Asking for more information: Can recognition-based enhancements (RBE) made to internal facial features produce more accurate composites?

Abstract

Research focused around investigative interviewing techniques demonstrate that offering witnesses multiple opportunities to retrieve information about a crime can, considerably, improve the accuracy of information recalled from memory. However, this advantageous effect has yet to be explored in the context of composite construction. Taking this gap into consideration, the primary objective of this thesis is to investigate whether providing witnesses with further opportunity to make recognition-based enhancements (RBE) to the internal facial features of a composite, leads to subsequent improvements in composite accuracy. Across four experiments, participants were presented with an unfamiliar target face (for 60 seconds) and then constructed a composite (Experiment 1, 2 and 3: 24 hours later; Experiment 4: 3-4 hours later). Participants were randomly assigned to one of four Accumulated EvoFIT Stages (AEFS), each offering a different level of opportunity to make RBE to the face: AEFS A: involved choosing the “best” face from arrays: AEFS B: involved AEFS A, plus use of Holistic Scales: AEFS C: involved AEFS B, plus use of the Shape Tool: AEFS D: involved AEFS C, plus use of Holistic Scales for internal facial features. Experiment 1, 2 and 3 revealed that composites constructed with fewer opportunities to make RBE to internal facial features, tended to have a stronger likeness towards the target face. However, Experiment 4 revealed a noteworthy finding that, when used in combination with a holistic-cognitive interview (H-CI) and a short retention interval (3-4 hours), composites constructed with a greater opportunity to make RBE produced more accurate composites. The series of experiments suggest that the incorporation of particular interviewing techniques, and the careful optimisation of timing, can substantially enhance the accuracy of facial composites. An implication drawn from these findings is that the most accurate composites were produced when a retention interval of 3-4 hours was adopted, a timeframe that may not always be achievable to replicate in a real-world setting due to police resources.