Drawing the Artist: the Advantage of Artistic Ability for the Construction of Holistic Facial-Composite Images.

Abstract

Purpose
Building on Valentine’s (1991) face space theory and evidence suggesting artists possess superior internal-feature encoding (e.g. Devue and Barsics, 2016; Kozbelt et al., 2010), this study aims to investigate whether such an advantage would improve the construction of forensic facial composites.

Design/methodology/approach
Artists and non-artists described a previously-seen unfamiliar face using a cognitive interview and then constructed a composite of it using one of the holistic facial-composite systems, EvoFIT. The effectiveness of the composites was assessed by asking participants who were familiar with the target identities to name the composites; we also asked further participants to rate the composites for likeness. Furthermore, descriptive terms of the face provided by constructors during the cognitive interview were also analysed.

Findings
Artist-generated composites were named significantly more often and rated significantly higher for likeness than non-artist composites, despite artists using significantly fewer descriptive terms, supporting a link between perceptual expertise and enhanced composite performance. Practical implications of the research are discussed.

Research limitations/implications
Although Vogt and Magnussen (2007) proposed that artists scan images more broadly and encode abstract features, this was not directly tested here; nonetheless, the higher likeness scores suggest artists retained more diagnostic visual information. The focus on internal facial features, observed in prior studies (Bruce et al., 1999; Ellis et al., 1979; Kozbet et al., 2010), may account for enhanced recognisability, whereas the ability to flexibly shift between local and global perceptual strategies (Chamberlain and Wagemans, 2015) likely underpins both improved naming and likeness outcomes.

Practical implications
Artists used fewer descriptive units in their H-CIs, yet their composites achieved higher recognition and likeness scores. This indicates that detailed verbal articulation may not reflect the depth or quality of visual encoding; it may also be the case that artists create a more compact encoding, a suggestion that would be worthy of further investigation.

Originality/value
These findings also hold practical utility for law enforcement, underscoring that visual perceptual skill, rather than the volume of verbal descriptors, may be a stronger predictor of composite recognisability. Accordingly, practitioners should avoid assuming that limited verbal detail reflects poor facial recall. Instead, they should acknowledge that witnesses capable of compact yet accurate visual encoding may produce highly effective composites, even when verbal output is minimal. These insights support a more nuanced, evidence-informed approach to witness assessment, composite scheduling and practitioner training within investigative environments.