What Is Affective Computing?

Affective computing is a field of computer science and engineering concerned with systems that can recognize, interpret, simulate, and respond to human emotions and affect. The term was introduced by Rosalind Picard of MIT in her 1997 book, where she defined it as "computing that relates to, arises from, or deliberately influences emotions." The field proceeds from the observation that emotions are not peripheral to cognition but central to it: human decision-making, attention, and learning are all shaped by affective states, and systems that cannot perceive or respond to affect are limited in the quality of the interaction they can sustain.

Affective computing draws from psychology, cognitive science, human-computer interaction, machine learning, and signal processing. It addresses both the sensing problem, how to detect emotional states from observable signals, and the expression problem, how to generate affective outputs in agents or interfaces that produce appropriate responses. Work in the field spans wearable biosensors, facial analysis algorithms, speech emotion recognizers, and embodied agents capable of displaying affect-like behaviors.

Emotion Recognition

Emotion recognition is the task of inferring a person's affective state from one or more measurable signals. The most widely studied modalities include facial expressions, which follow patterns described by the Facial Action Coding System developed by Ekman and Friesen; vocal characteristics such as pitch, rate, and energy contour; and physiological signals including heart rate variability, galvanic skin response, respiration rate, and electroencephalographic activity. Multimodal recognition systems fuse evidence from several channels simultaneously, generally outperforming unimodal approaches because different emotions produce stronger signals in different modalities. Deep learning methods, particularly convolutional networks for visual input and recurrent or transformer architectures for temporal signals, have driven substantial accuracy gains since 2014. The MIT Press foundational text on Affective Computing by Picard remains an authoritative reference on the theoretical and sensing foundations of the field.

Affective Human-Computer Interaction

Affective human-computer interaction (HCI) applies emotion recognition and synthesis to the design of systems that adapt their behavior based on the user's emotional state. A tutoring system that detects frustration and adjusts problem difficulty, a vehicle that recognizes driver drowsiness and provides an alert, or a social robot that responds to a child's engagement level are all examples of affectively adaptive systems. The practical challenge is that real-world deployment requires low-latency inference, robustness to variations in lighting, background noise, and individual differences in emotional expression, and careful attention to the privacy implications of continuous affective monitoring. A detailed survey of recent advances, challenges, and future trends in affective computing published in Intelligent Computing covers the state of emotion recognition architectures, benchmark datasets, and open problems in the field as of the mid-2020s. The MIT News coverage of Picard's foundational lecture on affective potential documents the original framing of the field and its central argument for why emotional awareness is a prerequisite for genuinely intelligent systems.

Cognitive and Physiological Modeling

Affective computing research intersects with cognitive science through models that treat emotion as an information-processing state rather than only a behavioral output. Appraisal theories, which hold that emotions arise from an individual's evaluation of events relative to their goals, have been implemented in computational agents to generate contextually appropriate affective responses. Physiologically, work on affective computing uses the autonomic nervous system signals that are hard to voluntarily suppress, such as pupillary dilation, galvanic skin response, and cortisol levels, as indicators of stress, arousal, and valence. These physiological channels are particularly useful in applications where facial or vocal signals are unavailable or unreliable.

Applications

Affective computing has applications in a wide range of fields, including:

• Intelligent tutoring and adaptive educational systems
• Driver monitoring and vehicle safety systems
• Mental health monitoring and clinical support tools
• Social robots and companion systems for elderly care
• Marketing and product evaluation through emotional response measurement