Induction by reinforcer schedules.

Conant et al. (1984) wrote a theory paper. They looked at how past work tried to pull "intermittency" out of a schedule.

The authors said that move is invalid. They urged scientists to study the whole schedule as one unit instead.

The paper finds no data. It argues that schedule parts work together and cannot be split.

When you carve out "how often" alone, you lose the effect of delay, amount, and what happens next.

Angle (1970) showed rats only repeat an IRT if it actually produced food. That early hint says the full contingency, not just time, controls behavior.

Attwood et al. (1988) later bunched all food at the end of the session. VI rates dropped, proving that timing across the whole schedule matters.

Catania (2025) now frames reinforcers as parts of "consequence classes." That idea grows straight from the 1984 call to keep the schedule intact.

When you write a program, look at the whole schedule box, not just the FR or VI number. Ask: how long, how big, what next? Changing one knob changes the whole system. Map the full contingency first, then tweak.