New way to model how brain circuits recover from changes
Researchers used a multiscale approach to show that different brain circuits can look the same during steady activity but react differently to disturbances.
Scientists have developed a way to model how brain circuits function. This method helps explain how different pathways in the brain influence stability and recovery.
Modeling detailed neuronal networks is very difficult because they require high computational costs. This difficulty makes it hard to test many different biological hypotheses.
The researchers used a multiscale approach to solve this. They first calculated the responses of small groups of spiking neurons. They then reused these responses across a larger cortical circuit.
They used deep neural networks to approximate these local responses. This method achieved an estimated acceleration of the order for steady-state evaluation compared to spiking network simulation.
The study focused on the input layer of the macaque primary visual cortex (V1). Specifically, they looked at layer 4C (L4). This layer receives signals through the magnocellular pathway.
The model includes hypercolumns (HCs) that contain approximately 4000 cells. These cells respond to different orientations of visual stimuli. The model also includes orientation domains (ODs) where preferred angles change.
A nonequilibrium steady state (NESS) is a driven state where population statistics remain steady even though neurons continue to fire. The researchers wanted to see if activity patterns return to this state after a change.
The study found that different combinations of pathway strengths can produce similar responses to visual stimuli. However, these same combinations produce different recovery dynamics. This means steady responses and recovery from perturbations provide different constraints on possible circuits.
A perturbation is a small change or disturbance in brain activity. Recurrent interactions can temporarily increase some patterns of these disturbances. Some patterns may also decay slowly during the recovery process.
Key facts
- The new method is estimated to be times faster for steady-state evaluation than spiking network simulation.
- Different circuit pathways can produce identical steady visual responses but different recovery dynamics.
- The model simulates the layer 4C of the macaque primary visual cortex (V1).