Andrew L. Urban
Of my five books dealing with cases I regard as wrongful convictions, the most recent two examine cases of alleged historical sexual and physical abuse dating back over four decades. The only ‘evidence’ against the men was say-so based on memories of the complainants. They were found guilty. Now Ben Roberts-Smith is to face trial over alleged war crimes in Afghanistan about two decades ago also based on say so memories. Lawyers doubt his trial could be fair.
More books will be written about the failures of the legal system as a result of the Ben Roberts-Smith prosecution. No matter what the result. The furore began when federal police arrested him arriving with his daughters in an unwarranted public display at Sydney airport on Tuesday, April 7, 2026. It is not unreasonable to expect the subject of witness memory reliability at trial to be a feature in all of them. It certainly was in mine.*
The most recent commentary on the dangers inherent in a Roberts-Smith trial comes from Chris Merritt, vice-president of the Rule of Law Institute of Australia, writing in The Australian today (April 10, 2026).
This country needs to come to terms with the fact that there is a real risk that Ben Roberts-Smith will not be given a fair trial. This would be the worst possible outcome – far worse than having a decorated former soldier convicted on five charges of war crime murder.
If we cannot provide a fair trial for Roberts-Smith it would shake confidence in the justice system and send a terrible signal to those in the military as well as potential recruits.
That outcome is not inevitable. But the judge who presides over his trial will face an immense task in ensuring it does not lead to injustice.
There is no clear rule on how far judges will allow witnesses to go in attempting to recall what happened in the distant past, but it’s worth considering the view of the late Sir Laurence Street, another former Chief Justice of NSW.
Street conducted a royal commission in 1983 and expressed grave doubts about the reliability of testimony about events that were just five or six years in the past.
“In some cases, suspicion underwent subtle change to belief, which itself progressed to reconstruction, which in turn escalated to recollection,” Street wrote.
There is, of course, no suggestion that the witnesses who will give evidence against Shultz and Roberts-Smith are afflicted by such problems.
Nor is there any suggestion that judges are incapable of warning juries to treat testimony with caution – or even extreme caution.
Even so, these cases will take the courts into uncomfortable territory: how can they do justice over events in the distant past where the critical factor is human memory – which is not infallible?
It was clear in my own research that memory degrades and becomes reconstructive rather than photographic. Unreliable even after a few years, never mind decades. Prominent barrister and former president of the Law Council of Australia, Arthur Moses SC, has repeatedly highlighted the difficulty of relying on recollection in cases involving events from many years earlier. A criminal prosecution relying on soldiers recalling events from Afghanistan more than a decade ago will require careful scrutiny of witness reliability. Cross-examination will focus heavily on inconsistencies between recollections. And his views are universally held.
A former NSW Senior Crown Prosecutor has commented in media discussions about war-crimes prosecutions and delayed trials, noting that honest witnesses can still be mistaken.
By the way, I would gently edit Chris Merritt’s statement that “If we cannot provide a fair trial for Roberts-Smith it would shake confidence in the justice system…” I would add the word “further” before the word “shake”…
Adding to these challenges, the trial jury will not comprise Roberts-Smiths’ military-relevant peers. Former barrister Louise Clegg writes (The Spectator Australia, April 2026): In the United States, such allegations would be prosecuted by court martial.
This matters. The criminal law assumes access to crime scenes, forensic evidence, and witnesses who can be tested contemporaneously. Yet even the Office of the Special Investigator has acknowledged the realities: no access to Afghanistan, no post-mortems, no ballistic evidence, often little more than photographs and recollections.
None of this is to suggest serious allegations should not be investigated, or grave wrongdoing excused. But the system now deployed was not designed for these circumstances. It reflects a fusion of two impulses: the late-20th-century drive to universalise international (including war) criminal law, and the older constitutional commitment to trial by jury in ordinary courts.
I make no defence of Ben Roberts-Smith. But there are real questions about delay and cost, the institutional fit of a modern jury trial for wartime conduct, and whether parliament in 2002 fully appreciated the divisions it might create in a more polarised world when it copy and pasted the most expansive vision of international criminal justice into domestic law.
The politicisation and weaponisation by political actors and media of our most decorated living soldier through this very public process will be devastating to the morale of our defence forces – and to recruitment and retention – at the worst possible time. We are embarked on yet another legal circus that will be used to rip the country apart. Long story short, this process should be playing out in a system of military justice, not the ordinary criminal courts.
Well, that has not been an option since Roberts-Smith left the military. The judge and jury in this trial will carry a massive burden of responsibility to safeguard the trial that will, from the start, look like a case meant for text book study be law students. Every conceivable element of a fair trial is captive to the combination of circumstances – and time.
What a helluva way to head into ANZAC Day….
*Presumption of Evil – Noel Greenaway – sexual abuser or collateral damage?
Presumption of Evil 2 – Frank Valentine Guilty by Say-So
In a transcript, when I come across a witness being permitted to ‘refresh’ his or her memory by going to their notes, I tend to take that exercise with a grain of salt. What is being refreshed? One’s memory of what occurred or one’s memory of what one wrote in one’s notes after it occurred?
Another aspect of memory is seen in Sue Neill-Fraser’s observation during her trial, when she said, “The trouble is if you talk about something enough it becomes a memory” – Court Transcript [2010 10 11: p1182 2].
That had been in response to a question as to whether she remembered being shown the ‘red jacket’ which had been found on the brick fence across the road from the rowing-club spit.
The memory question: ‘Do you remember if…’ is one that is capable of producing ambiguous answers. The examination of Meaghan Vass by Mr Shapiro is a case in point:
It should be remembered that Vass’s evidence at the trial was largely of the ‘I don’t remember’ type … or replying ‘No’ to questions such as ‘Do you remember if you…’ type. These were deliberately framed Qs from Shapiro … all with the ‘do you remember?’ proviso attached. (The exception is the Q about ever being on the Four Winds).
Here’s a glorious example of a couple of the ‘Do you remember’ type questions that Mr Shapiro put to Meaghan Vass:
[Trial Transcript: p629 25-36]
And I can’t resist a second example from Mr Shapiro:
[Trial Transcript: p630 1-14]
Do you get the idea? Happy to get Vass outright to say she’s never been on the Four Winds but clouds other questions with the ‘Do you remember’ formula. Gunson in XXN does not break out of that straitjacket, except before the jury (after the initial ‘Basha Inquiry’) to tax her with her confusion about where she – a self-described homeless girl – had been living at the time (of Bob Chappell’s disappearance)… ‘Nasty badgering’ … as Mr Ellis later referred to it.
I acknowledge Steven Fennell’s contribution (in Comments) which touch on his own ‘matter’ and in which he makes various observations about memory-based evidence. For those wanting to understand Steven’s remarks concerning his own-matter, please see:
QLD Appeal – Dismissed
[ https://classic.austlii.edu.au/au/cases/qld/QCA/2017/154.html#fnB61 ]
HCA Leave to Appeal Application – Leave granted. [https://www.austlii.edu.au/cgi-bin/viewdoc/au/cases/cth/HCATrans/2019/58.html?context=1;query=fennell;mask_path=au/cases/cth/HCATrans ]
HCA Appeal – Acquitted
https://www.austlii.edu.au/cgi-bin/viewdoc/au/cases/cth/HCATrans/2019/186.html?context=1;query=fennell;mask_path=au/cases/cth/HCATrans
HCA Reasons for Acquittal
[ https://www.austlii.edu.au/cgi-bin/viewdoc/au/cases/cth/HCA/2019/37.html?context=1;query=fennell;mask_path=au/cases/cth/HCA ]
The Illusion of Memory and the Weight of Silence – A Lived Perspective
The arrest of Ben Roberts-Smith on 7 April 2026 forces a confrontation with an uncomfortable truth; the Australian legal system continues to treat human memory as though it were a photographic record, when in reality, it is closer to shifting sand. There is a dangerous myth at the heart of criminal justice: that memory is reliable. It is not. Memory is not fixed. It is not immune from influence. It is reconstructive capable of being reshaped, reinforced, and, at times, entirely reimagined. This is no longer controversial. It is settled science. Yet, in courtrooms across Australia, memory continues to be treated as evidence of fact.
The “Miracle of Memory”
In my own matter, I witnessed what can only be described as a miracle of memory. Witnesses who initially provided incomplete, contradictory, and at times factually incorrect accounts later returned with clearer, more detailed recollections. These changes did not occur in isolation they followed repeated interactions with police and prosecutors. Initial oversights were “corrected” through addendums. Contradictions evident at committal were smoothed over by trial. By the time these witnesses stood before a jury, their accounts were no longer raw recollections they were refined narratives, delivered with confidence. But confidence is not accuracy. Statements evolved. Versions shifted. Evidence contradicted earlier accounts and, at times, each other a husband and wife consorting, confiding, and I posit corrupting each others versions. Yet the jury accepted these recollections as credible. The prosecution, skilfully, presented factually flawed and inconsistent testimony as reliable “gold.” This is not an anomaly. It is precisely what science predicts.
What the Science Actually Says
Modern cognitive research has dismantled the assumption that memory is reliable. The work of Elizabeth Loftus demonstrates that memory can be altered or entirely fabricated through suggestion. Even subtle changes in questioning, word choice, tone, repetition can reshape recollection. The well known “misinformation effect” shows how easily post-event information can overwrite original memory. More importantly:
• Memory does not simply fade—it evolves
• Repeated recall strengthens whatever version is last expressed
• Exposure to external narratives alters recollection
• Confidence increases even when accuracy does not
Once altered, a false memory can become indistinguishable from a genuine one. In practical terms, a witness who initially recalls little or recalls inconsistently may later present with clarity, coherence, and certainty. Not because the memory improved. But because it was reconstructed. “Reconstructed evidence is about as useful as a screen door on a submarine.” For the courts and police I say that “Reconstructed memory is the legal equivalent of a rumour that’s been rehearsed.” The danger isn’t that reconstructed memory is weak; it’s that it sounds strong enough to convict.”
The Investigative Amplifier
What is less openly acknowledged is the role investigative processes play in shaping memory. Repeated interviews do not simply extract memory, they influence it.
Scientific literature confirms that:
• Suggestive questioning can introduce new details
• Repetition reinforces altered versions of events
• Interaction with authority figures can reshape recall
This creates what I experienced firsthand: memory becoming “better” over time. Not more accurate, just more persuasive.
The Jury Problem
Here is where the system fractures, despite overwhelming scientific evidence, juries continue to treat memory as highly persuasive, often more persuasive than it deserves to be. Research consistently shows that:
• Eyewitness testimony is compelling to juries
• Yet it is not particularly reliable
• Judicial warnings about memory have limited impact
This creates a dangerous paradox: The least reliable evidence is often the most convincing.
The Strategic Gamble: Silence
Against this backdrop sits one of the most entrenched strategies in criminal defence, make the prosecution prove its case. The logic is simple. If the prosecution lacks hard evidence DNA, fingerprints, ballistics, then silence preserves the advantage of speaking last. The “ducks guts” of closing. But closing submissions are not evidence. Judges remind juries of that explicitly. So what happens in practice?
• The prosecution presents a narrative
• Witnesses present confident memory
• The defence presents nothing
Juries, being human, fill the silence. Not with doubt but with acceptance. I vehemently disagreed with that approach in my own case. The High Court of Australia later dismantled key pillars of the prosecution’s argument, points that could have been challenged at trial through evidence and witnesses I had identified. Had those rebuttals been presented, the outcome may well have been different. Instead, silence handed the field to the Crown.
The “Tentacles” of Memory-Based Cases
In cases devoid of forensic evidence, prosecutions rely heavily on motive and memory. As Sir Laurence Street observed: “Suspicion underwent subtle change to belief, which itself progressed to reconstruction, which in turn escalated to recollection.”¹ That progression is not theoretical. It is exactly what occurs when memory is reinforced through repetition and authority. Memory-based cases develop “tentacles” only extending outward, drawing strength not from independent corroboration, but from internal consistency built over time.
When Memory Fails the Test of Law Australian courts have already confronted the dangers of memory-based prosecutions.
Fennell v The Queen [2019] HCA 37
The High Court unanimously overturned my conviction, finding that the circumstantial, memory-based evidence was insufficient to establish guilt beyond reasonable doubt.
Lesson: A jury’s acceptance of a narrative does not make it legally sustainable.
Pell v The Queen [2020] HCA 12
The High Court quashed Cardinal Pell’s convictions, ruling that the jury ought to have entertained reasonable doubt.
Issue: The case relied heavily on a single complainant’s memory, contradicted by substantial opportunity evidence. The Court found the account contained compounding improbabilities.
Mallard v The Queen [2005] HCA 68
Andrew Mallard’s conviction was overturned after 12 years in prison.
Issue: Investigative failures and reliance on flawed interpretations of evidence contributed to a miscarriage of justice.
Other Cases
Cases such as R v Button, R v Jama, and R v Keogh further demonstrate how untested or misunderstood evidence, often intertwined with memory, can lead to wrongful convictions.
The System Knows—But Does Not Adapt The science is not emerging. It is settled.
• Memory can be altered
• Memory can be contaminated
• Memory can become more convincing as it becomes less reliable
Yet the legal system continues to treat it as though it were evidence of fact, rather than a product of influence. That is the real danger. Because when memory is repeated, reinforced, and presented with confidence, it does not merely persuade; It convinces. And once it convinces a jury, the distinction between truth and reconstruction can disappear entirely.
Conclusion: A Two-Edged Sword
As high profile, memory based prosecutions continue, including those built on decades-old recollections, the system faces a fundamental question. Will it adapt to what science has already proven? Alternatively, will it continue to rely on assumptions that place undue weight on the most fragile form of evidence? Silence in the face of such evidence is not neutral. It is a strategic gamble. Sometimes it works. Sometimes it fails catastrophically.
My experience shows that when memory becomes the foundation of a case and is left unchallenged it can carry a jury to conviction, even when the underlying evidence is flawed. The burden must remain where it belongs; on the prosecution to prove guilt beyond reasonable doubt .But that burden means little if the evidence used to meet it is fundamentally misunderstood.
Will there be justice or will this be a courtroom circus led by witnesses’ trotting out memories which may or may not have assisted by the parties involved or worse for those that are looking for a book deal after the outcome which is a win-win for book sales.
Footnotes & Sources
1. Street, L. (1983). Royal Commission into Certain Committal Proceedings against K.E. Humphreys.
2. Fennell v The Queen [2019] HCA 37; 267 CLR 504.
3. Pell v The Queen [2020] HCA 12; 268 CLR 123.
4. Mallard v The Queen [2005] HCA 68; 224 CLR 125.
5. Loftus, E. F. — Research on false memory and reconstructive recall.
6. Loftus, E. F., & Palmer, J. C. (1974). Reconstruction of Automobile Destruction.
7. Research on misinformation effect and memory contamination.
8. Studies on confidence vs accuracy in eyewitness testimony.
9. Rakoff, J., & Loftus, E. — Jury response to memory evidence.
10. R v Button.
11. R v Jama.
12. R v Keogh.
Sadly, all the points in Steven Fennell’s comment are all too familiar to readers of wrongfulconvictionsreport.org. Clever minds in the legal fraternity should seek a way to safeguard against wrongful convictions where memory is the only ‘evidence’.