Poliheuristic decision theory describes how people actually choose when stakes are high: a two-stage process. First, they apply a non-compensatory screen — they eliminate any option that fails a critical dimension. No amount of upside on other dimensions can compensate. That dimension is often political or survival-related (will this cost me my job? my coalition? my credibility?). Stage one cuts the choice set down. Second, among the options that passed the screen, they use compensatory reasoning — trading off pros and cons to pick the best. Alex Mintz and others developed the model in political science to explain foreign-policy and leadership decisions; it applies wherever one dimension is "must not fail" and the rest are negotiable.
The practical point: decision-makers don't optimise over everything at once. They first exclude the unacceptable, then optimise within the acceptable. If you're designing a process or advising a decider, identify which dimension is non-compensatory for them. Is it regulatory risk? Reputational risk? A key relationship? Options that fail that test are dead before the trade-off analysis starts. The same logic works in reverse: if your proposal is being rejected, it may have failed an unstated non-compensatory screen. Find it.
Section 2
How to See It
Poliheuristic reasoning shows up when someone rules out options with a single criterion — "we can't do that because of X" — and then deliberates among the rest. The diagnostic: was there a hard filter before the trade-off discussion? Look for "that's off the table," "we're not even considering that," or "as long as we don't violate Y, we'll choose from these." When a stakeholder kills an option despite strong benefits elsewhere, they're applying a non-compensatory rule. The pattern: eliminate by critical dimension → choose among remainder. When you hear "we've narrowed it to three options," ask what eliminated the others.
Business
You're seeing Poliheuristic Decision Theory when a board rejects an acquisition that would boost revenue but dilute the founder's control. Control was the non-compensatory dimension — no amount of revenue could compensate. The board then compares only deals that preserve control. The same applies to "we won't enter markets where we can't be number one or two" or "we won't do anything that could trigger a regulatory review." One dimension screens; the rest are traded off.
Technology
You're seeing Poliheuristic Decision Theory when a CTO rules out any architecture that doesn't meet a latency SLA. Options that fail latency are eliminated before cost, complexity, or vendor lock-in are discussed. The non-compensatory dimension is "must not break the SLA." Among systems that pass, the team then optimises. The pattern appears in security ("we won't ship without X"), compliance, or scalability thresholds.
Investing
You're seeing Poliheuristic Decision Theory when an LP or allocator says "we don't invest in strategies that can blow up" or "we only consider managers above $X AUM." The blow-up risk or size is the screen; among strategies or managers that pass, they then compare returns, terms, and fit. The first stage is binary — pass or fail. The second stage is compensatory.
Markets
You're seeing Poliheuristic Decision Theory when a regulator or policy-maker rejects a class of solutions that violate a red line (e.g. consumer protection, fiscal neutrality) and then chooses among options that satisfy it. The red line is non-compensatory. Trade-offs happen only within the feasible set. Proposals that ignore the red line are dismissed before their merits are weighed.
Section 3
How to Use It
Decision filter
"Before a major decision, name the non-compensatory dimension: what would make an option unacceptable regardless of other benefits? Screen out options that fail it. Then choose among the remainder using normal trade-offs. If your proposal keeps getting rejected, ask what unstated screen it failed."
As a founder
Identify your non-compensatory dimensions — survival, control, brand, regulatory exposure — and screen first. Don't waste time optimising among options that fail a must-have. When you present options to co-founders or the board, surface the screen: "We're only considering paths that don't dilute below X" or "We've excluded anything that would require a new licence." That makes the process legible and avoids later surprises. The mistake is bringing a "great on paper" option that fails someone's unstated screen and then being puzzled when it's killed. Elicit the screen early.
As an investor
When a founder or partner rejects a path you favour, they may be applying a non-compensatory rule — reputation, relationship, or risk tolerance. Ask: "What would have to be true for that to be on the table?" The answer often reveals the dimension that can't be traded off. When you're the decider, state your screen explicitly so others don't waste time on options you'll never accept. "We only look at deals that don't require board approval" is a screen; then you optimise within that set.
As a decision-maker
Design processes that separate screening from selection. First round: does it pass the non-compensatory criteria? Second round: among those that pass, which is best? Mixing the two creates confusion — people will argue about trade-offs for options that were already eliminated by someone's screen. Make the screen explicit and get alignment on it before the compensatory debate.
Common misapplication: Treating every dimension as non-compensatory. If everything is a hard filter, you have no choice set left. Reserve the non-compensatory frame for the one or two dimensions that truly are "must not fail." Everything else enters the trade-off.
Second misapplication: Assuming your screen is obvious. Different stakeholders have different non-compensatory dimensions. The CEO's screen might be "won't damage the brand"; the CFO's might be "won't breach covenants." Elicit and align on screens before deep analysis, or you'll build options that die in the room.
Grove's "only the paranoid survive" and his focus on strategic inflection points implied a non-compensatory screen: options that left Intel vulnerable to existential threat were off the table. He would eliminate paths that failed the "will we survive?" test, then choose among the survivable options. His decision to exit memory and bet on microprocessors was preceded by a brutal screen — what could we do that keeps Intel in the game? — and then compensatory analysis among those options.
Bezos's "disagree and commit" and "type 2 decisions" can be read through a poliheuristic lens: some decisions have a non-compensatory dimension (e.g. customer trust, long-term value) that rules out options before the team debates the rest. Amazon's "working backwards" from the customer often sets a screen — if it doesn't serve the customer in the way we've defined, it's out — and then the team optimises within the acceptable set. The screen is explicit in the press release and principles.
Section 6
Visual Explanation
Poliheuristic Decision Theory: Stage 1 — eliminate options that fail the non-compensatory dimension. Stage 2 — choose among the remainder using compensatory trade-offs.
Section 7
Connected Models
Poliheuristic theory sits with models of bounded rationality, screening, and two-system thinking. The connections below either describe the same two-stage logic, the psychology of non-compensatory choice, or the limits that make screening necessary.
Reinforces
Bounded Rationality
Bounded rationality says we have limited cognitive capacity and simplify to cope. Poliheuristic theory is one simplification: instead of evaluating every option on every dimension, we screen first (reduce the set) then choose. The two-stage process is a way to make complex decisions tractable.
Reinforces
Satisficing
Satisficing is choosing the first option that meets a threshold rather than optimising. Poliheuristic screening is a form of satisficing on the critical dimension — "good enough" on that dimension is the bar; options that don't meet it are rejected. The second stage can then maximise over the remainder.
Reinforces
Elimination by Aspects
Elimination by aspects (Tversky) is a choice procedure: eliminate options that fail the most important attribute, then the next, until one remains. Poliheuristic theory is a two-attribute version: one non-compensatory dimension does the elimination; then compensatory reasoning picks among the rest. The same family of "eliminate then choose" models.
Leads-to
Decision Fatigue
Decision fatigue is the cost of making many choices. Screening reduces the number of options before the hard trade-offs, which can reduce fatigue. The connection: poliheuristic structure can make the final compensatory stage less taxing by shrinking the set first.
Section 8
One Key Quote
"Decision makers first eliminate alternatives that do not meet a certain threshold on a critical dimension … and then choose among the remaining alternatives using a more analytic, compensatory process."
— Alex Mintz, How Do Leaders Make Decisions? (2004)
The quote states the two stages. The critical dimension and the threshold are the keys — they define what gets eliminated. The practitioner's job: make the critical dimension and threshold explicit so the process is legible and so options are designed to pass or fail the screen clearly. Then run the compensatory analysis on what's left.
Section 9
Analyst's Take
Faster Than Normal — Editorial View
Name the non-compensatory dimension. Before a big decision, ask: what would make an option unacceptable no matter how good it is on other fronts? That's the screen. Get it on the table. If the CEO, board, or key stakeholder has an unstated "we will never do X," every option that does X is dead before the meeting. Elicit it early.
Screen first, optimise second. Don't spend cycles refining options that fail someone's must-not-fail test. Run the screen, get to the acceptable set, then do the trade-off analysis. Mixing the two wastes time and creates frustration when "great" options are killed for a reason that wasn't discussed up front.
If your proposal keeps dying, find the screen. When a proposal has strong support on paper but gets rejected, it likely failed an unstated non-compensatory criterion. Ask: "What would have to change for this to be acceptable?" The answer often reveals the dimension that can't be compensated. Fix that, or stop pushing.
Don't over-screen. If every dimension is a hard filter, the choice set is empty. Reserve non-compensatory treatment for the one or two dimensions that are truly existential or non-negotiable. Everything else goes into the compensatory stage.
Section 10
Test Yourself
Is this mental model at work here?
Scenario 1
A founder rejects a lucrative acquisition because it would require relocating the team. She says no amount of money compensates for losing the team. She then compares two other offers that both allow the team to stay.
Scenario 2
A board weighs three strategic options, scoring each on revenue impact, risk, and speed. They debate trade-offs among all three options on all dimensions.
Scenario 3
A regulator says 'we won't consider any proposal that increases systemic risk' and then evaluates five proposals that meet that bar on cost and effectiveness.
Section 11
Summary & Further Reading
Summary: Poliheuristic decision theory is a two-stage process: first eliminate options that fail a non-compensatory (critical) dimension; then choose among the remainder using compensatory trade-offs. Use it by naming the critical dimension, screening first, and optimising second. If a proposal is rejected despite strong benefits elsewhere, it likely failed an unstated screen — elicit it. Pair with bounded rationality, satisficing, elimination by aspects, and threshold rule.
Mintz and colleagues present the poliheuristic theory and apply it to foreign-policy decisions. The two-stage (non-compensatory then compensatory) framework is laid out clearly.
Kahneman on System 1 (heuristic) and System 2 (analytic) thinking. Poliheuristic screening maps to fast, heuristic rejection; compensatory stage to slower analysis. The book grounds the two-stage intuition in cognitive science.
Reinforces
Two-System Thinking
System 1 is fast and heuristic; System 2 is slow and analytic. Poliheuristic theory maps roughly: the screen (non-compensatory, fast) is System 1–like; the compensatory stage is System 2–like. The two-stage process uses both modes.
Tension
Threshold Rule
Threshold rules say "if X is below (or above) a cutoff, reject." Poliheuristic screening is a threshold on one dimension. The tension: if the threshold is too strict, nothing passes; if too loose, the screen doesn't help. The discipline is setting the non-compensatory bar where it should be — and only for the dimension that truly can't be traded off.
