"Replacing" problem behavior: an analysis of tactical alternatives.

The author looked at the common rule: 'Always teach a replacement behavior when you want to stop problem behavior.'

He checked the logic behind that rule. He used past studies and basic behavior principles, not new data.

The paper says the rule is more habit than science. Sometimes you need a replacement skill, sometimes you do not.

The author argues we should decide case-by-case instead of always adding a replacement response.

Meyer (1999) and Staats et al. (2000) show clear wins when a matched replacement skill was taught. Those papers are examples of the practice Johnston (2006) questions.

Potter et al. (2013) took a different path. They treated only mild precursor behaviors and severe problem behavior still dropped. That outcome supports the idea that full replacement is not always required.

Boyle et al. (2024) update the talk for multiply-controlled behavior. They still sequence treatments, but they do not claim every function needs its own topographically new response. The field is moving toward the lighter stance Johnston (2006) recommended.

Next time you write a plan, pause before you add 'teach replacement behavior' as a default line. Ask: Will removing the reinforcer be enough? Could I strengthen an existing skill instead of inventing a new one? This small shift can save teaching time and still give you a safe, durable result.