Eyewitness Memory: Unreliable or Misunderstood? New Research Challenges Decades of Skepticism

A Nature feature examines the shifting science of eyewitness memory, centering on the case of Charles Don Flores, a Texas death row inmate convicted largely on eyewitness testimony. In 1998, neighbor Jill Barganier initially failed to identify Flores in a photo line-up and described a white man with long hair, but later testified with 'over 100 percent' certainty that Flores was the perpetrator. Researchers like John Wixted of UC San Diego argue that initial high-confidence identifications can be highly accurate, challenging the long-held view that memory is inherently unreliable. Wixted's confidence-accuracy characteristic (CAC) analysis, based on signal detection theory, shows that confident witnesses can be up to 97% accurate under proper conditions. The article traces the history of eyewitness research from Elizabeth Loftus's 1970s experiments to Gary Wells's work on line-up biases, and highlights reforms like double-blind line-ups adopted by some jurisdictions. Wixted now hopes his arguments will aid a Supreme Court petition to reconsider Flores's conviction.