Delayed matching-to-sample performance of hens: Effects of sample duration and response requirements during the sample.

Researchers worked with hens on a delayed matching-to-sample task. The birds had to peck a sample key, wait 1.5 s, then pick the matching key.

The team changed two things each session: how long the sample stayed on (2–10 s) and how many pecks the hen had to give (0–10 pecks).

They tracked how often each bird picked the correct match after the short delay.

Accuracy moved up or down in a clear pattern as sample time and required pecks changed.

The data showed you can tune a DMTS task by tweaking these two knobs.

Thompson et al. (1974) saw the same duration effect with a dolphin. When the sample shrank below 0.2 s, the animal’s score crashed to chance, echoing the hen data.

Madden et al. (2003) kept the DMTS frame but added overt naming for humans. Requiring spoken names lifted accuracy, showing that what you do during the sample matters as much as how long it lasts.

Together the papers say: give enough sample time AND ask for active behavior—pecks for hens, naming for people—to get the best delayed matching.

When you set up a delayed matching program, pick a sample length that is neither too short nor too long, and add a clear response requirement—pecks, names, or clicks. Start with 3–5 s and require at least one overt action. Watch accuracy; if it drops, lengthen the sample or strengthen the response rule. This quick tune-up works for kids, staff training, or client assessments.