The article Ants Swarm Like Brains Think really helped me to understand the way that neurons which are pretty dumb on their own (like ants) can work together to create a pretty genius system (a brain or an ant-hive). The idea is that both the neurons and the ants function majorly via positive and negative feedback, which is really neat to understand.

However as shown in that article, sometimes groups of ants get caught in something called a feedback loop. They can repeat the same action over and over because the positive feedback is just being relayed from ant to ant to ant in a circle, and the ants can do something unhealthy such as walk in a circle for days. I don't know any examples of negative feedback loops but they would presumably also be unhealthy.

As noted in the comments, this is an analogy, but not without a good reason: The brain does work via positive and negative feedback. And that does create the potential for feedback loops.

So I'm curious about what biological features in neurons help to avoid these feedback loops in our brains. How does the brain avoid feedback loops?

  • $\begingroup$ The article is cool, but it's using an analogy. The brain doesn't really work, strickly speaking, as an ant swarm. $\endgroup$
    – Tivie
    Oct 15, 2014 at 18:40
  • $\begingroup$ @Tivie of course, but the analogy is not without a good reason: The brain does work via positive and negative feedback. And that does create the potential for feedback loops. $\endgroup$
    – J.Todd
    Oct 15, 2014 at 18:48
  • $\begingroup$ Well, sometimes, it does enter a kind of feedback loop, Epilepsy. $\endgroup$
    – Tivie
    Oct 15, 2014 at 19:01
  • 1
    $\begingroup$ I will try to post a simple answer $\endgroup$
    – Tivie
    Oct 15, 2014 at 19:07
  • 1
    $\begingroup$ @jt0dd This is interesting. I just realized that feedback loop is not a common network motif in neural networks. Feed-forwards/bi-fan are very common. There should be a reason for this: something theoretical. I shall post when I get an answer to this $\endgroup$
    Oct 16, 2014 at 6:45

1 Answer 1


Normally, after an excitatory neuron fires, it becomes more resistant to firing again for a period of time (refractory period). This is due to electrical changes within the excitatory neuron.

After depolarazing and reaching an action potential(firing), the neuron enters the refractory period, in which its charge will go back to normal (resting potential). This refractory period can be divided in two phases: Absolute and relative. During the Absolute refractory period, the neuron cannot be discharged again, since the sodium channels are inactive.

It's the relative refractory phase, however, that is more relevant to your question. During this period, the neuron enters a hyperpolarization states, meaning that it's resistant to firing again, unless it receives a greater stimulus.

So, basically, after firing and for a while, a neuron requires greater stimulus to be fired again. This works as a Feedback control, since even if you overstimulate a neuron it can fire only so often.

Inhibitory and excitatory neurons (and inhibitory and excitatory neurotransmitters) also play a role in the feedback control. Inhibitory neurons reduce the likelihood of a postsynaptic neuron to fire while excitatory neurons do the opposite.

However, sometimes these mechanism fail.

For instance, in Epilepsy, the resistance of excitatory neurons to fire during the refractory period is decreased. A group of neurons begin firing in an abnormal, excessive, and synchronized manner, resulting in a wave of depolarization known as a paroxysmal depolarizing shift, causing seizures.

In Parkinson disease, decreased dopamine causes increased inhibitory output (excess negative feedback) of the motor circuit, which leads to hypokinesia.

In contrast, excess dopamine and dopaminergic neuron activity seem to be related with Schizophrenia.

  • 2
    $\begingroup$ thank you for this explanation. Do you happen to know the relative timespan of neuron-neuron signal propagation and the refractory period? In other words, how many neurons would need to be in a positive feedback loop to moot the refractory effect. And is this much longer than a normal signal path? $\endgroup$ Jun 28, 2016 at 17:12
  • $\begingroup$ @FullDecent has a great followup question here. Say the refractory period is 250ms. Then there could be a feedback loop as long as the time the signal takes to return to the starting node is at least 250ms. $\endgroup$
    – Jamie
    Dec 8, 2016 at 6:10

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .