An Investigation of Behavioral Contrast in a Simulated Workplace

Ring et al. (2023) built a fake office on campus. College students typed numbers into a computer for pay. The task had two parts. In Part A, every correct response earned money. In Part B, the screen said "EXTINCTION" and no money came out. The team watched whether Part A typing speed jumped when Part B stopped paying.

Only one student typed faster in the paid part when the unpaid part was shut off. The rest showed no jump or typed slower. Behavioral contrast—the classic rise in the unchanged task—showed up just once.

Older lab studies found contrast almost every time. Henton (1972), de Rose (1986), and Ginsburg et al. (1971) all saw big jumps after extinction. The difference: they used pigeons or rats and short, simple lever pecks.

Rogers-Warren et al. (1976) already warned that contrast disappears when reinforcement is only reduced, not stopped. Ring’s team used full extinction, yet still saw little effect. The gap shows human workplace tasks are more complex than animal lever presses.

Boyle et al. (2023) surveyed BCBAs the same year. Practitioners reported that contrast hurts rapport and buy-in across settings. The lab failure and field headache now match: contrast is weak in humans but still matters to clients.

If you withhold reinforcement in one setting, do not assume performance will boom somewhere else. Check data before promising parents or bosses a bounce. When you fade rewards, add extra reinforcement to the target task instead of counting on contrast that may never come.