image: The schematic connects emotion processing and affective computing through brain signals, showing how neuroscience can guide computational models and how quantitative predictions can inform theory and applications.
Credit: Wang et al., SCIENCE CHINA Life Sciences (2026). © The Author(s) 2026.
How do we perceive emotion in a face or a smell? How does experience give events emotional significance and shape what we remember? Can emotion be identified from brain activity? A review in SCIENCE CHINA Life Sciences examines these questions across emotion perception, emotional learning and memory, and brain-based affective computing.
The review begins with major accounts of emotion, including basic emotion theories, dimensional models and constructionist approaches. It then examines visual and olfactory processing. Research on vision addresses facial expressions, dynamic social signals and neural pathways involved in threat processing. Research on smell explores pleasantness, arousal and the close links between odors, emotion and memory. Together, these findings illustrate the diversity of emotion perception and ongoing debates about how it is organized in the brain.
The next section considers how emotional significance is learned and how emotion influences memory. It describes how initially neutral cues become associated with threat or reward, and how emotional experiences affect the retention and retrieval of information. The review discusses the roles of the amygdala, hippocampus and prefrontal cortex, alongside consolidation, reactivation and updating, to explain how emotional memories form, persist and change with subsequent experience.
The final section surveys affective computing based on brain signals, particularly electroencephalography (EEG). It introduces signal features and machine learning and deep learning methods used to identify emotion. It also considers challenges involving emotion labels, individual differences and generalization across contexts. Distinguishing the emotional properties of a stimulus from a person’s subjective experience is essential for interpreting what a model predicts.
By bringing psychological theories, neural mechanisms and computational methods together, the review offers a broader account of emotion and its study. It discusses how naturalistic research, computational modeling and neuromodulation could inform one another and support future applications in mental health and human–machine interaction.
Journal
Science China Life Sciences
Method of Research
Literature review