Hacker News Reader

Comment by achileas | Hacker News Reader

That's oversimplified and usually how it's taught at the undergrad level. Neurons and especially the networks they form are more analogous to analog electronics, but both analogies end up failing at some point.

View on HN

IAMNANS (I am not a neuroscientist) but my very simple understanding is that to a reasonable approximation, inputs are analog (there’s a lot of nuance in if and how much signal makes it into a neuron) and outputs are digital (neuron fires or doesn’t), leading to all kinds of interesting circuitry.

From what I know, neurons typically click on or off, so they are digital in that way. Signal strength is implied by the frequency of those clicks, so that's analog. Transistors are analog electronics, but used to make digital switches so it's sort of the reverse.

I guess my point is that calling neurons analog or digital is not a very good analogy and should probably be avoided.

It's a mixed-signal, general-purpose ASIC done in wetware. Combines both digital and analog traits. Some of most interesting research I saw in hardware CompSci does the same. Never goes mainstream since industry prefers things that are easy to automate and scale up for obvious reasons.

But whether or not the incoming signals ever lead to a new action potential depends on so many factors (spacial, temporal, chemical, biological,... and it's all dynamic on many levels) that can all be called "analog" and are quite diverse that it's hard to call anything "digital". One action potential, that thing that brings to mind the "digital" analogy, really is a very small part in that system.

Transistors also shut on and off fully in, say, a switched mode power supply, or a class D amplifier, or any other pulse-width-modulation-like scenario. That is all considered analog.

The brain is analogous to an analog analogue of digital computing?

Reply on news.ycombinator.com