Eyewitnesses are misled by human but not robot interviewers

Abstract

This paper presents research results from a study to determine whether eyewitness memory was impacted by a human interviewer versus a robot interviewer when presented misleading post-event information. The study was conducted with 101 participants who viewed a slideshow depicting the events of a crime. All of the participants interacted with the humanoid robot, NAO, by playing a trivia game. Participants were then interviewed by either a human or a robot interviewer that presented either control or misleading information about the events depicted in the slideshow. This was followed by another filler interval task of trivia with the robot. Following the interview and robot interactions, participants completed a paperpencil post-event memory test to determine their recall of the events of the slideshow. The results indicated that eyewitnesses were misled by a human interviewer (t(46) = 2.79, p <; 0.01, d = 0.83) but not by a robot interviewer (t(46) = 0.34, p > 0.05). The results of this research could have strong implications for the gathering of sensitive information from an eyewitness about the events of a crime.