Competition: Some behavioral issues.

Schmitt (1986) wrote a think-piece, not an experiment.

The paper says we still don’t know when competition helps or hurts learning.

It lists four reinforcement details we must measure: size of skill gaps, how the prize is earned, how it is shared, and any cues that show who is ahead.

There is no new data.

The author shows that past “cooperation beats competition” claims skipped these four details, so the claims are premature.

Cullinan et al. (2001) and Grosch et al. (1981) ran pigeon studies showing that a simple light signaling “you will get food in 5 s” steers choice. Schmitt (1986) wants the same kind of signaling tested in people who can see a peer’s scoreboard.

Durand (1982) tracked how pigeons split time between two levers when one pays on a slow clock and the other on hard work. The results preview R’s point: the exact pay rule, not the label “competition,” decides where the bird goes.

Gomes‐Ng et al. (2023) later showed that paying for wrong answers can pull stimulus control to the wrong cue. The data back R’s warning: if you set up a contest and reward errors (say, second-place points), you may accidentally strengthen the very behavior you want to remove.

Next time you run a game or “table leader” program, stop guessing. Track the four variables: skill gap, earning rule, split of prizes, and peer cues. Start with a quick baseline of each learner’s rate, then adjust only one variable at a time. You will turn competition from a blunt tool into a measured intervention.