Eyewitness testimony research, pioneered by Elizabeth Loftus, has demonstrated that eyewitness memory is far less reliable than legal systems and jurors typically assume. Memory for witnessed events is not a video recording that can be played back; it is a constructive process influenced by encoding conditions, post-event information, retrieval procedures, and social factors. This research has had profound impact on the criminal justice system.
Factors Affecting Accuracy
Encoding factors: stress impairs memory for peripheral details while sometimes enhancing memory for central details; brief exposure limits encoding; poor lighting and distance reduce encoding quality; weapon focus narrows attention to the weapon at the expense of the perpetrator's face. Storage factors: the misinformation effect shows that post-event information (from leading questions, media reports, co-witness discussion) can alter memory for the original event. Retention interval allows forgetting and increases susceptibility to misinformation. Retrieval factors: suggestive identification procedures, interviewer behavior, and repeated questioning can shape reported memories.
Loftus's landmark misinformation studies showed that exposure to misleading post-event information can systematically alter eyewitness reports. When asked "How fast were the cars going when they smashed into each other?" (versus "hit"), witnesses reported higher speeds and were more likely to report (non-existent) broken glass a week later. This research established that memory is malleable and that seemingly innocuous questioning can alter recollection, with profound implications for police interviewing practices.
Improving Accuracy
The cognitive interview, developed by Fisher and Geiselman, applies memory research to improve witness recall: mentally reinstating the context of the event, reporting everything (even seemingly trivial details), recalling events in different temporal orders, and recalling from different perspectives. Research shows the cognitive interview increases correct information by 25-35% without increasing errors. For identification, best practices include double-blind administration, unbiased instructions, sequential presentation, and immediate confidence assessment.