Training pouched rats to find people.

Five giant pouched rats learned to find people buried under rubble. Trainers used shaping and a clicker. The rats had to run to the person, pause, then return to the trainer for food.

Each trial offered three choices: a hidden human, a bag of clothes, or an empty bag. The study ran outdoors on a rubble field that looked like a real disaster site.

The rats picked the person on 83 out of every 100 trials. They chose the clothes bag only 37 times and the empty bag only 11 times.

Human scent gained strong control over the rats’ behavior. The animals rarely wasted time on objects that smelled only of fabric or dust.

Poling et al. (2011) taught the same species to find land mines. Both studies used click-and-food shaping, showing operant conditioning works across very different rescue tasks.

Ohan et al. (2015) also used pouched rats for mine detection, but on tilled soil. Together the three papers build a line of evidence that these rats can learn any scent-based target if trainers start with easy steps and slowly raise the difficulty.

Brown et al. (2025) warns that rare targets can hurt accuracy. In the human-search study the person target appeared on about one third of trials, so prevalence stayed moderate and the rats kept their edge.

You can shape non-human helpers when speed and weight matter. A rat weighs one pound yet crosses rubble that would tire a 70-pound dog. If you ever consult on disaster teams, ask if pouched rats are an option. Use the same shaping rules you already know: start with clear cues, reinforce every success, and test against decoys early. The rubble task is just another discrimination program.