Avatars in child investigative interviews

: supporting disclosure and eyewitness memory through technology

Abstract

This thesis explored the use of human-operated avatars and how this affects the quality and quantity of child forensic disclosure, episodic recall, and susceptibility to false information during investigative interviews. The research also captured the concerns and recommendations from child forensics interviewers for applying avatar technologies in practice.

Recent legislative changes under the Scottish Child Interview Model have ushered in trauma-informed ‘safe spaces’ for children to be interviewed when they experience crime or abuse. The incorporation of technologies, such as the pre-recording of evidence, is intended to make the disclosure process more accessible for vulnerable witnesses, preventing the need for children to attend court, and avoid drawn-out trial proceedings. Remote, online interviewing has also been introduced in some courtroom-based contexts to reduce witness anxiety. In addition, virtual characters, e.g., avatars, are becoming increasingly common across many professional fields, including healthcare, counselling, and education. However, the use of avatars as investigative interviewers specifically for children and adolescents is comparatively under-researched.

This thesis investigated whether avatars might facilitate or hinder adolescents’ disclosure and memory recall, similar to other computer-mediated interactions and online disclosure methods. This research used a mixed-methods approach, comprising a survey study, two experimental studies, and a qualitative study.

Study 1, an online survey, collected information on children and adolescents’ preferred forensic disclosure recipients based on crime type and preferred disclosure recipient characteristics. The recipient characteristics included age, gender, and whether the disclosure interaction was online, face to face, with a human virtual character, or a non-human virtual character. The purpose of this study was to identify fruitful directions in designing avatar interviewers, based on preferred disclosure recipient traits. These findings confirmed specific preferences, such as disclosure recipients being predominantly adults, recipients being the same sex as children, and that human virtual characters were preferable to non-human virtual characters. Furthermore, these findings indicate that girls felt less positive about disclosing crimes, which is indicative of potential sex differences depending on the context of disclosure.

Study 2 used a mock investigative interview paradigm and compared human interviewers appearing via video chat with human-appearing avatar interviewers. This study also manipulated whether avatars and human interviewers were of the same or opposite sex as participants. Study 2 was split into two parts. Part 1 of this study investigated the impact of interview condition (avatar, human) and dyad (where interviewers were the same sex as the interviewee, or opposite sex as the interviewee) on the disclosure of positive and negative life experiences. The data revealed no significant differences. However, girls did report significantly lower fears of negative evaluation during their disclosure when being interviewed by an avatar, irrespective of the avatar/interviewer’s sex. Part 2 of this study investigated the impact of the interviewer format on the episodic recall and suggestibility of false information after watching a mock crime video. No significant differences emerged between humans and human-appearing avatars, or between same-sex and opposite-sex dyads.

Study 3 also used a mock investigative interview paradigm and included the same dependent variables as Study 2 but did not explore same-sex/opposite sex dyads. Due to the lack of significant findings between human and human-appearing avatars in Study 2, the purpose of Study 3 was to expand these findings by including a non-anthropomorphic avatar (faceless, abstract shape) for comparison alongside a face to face human interviewer and a human-appearing avatar. This study was also split into two parts. Part 1 of this study focused on disclosure quantity and quality and found no significant differences between interviewer conditions. Part 2 of this study focused on episodic recall and suggestibility. Similar to Study 2, Study 3 found no significant differences between interview conditions.

Study 4 captured child forensic interviewers’ perspectives on the benefits and challenges of using avatar technologies with children in practice. Qualitative findings generated mixed opinions. Interviewers appreciated children's familiarity with using such technologies and the potential rapport-building opportunities avatar technologies may foster. However, interviewers expressed concern that avatars would obscure non-verbal cues and create an emotional disconnect with the children they interviewed. Interviewers also anticipated resistance from the courts in accepting avatars as an interview tool. This was attributed to concerns over potential misleading or coercive influences.

Taken together, these findings suggest that human-controlled avatars can be as effective as human interviewers in eliciting disclosure and that complete and accurate memory accounts can be obtained from adolescent participants when appropriate interview questions are used. Despite this, insights from child forensic practitioners highlight several practical and organisational challenges of using avatar technologies as a means of gathering eyewitness testimony.

Date of Award	3 Nov 2025
Original language	English
Awarding Institution	Abertay University
Sponsors	Northwood Charitable Trust
Supervisor	Julie Gawrylowicz (Supervisor), Robin J. S. Sloan (Supervisor) & Andrea Szymkowiak (Supervisor)