The effect of modelling on drinking rate.

Three college students sat in a quiet room with a beer. A second person drank at a set speed. The researchers watched to see if the student matched that speed.

The confederate drank one-third faster or one-third slower than the student’s first-day rate. No one spoke. The only cue was the drinking pace.

Every student copied the new speed right away. When the model slowed, they slowed. When the model sped up, they sped up.

The change was exact and lasted the whole session. Modeling alone moved a simple operant behavior up or down by one-third.

Peterson et al. (2019) extended this idea to kids with autism in DTI. Varied modeling gave only brief flexibility and slowed learning for half the children. One clear model still works best when you need fast skill gain.

Marano et al. (2020) conceptually replicated the effect with staff training. Staff who watched and scored videos of preference assessments later ran the procedure more accurately. Modeling travels well from beer drinking to staff behavior.

Morris et al. (2025) moved the lab forward by testing matching-law sensitivity in young adults. The 1977 paper showed that modeled rate can override baseline drinking; the 2025 paper asks if absolute reinforcement rate changes sensitivity to relative rate.

You already use modeling, but this study reminds you how powerful it is. A single visible pace can reset a client’s response rate without extra prompts or tokens. Next time you teach a motor skill, try demonstrating the exact speed you want and watch the learner match it.