Probability of response and probability of reinforcement in a response-defined analogue of an interval schedule.

Nigro (1966) built a tiny computer model of a pigeon. The model pecked in chains. Each chain length had its own chance of food.

By changing those chances, the team made the model act like real birds on a variable-interval (VI) schedule. They wanted to see if VI patterns pop out from two simple things: the bird’s built-in chain bias and the set food odds.

The model did copy real VI performance. Long chains plus low food odds made steady, slow responding. Short chains plus higher odds made faster bursts.

The study says VI patterns are not magic. They are the product of the animal’s own response style mixed with the experimenter’s reinforcement probabilities.

Iwata et al. (1990) later showed the same math breaks if extra food waits outside the session. Their open-economy birds ignored the model’s neat rate curve, proving the interaction only holds in a closed economy.

Lecavalier et al. (2006) went further. They gave rats inverted-U feedback functions and found the animals climbed to the peak. This optimization view keeps the probability idea but adds a hill-climbing rule, so it partly supersedes the 1966 account.

Adkins et al. (1997) added history power. Birds trained on DRH first kept their high rate even when switched to VI. The 1966 model missed that past probabilities can override current ones.

You now have a quick lens for puzzling response patterns. Ask two questions: What chain length is this learner prone to emit? And what are the true odds of payoff right now? If the rate looks weird, check economy, history, and feedback shape before you tweak the schedule. Use closed economies during teaching, probe for DRH/DRL history in assessments, and remember that optimization beats simple probability when reinforcers are scarce.