Qin and Wheeler (2006) show that C. elegans can learn to travel through a T maze, and that their latency to reach the bacterial reward diminishes as the trial number increases. These nematodes only have 302 neurons, so any example of them learning associatively always fascinates me. They also have data showing that cat-2 dopamine deficient mutants consistently display less exploratory behavior than wildtype worms, which makes sense because dopaminergic neurons are often linked to reward-seeking behavior. Here is their description, a short video (you get to see a worm in “traveling mode”) and a link to the actual paper.
Qin J, Wheeler AR. 2006 Maze exploration and learning in C. elegans. DOI: 10.1039/b613414a.