Performance in continuously available multiple schedules.
When schedules never pause, organisms overshoot the rich option.
01Research in Context
What this study did
Researchers let pigeons peck two keys for food around the clock. The keys followed different VI schedules that never turned off.
Birds could switch any time. Sessions ran continuously, not the usual one-hour lab slot.
What they found
Pigeons overmatched. They pecked the richer key even more than the reinforcer rate said they should.
Typical short-session studies show undermatching or close matching. Continuous access flipped the pattern.
How this fits with other research
Locurto et al. (1976) and Smith et al. (1975) saw undermatching in standard one-hour sessions. Same birds, same VI-VI schedules, opposite result.
The clash is only on the surface. The 1985 study kept the clock running. The earlier studies gave breaks. Breaks let preference reset; nonstop access let bias snowball.
Siegel et al. (1986) one year later asked which setup birds like better. Short links favored multiple schedules, long links did not. Together the papers show that schedule length tweaks both choice and matching.
Why it matters
If you run concurrent schedules all day—think classroom centers or shared tablets—watch for overmatching. Learners may lock onto the richer option harder than reinforcement rates predict. Build in brief breaks or rotate tasks to keep behavior balanced.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Add brief resets or timer cues every 10–15 min during long concurrent sessions.
02At a glance
03Original abstract
Three pigeons were given continuous access in their home cages to food reinforcement on two-component multiple variable-interval variable-interval schedules. The reinforcer rates in the two components were varied over seven experimental conditions, and a partial replication over five conditions was arranged one year later. When component reinforcer rates were unequal, ratios of component response rates were more extreme than ratios of obtained component reinforcer rates, a result which in a generalized-matching analysis is termed overmatching. This finding contrasts sharply with results obtained when multiple schedules are arranged in shorter sessions, in which performance is characterized by undermatching when subjects are deprived of food, and by matching, or equality between component response- and reinforcer-rate ratios, when deprivation is minimal. More molecular data obtained in two subsequent conditions suggested that this finding did not reflect local fluctuations or asymmetries in deprivation. Theories of multiple-schedule performance that predict that matching cannot be exceeded are disconfirmed by the present results.
Journal of the experimental analysis of behavior, 1985 · doi:10.1901/jeab.1985.44-343