Emotion recognition training using composite faces generalises across identities but not all emotions

Michael Dalili, Lawrence Schofield-Toloza, Marcus Munafo, Ian Penton-Voak

Research output: Contribution to journalArticle (Academic Journal)peer-review

7 Citations (Scopus)
330 Downloads (Pure)


Many cognitive bias modification (CBM) tasks use facial expressions of emotion as stimuli. Some tasks use unique facial stimuli, while others use composite stimuli, given evidence that emotion is encoded prototypically. However, CBM using composite stimuli may be identity- or emotion-specific, and may not generalize to other stimuli. We investigated the generalisability of effects using composite faces in two experiments. Healthy adults in each study were randomised to one of four training conditions: two stimulus-congruent conditions, where same faces were used during all phases of the task, and two stimulus-incongruent conditions, where faces of the opposite sex (experiment 1) or faces depicting another emotion (experiment 2) were used after the modification phase. Our results suggested that training effects generalised across identities. However, our results indicated only partial generalisation across emotions. These findings suggest effects obtained using composite stimuli may extend beyond the stimuli used in the task but remain emotion-specific.
Original languageEnglish
Pages (from-to)858-867
Number of pages10
JournalCognition and Emotion
Issue number5
Early online date12 Apr 2016
Publication statusPublished - 4 Jul 2017

Structured keywords

  • CRICBristol


  • cognitive bias modification
  • emotion recognition
  • Generalisation


Dive into the research topics of 'Emotion recognition training using composite faces generalises across identities but not all emotions'. Together they form a unique fingerprint.

Cite this