My knowledge of logic and naive bayesian probability is naive and bayes-ic compared to a lot of much smarter people I know around here.
But I found this an interesting concept: Flipping through “The De-Mathematisation of Logic” By Barry H. Slater and I came across this interesting set up:
Bayes is used in logic for “belief” but in the literature they talk about the “Zero problem” – that is, it’s a bad thing when it’s zero because there’s no updating.
But the author makes a good simple point. If 1=Belief and -1=Disbelief then what can 0 be?
Wonder. 0=Wonder
I like this. Three-valued, allows switch between “continuity of belief” (Bayes but where a 0 was undesirable) and classic logics which assign hard settings of TRUE / FALSE or “BELIEF / DISBELIEF” (although there’s plenty of inbetween ones in some folks’ formulations of belief, but this is the first time I’ve seen 0 for wonder before – or 0 being allowed at all. It’s like an inertia. Everything keeps going as it was, whatever it was, which in this case is “I wonder”, neither believing nor disbelieving.
—
