Contingency Discrimination Training and Resurgence: Effects of Reduced Extinction Session Durations
When we teach a new skill to replace problem behavior, that behavior can return if the new skill stops working—a phenomenon called resurgence. This study examines whether a training approach called contingency discrimination training (CDT) can reduce resurgence, and crucially, whether shorter “unavailability” periods can still provide protection. For clinicians using DRA or FCT, understanding how to build durable skills that hold up when reinforcement breaks down is essential.
What is the research question being asked and why does it matter?
This study asked a practical question about relapse. When we use DRA or FCT, problem behavior can return if the alternative behavior stops working—maybe the caregiver cannot provide attention, or the communication device is unavailable. That return is called resurgence. Clinicians see this when a learner was doing well, reinforcement for the new skill drops, and the old behavior reappears.
A treatment add-on called contingency discrimination training (CDT) has reduced resurgence in past studies. CDT involves planning times when the alternative response gets reinforced (“on”) and times when it does not (“off”), while the old target behavior stays on extinction. The idea: the learner practices the real-life pattern where sometimes the new response works and sometimes it does not—and that does not mean the old behavior will work again.
The problem is that “off” time is essentially extinction for the alternative response. In the clinic, we usually want to limit extinction, especially for a valuable skill like communication. So the key question was: can we keep the relapse-protection benefits of CDT while making the “off” periods shorter?
What did the researchers do to answer that question?
The researchers used rats pressing levers for food. This is not the same as teaching a child communication, but it allows tight control over the schedule. First, all rats learned a “target” lever that earned food (baseline). Then the study moved into “treatment,” where the target lever stopped earning food (extinction) and a new “alternative” lever earned food (a DRA-like setup).
One group had alternative reinforcement available in every treatment session (the “All On” group). The other groups had CDT, meaning sessions alternated between “on” (alternative reinforced) and “off” (alternative not reinforced). The key difference between CDT groups was the length of “off” sessions: some had long sessions (30 minutes), some medium (15 minutes), some very short (5 minutes), and one group had sessions that started short and increased over time.
After treatment, all groups entered a “test” phase where both levers were on extinction. This mimics a real-life moment when reinforcement for the alternative behavior drops out, letting researchers measure whether the old behavior returns—especially in the first test session.
How you can use this in your day-to-day clinical practice
If you use DRA or FCT and worry about relapse when reinforcement gets thin or breaks down, this study supports a training idea: plan practice periods where the alternative response does not pay off, but do it in a controlled way. The point is not to withhold reinforcement to be tough. The point is to teach a clear rule through experience: “Sometimes this new response won’t work right now, and the old behavior still won’t work either.”
In real settings, that can look like short, planned times when the learner is signaled that the reinforcer is unavailable, paired with support for safe waiting, coping, or switching to another valid option.
The most useful detail from this study is that the length of “off” periods mattered considerably. When “off” periods were very short (5 minutes), CDT did not reliably reduce resurgence compared to always reinforcing the alternative. When “off” periods were moderate (15 minutes), resurgence was lower but not eliminated. When “off” periods were longer (30 minutes), or started short and then increased, resurgence was essentially absent in the first test.
For clinicians, the takeaway: tiny “off” exposures may not provide enough practice to build the discrimination you want, but you may not need extremely long “off” times to see benefit.
A practical way to translate this is to think in terms of “dosage” of unavailability practice. If your current plan is “we never let the alternative contact extinction,” you may be setting up a situation where the learner only learns one world: “my new response always works.” When real life happens and it does not work, the learner may be more likely to try old behavior again.
This study suggests you can reduce that risk by adding planned unavailability periods. But it also suggests a 30-second or 1-minute “not available” practice will not do the job alone. You may need enough time and enough repetitions for the learner to truly contact “not available” conditions and learn what to do instead.
Watch what happens to the alternative behavior during “off” times. In the study, alternative responding dropped during “off” sessions and returned during “on” sessions—exactly what you would want if teaching flexible, context-sensitive behavior. In practice, aim for the same: the learner uses the alternative response when available, and when it is not, they shift to another taught response (wait, ask later, ask for help, choose another activity) rather than escalating.
If your learner stops using the alternative response even when reinforcement is available again, that is a warning sign that your “off” practice is too hard, too long, or not well supported.
This study also addressed a common concern: “If I add off periods, will I cause more problem behavior during treatment?” Here, the total amount of target responding during treatment did not increase for CDT groups compared to the constant alternative reinforcement group. That does not guarantee the same in your case, especially with severe behavior, but it suggests CDT-style programming does not automatically mean more problem behavior overall.
Clinically, you would still plan for safety: start with low-risk targets, strong reinforcement histories for the alternative, high-quality prompting, and clear signals for availability.
The “escalating off duration” condition is a useful clinical idea, even as early evidence. It suggests a shaping approach: start with shorter “not available” periods and gradually increase them if the learner is stable. If you try this, make changes based on data, not a calendar. If the learner shows distress, aggression, or a spike in problem behavior, pause, shorten the “off” time, improve supports, or teach missing skills (like tolerating delay) before increasing again.
Be careful not to overapply these findings. This was a rat lever-press study with food, not children with complex learning histories, trauma histories, or dangerous problem behavior. The study used clean, separate sessions with clear contingency shifts. In homes and schools, unavailability is often messy and inconsistent, which may change the effect. Use this as a planning idea to test thoughtfully, not as a rule that “15 minutes is enough” or “5 minutes is useless” for every learner.
If you try a CDT-like approach, your clinical decision points should include: whether the learner has a safe way to respond during unavailability, whether caregivers can implement signals consistently, and whether the “off” periods are ethically acceptable given the learner’s needs.
The goal is not compliance with “no.” The goal is durable skills with choice and dignity, so that when reinforcement is unavailable, the learner still has workable, respectful options and does not need to return to unsafe behavior to get needs met.
Works Cited
Shahan, T. A., Hiltz, J. B., Avellaneda, M., & Greer, B. D. (2026). Contingency discrimination training and resurgence: Effects of reduced extinction session durations. Journal of the Experimental Analysis of Behavior, *125*(1), e70072. https://doi.org/10.1002/jeab.70072
