Additional-delay schedules: A continuum of temporal contingencies by varying food delay.
Extra delay only lengthens response pauses when the baseline food delay is short.
01Research in Context
What this study did
Manabe (1990) built a new lab schedule. After each peck, food arrived only after an extra delay.
The extra delay, called A, could be 0, 5, 10, 20, or 40 seconds. The regular food delay t was set at either 10 or 30 seconds.
Birds pecked a key under a variable-interval program. The team tracked how long each peck took to occur.
What they found
When t was 10 s, peck latencies grew steadily as A increased. The birds waited almost the full 10 s before pecking again.
When t was 30 s, the same A steps barely changed latency. The birds still pecked quickly.
Only the short t = 10 s condition let the extra delay control the pause length.
How this fits with other research
Rogers-Warren et al. (1976) saw response rates fall as delay grew. Manabe (1990) shows the same delay rule can lengthen pauses instead, depending on the baseline t value.
CHUNG (1965) found that longer delays push pigeons away from a key. Manabe (1990) tightens that idea: the push only works when the baseline delay is already short.
Schaal et al. (1990) added signals during delay and got faster pecking. Manabe (1990) removed signals and still mapped a clean latency curve, proving the delay itself drives the pause.
Why it matters
If you run DRL or delay-to-reinforcement programs, check the baseline wait time first. A small added pause can backfire when the learner already waits a long time. Start with short baseline intervals if you want extra delay to stretch pauses.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Set your DRL baseline at 10 s or less before you add any extra delay.
02At a glance
03Original abstract
Pigeons performed on discrete-trial, temporally defined schedules in which the food delay (D) was adjusted according to the latency of the key peck (X) and two schedule parameters (t and A). The schedule function was D = A(t - X), where D is the experienced delay between a response and a reinforcer. The schedule parameter t is the maximum value below which the present contingencies occur. A is the additional delay to reinforcement for each second the response latency is shorter than the t value. When A = 0 s, the schedule is a continuous reinforcement schedule with immediate reinforcement. When A = 1 s, the schedule is a conjunctive fixed-ratio 1 fixed-time t-s schedule. When A approaches infinity, the schedule becomes a differential reinforcement of long latency schedule. The latencies for subjects with t = 10 s and t = 30 s were observed with the present schedules having seven values for A between 0 s and 11 s. In addition, the latencies for subjects for which t = 30 s were observed at an A value of 31 s to 41 s. As the A value increased, the latencies approached the t value for subjects for which t = 10 s. The latencies for 30-s-t subjects did not approach t, even when the A value was 41 s. The latencies for 10-s-t subjects at 11-s A value were longer than those under yoked conditions having exactly the same delays/interreinforcement intervals. These results demonstrated a continuum of latency related to the schedule continuum (value of A) at a small t value.
Journal of the experimental analysis of behavior, 1990 · doi:10.1901/jeab.1990.54-85