The claim under examination in this essay is specific and factual, not merely interpretive. It is the apologetic assertion that the four canonical Gospels constitute reliable eyewitness testimony to the life, death, and resurrection of Jesus of Nazareth, and that this testimony provides sufficient grounds for treating the miraculous claims within them as historically credible. Not the broader question of whether Jesus existed, not the literary merits of the Gospel narratives, not the sociological function of early Christian communities. The claim is an epistemological one: that testimony of this particular kind, from this particular period, transmitted in this particular way, is reliable enough to ground belief in events that would otherwise violate everything we know about the natural world.
The reason to examine it carefully is that this argument is everywhere in contemporary apologetics. It surfaces in the work of N. T. Wright, in the popular writings of Lee Strobel, in the lectures of William Lane Craig, and in the confident Sunday-school assertion that Matthew, Mark, Luke, and John were there and wrote it down. The argument moves quickly from testimony to credibility to miracle, and it relies at every step on assumptions about eyewitness memory that the science does not support. When psychology and the law are brought to bear on the question, the foundations of the argument crack very visibly. That cracking is not an accident of framing or an expression of anti-Christian bias. It reflects what we have learned, over several decades of painstaking research, about how human memory actually works, and about how dramatically it departs from the recording device that most people imagine it to be.
Before proceeding, a word on the sceptic’s alleged bias, because the presuppositional charge is usually levelled before the evidence is even examined. The argument runs roughly as follows: the sceptic begins with a commitment to naturalism and therefore dismisses miraculous testimony in advance, not because the evidence is weak but because the sceptic has already decided miracles cannot happen. It is worth addressing directly, because if it were true it would undermine everything that follows. The short answer is that it inverts the actual burden. Sceptics do not refuse to consider miraculous claims; they require that such claims be supported by evidence proportionate to their extraordinary character. That is not bias; it is the ordinary standard applied to every other extraordinary claim in every other domain of inquiry. The longer answer will come toward the end of this essay, when the tu quoque about faith in science is examined on its merits. The evidence comes first.
1. What Eyewitness Memory Actually Is
The popular understanding of memory is that it functions something like a video recording. An experience is encoded at the moment it occurs, stored somewhere in the brain, and retrieved more or less intact when called upon. On this model, a sincere eyewitness who reports what they saw is simply playing back a stored file. If they are not lying, they are telling the truth, and the only two variables are honesty and the quality of perception at the moment of encoding.
Decades of cognitive psychology have established that this picture is entirely wrong. Beginning seriously with the pioneering work of Elizabeth Loftus in the 1970s and expanded by an enormous research literature since, the field has demonstrated that memory is not reproductive but reconstructive. When a person recalls an event, they are not retrieving a stored recording; they are actively rebuilding the experience from partial, fragmentary, and sometimes competing sources. They draw on what they actually perceived, on what they have subsequently learned about the event, on what others have told them, on their expectations about how such events tend to unfold, and on a range of post-encoding experiences that may have nothing to do with the original event at all. Every time a memory is retrieved, it becomes slightly plastic again, susceptible to revision before being re-stored.
Elizabeth Loftus demonstrated this in a series of now-classic experiments. In one paradigm, participants watched a film of a car accident and were then asked questions about what they had seen. The precise wording of the question mattered enormously. Participants who were asked how fast the cars were going when they “smashed” into each other reported significantly higher speeds and were significantly more likely to report seeing broken glass, which was not present in the film, than participants who were asked how fast the cars were going when they “contacted” each other. The information embedded in the question was incorporated into the memory itself. The participants were not lying. They genuinely believed they had seen broken glass. Their memory had been silently rewritten by a single word.
Subsequent research has extended these findings far beyond the laboratory. It is now possible to implant entirely false memories of events that never happened, memories sufficiently detailed and emotionally vivid that the person holding them will defend them under challenge. In one experiment, participants were convinced they had been lost in a shopping mall as young children, despite this having never occurred. In another paradigm, a significant proportion of participants came to believe they had committed a crime as a teenager, after brief sessions designed to encourage imaginative reconstruction of the period. Those reconstructed memories felt entirely genuine to the people holding them. The participants reported distress at recalling them. None of those remembered events had happened.
These findings are not peripheral results from a single laboratory. They have been replicated internationally, across cultures, and across widely varying experimental designs. The Innocence Project, which has used DNA evidence to exonerate more than 375 wrongfully convicted individuals in the United States alone since its founding in 1992, reports that mistaken eyewitness identification was a contributing factor in approximately 69 percent of those wrongful convictions. These were not cases of perjury. Witnesses who sent innocent people to prison for years, in some cases decades, were sincere. They believed with absolute certainty in what they had seen, and their certainty was catastrophically misplaced.
The conditions that increase the probability of false or distorted memory are well documented. They include: high emotional arousal at the time of encoding, because stress narrows attentional focus and disrupts the encoding of peripheral detail; substantial elapsed time between the event and the report, because memory decays and is revised during the interval; post-event discussion with other witnesses, because accounts contaminate each other through a process sometimes called memory conformity; leading questions or suggestive framing during recall; and strong prior expectations or beliefs that shape how ambiguous information is interpreted and stored. Most of these conditions are not merely present in the Gospel accounts. They are overwhelmingly present, as will become clear when the transmission history of those accounts is examined in detail.
2. The Transmission Problem: From Event to Text
The apologetic case for Gospel reliability typically runs as follows. The disciples were present during Jesus’s ministry. They witnessed the events recorded in the Gospels. Some of them, or people in close contact with them, wrote those events down, producing the texts we now have. The texts are therefore eyewitness testimony, and eyewitness testimony from sincere reporters is reliable evidence. There are several serious problems with each step of this argument, beginning with the most basic chronological one.
The scholarly consensus, shared by both secular historians and by the majority of New Testament scholars including many who are themselves Christian, places the composition of the earliest Gospel, that of Mark, at around 70 CE, approximately four decades after the crucifixion, which is conventionally dated to around 30 or 33 CE. Matthew and Luke are generally dated to the 80s CE. John is most commonly placed in the 90s or even the early second century, more than sixty years after the events it describes. The letters of Paul, which do not purport to be biographical accounts and which make no claim to have known Jesus personally before the crucifixion, are the earliest surviving Christian texts, written in the 50s CE.
What exists in the gap between the events and the texts is oral tradition. For decades, the stories, teachings, passion narratives, and resurrection accounts that would eventually find their way into the Gospels were transmitted by word of mouth through communities of believers scattered across the eastern Mediterranean. This is not a sceptical speculation; it is what New Testament scholars across the theological spectrum accept. The stories passed through multiple hands, were told in varying contexts to varying audiences, were shaped by the apologetic and liturgical needs of the communities that transmitted them, and were subject to exactly the kind of post-event social influence that the psychology of memory identifies as maximally corrosive to accuracy.
The anonymity problem compounds this significantly. The Gospels are anonymous documents. The names Matthew, Mark, Luke, and John were attributed to them by early church fathers in the second century; they do not appear in the texts themselves as authorial signatures. The scholarly argument that Mark was written by a companion of Peter, or that Matthew was written by the apostle of that name, is not established by any contemporary evidence. It rests on patristic testimony from Papias, writing around 120 to 130 CE, who is not regarded by modern historians as a uniformly reliable source. The claim that the Gospels represent eyewitness accounts depends substantially on these second-century attributions, not on anything the texts themselves demonstrate.
This matters enormously for the epistemological question. If a witness account is to carry evidential weight, we need to be able to trace it back to a specific person who was present at the events in question, assess that person’s perceptual vantage point, and evaluate the fidelity of the transmission from the original experience to the written record. With the Gospels, none of these things can be done with any confidence. What we have are anonymous texts, composed decades after the events, in a different language from the one spoken by the historical Jesus, who spoke Aramaic while the Gospels are written in Greek, drawing on oral traditions whose shape and content cannot be independently verified, produced in communities with strong theological commitments to particular conclusions about who Jesus was.
Defenders of Gospel reliability sometimes appeal to the work of scholars like Richard Bauckham, whose 2006 book Jesus and the Eyewitnesses argues that the Gospels preserve testimony traceable back to named individuals. Bauckham’s work is serious and deserves engagement rather than dismissal. His argument rests partly on the observation that certain names appearing in the Gospel narratives function as what he calls “inclusios,” framing devices that signal the testimony of a named eyewitness. It is a genuine contribution to the scholarly conversation. But even granting Bauckham’s framework, the fundamental problem of transmission remains. Even if the stories were originally associated with named eyewitnesses, the decades of oral transmission before those stories were written down subjected them to all the reconstructive, socially influenced revision that the psychology of memory would predict. The problem is not only with the source; it is with the channel through which that source’s account was carried.
3. Confabulation and the Sincere Witness
One of the most important concepts in the psychology of memory for this discussion is confabulation. In clinical neuroscience, confabulation refers to the production of fabricated or distorted memories without any conscious intention to deceive. It is most dramatically observed in patients with certain kinds of brain injury or dementia, who produce elaborate accounts of events that never occurred, accounts they believe entirely and defend with confidence when challenged. But milder forms of confabulation are universal features of normal human memory, not pathological exceptions to it.
When ordinary people fill gaps in their memory, they do so seamlessly and unconsciously. The reconstruction feels indistinguishable from genuine recall, because the brain does not flag the difference between a remembered detail and a plausible detail inserted to maintain narrative coherence. This is why eyewitness testimony is so dangerous in legal contexts: the witness is typically the last person to know that their account contains confabulated material. Their confidence is not a reliable guide to their accuracy. Research consistently shows that the correlation between a witness’s expressed certainty and the actual accuracy of their account is weak, particularly for fine-grained details, and that high confidence can be produced by social factors entirely independent of accuracy, including simply being asked the same question a second time, or being told that other witnesses have given a similar account.
Now consider the situation of the people transmitting the stories that became the Gospels. They were members of communities built around specific theological beliefs about the identity of Jesus. They were telling stories that had already been told many times, by many people, across communities that were under social, political, and sometimes physical pressure. The stories they told had liturgical functions; they were used in worship, in argument with opponents, and in the instruction of new members. Every one of these conditions is a known driver of the kind of unconscious narrative revision that produces confabulated detail. Stories grow more vivid and more theologically coherent with each retelling, not because anyone is lying but because human memory supplies what expectation, belief, and community require.
The textual evidence bears this out in instructive ways. The resurrection narratives, which carry the greatest theological weight and which are the most central miraculous claims in the Christian case, are also the most internally inconsistent across the four Gospels. The accounts differ on who went to the tomb and when, on what they found there, on who or what they encountered, on what they were told, on where the risen Jesus first appeared, and on what he said. These are not trivial variations in peripheral detail. If four witnesses to the same pivotal event produced accounts this divergent, a competent forensic psychologist would flag serious reliability concerns. Harmonisation of these accounts, the standard apologetic response, requires a level of textual ingenuity that itself signals the problem: if the accounts were straightforwardly reliable, harmonisation would be unnecessary.
4. The Crucible of Social Influence
The psychology of memory conformity deserves sustained attention in this context because it speaks directly to the communal transmission process through which the Gospel material was shaped. Memory conformity is the process by which individuals who discuss an event together come to share a more uniform account of it than they would have produced independently, with the dominant or more socially confident account tending to suppress minority recollections. It has been studied extensively by Kimberley Wade, Maryanne Garry, and their colleagues, among others, and it produces reliable and substantial effects under laboratory conditions.
In the context of the early Christian communities, the dynamics of memory conformity would have been operating at maximum intensity. These were close-knit groups whose social cohesion depended substantially on shared narratives about Jesus. Dissenting accounts, accounts that departed from the community’s central theological conclusions, faced powerful social pressure. The sociological evidence of Paul’s letters, which repeatedly address conflicts over correct belief and correct practice, illustrates how actively the early communities policed the boundaries of their narratives. Under these conditions, the convergence of accounts over time is not evidence of accuracy; it is precisely what memory conformity research would predict, regardless of whether the original events occurred as described.
There is also the problem of motivated reasoning, which operates differently from confabulation but produces similarly distorted outcomes. As Jerry Coyne observes: “In the end, religious investigations of ‘truth,’ unlike those of science, are deeply dependent on confirmation bias. You start with what you were taught to believe, or what you want to believe, and then accept only those facts that support your prejudices. This is the basis for the theological practice of ‘apologetics,’ designed to defend religion against counterarguments and disconfirming evidence…. In contrast, science has no apologetics, for we test our conclusions by trying to find counterevidence.” The people transmitting the Gospel material were not disinterested reporters. They were, by definition, people who had already concluded that Jesus was the Son of God and that the resurrection had occurred. Their transmissions were shaped, at every stage, by the theological framework they brought to them.
This is not an accusation of bad faith. It is a description of a cognitive process that operates below the threshold of conscious intention. Believers in the early communities would have selected, emphasised, and unconsciously embellished details consistent with their theological convictions while allowing inconsistent details to fade. This is what motivated cognition does, and it does it to all of us, in all domains. The only systematic defence against it is the kind of external checking mechanism that science provides and that the oral transmission of religious narrative does not. The early communities had every motivation to remember the resurrection accounts in a particular way, and the psychology of memory tells us precisely what happens when motivation and memory are placed in such close proximity over such an extended period.
5. What the Law Has Learned
The legal system’s encounter with eyewitness reliability over the last four decades is instructive not only for what it has established about memory but for the institutional response it has forced. Courts in both the United States and the United Kingdom spent most of the twentieth century treating eyewitness testimony as highly probative evidence, particularly when the witness expressed strong confidence. That confidence was taken by judges and juries as a reliable signal of accuracy. The Innocence Project’s DNA exonerations began to demonstrate, case after case, that this assumption was catastrophically wrong.
The response from the legal establishment has been substantial and ongoing. In the United States, the New Jersey Supreme Court issued revised jury instructions in 2012, following the landmark case of State v. Henderson, requiring jurors to be informed of the specific factors that research shows can compromise eyewitness reliability: cross-racial identification, viewing conditions, stress, weapon focus, post-event information, and the effects of time on memory. Similar reforms have been pursued in the United Kingdom, where the Court of Appeal has repeatedly emphasised the need for caution around eyewitness identification evidence, and the Police and Criminal Evidence Act codifies specific safeguards around identification procedures for exactly this reason.
The institutional response is significant because it represents the considered collective judgment of legal systems that have seen the evidence in the worst possible context: innocent people imprisoned for years on the basis of sincere, confident, and wrong eyewitness reports. The legal reform movement is not driven by philosophical scepticism about perception; it is driven by the brute empirical fact that eyewitness testimony, even from sincere, well-meaning, emotionally invested witnesses, fails at rates that no honest legal system can afford to ignore.
Translate this back to the apologetic claim. The Gospels offer testimony that post-dates the events by decades, transmitted through communities with strong beliefs about what those events meant, in a language different from the one in which the events occurred, attributed to authors whose identities were assigned by later tradition rather than demonstrated by contemporary evidence. If this testimony were offered in a modern court of law, not even as proof of the miraculous but simply as reliable historical reporting, it would face a level of scrutiny that the straightforward apologetic argument entirely evades. The apologist says: these were sincere witnesses, and sincere witnesses are reliable. The legal research of the last forty years says: sincerity is necessary but nowhere near sufficient, and here is the systematic evidence for why that is so.
There is a particular irony in the fact that the legal standard, which is far lower than the standard of scientific evidence, is itself sufficient to cast serious doubt on the Gospel accounts. We are not even asking whether the resurrection can be demonstrated by controlled experiment; we are asking whether it is supported by testimony of the quality that a modern court would regard as reliable. The answer, by any honest application of the standards that modern forensic psychology has established, is that it is not.
6. Testimony Contrasted with Repeatable, Checkable Evidence
The broader epistemological question lurking behind the specific discussion of the Gospels is this: what kind of evidence should we trust most, and why? Testimony occupies a curious position in this hierarchy. We rely on it constantly and necessarily; no individual can verify every claim they act on from first principles. The question is not whether testimony is ever reliable, but what factors increase or decrease its reliability, and how it compares with other forms of evidence when the claims being assessed are extraordinary ones.
The decisive contrast is between testimony and what we might call repeatable, publicly checkable evidence. When a physicist claims that a particle has a particular mass, or that a quantum effect has been observed in a new material, the claim is subject to independent replication. Other laboratories, using independently calibrated equipment, following independently documented procedures, will attempt to reproduce the finding. If they succeed, confidence in the claim rises. If they fail, confidence falls and the original claim is revisited. This is the mechanism that distinguishes scientific knowledge from other forms of knowledge-claiming. It is not a perfect mechanism; it is prone to its own biases, including publication bias and the difficulty of publishing null results. But it has a self-correcting structure that testimony, by its nature, does not.
The resurrection claim is, by definition, not repeatable. It happened once, at a specific time and place, and has not been reproduced under any conditions that allow independent verification. The only evidence for it is testimonial: people say it happened. The question is whether that testimony is of the quality and character that would licence confidence in an event of this magnitude, and the answer, when the transmission history, the anonymity of the sources, the divergence of the accounts, and the known properties of human memory under precisely the conditions present in the early Christian communities are taken together, is that it is not.
This is not a logically airtight disproof of the resurrection. It is a judgment about evidence, and the proper response to inadequate evidence for an extraordinary claim is not refusal but suspension. The sceptic is not asserting that Jesus definitely did not rise from the dead; the sceptic is asserting that the evidence offered for the claim does not meet the standard that the claim requires. That is a different position, and it is the honest one. It is also, incidentally, the position that any of us would take if the extraordinary claim were drawn from a different religious tradition. If someone presented accounts of the same quality for a miracle attributed to a deity in whom Christians do not believe, Christians would immediately identify all the weaknesses that the sceptic identifies in the Gospel accounts. The principle of proportionate evidence is universal; only the application of it varies according to which tradition one has inherited.
For a more developed discussion of how evidence and faith operate as competing epistemological frameworks, the earlier essay on evidence versus faith covers the ground in some detail, and the argument there complements the more specific case being made here about the limits of testimonial reasoning.
7. Answering the Tu Quoque: “You Have Faith in Science Too”
The charge that the sceptic is guilty of the same kind of faith they criticise in the believer is among the most frequently deployed moves in presuppositional apologetics. It has an appealing rhetorical symmetry: you say I have faith without evidence; I say you have faith in science without evidence. We are both, the argument concludes, in the same epistemic boat, and therefore neither of us has grounds to criticise the other’s credulity. This argument fails, and it fails for a specific and demonstrable reason, not because sceptics are cleverer or better informed, but because the two things being compared are not the same kind of thing at all.
The trust that an educated layperson places in scientific consensus is not structurally identical to the faith that religious tradition demands in miraculous testimony. The differences are several and they matter enormously. Scientific trust is provisional; it is explicitly held subject to revision if better evidence emerges, and the history of science is largely the history of such revisions. Religious faith, in the tradition the apologist is typically defending, is held to be a virtue precisely because it persists in the absence of, or in the face of, contrary evidence. These are not the same epistemic posture wearing different hats. One is designed to track evidence; the other is designed to persist regardless of it.
Scientific trust is also earned by a track record that is publicly available and independently assessable. Vaccination works; we know this because the rates of the diseases it targets have been measured, across many populations, over many decades, and have declined in consistent and predictable ways as vaccination coverage has increased. Antibiotics work; the mechanism is understood at a molecular level, the effects are measurable and have been measured, and the evidence is available to anyone who looks for it. The trust we place in scientific medicine is not blind; it is proportionate to a history of successful prediction and intervention. When that track record falters, as it sometimes does, the scientific community revises its position. This self-correcting mechanism is not a weakness of science; it is its defining strength, and it is precisely what religious testimonial tradition lacks.
Coyne makes the relevant contrast with characteristic economy: “Understanding reality, in the sense of being able to use what we know to predict what we don’t, is best achieved using the tools of science, and is never achieved using the methods of faith. That is attested by the acknowledged success of science in telling us about everything from the smallest bits of matter to the origin of the universe itself, compared with the abject failure of religion to tell us anything about gods, including whether they exist.” The track record is the decisive point. It is not that scientists are uniquely wise or that sceptics are specially gifted with reason. It is that the method of provisional, publicly checkable, self-correcting inquiry has a demonstrated record of producing reliable knowledge, and the method of deference to ancient testimonial tradition does not.
The tu quoque also typically misrepresents the sceptic’s position by implying that science is accepted on authority in exactly the way that religious testimony is. A scientifically educated person does not simply take physicists’ word for quantum mechanics in the way a believer takes the church’s word for the resurrection. They understand the general features of how the evidence was gathered, what would count as disconfirmation, and what the track record of the relevant field is. They have access, in principle, to the methods and data. The chain of epistemic warrant runs from evidence through method to conclusion, and at every link in that chain there are checks. The testimonial chain for the resurrection runs from an event through decades of oral tradition through anonymous authorship through second-century attribution, and the checks are minimal. Calling these two chains equivalent is not a philosophical observation; it is a category error that flatters the weaker argument by associating it with the stronger one.
8. The Specific Weaknesses of the Resurrection Testimony
Having established the general framework, it is worth dwelling on the specific features of the resurrection testimony that make it particularly weak by the standards the psychology of memory and the law of evidence have established. Five features stand out above the others, and they compound each other rather than sitting independently.
The first is the emotional intensity of the circumstances. The disciples, if they were among the witnesses to the crucifixion, were experiencing an event of enormous psychological and emotional significance. The man they believed to be the Messiah had been publicly executed, their movement appeared to have been destroyed, and they were living under the kind of threat that produces exactly the stress-induced perceptual distortion and post-event memory revision that the research literature identifies. High emotional arousal is consistently associated with reduced reliability in the peripheral details of memory, with a narrowing of attentional focus, and with increased susceptibility to post-event suggestion. The conditions of the disciples’ experience were not favourable for accurate encoding, and the apologetic tradition’s insistence on the emotional power of their conviction actually works against them epistemologically, because emotional intensity is precisely one of the factors that degrades rather than enhances the reliability of subsequent testimony.
The second is the elapsed time. As already noted, the Gospels were written forty to sixty years after the events. Within that period, the stories were transmitted orally, in communities with strong theological interests. Memory research is unambiguous on the effect of time: memories become less accurate, more schematic, and more susceptible to post-event revision as time passes. The longer the interval, the greater the expected departure from the original experience. Forty years is an enormous interval by the standards of the research, far beyond the range within which eyewitness accounts retain the kind of reliability that the apologetic claim requires, and sixty years is simply extraordinary as a baseline for confident historical testimony about specific events.
The third is the language gap. The historical Jesus spoke Aramaic. The Gospels are written in Greek. Whatever the original oral accounts were, they have been translated, at least once and probably more than once, in the process of transmission. Translation is not a neutral process; it involves interpretive choices that shape the content and emphasis of what is transmitted. The chain of transmission from an Aramaic-speaking fisherman’s account to a Greek-language text composed in a different cultural context decades later is long enough, and involves sufficient interpretive mediation, to constitute a serious evidential problem that the apologetic literature tends to treat as a footnote rather than the obstacle it actually represents.
The fourth is the theological shaping of the accounts. The Gospel writers were not attempting to produce neutral historical reports. They were writing, as John 20:31 makes explicit, “that you may believe that Jesus is the Messiah, the Son of God.” Their accounts are organised around a theological conclusion that was already settled before they wrote. The selection of what to include, what to emphasise, and how to present events was governed by this prior theological commitment. This is not an accusation against their sincerity; it is a description of the genre they were working in. But it means that the accounts cannot be read as disinterested testimony, and the theological shaping is exactly what motivated cognition research would predict from sincere believers reconstructing and transmitting the stories of their movement’s founder across decades of communal life.
The fifth, and perhaps most decisive, is the absence of independent corroboration. For a claim of this magnitude, the evidential bar must be set proportionately. The risen Jesus appeared, according to Paul’s first letter to the Corinthians, to more than five hundred people. If this occurred in a Roman-administered province with literate administrators, scribes, and historians, we would expect some trace of it in independent contemporary sources. No Roman or Jewish source contemporary with the events provides any corroboration. Josephus’s references to Jesus are disputed, with the Testimonium Flavianum almost certainly having been partially or wholly interpolated by later Christian scribes, and no Roman source describes anything consistent with a resurrection. The silence of independent contemporary sources is not, by itself, conclusive, but it is significant, and it stands in sharp contrast to what we would expect if an event of this character had actually occurred in the way the Gospels describe.
9. Steelmanning the Apologetic Position
It would be intellectually dishonest to proceed without engaging with the strongest version of the apologetic argument for Gospel reliability, because some of the arguments made by serious New Testament scholars are considerably more sophisticated than the popular presentation of them suggests. The case deserves a fair hearing before the conclusion is drawn.
The most sophisticated argument for reliability rests on the criterion of embarrassment. This criterion, used by historians of all theological persuasions, holds that details which would have been embarrassing to the early Christian community are unlikely to have been invented and are therefore more probably authentic. The most frequently cited examples are the role of women as the first witnesses to the resurrection, given that women’s testimony was not held in high esteem in first-century Jewish culture and that no one fabricating a resurrection story for persuasive effect in that context would have chosen women as the first witnesses; the baptism of Jesus by John, which implies submission to John’s authority and was awkward for early Christology; and the crucifixion itself, a mode of death associated with criminals and deeply incompatible with messianic expectations. The argument is that the presence of these inconvenient details is evidence that the tradition was not simply invented to serve theological purposes, because anyone inventing it would have avoided them.
This is a genuine argument with genuine force within its own domain. The criterion of embarrassment is a legitimate historical tool, and its application to the Gospel material is not without value. But its reach is limited in a precise and important way. It may establish that some core elements of the tradition have a basis in historical memory; it says nothing about whether the miraculous claims associated with those elements are reliable. A tradition can preserve an accurate recollection that a crucifixion occurred and still radically misrepresent, through decades of theologically motivated reconstruction, what happened in the days that followed it. The authentic scaffold does not validate the structure built upon it, and the criterion of embarrassment establishes only the scaffold.
A second serious apologetic argument concerns the rapidity of the development of resurrection belief. Paul’s letter to the Corinthians, which most scholars date to the mid-50s CE and which contains the creedal formula of 1 Corinthians 15, is thought by many scholars to preserve a tradition that dates to within a few years of the crucifixion, perhaps to Paul’s visit to Jerusalem around 36 CE when he met Peter and James. This is significantly earlier than the composition of the Gospels, and it establishes that resurrection belief was present in the movement very shortly after the events. The apologetic argument is that this rapid emergence is difficult to explain without a genuine historical basis for it.
The cognitive science of religion, however, provides several plausible alternative explanations for the rapid emergence of resurrection belief that do not require the literal resurrection to have occurred. The disciples were members of a milieu saturated with eschatological expectation, in which the resurrection of the dead was a widely held belief within certain Jewish traditions, particularly Pharisaic Judaism. The experience of visionary states in grief-stricken communities, the phenomenon of post-bereavement hallucinations, which are documented, common, and experienced as entirely real by those who undergo them, and the process by which charismatic religious movements generate conviction narratives in the aftermath of a founder’s death are all well-documented phenomena that require no supernatural explanation. Taken together, they constitute a set of perfectly naturalistic mechanisms capable of generating exactly the kind of early resurrection belief that Paul attests.
None of this proves the resurrection did not occur. The point is that the apologetic argument from early belief, like the argument from embarrassment, provides considerably less support for the miraculous claim than it is typically presented as providing. It establishes that something happened in the movement shortly after the crucifixion that generated intense conviction. It does not establish what that something was, and it does not establish that the written accounts produced decades later accurately describe it. The gap between “something caused this conviction” and “the resurrection occurred and the accounts are reliable” is the entire evidential gap that needs to be bridged, and neither of these arguments comes close to bridging it.
10. The Broader Pattern of Testimonial Religion
The specific case of the Gospels opens onto a broader question about the epistemological structure of testimonial religion in general. Christianity is far from unique in resting miraculous claims on testimonial foundations. Islam’s case for the prophethood of Muhammad rests on the testimony of those who knew him and of those who transmitted the hadith literature across the generations. Mormonism’s founding claims rest on the testimony of Joseph Smith and a small number of witnesses to the golden plates. Virtually every religion that makes historical miracle claims makes them on a testimonial basis, because miracles, by definition, are not the kinds of events that admit of independent, repeatable verification.
The wider pattern is revealing in a specific and uncomfortable way. People who are Christians because of the testimonial evidence for the resurrection typically do not find the testimonial evidence for the golden plates convincing, and vice versa. People who are Muslims because of the testimony to Muhammad’s prophethood typically do not find the testimonial evidence for the resurrection convincing. The symmetry is exact, and it reveals something important: the weight given to testimonial evidence in the religious domain is not determined by a consistent epistemic standard applied across traditions. It is determined by which tradition one has been born into, educated in, and formed by. The apologetic argument for Gospel reliability, if it were genuinely an argument from evidence, would have to explain why it does not also establish the truth of competing religions that deploy equally strong testimonial arguments. It never does explain this, because it cannot without abandoning the pretence that evidence alone is doing the work.
The question of historical accuracy in the Bible more broadly, beyond the specific resurrection claim, is addressed in depth in the essay on biblical historical accuracy, which covers the archaeological and documentary evidence across both testaments and reaches conclusions that reinforce the case being made here about the gap between testimonial tradition and verifiable historical record.
11. The Charge of Bias, Revisited
Return now to the presuppositional charge that the sceptic is simply biased against the miraculous. The charge is usually articulated as follows: you refuse to accept the resurrection testimony because you have already committed to naturalism, and within naturalism miracles are definitionally impossible, so you are not actually examining the evidence; you are dismissing it in advance by definitional fiat. This is a more sophisticated version of the charge than the simple accusation of bad faith, and it deserves a direct response rather than a wave of the hand.
The sceptic’s position, properly understood, is not that miracles are definitionally impossible but that extraordinary claims require extraordinary evidence. This principle, associated most famously with Carl Sagan but rooted in the older formulations of Laplace and Hume, is not a prejudice imported from outside the inquiry; it is the standard epistemic principle applied to all extraordinary claims in all domains. When someone reports having seen a large carnivorous mammal in an area where no such animal is thought to exist, we do not dismiss the report out of hand. We apply a level of scrutiny proportionate to the extraordinary character of the claim: we ask about the quality of the observation, the reliability of the observer, the existence of corroborating evidence, and the plausibility of alternative explanations. If the evidence is strong enough, the extraordinary claim is accepted. The discovery of populations of species previously thought extinct has occurred on exactly this basis. The sceptic is not operating with a special anti-miracle rule; they are operating with a consistent evidential rule that applies to all extraordinary claims, miraculous or otherwise.
The resurrection claim fails this test not because it is miraculous but because the evidence offered for it is of the quality and character that, by the standards the psychology of memory and the law of evidence have established, would be insufficient to support confident belief in an ordinary historical claim, let alone an extraordinary one. The sceptic is not raising the bar for miracles specifically; the bar is already set by the nature of the claim, and the evidence falls below it by the standards that anyone, of any religious persuasion, would apply to a claim of equivalent magnitude made on behalf of a different tradition.
There is also a form of bias worth naming on the apologetic side of this exchange. The apologist begins with the conclusion, namely the resurrection as a historical fact and as the cornerstone of Christian theology, and works backwards to assess the evidence. This is not a criticism of the individual apologist’s sincerity or intellectual capacity; it is a structural feature of apologetics as a project. Its entire purpose is the defence of a prior conclusion, not the dispassionate assessment of evidence. The researcher who begins an inquiry without a predetermined conclusion, applying consistent standards across all the evidence including disconfirming evidence, is practising a fundamentally different epistemic activity from the one the apologist is practising. The charge of bias, honestly assessed, lands far more squarely on the project of apologetics than on the project of critical historical inquiry, and the intellectual honesty of the sceptical position consists in acknowledging this asymmetry plainly.
12. What Would Constitute Convincing Evidence
A sceptical argument is only fully honest if it can say what evidence would change its conclusion, because a position immune to all possible evidence has stepped outside rational discourse entirely. The question of what would constitute convincing evidence for the resurrection is therefore worth addressing directly, not as an exercise in hypothetical theology but as a clarification of what the evidential standard actually requires and why the existing evidence falls short of it.
In principle, the resurrection could have left physical evidence. It was claimed to have occurred in a specific place and time, within a literate society with administrative records, in a period for which we have substantial documentary and archaeological remains. If multiple independent, contemporary sources, sources with no evident prior commitment to the theological conclusion, had recorded the event with consistent and corroborating detail, that would be significant evidence, not conclusive, but significant. If the tomb were located and its contents, or the absence of them, could be independently verified, that would be significant. If the post-resurrection appearances were attested by sources with no prior stake in the outcome, writing close to the events in their own language and from an identifiable position of direct observation, that would constitute a qualitatively different evidential base from the one we actually have.
None of these conditions are met. Their absence does not, by itself, prove that the resurrection did not occur. Absence of evidence is not always evidence of absence, as any philosopher will note. But for a claim that is being offered as the foundation of the most important religious and moral commitment a person can make, the absence of the kind of evidence we would normally require for far lesser claims is not a trivial epistemological problem. It is the central one, and the apologetic tradition’s response to it, which generally involves shifting the burden back to the sceptic or invoking the sincerity of the witnesses, does not engage with it at the level of rigour it demands.
The essay on the proof that never comes develops the related argument about why religious claims persistently fail to produce the kind of evidence that would settle the question, and what that persistent failure tells us about the epistemic status of those claims when held up to the light of consistent evidential standards.
13. Why the Stakes Matter Beyond the Academy
The evidential question about the Gospels might seem, at first glance, to be an arcane dispute between historians and theologians with little practical consequence for the rest of us. This impression is seriously mistaken. The claim that the Gospels constitute reliable eyewitness testimony is not merely an academic position; it is the foundation upon which enormous institutions of political, social, and moral authority rest. If the resurrection testimony is as weak as the evidence suggests, then those institutions are built on a claim that cannot withstand scrutiny, and the policies, laws, and social attitudes they generate in the name of that claim deserve to be challenged on exactly that basis.
Consider the practical consequences of treating the resurrection as established historical fact rather than as a claim that fails basic evidential standards. Doctrines about sexual morality, the status of women, the nature of marriage, the treatment of those who dissent from the faith, and the appropriate relationship between religious authority and civil law are all derived, in the traditions that maintain them, from the authority of texts whose reliability is precisely what is in question. When a church uses the Gospels to sanction particular moral positions, it is making an implicit empirical claim: that these texts accurately preserve the teachings and authority of a figure whose resurrection established him as divine. If that empirical claim is false, or more precisely if it is unsupported by evidence of the quality the claim requires, then the moral authority derived from it is similarly unsupported. The evidential question is not separate from the practical one; it is prior to it.
This is not an argument that religious people cannot be moral, or that religious institutions have produced nothing of value. Both of those assertions would be false and beside the point. It is an argument that the specific claim of authority derived from reliably witnessed divine revelation requires that the revelation in question was actually reliably witnessed, and that the evidence for this, examined honestly, does not sustain the weight of authority that has been placed upon it. The implications of that gap are significant, and they extend well beyond the theological seminar into the lives of people who are subject to the institutions that claim their authority from a testimonial foundation that will not bear the load.
Conclusion
The apologetic claim examined throughout this essay is that the Gospels constitute reliable eyewitness testimony and that this testimony provides sufficient grounds for belief in the miraculous claims they contain. Having examined the claim against the evidence that the psychology of memory, the history of legal reform, and the transmission history of the texts together provide, the conclusion is clear: the claim fails on multiple converging grounds, and it fails not because of anti-Christian bias or presuppositional naturalism but because of what we have learned, through decades of rigorous research, about the conditions under which human testimony is and is not reliable.
Human memory is reconstructive, not reproductive, and is subject to systematic distortion by precisely the conditions present in the transmission of the Gospel material: emotional intensity at encoding, substantial elapsed time before writing, post-event social influence operating through close communities, motivated cognition operating in communities with settled theological conclusions, and the absence of the kind of independent checking mechanism that the science of memory identifies as essential for maintaining accuracy over time. The Gospels are anonymous documents composed decades after the events they describe, in a different language, after extended oral transmission through communities with strong theological interests and strong social incentives for conformity of account. The resurrection narratives are internally inconsistent across the four accounts in ways that are exactly consistent with what the psychology of memory predicts from multiple independent reconstructions of emotionally charged events, filtered through decades of communal retelling.
The legal system’s response to the failures of eyewitness testimony provides an additional and sobering perspective. Courts that once treated sincere, confident eyewitness accounts as among the most powerful forms of evidence have been forced, by the evidence of wrongful convictions and by the sustained work of memory researchers, to institutionalise the very scepticism about testimonial reliability that critical examination of the Gospels demands. The standard that modern courts now apply, which is considerably less rigorous than the standard of scientific evidence, is itself sufficient to cast serious doubt on accounts that are decades removed from the events, shaped by strong community beliefs, and supported by no independent contemporary corroboration.
The strongest apologetic arguments, those based on the criterion of embarrassment and the rapidity of resurrection belief, establish that something historically real underlies parts of the tradition. They do not establish that the miraculous accounts built upon that historical core are reliably transmitted, because the psychological and social mechanisms that generate sincere conviction in communities of committed believers are sufficient to explain the development of those accounts without requiring that they accurately describe the events. The gap between “a historical person and some genuine events” and “reliably transmitted miraculous testimony” is precisely the gap that the apologetic literature consistently fails to close.
None of this constitutes a logically conclusive refutation of the resurrection. That is not the sceptic’s position, and a sceptic who claimed logical certainty on this question would be overreaching in exactly the way they criticise the believer for overreaching. The honest position is the evidential one: the testimony offered for the resurrection is of insufficient quality, by the consistent standards we apply to extraordinary claims in all other domains, to warrant confident belief. The evidence is insufficient. The responsible conclusion is suspension, not affirmation. And the institutions that derive their authority from the claim that this testimony is reliable are building on a foundation that careful examination does not support.
The larger lesson extends beyond Christianity to the general epistemological status of testimony as evidence for extraordinary claims. Testimony is necessary, valuable, and pervasive in human life, and we cannot function without it. But its reliability is a function of conditions that can be assessed, and when those conditions are poor, testimony is the most dangerous form of evidence precisely because it is the most persuasive. The sincere witness is more dangerous to accurate belief than the obvious liar, because sincerity disarms scepticism. The science of memory, hard-won through decades of research and confirmed through the tragedy of wrongful convictions, has given us the tools to assess those conditions systematically. Applying those tools honestly to the question of the Gospels as eyewitness testimony yields a conclusion that is uncomfortable for a great many people and for institutions of considerable power. Discomfort, however, has never been a good argument against an honest conclusion, and the discomfort here belongs to the apologetic tradition, not to the standard of evidence that exposes its weakness.