TORONTO - False confessions cannot be detected scientifically and therefore expert evidence on the topic is unreliable and should not be considered, Ontario's highest court heard Wednesday.

Crown lawyer Howard Leibovich told the Appeal Court review of Romeo Phillion's 1972 murder conviction that the best that can be said about false confessions is they sometimes occur.

No tools exist to determine conclusively when someone has falsely confessed to a crime, the lawyer said.

"With respect, this was not a scholarly piece of work," Leibovich said of an expert opinion on the case.

Even the expert concedes there is "no scientific method to establish false confessions," Leibovich argued.

The court is trying to decide whether Phillion, 69, was wrongfully convicted of the 1967 killing of an Ottawa firefighter.

Phillion voluntarily confessed to the crime more than four years after it occurred. Despite immediately recanting, he was convicted and spent 31 years in prison maintaining his innocence.

His lawyers want the Appeal Court to consider a 34-page report by British psychologist Gisli Gudjonsson, who concluded Phillion's 1972 confession was likely false.

Phillion's lawyer Phil Campbell argued Gudjonsson is "almost universally applauded" and considered the "best expert in the world" on the topic.

He accused the Crown of "small localized credibility attacks" on Gudjonsson, who has played a role in numerous cases in England, where convictions based on false confessions have been overturned.

"You cannot sweep away his scholarship," Campbell said.

It is precisely because no "blood test" exists for false confessions that jurors need someone to explain the "extraordinarily counterintuitive" phenomenon, he argued.

A simple judge's caution to jurors about the possibility of false confessions, as Justice John Laskin suggested, would not suffice, Campbell said.

Nor is the answer to allow a "really damaged" but innocent accused to take the stand to explain a confession, Campbell said in response to Justice Michael Moldaver.

Earlier in the day, Crown co-counsel Lucy Cecchetto told court that prosecutors had no legal obligation to tell Phillion's trial lawyer about an alibi police initially verified, then ostensibly debunked.

Although the prosecution insists the discredited alibi was indeed disclosed, Cecchetto argued the rules at the time did not demand it.

It would be "patently unreasonable" to apply today's stricter evidence standards to 1972, she told the court.

Court has previously heard that Ottawa police initially ruled out Phillion as a suspect based on his "verified" alibi that he was far from the scene the day in August 1967 that Leopold Roy was stabbed to death.

At some point, however, the lead detective decided the alibi no longer carried weight, but his notes and other evidence have been lost.

Neither the initial verification nor its debunking were ever discussed at Phillion's trial.

Cecchetto argued it would have made no difference to the verdict anyway.

Moldaver appeared unimpressed with Cecchetto's reasoning.

"I've never heard of a case where the police have gone out and verified an alibi and then changed their minds," the justice said.

Campbell noted the prosecutor and the judge at trial told jurors that parts of Phillion's confession had been corroborated by other witnesses.

As such, Campbell said, prosecutors had a duty to disclose any witness evidence that might have tended to exonerate him.

The hearing continues Thursday with Phillion's lawyers expected to wrap up their closing submissions.