Anyone who thinks about the law of evidence knows that there are gaps between cognitive psychology and the rules we try cases by. Excitement may distort or inhibit perception, but we permit and rely upon excited utterances as statements valued for their sincerity and truth. The dangers of limiting instructions are clear—they sometimes exacerbate the impact of the problematic proof, yet we teach Rule 105 as a fix to almost all risky evidence and then proceed under the myth that jurors will follow them to the "t."

Yet nowhere may the gap between what science teaches and what the law allows be greater than on the subject of memory. We regularly see lawyers refreshing memory as if it really works and generates accurate recall. Science suggests otherwise.

Rule 612 presumes the validity of using writings—even of third parties—to refresh witness memory. And a brief LEXIS search found no cases—except where hypnosis is involved—where a court concluded that the refreshing process produced an unreliable [read "inadmissible"] memory. Yet this can't be correct.

What do we know from memory science? Detail memory drops significantly within hours of an event; and how questions are asked (if the document was generated in an interview process) can itself create false memories. Consider the classic memory research, the car accident study of Elizabeth Loftus.

What did she test? As described in one summary, seven films of traffic accidents, ranging in duration from five to 30 seconds, were presented in a random order to each group.

After watching the film, participants were asked to describe what had happened as if they were eyewitnesses. They were then asked specific questions, including the question "About how fast were the cars going when they (smashed/collided/bumped/hit/contacted) each other?"

Thus, the Independent variable was the wording of the question and the dependent variable was the speed reported by the participants. The findings were stark—as the word changed, the speed estimate changed as well.

The suggestivity was not limited to analyzing what was actually seen; it extended to what was not seen but what was believed. One week later those who viewed the film were asked questions including, "Did you see any broken glass—yes or no?" There was no broken glass in the filmed event. Those who were questioned with the more potent word—smashed—now had a new memory entirely: What does this mean? When lawyers read/use interviews, they must understand that the interviewer may have had as much to do with the memory as the speaker. And how does this link back to refreshing recollection? It allows a technique that has serious problems:

• The doctrine permits refreshing no matter how many weeks, months or even years after the event the writing was created. So the refreshing document itself may be neither fresh nor accurate.
• No one will know how much of the original was suggested in the way the Loftus experiment demonstrated.
• Ultimately, it is impossible to tell if a memory has actually been refreshed, if the memory testified to in court is the memory of the event or the memory created when the document was prepared, or whether the witness is just acquiescing because "if it was on the paper I'll say it."

There is a second lesson, one apart from the law of evidence. Is teaching a better interviewing technique? The lessons are being applied in law enforcement, where the process known as the cognitive interview produces more detailed and accurate statements from witnesses to crimes and as a result improving police crime solving. See Satin and Fisher, "Investigative Utility of the Cognitive Interview: Describing and Finding Perpetrators," 43 Law & Human Behavior 491(2019).

How does the process work? And can lawyers use it? Both questions were answered 25 years ago, in Wydick, "The Ethics of Witness Coaching," 17 Cardozo L. Rev. 1 (September, 1995). As a process, Wydick explains the four steps of the original cognitive interview process as a tool to be used by lawyers:

• Take the witness back to the scene, establishing context. This means placing the witness at or in the scene by asking the person to visualize and feel it.
• Tell everything. Sorting facts into categories of important, not so important and unimportant can come later—first just let the details pour forth.
• Recall the event in different order (often from the end, back to beginning) or from the occurrence or observation that suck out the most.
• The event is then recalled from the perspective of various participants/observers.

Wydick goes on to detail how the approach was then updated, with the interview process built around stages:

• Introduction, rapport building and making the witness comfortable.
• The open-ended narration stage.
• The probing stage—the repeat of the cognitive interview process, but now turning to each significant event or topic the witness offered during the narration.
• The review stage, one where the interviewed summarizes what has been learned and then sees if the witness recalls or volunteers more.

How successful is the technique? Satin and Fisher report that "the CI has been tested in more than 100 laboratory and field studies in the United States, England, Australia and Germany, and has been shown empirically to enhance witness recall, typically eliciting between 25% and 50% more correct statements than a standard police interview …".

And the takeaway for lawyers and judges? Unless we train on better interviewing techniques, we will neither recognize which interviews have value for determining what occurred or for refreshing recall, nor ourselves elicit more accurate and necessary facts when we conduct interviews ourselves.

Jules Epstein is professor of law and director of advocacy programs at Temple University Beasley School of Law.