Kazushi Tsutsui

This paper induces pressure through a fragile streak goal in a force-control task, showing an inverted-U performance pattern and increased physiological responses, but no consistent group-level signatures of pressure effects.

This study shows that human bounded rationality reduces performance in human-AI cooperative tasks compared to all-AI groups.

The paper shows that humans generate sequences to evade prediction, but these sequences remain predictable due to biases in how people estimate others' predictions.