FURTHER EXPERIMENTS ON PROBABILITY-MATCHING IN THE PIGEON.

Researchers put pigeons in a box with two keys.

Each key gave grain on a different schedule.

Some birds got a correction trial after wrong pecks.

Others saw both keys light up at the same time.

The team changed these small rules across days.

Birds did not stick to one rule.

With correction trials they almost always picked the richer key.

Without correction they split pecks to match the payoff odds.

A tiny switch—do-over or no do-over—flipped the whole pattern.

Mellitz et al. (1983) later showed one hill-climbing rule can explain both looks.

The birds just took the best local option moment to moment.

Schenk et al. (2020) found the same math fits kids shooting 2- or 3-point shots in a video game.

Same equation, feathers or fingers.

Frank-Crawford et al. (2018) also saw choice swing when work demands changed.

Together the papers say: watch the task details, not just the numbers.

Your client may “match” or “maximize” depending on tiny setup cues.

If you give an error-correction trial after a wrong response, you may push the child toward the richer side every time.

Remove that trial and you might see a 70-30 split that mirrors the payoff rate.

Check your procedure before you label a choice pattern as a trait.