In neuroscience research, differences in brain activity patterns between individuals pose a challenge for large-scale data analysis. In this study, we will develop a technology to convert brain activity patterns across individuals without using common stimuli or tasks via latent representations of "content." We aim to (1) achieve high-accuracy brain information decoding from limited data, (2) integrate brain information from different modalities in a unified analysis, and (3) compare brain activity patterns between humans and model animals.