Day 3: Trial Goes Down Memory Hole

Retired professor Nancy Franklin transformed a courtroom into a psychology classroom as she weighed the reliability of the eyewitness statements that helped convict Adam Carmon for the 1994 murder of a 7‑month-old baby — and prompted broader discussion about the role of memory in criminal justice.

Franklin, a cognitive science researcher who specialized in memory at Stony Brook University, spent hours on the witness stand of Courtroom 5A at the Church Street courthouse on Wednesday. She was testifying as an expert witness called by lawyers representing Adam Carmon, the man convicted of killing 7‑month-old Danielle Taft and paralyzing Taft’s grandmother in the 1994 shooting.

Carmon is pursuing his fourth habeas petition, a lawsuit against the state seeking to win him a new trial, based on the argument that prosecutors at his original trial withheld key information from defense and that subsequent scientific developments shed new light on evidence at the time.

Carmon’s conviction rested on two eyewitness identifications from Jaime Stanley and Raymond Jones, who testified that they’d seen the shooting while driving past Taft’s apartment building at 810 Orchard St.

Franklin had reviewed case materials before the trial in order to analyze those eyewitness accounts. She explained on Wednesday that her goal as an expert witness was to opine not on credibility, on whether Stanley and Jones had been honest; but on reliability: on whether their memories of the shooting could be trusted given the conditions of that night and the conditions under which they had been asked to recall the event.

Memory, Franklin testified, is both imperfect and malleable over time.

“Memory is not a video camera. We only notice some subset of what’s around us,” she said.

Remembering something doesn’t simply retrieve a mental image from the back of a person’s mind, Franklin said; ​“Memory is reconstructed at recall.” It’s stitched back together with unintentional edits, affected by both time passed and new information that re-contextualizes the event in question. 

“What you start with is a memory with holes, which gets weaker over time,” she said. ​“We are highly prone to filling those holes” with new information.

Faces are particularly hard to remember accurately, Franklin said, simply because there is so much ​“information” that human minds need to store about a face.

“Memory begs to be altered,” she later said. ​“It begs to have the holes filled in.”

In response to questions from an attorney on Carmon’s team, Doug Lieb, Franklin observed that the three seconds for which both Jones and Stanley said they saw the perpetrator’s face offered a ​“very brief view.” One 2003 study found that exposures as long as 12 seconds yielded a 90 percent rate of false identifications, Franklin said.

Franklin noted the lighting conditions at 810 Orchard St. on the night of the murder could have affected Stanley’s and Jones’ perception of the shooter’s face. Not only did the murder occur at around 10:30 at night, in the dark; but a lamppost appears to have been lit in front of the building. Light concentrated in a particular spot in the dark can sometimes lead to poorer perception because of the contrast and shadows it can produce, Franklin testified.

Critically, Franklin testified that studies attest to memory’s weakened state during moments of intense stress, particularly when a weapon is involved. Weapons not only add to a witness’ sense of danger, she said; they tend to attract the witnesses’ attention — and draw that attention away from the perpetrator’s identifying features.

Judge Jon Alander jumped in. 

“According to conventional wisdom, adrenaline kicks in” in stressful situations, and ​“you’re more aware of your surroundings. Is that wrong?” he asked.

The purpose of that adrenaline is to promote survival, Franklin said. ​“The surroundings you become most aware of are escape routes, or places to hide.”

“Wouldn’t you also want to know your attacker?” Alander asked. 

It’s useful to know that an attacker is present, Franklin responded, but in moments of danger, the brain does not tend to ​“spend cognitive resources memorizing the attacker’s eye color … the specific details that make that person different from other people.”

The police interviews leading to suspect identification also affect the reliability of a witness, according to Franklin.

As established by contemporary police records and new testimony from now-retired New Haven Detective Michael Sweeney, investigators showed Stanley a series of 18 photographs as part of their efforts to identify the shooter. Among those 18 photographs were two images of Carmon, neither of which Stanley pointed out. In fact, a note on the back of one of the photographs indicates that Stanley had identified someone else, apparently unconnected to the murder, as a ​“look-alike” of the shooter.

Later, a detective brought Stanley to a session of arraignment court, where 14 men — including Carmon — had been arrested and detained. Carmon’s attorneys were able to obtain photographs of 11 of those men, a group that included individuals of different races and ages, some of whom had facial hair. 

According to a police description, when Carmon saw Stanley, he said ​“Oh, shit,” and put his head in his hands.

In preparation for the trial, Franklin recruited 44 people with no knowledge of the case details to read Stanley’s initial description of the shooter. She then asked each of those 44 people to examine the 11 photographs of the men in arraignment court and to identify the photograph that best fit Stanley’s description. According to Franklin, an ideal lineup or photo array would include faces that equally match the description; in that preferable scenario, her study would have yielded a random distribution of identifications among the 11 men.

Instead, 20 of the 44 participants identified Carmon as the best fit for Stanley’s description, Franklin said. One other individual garnered 19 identifications. The results were ​“astronomically significant,” Franklin said. ​“This is a highly biased set of photos.”

That the first photo array contained two photos of Carmon was ​“bad practice” on its own, Franklin said, since the multiple photographs would have reinforced his image in her memory. That set is ​“highly biased already toward Mr. Carmon, and still he was not identified.”

Later, when Stanley saw Carmon in person, he might have seemed familiar simply because she had seen his photograph twice in a lineup.

In addition, Franklin said, the presence of so many individuals who did not fit Stanley’s description well — and especially a handful of arrestees whose appearances diverged widely from her description — did not only narrow the set of people whom Stanley would have likely been examining. Those outlier arrestees could have made Carmon and the other most likely fit look more like the perpetrator by contrast. Stanley would have been drawn more strongly to Carmon under those circumstances, even if ​“she would not have known why,” Franklin said.

The whole process, Franklin said, was ​“highly suggestive” toward Carmon.

But What's New?

The question at stake Wednesday was whether any new science — unavailable at the 1995 trial — would have meaningfully established the witnesses’ unreliability. 

In her cross-examination of Franklin, Assistant State’s Attorney Lisa D’Angelo noted that approximately 13 of Franklin’s citations in her case report dated back to before 1995.

Judge Alander also pressed Franklin on the issue. Franklin responded that advancements in technology have allowed researchers to better simulate the process of witnessing an event under stress in scientific studies. She noted that findings on the likelihood of false identifications as a specific subset of inaccurate eyewitness interviews had grown. She spoke to a changing ​“consensus” among psychologists about eyewitness reliability. And she said that certain concepts, such as the notion that the presence of a weapon could affect a witness’ perception, developed after 1995.

D’Angelo also questioned Franklin about the notion that Stanley’s view of a gun had affected her memory of events.

She asked Franklin to read from the original trial transcript, returning to a moment in which Carmon’s attorney at the time, Richard Silverstein, questioned Stanley about her narrative of events.

“Weren’t you looking at the gun?” Silverstein asked.

“No, I was looking at his face,” Stanley said. The pair repeated versions of the exchange several times in a row.

Franklin offered a distinction between what Stanley was looking at and what her mind was paying attention to. Stanley could have been staring at the shooter’s face while absorbing fewer details than usual, as her mind processed the threat of the gun, Franklin argued.

As the science on eyewitness identification has evolved, so has state law on the role of experts in assessing eyewitness identifications in court. In Connecticut, experts have been allowed to testify and contextualize eyewitness accounts in perception and memory science only as of a 2012 state Supreme Court ruling on State v. Guilbert.

“Would you have been able to share your testimony with the court at the time of trial?” asked Leib.

“No,” Franklin responded simply.

Click here to read articles published at the time of Danielle’s murder about the circumstances both in the case and the surrounding neighborhood, as well as an interview with the man who owned the gun stolen to commit the killing. Click here for a story about how the family and neighborhood were faring 20 years after the murder, and here for an account of Danielle’s posthumous 21st birthday birthday.

Previous coverage of Adam Carmon habeas trial:

• Day 1: New Trial Sought In Baby’s Murder
• Day 2: Lost Years Flow Into Courtroom 5a