STIMULUS GENERALIZATION ALONG A LIGHT FLICKER RATE CONTINUUM AFTER DISCRIMINATION TRAINING WITH SEVERAL S-'S.
Each new S- nudes the whole gradient, so sequence your "not this" stimuli with care.
01Research in Context
What this study did
Researchers worked with pigeons to see how adding new "no-go" flicker rates changed their pecking. The birds first learned to peck only when a light flickered at one target speed. Then the team slowly added nearby flicker speeds as S- (no-peck) signals. They tracked where the birds still pecked after each new S- was added.
What they found
Each time a new flicker speed became an S-, the pigeons' pecking slid away from that spot. The gradient peak moved, and the far end of the scale flattened out. By the end, responding looked like a lopsided hill that had been pushed along the line.
How this fits with other research
Howard (1979) later showed the same sliding effect happens with people and a verbal "job status" line, proving the rule works outside the lab and with words, not just lights. Reed et al. (2003) found rats treat speech sounds the same way, so the idea crosses senses and species.
Mello (1966) used the same pigeon lab setup but with punishment instead of flicker. That study also saw flat gradients when birds lacked clear S- training, backing up the need for sharp contrast cues.
Stretch et al. (1966) asked how to measure these gradients. They learned that rate after the first peck drives the shape. So when you plot your own data, focus on steady response speed, not how long the learner takes to start.
Why it matters
When you teach a client to tell "correct" from "almost correct," add your S- steps one at a time and watch where the skill drifts. If you see responding spread too wide, tighten the range or insert another S- to push the peak back. Map the gradient early and pick the next S- to park the behavior where you want it.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Run a quick probe of five close stimulus values, plot the responses, and add an S- at the steepest drop to sharpen the peak.
02At a glance
03Original abstract
Four pigeons were trained with VI reinforcement to peck a key which was briefly illuminated by a flickering light. Generalization gradients were then obtained with nine different rates of flicker, four faster than S+ and four slower. Two birds were then trained to discriminate between S+ and the fastest stimulus (S-). These birds were then trained to discriminate between S+ and the two fastest stimuli, alternated as S-'s. This procedure was continued, adding one new S- at a time, until all four stimuli faster than S+ were S-'s. The remaining two birds were trained on this latter discrimination without intervening training. In a final stage, using the first two birds, the slowest stimulus was added as a fifth S-. Generalization gradients in extinction were obtained from each bird after each stage of training. As more stimuli from one end of the continuum served as S-'s, responding increased in the presence of stimuli from the other end of the continuum, and the gradient tended to become flattened at this end.
Journal of the experimental analysis of behavior, 1964 · doi:10.1901/jeab.1964.7-217