Accuracy of discrimination, rate of responding, and resistance to change.
Higher reinforcement rates make both response speed and discrimination accuracy stronger and harder to disrupt.
01Research in Context
What this study did
The team worked with pigeons in a lab.
They changed how often food came after correct pecks.
Some birds got food every 4 pecks. Others got food every 32 pecks.
They then tested how well the birds told colors apart and how fast they pecked.
What they found
Rich schedules made birds peck faster and pick colors more accurately.
When they added free food or made the room dark, rich-schedule birds kept pecking longer.
Lean-schedule birds stopped sooner.
Both speed and accuracy were tougher to disrupt under rich reinforcement.
How this fits with other research
Gulley et al. (1997) showed the same pattern earlier. Higher reinforcer rates made pigeons match the correct sample more often.
Thrailkill et al. (2018) adds a twist. They found rich schedules later cause bigger relapse after you stop reinforcement.
Johnson et al. (2009) extends the idea. Even lights that do not give food can make behavior resist change if they appear during rich schedules.
Gentry et al. (1980) seems to disagree. They said reinforcement rate only changes bias, not true discrimination. The key difference: D used signal-detection tasks with fixed stimuli, while A varied the actual schedule richness.
Why it matters
When you teach a new skill, thicker reinforcement not only builds speed but also locks in accuracy.
Plan for the downside: rich schedules can make the skill harder to fade later.
Start thick, then thin quickly to balance strength and flexibility.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Run your next discrimination lesson on a dense schedule, then thin to lean within the same session to test retention.
02At a glance
03Original abstract
Pigeons were trained on multiple schedules in which responding on a center key produced matching-to-sample trials according to the same variable-interval 30-s schedules in both components. Matching trials consisted of a vertical or tilted line sample on the center key followed by vertical and tilted comparisons on the side keys. Correct responses to comparison stimuli were reinforced with probability .80 in the rich component and .20 in the lean component. Baseline response rates and matching accuracies generally were higher in the rich component, consistent with previous research. When performance was disrupted by prefeeding, response-independent food during intercomponent intervals, intrusion of a delay between sample and comparison stimuli, or extinction, both response rates and matching accuracies generally decreased. Proportions of baseline response rate were greater in the rich component for all disrupters except delay, which had relatively small and inconsistent effects on response rate. By contrast, delay had large and consistent effects on matching accuracy, and proportions of baseline matching accuracy were greater in the rich component for all four disrupters. The dissociation of response rate and accuracy with delay reflects the localized impact of delay on matching performance. The similarity of the data for response rate and accuracy with prefeeding, response-independent food, and extinction shows that matching performance, like response rate, is more resistant to change in a rich than in a lean component. This result extends resistance to change analyses from the frequency of response emission to the degree of stimulus control, and suggests that the strength of discriminating, like the strength of responding, is positively related to rate of reinforcement.
Journal of the experimental analysis of behavior, 2003 · doi:10.1901/jeab.2003.79-307