Glossary // Glossary term Technical
Multimodality
The combination of several kinds of data, such as self-report, facial movement, voice, text and physiology, in one analysis. Combining modalities can reduce reliance on any single ambiguous channel, but it does not remove the underlying problem that each channel relates to emotional states only probabilistically and context-dependently. Multimodal systems also collect more personal data, which increases the importance of consent and data minimisation. Affective computing promises to infer emotion as if it were a system metric. The evidence on facial movements and autonomic signals, together with questions of consent, explains why every metric in this artifact is an artistic value. How one could reason about recording a catharsis-like change of state: signals, features, inference and context, baselines, self-report, error costs and ethics, and why every step stays uncertain. Nothing on this site measures anyone. The works the definition rests on, with the level of evidence each was appraised at. (1997). Affective computing. Supports the historical claim that affective computing was framed as a research field in 1997. It is not evidence that emotion can be validly inferred from signals. (2021). The ethics of emotion in artificial intelligence systems. doi:10.1145/3442188.3445939 Supports the claim that emotion-AI systems depend on contested models of emotion and proxy data, which shapes their ethical implications. A conceptual argument, not an empirical audit of any product.Where the site uses it
Studies behind it