Supplementary knowledge of results.

The team added a counter that clicked each time a college student got a trial right.

Students also kept the usual money for correct answers.

Later the group added a small fine for every wrong answer.

The counter alone made students answer faster but they made more mistakes.

When the small fine was added, accuracy bounced back while speed stayed high.

Extra rewards can boost speed but hurt quality unless you also penalize errors.

Davis et al. (1972) saw the same drop in baseline when pigeons earned extra grain.

Their birds slowed down; the students sped up and slipped—same warning: extra pay can warp performance.

Parsons et al. (1981) later showed the fix is to weave the reward into the task itself.

They had autistic children open a lid to reach candy; learning shot up without extra trinkets.

Together the three papers say: if the bonus sits outside the response, watch for trouble; if it’s built in, you usually stay safe.

If you run a token board or point counter, check accuracy the first week.

If errors climb, add a small cost—lose one token, pay one point—while keeping the counter.

You keep the momentum and protect the skill.