This summary of the meeting was created from the audio data using OpenAI's Whisper and ChatGPT by Ken Nakae. Due to the accuracy of the audio data readout and the limitation of the ChatGPT summary to 8196 tokens, this summary are not necessarily accurate at this time (and have been slightly modified by hand). Please refer to this as a recognition of those assumptions.

In this presentation, the speaker introduces two different ways to think about and represent brain structure and dynamics, focusing on the whole brain scale.

The first approach is to treat the brain as a discrete system, turning each part of the brain into a node and connecting them through structural connections to form a network. This approach makes sense as it represents the individual neurons with axons connecting to other neurons, but it has its drawbacks, such as losing the physical embedding of the neurons’ locations and making boundaries between different brain areas.

The second approach is a continuous one, representing the brain as a spatially continuous system with excitatory and inhibitory populations and their approximate interaction between them. This approach, although making several strong assumptions, simplifies the geometry and aims to find simple mechanisms to explain the brain’s dynamics. The prediction of this set of assumptions is that the dynamics play out like a damped wave equation.

The speaker argues that the second, more continuous approach is surprisingly good in representing brain dynamics and encourages researchers to consider this approach when studying the brain.When the brain experiences high activity in one area, it excites the adjacent parts, and they pass on the excitation to their neighbors. This sets off complex dynamics throughout the brain, which can be likened to how physicists study cross-scale models. At the micro-scale, interactions between local atoms, electron spins, and various other forces are intricate and complicated. Surprisingly, when these interactions are zoomed out, emergent macro-scale behavior tends to be governed by simpler laws, despite micro-scale complexity.

The same optimism can be applied to studying the brain, where neuroscientists and biologists pay great attention to details such as synaptic transmission and neurotransmitters. Physicists propose that by looking at population activity, the brain can be modeled with much simpler rules. Different modeling approaches must be applied at different scales, but at longer lengths, more straightforward assumptions can capture emergent dynamics.

An isotropic distance-dependent connectivity rule assumption allows researchers to visualize brain dynamics on a cortical surface and represents more traditional methods of looking at structural connectivity. A recent study provided evidence that this simple assumption of damped wave propagation on the cortical surface could predict functional connectivity better than previous models, despite only having one free parameter. This suggests that it’s possible to predict functional connectivity without any knowledge of structural connectivity.To address the question of how we go from structural connectivity to functional connectivity, a simple wave propagation mechanism is used to get straight to functional connectivity without knowing any of the long-range connectome projection. There are various parameters in the field, such as the gamma parameter, which is the strength of damping, and the global coupling parameter in the mass model. This model shows that with the geometry alone, we can capture much of the functional connectivity observed in the brain.

Various studies have used high-resolution functional connectivity data to capture the variance in subcortical structures, but this approach can be simplified by just using the geometry of the brain. This simplified method is capable of reproducing similar results as more complex approaches, emphasizing that simpler, more parsimonious accounts can often be sufficient.

The brain’s hierarchical structure is also correlated with its geometric positioning, suggesting that the brain has evolved to optimally arrange areas of differing functional specialization given the physical constraints on the brain. Furthermore, the geometric model can reconstruct a variety of functional MRI task activations, performing better than alternative models based on connectome information or assuming an exponential distance rule. This further supports the effectiveness of the simplified wave model approach.

However, it is important to note that the relationship between these simplified models and neural activity still needs to be explored. With new brain stimulation experiments and faster recordings, it is possible that more accurate models could be developed, potentially integrating both the simplified approach and neural activity for an even more comprehensive understanding of brain function. Simple isotropic connectivity and propagation mechanism can explain a surprising amount of phenomena measured from human FMRI. Curvature plays a role in the cortical geometry, but it locks modes into certain orientations relative to sulci and gyri. The relationships between the eigenmode connectivity in EEG and FMRI are difficult to determine without further study. The spatial constraints on the brain are often neglected in analysis approaches due to assumptions and data limitations. Evaluating feedback activity would require a more detailed model that incorporates various brain layers and their properties. Connectivity between regions in the brain, such as the distance dependence of connection probability, indicates the importance of considering physical constraints in neural models and analyses.The spatial embedding of brain physiology is essential, with organisms having a strong spatial constraint on their wiring. In the mouse connectome, connection probability decreases exponentially as a function of the distance between brain areas, and similar patterns emerge with gene expression.

Nearby pairs of brain regions have more similar gene expression patterns, which are influenced by factors such as cell types, dendrite properties, and neurotransmitter expression. Molecular gradients are set early in development, and brain structure continues to spatially expand over time. The spatial scale of connectivity also shows a proportional relationship to brain size. Various properties of brain structure, such as MRI measurements, cortical architecture, and gene expression patterns, are all influenced by spatial embedding.Various brain-related structures can be grouped based on their value ranges, indicating that these properties are not independent but vary together. This variation, referred to as the Mesulam (1998) picture of sensation to cognition, presents a functional hierarchical picture where sensory areas take in information from the world and integrate it in prefrontal areas. These properties co-vary, which should be considered when analyzing the relationship between pairs of measurements.

There are commonalities between mouse and human brains, with gradients of variation in key interneuron markers related to T1 and T2 MRI measurements. These patterns persist across scales and species, which could be evidence that evolution may be selecting for these neural architectural level properties. However, when processing gene expression data, there are many choices available, and these can lead to different or even opposite results. It is essential to be cautious when working with open data and understand the subtleties of the data being used. Tackling these challenges requires careful consideration of available choices and recommended methods. This model focuses on the impact of distance in connectivity, treating the cortex as a continuous sheet and the subcortex as 3D volumes. Both long-range connections and local interactions are known to play roles in brain activity, and considering time scales and length scales can provide a more detailed understanding of these mechanisms. The isotropic connectivity model assumes that the strength of connectivity between brain points A and B depends only on their distance apart and decreases exponentially with increasing distance. While it still involves connectivity, it does not specify direction and only considers distance decay, similar to how a spring, violin string, or water wave have local effects on neighboring areas.In this study, we discussed the similarities between evoked and spontaneous brain activity and how they may be meaningful. One possibility is that the brain’s geometry strongly constrains its activity, leading to similarities in activity patterns in both cases. A physical analogy could be how raindrops on a pond create patterns that may resemble those from a thrown rock, once the energy dissipates. There is correlational evidence for specialization in mouse and human brains, but further studies are needed.