Hardy Jonck gives a good answer, and like most things, the answer to this question is highly context sensitive.
Everyone has to start out with simple understandings.
Everyone has to start out believing ideas like True/False.
The more time one spends investigating all the different branches of science and understanding; the more likely it is that one will shift away from such simple hard binary distinctions, to a more “relaxed” kind of knowledge that is based in contextually relevant sets of probability distributions.
In many practical contexts, the probabilities are very close approximations to “True” or “False”, such that one doesn’t normally consider the difference in practice, even as one is always intellectually conscious of the difference.
So in that sort of context, where one is conscious of many levels of eternal uncertainty in all things, from domains like non-binary logics, Goedel incompleteness, irrational numbers, maximal computational complexity of systems, halting problems, etc (it seems probable that the class of classes of such things may be infinite); then one moves into a modern scientific understanding – which is not about Truth, but about getting models of whatever this reality we find ourselves in actually is, that are sufficiently useful in the context that they work in all experimental situations (right up to the point that they don’t, thus requiring further exploration of the set of all possible explanatory frameworks to find one that works with the newest set of data as well as all that came before).
In this context, David Snowden’s Cynefin Framework for the management of complexity is the best simplification of what seems to be an infinitely complex domain space that I have come across. If one applies it, to all levels of understanding, then one can start to build some useful models.
“True” science always involves both evidence and arguments, and sometimes the arguments about the nature of evidence, and the nature of understanding are every bit as important as the evidence itself.
And when one does the hard work of deeply investigating evolution, the geological evidence, the games theoretic contexts, the biochemistry, the animal and plant behaviours, AI and neural networks, etc; then one can begin to get a useful model of the deeply nested sets of complex adaptive (and contextually sensitive) systems that go into producing this experience we have of being human – of having the understandings (models) that we do. And it seems that every conscious individual lives in the experiential reality created by their subconscious brains, and every brain is in a body that is in a “reality” that is complex beyond the ability of any brain to deal with in detail. So reality demands that we make simplifying assumptions at many different levels. So nobody actually “Knows” what reality is, and some models are much more useful and likely to survive in some contexts than other models (more complex models take a lot of time and energy to develop, and sometimes that cost is too high in some contexts – thus simpler models survive better – it is extremely complex at many different levels).
Post modern philosophy gets some aspects of this complexity, but many tend to take that and “spin” it into a “nihilism” that displays no appreciation of the evolutionary context in which they exist. Every one of us is from a lineage of survivors. All of our ancestors survived at least long enough to leave offspring that also left offspring in the particular contexts of their specific existence. That is a non-trivial filter on the sort of stories we tend to tell ourselves. People who go down the Nihilist path tend to ignore that very important aspect of reality.
In some very fundamental and very important ways, every aspect of our intuitions, our emotions, our religions, our culture, our ethics, our stories, have been deeply tuned by the survival of things. Ignoring that is simple stupidity, often with large overlays of egoistic hubris.
We are a socially cooperative species.
We are the most socially cooperative species this planet has produced.
The evolution of complexity is predicated on cooperative behaviour that has associated behaviours that detect and effectively remove non-cooperative behaviour.
And we live in very complex times, when many of the old models, and many of their over simplifications of complex systems, are now causing existential level risk to us all. Perhaps chief among them is the idea that greed is good. Not true!!! It is deeply more complex than that, and the evidence for that is beyond any shadow of reasonable doubt.