Mammalian cortex has long been one of the most widely studied systems in neuroscience, dating back to the pioneering work of Santiago Ramon y Cajal in the late 19^th century. The cortex is much larger in primates than other mammals, and is thought to be responsible for the advanced cognitive abilities of humans. Today, models of cortical connections and computations form the basis for some of the most powerful deep learning paradigms. However, despite this success, there is still much that is unknown about how cortex functions. One feature of cortex that has recently been discovered is that neurons that appear to be similar to each other can have very different baseline activity levels: some neurons are 100x more active than their neighbors. We don’t know how neurons that are otherwise highly similar in shape and genetic makeup can maintain such different activity levels, or if the neurons with high and low activity levels have different functions in the brain. These neurons are otherwise so similar to each other that it is difficult to tell them apart without recording their activity directly, and current techniques for recording the activity of many neurons simultaneously in live animals do not allow us to later re-identify them for further study.

In a paper recently published in Neuron, the Turrigiano lab, led by postdoctoral researcher Nick Trojanowski, reported a new approach for permanently labeling high and low activity neurons in live animals, and then determining what makes them different. To do this they used a fluorescent protein called CaMPARI2 that changes from green to red as activity increases, but only when exposed to UV light. By shining UV light into the brain, they caused neurons with high activity to turn red, while neurons with low activity remained green. This procedure allowed them to run a series of tests on high and low activity neurons to identify differences between them. They found that high activity neurons would intrinsically generate more activity than low activity neurons when presented with the same stimulus. These high activity neurons also receive more excitatory input specifically from nearby neurons of the same type. Surprisingly, however, they found that the total amount of excitatory and inhibitory input that high and low activity neurons received from other neurons was not a major factor in determining their activity levels. Together, these results tell us that the differences in activity between neurons are due to intrinsic differences, as well as their pattern of connectivity to their nearby partners. This has deep implications for how the networks that underlie cortical computations are built and maintained.

With these tools in hand, it is now possible to further explore the differences between high and low activity neurons. Do these neurons serve different functions? Are the baseline activity levels specified from birth? How do these activity levels affect the mechanisms of plasticity that are responsible for learning and memory? The recently published results represent just the tip of the iceberg of information that can be learned with this new technique, in the mammalian cortex as well as other brain regions.