Local rates of responding and reinforcement during concurrent schedules.

Reid et al. (1983) pulled together every lab study that tracked moment-to-moment responding on concurrent schedules. They asked: do local response rates follow the same matching rule we see in overall rates?

The review covered pigeons, rats, and a few human studies. All used two keys or levers running side-by-side VI or VR schedules while computers counted every peck or press.

With long changeover delays, local reinforcement rates quickly equalized between keys. Local response rates, however, stayed different when one schedule was a ratio and the other an interval.

The data fit two ideas: the Equalizing Principle (reinforcement evens out locally) and the Melioration Principle (animals shift until payoff is the same). The classic Matching Law needed these tweaks to handle short time windows.

Emmelkamp et al. (1986) ran a direct replication and saw the same pattern: pigeons matched even when response effort was identical, proving matching is not just a cost trade-off.

Cohen (1975) had already shown that time allocation, not response form, drives concurrent performance. K et al. folded that finding into their refined Matching Law.

Malone (1999) took the idea further, building a local stay/switch model that predicts exact run lengths from reinforcement ratios—no across-key comparison needed.

Geckeler et al. (2000) moved the lab principle to children with autism. They showed that reinforcer choice boosts responding only when two response options are present, echoing the local-rate effects K et al. summarized.

When you run concurrent reinforcement programs, watch local rates, not just overall totals. If you mix a ratio schedule with an interval schedule, expect different local response speeds even when reinforcement evens out. Use a changeover delay to keep each schedule’s pattern clean. For clients who get reinforcer choice, make sure two clear response options exist; choice alone won’t boost single-response tasks.