Hacker News Reader

Comment by Waterluvian | Hacker News Reader

Tangentially related to this is when I installed a patio in my yard two summers ago. Okay, let me explain:

I had a sugar maple about 3 feet away from where I had to dig the foundation for the patio. An arborist said that it’s a bit close but the maple is strong and it should recover fine.

I cut out a bunch of roots and within a month the exact sections of the canopy corresponding to the roots I cut went brown.

It was incredible to me how the roots relate directly to the canopy. I always kinda thought it was one big circulatory system, where everything supports everything. I expected the whole tree to struggle a bit.

Next season the tree was fully recovered.

The neurological system seems to work the same way? It’s a directed graph where one root supports a very specific set of branches? I guess the circulatory system is like that too if you separate the two sets. Nature doesn’t really like cyclic graphs, does it?

View on HN

> I cut out a bunch of roots and within a month the exact sections of the canopy corresponding to the roots I cut went brown.

Root systems in general (for trees at least) mirror what's happening above ground. Pruning trees is a beneficial intervention, as it causes the corresponding roots to die and decompose. This not only makes precious biomatter available for recycling by the microorganisms in the soil, but also releases chemical signals that cause the tree itself and neighbouring plants to send out new growth.

When I worked for the park service, my botanist boss told me that the reason junipers and that sort of scrubby, stress-tolerant tree have twisty trunks is so that if they lose a section of roots they don't lose a sector of sunlight in the canopy. Because sunlight can be focused on one side of a tree (especially if growing on a steep slope), a twisting trunk can save a tree in the case of root damage on one side.

> Nature doesn’t really like cyclic graphs, does it?

I had a course in computational neuroscience as part of my bachelor's and one of the things that we covered was that the timing of fires is important, in that depending on how soon before or after a neighbouring neuron fires, the connection may be weakened or strengthened. This is called Spike-timing-dependent plasticity:

> Under the STDP process, if an input spike to a neuron tends, on average, to occur immediately before that neuron's output spike, then that particular input is made somewhat stronger. If an input spike tends, on average, to occur immediately after an output spike, then that particular input is made somewhat weaker hence: "spike-timing-dependent plasticity"

From [1].

The implication of that, I believe, is that it prevents short cyclic graphs, for the sole reason of avoiding feedback loops that can cause the brain to go haywire (lol) due to the feedback loop. It sounds like an evolutionary adaptation to prevent short-circuiting.

From Hebbian Learning, we have that the cycles would become easier to trigger, meaning that it is a feedback loop that increases efficiency, however, without a mechanism to prevent this cyclical feedback loop, the brain could be filled with cycles that eventually turn to just rings, which is probably not a desirable property.

If anyone knows more about this please tell me. If it's a new idea, please remember to add my name :')

[1] https://en.wikipedia.org/wiki/Spike-timing-dependent_plastic...

just keep inmind, this, and other phenomenon dont happen in all neurons, or in any particular neurons 100% of the time.

neurons change functional, and structural state, depending on past events [hysterisis] and will shut down/modify state activities depending on feedback from post synapic neurons.

also neurons will get tired and handoff activity to similar neurons in a cohort.

some old classmates of mine published a paper related to feedback loops:

https://www.frontiersin.org/articles/10.3389/fncom.2011.0002...

Its very much not a directed graph. We do have brain loops. That's where brain waves come from, circular paths of neurons activating themselves in a circular firing squad.

Directed graphs can be cyclic.

They probably both meant directed-acyclic-graph?

> I guess the circulatory system is like that too...

I have some news for you: https://en.wikipedia.org/wiki/Circle_of_Willis

i think of it like a plinko ball setup, in analog.

neural convergence, and divergence, produce logic arrays that integrate many parameters to one integrated decision, vice versa

If your model includes a tiny Bob Barker commentating on the transmission of neural signals, I’m on board.

Reply on news.ycombinator.com