Evolutionary optimization of neuronal network dynamics for recipient-optimized cross-area communication

Moritz Helias – FZ Jülich
Simon Musall – FZ Jülich

Evolution constantly optimizes neuronal information processing to promote animal survival within its habitat. Such optimization of neuronal networks faces severe challenges: individual cells have limited processing capabilities, requiring interconnected areas to carry out higher functions. Moreover, evolutionary optimization must support the integration of new areas into an existing system of already functional and previously optimized areas. A fundamental question is therefore how evolution has shaped the communication between areas; in particular, whether communication patterns differ between areas mostly due to different evolutionary ages (subcortical versus neocortical), promoting the view that each area has undergone a separate evolutionary optimization; or whether communication patterns are predominantly shaped by functional demands, suggesting a co-optimization of the joint system as a whole to optimally perform a set of tasks. Our proposal follows the latter hypothesis. Concretely, we hypothesize that evolution has optimized the communication patterns between areas so that the sending areas' signals optimally excite the receiving area. This hypothesis derives from two so far independent streams of work: the first has shown that operation of neuronal networks close to a dynamical critical point offers beneficial properties in terms of memory and signal processing due to the presence of a rich set of neuronal modes that span a diverse set of time scales. The second has established that communication between areas takes place within low-dimensional communication sub-spaces, which are putatively implemented by specialized populations of long-range projection neurons. We plan to employ a combination of network theory and simulation with the analysis of a large dataset of spiking neural data from different cortical regions of primates, as well as cortical and subcortical recordings in mice. First, we will characterize the neuronal modes and associated time scales in different areas of the macaque and mouse neocortex as well as the mouse thalamus. This will reveal differences in the functional operation of neural networks in evolutionary distinct species and across areas of different evolutionary ages. Second, we will theoretically investigate the consequences of our main hypothesis, the presence of optimized communication patterns between interconnected areas that preferentially couple modes with similar time scales to facilitate information transfer. Lastly, we will experimentally test our theoretical predictions by regression analyses of simultaneous recordings of cortical and subcortical regions in mice, while using retrograde labeling and optogenetic stimulation to isolate the activity of specific projection neurons. Recordings will be done under spontaneous conditions and during a tactile discrimination task that requires different time scales in terms of working memory and evidence accumulation.