A human study participant sends feedback on a robot's success rate as it sorts objects. For many of us, picking up on someone else's disapproval can lead to anxiety or defensiveness, but for a new mind-monitoring robot, unspoken criticism is the very best kind. In their pursuit of a seamless brain-bot interface for letting humans control machines with their minds, researchers from MIT's Computer Science and Artificial Intelligence Laboratory have created a feedback system that lets robots detect and correct their errors based solely on a human's unspoken observation. Based on brain activity from an electroencephalography (EEG) monitor, the system detects when hooked-in persons notice an error in its sorting work, and adjusts its behavior accordingly within milliseconds. An illustration shows the communication loop between human subjects and ErrP-sensitive robots.