A01

The analysis of cognition and mind using deep generative models has significantly advanced over the past 13 years due to technological innovations and has recently been introduced into cognitive information processing for mind analysis.

To derive psychological aspects from vast datasets of objects, an integration of objects, cognition, and deep learning models will be conducted.

Human impressions of objects will be measured using Mixed Reality (MR) devices, deep models be trained, and "Material Mind" be reconstructed. For training, MR devices will be used to collect sensory impressions via the Semantic Differential (SD) method, subjective verbal impressions, and gaze data while viewing 85 Jomon pottery artifacts and 8 clay figures. Additionally, 3D-RGB data of the artifacts will be incorporated into a deep learning model embedded with a large language model.

The trained model is then used to analyze newly inputted 3D-RGB data of 354 additional pottery artifacts, 154 clay figures, and 173 triangular clay tablets. From the generated impression texts and gaze heatmaps, which simulate human perception of objects, the factors influencing Material Mind will be clarified.

Identical participant experiments are conducted at overseas universities, including those in the United Kingdom, to examine differences in Material Mind and its influencing factors, which are then presented to the global community at the international conferences of CAA.