Such functions can be combined into 16 types, each with a four-letter code. If interested then search online for the cognitive function stacks to see what orderings are associated with each type. Everyone uses all eight functions, but each individual tends to specialize in one perception axis and one judging axis (..which tend to be a bit asymmetrical). Personality theory is based on the principle of exclusivity, meaning that there is always a trade-off associated with wiring the brain in a certain way. I will probably go into this in more detail later.
Knowledge functions are ways of knowing. Particular instances in memory can serve to validate a hypothesis on the metaphysical plane. A universal truth is an instinctual leap to a conclusion based on prior action on the physical plane.
Action can be done on either the physical or metaphysical plane. The metaphysical plane references the physical plane with metaphor. Any symbolic or abstract form is rooted in metaphor.
Interfacing can be done with either agents or objects. Agents are assumed to be capable of making selections. Objects can be referenced from conditions generated by agents.
Deliberation functions are methods of deliberation. Selections are unconditional during agent interfacing (though may be influenced by conditioning). By unconditional, I mean that there is no apparent mechanism behind it (this does not mean the mechanism is non-existent). Generated conditions eventually get selected (or not) by agents while engaged in object interfacing.
This is just my particular take on the model. If you have an alternative suggestion that still retains symmetry, then feel free to leave a comment explaining what and why. Ultimately, the accuracy of such models is dependent on how the user trains them with data. Be wary of over/under fitting, and keep in mind that models are not the real thing.