What memory science establishes
Cognitive psychology has, over decades of research, established several consistent findings about the reliability of long-delay witness testimony:
- Memories are reconstructive, not recording. The brain does not store events like a video recorder. Memories are reconstructed each time they are recalled, and each reconstruction can alter the stored memory.
- Memory accuracy declines with time. Long-delay memories are substantially less accurate for specific details (timing, sequence, peripheral observations) than short-delay memories, even when witnesses report high confidence.
- Repeated retelling modifies memory. Each time a witness tells the story of an event, the story itself changes what the witness remembers. By the time the story is told in court years later, it may differ materially from the original event even in the witness’s own recollection.
- External information integrates with memory. Information the witness learns after the event — from press coverage, colleagues, police interview, legal preparation — can integrate with the original memory in ways the witness cannot detect or correct.
- Confidence and accuracy are weakly correlated. A witness who is highly confident about a long-delay memory is not necessarily more accurate than one who is less confident. The confidence-accuracy relationship weakens with time.
How the memory-science framework applies to Letby witness testimony
Dr Ravi Jayaram on Child K
The single retrial conviction in July 2024 rested substantially on Dr Jayaram’s eyewitness account of the ET-tube dislodgement. The event was in February 2016. The retrial was in June–July 2024 — eight years later. Independent commentators have pointed out differences between Dr Jayaram’s contemporaneous 2016 notes and his 2024 oral testimony. This is not a claim of dishonesty; it is the memory-science-expected pattern. See our Child K ET-tube detail analysis.
The consultant team’s 2022–2023 testimony
The consultant paediatricians who raised the 2015–2016 concerns gave testimony at the 2022–2023 trial — six to eight years after events. Between the events and the testimony, the consultants had: participated in multiple internal reviews; been interviewed by Cheshire Police; consulted legal representatives; read press coverage; prepared witness statements; conferred with each other. Each of these is, in memory-science terms, a source of post-event information that would have integrated with the original memory.
The nursing-workforce testimony at Thirlwall
Nursing witnesses at the 2024 Thirlwall Inquiry were giving evidence nine to ten years after the cluster events. The specific operational details they could recall were necessarily subject to substantial memory-reconstruction. The picture they collectively reconstructed (unit under severe operational strain, Letby as competent and caring colleague) is consistent across witnesses, which is evidentially more reliable than any individual long-delay recollection — but still within the memory-reconstruction frame.
Contemporaneous records are more reliable
What the memory-science framework establishes is not that long-delay testimony is worthless; it is that contemporaneous records are substantially more reliable. The Countess of Chester Datix logs, internal-review minutes, contemporaneous clinical notes, and 2015–2016 email correspondence are more evidentially weighty, in memory-science terms, than the 2022–2023 witness recollections of the same events. See our contemporaneous medical notes evidence.
The Thirlwall Inquiry’s documentary-record exercise is therefore evidentially more reliable than the trial’s witness-recollection exercise. This is part of why the post-conviction record, drawing on contemporaneous documents, reaches different conclusions from the trial, drawing on long-delay testimony.
What UK courts know about long-delay memory
UK criminal courts routinely hear long-delay testimony in historic-abuse cases, and the memory-science literature on those cases has entered judicial awareness. The specific principle that long-delay testimony needs corroboration is well-established in the authorities on historic abuse. In the Letby case, the long-delay witness testimony was not systematically corroborated against the contemporaneous documentary record in the way modern best practice would require.
Why this matters for conviction safety
The CCRC review can weight the evidence by reliability. Long-delay witness testimony is less reliable than contemporaneous documentary evidence. Where the two conflict, the contemporaneous record should prevail. On the Letby record, the contemporaneous documents support the systems-failure reading that the Panel has developed; the long-delay witness testimony was pressed into the single-actor reading at trial. The CCRC can apply the memory-science reliability hierarchy in re-evaluating conviction safety.