So I've been reading a lot of papers on the reward pathway. But since I'm not schooled in any relevant knowledge I'm having trouble grasping the chain of events. Most papers detail just bits and pieces, but not the full sequence of events.

I'm trying to grasp how the error prediction signal works.

This is my guess so far:


Photosensitice cell A
Light processing neuron B
Dopamine neuron C
Motorneuron D


Light strikes A
A triggers B
B spikes after N triggers.
B triggers C
C spikes after N triggers.
C keeps repeadetly spiking for as long as its internally coded reward size/duration expectation.
D receives predecition triggers spikes causing action.
Reward comes, neural connections are strengthened.
If reward doesnt come in predicted time connection degrades.

Am I correct in this deduction or way off?

How does the actual reward get communicated? By dopamine release in blood?
By the same neurons that coded the reward prediction? If so, how do those exact neurons get triggered to release reward?

  • $\begingroup$ just within the visual processing you will have many neurons processing the information before it goes on to other areas. after all, you are not afraid of a black dot at a certain location of your retina, but of the spider (or whatever). Your 'neuron B' is not a single neuron, but a whole brain area. However I don't know how far up in the processing cascade a signal goes before the information spreads to other brain areas. It might also differ between signals depending on their abstractness. e.g. I expect that being afraid of speaking in public will involve more/different processing. $\endgroup$ – a tiger Jun 2 '17 at 9:04
  • $\begingroup$ i understand that, i tried to simplify the process to understand what happens within the reward pathway. Error prediction signal, reward. How those two are communicated by an event source, and unrelated reward trigger. I see food, i get appetite. Food is in mouth, im happy. How does the food in mouth, hot relaxing bath, sunlight on face get translated to a reward signal to the neurons that were triggered by the prediction signal $\endgroup$ – Tschallacka Jun 2 '17 at 9:48
  • $\begingroup$ I think it's too much to explain the whole reward circuitry here. You would be best off searching for review articles about reward pathways. For example, searching something like this on pubmed: ncbi.nlm.nih.gov/pubmed/?term=reward+prediction+signal+review There are several results that answer your question but do note that this is still an area of active research, it is possible to present hypotheses but the full circuitry is not perfectly known, because there are multiple contributing pathways. $\endgroup$ – Bryan Krause Jun 2 '17 at 16:17
  • $\begingroup$ @BryanKrause I realize it's a lot to explain in one answer, but it's really hard to find one article that summarizes the path from a to be in a "flowchart" type of style. They usually just focus one one piece, making it really hard to figure out how it all fits together as one coherent piece. Ideally i'd have a simple flowchart that would explain in very rough gist how the current theories are how a trigger ends up in a response and how an error correction signal works in it all. $\endgroup$ – Tschallacka Aug 27 '18 at 12:32

Essentially that's what I remember, but there can be a lot more steps between A and C, i.e. more pathways interacting in different systems to produce an associative memory. Pavlov's dogs is the classic example:

1: Pavlov rings bell, dogs hear bell

2: Pavlov feeds dogs, (dogs rewarded)

3: Repeat 1 and 2 for a few days.

4: Ring bell, dogs will salivate and expect food, even if no food is on the way.

So, in the skull, the dogs are making a link between hearing a bell and a reward.

Dopamine is released at dopaminergic synapses, the limbic system is probably the best understood part of the memory/reward system, particularly the mesolimbic pathway for reward.

Neurons get triggered by presynaptic neurotransmitters, some require convergence of signals to reach threshold and trigger an action potential. Connections are strengthened by long-term potentiation, which has been argued to be the basis for memories, but that's a massive topic I'm not qualified to discuss.

I haven't heard of the "error prediction signal" before, might be a new discovery/thinking that I'm a bit behind on.

| improve this answer | |
  • 1
    $\begingroup$ Yes indeed error prediction signals have been discussed for decades now, it is not a new concept. I appreciate the attempt at an answer but there is really very little here that actually answers the question, instead you are explaining a behavioral paradigm. Everything you say is correct, though. I think the original question is just not answerable in a SE format, it is much too broad and the circuitry is not fully understood yet. $\endgroup$ – Bryan Krause Jun 2 '17 at 16:20
  • $\begingroup$ It is a bit of tangent but one current idea is that humor itself exists as an error correcting mechanism. a way of dealing with pathways that make bad predictions $\endgroup$ – John Nov 29 '17 at 17:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.