I wouldn't agree with that. Its true that a hypothesis can never be entirely rejected or accepted. Instead we assign levels of confidence to our hypotheses. Its a problem thats plagued the social sciences, how do you quantify inherently qualitative data. Indeed as I type im working on a non-parametric hypothesis test in 3D data (cumulative freq. distributions don't work well in greater than 1D). Fortunately for me physics is my thing and my data tends to be lots of numbers so the idea of assigning probabilities and likelihoods is natural. Its harder in 'real life' but the principle is the same. I largely agree with your statement,
"Those who care about reason and logic do not accept hypothesis."
I would simply add the clause; 'or reject them'. Instead we assign some kind of subjective confidence. This can be based on many things, simple things like number of independent sources or clarity and consistency. A lot of incoherent sources maybe less convincing then a few independent but consistent ones. Theres always an element of subjectivity, we always weight our personal confidence level with an element of personal experience.
Oh and one thing you never hear "those who care about reason and logic" do, is make an absolutionist statement like 'do not accept hypothesis', event my modification is probably still to absolute.