A CNN-based cognitive emotion recognition method

As human error remains a primary cause of failures in increasingly complex systems, this study addresses the role of cognitive emotions-which reflect underlying cognitive states-in influencing human performance. To mitigate such errors, we propose a deep learning-based framework for the objective recognition of cognitive emotions from facial expressions. The approach involves three key steps: first, refining traditional emotion classification by introducing a taxonomy of cognitive emotions; second, extracting facial landmarks using a Dlibbased model to capture emotion-relevant features; and third, training a Convolutional Neural Network (CNN) to classify cognitive emotional states. A case study demonstrates that the model can effectively identify cognitive emotions during task execution. By enabling real-time emotional monitoring and operator alerts, this work offers a practical pathway to enhance human reliability and reduce error rates in safety-critical environments. The findings underscore the potential of emotion-aware systems to support human performance in complex operational settings.

Keywords: Human error, cognitive emotion, emotion recognition, affective computing, facial expression, convolutional neural network.