Final Report Summary - ATTENTIONCIRCUITS (Modulation of neocortical microcircuits for attention)
One of the most fascinating and essential capacities of the brain is its enormous functional flexibility, which for instance enables us to focus our entire attention on a text while at the same time still being able to rapidly react to relevant changes in our environment, such as the ringing of our phone or a fire alarm going off. Such adaptability is critical for survival in an unpredictable world, and the brain by far surpasses even the latest artificial intelligence algorithms in this respect. To identify the mechanisms that enable this flexibility, our research focuses on neocortex, the largest and most complex brain area which expanded and differentiated substantially during mammalian evolution, mediates many of the capacities that distinguish humans from their closest relatives, and plays a central role in many psychiatric disorders. However, the brain functions for which neocortex has evolved are not well understood. Our work has now revealed that neocortical processing is indispensable for behavior in response to complex, naturalistic sensory information but not to the reductionist, simple stimuli that are typically used in this research. These results indicate that only naturalistic information is sufficiently complex to necessitate neocortical processing. Given that learning paradigms using complex stimuli require very long training times, we have developed a fully-automated, naturalistic training system for these experiments. Moreover, these insights provide a robust basis for identifying the mechanisms by which neocortex mediates behavioral flexibility. Neocortical circuits are composed of excitatory principal neurons, which convey information to distant areas and ultimately produce all brain functions, and interneurons that inhibit, regulate and control these principal cells. The fact that excitation is closely matched to and often almost cancelled out by inhibition is at first sight counterintuitive, and in addition claims a major part of the brain’s energy budget. However, this organization also endows the brain with a number of distinct advantages, and our work suggests that computational flexibility is one of these benefits: The tight relationship between excitation and inhibition can be transiently broken when new information needs to be stored in the network. Thus, when an animal forms new associations between different external stimuli, inhibition is briefly reduced which in turn strongly boosts activity in the surrounding principal neurons and likely gates induction of synaptic plasticity that underlies the memory trace. This so called disinhibition is a conserved circuit mechanism for learning and memory that has by now been observed in a variety of behavioral tasks and brain areas. However, inhibitory interneurons are a heterogeneous class of cells that falls apart into many diverse types with dedicated morphology, physiology and function in the network. Our work has now defined the first genetic marker for interneurons in the outermost layer of neocortex, a previously enigmatic cell type. This marker is expressed selectively in these layer 1 interneurons in both mouse and human neocortex, opening up the possibility for translational work. Furthermore, these cells are under dominant control by similar neuromodulatory systems in both species, which likely endows them with functional flexibility. In addition to this input, these layer 1 interneurons receive information from a strikingly large number of distant brain areas, as well as from local inhibitory cells, suggesting that they may constitute flexibility hubs in the neocortical network. In line with this hypothesis, our results demonstrate that these layer 1 interneurons display remarkably flexible and experience-dependent information processing during different behavioral functions. In conclusion, work funded by this ERC Starting Grant has identified inhibition and in particular interneurons in layer 1 as exquisitely adaptive processing units in neocortex. In addition to informing our understanding of how these circuits produce perception, learning and memory, our data also have important implications for the design of artificial intelligence algorithms, and for translation to the healthy and diseased human brain.