Some effects of discriminative training with equated frequency of reinforcement.

Pigeons learned to peck one key when they saw a green light and to peck a different key when they saw a red light.

Both colors paid off with food the same number of times, so reinforcement was perfectly balanced.

The team then tested how the birds spread their pecks to colors that looked like the training green and red.

Even though green and red gave food equally, the birds pecked more to colors near the color that had been tied to their higher peck rate.

In other words, response history, not just payoff, nudged where the birds generalized.

Jensen et al. (1973) later showed the same thing can happen without any payoff change at all. They saw contrast effects that echoed M et al.'s pattern, proving that past response rate alone can bend stimulus control.

Kelly (1973) flipped the setup: when reinforcement was no longer equal, the steepness of the generalization gradient tracked the new payoff ratio. Together, the two papers bracket the 1968 result—equal pay gives response history the wheel, unequal pay hands the wheel to reinforcement.

Reberg et al. (1979) went further, showing that in detection tasks you can shift bias without touching how well the birds tell stimuli apart. This sharpens the 1968 takeaway: reinforcement mainly moves the decision line, not sensory clarity.

When you run discrimination lessons, remember that clients bring response habits, not just reinforcement histories. If two cues pay the same but one cue has sparked faster responding in the past, expect generalization to lean that way. To get even control, you may need to balance both reinforcer rates and response rates across stimuli.