Description
Defines the set of affective states that matter in a product context, with dimensional coordinates and observable indicators, so emotion can be measured consistently across channels.
Emotion measurement fails when teams do not define what they are measuring. Basic emotion categories such as anger, fear and joy rarely occur during product use; states such as confusion, frustration, boredom, engagement, anxiety and relief are far more frequent, as D'Mello and Graesser showed in learning technologies. An affective state taxonomy specifies the states relevant to the domain, locates each on dimensions such as valence and arousal (and sometimes dominance), and defines the indicators expected in each channel: self-report wording, facial action units, physiological markers and behavioural signals. It also defines which states are desirable, which are acceptable and which are failure signals. Rigour comes from grounding the taxonomy in observation of the domain, operational definitions, and checking inter-rater agreement when coding states.
Attributes
- Discipline
- Neuro, Biometrics & Affect
- Loss type
- Siloed
- Stage
- Definition, Discovery
- Data
- Mixed
- Effort
- Days
Steps
- Collect candidate states from domain observation, interviews and literature, aiming for 6 to 12 states relevant to the product (for example confusion, frustration, anxiety, relief, confidence, boredom).
- Define each state operationally, and place it on valence and arousal dimensions (and dominance if control matters, as in financial decisions).
- For each state, specify expected indicators per channel: self-report items, facial cues, physiological signals such as heart rate variability or electrodermal activity, and behaviours such as hesitation or rage clicks.
- Classify states as target, acceptable or failure states for each journey stage, for example anxiety acceptable at loan submission but a failure state after approval.
- Pilot the taxonomy by having 2 coders label 30 to 50 video segments, aiming for Cohen's kappa of 0.60 or above, and refine definitions.
- Use the taxonomy to structure multi-channel affective studies and to report results consistently over time.
Sources
- Affective Computing (Picard)
- Methods Atlas — affective measurement (Gopinath)
- The Inquiry Series, Ch 37 — Picard's Affective Framework — Multi-Channel Emotion Measurement
Also known as
- Affect taxonomy
- Emotion coding scheme
- Target affective states model
Categories
- Analytics & Behaviour
- Neuro, Biometrics & Affect