Determinants of contrast in the signal-key procedure: Evidence against additivity theory.
Contrast only appears when you remove accidental reinforcement and give clear, brief cues.
01Research in Context
What this study did
Pigeons pecked a key that sometimes changed color. The color told them if food was coming soon or later.
The team tweaked the set-up step by step. They removed accidental food deliveries and kept the color change separate from the food rate change.
They wanted to see when contrast—extra pecking—would truly appear.
What they found
Contrast only showed up after the clean-up steps. Before that, the birds pecked for the wrong reasons.
Once the set-up was tight, extra pecking followed the color cue, not the food rate itself.
How this fits with other research
Dews (1978) saw strong contrast using the same key-color trick. The difference: B kept the accidental food. Parsons et al. (1981) removed it and the contrast shrank. The papers seem to fight, but they really show contrast needs clean cues.
Reed (1991) later found contrast only in the first few seconds of a schedule. That lines up: short, clear cues give clean data.
Killeen (2023) folded both findings into one math rule: richer schedules build momentum. The 1981 data now sit inside a bigger equation.
Why it matters
When you run multiple schedules, strip out accidental reinforcers. Use clear, brief cues and watch early responses. Clean data let you see true contrast and adjust reinforcement faster.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Check if your SD lights or sounds accidentally pair with extra reinforcers—remove any and watch response rates jump.
02At a glance
03Original abstract
Two experiments are reported that challenge the interpretation of previous results with the signal-key procedure, in which the discriminative stimuli are located on a response key different from the key associated with the operant response requirement. Experiment 1 replicated the procedure of Keller (1974), and found that contrast effects on the operant key occurred reliably for only one of four subjects. High rates to the signal key initially occurred for only one subject, but modifications of the procedure produced substantial rates to the signal key for all subjects. In all cases, however, signal-key behavior was greatly reduced by the addition of a changeover delay which prevented reinforcement within 2 seconds of the last peck to the signal key, suggesting that signal-key pecking was maintained primarily by adventitious reinforcement. Experiment 2 modified the signal-key procedure by using three response keys, so that the discriminative stimuli on the signal key controlled different responses during all phases of training. With this modification, reliable contrast effects on the operant key occurred for all subjects, suggesting that the failure to find contrast in previous studies has been due to the confounding of changes in the discrimination requirements with changes in relative rate of reinforcement. The results challenge the additivity theory of contrast, and suggest that "elicited" behavior plays a minor role, if any, in the determination of contrast effects in multiple schedules.
Journal of the experimental analysis of behavior, 1981 · doi:10.1901/jeab.1981.35-161