Teaching serial position sequences to monkeys with a delayed matching-to-sample procedure.

Researchers taught monkeys to remember the order of pictures. They used a delayed matching-to-sample task. Brightness cues helped the animals pick the right sequence.

One group got the cues faded out slowly. The other group lost the cues all at once. The team watched who learned faster and who got stuck.

Slow fading won. Monkeys with gradual cues learned the order quickly. Monkeys who lost cues in one jump kept making the same errors.

Those errors stuck because wrong guesses were still rewarded some of the time. Intermittent reinforcement made the bad pattern hard to break.

Saunders et al. (2016) refined the same idea. They cut fading steps to 5 % and added a five-trial step-back rule. That package stopped the very error patterns A et al. saw when cues vanished too fast.

Herrnstein et al. (1979) stretched the logic to preschool kids. They also used brightness fades, but started with easy discriminations so children did not cling to the prompt.

Murphy et al. (2014) attacked the same problem from another angle. Instead of slower fades, they made consequences clearer for kids with autism. Both tactics broke stubborn errors; one changed the prompt speed, the other changed feedback salience.

When you remove prompts, take tiny steps and check often. Big jumps teach the wrong chain of responses. If a learner keeps repeating the same mistake, look at how fast you pulled the help and how often that mistake still pays off. Shrink the fade steps, add quick step-backs, or make correct answers more obvious. Either route stops the monkey trap in humans too.