Animals constantly have to adapt to varying environmental conditions, explore new situations and figure out new strategies to catch prey or avoid predators. On the other hand, they need to be able to behave consistently in a largely deterministic environment. Brains reflect the complex mixture of chance and necessity in the environment in thier structure and function.
One great example of this was just explained to me here on a poster at the annual meeting of the Society for Neuroscience ( SfN). The poster was entitled "Distinct inhibitory neurons exert temporally specific control over activity of a motoneuron receiving concurrent excitation and inhibition" and came out of the lab of Klaude Weiss at Mount Sinai.
The researchers work on the neurophysiological model system Aplysia. They study how the neurons in the buccal ganglia control the movement of the radula (a tongue-like organ) during feeding. Other scientists in the lab have shown previously that Aplysia feeding behavior is highly variable, probably to be able to adapt to a large variety of different food sources. What was unknown until now is how the neurons in buccal ganglia which control feeding behavior generate this variability. This is what this poster was about.
They showed that the seemingly simple, two-phase behavior of the radula coming out of the jaws (protraction) in the open state, and pulling food into the mouth (retraction) in the closed state is subdivided neuronally into smaller chunks of behavior. For instance, using electrodes in multiple neurons at once, they figured out that the closure state of the radula during retraction consists of an early phase and a late phase. The motor neuron which controls the closure of the radula is identified and is called B8. By recording the activity of various identified neurons in the buccal ganglia, as well as by de- or hyperpolarizing them in a quiescent preparation or during feeding motor programs actively generated by the ganglia in the dish, Sasaki et al. teased apart how excitatory and inhibitory input converges on B8 during retraction. The excitatory input makes B8 fire and the radula closes. Two neurons provide inhibitory input for B8. Depending on how rapid they fire and when, B8 fires less strongly and the radula closes less strongly or stays open. One of these neurons, called B70, fires in the early phase of retraction and the other, B4/5, fires late during retraction. On the poster, the researchers show how the delicate balance between tonic excitation and early/late inhibition by these two neurons allows the animal to close the radula more or less, earlier or later, depending on what is the most successful strategy with whatever food the animal is currently attempting to feed on.
This delicate balance is a great example of how brains keep their behavior flexible and variable: by keeping the behavior always on the edge of one or another state (open or closed in the case of the radula), it only takes the smallest variations in any neural activity and it may have large effects on the behavior. The researchers also showed that there are peptidergic modulations going on in this balance, allowing for longer-term fine-tuning of the balance.
I find these results very exciting, because the provide a compelling example of a potential mechanism for how behavioral variability is generated, which we have studied in the fruit fly Drosophila. It's exactly this 'critical' state in a labile balance which may serve as one of the mechanisms which generate the sort of variability we have also observed in Drosophila.This research has just been published in the Journal of Neuroscience, so you can have a lok at all the exciting results for yourself:

Sasaki, K., Brezina, V., Weiss, K., & Jing, J. (2009). Distinct Inhibitory Neurons Exert Temporally Specific Control over Activity of a Motoneuron Receiving Concurrent Excitation and Inhibition Journal of Neuroscience, 29 (38), 11732-11744 DOI: 10.1523/JNEUROSCI.3051-09.2009