The concept of belief is intrinsically gradient. It is not binary in spite of the terms employed to reflect the concept of belief suggesting a binary quality.
Belief is intrinsically gradient. Consider another concept that is intrinsically gradient. Love is a concept that intrinsically falls on a gradient.
Let’s imagine someone says the following.
“You either love me or you don’t. Which is it?”
We would consider this person immature and inappropriately forcing you to make love binary when we all know it is not. We are not forced to say “I am a lover” or “I am not a lover”. Love comes in degrees, and to treat it as binary is to unnecessarily default to a very low-resolution, as the tag “love” can be used to cover a multitude of positions along the gradient of love.
Consider another person saying…
“You either like James Brown or you don’t. You have to decide whether you do or you don’t.”
We know “liking”, just as “believing” and “loving”, are gradient concepts. We would consider this person philosophically unsophisticated at best.
The problem seems to be the misguided notion that, because the linguistic tags are binary (e.g., belief/disbelief, love/hate, like/dislike) that these somehow inform the concept.
But linguistic tags are not prior to the concepts they are employed to represent. The concept is always ontologically prior to any subjectively derived linguistic tags that might be recruited to convey that concept.
So forcing a clearly gradient concept into a binary linguistic mold is misguided and inappropriate.
I currently give an Einsteinian god a 15% degree of certainty, but this certainty was once closer to 80%, then gradually declined.
There was nothing magical when that degree of certainty reached 50%. This 50% has no more significance than you reaching your 50th birthday. It is an arbitrary number that may be aesthetically pleasing, but it remains arbitrary. We don’t, if we are rational, give concepts significance based on arbitrary thresholds.
Imagine Mr X suggesting that he is “bald” since he has only 49% of his original hair, while calling Mr Y with 51% of his original hair “not-bald”. Imagine Mr X running around suggesting to men who have <50% of their original hair that they must call themselves “bald”.
Observe the silliness of inverting the proper heirarchy between the concept and its linguistic tag.
If a concept is intrinsically gradient, you can’t invoke linguistic use or limitations associated with that concept to modify the concept. The concept remains as it is, and linguistics, if they are to facilitate a meaningful and accurate communication of that concept, must reflect with as high a precision as possible that concept.
If “disbelief” and “belief” are inadequate to reflect the vast range of possible belief positions between 0% and 100% on the epistemic gradient, we should not create an arbitrary threshold of 50% to validate the terms “disbelief” and “belief”, but rather add the modifiers that will best reflect our actual epistemic position. These modifiers can be terms such as “slight(ly)”, “significant(ly) and “extreme(ly)”, or can be an actual percentage as I attempt to do when expressing my Einsteinian god belief.
In summary, concepts are always logically prior and ontologically foundational to any linguistic tags that we employ to reflect those concepts. Inverting this heirarchy for gradient concepts such as loving, believing and liking is, at best, to pixelate the concept, and in most cases to distort the concept, a result not consistent with the presumed goal of accurate communication.