Dynamical Neuroscience

Pythagorean

When I introduced myself to the neuroscience/brain/mind community here on physicsforums, I was really interested in the higher-order problems like consciousness and cognition, so you may be mixing those connotations in with my writing, knowing that I'm the author (of course, it's possible that I'm also "leaking philosophy" in my writing).

Dynamical neuroscience doesn't serve to be holistic. There are, of course, holistic dynamical approaches.

In some respects, the field is a lot like a blacksmith. The blacksmith makes lots of different things out of metal, mostly because he's good at working with metal. The things he makes may applied all kinds of different way, from helping, to killing, to hanging on your wall as art. He's not very concerned with how people apply it. As the market grows and diversifies, the blacksmiths may specialize (like focusing on a market that uses particular metals and cuts that are safe for children in nursery construction). So then the blacksmith begins to learn more about child care and nurseries since the market is there, and he can provide a higher quality product, tailored for a specific demographic.

a dynamical scientist works with dynamical systems, because they know about the machinery of dynamical systems (specifically, they're versed in nonlinear dynamics, which requires a good mathematical background). Many dynamical scientists are like your unfocused blacksmith, they are only interested in dynamical systems in general, they don't cater to one particular group. But neuroscience, obviously, has exploded with interest and technology in the last couple decades and so there is now a demographic for dynamical scientists, so it has become more efficient to specialize in neuroscience, and learn the subject along with it.

Dynamic neueroscience has a very large medical and physiological component too that make no guesses about the consciousness or other cognitive aspects. At a certain point, your questions will push you over the line to dynamical psychology (yeah, it's out there), which is not the topic I'm covering. Cognitive neuroscience is the acceptable in-between.

addendum:

and ANN's are still acceptable dynamical systems if their global constraints are such. It's just that each individual neuron is not dynamical. The interaction dynamics can be very dynamical, depending on the model.

addendum2:
That's what the morris-lecar (which I summarized with equtions) and hodgkin-huxley models are, they focus more on the dynamics of singular neurons. They are based on resonance, not the digital all-or-none firing, but they still exhibit excitability (similar to all-or-none firing, but not quite the same) and can still be coupled together in a meaningful network.

apeiron

Your focus still seems to be just on the cellular level, so it is not "neuroscience". That is one big source of confusion here. And there is already a wiki on biological neurons that you link to which is about a dynamical sub-discipline.
http://en.wikipedia.org/wiki/Biological_neuron_model


And to call it dynamical, I would expect a justification. Is is dynamical that ignores the emergence of computational features (which it sounds as though you are saying)? Is it dynamical as the way to explain emergent computational features (which is what people would expect)?

Pythagorean

This is a matter of personal availability, not focus. See "attractor networks" in my article and "cognitive neuroscience" with "hopfield networks". This is where I need input from people (that's why I noted that i need more for the cognitive neuroscience section) but remember that it has to have a dynamical basis (it has to have a nonlinear mathematical formulation that exhibits rich dynamics) and it has to be based on empirical observation (to be science, of course).

Remember that I take a reductionist approach (and also remember that I do not think our approaches are mutually exclusive, and in fact think they're beneficial in terms of synergy) so I'm already spent on what I can offer the page. I've had to do a lot of research to expand it as much as I have, and it will take more research to expand it more, but this is why I'm asking for input, because I'm not completely sure where to look.

Note also, that I'm still reviewing old discussions from you. I just came across Scott Kelso, which I'm going to look into to add.

I'm not sure those are directly relevant. Dynamical refers to the mathematics. This is mathematical biology, but more specified. Dynamics is a subject of math, neuroscience as a subject of biology. Of course, nowadays, (neuroscience is interdisciplinary).

It doesn't directly make a judgments about computationalism, emergence, etc. It does the actual analysis on the models and looks for realistic regimes that explain observed behavior. The problem is that the nature of the equations is not simple, there are thousands of regimes to look in (depending on how many dimensions and parameters your model has).

Pythagorean

also note the text book:
http://books.google.com/books?id=kVj...page&q&f=false

apeiron

OK, I understand. But my point is that it is a fundamental mistake (arguably) to believe that it is possible to create a "proper" dynamical description, and from that derive the computational (or rather informational) aspects of the system in question.

This is a huge issue, widely discussed, most especially in mathematical biology (Rosen, Patttee, Salthe, Brier, etc). And also in developmental systems theory (Oyama, etc) - http://en.wikipedia.org/wiki/Develop...systems_theory

So to put up a page that is "just dynamics" needs justification to make sense. Otherwise it sounds like you are taking a regressive step.

Again, to repeat, it has been argued as a no-go theorem that "just dynamics" can't give us a full story in neuroscience. Robert Rosen and Howard Pattee are the best sources here.

If you just want to highlight a class of nonlinear mathematical modelling, then that's great. I'm merely saying that from my perspective, it sounds really odd not to acknowledge the wider context of debate.

The cutting edge of modelling would be about how dynamical processes (which are well-modelled in terms of attractors, metastability, etc) are harnessed by informational ones (which is where semiosis, symbol-grounding, the epistemic cut, etc, come in).

Pythagorean

I don't believe that. I believe in an interdisciplinary approach. What you call a "systems approach" (btw, 'systems neuroscience' academically is a synthesis class, it's a lot like a 'final lesson', you integrate all the subdisciplines for research, you collaborate across the fields, so most actively researching neuroscientist are "systems" neuroscientists by default.

For instance, follow the Dynamical Neuroscience conferences which I cite:
http://neuro.dgimeetings.com/Home.aspx
to see the broad spectrum of contributing fields.

Anyway, with respect the viewpoint you keep projecting on me, I believe that all fields follow from each other in a... well, a dynamical way. It's not like all of psychology is going to be discovered from neurons or all off neurons will be understood solely from psychology; it's that development of both fields will provide insights to each other.

Dynamical sciences been successful in employing "reductionist" (that's a relative term) modeling to describe emergent cognitive proerties (again, see attractor networks in the article).

And as I've also demonstrated in the article (Gluck, in the cognitive neuroscience section) it works in reverse too, but that's not a dynamical example that I know of. I still mentioned hopfield networks though.

And in the intro:

"Information theory draws on thermodynamics in the development of infodynamics which can involve nonlinear systems, especially with regards to the brain."

So please point to specifically where you're confused.

apeiron

You think you know what you are talking about? Fine.

Pythagorean

Yes, I do know what I'm talking about here: my perspective. Which you are misrepresenting, and which is more aligned with your perspective than you realize.

The wiki is based on scientific contributions, which will take time to research and digest before all major perspectives are represented.

Pythagorean

Here's the agenda for the 18th Dynamical Neuroscience Conference that might help clarify some of the topic of dynamical neurosicence:

http://neuro.dgimeetings.com/Home.aspx

JFGariepy

These statements are so contrary to the Wikipedia spirit that only this quotation is enough to justify a request for deletion of the article. Wikipedia is not a place where perspectives meet like in a debate, it is not a public place to express your views. I proposed the merge to Computational Neuroscience and the deletion of most of the content. Even the text book you provide uses "dynamical systems in neuroscience" rather than "dynamical neuroscience". There is no justification in the litterature referenced to create a separate page for dynamical systems, which have been used throughout the history of what has been called computational neuroscience. The whole discussion here proves that you are trying to introduce your personal views in Wikipedia, which in itself is unnacceptable on top of all the fundamental debate behind it. Most people using non-linear equations in computational neuroscience do not define themselves as "dynamical neuroscientists".

Now you have it :)


You can participate to the merger proposal vote here : http://en.wikipedia.org/wiki/Talk:Co...erger_proposal

jackmell

Thank you.

I have much faith in mankind one day solving the problem of mind, consciousness, and artificial intelligence and believe non-linear dynamics will play an important role.

Pythagorean

OK, I've done some research, I found the proper name for my field: Neurodynamics.

Here's a bit about it's history:
http://resources.metapress.com/pdf-p...6&size=largest

Here's what people are saying about it now (well, 9 years ago anyway):

Current Opinion in Neurobiology, August 2001, Volume 11, Issue 4. Neurodynamics: nonlinear dynamics and neurobiology: Henry D. I. Abarbanel, a and Michael I. Rabinovich

-----

You took that post out of context. I differentiated between the article and my post. In my POST I was talking about my perspective. The WIKI article is supposed to be about the collective perspective of the members of the field (which I'm still struggling to perfect through my research, admittedly).

JFGariepy

Well at least you see that the questions I raised concerning the name of your field and its exact definition was justified and it wasn't bad faith on my part. It's very important to watch for these things on Wikipedia if we don't want the encyclopedia to become a live discussion forum where people define things the way they want. Good luck with your future contributions.

Pythagorean

Here's a paper highlghting the specific advantages of a dynamical view:

http://sulcus.berkeley.edu/FreemanWW...s/IC13/90.html

jackmell

That was 20 years ago and the other, nine. Can you cite a more recent opinion of neuroscientist on the matter of using non-linear dynamics, chaos theory, strange attractors, emergence, and self-organization as tools for understanding brain function?

Have you seen the brain series on Charlie Rose? I just caught a part of episode 10 yesterday. I'd be interested in what those guys think about neurodynamics. Maybe even you can contact Charlie and ask him if he could create an episode about neurodyanmics. Get Freeman in there maybe. Here's the link:

http://www.charlierose.com/view/collection/10702

apeiron

Freeman actually impressed me the most out of all the "dynamicists" who sprang up in the 80s/90s. But equally, he showed that attractors and other "straight non-linear models" lacked real bite. They are good for making analogies, but not then for producing actual predictive models.

So just as we would say the brain is not a Turing computer, we can also say it is not a straight dynamical system either.

Pythagorean

jackmell:

strictly neuron behavior
Here's some more recent papers that make use of nonlinear dynamics to understand neuron behavior itself (generally, drawing no conclusions about cognitive aspects). These have been productive in medical and general physiological understanding.

from the nonlinear dynamics journal:
http://www.springerlink.com/content/n2567128x6372603/

Izhikevich gives his opinions in the text, "Dynamical Systems in Neuroscience" written in 2007:
http://mitpress.mit.edu/catalog/item...pe=2&tid=11063

http://www2.gsu.edu/~matals/ashilnikov_cv.pdf - Shilinikov's CV (scroll down to publications. You see a lot of direct medical applications for nonlinear dynamics)


On the more cognitive side:

http://www.mitpressjournals.org/doi/...n.1995.7.4.512
(A Dynamic Systems Approach to the Development of Cognition and Action)

Lewis, Marc D. (2005) Bridging emotion theory and neurobiology through dynamic systems modeling. BEHAVIORAL AND BRAIN SCIENCES; 28, 169–245