Behavioral effects of pairing an S-D with a decreasing limited-hold reinforcement schedule.

Three pigeons pecked a key for grain. A green light (S-D) meant grain was ready on a tight timer. The timer window shrank across days: 8 s, 4 s, 2 s, 1 s.

Birds also saw a red light (S-delta) that never paid off. Researchers tracked how often each bird pecked under each color.

When the grain window dropped to 2 s or 1 s, the birds lost the plot. They pecked like crazy under the red light even though it never paid.

Short limited-hold windows broke the stimulus control. The birds could no longer tell which light meant pay time.

Reid et al. (2005) extends this work to kids. Preschoolers first asked for teacher attention only when a yellow card was present. After weeks, the card was removed and the kids still asked only at the right times. The discrimination stuck even without the cue.

The two studies seem to clash. WEISSMAN (1963) says short limited-hold kills stimulus control; Reid et al. (2005) says you can remove the cue entirely and control stays. The gap is the training dose. The pigeons got minutes of practice; the kids got days. More training lets you fade or tighten cues without losing control.

Reed (2023) adds a twist. He showed that adding a brief signal right before grain boosts response variability when variability is required. Together, the papers warn: timing and signaling must match the skill you want. Tight windows or missing signals both erode control.

If a client starts working during the wrong cue, check the reinforcement window. A 1-s response window can wipe out discrimination you thought was solid. Try lengthening the window first, then thin it slowly. And give plenty of practice before you fade cues—kids and birds both need reps to lock in the difference.