Eyewitness Sketches Provide ‘Stormy’ Results, Can Lead to False Convictions

Posted on by

Yves here. I’ve always assumed forensic sketches were lousy. How can anyone describe a face accurately enough for another party to make a rendering? And then the individual giving the description can too easily have their memory distorted by the process of creating the sketch, that they will become invested in believing that it is a good representation and the sketch will supplant their memory, even assuming their internal image was highly accurate in the first place.

By Steve Horn, a San Diego, CA-based Research Fellow for DeSmogBlog and a freelance investigative journalist. Originally published at Criminal Legal News

Stephanie Clifford, the adult film actress known as Stormy Daniels, has set off a flood of media coverage on her claims of having a sexual relationship with President Donald Trump and the subsequent non-disclosure agreement and $130,000 paid to her by Trump’s personal attorney, Michael Cohen, during the last weeks of the 2016 presidential campaign.

The saga catapulted to the forefront of U.S. political discourse in the aftermath of a January article published in The Wall Street Journal about the $130,000 payment and the 2006 sexual affair, which took place just four months after Trump’s wife, Melania Trump, gave birth to their son Barron. Some campaign finance experts have said this is an illegal in-kind campaign contribution. Cohen has come under FBI investigation and could potentially face charges in the U.S. District Court for the Southern District of New York for his actions.

On a parallel track, in a March interview on CBS’ 60 Minutes with Anderson Cooper, Daniels said she was threatened while in Las Vegas in 2011 by a man who did not give her a name, saying, “Leave Trump alone. Forget the story.” That man then turned toward Daniels’ infant daughter and said, “That’s a beautiful little girl. It’d be a shame if something happened to her mom.” Daniels told Cooper in the interview that she could remember his face perfectly some seven years later and would recognize him instantly if she were to see him today.

After dozens upon dozens of media appearances, Daniels’ attorney and former Democratic Party opposition researcher Michael Avenatti, put up a $131,000 offer for the #thugsearch (the hashtag he used on Twitter) in the aftermath of hiring famed forensic sketch artist Lois Gibson to draw a picture of the man based on Daniels’ description of him. The picture, first made public on an episode of the television show The View during a joint appearance on the show by Avenatti and Daniels, has since gone viral online. Some have joked that the man in Gibson’s sketch looks a bit like Daniels’ husband, Glendon Crain, or a bit like New England Patriots quarterback Tom Brady. In those jokes lay a hidden truth about the lack of reliability of those kinds of sketches.

Avenatti has since filed a defamation lawsuit on behalf of Daniels in the U.S. District Court for the Southern District of New York in the aftermath of President Trump saying the photo looks like her husband, Crain, and him calling it a “total con job” in a tweet.

Lost in this soap opera and wire-to-wire coverage of the escapades, though, has been any substantive conversation about the accuracy (or lack thereof) of sketches of the sort Gibson did for Avenatti and Daniels. But, in fact, many legal scholars have pointed to their fallibility as evidence used in the U.S. criminal justice system, which raises questions about the $131,000 payment offer put out by Avenatti. Legal scholars have raised questions about the accuracy of the sketches themselves, which have been responsible for over 70 percent of the 349 of wrongful convictions overturned by The Innocence Project since 1989. They also have called into question people’s ability to recall what someone looks like some seven years after an incident occurred, let alone seven weeks.

Faded Memory Under Stress

In their chapter titled “Eyewitness Memory for People and Events,” which appears in a broader academic textbook about the topic titled, Handbook of Forensic Psychology, psychology professors Gary Wells and Elizabeth Loftus — of Iowa State University and University of California-Irvine, respectively — write that physical evidence and memory-based evidence have something key in common: They both can be contaminated.

“Like physical evidence, memory trace evidence can be contaminated, lost, destroyed, or otherwise made to produce results that can lead to an incorrect reconstruction of the event in question,” they write. “Like physical trace evidence, the manner in which memory trace evidence is collected can have important consequences for the accuracy of the results.”

At its root, Wells and Loftus write that bad science drives bad eyewitness accounts, and thus bad sketches. They say that, at its root, courts and law enforcement have no central theory to anchor how memory works at a scientific level, and that’s what generates bad outcomes.

“It appears that the justice system is assuming implicitly that information stored in the brain remains largely unchanged as a function of post-event information and is relatively impervious to suggestion, and that memory failures are primarily failures to retrieve information,” they write. “In fact, however, memory reports are readily influenced by post-event information, are very susceptible to suggestion, and can err in numerous ways.”

In another 2017 article by Wells and his colleague John Wixted of the University of California-San Diego, the two of them explain that eyewitness testimony is not inherently unreliable, but that law enforcement agencies do not utilize the actual scientific method to ensure that innocent people are not wrongfully convicted. For law enforcement and prosecutors, doing real scientific method would be time-consuming and labor-intensive, the procedure for which is drawn out by the authors in their paper titled, “The Relationship Between Eyewitness Confidence and Identification Accuracy: A New Synthesis.”

The academic debate, then, says it is at least feasible that eyewitness identification can work if a criminal justice investigation and pre-prosecution investigation are conducted in a thorough manner. But then there is the reality, as is the case with Stormy Daniels, that her eyewitness event occurred under self-described duress.

“I remember going into [a] workout class [after the incident]. And my hands are shaking so much, I was afraid I was gonna drop [my daughter],” Daniels told 60 Minutes .

Daniels also said she was “100 percent” sure she could recognize the person who threatened her if she ever saw him again.

“Even now, all these years later. If he walked in this door right now, I would instantly know,” Daniels proclaimed.

A seminal study published in 2008 by Tin Valentine and Jan Mesout, psychology scholars at the Goldsmiths, University of London, concluded that eyewitness identification while under duress can be seen as a de facto fatal flaw in terms of its accuracy even in the short aftermath of a stressful event. For that study, Valentine and Mesout surveyed a random sample of people who went into the Horror Labyrinth in London by getting informed consent and strapping a heart rate monitor around their chests.

“These data suggest that the accuracy of eyewitness identification by some victims of crime has been overestimated in data from laboratory studies that do not induce an activational mode of attentional control,” the study concluded. “This conclusion suggests that the potential for error associated with eyewitness identification evidence is even greater than laboratory studies have suggested.”

Sketchy Sketches

Even if someone’s memory has perfect recollection of what a person looked like, it’s still up to a sketch artist to draw the sketch according to the description. In the case of Daniels and her attorney Michael Avenatti, they hired sketch artist Gibson, who formerly worked for the Houston Police Department.

Sketches, Iowa State University psychology professors Gary Wells and Lisa Hasel wrote in a 2007 paper, “produce poor likenesses of intended faces.”

“The difficulty that people have with constructing face composites is not the result of weak memory for faces per se,” they wrote. “Instead, it appears that human face processing is designed more for face recognition, which is facilitated by holistic representations, than it is for face recall, which requires individual feature representations.”

Put another way, even if Daniels had a fairly clear memory of the man’s face, it would unlikely to be specific enough of a description to yield a foolproof sketch by Gibson. And, even if the description was perfectly solid by Daniels, there’s no guarantee that the sketch would be up to snuff because of a concept called “piecewise reconstruction.” That’s a fancy way of saying that in verbalizing a face’s description piece by piece to a sketch artist, feedback received by the person describing the face could bias one’s memory and subsequent description of that face to the artist.

A “verbal description itself degrades the eyewitness’ visual memory as well as the recognizability of the target face, in a phenomenon known as verbal overshadowing,” explains a 2011 scholarly study titled, “Do You See What I See? A More Realistic Eyewitness Sketch Recognition,” published by scholars at National University of Singapore and Ramon Llull University in Barcelona, Spain. And “piecewise reconstruction of the target face, especially in composite sketches, degrades the eyewitness’ mental image of the target face and the recognition result.”

“Memory Fades”

In an email to Criminal Legal News, University of California-Irvine Professor Elizabeth Loftus said it is possible that Daniels has a great memory. But that does not mean the sketch will end up as an accurate or actionable one for Daniels’ legal team.

“Maybe she has a strong memory for the event itself and that convinces her she will be able to identify him,” said Loftus. “But we know how much memory fades over seven years. A huge number of mistakes of identification have been made after much shorter periods of time.”

In response to questions sent via email, Gibson told Criminal Legal News that “I do not do interviews or answer questions about Stormy Daniels.” She also did not respond to general questions asked about the reliability of eyewitness sketches.
Avenatti also did not respond to multiple emails and tweets seeking comment on the issue. He has since made his Twitter profile private, citing “bots” and “abusive trolls.” Daniels’ attorney has appeared hundreds of times in the media since the initial “60 Minutes” interview, but has not been asked a single time in any of those appearances about the scientific reliability of eyewitness sketches.
Print Friendly, PDF & Email


  1. David

    Not exactly new – Elizabeth Loftus has been writing about this for decades. Eyewitness evidence is no longer used in prosecutions in many countries without some other kind of support: there have been too may miscarriages of justice. More generally, psychological experiments have shown that people can easily be induced to remember things that never happened to them, or to forget or misremember things that did, even extending to historical events that never happened. Here is a popular summary. What’s worse is that studies show that events which are stressful are much less likely to be remembered correctly than those that are not – for example few people will remember the details of an assault well enough to give useful evidence to police. In practice, as any psychologist will tell you, memories are not like data stored on a hard drive: they are constructed and can be edited and changed one time, by all sorts of external and internal factors.

    1. pretzelattack

      she did a lot of good opposing that idiotic “recovered memory” movement in psychology.

      1. Ur-Blintz


        “I’m a very private person and not at all interested in public attention. But, given the incredibly inaccurate and misleading attacks on my father, Woody Allen, I feel that I can no longer stay silent as he continues to be condemned for a crime he did not commit.

        I was present for everything that transpired in our house before, during, and after the alleged event. Now that the public hysteria of earlier this year has died down a little and I have some hope that the truth can get a fair hearing, I want to share my story.”

        1. pretzelattack

          was there a recovered memory issue here? nope. from your own quote, he was never changed his story, so what’s your point?

          1. Plenue

            I think this part is relevant:

            “In her widely-circulated 2014 open letter in The New York Times, the adult Dylan suddenly seemed to remember every moment of the alleged assault, writing, “He told me to lay on my stomach and play with my brother’s electric train set. Then he sexually assaulted me. He talked to me while he did it, whispering that I was a good girl, that this was our secret, promising that we’d go to Paris and I’d be a star in his movies. I remember staring at that toy train, focusing on it as it traveled in its circle around the attic. To this day, I find it difficult to look at toy trains.”

            It’s a precise and compelling narrative, but there’s a major problem: there was no electric train set in that attic. There was, in fact, no way for kids to play up there, even if we had wanted to. It was an unfinished crawl space, under a steeply-angled gabled roof, with exposed nails and floorboards, billows of fiberglass insulation, filled with mousetraps and droppings and stinking of mothballs, and crammed with trunks full of hand-me-down clothes and my mother’s old wardrobes.”

    2. PlutoniumKun

      Capturing the Friedmans is an eyeopening film documentary about the nature of memory and abuse. The Akira Kurosawa film Rashomon is a brilliant fictional version of the same thing, before the phenomenon became widely known (which shows artists are often ahead of scientists). As others have noted with the Woody Allen case, its entirely possible that nobody is lying, despite their being entirely contrary versions of what supposedly happened between him and his daughter.

      The horrifying thing is the number of lives destroyed by it – in the 1980’s there was a wave of convictions of school and nursery workers for abuse because of the satanic abuse panic. Its a multi level problem – it means false convictions and sometimes real criminals going free.

      1. pretzelattack

        yes, based on recovered memories. loftus was a witness for the defense in some of those trials. the kids were coached for months in some cases “don’t you remember x…” and were rewarded for remembering the story their adult interlocutors wanted to hear.

      2. David

        The same thing happened in the UK and Europe. The worrying thing though is that the criminal justice system and members of juries prosecuted and convicted innocent people, even on charges that were actually impossible, because the “victims” were obviously telling the truth, and it was considered wrong to disbelieve them. In fact, different people can have, quite genuinely, different memories of what happened. No-one is “lying.”
        Another interesting case is that of Jean-Claude Menezes, the Brazilian shot dead by police the day after the 2005 London bombings. Eyewitnesses in the same carriage gave completely different accounts of how the (innocent) man behaved, what he was wearing, whether he was carrying a bag, how the police behaved and how many shots they fired.

        1. PlutoniumKun

          I remember well the day after Menenzes was shot reading an interview with someone who was in the train insisting that he had ‘a backpack with wires sticking out’. Of course, we know now he didn’t even have a bag with him.

          1. The Rev Kev

            I seem to recall that the CTV cameras in the carriage could have proven whether the police were lying or not but ‘something’ happened to them. It was murder most foul.

      3. vlade

        I’d change “nobody is lying” to “nobody is knowingly lying” :)

        I thoroughly recommend reading “Mistakes were made (but not by me)”. It’s a bit repetitive at times, but still fascinating.

        A number of people has been put in jail after pleading guilty to murders, which they believed they commited, after some pressure from police who told them repeadetly “you did it, we have proofs, you just don’t want to remember”.

        Worse yet, when some cases were proven later on to be wrong (for example because the real murderer admitted it, with facts that only the murder could know), the cops that put the innocent in jails refused to believe the new evidence.

        1. David

          Yes, excellent book. More scholarly, but absolutely fascinating, is Daniel Kahneman’s “Thinking Fast and Slow” based on his Nobel Prize winning work.
          The real danger is the false dichotomy (“so you’re saying the victim is lying then?”) which completely ignores how the human mind actually functions.

        2. Jimeny Cricket

          Yes. See one of the most excellent documentaries ever made… you probably have…

          ‘A Thin Blue Line’ by Errol Morris

          Goes beyond policing, law and order, to society, to behaviour/nature of the human being himself, herself.

        3. Jimeny Cricket

          Seen one of the most excellent documentaries ever made…

          ‘A Thin Blue Line’ by Errol Morris?

          Goes beyond policing, law and order, to society, to behaviour/nature of the human being himself, herself.

    3. Lee

      I recall a book on the topic, Making Monsters, that debunked the recovered memory methodology and its results. The McMartin pre-school trial, in which Loftus was involved, was one of the most egregious abuses of the phenomenon. https://en.wikipedia.org/wiki/McMartin_preschool_trial

      I recall a conversation about recovered memory with a woman who claimed, quite credibly I believe, to have been abused by her own father. “We never forget,” was her emphatic response. I know, it’s only one instance, but it did confirm what I had come to suspect.

  2. skippy

    One has to be trained to be an observer and even with such there is fields of view, as such a trained observer understand their limitations, albeit can extenuate from experience and intuition.

  3. Eustache De Saint Pierre

    In portrait sculpture the margin for error is about 5%, anything above that is what amounts to at best, looks like. The great French portrait sculptor Jean- Antoinne Houdon who sculpted the likes of Jefferson & Washington, used multiple measurements from life, although in the latter’s case I believe he worked form a death mask.

    From my own experience there is nothing in anatomy anywhere near as complex as the human face & unless you are one of the beautiful people it is largely distortion that gives you your facial character, & is part of the reason why that although within ethnic generalities we have such a massive variation. There is also a huge gulf between two & three dimensions but slight error can cause huge difference in both.

    Portrait sculptors, especially if restricted to photographic reference, usually produce only clever caricatures & tend to prefer subjects with large noses & other highly distinctive features that help with recognisability. There are around a hundred & twenty basic planes to a human face which would give you the aspect of a robot or a doll if merged together – as you go deeper they increasingly multiply in complexity like a coastline as you zoom in from above.

    1. Eustache De Saint Pierre

      Thank you for that.

      I had seen the original which is a lesson in too much distortion, the second is much better but I see some David Beckham in it which might explain the accusations of being too handsome.

      On occasion non-figurative sculptors get commissions based on the success of their name as abstract sculptors. The results are usually awful, but other times those who do figurative work fall foul when it comes to portraits, which are a totally different ballgame.

      This one is a good example of someone getting in over someone else’s head :


  4. Jim A.

    Sketches are nigh worthless. At a fundamental level, language and facial recognition reside in different hemispheres of the brain, which makes describing faces and drawing them from descriptions very difficult.

Comments are closed.