a person, which is not yet empirically defined. How, then, is the numerical value of the degree of belief determined?