Shaping by automated tracking of an arbitrary operant response.

Three pigeons wore tiny sensors on their heads. A computer watched every move.

When a bird’s head got close to a 3 cm sphere, the machine dropped food. The gap needed shrank bit by bit. No human clicked a feeder.

All three birds learned to bonk the sphere in minutes. The rig shaped a brand-new move with millimeter accuracy.

Vostanis et al. (2024) took the same idea into a kindergarten. They used auto-tracking to teach autistic kids to look where a classmate pointed. Birds or children — the code still rewards closer and closer tries.

Shih et al. (2010) did it with cursors. Adults with developmental disabilities earned tokens when two mice met on screen. Again, the computer set the bar, not the teacher.

Szempruch et al. (1993) ran an earlier pigeon lab. Their box auto-advanced trials but still needed hand-made lesson plans. The 1987 gear closed the loop: sense, judge, and pay in real time.

If you can track it, you can shape it. Head angle, eye shift, finger hover — any micro-movement becomes a target. Let the software tighten the rule while you watch data. Free hands, sharp timing, and one-to-one precision for every client.