On the effects of food deprivation and component reinforcer rates on multiple-schedule performance.

Wallander et al. (1983) worked with lab rats on a two-key multiple schedule. Each key paid off on its own variable-interval plan. The team first let the rats eat freely until body weight hit 100%. Then they repeated sessions at 90% and 80% of that free-feeding weight. Throughout, they counted how often the rats pecked each key and how many food pellets each side produced.

When the rats were well-fed at 100% weight, their response ratios almost perfectly matched the reinforcer ratios. At 90% weight the match was looser. At 80% weight the two ratios barely lined up; response rates on the leaner key stayed higher than the payoff rate predicted. Hunger made the animals less sensitive to the real payoff odds.

Dove et al. (1974) saw a similar hunger effect using concurrent food versus water keys. Pre-feeding pushed the rats toward the water key, just as lower body weight in the 1983 study pushed rats toward the poorer food key. Leslie (1981) had already shown that local response rate tracks local reinforcement probability moment to moment. L et al. extend that idea: deprivation changes how strongly those local odds steer overall choice. Williams (1971) found that concurrent FI/VI response ratios follow a power function with exponent 0.5. The 1983 data add a new knob: the exponent itself shrinks when body weight drops, flattening the matching curve.

If you run concurrent or multiple schedules, remember that a hungry client will not 'play the odds' the same way. Before session, check meal timing and snack access. A child who skipped lunch may respond more on the leaner schedule than your data sheet predicts. Adjust reinforcer rates or add a quick snack so the matching relation you expect actually shows up.