Cognitive Functions: Find Your Stack in 4 Questions

Most function guides hand you eight definitions and wish you luck. The eight cognitive functions are really four axes — Ti–Fe, Te–Fi, Ni–Se, Ne–Si — and nobody mixes and matches across them. Pick one perceiving axis, one judging axis and the direction your strongest process points, and all four slots are decided. Here is the derivation, the 16-row table, and the three places it quietly goes wrong.

Eight functions, four axes

Every function has a partner doing the mirror job in the mirror direction, and the pair works like a see-saw: effort spent on one end is effort the other does not get. Ti chases internal consistency, Fe chases the room’s — both asking “does this hold together?”, one about the model, one about the people. Note that each judging axis holds a T and an F, and each perceiving axis an S and an N. There is no such thing as a T axis.

The rule that kills 1,664 of the 1,680 possible stacks

Eight functions in four slots gives 8 × 7 × 6 × 5 = 1,680 orderings. Exactly sixteen are legal types. Two rules do all the killing:

1. You get two whole axes, never halves of four. Your top four holds both ends of one judging axis and both ends of one perceiving axis. If Ti is in there, so is Fe. If Ni is, so is Se. What never happens is Ti beside Te, or Ni beside Ne — the other two axes are simply absent.
2. Direction alternates down the stack. Dominant inward means auxiliary outward, tertiary inward, inferior outward. So dominant and inferior are the two ends of one axis, and auxiliary and tertiary are the two ends of the other.

Two axes give you a quadrant, not a type

Choosing a perceiving axis and a judging axis narrows sixteen types to four. Those four own an identical set of functions and differ only in the order:

Here is the part that surprises people: types inside a quadrant do not resemble each other at all. ISTP and ENFJ hold the same four functions in perfectly reversed order, as do INTP and ESFJ, INTJ and ESFP, ISTJ and ENFP. Read any stack backwards and you land on a real type — the one whose dominant is your inferior, which is why those pairs find each other maddening and quietly impressive at once.

Worked case: the same four functions, two opposite people

Run one quadrant twice. Both people below settle on Ni–Se and Ti–Fe, so both own exactly {Ti, Se, Ni, Fe}. Only the entry point differs.

Case one: lands on ISTP

Perceiving axis. Given an open problem and no deadline, this person goes and touches the thing — pulls the panel off, drives the route, runs the actual query — and their hunches arrive as one read they cannot justify rather than six branching what-ifs. Ni–Se. Judging axis. They will argue a definition for twenty minutes, then be visibly awkward about the mood they just wrecked: logic private, care public. Ti–Fe. Which end leads? The internal model runs unprompted, so the dominant points inward and it is the judging one: Ti. Now the slots are forced. Auxiliary points outward from the other axis (Se), tertiary is the far end of that axis (Ni), inferior the far end of the dominant’s (Fe). Ti → Se → Ni → Fe is ISTP, derived without once looking at the letters I, S, T or P.

Case two: lands on ENFJ

Perceiving axis. Same answer, different evidence: this person tracks where a group is heading long before there is proof, arrives at one conviction rather than a menu, and under stress either forgets the physical world or over-indulges it. Ni–Se. Judging axis. Care is the public instrument — tone, timing, who got left out — while the logic stays private and turns pedantic under pressure. Ti–Fe. Which end leads? Here the split happens: the process running unprompted points outward, at people, and it is a judging function. Dominant Fe, auxiliary inward from the other axis (Ni), tertiary its far end (Se), inferior the far end of the dominant’s (Ti). Fe → Ni → Se → Ti is ENFJ — the exact reverse of case one, from identical answers to the first two questions.

What one axis sounds like from both ends

The cleanest confirmation of an axis is not your strength but your inferior misfiring — incompetence is louder than competence. Watch Ti–Fe run in both directions:

Both use Ti and both use Fe. The ISTP’s Fe arrives as an anxious demand to know the emotional score; the ENFJ’s Ti as brittle pedantry about definitions. That mirrored clumsiness is the axis showing itself, and it is much harder to fake than a strength.

The four-question procedure

1. Which perceiving axis? Ask: with an unformed problem and no deadline, does my attention branch or converge? Ne–Si fans one input into options, then checks them against precedent. Ni–Se collapses to a single read, then goes and touches reality to test it. Wrong answer: describing what your job forces on you. A client deck makes everyone branch; ask what your attention does on a Sunday.
2. Which judging axis? Ask: which settling instinct do I do out loud? Ti–Fe keeps the logic private and the care public — definitions refined internally, the room managed openly. Te–Fi reverses it: publish the plan, the metric and the owner, and never mention the value that actually drove the call. Wrong answer: picking F because you are warm or T because you are sharp. Both axes contain both letters.
3. Which end leads? Two sub-questions. Which process runs when nobody needs anything from me? That one is dominant; the other is the auxiliary you deploy on purpose. Then: dropped into a mess, is my first move to take in more or to settle something?Wrong answer: using social confidence as a proxy for direction. Plenty of dominant-introverted people are funny at parties.
4. Read the last two slots off the axes. No new question — tertiary is the far end of the auxiliary’s axis, inferior the far end of the dominant’s. Then falsify: a correct stack should name your worst days, not just your best. Wrong answer: shrugging when the inferior does not fit. A step above it is wrong, almost always step three — swap dominant and auxiliary first.

All 16 stacks

For checking your derivation, or jumping straight to a type. Every row obeys both rules: dominant and inferior share an axis, auxiliary and tertiary share the other, direction alternates.

The eight functions, one line each

• Introverted Thinking (Ti) — takes the machine apart to find the principle everything rests on, and will not use a framework it cannot rebuild.
• Extraverted Thinking (Te) — drags the world into owners, dates and measures, and treats an unmeasured plan as no plan.
• Introverted Feeling (Fi) — checks options against a private ledger it rarely reads aloud, and knows “no” before it can explain why.
• Extraverted Feeling (Fe) — reads the room’s emotional weather and adjusts its own tone to keep the room usable.
• Introverted Sensing (Si) — compares now against a detailed archive of before, and notices the two degrees this version is off.
• Extraverted Sensing (Se) — lives at full resolution in the physical present and responds before the analysis lands.
• Introverted Intuition (Ni) — collapses scattered input into one direction and then cannot show its work, which is exactly why it gets over-claimed.
• Extraverted Intuition (Ne) — fans one input into six alternatives and enjoys the fanning more than the choosing.

Three ways this goes wrong

1. Typing the aspiration instead of the default

Asked which process runs unprompted, most people name the one they respect. The tell: every example comes from a good day — the project you finally planned properly, the argument where you stayed warm. Defaults do not need good days. Describe your last three disagreements instead, the tired ones, and watch which process ran uninvited. No Tuesday-afternoon example, no dominant.

2. Back-deriving functions from the four letters

This one survives because it always produces an answer. You take INFJ, map I to introverted, N to intuition, F to feeling, and reason your way to a stack. The trap is the last letter: for introverts, J or P describes the extraverted function, not the dominant. INFJ’s J means the outward-facing function (Fe) is a judging one, while the dominant, Ni, is a perceiving function. So INFJ leads with perception and INFP leads with judgment, despite their final letters. The deeper problem is circularity: the letters summarise the stack, so deriving the stack from them returns what you put in and validates nothing.

3. Over-claiming Ni

“I just know things” is not evidence. Ne users get instant hunches too; theirs arrive in threes. Ne branches and enjoys the branching, Ni converges to one read and cannot unpack it, so certainty that arrives with a list of alternatives attached is Ne. Better, use the axis rule as a check: Ni and Se always travel together, so a genuine Ni stack shows a distinctive relationship with the immediate physical present \u2014 strained where Se sits inferior or tertiary (INFJ, INTJ, ENFJ, ENTJ), fluent where it leads (ESTP, ESFP, ISTP, ISFP). What you should not find is Ni with no Se anywhere. If neither shows up, Ni is probably not relationship with the physical present — blanking on your own body and surroundings, or compulsively over-indulging them under stress. If that relationship is unremarkable, Ni is probably not in your top four. A claim about one end is always a claim about the other.

What our own numbers say about self-typing

We can be specific here, because we ran it on ourselves. Our typing study scored 1,179 cleaned conversations and returned INTP for 42.3% and INFP for 23.3% — two types, roughly two-thirds of the sample — with INFJ at 1.2%. That is not a population estimate; it is a portrait of who uploads chat logs to an AI typing tool. Two things follow: do not use internet type distributions as a prior on yourself, and treat the gap between what forums self-report and what a scorer returns from real messages as a standing warning about step one.

The same study cross-checked two independent scorers. On introversion they agreed on 80.2% of individuals — the sturdiest signal in the set. On conscientiousness they did not: MBTI scoring called 16.1% of people Judging while Big Five scoring called 35.4% of the same people Organized, a 19.3 point gap. Notice which letter that is. J/P decides which middle function faces outward, making it the most load-bearing letter in the stack — and it is the least stable thing we measure. That is the whole case for deriving stacks from axes and behaviour rather than from a code, and for pairing any function read with the measurable traits in the Big Five guide.

Where to go next

This page is a procedure, not a dictionary. For the plain-English definitions — what each function is, how the hierarchy works, loops and grips, how each one surfaces in your prose — start with MBTI cognitive functions, explained. Read that first if the terms are new; come back here to work out which four are actually yours. Then go deep on the result: the INTP, INFJ, ENTJ and ESFP pages take one stack apart slot by slot, and the type library covers the behaviour around it.

Want a starting hypothesis rather than a blank page? The free MBTI test hands you four letters to run through the procedure above, and all five ungated quizzes live at the test index. Just remember what step four was for: a type you cannot falsify is one you are wearing, not one you found.