Comparing positive and negative reinforcement: A fantasy experiment

Nevin et al. (2017) wrote a blueprint for a fantasy experiment. They asked: do positive and negative reinforcement change behavior in the same way?

To find out, they sketched a lab with two matched groups. One group earns candy for pressing a lever. The other group avoids loud noise by pressing the same lever.

Both groups would get the same amount of reinforcement and work under the same schedule. The authors then list the ethical roadblocks that stop the study from ever running.

The paper does not give new data. Instead, it shows that no one has yet run a clean, fair comparison of the two reinforcement types.

The authors warn that any real trial would face heavy ethics review. Taking away something a person needs, like food or quiet, can harm the participant.

Cao et al. (2026) later ran a real mouse study that acts out part of the fantasy. They tracked five separate ways persistence can show up: effort, endurance, resistance to extinction, consistency, and sequence stability. Their data give the fine-grained metrics the target paper says we need.

Last et al. (1984) and Kohlenberg (1973) already tweak reinforcement parameters in the lab. They change stimulus length or reinforcer size and watch response rates shift. These studies prove the tight experimental control demanded by Nevin et al. (2017) is possible, but they stay within positive reinforcement only.

Iqbal (2002) shows the ethical worry is real. A clinical team tried a differential reinforcement plan for an adult with disabilities. Staff stopped the plan because taking away preferred items felt wrong, and the behavior returned. This case mirrors the ethical barrier the target paper predicts.

If you design or review behavior plans, this paper reminds you to ask: are we choosing positive or negative reinforcement because it works better, or because it feels safer? The fantasy design gives you a checklist: match reinforcer size, keep schedules equal, and track response rate. Until someone runs the study, treat claims that one type is stronger with caution, and always weigh client rights against experimental curiosity.