I would like to begin a literature review. Not for any specific project, rather I want to just do a very general literature review of Artificial Intelligence / Cognitive Neuroscience. My goal is to try and develop a model for the brain. I want to build a foundation first though. What topics would you suggest I start going through? I'm thinking the first thing I should do is review most of the previous models developed and also read as much literature on the biology of the brain as possible (especially neurons and clusters of neurons). Pardon me if I sound foolish, and pointers would be appreciated.

You'll have to narrow your focus -- doing a literature review over such a broad field even as AI probably would only be an exercise in frustration! Each subfield has its own jargon and mathematical prereqs that you should be familiar with to understand modern literature. If you really do want a broad overview, I recommend a textbook. If some areas in it particularly spark your interest, then you could maybe consider a literature review of those subareas. To jump aboard a subarea, take a look at Readers (collections of important literature for a field) or Review Journals (offer summaries of each subfield with references to essential past literature and modern literature). I recommend learning your textbook material of a subfield before jumping into the literature since the literature often assumes the readership knows about the field, so you won't find good explanations (if they're given at all!).

You might want to look at your terminology a bit as well. Believe it or not, modern AI has very little to do with biology, and it really doesn't even use biological principles ("biologically-inspired" is a buzzword in AI/robotics -- however, these projects are rarely based on much bio). And, AI ESPECIALLY has very little to do with Cognitive Neuroscience (except perhaps with analyzing/finding patterns in fMRI data). Further, looking at literature on "neurons and clusters of neurons" would be Cellular/Molecular Neuroscience, again which has little to do with the previous two categories you mentioned :-) Just to be complete, Cognitive Neuroscience looks at very high-level things like the effect of alcohol on the brain, fMRIs, etc.

Even more importantly than looking for a model is to search for a framework, in my opinion. A framework is something that can be used as a basis for future research which allows for prediction and elaboration. The largest flaw with brain science nowadays is that there is no good framework, so approaches to the brain are very divergent (philosophy, CS, biology, etc) and frequently unproductive. Some of the frameworks and models that have been proposed in the past:

Neural Networks (see Grossberg's papers or Minsky's "Perceptrons" which killed the field in the late 70s)
Logical (old-style AI, see Hofstadter's "Godel, Escher, Bach")
Prediction Model (Cognitive Psychologists, see Jeff Hawkins' "On Intelligence" for a more popular science read on it)
Simulation (once we can model a neuron perfectly, we can work our way up)
Computational (the brain is like a computer -- very modular)
Quantum Theory (it's inherently unpredictable)

Most NN researchers are now looking at applications of NNs rather than how the brain uses them due to their limitations. Biophysical NN models are well understood (on the Simulation side), but we just don't know how to hook them together.

If you want the biology of neurons and "clusters" of neurons ("networks" might be a better word), you might consider some of these texts: "Principles of Neural Science" by Kandel is considered the "Bible" of Cellular/Molec Neuroscience. "From Neuron to Brain" is an advanced undergrad text which is almost entirely about neurons -- this may be a bit more accessible while still being detailed. If you delve into some of these more advanced biology texts, you'll find molecular concepts used frequently that you may not be familiar with and which aren't explained -- e.g. how a G-Protein works, or ATP->ADP, L- vs. D-, organic chemistry, etc. These are explained well in more introductory biology and organic chem texts. Unfortunately, biologists aren't mathematicians (generally) and vice versa, so you will only find "vague descriptions" (to a computer scientist) of mechanisms in biology texts, and very little math (many biologists don't have a good math background).

Good luck!

Quote: Original post by Sagar_Indurkhya
I would like to begin a literature review. Not for any specific project, rather I want to just do a very general literature review of Artificial Intelligence / Cognitive Neuroscience.

My goal is to try and develop a model for the brain. I want to build a foundation first though. What topics would you suggest I start going through? I'm thinking the first thing I should do is review most of the previous models developed and also read as much literature on the biology of the brain as possible (especially neurons and clusters of neurons).

Pardon me if I sound foolish, and pointers would be appreciated.

Seeing as we dont really know how the brain works yet, any attempt to imitate it IS rather foolish. All the models claiming to imitate biology, like neural networks, genetic algorithms and the likes, well, they dont. Its just a pretty name.

If you still want to go with it, google schoolar is your best friend. Search for reviews of connectivism techniques. Just be prepare for fruitless work and desilusions.