Overview
- A peer-reviewed paper in Science by Amalia P. M. Bastos and Christopher Krupenye reports controlled “tea-party” experiments with transparent cups and jars to test pretend-object tracking.
- Kanzi identified the location of imagined juice and grapes at roughly 68–69% accuracy in critical trials that delivered no rewards, performing above chance.
- When offered a real-juice cup versus one associated with pretend pouring, he chose the real option 77.8% of the time, indicating he distinguished fiction from reality.
- Safeguards included counterbalancing order and side of actions and checking for frustration, reducing the likelihood of cue-following or simple reinforcement explanations.
- The team highlights Kanzi’s extensive language training and human exposure at the Ape Initiative as a limitation and proposes that such capacities may trace to a human–ape common ancestor 6–9 million years ago.