Humans' sensitivity to variation in reinforcer amount: Effects of the method of reinforcer delivery.

King et al. (1990) asked adults to pick between two buttons. One button gave a few points right away. The other gave more points after a short wait.

Sometimes the adults had to press a third 'eat' button to collect the points. Other times the points just piled on by themselves. The team watched how often people chose the bigger, later reward under each setup.

When people had to press the 'eat' button, they usually picked the larger reward. They acted with self-control.

When the points arrived automatically, the same people grabbed the smaller-sooner pile far more often. Sensitivity to amount dropped sharply.

Rutherford et al. (2003) later showed pigeons also track token amount, but only when the final grain delays are equal. Their lights worked like our 'eat' button, acting as extra conditioned reinforcers. The two studies line up: a brief action at the end sharpens amount discrimination.

Austin et al. (2015) go further, proving tokens can wear several hats at once—reinforcer, signal, or elicitor—depending on how you schedule them. That warns us the 'eat' step is doing more than we might think.

Ainslie et al. (2003) seem to disagree at first glance. Bundling several rewards in a row helped rats wait longer, even without an extra response. The key difference is species and task: rats got many small deliveries during the delay, while our humans lost contact with amount when delivery became passive. Method, not magic, explains the gap.

Keep a quick consummatory response in your token board or point system. Let the learner click a tally, drop a chip in a jar, or swipe a card. That tiny action keeps the brain tied to how much was just earned and supports waiting for the bigger payoff. Skip the step and you risk sliding back to impulsive picks.