Miracles and Hume.

▸ The Catholic Position

A miracle is not a violation of nature — it is the Author of nature acting within the order He freely authored. The regularities we call "laws of nature" are not self-existing constraints binding God from the outside; they are the habitual expression of God's ordinary will, and the same will that sustains them can, on rare occasion and for a revelatory purpose, act in a higher mode. Scripture calls these acts sēmeia (signs) and dynameis (mighty works) precisely because their function is to point — to authenticate the One who works them and to invite, never to compel, the assent of faith.

The Catholic Church therefore holds two things at once that the skeptic assumes are in tension: that miracles are really possible and really knowable as signs of divine revelation, and that God is a free personal agent who will not be reduced to a force reproducible on demand in a laboratory. The Church's own evidentiary practice is not credulous but ruthlessly skeptical — she rejects the overwhelming majority of alleged wonders and certifies a cure only after every natural explanation has been exhausted. Faith built on signs is not blind: it is reasoned assent to credible testimony.

▸ Catholic Rebuttal · MIR.1.R

Hume's argument is not a proof. It is a definition smuggled in as a proof — and beneath the elegance it is viciously circular. Examine the load-bearing claim: a "firm and unalterable experience" has established the laws, so the evidence against any miracle is "as entire as any argument from experience can possibly be imagined." But how does Hume know experience is unalterable and uniform against miracles? Only by assuming in advance that no miracle has ever occurred. If even one well-attested miracle has happened, then experience is not uniform against miracles, and Hume's "entire proof" evaporates. He counts the testimony for miracles out of the evidence pool before weighing it — which is to assume his conclusion as his premise.

First — the equivocation on "law." Hume treats a law of nature as if it were a metaphysical wall that nothing could ever cross. But a law of nature is a generalization about how nature ordinarily behaves when left to itself. It says nothing about what happens when a free agent — human or divine — intervenes. A man who catches a falling apple does not "violate" gravity; he introduces a new cause. The miracle is not nature breaking its own rule; it is a higher cause acting within a regularity that was always contingent on the will sustaining it. Augustine saw this fifteen centuries before Hume: a wonder is contrary to what we know of nature, not to nature itself, whose full order is God's will.

Second — the mathematics refute the blanket "never." Hume wrote before the probability calculus matured, and his maxim does not survive contact with it. Bayes's theorem shows that a sufficiently strong body of multiple, independent testimonies can overcome an arbitrarily low prior probability — and that with enough independent witnesses, no single witness need even be highly reliable. The philosopher of science John Earman — himself no Christian apologist — argued exactly this in Hume's Abject Failure (Oxford, 2000): Hume's argument, even on the most charitable reconstruction, cannot survive a proper Bayesian treatment, and is "largely derivative, almost wholly without merit where it is original." Hume's universal "no testimony is sufficient" is not a discovery about the world; it is a mathematical error he could have avoided had he used the developing probability calculus of his own contemporaries.

▸ Catholic Counter-Counter · MIR.1.R.S.R

The chastened argument is honest, and it deserves an honest answer rather than a triumphalist one. But it overstates the uniformity of the evidential record and understates the Church's own rigor.

On "pre-scientific, motivated, non-independent witnesses": this describes some claims and not others. The Lourdes Medical Bureau has, since 1858, examined more than 7,000 claimed cures and certified only 72 as medically inexplicable — under examination by physicians of any faith or none, applying modern diagnostic medicine, after exhaustive exclusion of natural causes. These are not first-century rumors; many are 20th- and 21st-century cases (the most recent recognized in 2025) with imaging, pathology reports, and hostile review. The witnesses include the very physicians whose professional interest is to find a natural explanation. That is the structural opposite of a closed, motivated, contaminated testimonial chain.

On the circularity charge — that the theist assumes a non-zero prior only because he already believes: this cuts in the atheist's favor only if the atheist's prior is itself neutral. It is not. To set the prior probability of any miracle at zero — which is what "no testimony could ever suffice" requires — is not the conclusion of an investigation but the refusal to begin one. Earman makes precisely this point: a prior of exactly zero for a hypothesis makes it unrevisable by any conceivable evidence, which is illegitimate; it is rationally open, and sometimes necessary, to adopt a small but non-zero prior. The Catholic does not assume the prior is high. He insists only that it is not zero — that the question of whether a personal Cause of the universe could act within His own order is left open, to be settled by evidence, not foreclosed by definition. The honest agnostic and the Catholic agree on this; only the doctrinaire naturalist does not.

On the practical bar: the Church does not rest the case on Lourdes. The decisive miracle is the Resurrection — public, early, multiply and independently attested, attested by hostile and initially unbelieving witnesses (Paul the persecutor, James the skeptical brother), and world-altering in a way no statue-rumor is. That is the dossier the chastened Humean must actually engage, and it is treated in its own cluster.

▸ Catholic Response to the AI · MIR.1.R.S.R.AI.R

The AI's three moves are the strongest secular package available, and each fails on inspection.

On (1), "medically inexplicable means not yet explained" and the base-rate point: this conflates two different claims. The Lourdes Bureau does not certify a cure as miraculous merely because current medicine cannot explain it. It requires that the disease be objectively diagnosed and serious, that the cure be instantaneous or near-instantaneous, complete, lasting (verified over years), and without convalescence — features that spontaneous remission does not display. Spontaneous remission of, say, multiple sclerosis or advanced tuberculosis is gradual and partial; the certified Lourdes cures are abrupt and total. The base-rate appeal also commits a statistical error: you cannot compare 72 certified cures against 200 million pilgrims, because the 72 were selected against the base rate — every case explicable by remission was already excluded. The comparison class is not 'all pilgrims' but 'cures that survived a multi-decade naturalistic elimination,' and against that class the residue is precisely what a non-zero-prior framework predicts.

On (2), "unfalsifiable, therefore methodological naturalism is the only workable epistemics": this is the move's fatal overreach. The Catholic claim is not that every anomaly is a miracle — it is rigorously falsifiable, because the Church's own procedure is built to falsify miracle claims and succeeds in the overwhelming majority of cases. A hypothesis that is rejected in roughly 6,928 of 7,000 cases is the opposite of one that 'absorbs any data.' And the demand that all inquiry be confined to naturalistic causes is not a finding of science; it is a philosophical stipulation imported before any evidence is examined. To say 'only natural causes may be considered, therefore no evidence can ever point to a supernatural cause' is the very circularity the AI accused the theist of — methodological naturalism, asserted as exhaustive, simply is Hume's circle restated.

On (3), the Resurrection "defers and circles": deferring to the strongest case is not circularity; it is locating the argument where it belongs. The Resurrection witnesses include the demonstrably hostile (Saul of Tarsus, who was persecuting the Church when he was converted by what he reported as an encounter with the risen Christ — Acts 9, Gal 1:13-16) and the initially unbelieving (James, named among the resurrection witnesses in 1 Cor 15:7, whom John 7:5 records as not believing during the ministry). That is exactly the independent, adverse-interest testimony the chastened-Hume argument demands and claims never exists. The AI's confident 'never cleared the bar' is a conclusion it has not earned — it has assumed the bar is unclearable, which is Hume's circularity wearing a 21st-century coat.

▸ Catholic Rebuttal · MIR.2.R

The STEP study is excellent science and it refutes a theology no Catholic holds. It tests, and disproves, the proposition that God is a vending machine — a mechanical force that dispenses a statistically reproducible output when the correct petitionary input is supplied. That proposition is not Catholic doctrine. It is something Scripture explicitly forbids.

First — the test-God experiment is condemned by Christ Himself. When Satan urges Jesus to throw Himself from the Temple to force a demonstration of divine protection, Jesus answers: "You shall not put the Lord your God to the test" (Mt 4:7, quoting Dt 6:16). A double-blind trial designed to coerce God into a measurable signal on demand is, structurally, exactly the test Christ refused. The Church classifies this under the sin of tempting God — putting His goodness and power to a contrived proof. A God who consented to be the dependent variable in an RCT would not be God; He would be a force of nature, which is precisely what He is not.

Second — God is a free personal agent, not a mechanism. A person's love is not extractable by averaging over trials. If a researcher tried to prove a husband loves his wife by measuring whether saying 'I love you' on a fixed schedule produces a reproducible physiological response, the failure of the experiment would prove nothing about the love — it would prove the experiment misunderstood what love is. Petitionary prayer is the address of a child to a Father whose response is ordered to the child's true good (which is often not the absence of complications after surgery), not the operation of a lever. Christ repeatedly refused to give signs on demand to those who came testing rather than seeking.

Third — and decisively — the Church's actual evidentiary practice is the opposite of credulous. The skeptic imagines the Church gobbling up anecdotes. In fact, the Lourdes Medical Bureau has examined more than 7,000 claimed cures over more than 160 years and certified 72, under skeptical medical review, after exhaustive exclusion of natural causes. The Church built an institution to disbelieve miracle claims. The survivorship-bias charge is answered not by anecdote but by a body of physicians whose job is to find the natural explanation first.

▸ Catholic Counter-Counter · MIR.2.R.S.R

The 'heads I win' charge is the serious one, and it deserves a precise answer rather than a dodge. The answer is that the Catholic position is not symmetric between the two cases, and it is not unfalsifiable.

The asymmetry is principled and was fixed in advance. The Church did not invent 'God won't be tested' after STEP failed; it is a doctrine articulated by Christ at the outset of His ministry and codified for centuries before any RCT existed. The rule is stated ahead of the data: God, being a free personal agent, does not deliver coerced, reproducible signals on demand. This is a substantive theological commitment that makes a prediction — namely, that no test designed to compel God's hand will succeed. STEP confirmed that prediction. Far from being embarrassed by the null result, classical theism expected it. An RCT proving prayer 'works' on demand would actually have falsified the Catholic doctrine of God by reducing Him to a manipulable force.

The position is falsifiable — and is in fact falsified thousands of times. The Lourdes Bureau rejects the overwhelming majority of the claims it examines. A claim is falsified the moment a physician identifies a plausible natural cause, a misdiagnosis, an incomplete cure, or a relapse. A standard that throws out more than 6,900 of over 7,000 cases is not a god-of-the-gaps that 'absorbs any data'; it is a sieve calibrated to catch almost nothing. The 72 are the residue that survived active attempts to explain them away.

The base-rate objection is statistically backwards. You cannot divide 72 certified cures by 200 million pilgrims, because the 72 were not drawn at random from the pilgrim population — they were drawn from the tiny subset that presented serious organic disease, was medically documented before and after, recovered suddenly and completely, and survived decades of relapse-monitoring. The base rate of spontaneous remission applies to gradual, partial recoveries; the certified cures are categorically different — instantaneous and total, of conditions (advanced tuberculosis, organic blindness, multiple sclerosis) that do not remit that way. The correct comparison class is not 'all pilgrims' but 'documented cures of grave organic disease that no physician of any faith could explain after exhaustive review,' and against that class the residue is not noise.

▸ Catholic Response to the AI · MIR.2.R.S.R.AI.R

The AI's package is the best secular version of the objection, and each strand unravels on its own terms.

On (1), 'post-hoc, predicts every outcome': this is false, and the falsity is checkable. The doctrine does not absorb every outcome — it specifically forbids the interpretation the AI claims it would have embraced. A statistically reproducible prayer effect on demand would not have been welcomed as 'God rewarding faith'; it would have contradicted the doctrine, because it would make God a coercible mechanism, which classical theism denies He is. The doctrine of divine freedom makes a genuine forbidden prediction: no experiment engineered to compel a measurable divine output will succeed. That is falsifiable in the strict sense — produce one such experiment that robustly succeeds and the doctrine is in crisis. None has. The prediction is not retrofitted; it is two thousand years old (Mt 4:7) and it held.

On (2), 'no known natural cause is not a supernatural cause / yesterday's miracles become today's mechanisms': this proves too much and too little. Too much, because by the same logic no historical inference to any unobserved cause could ever be warranted — yet science routinely infers unobserved causes from their effects. Too little, because the Lourdes criteria are not merely 'currently unexplained' — they require features (instantaneity, completeness, permanence, of grave organic disease) that no future mechanism of spontaneous recovery could supply without ceasing to be spontaneous. 'Yesterday's miracles became mechanisms' is true for lightning and eclipses, which were always lawful regularities misread as portents. It is not true for the sudden, total, permanent disappearance of organic pathology — that has never been folded into a natural mechanism, because spontaneous remission, by its nature, is gradual and partial. The AI is asserting an inductive promissory note it cannot cash.

On (3), 'gerrymandered reference class / special pleading': the charge inverts the actual logic. The Church did not draw the boundaries of the reference class to keep the cures it wanted; it drew them to exclude everything explicable — and the overwhelming majority of claims fall outside the line. A standard built to reject its own favored conclusion in almost every case is the antithesis of special pleading. The honest naturalist position is not 'the evidence is null'; it is 'there exists a small, hard residue of sudden total cures of grave organic disease that no physician of any faith has been able to explain, occurring in a religious context, and I choose to bet they will someday be explained naturally.' That is a faith commitment too — a bet on future naturalistic explanation that the evidence does not yet underwrite. The Catholic and the naturalist are both reasoning past the data; only one of them admits it. And the deepest answer is Luke 16:31: a will closed to God will not be convinced even by a man rising from the dead — which is exactly why God authenticates Himself by the Resurrection and not by an RCT.

▸ Catholic Rebuttal · MIR.3.R

The objection assumes the Catholic standard is 'any reported wonder authenticates my religion.' That is not, and has never been, the Catholic standard. The Catholic standard is demanding, consistent, and openly stated in Scripture — and it is more skeptical of Catholic claims than the objection imagines.

First — the standard is not evidence-alone; it never was, and Scripture says so explicitly. Deuteronomy 13 commands that even a prophet whose sign genuinely comes to pass must be rejected if he leads toward false gods. The criterion of authenticity was always twofold: evidential and doctrinal. A wonder must be real and must point to the true God in continuity with revealed truth. This is not an ad hoc filter invented to protect Catholic miracles; it is an ancient biblical rule that the Catholic applies to Catholic claims first.

Second — Catholicism positively expects counterfeit wonders, including impressive ones. Christ warns that "false christs and false prophets will arise and show great signs and wonders" (Mt 24:24). Paul warns of the lawless one who comes "with all power and with pretended signs and wonders" (2 Thess 2:9). The Catholic does not need to deny that a Ganesha statue appeared to drink milk or that a faith-healer produced a striking event. The Catholic can grant a preternatural occurrence and still reject the system it serves — because Scripture predicted exactly such signs and commanded their rejection on doctrinal grounds. The Church claims no monopoly on the anomalous.

Third — the Church is the harshest skeptic of her own claims. The same Bureau that certified 72 Lourdes cures rejected the other thousands examined. The Church rejects the overwhelming majority of alleged apparitions; she demands medical exclusion of natural causes, doctrinal orthodoxy, and good moral fruits. The objection's picture of a credulous Catholic waving through any wonder is a caricature; the documented reality is an institution engineered to disbelieve. The standard is even. It simply yields different verdicts because the evidence and the doctrine differ from case to case.

▸ Catholic Counter-Counter · MIR.3.R.S.R

The 'rigging' charge confuses two different things: using a conclusion as a premise (which is circular) and using independently grounded criteria that happen to favor one tradition (which is not). The Catholic doctrinal filter is not 'whatever confirms Catholicism'; it is a set of criteria — internal coherence, moral fruit, historical credibility, continuity with prior revelation — that can be stated and tested without presupposing Catholicism, and that the rival claims fail on their own merits.

First — the criteria are publicly specifiable, not tribal. Take the cases. The Ganesha milk phenomenon (1995) was, within days, given a fully natural explanation: capillary action drawing liquid up the porous stone and spreading it across the surface, reproducible on any non-sacred object. It fails the evidential test, not merely the doctrinal one — no Deuteronomy 13 filter is even needed. The moon-splitting of Qur'an 54:1 rests on a single tradition with no independent contemporaneous corroboration and an exegetical dispute within Islam itself over whether a past event or a future eschatological sign is meant; it fails the multiple-independent-attestation test. The doctrinal filter is the last screen, applied only to wonders that have already passed the evidential ones — and most rival claims never reach it.

Second — the doctrinal screen is itself rationally grounded, not arbitrary. The principle 'a sign that contradicts established truth or produces evil fruit is to be rejected' is not 'reject what isn't mine.' It is the recognition that truth cannot contradict truth: a wonder that would authenticate a self-contradictory or morally corrupt system is, for that reason, not a divine authentication, because God does not certify falsehood. This criterion is available to any rational inquirer and is exactly what the AI's own demand for 'good fruits' and 'coherence' would require. The Hindu can run the symmetric move only by claiming his system is the coherent, fruitful, historically grounded one — which is an argument to be had on the merits, not a stalemate. Symmetry of form is not symmetry of strength.

Third — the Resurrection is not 'just another contested case'; it is evidentially unlike a statue rumor. It is public (not a private vision), early (proclaimed in Jerusalem within the lifetime of hostile eyewitnesses who could have produced the body), multiply and independently attested (the 1 Cor 15 creed, the four Gospels, the explosive growth of a movement built on it), and attested by adverse-interest witnesses (Paul, James). No rival tradition's foundational miracle has this evidential profile. The Catholic does not say 'trust my filter'; he says 'examine the strongest case by neutral historical criteria.' That is the opposite of question-begging.

▸ Catholic Response to the AI · MIR.3.R.S.R.AI.R

The AI has stated the objection at full strength. It still fails, point for point.

On (1), 'which propositions count as fixed truth is itself the disputed content': not all of it is. Some of the criteria the doctrinal screen enforces are not sectarian at all — they are the laws of logic and the moral law accessible to natural reason. A wonder offered to authenticate a self-contradictory claim (e.g., 'God is and is not one') is rejected not because it's un-Catholic but because it's incoherent, and incoherence is a tradition-neutral defeater. A wonder offered to authenticate manifest evil (Dt 13's 'go after other gods' that demand child sacrifice; the 'pretended signs' of 2 Thess 2:9 serving the 'lawless one') is rejected on the moral law any conscience can read. The screen is not purely doctrinal; its outer layer is reason and conscience, which the Hindu and the Muslim and the atheist share. The AI's 'it's all sectarian' is false: part of the filter is public reason.

On (2), 'selective rigor — you leave Lourdes unexplained but debunk Ganesha': the asymmetry is in the evidence, not the rigor. The same naturalistic hypotheses the AI wants applied to Lourdes are applied — by the Lourdes Bureau itself, staffed by physicians of any or no faith, which rejects the great majority of claims precisely by finding natural explanations. Ganesha was explained in days because capillary action explained it; the 72 Lourdes cures remain after the natural hypotheses were tried and failed. 'Apply equally aggressive naturalism' is exactly what the Church does — and the AI cannot cite the natural mechanism that explains the certified cures, because none has been produced. Demanding that the cures be called 'natural' with no mechanism on offer is not rigor; it is the naturalism-of-the-gaps mirror image of the god-of-the-gaps it condemns.

On (3), 'the Resurrection sources are in-house, not neutral': this misunderstands what historical method requires. All testimony about any ancient event comes from someone with a standpoint; the historian does not discard sources for having a viewpoint but weighs them for independence, proximity, and adverse interest. By those neutral criteria the Resurrection dossier is strong precisely where 'in-house' sources are usually weak: it includes a hostile convert (Paul, who was killing Christians), an initially unbelieving family member (James), a creed (1 Cor 15:3-7) widely dated to within a few years and naming witnesses then alive to be cross-examined, and a movement that arose in the very city where the tomb was and the body could have been produced. The Hindu's 'structurally identical appeal' is structurally identical in form only — produce the rival foundational miracle with a hostile-witness conversion, an early creed naming living checkable witnesses, and a public-square setting, and the Catholic will weigh it by the same rules. None exists. The contrary-miracles argument assumes all the dossiers are evidentially equal; examined, they are not. That examination is the opposite of methodological naturalism's refusal to look.