Atsushi Masumori
The University of Tokyo, Fuji Machine (Japan), Tokyo University of the Arts
Papers
7
Total Citations
76
H-Index
4
About
Atsushi Masumori is a pioneering researcher at the intersection of artificial life, embodied cognition, and neurorobotics. His work centers on two transformative research areas: developing biologically inspired learning principles for neural networks, and grounding large language models in physical robotic bodies. Masumori’s most influential contribution is the “Stimulus Avoidance Principle,” a general learning rule for spiking neural networks that enables adaptive behavior without explicit supervision. This principle was experimentally validated using real cultured neuronal cells controlling a robot (15 citations), demonstrating how sensorimotor coupling can shape neural circuits. His work on the humanoid robot Alter3 has been equally impactful, with a 2021 study on personogenesis through imitation (17 citations) and a 2025 paper integrating GPT-4 for spontaneous motion generation (9 citations). This latter achievement—grounding a large language model in a physical robot—represents a breakthrough in overcoming the hardware-software gap that typically limits LLM applications in robotics. Masumori’s research has accumulated over 76 citations, with his 2017 paper on learning by stimulation avoidance (27 citations) being his most cited work. His ongoing exploration of minimal selfhood in robots, as demonstrated in Alter3, continues to push the boundaries of artificial consciousness and autonomous agency.
Research Focus
Key Achievements
Top Papers
- 1
- 2Personogenesis Through Imitating Human Behavior in a Humanoid Robot “Alter3”17 citations · 2021
- 3
- 4From text to motion: grounding GPT-4 in a humanoid robot “Alter3”9 citations · 2025
- 5A new design principle for an autonomous robot4 citations · 2017
- 6Minimal Self in Humanoid Robot “Alter3” Driven by Large Language Model2 citations · 2024
- 7From Text to Motion: Grounding GPT-4 in a Humanoid Robot "Alter3"2 citations · 2023