A comparison of two types of extinction following fixed-ratio training.

Weissman et al. (1966) worked with pigeons that had learned to peck on a fixed-ratio 50 schedule. After every 50 pecks the birds got grain.

The team then split the birds into two extinction groups. One group saw the usual dark chamber. The other group saw the same chamber plus a flashing light that had earlier signaled grain.

Both groups kept the same burst-pause pattern they had used during training. The flashing light did not cut the total number of extinction pecks.

The light only kept the ratio chunking intact. It gave no bonus protection against the drop to zero reinforcement.

Rodriguez-Seijas et al. (2020) later showed that hopping back to the old grain context during extinction does lower renewal. Their birds got brief grain sessions in Context A while extinction ran in Context B. This on-off design cut relapse when the birds later saw a brand-new Context C.

King et al. (2025) took the idea to children with autism. They found that giving the original reinforcer non-contingently during extinction tamped down resurgence better than a novel reinforcer. The 1966 null result looks like a first step that these later studies refined.

Storch et al. (2012) also built on the context theme. They trained an alternative response in a separate room before mixing the two contexts. That move lowered both resistance to extinction and later relapse. Again, the 1966 paper showed the baseline: a simple stimulus swap alone is not enough.

If you are fading reinforcement for a client, adding a favorite toy or video that used to signal treats may keep the old response pattern alive without cutting the total burst of problem behavior. To get real relapse protection, pair that stimulus with brief, scheduled reinforcement breaks or train a new skill in a different room before you mix the contexts. Use the 1966 finding as a warning: a cue alone is neutral; what matters is how you schedule the reinforcement around it.