Multiple determinants of the effects of reinforcement magnitude on free-operant response rates.
Bigger reinforcers speed VR responding but can depress VI responding, so match magnitude to schedule type.
01Research in Context
What this study did
Reed (1991) tested how the size of a food reinforcer changes how fast rats press a lever.
Some rats worked on a VI schedule where the first press after an unpredictable time produced pellets. Others worked on a VR schedule where every few presses produced pellets. A third group had two levers at once.
The team varied the number of pellets per press from one to four while counting every response.
What they found
On VI schedules, bigger pay-offs actually slowed the rats down. On VR schedules, the same bigger pay-offs sped the rats up.
When both schedules ran together, the rats pressed the richer side more. The same reinforcer size can hurt or help depending on the schedule.
How this fits with other research
CATANIA (1963) saw the same VI drop decades earlier. That team used concurrent VI only and warned that single VI hides magnitude effects. Reed (1991) now shows the drop can flip to a boost just by switching to VR.
Reed et al. (1988) ran a near-copy of the VR part and found the same speed-up, giving a direct replication.
Lowe et al. (1974) looked at pauses instead of speed. They found bigger pellets always lengthened the post-reinforcement pause. Reed (1991) agrees on VR but adds that the pause can win so completely on VI that overall rate falls.
Davison et al. (2003) zoomed in on single visits. Larger pellets kept pigeons in one spot longer, showing the magnitude signal works moment-to-moment as well as across the whole session.
Why it matters
If a client slows down when you add extra praise or candy, check the schedule before you drop the size. On rich VR-like tasks such as discrete-trial drilling, bigger rewards will probably push faster responding. On loose VI-like tasks such as casual play or DRL pacing, the same reward may create contrast and slow the rate. Try schedule tweaks first, then adjust magnitude.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Track response rate across two tasks: one fast-paced (VR-like) and one slow-paced (VI-like), then test whether doubling the token or edible raises or lowers speed in each task.
02At a glance
03Original abstract
Four experiments examined the effects of increasing the number of food pellets given to hungry rats for a lever-press response. On a simple variable-interval 60-s schedule, increased number of pellets depressed response rates (Experiment 1). In Experiment 2, the decrease in response rate as a function of increased reinforcement magnitude was demonstrated on a variable-interval 30-s schedule, but enhanced rates of response were obtained with the same increase in reinforcement magnitude on a variable-ratio 30 schedule. In Experiment 3, higher rates of responding were maintained by the component of a concurrent variable-interval 60-s variable-interval 60-s schedule associated with a higher reinforcement magnitude. In Experiment 4, higher rates of response were produced in the component of a multiple variable-interval 60-s variable-interval 60-s schedule associated with the higher reinforcement magnitude. It is suggested that on simple schedules greater reinforcer magnitudes shape the reinforced pattern of responding more effectively than do smaller reinforcement magnitudes. This effect is, however, overridden by another process, such a contrast, when two magnitudes are presented within a single session on two-component schedules.
Journal of the experimental analysis of behavior, 1991 · doi:10.1901/jeab.1991.55-109