In her bid to gain insight into the complex relationship between learning and behaviour, Cornelia Bargmann opted to keep things simple: she studied the nematode worm Caenorhabditis elegans (see page 179).

Earlier research had shown that these worms can learn, but Bargmann, based at Rockefeller University in New York, wanted to know how and why. She knew that worms fed in an environment kept at a particular temperature associated that temperature with food. If they were then placed in another environment, they would head for an area at the same temperature as where they were fed.

To expand the boundaries of what could be studied in the organism, Bargmann decided she needed to “think like a worm”. That led her to consider a worm's natural environment: soil. This is teeming with thousands of organisms — mostly bacteria. Some of those bacteria are beneficial to the worms, others can make them sick, or even kill them. “For the worm, this would be something worth learning about,” thought Bargmann.

To find out more, she looked at ‘conditioning’. Common in many organisms, this form of learning occurs when, for example, an organism eats something with a novel flavour and gets sick shortly afterwards. As a result, the organism forms a strong memory of that flavour and rejects it in the future. The effect can be so strong that one bad taste can be enough to turn an organism away from things associated with that experience.

Credit: K. CHERNUSH/HHMI

Bargmann's postdoc, Yun Zhang, did some simple experiments exposing worms to a ‘good’ strain of bacteria and a ‘bad’ one. Very quickly, the worms showed some strong preferences. But Bargmann was worried that giving the worms two choices was like giving someone a true/false test; guessing would give a correct answer half the time. She wanted more definitive results, but wasn't quite sure how to produce them.

Fortunately, she had just begun an interdisciplinary project: Bargmann had invited a few engineering postdocs to work with her, although she wasn't yet sure how they would fit in. One of them, Hang Lu, who did microfluidics at the Massachusetts Institute of Technology, took an interest in the worm project.

Lu's work had focused on how single molecules interact in a circuit-like system, and she thought that scaling this idea up could help the group understand the worms' learning behaviour. “Basically, she made a worm maze,” Bargmann says.

Lu built a maze to provide statistically significant results. It had eight arms, with four types of bacteria — two malevolent and two benign. One of each kind was a type the worms had been exposed to before and one of each type was new.

The worms' behaviour was very distinctive. “They learned to like the good bacteria,” Bargmann says. “But more clearly they avoided the bad bacteria.” The next step brought Bargmann to more familiar ground: linking the learning to specific genes and cells. Bargmann says that she now wants to find out how the worms make the associations with good and bad bacteria. This will take her into the world of neuroscience — and no doubt require another interdisciplinary effort.