Abstract: This study introduces ARTgram, an AR adaptation of a validated tangram matching task, aiming to delve into human collaboration within the metaverse. By employing the Muse EEG system, this study also pioneers low-cost hyperscanning to investigate similar neural responses expressed during the course of an AR cooperative task. Results demonstrate that tangram performance and partner trust increase as a function of neural similarities among interacting participants. These results signal a promise for low-cost hyperscanning in naturalistic AR communication. By validating various hyperscanning preprocessing routines and offering a high-control yet naturalistic AR task, this project sets the groundwork for theoretical and methodological advancements in understanding metaverse-mediated human interaction.