Concurrent schedule assessment of food preference in cows.

Lutzker et al. (1979) let cows choose between two foods. Each food sat at the end of its own lever. Pressing a lever delivered a bite. The team changed how often each bite was ready. They watched which lever the cow pressed most.

The setup is called a concurrent schedule. It tests the matching law. The law says response ratios should match reinforcement ratios.

Cows followed the matching law, but not perfectly. They pressed the rich lever more, yet not as much as the payoff rate predicted. This mild gap is called undermatching.

In plain words, cows showed clear food preference, and their choice ratios tracked the bite ratios.

Macdonald et al. (1973) first showed perfect matching in pigeons pecking keys. Lutzker et al. (1979) stretched the law to a new species and response: cows pressing levers for food.

Ghaziuddin et al. (1996) ran almost the same cow study years later. They stripped out pauses and side activities. Undermatching still showed up. This confirms the 1979 result was not a fluke.

Hall (2005) later showed the basic matching law can break down when the work needed to earn food differs across levers. The cow data sit in the middle: matching holds, but only if earning effort is equal.

The matching law is a ruler for choice. If you track how often each option pays off, you can forecast how a client will split their time. The cow study tells us the ruler works outside the pigeon lab and outside human speech. It also warns us to expect undermatching: people or animals may not chase the richest choice as hard as the math says. When you see this mild drift in your data, do not toss the law—note it, then check for extra effort, delays, or bias that could explain the gap.