While emotions have proven to play a crucial role in decision-making process, the influence of emotions is often overlooked in current decision prediction models. To compensate for the lack of emotional integration, this paper proposes a single trial decision prediction model, MeCM-EDNet, an emotional decision network (EDNet) that incorporates meiosis data augmentation (Me), contrastive learning (C), and multi-task learning (M). It consists of four blocks: data augmentation block, time learning block, graph learning block, and multi-task learning block. Among them, the multi-task learning block integrates decision prediction, emotion recognition, and supervised contrast learning tasks, effectively achieving the emotional integration of decision model. Moreover, we embed the neural mechanisms of emotion-driven decision-making into our graph learning module to overcome the limitations of decision models lacking neuroscientific knowledge. To validate our model, the paper designs an experiment to investigate how emotions influence spatial decision making and constructs a dataset. Comparative experiments in our dataset and Emoback dataset demonstrate that MeCM EDNet achieves state-of-the-art performance. The ablation studies confirm the effectiveness of each block of our model, particularly highlighting the important role of the integration of emotional information and neuroscience knowledge. MeCM-EDNet underscores the importance of considering emotional influence in decision research, addressing the current lack of emotional integration in most decision prediction models. The code of our paper is freely available at https://github.com/qimingzitainanla/MECM.

Existing deep learning models for electroencephalogram (EEG) are typically tailored for specific tasks, datasets, or even subjects. This specialization restricts their applicability, reducing both perceptual capabilities and overall generalizability. Recent research has begun to explore large-scale pretraining for EEG representation learning. These efforts often focus on enabling data from different domains to be encoded by the same model through unifying the data format. However, this approach tends to overlook the importance of decoupling EEG representations across various domains, leading to the missing of valuable domain-specific details. Given the diverse distribution of large-scale EEG data forming multiple distinct domains, we hope to learn not only domain-shared representations but also to learn domain-specific representations. Therefore, in this article, we propose EEG mixture of experts (EEGMoE), a domain-decoupled self-supervised pretraining model for EEG representation learning. Specifically, EEGMoE introduces a Transformer-based domain-decoupled encoder, featuring specific and shared expert groups within our mixture-of-experts (MoE) block. The specific expert group adopts Top- $K$ routing to select the most appropriate $K$ experts for each token, while the shared expert group employs soft routing, leveraging all experts to learn for each token. Pretrained on various EEG datasets from multiple tasks, EEGMoE is then fine-tuned and validated on new datasets covering three mainstream EEG tasks, including emotion recognition (ER), motor imagery (MI) classification, and mental workload detection. Experimental results show that EEGMoE outperforms state-of-the-art models on three public datasets, demonstrating its strong generalization ability to new domains. Extensive experiments and visualizations further highlight the importance and effectiveness of disentangling domain-specific representations. Our code and model will be released.

Recent advances in Large Language Models (LLMs) have shown potential in tasks like Theory of Mind (ToM) and social reasoning, but their application in sequential multi-party multi-turn (MPMT) dialogues is limited, particularly in detecting cognitive biases that potentially impact group decision-making quality, which we term Non-Performing Mind Detection (NPMD). Therefore, to better validate the capability of LLMs in the area, we propose NPMTiny, a novel benchmark comprising 77 annotated Chinese meeting transcripts designed to evaluate LLMs’ ability to detect cognitive biases that potentially impact decision quality in group decision-making processes within MPMT dialogue environments. Our experiments reveal that although LLMs excel in traditional tasks, they fall short in the NPMD task. Qwen2.5-7B-Instruct fine-tuned with NPMTiny significantly improves detection capabilities, achieving performance comparable to leading commercial models. This study underscores the potential of LLMs in mitigating cognitive biases and displays a possible new paradigm to computer-aided group decision-making for humans. Resource is available at: https://github.com/HWME-funder/NPMTiny.

In order to cultivate children’s computational thinking, researchers have developed many excellent programming systems. Among them, the tangible programming systems combined with graphic output have been widely accepted because of the intuitive input method and diversified visual feedback. However, few of them support collaborative learning or programming. Little is known about children’s experiences, emotional states and behaviours on collaborative programming. Motivated by this gap, we present a novel tangible and collaborative enabled programming system named Lighters, which designed for children aged 7–10 and had two types of collaboration modes (block-based and role-based). We conducted an user experiment with 24 children, and collected physiological (EDA), questionnaire, video recording and interview data for analysis. Based on our experiment results, Lighters are effective in helping children learn to program collaboratively. In addition, Lighters can mobilize children’s positive emotions and enthusiasm to learn programming. Compared with block-based collaboration, role-based collaboration is more likely to stimulate children’s emotional states and has a better effect on learning programming.

Emotions are important factors in decision-making. With the advent of brain-computer interface (BCI) techniques, researchers developed a strong interest in predicting decisions based on emotions, which is a challenging task. To predict decision-making performance using emotion, we have proposed the Maximizing Mutual Information between Emotion and Decision relevant features (MMI-ED) method, with three modules: (1) Temporal-spatial encoding module captures spatial correlation and temporal dependence from electroencephalogram (EEG) signals; (2) Relevant feature decomposition module extracts emotion-relevant features and decision-relevant features; (3) Relevant feature fusion module maximizes the mutual information to incorporate useful emotion-related feature information during the decision-making prediction process. To construct a dataset that uses emotions to predict decision-making performance, we designed an experiment involving emotion elicitation and decision-making tasks and collected EEG, behavioral, and subjective data. We performed a comparison of our model with several emotion recognition and motion imagery models using our dataset. The results demonstrate that our model achieved state-of-the-art performance, achieving a classification accuracy of 92.96%. This accuracy is 6.83% higher than the best-performing model. Furthermore, we conducted an ablation study to demonstrate the validity of each module and provided explanations for the brain regions associated with the relevant features.

Integrating music into Computer Science (CS) education can stimulate children’s creativity, change the stereotypical perspective of CS, and encourage women, ethnic or cultural minorities involved in the Computer Science area. In this paper, we use Augmented Reality (AR) technology to design a tangible programming system - AR-MPro for children, acting as a bridge between programming and music. It allows children to create customized AR effects to visualize music with low-cost materials by constructing tangible program sequences. AR-MPro is expected to broaden participation in computing, and be more intuitive, intriguing and instructional to enrich children’s creating and programming experiences.

Due to the prevalence of group activities on social networks, group recommendations have received an increasing number of attentions. Most group recommendation methods concentrated on dealing with persistent groups, while little attention has paid to ephemeral groups. Ephemeral groups are formed ad-hoc for one-time activities, and therefore they suffer severely from data sparsity and cold-start problems. To deal with such problems, we propose a pre-training and fine-tuning method called GBERT for improved group recommendations, which employs BERT to enhance the expressivity and capture group-specific preferences of members. In the pre-training stage, GBERT employs three pre-training tasks to alleviate data sparsity and cold-start problem, and learn better user representations. In the fine-tuning stage, an influence-based regulation objective is designed to regulate user and group representations by allocating weights according to each member's influence. Extensive experiments on three public datasets demonstrate its superiority over the state-of-the-art methods for ephemeral group recommendations.

Stereoscopic visual fatigue (SVF) due to prolonged immersion in the virtual environment can lead to negative user experience, thus hindering the development of virtual reality (VR) industry. Previous studies have focused on investigating the evaluation indicators associated with SVF, while few studies have been conducted to reveal the underlying neural mechanism, especially in VR applications. In this paper, a modified Go/NoGo paradigm was adopted to induce SVF in VR environment with Go trials for maintaining participants' attention and NoGo trials for investigating the neural effects under SVF. Random dot stereograms (RDSs) with 11 disparities were presented to evoke the depth-related visual evoked potentials (DVEPs) during 64-channel EEG recordings. EEG datasets collected from 15 participants in NoGo trials were selected to conduct individual processing and group analysis, in which the characteristics of the DVEPs components for various fatigue degrees were compared and independent components were clustered to explore the original cortex areas related to SVF. Point-by-point permutation statistics revealed that DVEPs sample points from 230 ms to 280 ms (component P2) in most brain areas changed significantly when SVF increased. Additionally, independent component analysis (ICA) identified that component P2 which originated from posterior cingulate cortex and precuneus, was associated statistically with SVF. We believe that SVF is rather a conscious status concerning the changes of self-awareness or self-location awareness than the performance reduction of retinal image processing. Moreover, we suggest that indicators representing higher conscious state may be a better indicator for SVF evaluation in VR environments.

In this paper we present ARCat, a tangible programming tool designed to help children learn Depth First Search (DFS) algorithm with augmented reality (AR) technology. With this tool, children could use tangible programming cards to control a search process, rather than control virtual characters directly. With the special design of card semantics and real-time feedback, the cognitive load of the learning process had been proved to be affordable to children (ages 8-9) with the result of our preliminary evaluation, which shows the possibility of basic algorithm education for young children with tangible interface.

In the past decade, Electroencephalogram (EEG) has been applied in many fields, such as Motor Imagery (MI) and Emotion Recognition. Traditionally, for classification tasks based on EEG, researchers would extract features from raw signals manually which is often time consuming and requires adequate domain knowledge. Besides that, features manually extracted and selected may not generalize well due to the limitation of human. Convolutional Neural Networks (CNNs) plays an important role in the wave of deep learning and achieve amazing results in many areas. One of the most attractive features of deep learning for EEG-based tasks is the end-to-end learning. Features are learned from raw signals automatically and the feature extractor and classifier are optimized simultaneously. There are some researchers applying deep learning methods to EEG analysis and achieving promising performances. However, supervised deep learning methods often require large-scale annotated dataset, which is almost impossible to acquire in EEG-based tasks. This problem limits the further improvements of deep learning models for classification based on EEG. In this paper, we propose a novel deep learning method DMTL-BCI based on Multi-Task Learning framework for EEG-based classification tasks. The proposed model consists of three modules, the representation module, the reconstruction module and the classification module. Our model is proposed to improve the classification performance with limited EEG data. Experimental results on benchmark dataset, BCI Competition IV dataset 2a, show that our proposed method outperforms the state-of-the-art method by 3.0%, which demonstrates the effectiveness of our model.

Programming is an effective way to foster children's computational thinking. We present AR-Maze, which is a novel tangible programming tool using Augmented Reality (AR) technology for young children. AR-Maze superposes constant feedback on the physical world and maintains a positive, low-cost learning environment. Using this system, children could create their own programs by arranging programming blocks and debug or execute the code with a mobile device. In addition, they will be able to learn fundamental programming concepts, such as parameters, loop logic, debug, etc. We design and implement this system, as well as conduct a preliminary user study and analyze the results, which can guide a better design of AR-Maze. With this work, we intend to help children programming in an interesting and intuitive way.

Optical tracking has become the most commonly used virtual reality (VR) tracking technology because of its high precision and non-contact characteristics. The optical tracking system represented by HTC VIVE has the problem that signals of base stations interfere with each other, and the number of base stations cannot be extended by cascades, thereby limiting the scope of its work. In this paper, an extensible optical tracking system is proposed, which can distinguish the signals from different base stations and support the simultaneous operation of multiple base stations. Furthermore, we designed an encoding scheme to generate independent code for up to 32 base stations and proposed a highspeed decoding method. Experiments demonstrate that the system has high tracking accuracy and low system latency. Users can adjust the number and layout of base stations according to the actual demand, which greatly improves the flexibility of the system, and benefits promoting the development of large scale optical tracking equipment with low cost and high precision.

Learning programming has positive effect on children's development, and Tangible User Interfaces (TUIs) is a convenient way for teaching young children programming. TanProRobot 2.0 is a tangible system as well as a small-scale distributed embedded system designed for children at grades 1-2 to learn programming concepts. The system consists of three parts: tangible programming blocks, a robot car and several manipulatives. The input and output of the system are both tangible. Children can program the robot car to act certain actions by arranging the programming blocks. Also, children can interact with the car with manipulatives. TanProRobot 2.0 aims to introduce event handling concept and sensors to children. Through a user study with 11 children, we found that TanProRobot 2.0 is an interesting programming system for children, and it is easy to learn and to use. Furthermore, it could help children get a preliminary understanding of event handling concepts.

Tangible User Interfaces (TUIs) can create opportunities to learn programming for children, which have positive effect on children's development. TanProRobot is a tangible system designed for children at grade 1-2 to learn programming concepts. It consists of three parts, tangible programming blocks, the robot car and the manipulatives. The input and output of the system are both tangible. Children can program the robot car to act certain actions by arranging programming blocks. Also, children can interact with the car with manipulatives. TanProRobot aims to introduce event handling concept and sensors to children. During the game, children can also learn some traffic rules, which are import to self-security.

TanPro-Kit 2.0 is a programming tool for children aged 6 to 8, which is based on tangible user interface (TUI). It consists of programming blocks and a LED pad. The pad presents visual and audible feedback according to the arrangement of programming blocks with which children construct programs to play a maze game. Based on TanPro-Kit, we expanded three important programming concepts: parameters, Boolean logic and branch, and improved the system to support two-dimensional connection, which aims to make the program structure clearer. To realize these new features, we added three kinds of programming blocks accordingly, improved the process of block sequence in Arduino, and modified the infrared and wireless communications of the Single-Chip Microcomputer (SCM). A lab-based user study with 15 children was conducted, and the results show that the children can use the system to complete tasks easily and have a basic understanding of the related programming concepts.

This paper describes a new tangible programming tool--- TanPro-Kit which was designed for children aged 5 to 9. It consists of programming blocks and a LED pad. The LED pad presents visual animations and audible feedback according to the arrangement of blocks with which children make program to play a maze game. Aiming at lowering the cost of TanPro-Kit, we adopted LED, RFID, wireless and infrared technology to develop the whole system. The system acquires the programming blocks' physical information which is then translated into programming semantic. TanPro-Kit is low-cost, which is more acceptable in developing countries. We ran a user study with 16 children involved, which showed TanPro-Kit to be attractive to children and easy to learn and use.

In this paper, we introduce the design and implementation of TempoString, an easy-to-use tool which assists children with music creation. It provides such a fun and novel platform by allowing children to "draw" music on a canvas and then edit it using a rope. The main contribution of our work is the novel access which allows children to "paint" music on a canvas and then edit using a rope.

This paper presents a tangible programming tool 'T-Maze' for children aged 5 to 9. Children could use T-Maze to create their own maze maps and complete some maze escaping tasks by the tangible programming blocks and sensors. T-Maze uses a camera to, in real-time, catch the programming sequence of the wooden blocks' arrangement, which will be used to analyze the semantic correctness and enable the children to receive feedbacks immediately. And children could join in the game by controlling the sensors during program's running. A user study shows that T-Maze is an interesting programming approach for children and easy to learn and use.

In this paper, we describe CoolMag, a tangible interaction tool to enable children to create different instruments collaboratively in music education. With CoolMag, children could learn the basic playing methods of different instruments. It also has the potential to inspire children's creativity, because children could adopt objects in daily life (broom, cup, pen etc.) as the carrier of their novel instruments whose appearance may differ from the traditional one.