/sci/ - Science & Math

>>5386198
I'm not saying that the majority of computational work is large scale neural network simulations - my earlier comment was to give an example of particularly bad work in computational neuroscience.

I'm saying that a lot of the work is poor, particularly because much of it ends up being more descriptive than predictive. Also, the abstractions the models make can be a problem. Again, this is not ALL of the work (e.g., XJ Wang).

>>5386193
I guess rigor was not the right word to use. Explanatory is closer to what I meant. Although, I would contend that the limitations of fMRI are unclear in general (partly due to the abomination that is Logothetis et al., 2001), because it's unclear what blood flow tells us about neural activity.

>>5386183
Xiao-Jing Wang, you say? Yeah, he's down the hall from me. His spiking attractor network model is solid and the predictions about relative NMDA concentration are surprising and impressive.

I'm not saying that all of the work is poor, just a lot of it.

>>5386175
Any work in any field is poor if the wrong analytical tools are used, or the correct analytical tools are used poorly.

My argument is orthogonal to the problem of using the correct statistics. fMRI gives insufficient spatial and temporal resolution of a signal that is not known to directly correspond to any particular neural process. Neurophysiology, while restricted to studying small volumes of tissue in vivo, at least makes measurements with a known relationship to neuronal activity (save for LFP i guess).

>>5386095
>Computational guys - Getting increasingly more useful as it starts to intersect with relevant research, but at present it's mostly hype.

Sad truth. Did you read that recent science paper about the "large scale model of the brain?" Pathetic that such useless work can get so much attention.

>>5382944
Depends on the technique used. You can't really lump together neurophysiologists with imaging researchers because their findings can be inadequately reconciled. Additionally, there is a disparity in the relative rigor of the two disciplines (with neurophysiology being vastly more rigorous).

Not every neuroscientist is what the media sells nowadays - an imaging researcher trying to find God, or language, in a "local" (read: 10*1.25mm^3) region of the brain.

Also, I'm a neuroscience grad student at an Ivy.

>>5380860
It's not shit tier, its top ten in the country for my program.

They're also top fifteen most funded institutions by the NIH. I can't imagine where you're getting your stats from.

I got an offer for grad school from vanderbilt, OP. The PIs whose labs i applied to literally begged me to go.

That's pretty neat in principle. At present, I think that it'd be difficult to get a slice of brain that could make that robot do anything interesting. The prospects are even bleaker if we're using cultured neurons.

It's a cool idea, though. Definitely promising if we can reliably grow and large networks of neurons in vitro with custom wiring.

>>5240243
>hurr durr neuron

My bad, Nature Neuroscience.

>>5240238
Um, Neuroscience isn't a neurology journal, it's a neuroscience journal. If you think I'm being pedantic, then lolyou.

Also, Neuron has the highest impact factor of any dedicated neuroscience journal, so you're flat out wrong.

I drink coffee every day. My lab has a Jura impressa j6 (http://www.amazon.com/Jura-Capresso-13548-Impressa-Automatic-Espresso/dp/B003XPHFY8).).

I take it black, and I drink about four cups (a quart) of coffee every day.

Well fuck, guess I'm late. Seems like the consciousness "intellectuals" hijacked this thread, too.

I'm a Neuro Ph.D student at a top ten institution. If there's any questions about neuroscience (i.e., not useless dribble like consciousness), I'll be monitoring the thread.

>>5229128
I'm around. I'm leaving the thread, too.

Acceptance is to grievance as calling others trolls is to shitposting.

>>5229098
Rather than continuously and unhelpfully asserting that something is "not science," I thought I'd do the plebs a favor by explaining why it's not science, and why it's not necessary to evoke qualia as an explaination. Doing the equivalent in an evolution classroom containing creationists is a lot more helpful and convincing than turning red with rage and failing everyone who believes in god.

A big part of science is sharing knowledge and findings, my friend. I think defeating ignorance with facts is more valuable than most of the threads on this board. Besides, every neuroscience thread degenerates to this bullshit anyway.

>>5229045
I was never shitposting metaphysical nonsense. On the contrary, I provided examples where real science has asked real questions and received real answers about questions that are (naively) thought to be fundamentally subjective and unanswerable.

>>5229059
A photodiode detects light in the environment in much the same way that a rod cell in my retina detects light in the environment. The difference between me and a camera is that I have a massive feedforward network that converts the detection of environmental light into an extremely large array of muscle contractions. With complete knowledge, I could measure the state of the network and ask which of many environmental states my nervous system is acting as though it's in - I could ask what my brain "believes" is going on. There's no need to talk about subjective and objective here. The brain is a real object doing real computations on its environment.

>>5229039
Then it is likely that your definition of perception is naive, or at best, ill-formed. You can't meaningfully define a phenomenon in terms of something that you can't observe. The ability of an animal to detect (i.e. perceive) a stimulus is observable by requiring that the animal report the presence of the stimulus. Perception (detection by sensory systems) is present when the animal correctly reports the presence of the stimulus above chance.

>>5229035
I can't give you a scientific defnition of consciousness, I never will be able to, and I've asserted long before you joined this thread that I cannot do that.

>>5229022
Perception has been defined many many many times over in terms of the limits of a response (motor) of an animal to a sensory stimulus.

>>5229010
So? My point still holds that it's not needed to explain any physical phenomenon.

>>5228947
This is a very good point, the one that I was trying to make throughout the thread. I find it ironic that this fella questions my credibility and proceeds to tote the exact same argument...

Conciousness is not required to explain any physical phenomenon, and no aspect of any physical phenomenon is best explained by conciousness. There's no reason to assume that it exists, and there's no reason to make decisions as though it does. A zombie with an identical behavioral and neural assay to a conscious entity (whatever that is) is equivalent to that conscious entity.

Perhaps of interest to those in this thread, it's worth noting that it's been mathematically proven that a feedforward network (one with an input layer and an output layer) can approximate any mathematical function. A multilayer feedforward network, sounds familiar... note that a consciousness axiom is not required for their proof.

Hornik, K., Stinchcombe, M., & White, H. (1989). Multilayer feedforward networks are universal approximators. Neural networks, 2(5), 359-366.

>>5228827
When it comes down to it, I cannot disprove (nor can I prove) the existence of conciousness. In fact, I believe that it exists, perhaps as baselessly as a creationist believes in God. After all, the best argument I've heard for both ideas is "just look around you."

However, much like God, I don't think that conciousness offers an explanation for anything and I think that it's fundamentally impossible to study. I think that "why conciousness?" would be impossible to answer even with absolutely perfect information about the state of the brain sampled at a million hertz.

>>5228807
citation so you know I'm not full of shit.

http://tools.medicine.yale.edu/ncs/www/documents/lecture12/wang.neuron2002.pdf

Wang, X. J. (2002). Probabilistic decision making by slow reverberation in cortical circuits. Neuron, 36(5), 955-968.

>>5228730
Any model of any phenomenon can be said to be descriptive if all my model needs to do is map one space to another. The problem is that qualia doesn't even have a meaningful description, let alone an explanation short of "the brain did it."

>>5228743
Although we don't have models of the brain that are nearly complete, there are definitely spiking neural network models of perceptual decision making that can reproduce population firing rates observed in various regions of the brain (LIP, FEF, etc...), while simultaneously mapping stimuli to behavior in a way that accuately reproduces the reaction times and accuracy of real primates in a perceptual discrimination task. As far as experimental neuroscience goes, I'd say that perceptual discrimination is as close to studying consciousness as we'll get. These models are also the closest things we have to explaining a concious process. Hell, you can even do a state space analysis on the population of spiking units of the model to get the model's beliefs about the identity of the stimulus at any given instant of a simulated "trial."

What more is there to understand about that decision? What rocks are left unturned that must be explained by, or comprise, subjective experience?

>>5228662
Okay, thats cool, they can exist. I know that I'm experiencing the feeling of my mechanical keys beneath my fingers or the taste of this (extremely shitty) coffee. I'm not gonna pretend that I'm not, although I could and you'd have no idea if I was lying to you or not.

If you ask me to define "experience," I'll shrug. Because I can't define experience, I won't use it to explain anything because it doesn't have a definition, and is therefore a completely trivial explanation for anything. Unless one can give a formal and tractable definition of qualia, it's as useless a concept as phlogiston or vital force. Qualia aren't anything, and they can't explain any observations.

So why talk about them?