FACS-Lite UX Protocol

Description

Codes a small set of facial action units, such as brow-lowering and lip-corner pull, during usability sessions to locate moments of confusion, frustration and enjoyment.

The Facial Action Coding System decomposes any facial expression into anatomically defined action units (AUs). A full FACS certification takes around 100 hours, so the UX adaptation restricts coding to a handful of diagnostically useful units: AU4 (brow lowerer) for concentration or confusion, AU12 (lip-corner puller) for enjoyment, AU17 (chin raiser) and AU14 (dimpler) for doubt or frustration. Coders review video frame by frame around pre-identified task moments and align codes with screen-recording timestamps. Rigour comes from coding behaviour rather than inferring emotion, from two independent coders with kappa above 0.70, and from treating automated tools such as OpenFace or FaceReader as a first pass that must be validated against human coding rather than a ground truth.

Attributes

Loss type
Siloed
Stage
Evaluation
Data
Mixed
Effort
Weeks

Steps

  1. Core action units for UX: AU4+AU5 = confusion/concentration. AU12 = enjoyment. AU17+AU23 = frustration. AU1+2 = surprise.
  2. Coding protocol: Frame-by-frame review at task difficulty moments. Code AU presence/absence. Intensity A (trace) through E (maximum).
  3. IRR: Two independent FACS coders. Calculate κ. Minimum κ > 0.70 before proceeding.
  4. Map to interface events: Align FACS codes with screen recording timestamps. Interface events preceding frustration expressions = usability problem indicators.
  5. Automated fallback: OpenFace or FaceReader as first-pass tool. Validate automatically on 20% sample.

Sources

  • Unmasking the Face (Ekman & Friesen)
  • Affective Computing (Picard)
  • Designing Across Senses (Noessel)
  • Facial Action Coding System (Ekman & Friesen)
  • The Inquiry Series, Ch 38 — FACS, Micro-Expression Analysis & Psychophysiology

Also known as

  • Facial Action Coding for usability
  • Micro-expression coding
  • Simplified FACS

Categories

  • Analytics & Behaviour
  • Neuro, Biometrics & Affect