CoSuBio: Confidence and success dataset based on multimodal biosignals.

Area of Science:

• Cognitive Neuroscience
• Computational Psychology
• Affective Computing

Background:

• Understanding self-confidence fluctuations during cognitive tasks and their physiological correlates is a research challenge.
• Existing multimodal datasets lack systematic manipulation of self-confidence states.
• The CoSuBio dataset bridges this gap with controlled confidence induction.

Purpose of the Study:

• To introduce the CoSuBio dataset, the first multimodal resource for studying confidence-induced cognitive and behavioral changes.
• To enable research on confidence estimation, performance prediction, and multimodal learning.
• To support applications in adaptive learning, clinical psychology, and human-computer interaction.

Main Methods:

• Collected synchronized electroencephalography (EEG), peripheral physiological signals (EDA, BVP, temperature, accelerometer), and behavioral data from 34 participants.
• Implemented within-subject neutral, confidence-reducing, and confidence-enhancing phases.
• Included self-reported confidence measures and decision-making/cognitive tasks.

Main Results:

• The CoSuBio dataset provides synchronized multimodal biosignals and behavioral data under controlled confidence states.
• It facilitates the investigation of how experimentally manipulated confidence affects cognitive processes and physiology.
• The dataset enables the development of models for confidence estimation and performance prediction.

Conclusions:

• CoSuBio is a unique, open-access dataset for advancing research on self-confidence and its physiological underpinnings.
• It supports novel investigations into confidence dynamics during cognitive activity.
• The dataset has broad implications for fields ranging from affective computing to adaptive learning systems.