Hacker News new | ask | show | jobs
by thelogos 3373 days ago
It is not about how big the brain is. It is about the number of neurons, latency, signal reliability and number of connections.

Voltage-gated ion channels are non-deterministic. Meaning they don't always open (or not open) when they should. In order to stuff more neurons into the same volume of space, you have to shrink them. The problem is, those ion-channels become more and more unreliable as the size decreases. I would argue that they're already too unreliable in many humans.

Second problem, as the size of the axon and myelin sheath decrease, signal reliability and latency will suffer. Yes, the current can die out part-way to its destination. As the brain is less globally connected due to the sheer lack of space, poor signal reliability and increased latency, it will begin to favor local connections over global ones. In other words, specialization and usage of signal superhighways to compensate, just like a crowded city. The problem with a crowded city is, even with great public transport, many people never leave their neighborhoods.

So what to do about it? You can leave neurons the same size and make more room instead of trying to shrink them.

First problem with this, difficulty of childbirth due to skull size. Second, increased development time, it's already too long as it is. Third, latency and signal reliability will still suffer due to increased distance. Fourth, increased use of resource. You also need to support those neurons and that support system will eat up more and more space.

If you try to blow up the size of the axon and myelin sheath to fix the latency and reliability problems, it will eat up even more space. In other words, less room for neurons and you're back to square one. Another problem is, you need a bigger body to support that huge brain. More neurons will be dedicated to processing touch instead of higher-level abstract thoughts.

One last thing you can try is, decrease body size, increase brain volume slightly. The lower level of violence, abundance of food, modern healthcare (c-section) and longer lifespan (more time to mature) in modern human societies allow us to do this already. Dedicate more resource to higher levels of the brain associated with abstract thoughts, planning, reasoning, etc. Over-myelinate those areas to increase signal speed and reliability.

At the end of the day, there's not much more mother nature can do without deep structural and material change. Reengineer the myelin materials to increase their insulating property and decrease the size. Make the ion-channels more reliable so you can shrink neurons even more, although you still have to worry about quantum tunneling. Or better yet, do away with ions completely and switch to photonic computing.

3 comments

I dont think neurons have a reliability problem. Most synapses use ligand-gated channels, and neuronal homeostasis makes sure neurons do fire within their physiological range. Stochastic channel opening has not been found to have conclusive effects in reliability. Cortical neurons live near the surface and the brain solves the surface area problem by creating folds - the bulk of the brain volume is axons which form the white matter. If there's a tradeoff that evolution had to make, i speculate it would be between head size and vaginal opening size.
I wonder if there will be a human-computer brain interface so that we can augment our cognitive powers.
There's already one, actually just appeared :) https://en.wikipedia.org/wiki/Neuralink
He asked if there was an interface, not a startup that claims to be developing such an interface.
Upvoted, asuming you know what your talking about. I hope you do ...

Just one thing, photonic computing, is there more to it? I mean are there ideas how that might actually work?