Choice between fixed-interval schedules: Graded versus step-like choice functions.

Pigeons pecked two keys. Each key paid off on a fixed-interval schedule. One key gave grain after 30 seconds. The other gave grain after 60 seconds. The birds could switch any time. The team watched which key the bird pecked second-by-second.

They wanted to know if birds would always pick the shorter 30-second side. They also wanted to see if choice changed smoothly as the session went on.

The birds did not lock onto the 30-second key. They split their pecks in a smooth curve. Early in the interval they pecked both keys. As time passed they slowly shifted toward the side that was closer to payoff.

Choice looked like a ramp, not a switch. Tiny changes in the interval ratio produced tiny changes in preference. The data formed a gentle hill, not a steep cliff.

Nevin (1969) first drew the two-state picture: long pause, then sudden jump to fast pecking. Palya (1992) keeps the pigeons and the FI schedules, but adds a second key. The pause-jump pattern still shows up, yet choice across keys stays smooth. The old model holds for single-key timing; the new one shows how timing guides choice.

Neef et al. (1986) saw pigeons break the matching law under fixed schedules. They expected exclusive preference, but birds stayed wishy-washy. Palya (1992) finds the same wishy-washy curve under FI schedules. Together they warn us: fixed time or fixed ratio requirements can keep organisms from swinging all the way to one alternative.

Sanders et al. (1971) used concurrent variable-interval schedules and got neat matching. Palya (1992) uses concurrent fixed-interval schedules and gets gentle curves instead. The difference is the schedule type. VI gives steady payoff rates; FI gives one moment of payoff. That single moment spreads preference out instead of sharpening it.

When you run concurrent schedules with kids, do not assume they will pick the rich side 100%. If the schedules are fixed-interval—like a token every 5 minutes—expect smooth, partial preference. Track responses minute-by-minute. You will see a ramp, not a jump. Use that ramp to adjust timing, not to blame the learner for 'poor choice'.