Constructing Facial Composites: Increasing the Forensic Relevance of Laboratory Research

Lead Research Organisation: University of Leeds
Department Name: Institute of Psychological Sciences

Abstract

Abstracts are not currently available in GtR for all funded research. This is normally because the abstract was not required at the time of proposal submission, but may be because it included sensitive information such as personal details.

Publications

10 25 50

publication icon
Charity Brown (Author) (2012) Applied issues in facial composite construction : the effects of verbal recall, post-encoding delay and intentional and incidental face encoding on the quality of facial composites in Fifth Annual Conference of the International Investigative Interviewing Research Group

publication icon
Charity Brown (Author) (2012) The effects of holistic attribution on the construction of facial-composites in Experimental Psychology Society

publication icon
Charity Brown (Co-Author) Constructing Facial Composites: Increasing the Forensic Relevance of Laboratory Research. in European Association of Psychology and Law

publication icon
Fodarella C (2015) Cross-age effects on forensic face construction. in Frontiers in psychology

publication icon
Frowd CD (2013) Whole-face procedures for recovering facial images from memory. in Science & justice : journal of the Forensic Science Society

 
Title Installation for National Media Museum Exhibition: In Your Face (July to October 2016) 
Description To accompany the In Your Face exhibition we developed a stand-alone installation that allowed participants to try out first hand some of the technology police forces across the country are using to make facial likenesses of criminals. Participants of the installation watched a staged crime and then were taken through the process of evolving a composite of the "perpetrator" from memory using a cut-down version of EvoFIT. EvoFIT is a 'holistic' software system (developed by the co-investigator) that works by creating a composite on the basis of 'breeding' together selections of whole faces. The witness chooses these whole faces from randomly generated arrays on the basis of their similarity to the remembered face. The installation was available within the museum exhibition from July to October along with accompanying background information concerning our on-going project relating to improving both the use of feature-based and holistic-based systems with witnesses. 
Type Of Art Artistic/Creative Exhibition 
Year Produced 2016 
Impact During its life (from July to October) the installation was run all the way through a total of 5,800 times, each time resulting in a finished composite along with a rating given by the visitor as to how alike the composite was to the face of the perpetrator. Across the 5,800 composites, the overall mean rating given by visitors was 2.8, on a scale of 1 (low likeness) to 5 (high likeness). Thus despite working through relatively few cycles of selection and breeding to evolve the composite (compared to the full EvoFIT system used by police forces), visitors seemed to be positive about the composite produced. 
 
Description An eyewitness may produce a visual likeness of an offender. Good quality facial composites prompt recognition by people familiar with the offender. However, research shows that feature-based composite systems, still in extensive use worldwide, generally produce poor likenesses. In standard police practice an eyewitness describes the offender using a Cognitive Interview (CI). This description is then used to select facial features (e.g. narrow noses) from a database and the witness chooses those that best match their memory (e.g. eyes, nose). Describing facial features can interfere with face recognition, perhaps because recognition is best achieved by emphasizing holistic properties of the face. This project examined the effectiveness of two techniques proposed to overcome this problem when using feature-based composite systems. First, following a (face-recall) CI, attributing personality characteristics to the face, thereby focusing on the whole face, can improve composite quality. Our results indicate this Holistic-Cognitive Interview (H-CI) is likely to be effective when the witness reports having attended to the perpetrator during the crime and/or provides a good level of detail about the face. In these circumstances, we found that the H-CI worked when administered 2 days after encoding the target, a delay typical in criminal investigations. Second, inserting a 30 minute delay between CI and composite construction was assumed to allow interference from describing a face to subside, leading to the production of a better facial likeness. However, when interviewed after 2 days, we found worse quality composites were produced with a 30 minute delay between interview and construction, an unexpected finding. We have since found in further research that the detriment of a 30 minute delay between interview and construction to also extend to a holistic-based system. We further illustrate how a novel technique for presenting composites (a perceptual stretch technique that involves showing the facial composite angled to the side) can increase the likelihood that composites are successfully recognised.
Exploitation Route The main objective was to evaluate the effectiveness of two novel techniques for improving the quality of facial composites produced by witnesses using a typical modern feature-based composite system. The benefits of both techniques required replication and their effectiveness under more forensically-relevant conditions had not yet been tested. The findings have contributed to both theory and practice. They have provided important information concerning when the police should use these techniques with witnesses to improve composite quality. Data arising from our project have allowed us to make specific recommendations to relevant practitioners within the police concerning: (1) when use of the Holistic-Cognitive Interview (H-CI) with witnesses is likely to be most effective, and (2) that further benefit from the H-CI can be obtained by circulating the finished composite using a perceptual stretch technique (i.e. showing the facial composite angled to the side). Importantly, we have shown that this technique benefits the recognition of composites created with both feature-based and holistic-based composite systems. Notably, the results indicate that memory strength for the target face is important in determining the success of the Holistic-Cognitive Interview (H-CI) for improving composite quality. If a witness struggles to recall the face then it is likely they will be unable to apply the H-CI as intended and will not benefit from its use. Our work further indicated that inserting a 30 minute delay between the face description (elicited using a CI) and constructing a feature-based composite to be detrimental to composite quality. We have since found this detrimental effect to extend to a holistic-based composite system. Thus describing a face does not necessarily interfere with the ability to produce a composite, and may in fact be useful for composite construction if undertaken immediately before (for both feature and holistic-based systems). Our on-going work is directed at further understanding how these interview techniques work and how to tailor (1) the interview; (2) the composite system; and (3) the presentation format of the final composite to best aid witnesses to produce recognisable composites from both feature and holistic-based systems under varying forensic conditions. For example, during 2016 we have gathered data from the general public (visitors to the National Media Museum, Bradford) on their perceived effectiveness of a cut-down version of a holistic-composite system. Visitors watched a film of a staged crime and by following instructions on the screen were guided through the process of creating a composite of the perpetrator. Overall, 5,800 composites were created and the mean rating given by visitors as to how alike the composite seemed to their memory of the perpetrator was 2.8 (on a scale of 1, low likeness to 5, high likeness). Thus, despite working through relatively few selection and breeding screens, and the absence of any detailed artwork additions to the composite, visitors seemed to be positive about the composite produced. Further work can now investigate whether streamlining the full composite system to make it less time-intensive for police officers to administer can lead to composites that are effectively recognised.
Sectors Digital/Communication/Information Technologies (including Software),Education,Government, Democracy and Justice,Security and Diplomacy

 
Description Principally, facial composites are used in police investigations of serious crime (e.g. rape, burglary). Good quality composites can generate leads when a suspect has not been identified, and there is evidence that composites are used extensively. For example, based on feedback about one system we estimated that across 14 UK police forces 3000 composites have been constructed over 5 years. This project improves our knowledge of how to best create recognisable composites with witnesses and victims, and thus will contribute to the increased identification of offenders and the lowering of crime. Primarily, the effectiveness of two techniques previously identified as enhancing a witness's ability to produce a recognisable composite were evaluated when approximating conditions typical of forensic situations: (1) asking a witness to attribute personality characteristics to the face of the perpetrator they image in memory prior to composite construction (the H-CI); and (2) inserting a 30-minute delay between the witness seeing the perpetrator and producing a composite. The findings highlight the importance of examining the specific characteristics of a witness's statement about their memory: Obtaining information about their self-reported intention to remember the perpetrator, the length of the retention interval and assessing the amount of information they have recalled about the face, can prove useful in deciding whether applying these techniques are likely to improve or impair composite quality. We also showed that changing the way in which composites are viewed by potential recognisers can improve the likelihood that those composites are successfully named. Viewing composites side-on, the perceptual stretch technique, improved recognition of composites under some conditions and appears to be a promising technique to apply in forensic situations.
First Year Of Impact 2013
Sector Government, Democracy and Justice,Security and Diplomacy
Impact Types Societal,Economic

 
Description Contribution to the training of facial composite operators within the police
Geographic Reach National 
Policy Influence Type Influenced training of practitioners or researchers
Impact Our grant based work evaluated the effectiveness of a quick to administer interview technique (the Holistic-Cognitive Interview, H-CI). This technique had previously been shown to improve the identification of composites within the laboratory setting and we set about evaluating this technique under laboratory conditions that were more relevant to the forensic setting. Our findings demonstrated that the H-CI is likely to be of benefit when a witness reports trying to remember the face during the crime, when there is a short delay between seeing the face and composite construction, or when the witness provides a fairly detailed face description. The grant based work predominantly focused on a feature-based composite system (PRO-fit). However, a recent meta-analysis, confirms the effectiveness of the H-CI across both feature (E-fit, PRO-fit and Sketch) and holistic-based composite systems (Evofit) (Frowd et al., 2015, Journal of Forensic Practice). Shortly following the grant end we surveyed police officers in their use of the H-CI following a training workshop including 37 composite operators from 13 forces. Those not using the technique reported that they lacked understanding and confidence in what was required. This suggested to us that hands-on training sessions to promote the H-CI would be useful. Approximately one hundred police personnel across 10 UK forces have now been trained in how to use the H-CI. Training has also been extended to police forces in Romania, Israel and the US (Boston). Feedback remains positive: it is used as a matter of course with witnesses and victims provided that verbal recall of the face is fairly substantive, and operatives think that the witness can suitably understand the two additional memory aids making up the H-CI. We have provided a short guide for police operatives outlining use of the H-CI which is publically available at the following link: http://www.evofit.co.uk/training-support/.
 
Description Practical, evidence-based information for improving how composites are produced and subsequently used to catch criminals.
Geographic Reach National 
Policy Influence Type Influenced training of practitioners or researchers
 
Title The Perceptual Backdrop Image Format for improving the effectiveness of facial composites 
Description Our grant based work showed that asking participants to turn a facial composite to the side, thereby viewing the composite side-on (the perceptual stretch technique), improved the likelihood that composites constructed with a feature-based composite system 2 days after seeing the perpetrator would be successfully recognised. A 2 day delay before composite construction is typical in forensic investigations, and we suggested that applying this technique as standard is likely to be advantageous. Further, we also demonstrated the combined effectiveness of the Holistic-Cognitive Interview technique and perceptual stretch for both a feature-based (PRO-fit) and holistic-based (EvoFIT) system. This benefit of perceptual stretch has now been independently shown to generalise to two alternative composite systems (EFIT-V, a holistic system and E-FIT a feature-based system) by an alternative laboratory (Davis, 2015, Journal of Forensic Practice). However, feedback we gathered from individual operators expressed concern that publicising this technique may invite ridicule and undermine public engagement. Therefore a Perceptual Backdrop Image was created. This presents a composite image against a perceptual backdrop so that it can be presented directly to the viewer, without asking them to angle the composite. It has the same effect on the visual appearance of the image as physically turning the composite to the side. Software has been developed to allow a perceptual backdrop image to be produced as a standard output from EvoFIT and the same effect can easily be achieved using Photoshop for composites generated with other composite systems (e.g., PRO-Fit, EFIT-V, E-FIT). This has proved more acceptable to composite operators. Using a perceptual backdrop also ensures that the perceptual effect we are striving for is maintained (and the image is not altered) when images are published by alternative outlets other than the police (e.g., in newspapers). 
Type Of Technology New/Improved Technique/Technology 
Year Produced 2012 
Impact The Police Service of Northern Ireland have been using the Perceptual Backdrop Image extensively when circulating composites for identification both within and external to their force. However, in spite of the benefits some forces are still expressing the concern that it seems too unusual. We have discussed alternatives with police operatives who think it would be acceptable to include the technique as part of a sequence of representations that should help recognition. These include caricature, Gaussian blue and contrast enhancement, all found in past research to improve recognition rates. We have provided a short guide for police operatives outlining use of the Perceptual Backdrop Image format which is publically available at the following link: http://www.evofit.co.uk/training-support/. 
 
Description BBC Radio Leeds: Wes Butters mid morning programme 9th August 2012 
Form Of Engagement Activity A press release, press conference or response to a media enquiry/interview
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Media (as a channel to the public)
Results and Impact Dr Charity Brown was interviewed live concerning research investigating how to improve the interview techniques the police use when working with witnesses to create facial composites of perpetrators.

This was used as a route to advertise to and recruit research participants to take part in experimental work associated with this research.
Year(s) Of Engagement Activity 2012
 
Description Holistic-Cognitive Interview Factsheet 
Form Of Engagement Activity A magazine, newsletter or online publication
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Policymakers/politicians
Results and Impact A factsheet we created for facial composite operators that provides background to the development of the Holistic-Cognitive Interview and which outlines its procedure.

Composite operators from several police forces are currently using this interview technique when creating facial composites with witnesses.
Year(s) Of Engagement Activity 2013
URL http://tiny.cc/cfrowd
 
Description Identifying the Suspect: Improving Facial Composites 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Type Of Presentation workshop facilitator
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact Witnesses can be asked by the police to help create a facial composite but research shows that they are often a poor likeness for the target person. This workshop brought together specialists from across the UK in facial composite research to discuss:

• Who makes a good witness?

• When is the best time to make a facial composite?

• What is the value of an accidental witness?

• What are the best questions to ask the witness?

• What techniques during facial composite construction lead to a better likeness?

• How should composites be best displayed in the media?

Forty-five practitioners registered for the event, with 37 able to attend. Officers received funding to attend from their forces. In a follow-up survey 19/19 respondents reported finding the workshop useful for their job and 95% would recommend the workshop to a colleague.

Following the workshop a subset of operators reported using the Holistic-Cognitive Interview (H-CI) technique with witnesses. Those not using the technique reported that they lacked understanding and confidence in what was required. This suggests that hands-on training sessions to promote the H-CI would be useful. Feedback from individual operators also informed how best to incorporate techniques for presenting composites within real-world investigations. A concern expressed was that despite evi
Year(s) Of Engagement Activity 2013
 
Description Pop-up activity session, National Media Museum Lates: Faces Event 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Public/other audiences
Results and Impact A pop-up activity session was held at the National Media Museum Lates: Faces Evening Event on 29th October, 2016. This event was for adults (18+) to explore the museum afterhours and 600 hundred visitors attended. We gave a talk concerning recent research developments in facial composite systems and 50 visitors volunteered to take part in a real psychology experiment that further contributed to our aim of helping the police to improve the way in which they construct facial likenesses with witnesses and victims of crime. The activities sparked questions and discussion from the participants (i.e., the members of the general public visiting the museum). Our presence at this event has led to an invitation to contribute a similar activity to an exhibition at the National Railway Museum, York, alongside the British Transport Police.
Year(s) Of Engagement Activity 2016
 
Description Pop-up activity session, National Railway Museum: British Transport Weekend 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Public/other audiences
Results and Impact This event was aimed at families. We provided an opportunity for participants to try out first hand some of the technology police forces across the country are using to make facial likenesses of criminals (using our stand alone software installation). We also exhibited banners detailing the applied nature of our work. These activities led to discussion with members of the general public and British Transport Police attendees about broader principals concerning how the process of face recognition in everyday settings works. We aimed to increase public trust in research by describing how our research is relevant to people's life. As a result of this event 14 visitors attending the event volunteered to take part in our research project.
Year(s) Of Engagement Activity 2017
URL https://www.yorkshire.com/view/events/york/mystery-on-the-rails-2069721
 
Description Presentation to police practitioners 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact This involved a presentation of research findings to a number of police practitioners from across the country at a workshop organised by Dr Karen Lander at the University of Manchester (funded by the Leverhulme Trust). This was the first workshop held for this specific area of policing expertise (composite operators) since the ESRC funded workshop we had previously held in 2013. The presentation sparked discussion concerning the types of interventions in the process of building a composite that practitioners felt would be perceived positively (or negatively) by witnesses. This has led to plans among the academics attending the event to take forward the study of individual differences among witnesses in their ability to construct effective composites.
Year(s) Of Engagement Activity 2017
 
Description Researching eyewitness memory 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Schools
Results and Impact A short presentation given to approximately 60 GCSE and A level Psychology students at a local school.

This ESRC-funded project provided students with a good example of how laboratory research in psychology can be directly applied to real world problems within the criminal justice system and highlights how recommendations can be derived from laboratory research that can inform police working practices. Feedback from the school indicated that the work was positively received by the students.
Year(s) Of Engagement Activity 2012
 
Description The Perceptual Stretch Composite Technique Factsheet 
Form Of Engagement Activity A magazine, newsletter or online publication
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Policymakers/politicians
Results and Impact A factsheet outlining the background to developing the perceptual stretch and perceptual backdrop image techniques and the procedure for using these techniques

A factsheet for facial composite operators
Year(s) Of Engagement Activity 2013
URL http://tiny.cc/cfrowd