Fast and complex neural computation at the basis of cognition

Abstract

In chapter 2, we recorded AP shapes and underlying Na+ and K+ currents in human cortical pyramidal neurons and compared them to rodents. We found that not only the AP shape, but also the underlying currents have remarkable stability, even when firing hundreds of consecutive APs at frequencies of 60 Hz. We then argued that biophysical adaptations in voltage-gated sodium and potassium channels must be responsible, and tested this by characterizing the voltage-dependent properties of the Na+ and K+ currents. We find that human Na+ and K+ currents had different biophysical features than their mouse counterparts. Then we used in silico models and found that indeed, the observed gating properties result in more stable APs. Furthermore, the models with human-like channel biophysics were able to reliably encode a larger range of inputs than models with rodent-like biophysics. Based on this we concluded that adaptations in the biophysical gating of Na+ and K+ channels contribute to the computational properties of human neurons. Fast Spiking interneurons (FSINs) are specialized in fast input-to-output conversion. The whole neuron is filled with adaptations that support this fast function. Therefore, this is the perfect neuron to study when interested in the speed limits of neural processing. In the human cortex, neurons are spaced further apart to make space for all the elaborate dendritic and axonal trees, which are much more interconnected than in rodents. This could potentially slow down neural signaling, as signals take time to travel and become weaker and slower over distance. Yet neural signaling seems not to be slow in human neurons. In this chapter, we investigated whether and how human FSINs can retain their fast function. We first characterized the dendritic morphology of these neurons and found that indeed, their dendritic paths are much longer. However, when we made paired recordings we did not find a slowdown of input and output, indicating that some compensatory mechanism must be at play. This could not be explained by sodium channel gating, as we find similar sodium current properties in human and mouse FSINs. Using biophysical modeling we find that a combination of enlarged synaptic inputs, reduced dendritic complexity and fast outputs in human FSINs are responsible for conserving a fast signaling speed even over large distances. In this chapter, we take the concept of cellular properties as important determinant for computational power of neurons and see how that relates to the function of the brain as a whole. Although we know that genetics and thickness of the cortex play an important role in IQ differences between individual humans, we do not know it’s structural and functional basis. In this first-of-its-kind study, we obtained IQ scores as well as cellular measurements from individuals. We find that high IQ score does not only go together with large cortical thickness of the temporal cortex, but also with increased dendritic complexity and fast AP shapes. This provided the first evidence that human intelligence is related to complexity and speed of information exchange between individual neurons. Strikingly, the increase in cortical thickness with IQ was found to be purely specific to layers 2 and 3 of the cortex. Apparently, these cortical layers, which serve as important computational layers between input and output layers, are pivotal for intelligence. We then confirmed that high IQ relates to lower cell density, larger cells, more complex dendrites and faster AP signaling all specifically to the pyramidal neurons in layers 2 and 3. Furthermore, we find that verbal intelligence correlates with all these microstructural and cellular features, but only on left, not right, temporal cortex.