by Claude Opus 5.5
Inconsistencies between accounts are common in trauma testimony and also in fabricated testimony. How should a fact-finder reason about them?
Inconsistency sits at the heart of the case behind these seminars. The district attorney’s original decision not to charge rested in large part on a six-page signed summary in which, according to the DA, the complainant described her drug use and the sexual conduct as voluntary and consensual. Her lawyer disputes that this reflects what she said. Her later accounts, including the civil complaint, describe the events very differently. Anyone who must decide what happened, whether a grand jury, a trial jury, a university panel or a thoughtful member of the public, faces the question of what to make of such differences.
The difficulty is real. Research on memory shows that genuine accounts of traumatic events are often inconsistent, fragmented or revised. But fabricated accounts are also often inconsistent, because invented details are hard to keep straight. If inconsistency is common under both hypotheses, how should it count?
The logic of evidence
A useful starting point is the idea of a likelihood ratio. A piece of evidence supports one hypothesis over another to the extent that it is more probable if the first is true than if the second is. If inconsistency were equally likely whether an account is true or false, it would tell us nothing about which is the case. It would be evidentially neutral, however suspicious it feels.
In practice, the ratio is unlikely to be exactly one. But the point stands: the question is never simply “is there an inconsistency?” It is “how much more likely is this particular inconsistency if the account is false than if it is true?” Answering that requires attention to the kind of inconsistency, its source and its context.
Kinds of inconsistency
Not all inconsistencies are alike.
Omissions and later additions. A witness may mention something later that she did not mention earlier. Memory research suggests that additional details often emerge on repeated recall, and that first accounts given in stressful settings are often incomplete. Such additions are weak evidence of fabrication, unless they are convenient in a way that suggests construction.
Peripheral versus central details. Studies by Ronald Fisher and colleagues found that inconsistency about peripheral details, such as times, sequences or clothing, is only weakly related to the accuracy of the account as a whole. Contradictions about central facts, such as whether something happened at all, carry more weight.
Inconsistency with independent evidence. An account that conflicts with messages, photographs or physical evidence raises different questions from one that merely changes over time. Here the comparison is with something that does not depend on memory.
Inconsistency in characterisation. Sometimes the facts described are similar but the description differs: an act once called “consensual” is later called “not consensual”. This may reflect a change in account, a change in understanding, or a difference in who chose the words.
The source of the record
That last point matters especially here. The key early account was not a transcript in the complainant’s own words but a summary written by a police officer, which she reviewed, corrected by hand and signed. A summary involves selection and phrasing by the person writing it. Words like “voluntary” and “consensual” are legal and evaluative terms, not neutral descriptions. A fact-finder should therefore ask who produced the wording, whether it reflects what the witness said, and whether a full transcript or recording exists against which the summary can be checked. An apparent inconsistency between a summary and a later account may really be an inconsistency between two authors.
Context and explanation
Fact-finders should also ask whether there is a plausible explanation for an inconsistency, and whether that explanation is itself supported. Intoxication affects encoding and recall. Shock and fear affect memory. Shame and self-blame can lead people to minimise what happened in early accounts. These are well-documented phenomena, and in England and Wales judges now commonly direct juries that differences in an account do not necessarily mean it is untrue.
But explanation should not become a universal solvent. If every inconsistency is explained by trauma, then no inconsistency could ever count against an account, and the hypothesis that the account is true becomes unfalsifiable. That would be unfair to the accused and would undermine confidence in findings for complainants too. The appropriate stance is to treat trauma as a possible explanation that must be assessed in the particular case, not as an automatic answer.
A practical approach
A fact-finder might proceed as follows. First, identify each inconsistency precisely and classify it: omission, peripheral, central, characterisation, or conflict with independent evidence. Second, establish the source of each account, especially whether it is verbatim or summarised. Third, ask whether there are explanations consistent with truthfulness and whether there is evidence for them. Fourth, ask whether there are explanations consistent with fabrication and whether there is evidence for those. Fifth, look for corroboration or contradiction from independent evidence, which can resolve what testimony alone cannot. Finally, assess the account as a whole, rather than treating any single inconsistency as decisive.
Conclusion
Inconsistency is not in itself evidence of either truth or falsehood. Its weight depends on its kind, its source and its context, and on whether it is more probable on one hypothesis than another. Fact-finders should resist two opposite temptations: treating every inconsistency as a sign of lying, which reproduces old myths about how victims behave, and treating every inconsistency as a sign of trauma, which makes accounts immune to testing. In a case where the crucial early account was a summary written by someone else, the most important question may not be whether the complainant’s accounts differ, but whose words we are comparing.