Description
Measures users' emotional states during interaction by combining self-report on valence and arousal with physiological and behavioural signals such as skin conductance and facial expression.
Grounded in Picard's affective computing and Russell's circumplex model, which locates emotions on dimensions of valence (pleasant to unpleasant) and arousal (calm to activated), this method measures emotional responses during interaction. It triangulates three channels: self-report (e.g. Self-Assessment Manikin or moment-by-moment ratings), physiological signals (electrodermal activity, heart rate variability) and expressive behaviour (facial expression, voice). Signals are time-aligned with interaction logs so peaks can be attributed to specific events. Rigour requires baselines per participant, artefact handling, awareness that physiological arousal does not indicate valence alone, validation of automated classifiers against human coding, and careful ethics given the intimacy of emotional data.
Attributes
- Discipline
- Neuro, Biometrics & Affect
- Loss type
- Unmigrated
- Stage
- Evaluation
- Data
- Quantitative
- Effort
- Weeks
Steps
- Multi-channel affect measurement protocol:
- Physiological: EDA (electrodermal activity) = arousal level. HRV = cognitive load and stress.
- (2) Facial: FACs (facial action coding system) or automated tools (Affectiva, Microsoft FER).
- (3) Voice: pitch, rate, and energy features from audio.
- (4) Self-report: SAM (Self-Assessment Manikin — valence and arousal). Triangulate: agreement across channels = high confidence. Disagreement = the most interesting data (which channel is wrong, and why?).
Sources
- Affective Computing — Rosalind Picard
Also known as
- Affective computing measurement
- Emotion measurement in UX
- Valence-arousal measurement
Categories
- Analytics & Behaviour
- Neuro, Biometrics & Affect