Recently, convolutional neural network (CNN)-based classification models have shown good performance for motor imagery (MI) brain-computer interfaces (BCI) using electroencephalogram (EEG) in end-to-end learning. Although a few explainable artificial intelligence (XAI) techniques have been developed, it is still challenging to interpret the CNN models for EEG-based BCI classification effectively. In this research, we propose 3D-EEGNet as a 3D CNN model to improve both the explainability and performance of MI EEG classification. The proposed approach exhibited better performances on two MI EEG datasets than the existing EEGNet, which uses a 2D input shape. The MI classification accuracies are improved around 1.8% and 6.1% point in average on the datasets, respectively. The permutation-based XAI method is first applied for the reliable explanation of the 3D-EEGNet. Next, to find a faster XAI method for spatio-temporal explanation, we design a novel technique based on the normalized discounted cumulative gain (NDCG) for selecting the best among a few saliency-based methods due to their higher time complexity than the permutation-based method. Among the saliency-based methods, DeepLIFT was selected because the NDCG scores indicated its results are the most similar to the permutation-based results. Finally, the fast spatio-temporal explanation using DeepLIFT provides deeper understanding for the classification results of the 3D-EEGNet and the important properties in the MI EEG experiments.

In perceiving a virtual environment (VE), distance perception is important in that it affects users’ behaviors and interactions. This study addresses an on-going issue of virtual reality (VR), distance underestimation within a VE. For this purpose, a 2x2x2x5 mixed design was used with two between-subject variables [visual cue for perception (provided and non-provided) and visual cue for action (provided and non-provided)] and two within-subject variables [response measure (verbal estimation and triangulated blind walking) and distance (1 m, 5 m, 10 m, 20 m, and 35 m)]. Sixty-four undergraduate or graduate students participated in the experiment, and a cube was chosen as the target object. The experimental results showed that visual cues for action mitigated distance underestimation and improved the accuracy of perceived distances. In general, distance underestimation was more obvious in the triangulated blind walking than the verbal estimation, and in the action/vista space than personal space, with relatively lower accuracies. Meanwhile, when visual cue for action were provided, the degree of underestimation decreased significantly with higher accuracy particularly in the triangulated blind walking. Although overestimation was found across all the distances in the verbal estimation, overestimation was found only for the personal distance, and underestimation was found for the action/vista distances in the triangulated blind walking. In both response measures, the accuracy of distance perceptions were the worst in the 10 m condition. These findings can be expected to provide insight into the problem of distance underestimation and can help guide design of VEs.