Within-session response rates when reinforcement rate is changed within each session.

The team worked with rats and pigeons in a small lab. They used a variable-interval (VI) schedule that changed its speed part-way through each session.

Animals pressed a lever or pecked a key. The researchers counted how fast the animals responded while the schedule shifted from rich to lean or lean to rich within the same hour.

Response rate moved with the schedule. When reinforcement got richer, the animals sped up right away. When it got leaner, they slowed down within minutes.

A simple equation (Herrnstein’s hyperbola) fit the rat data almost perfectly. It fit the pigeon data less well, showing species can tweak the same rule.

Silverman et al. (1994) ran the same VI switch one year earlier and saw the same quick within-session shifts. The 1995 paper is a direct replication with new numbers, so the effect is solid.

Buitelaar et al. (1999) later used the same drift to show economic demand curves also slide within sessions. The 1995 finding became the base for that bigger story.

LeBlanc et al. (2003) looked at pigeons under rich versus lean VI schedules and found both higher response rates and sharper stimulus discrimination. They extended the 1995 result from rate alone to rate plus accuracy.

If you run VI schedules in skill-building or maintenance sessions, know that behavior keeps drifting while the clock runs. A child’s correct-rate may look flat, but the true trend could be a slow slide because the schedule feels leaner by minute 20. Keep sessions short or swap in brief rich components to reset the drift. When you graph data, slice the session into thirds and watch for hidden trends instead of trusting one overall count.