5th Edition of International Neurology Conference 2026

Abstract

Electroencephalography (EEG) is one of the major fields of study within affective computing for emotion recognition purposes. EEG allows for the measuring of electrical activity produced by the brain to infer an individual's emotional state. Many existing systems for recognizing emotion using EEG data apply to datasets containing young adults therefore limiting our understanding of how age-related differences in emotional processing and brain activity impact the generalizability, robustness, and ethical implications of these models across different populations. By using information from developmental psychology, we can address most of the problems above by providing a basis for understanding how we create, regulate and express emotion over time and how these processes change throughout the life cycle.

The current research project will investigate how developmental psychology can enhance EEG emotion recognition systems by considering the cognitive, emotional, and neurophysiological characteristics unique to each age group. There are profound differences between developing brains and mature brains that impact the properties of EEG data, as well as how we express our emotions. For example, infants and toddlers tend to produce higher levels of theta brain activity (4 to 7 hertz) with little distinct recognition of their emotional state, while adolescents produce heightened levels of emotional reactivity due to differences between limbic system sensitivity and the prefrontal cortical development. Adult brain structures produce more mature EEG patterns compared to the developing brain and have a greater degree of cognitive/emotional regulation, while older adults experience slower processing speeds in their brains and often express positively compared to younger populations. The differences in how emotions are expressed at different ages significantly influence feature extraction, model training, and prediction performance of emotion classification through EEG.

In order to improve emotion recognition models, the presented framework underscores and provides a rationale for developing age-adaptive and dynamic emotion recognition systems with capabilities to change according to individual developmental differences. It is noted that there are distinct benefits to how Electroencephalogram (EEG)-based systems function by using similar properties (e.g., selection of similar EEG feature types), but for different age groups (e.g., frequency band power and frontal asymmetry). The study also provides perspective on age differences in labelling emotions from children and adults. For children, observational and/or multimodal approaches to labelling are preferred; while for adults, self-report measures are often utilised. In the course of the analysis, bias in emotion detection will be explored and it was determined that comparatively underrepresented populations (children and older adults) create challenges for generalizing the current emotion recognition models (due to lack of adequate representation).

An examination of the use of multimodal signals corresponding to EEG signals (e.g. facial expression, speech) has been conducted in order to improve accuracy of emotion recognition when self-reporting cannot be counted on to reliably represent a person’s emotions. Privacy of data, informed consent and changes from an ethical standpoint (i.e., misinterpretation of a person’s stated emotions) must also be considered due to the vulnerable context of the population samples studied (e.g. children and elderly adults). A developmental perspective will enhance the use of computational models used in the emotion recognition process which will improve the accuracy and inclusivity of emotion recognition systems.

Creating emotion recognition models within the framework of developmental psychology leads to better quality/longer lasting emotion recognition models and creates methods of emotion recognition that are fair, developmentally appropriate, and adaptable in addition to having real-world implications. Expanding this perspective will support the development of affective computing systems that are developmentally diverse and develop a more personalised and contextually relevant understanding of a person's emotional state.