Objective: Confusion is the primary epistemic emotion in the learning process, influencing students' engagement and whether they become frustrated or bored. However, research on confusion in learning is still in its early stages, and there is a need to better understand how to recognize it and what EEG signals indicate its occurrence. The present work investigates confusion during reasoning learning using EEG, and aims to fill this gap with a multidisciplinary approach combining educational psychology, neuroscience and computer science. 
Approach: First, we design an experiment to actively and accurately induce confusion in reasoning. Second, we propose a subjective and objective joint labeling technique to address the label noise issue. Third, to confirm that the confused state can be distinguished from the non-confused state, we compare and analyze the mean band power of confused and unconfused states across five typical bands. Finally, we present an EEG database for confusion analysis, together with benchmark results from conventional (Naive Bayes, SVM, Random Forest, and ANN) and end-to-end (LSTM, ResNet, and EEGNet) machine learning methods.
Main results: Findings revealed: 1. Significant differences in the power of delta, theta, alpha, beta and lower gamma between confused and non-confused conditions; 2. A higher attentional and cognitive load when participants were confused; and 3. The Random Forest algorithm with time-domain features achieved a high accuracy/F1 score (88.06%/0.88 for the subject-dependent approach and 84.43%/0.84 for the subject-independent approach) in the binary classification of the confused and non-confused states.
Significance: The study advances our understanding of confusion and provides practical insights for recognizing and analyzing it in the learning process. It extends existing theories on the differences between confused and non-confused states during learning and contributes to the cognitive-affective model. The research enables researchers, educators, and practitioners to monitor confusion, develop adaptive systems, and test recognition approaches.

Objective. Due to individual differences in EEG signals, the learning model built by the subject-dependent technique from one person's data would be inaccurate when applied to another person for emotion recognition. Thus, the subject-dependent approach for emotion recognition may result in poor generalization performance when compared to the subject-independent approach. Existing studies, However, have attempted but not fully utilized EEG's topology, nor have they solved the problem caused by the difference in data distribution between the source and target domains. Approach. To eliminate individual differences in EEG signals, this paper proposes the domain adversarial graph attention model (DAGAM), a novel EEG-based emotion recognition model. The basic idea is to generate a graph using biological topology to model multichannel EEG signals. Graph theory can topologically describe and analyze EEG channel relationships and mutual dependencies. Then, unlike other graph convolutional networks, self-attention pooling is used to benefit the extraction of salient EEG features from the graph, effectively improving performance. Finally, following graph pooling, the domain adversarial based on the graph is used to identify and handle EEG variation across subjects, achieving good generalizability efficiently. Main Results. We conduct extensive evaluations on two benchmark datasets (SEED and SEED IV) and obtain cutting-edge results in subject-independent emotion recognition. Our model boosts the SEED accuracy to 92.59% (4.06% improvement) with the lowest standard deviation of 3.21% (2.46% decrements) and SEED IV accuracy to 80.74% (6.90% improvement) with the lowest standard deviation of 4.14% (3.88% decrements) respectively. The computational complexity is drastically reduced in comparison to similar efforts (33 times lower). Significance. We have developed a model that significantly reduces the computation time while maintaining accuracy, making EEG-based emotion decoding more practical and generalizable.