Choice behavior and the accessibility of the reinforcer.

The team ran two pigeons on two side-by-side VI schedules.

They changed how long the birds had to wait after switching keys.

They also tested if matching held when one key paid off for 3 s and the other for 7 s.

Matching stayed tight when the delay was 0 or 1.5 s.

At 3 s the birds’ choices drifted away from the payoff rates.

Duration did not guide choice; the birds still matched rate, not time.

Santi (1978) later showed the same rule works for negative reinforcement.

Rats matched timeout-from-shock rates just like pigeons matched food rates.

Davison et al. (1984) then warned that VI VR mixes can fake matching.

Their model said the schedule itself, not the bird, can push the ratio.

Green et al. (1999) added that even when the ratio looks right, short hops to the lean key may be optimizing, not matching.

Together the story is: matching holds on pure VI, but watch the delay and watch the schedule type.

If you run concurrent reinforcement, keep the change-over delay under 1.5 s.

Longer waits break the contact between choice and payoff.

Also, stick with VI VI pairs when you want clean matching data; VI VR can fool you.