Association for Behavior Analysis International

When an organism is seen to behave in one way in the presence of one stimulus, and not in the presence of others, stimulus control is said to have taken place. To teach stimulus control, many trials of discrimination training must occur, and outcomes must be reliably measured. When humans are teaching these skills to animals, humans may not be able to themselves experience what the discriminative stimulus may actually be; humans cannot see scent, they cannot smell it at the same level as their subject, and they may not be able to tell when the animal is under the control of a particular odor. Testing these contingencies is a challenge when resources are slim, however, benefit exists for humans and non-human animals alike when these missions are undertaken.

Unlike light or sound, scent is a chemical, not a physical, phenomenon. Therefore research with this stimulus modality represents additional difficulties. APOPO, an NGO based in east Africa, has been training giant rats to sniff out landmines, TB and other scents of humanitarian interest for over a decade. Their TB operations have markedly increased case findings in collaborating clinics, but issues around stimuli confirmation render tightly controlled research difficult. Using a library of scents common in the fragrance industry, research was planned to replicate the TB-detecting training process, but with stimuli whose properties and status was known and could be manipulated to a high degree of accuracy. After successfully training 9 rats to indicate a target scent from 9 controls, the next step was to reduce accuracy to approximately 75%. This would mimic current TB-rat accuracy and allow measurement of the effect on accuracy of manipulating parameters such as reinforcement rate and target prevalence. After performance was still suspiciously (=98%) accurate with dilutions as low as 0.0000001%, this seemingly simple step evolved into a series of manipulations of sample preparation and session procedure in an attempt to demonstrate stimulus control. This presentation will detail these manipulations and the challenges, pitfalls and successes that can be encountered when conducting research in resource-poor areas.

The exquisite abilities of the dog's nose have been exploited by humans for centuries. From tracking prey with hominids thousands of years ago to tracking cancer cells and endangered species, the noses of dogs have served us well. However, since their sense of smell outdistances ours by many, many zeros, we often do not know exactly what we are teaching them when scent discrimination is the lesson. For homeless dogs in a shelter environment, the opportunity to learn scent discrimination can illuminate science and enrich an institutional environment. In this experiment, we compared two methods of teaching discrimination: in one group, the S+, sweet birch (betula lenta) was paired with food and hidden in 5" x5" x 2" cardboard boxes. The dog's indication was simply to paw at or try to get into the box; the food was then faded and pawing at a box with only birch in it was reinforced. In the other group, large amounts of the same odor were hidden in a box, and any interaction with that box was reinforced with food. Latency to interacting with a box for 5 seconds was measured and compared across conditions.