Fading and errorless transfer in successive discriminations.

Fields (1978) tested how to fade color cues so learners switch to line cues with zero errors. The team used a lab setup with color-plus-line cards. They tried three fade orders: drop color from the correct card first, drop color from the wrong card first, or drop color from both cards together.

Each test looked at whether the learner could pick the right card after the color was fully gone.

Taking color away from the correct card, alone or with the wrong card, gave almost perfect accuracy. Taking color away from the wrong card first left the learner lost and errors showed up.

Bottom line: fade the positive cue, not the negative one, for errorless transfer.

MOORHEARSKELLEHER et al. (1964) did the first errorless fading study. They moved from bright-to-dim and reached near-zero errors. Fields (1978) keeps the errorless goal but asks which cue to fade, adding a fine-grain rule.

Berkowitz (1990) seems to clash. Autistic students learned picture discriminations faster with delayed prompting than with prompt fading. The clash is only surface: S compared two prompt-removal tactics, while L shows the best order inside a fading sequence. Both papers tell you to cut errors, but S says “try delayed prompting first” and L says “if you fade, start with the positive cue.”

Emmelkamp et al. (1986) and Gutierrez et al. (1998) took L’s positive-cue rule into the real world. They faded arm splints and tubes for kids who hurt themselves. Removing the most salient part of the restraint first kept self-injury at zero, just like removing color kept errors at zero in the lab.

Next time you fade prompts, strip the positive cue first. If you are moving from color to shape, drop the color on the correct card. If you are moving from hand-over-hand to a light touch, reduce contact on the correct response first. This tiny tweak keeps errors and problem behavior low across settings, from DTT tables to restraint reduction plans.