Study finds errors by expressive robots lower trust in a distinct way

In a Science Robotics study, 50 people talked with a Pepper robot that sometimes erred while being either animated or still. Mistakes cut trust in both cases, but brain imaging showed errors from the animated robot were processed more like social violations, and higher oxytocin predicted lower trust, a finding relevant to humanoid interface design.

Summary written by DroidBazaar. The original article belongs to its publisher.