Behavioral contrast in the pigeon: a study of the duration of key pecking maintained on multiple schedules of reinforcement.

The team watched how long pigeons held each key peck during a two-part schedule. One part paid every 30 s. The other part paid every 90 s. They timed every peck to the millisecond.

Birds worked in 90-minute sessions. Sessions flipped between the two parts every 60 s. No food ever came from the second key, only from the first.

Overall response rate on the food key never changed. Peck length told a different story. Right after each switch, birds gave a burst of very short pecks on the signal key.

Those short pecks stopped after a few seconds. The authors say the burst is two response classes adding together, not true contrast.

Green et al. (1975) ran almost the same birds and same boxes the same year. They saw a late-component jump in rate that the summation idea cannot explain. The two papers look opposite, but they measure different moments. B et al. caught the first three seconds; L et al. caught the last ten.

Van Hemel (1973) had already shown that early-component rate surges drive contrast. The 1975 data narrow the surge to ultra-short pecks, tightening the timeline.

Gentry et al. (1980) later proved you can shape peck length up or down with differential reinforcement. That gives clinicians a lever: if duration matters, you can train it.

When behavior seems to "contrast," check the micro-structure before you re-design the program. Split your data into one-second bins. If you see a brief spike that dies out, the change may be two topographies summing, not a true reinforcement effect. Target the spike itself or wait for it to pass instead of altering the whole schedule.