Assuming that the brain (structure) is only constructed using neurons and synapses, then what is the difference between the following linguistic statements from a 'mechanical point of view'?

  • Bob is a builder.
  • Bob should be a builder.
  • Bob must be a builder.

In each statement, there are the same "neurons" involved: "Bob" and "builder". I am trying to understand how both neurons could be connected/modelled in each statement in order to differentiate between the semantics.

  • $\begingroup$ I have a strong suspicion that linguists studying modality are not talking to neural network modelers at all. $\endgroup$ – Russell Richie Feb 15 '16 at 6:24
  • 1
    $\begingroup$ Delete the question and make new ones beginning with an investigation of your premises (which are false). For example: In a sentence "Bob is a builder", is there a "Bob neuron" and a "builder neuron"? $\endgroup$ – jona Feb 15 '16 at 13:33
  • $\begingroup$ @jona would be great if you could tell me in what way my assumption is wrong: the question is is about modeling at a higher abstract level, I am aware of "bob" not being ONE neuron but instead may be a neural net that represents "bob". $\endgroup$ – erdal.karaca Feb 15 '16 at 15:12
  • 2
    $\begingroup$ Then possibly: Are words in the brain 'represented' as a network of neurons? In case they are and you know about it, add it as a motivation to the question. In case you assume they are, don't phrase a different question based on this assumption. In case what I stated is your question, motivate why you believe this to be the case, digging down until you no longer require any unfounded 'assumptions' to formulate your question. $\endgroup$ – Steven Jeuris Jun 2 '16 at 19:36

Your question reduces to "what are the neural mechanisms behind language?", which is very much a work in progress.

The only neural model of language that I currently know of is the Semantic Pointer Architecture (SPA), which is largely theoretical and only has some super basic examples.

Basically, the SPA represents language as vector manipulation. The vectors are represented in neurons using the Neural Engineering Framework.

For more details on how this is done, check out "How to Build a Brain" by Chris Eliasmith and the work by Peter Blouw who is trying to solve part of this problem with his PhD thesis.

| improve this answer | |

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