BRAIN CELLS' BRATS : ' UNUSUAL ' GLOBAL ESSAY

 

THERE is compelling evidence that misfunctioning astrocytes play a role in mood disorders such as depression and anxiety and neurodegenerative diseases like Alzheimer's.

Most strikingly, in 2021 Liam O'Leary at McGill University in Montreal reported that the brains of depressed suicide victims had markedly reduced densities of astrocytes, compared with healthy brains, in parts of the prefrontal cortex [ the brain's executive ], the caudate nucleus [ which helps control goal-directed behaviour ] and the thalamus [ which passes sensory information to the cortex.]

In the past decade, several lines of evidence, including brain-imaging, post mortem and genetic studies, have pointed to dysfunctional oligodendrocytes as the cause of psychosis in conditions like multiple sclerosis, bipolar disorder and schizophrenia.

Such dysfunction disturbs the myelin on axons, disrupting the timing of their signals. The hypothesis is that this results in hallucinations - imaginary sights and sounds - that are the defining feature of psychosis.

In 2017 Ishizuka Kanako of the Nagoya Institute of Technology in Japan, found a link between an increased risk of autism and the presence of a pair of genetic variants known to disrupt, in microglia, the expression of a protein called CX3CR1.

And in 2020 Xu Zhixiang of Scripps Research, in San Diego, showed a range of microglial protein-synthesis problems cause autism-like symptoms in mice.

Current thinking is that misfiring microglia in people with autism fail to prune synapses thoroughly enough during brain development, resulting in overconnected brains with heightened sensitivity to stimuli, both sensory and emotional.

Moreover, the effect Dr Xu found disproportionately affects male mice - a bias that, perhaps not coincidentally, is also a feature of autism in human beings.

And it is not only psychiatrists who are inspired by the newly discovered roles of glia. Computer scientists are getting in on the act, too.

Artificial neural networks are based on an early model of how neurons work - and, though subsequent has shown this was simplistic, these networks' organisation into interconnected layers of neuron analogues does reflect that of the central cortex.

Understandably, therefore, some computer scientists have tried adding artificial glia to networks to see if their performance improves.

' Transistorised, at half the price ' : It does. Several groups have discovered independently that getting rid of rarely used synapses, the job of microglia, helps artificial neural networks to encode new information and store memories. Coming up with ways to make neural networks sparser is now an important field.

Artificial astrocytes are also being investigated, along with the idea of artificial neuron-gila networks [ ANGNS ].

These imitate tripartite synapses by using astrocyte analogues to strengthen and weaken synapses in response to how the rate at which those synapses fire changes over the course of time.

When tested alongside conventional networks, ANGNS consistently outperforms them. As with many things in human engineering, it seems that nature got their first.

This Unusual Essay Publishing continues. The World Students Society thanks The Economist.