Definition
Plain language
Free software for running psychology experiments, like reaction-time tests.
As stated in the literature
The Psychology Experiment Building Language platform for cognitive-task experiments; used as a computer-use agent environment where a task's stated controls (arrow keys) didn't match the software's actual key bindings (shift keys), making the task uncompletable as written.
Why it matters: It matters because it revealed a benchmark task that was literally impossible as written, highlighting how flawed instructions can unfairly trip up agents.
For example, a researcher might use PEBL to run a reaction-time test where volunteers press a key as fast as they can when a shape appears.