Development and modification of a response class via positive and negative reinforcement: a translational approach.

Smith et al. (2010) built a lab game with neurotypical adults. Players could pick one of three keys. Each key needed a different number of presses.

The team paid people with either candy (positive reinforcement) or by turning off loud noise (negative reinforcement). They watched which key players chose most.

No matter how they got paid, players quickly formed a response class. They almost always picked the key that needed the fewest presses.

Low effort won. The result was the same under candy pay or noise-stop pay.

McGonigle et al. (1982) used the same lab and showed the flip side. When they punished or blocked the easy key, people switched to the next-easiest option. Together the studies show a full picture: reinforcement builds low-effort classes; punishment or restriction breaks them.

Critchfield et al. (2003) ran a similar choice game but added punishment. Their data fit a one-factor model: punishment directly cuts the target response. Smith et al. (2010) fits that view because negative reinforcement (noise stop) raised only the cheapest response, not a rival one.

Edwards et al. (2020) later argued that escape behavior is more about learned stimulus end than about vague motivating operations. Smith et al. (2010) gives lab proof: participants escaped noise by landing on the same low-effort key they used for candy, showing the response class, not a special escape drive, drove the choice.

You now have clean lab evidence that both candy and escape build the same low-effort response class. When you write behavior plans, assume clients will land on the easiest way to get what they want. Build the desired response into that easy slot, or the client will find a quicker workaround. If you must block an unwanted shortcut, have an easy replacement ready—McGonigle et al. (1982) shows only the next-easiest behavior will rise.