Parametric analysis of delayed primary and conditioned reinforcers

Leon et al. (2016) compared two kinds of delayed reinforcement with adults who have intellectual disabilities.

They used an alternating-treatments design. Sometimes a food treat came after a short wait. Other times a plastic token came after the same wait; the person could trade tokens for food later.

Immediate food served as a control condition to check that the task itself worked.

When reinforcement had to wait, food kept responding stronger than tokens did.

Participants kept pressing the button longer for delayed food than for delayed tokens.

Immediate food worked best, but delayed food was clearly better than delayed tokens.

Einfeld et al. (1996) seems to disagree. They found that children with autism ate more new foods when reinforcement came right away, not after a delay. The two studies clash because Leon shows delayed food still works while L shows it fails. The gap closes when you see the settings: Leon taught a simple button press; L treated feeding problems. Feeding is more sensitive to delay, so immediacy won there.

Reed et al. (1988) set the stage. They showed that adults with ID pick smaller-sooner rewards when delays grow, the classic delay-discount curve. Leon adds a twist: even within delayed options, edible beats token.

Clarke et al. (2003) extends the idea. They gave one adult with brain injury a simple hand-open task during the wait and saw choice swing back toward larger delayed rewards. Pairing Leon’s finding with this trick gives you a toolkit: use food if you must delay, or add a motor task if you want tokens to survive the wait.

If you have to pause reinforcement—while you walk to the cabinet, count responses, or help another client—hand the client an edible right after the wait, not a token. Tokens lose value faster during empty time. Save token boards for moments you can deliver immediately or when you can add a simple concurrent task to bridge the gap.