Rationality Without Foundations

Pancritical rationalism and evolutionary epistemology

Justify a belief and you have used other beliefs to do it. Justify those and the process repeats. Every chain of justification faces the same three exits, and all of them are bad: stop at some belief declared exempt from the demand (dogmatism), keep the chain going forever (infinite regress), or route it back into itself (circularity). The trilemma is ancient, and the official response of Western epistemology has been to take the first exit and try to make it respectable — to find stopping points so secure that exempting them from doubt is not dogmatism but insight. Descartes nominated the Cogito. Kant nominated the categories. The logical positivists nominated elementary reports of sense experience. The program is called foundationalism, and its record is a series of confident anchorings followed by embarrassed retreats, because the fatal question is always askable and never answerable from inside: why should that foundation be immune? Whatever is offered in reply becomes the new bottom of the chain, and the question moves down one floor.

In What Is Philosophy For? I argued that the regress of justification never terminates in observation or data; it terminates in commitments — and I promised to show that it need not terminate in foundations. This chapter pays that debt. The escape route runs through Karl Popper, who found the door, and William Warren Bartley, who walked through it.

Popper’s Half-Escape

Popper’s great move was to reject the demand for justification in science. Theories are conjectures. They are never proven, never even rendered probable by accumulating confirmations; they are tested, and the ones that fail are discarded. Knowledge grows not by building upward from secure foundations but by exposing bold guesses to refutation and keeping whatever survives. This is a genuine escape from justificationism — for scientific theories.

But Popper left one thing standing in the old posture: the rationality of criticism itself. Ask the critical rationalist whether his own commitment to reason is open to the treatment he prescribes for everything else, and the position wobbles. If the commitment is exempt, then rationalism rests on exactly the kind of unquestionable foundation it denies to everyone else — and the fideist wins the exchange. You too, he says: your allegiance to reason is a faith like mine, adopted without grounds, protected from scrutiny at the core. Popper himself conceded that the choice of rationalism must in the end be made without reasons — a minimal concession, he thought, confined to a single point. But a foundationalism of one axiom is still foundationalism, and a critic needs only one immune belief to press the charge.

Everything Is Criticizable

Bartley’s Pancritical Rationalism (PCR) eliminates the last vestige. It rests on three commitments:

1. Universal fallibilism. Every proposition, belief, or framework is open to rational criticism. No exceptions — not logic, not the standards of criticism, not PCR itself.
2. Rejection of justificationism. Rationality does not consist in grounding beliefs on something secure. It consists in exposing them to everything that might overthrow them.
3. Reflexive criticism. PCR explicitly applies to itself. It invites its own refutation and stakes its standing on withstanding it.

The insight is radical and simple: there are no ultimate foundations, only provisional positions under continuous test. Bartley separated two things that had been fused since Plato — holding a belief rationally and justifying it. Justificationism demands that a rational belief be derived from something already secure, which is what generates the trilemma in the first place. Criticism demands only that the belief be exposed: held in a way that lets counterargument and counterevidence reach it. A position is rational not because of where it came from but because of what it has survived and what it remains open to.

Notice what this does to the regress. It does not terminate the regress; it dissolves it. The regress was powered entirely by the demand for justification — each answer had to be grounded, so each answer raised the question again. Withdraw the demand and there is nothing left to regress on. Rational assessment becomes comparative rather than absolute: the question is never “is this theory grounded?” — nothing is — but “does this theory withstand criticism better than its available rivals?” That question can be asked and answered without ever touching bottom, because it does not need a bottom.

And notice what it does to the reflexivity problem. Earlier rationalisms tried to evade the circle — reason vouching for reason — and were embarrassed by it. PCR steps into the circle deliberately. The commitment to criticism is itself a position under test: if someone shows that unbounded criticism is incoherent, or self-undermining, or inferior to some rival stance, PCR loses by its own standard. That exposure is not a weakness to be papered over. It is the entire content of the position. A rationalism that could not lose would be one more dogma; this one holds itself hostage to its own rule, and that is precisely why the fideist’s you too no longer lands. There is no immune core left to point at.

Three Objections

The regress of criticisms. Doesn’t unlimited criticizability just relocate the regress — critique of the critique of the critique, forever? No, because nothing requires the series to be completed. Under justificationism the regress was mandatory: a belief was rational only if its grounds were, so the whole chain had to be secured before the first link counted. Under PCR criticism proceeds where there is a live challenge and rests where there is not. An unexamined belief is not thereby irrational; a belief shielded from examination is.

Paralysis. If nothing is ever settled, how does anyone act? By acting on the best-tested position available while leaving it open to revision — which is what every competent practitioner already does. The demand that action wait on certainty is justificationism’s demand, not reason’s. High confidence is fine; the pilot descending through fog trusts her instruments and bets her life on them. What she does not do is freeze the update rule — and that distinction, between confidence and immunity, is the same one that separates trust from faith in Against Faith. Rationality guides action through continuous improvement, not through guarantees.

Who defines legitimate criticism? If everything is criticizable, the standards of criticism are too — so aren’t we adrift, with no fixed rule for what counts as a good objection? The response is to accept the premise and deny the panic. Standards of criticism evolve under criticism, exactly like the beliefs they scrutinize. Logic, statistical method, experimental controls, peer review — every one of these is a technology that was proposed, attacked, refined, and partially replaced, and each remains in service because it keeps surviving the treatment. The standards are dynamic because they are participants in the game, not referees standing outside it. This is not a defect of the position. It is the same story told one level up.

Rationality as an Evolutionary Strategy

Put the pieces side by side and a familiar structure appears: variation, selection, retention. Conjectures are variants; criticism is selection pressure; what survives is retained — provisionally, until a fitter rival or a sharper test arrives. PCR is not merely compatible with evolutionary epistemology; it is evolutionary epistemology stated as a norm. Ideas are evolutionary entities. They earn their keep not by being certified at birth but by surviving in an environment that is trying to kill them.

Taking the evolutionary framing seriously explains something that otherwise looks like a paradox. Matthew Pirkowski observes that people trained most rigorously in formal logic often struggle most when they confront complex adaptive systems — markets, ecosystems, cultures, minds. The rigor becomes the weakness. In mathematics, a single counterexample kills a theorem, and should: the domain is fixed by stipulation, so failure anywhere is failure everywhere. But a complex adaptive system has no single fixed domain. Its populations live across many niches at once, and a strategy that is fatal in one context can be an adaptive breakthrough in another — evolution demonstrates this constantly, rescuing tomorrow’s innovation from today’s disaster. Against that kind of system, the logician’s absolutes — solution, proof, refuted — lose their grip. The concepts that keep their grip are navigation, experimentation, and adaptation.

Rationality itself should be understood the same way: not as a fixed logic engine dispensing eternal verdicts, but as an adaptive search strategy — an iterative algorithm that generates variants, harvests feedback, updates, and moves, without ever anchoring itself to a certainty it would then have to defend. (What that updating demands of an actual fallible agent — the priors, the rationalization traps, the courage — is the business of The Discipline of Updating.) This is why science succeeds. Not because it delivers final answers — it never has — but because it is the most ruthlessly evolutionary institution we have ever built: variation by conjecture, selection by experiment, retention by replication. Certainty is not science’s product; certainty is what degrades science, because a community that believes “Science Will Solve X” has adopted a dogma and stopped exploring, which is the one thing an adaptive system cannot afford. David Deutsch draws the optimistic corollary: precisely because no explanation is ever final, there is no ceiling — explanations can improve through criticism without limit.

None of this demotes logic. Deduction remains the sharpest selection instrument we possess; within any fixed framework it is ruthless and irreplaceable. Evolutionary rationalism simply situates it: logic serves the navigation, it does not dictate the destination. Truth — conditional validity, as All Truth Is Conditional defines it — does not change under this framing. What changes is our picture of how creatures like us get more of it: not by arriving, but by climbing. Truth-seeking is navigation, and what we call our knowledge at any moment is just the current front of the climb — the set of positions that have so far survived the selection pressure of criticism.

Commitments, Not Foundations

Now I can discharge the worry this volume has been accumulating. Conditionalism says every truth claim holds only relative to conditions. The conditions themselves rest on further conditions, and Truth Machines closed the last exit: Gödel and Turing guarantee that no interpreter can certify itself, so there is no ultimate vantage — truth is locally consistent and globally ungrounded. The natural response is vertigo. If it is conditions all the way down, what holds the edifice up?

Bartley’s answer, and mine: nothing holds it up, and nothing needs to. The demand for a ground floor is justificationism’s demand, and justificationism is false — not sadly, regrettably false, but well lost, because the demand was incoherent from the start and every attempt to meet it produced either dogma or despair. Frameworks are not buildings, which stand only if something bears their weight. They are organisms, which stand only if they survive — and survival is a matter of exposure, not support. A conditional framework is held rationally when it is held open: its conditions stated, its rivals engaged, its update rule live.

So the regress terminates in commitments, exactly as promised — but a commitment is not a foundation. A foundation is a stopping point exempted from criticism; a commitment is a starting point offered up to it. The two can look identical from outside, held with the same confidence, acted on with the same resolve. The difference is structural, and it is the whole difference between dogmatism and rationality: not what you hold or how hard you hold it, but whether anything is permitted to change your grip.

There is no ultimate vantage, and now no ultimate ground either — and knowledge grows anyway, faster than it ever did when we thought it needed both. Rationality was never a matter of standing on something solid. It is a way of moving: conjecture, exposure, selection, revision, again. Truth is not a destination we will someday reach and rest at. It is a path, and rationality is how we navigate it.