Unpredictable rewards keep you trying long after the rewards stop

Picture two gamblers at two slot machines. One machine has paid out on every single pull. The other has paid out only sometimes, on no pattern the player can spot. Then both machines go cold and stop paying at all.

The first gambler notices almost right away and walks off. The second gambler keeps pulling the lever, unrewarded, for a long time.

That difference is why psychologists say that rewarding every time is good for teaching something new, but rewarding only sometimes is the most efficient way to keep a learned behavior going. They call the first approach continuous reinforcement, and the second intermittent reinforcement, meaning a reward given only on some of the times the desired behavior happens.

It works for three practical reasons. Stopping to hand over a reward takes time away from the work itself, since you cannot pay a factory worker after every single piece without stopping the line. It is cheaper not to pay every time. And unpredictability itself creates excitement, so you get more repeats without spending more on rewards. Casinos run on that logic, because they want people to keep gambling without being able to afford big wins very often.

The slot machine story also shows the strangest part, what researchers call resistance to extinction. Extinction simply means the behavior stops once the reward is removed. If you have always been rewarded, a sudden dry spell stands out as a clear change. If you have been rewarded only now and then, a dry spell looks normal.

So an occasional, unpredictable payoff trains you to ride out the losses. Once you learn that sometimes means sometimes, no reward feels like just another gap before the next win, and you keep going.

The takeaway is to match the timing to the goal. Reward every correct try while someone is learning, then switch to occasional and unpredictable rewards once the habit is set, if you want it to last.