Can someone explain how neurons work to a programmer? Are they

>>8639593
it's... fuking complicated, and not so well known.It's been a long time since my last biology class, but I'll try.

globally, a neuron is a transmission device. It recieves an electric impulse from one or more other neurons and transmit to one or more other neurons.
Two characteristics
-the neurone/neuron transmission (aka the synapse) is an activation treshold, which can vary between neurons.
-The signal may be amplified by the neuron.

no new neurons are created after adolescence
the creation of new neuronal connections I don't know or forgot.

If you study neural artificial intelligence in uni you have to take a course in biology specifically in neural networks in the brain. Cool stuff, might major in it myself

>>8639725
So, it's like a more-than logic gate, but the right side number is part of the synapse and also it's analog rather than binary?

Makes me wonder if that alone can be turing-complete.

>>8639923
the intensity that a synapse let through is a binary either the treshold is met, or nothing passes. However there is a frequency coding : the higher the concentration of neurotransmitters released at the synapse, the higher the frequency of the signal in the recieving neurone, and thus the higher the concentration of neuro transmitters this signal will release when arriving at thenext synapse.

Also, the neurons are one way only : there is a recieving part and a transmitting part, and the signal can't travel backward.
You also have to consider the system isn't linear : one neuron can recieve signal from different neurons atthe same time, which will add up into one transmitted signal, which can be transmitted to more than one neuron.

Once again, this is a very simplified model and can't actually fully picture the complexity and diversity of a brain.

>>8639923
Usually if you have a programmable system you have to try to make it not Turing complete.

Here's a better metaphor for you OP. You have an object in a program in an OOP paradigm. It receives input from other objects. The object has an internal variable that increases toward a threshold, beyond which it will output a signal to another object.

Every time the programs loops through its main loop, the object takes stock of the positive and megative inputs it is receiving. Each positive and negative input has its own intensity value. If the sum of all those inputs is greater than some given threshold, it sends an output signal to another object. If it is lower than the threshold, then no signal is sent and the input decays a little.

After sending a signal, the object puts itself in a state where it cannot send or receive signals for a short period of time.

Depending on how you structure the network of such objects and what inputs the system receives and other parameters, you get a brain.

>>8639593
You could fill several textbooks with descriptions of what goes on inside of neurons. Simplistic descriptions like what has been posted in this thread so far are meme shit that wannabes tell themselves to justify simplistic models that they've become heavily emotionally invested in.

Don't worry about it OP, just use matrix products and ReLU like everybody else.

>>8639725
>no new neurons are created after adolescence

File: neurons.jpg (96KB, 600x446px) Image search: [Google]

96KB, 600x446px

>>8640040
10/10 made me respond.

>>8639725
A neuron it's not an electric circuit, it's a cell.

>>8639923
Regardless of theirs on and off states, they also have gradient, there's also an interrupted synapsis, it has a limit, for the impulse to be transmitted that's also variable.

That not counting architecture differences between neurons.

>>8639968
One way only, it's a stale meme.

>>8640023
>Megative inputs.

>>8640054
>>8640059
did you two morons ever heard of simplification for the sake of vulgarisation. Iwastrying to give OP what he was asking for, aka a simplified model of a generic neuron.

ofc, neurons are cells, not electronic components, still, as the op was asking the comparison with the transistor, Iexpressed a neuron as close as possible to an electronics equivalent.

You see, intelligence is adapting to the need of the situation. not about blabbering your heart learned useless knowledge

>>8640073
Why couldn't you at least say
>No new neuronal structure is created after adolescence

>>8640059
Thats awesome if thats a real pic and not an illustration

iirc there are three types of neurons: ones that are just an axon and a nucleus, ones that have two axons connected to the body and ones that have more than two axons connected to the body. The ones that are basically just an axon do no thinking at all and just transmit the impulse. The other ones can amplify or decrease the impulse depending on the proportional amounts of various neurotransmitters in their synapses.

>>8640092
Also, look up action potentials. That's first step to neurology.

Although there are many types of neurons, can anyone describe a generalised neuron which most neurons would be akin to with some parameter changes?

I'm trying to get into AI programming because there's a particular thing I want to test, so I don't need to know all types of neuron: Just one type with parameters.

>>8639593

Short answer: Nobody knows.

Can someone explain how transistors work to a neurobiologist?

>>8640015
>programmable system
>not turing complete

do you even understand turing machines?

>>8640073
t. buttnerd