Molecular contingencies: reinforcement probability.

Researchers put pigeons in a two-key box. Each key gave food at a different chance.

The chance changed every few seconds. Birds had to pick the key that paid off best right now.

The team watched which key the bird hit trial after trial.

Birds did not just copy the overall odds. They leaned toward the key with the better instant chance.

In other words, pigeons tracked moment-to-moment payoff, not long-run rate.

Emmelkamp et al. (1986) saw matching, not maximizing, when birds worked under a VT-VRT schedule. The clash fades once you see the 1975 study used quick, discrete trials while 1986 used steady concurrent schedules.

Iwata (1993) later showed that scheduled, never-seen rates still sway choice. That finding adds a layer: local payoff rules, yet the wider plan still whispers.

Alvarez et al. (1998) gave more proof that nearby cues, not global time gaps, steer picks. Together the papers say, watch the local moment, but keep the big picture on the radar.

Your client, like the pigeon, reacts to what just paid off. If you switch reinforcer odds within a session, expect rapid drift toward the richer moment, even if the overall plan stays the same. Track minute-by-minute data, not just session totals, to spot these micro-shifts and adjust teaching or thinning steps on the fly.