One interesting aspect of this is that if I read DeWitt correctly, he is saying that Epicurus put the strongest indicia of reliability on "deduction from first principles." And he is saying that analogy from the visible (observation?) to the invisible, and ordinary human intelligence, are less reliable.
So is DeWitt not saying that the most reliable kind of reasoning is deduction based on "first principles" which would be from something like "nothing comes from nothing?"
It is very difficult for me not to see that (deduction from a first principle) as not being well described in modern colloquial terminology with words like logic and reason.
That's why I think it's necessary to have a presentation to new students of Epicurus that would help them distinguish between acceptable forms of logic and reason and unacceptable forms.