Yu Kohno

Nihon University, Tokyo Denki University

Papers

Total Citations

H-Index

About

Yu Kohno is a researcher whose work lies at the intersection of cognitive science and robotics, focusing on how intelligent agents can efficiently learn and act under severe constraints. His key research areas include reinforcement learning, cognitive biases, and state-space compression for robotic action acquisition. Kohno’s major contribution is the concept of "Cognitive Satisficing," introduced in his 2016 paper of the same name (6 citations), which addresses the fundamental challenge of exploding state-action spaces in reinforcement learning. He argues that when physical and computational limits make exhaustive exploration impossible, agents must adopt satisficing strategies—settling for "good enough" solutions rather than optimal ones. This idea is further developed in his work on "Robotic action acquisition with cognitive biases in coarse-grained state space" (2 citations), where he demonstrates how incorporating human-like cognitive biases can help robots learn effectively in simplified, coarse-grained environments. Though his citation counts are modest, Kohno’s research offers a provocative alternative to traditional optimality-driven approaches, suggesting that bounded rationality and heuristic decision-making may be essential for scalable, real-world AI. His work is particularly relevant for researchers exploring resource-limited robotics and cognitively plausible learning algorithms.