Internally, I take the pattern and just keep cross checking it against newly observed patterns. So every idea I have really needs a percentage certainty added. In addition the ideas are never, ever complete. The are skeletons that-sort of like those graphs of mandlebrot sets-can be endlessly refined downwards with new patterns in old patterns. The first pattern can be built into larger patterns.
One of my most trusted patterns is the "ENXP Symmetry Model". This means that every thing I see an ENTP can be back predicted onto what an ENFP would do and visa-versa, but the action will be reflected in reverse via the two judging functions. Sounds cute until you find yourself looking at your own flaws-flaws you never knew you had till you found an ENTP's flaws. You learn a lot that way.
So this model gets a 95% certainty score. An ESXP Symmetry model gets a 85% certainty. An INXJ Symmetry model-A 30%-it should be there but I dont have data and I dont understand introverts and I dont understand how Ni...Thus it sits as a puzzle simmering over in the corner. Yet I will still externalize the possibility.