Connectome? Not so fast...

AS Seed's Featured Blogger of the week, I have written a short article about the Human Connectome Project, in response to a news story on the magazine's website, called Mapping the brain's highways, by Azeen Ghorayshi.

Several weeks ago, the National Institutes of Health announced the Human Connectome Project, an ambitious $30 million five-year initiative, which aims to map the connectivity of the human brain.

Is this feasible? In short, the answer is no. The idea that a complete connectivity map of the whole brain can be achieved within five years is unrealistic, and producing a microscale map at the level of single neurons and synapses within that time frame is impossible.

To find out why I think a whole brain connectivity map won't be achieved within five years, continue reading at the Seed Magazine website. Any comments, about either of the articles or about the connectome project itself, should be posted here.

More like this

Quite so.
Looking at the field of neuroscience as an amateur, I am still puzzled by the slow speed of progress. Way back in the golden age of cybernetics, when Lem read McCullough-Pitts and Shannon, the pioneers thought that electronic brains would soon be everywhere. Their optimism was well founded in recent advances in mathematics, electrical engineering and biology/biochemistry. And yet, more than half a century later, more than fifty years after the Mountcastle hypothesis, we are still not able to build artificial brains.
This strikes me as odd. Uncomfortably odd.
On the one hand, so much is known about individual neurons,their behaviour, and their connections; right down to the level of biochemistry/genetics. But we do not understand the function of a single cortical column. The blue brain project is a quite desperate project, an heroic effort the break the deadlock.

Really, very interesting topic. If possible, please more information. This is one of the better blogs that I read yeah

Thank you for the article, Mo. Really interesting project. Even a rough map of connection may help scientists unraveling the mechanisms behind neurological and psychiatric diseases, so why aim at consciousness right away ne ayak ..

The commenter who suggested that consciousness cannot be an object of scientific study is refuted by the fact that it is an object of perfectly legitimate and cool scientific studies (e.g., studies of rivalry).

On the other hand, to suggest that Dennett's 'multiple drafts' metaphor explains consciousness is a clear overstatement. The science of consciousness is a fragile green shoot, having just recently broken free from its reputation as fringe science. Consciousness is too interesting to leave in the hands of philosophers.

I can't resist mentioning that I am presently writing a series of posts on the biology/psychology of consciousness, the TOC is here.

The commenter who suggested that consciousness cannot be an object of scientific study is refuted by the fact that it is an object of perfectly legitimate and cool scientific studies (e.g., studies of rivalry).

On the other hand, to suggest that Dennett's 'multiple drafts' metaphor explains consciousness is a clear overstatement. The science of consciousness is a fragile green shoot, having just recently broken free from its reputation as fringe science. Consciousness is too interesting to leave in the hands of philosophers.

I can't resist mentioning that I am presently writing a series of posts on the biology/psychology of consciousness, the TOC is here.

Nate: I'm not the one making bold claims here. I'm not the one who believes that a complete complete connectome will lead to the secret of consciousness in 15 years' time.

I didn't mean that consciousness *is* a bunch of connectivity configurations. But it would certainly involve this, at the very least, because the activity patterns associated with that sequence of configurations are the neural correlates of consciousness.

I doubt we'll ever fully understand how consciousness emerges

1231 the arguably most profound topic of neuroscience, consciousness, has not much to do with plasticity and everything to do with the connections in a brain. The secret to consciousness will be in that wiring diagram
How could you possibly know that?

Part of the problem is that there is no such thing as "the connectome". The STG that Bjorn mentions is a perfect example. Regardless of the anatomical connectivity--i.e., patterns of morphological synaptic connections--there is so much functional plasticity that the anatomical connectome tells you almost nothing of functional relevance.

Probably not in 5 years. But even if, it would only be a step. A big and worthwhile step, mind you, but look at the C. elegans and stomatogastric people: they've had their connectomes for quite some time now (1990s) and still don't fully understand how it all works.

Part of the problem is that there is no such thing as "the connectome". The STG that Bjorn mentions is a perfect example. Regardless of the anatomical connectivity--i.e., patterns of morphological synaptic connections--there is so much functional plasticity that the anatomical connectome tells you almost nothing of functional relevance.

My friend just got a job in Germany trying to get this working for mice with this guy. I'd imagine that will be done in 5 years. Humans would probably be another 5 or 10 years after that, although it might take a ridiculous amount of storage, as my friend was already mentioning dealing with petabytes of data for mice.

Also, there's this in the article: "Undoubtedly, a whole-brain connectivity map will be useful to researchers once it is eventually completed. But what such a map can tell us about how the brain actually works is likely to be limited. This is because the connectome apparently ignores the phenomenon of neuroplasticity."

I just don't agree. Not about neuroplasticity not being important, but because the arguably most profound topic of neuroscience, consciousness, has not much to do with plasticity and everything to do with the connections in a brain. The secret to consciousness will be in that wiring diagram, although it make take use another decade to find it.

In the article, you write:

"Neurons can sprout new connections within minutes of a given stimulus..."

I found that interesting. I was under the impression that stimulus dependent plasticity is mainly the strengthening and weakening of synaptic connections as in Hebbian learning rules, since, afaik, we start of with more synapses than we need and gain function through the pruning of the ones not needed and the tuning of synaptic strength.

If neurons, throughout life, sprout new connections in response to stimuli, what guides the axons so as to facilitate optimal response to the kind of stimuli presented. I was under the impression that axonal guidance worked primarily through chemotaxis along glial structures in development. I guess my main questions is: How would new axons wire to increase functionality in light of the above?

On a side note - I completely agree with the points you make: Plasticity and glial contributions seem essential to understanding brain-function, and I can't see how we should have the technology to make a sufficiently detailed human connectome.

Also, I wonder what you make of these publications on connectomes:

* Hagmann P, Kurant M, Gigandet X, Thiran P, Wedeen VJ, Meuli R, Thiran JP (2007) Mapping human whole-brain structural networks with diffusion MRI. PLoS ONE 2, e597.

* Hagmann P, Cammoun L, Gigandet X, Meuli R, Honey CJ, Wedeen VJ, Sporns O (2008) Mapping the structural core of human cerebral cortex. PLoS Biology 6, e159.

* Izhikevich, EM, Edelman, GM (2008) Large-scale model of mammalian thalamocortical systems. PNAS 105, 3593-3598.

Thanks for editing in the links... I forgot that... okay, to be honest, I was just lazy. Sorry.

So, KevinH, you say 10-15 years for the connectome, and 10 years after that to solve the problem of consciousness?

Consciousness is far more complicated than the sum of connections in the brain, more like a continuous sequence of transient whole-brain connectivity configurations, each lasting just thousandths of a seconds.

I really doubt we'll find the secret to consciousness so quickly, but time will tell...

MPhil: Sprouting new connections doesn't involve axon guidance. It refers to the formation of dendritic spines, the tiny bulbous projections at which much synaptic signalling takes place. Pathways can become strengthened by the addition of new spines (increasing the number of synapses in the pathway), and weakened by their removal (which decreases the number of synapses).

Yes, it was very lazy indeed not to add links. I did so because all the papers are freely available - two in open access journals, and the third as a PDF. They're all key papers, I guess, in the emerging field of connectomics. There's also this 2005 paper by Sporns et al, in which the term "connectome" was first introduced (also open access), and here's an old post about a hi-res topological map of the human brain.

#4
the arguably most profound topic of neuroscience, consciousness, has not much to do with plasticity and everything to do with the connections in a brain. The secret to consciousness will be in that wiring diagram

How could you possibly know that?

#7
Consciousness is far more complicated than the sum of connections in the brain, more like a continuous sequence of transient whole-brain connectivity configurations, each lasting just thousandths of a seconds.

Or even that?

(Without knowing what consciousness is first, which, of course, we don't.)

Nigel: Of course I don't know what consciousness is, but I think it's more likely to involve transient connectivity configurations than being merely the sum of connections.

Mo: regardless of what you "think" consciousness is what you lack for your assumptions is any definitive proof to support your bold claims. If consciousness were so easily explained as simply (which of course is nowhere near simple) transient connectivity configurations then permit me to ask a question or two.

1. If it does turn out to be transient connectivity configurations, which of those account for what is "essentially" consciousness? Can simple bio-electrical currents and neuro connectivity truthfully and honestly be shown as a picture or even interactive map shown in real time, and pointed at and said about with any real certainty "This is what human consciousness is"?

2. Is human consciousness reducible to any single set of connectivity configurations? Because saying that it is more than one set of connectivity configurations makes one wonder just exactly which of the many numerous configurations actually have to do with consciousness and which are simply reflexes, if it IS in fact a single set of them, then, you're looking at a set of connectivity functions that is impossibly complex and could never be mapped or discovered.

I think this article is an ambitious goal, but will ultimately be of limited use pertaining to actual consciousness. Even if the claim was made about it that "this represents the human consciousness" then there will always be skeptics like Nigel above who will always be able to ask "How could you possibly know that?" and they will be right for asking. Furthermore, there is no answer anyone could give that will satisfy the question. You couldn't know that. No matter how much "scientific" proof you give, consciousness will likely fall into those cracks between certainty and questioning. It very well MIGHT be as you say. But even if it IS as you say, you could never know with any real certainty if it was or not.

Nate: I'm not the one making bold claims here. I'm not the one who believes that a complete complete connectome will lead to the secret of consciousness in 15 years' time.

I didn't mean that consciousness *is* a bunch of connectivity configurations. But it would certainly involve this, at the very least, because the activity patterns associated with that sequence of configurations are the neural correlates of consciousness.

I doubt we'll ever fully understand how consciousness emerges.

Mo:

"But it would certainly involve this, at the very least, because the activity patterns associated with that sequence of configurations are the neural correlates of consciousness."

This is an unprovable statement and so I find it hard to be as certain as you are about it. While I also agree that it is likely true, at least to a certain extent, there really isn't a way to dive into a persons "private subjectivity" to understand how their consciousness is reacting or interpreting or understanding, or feeling, or....ect. So even if it plays a role, what role, and how that role is even related to consciousness? I shall quote an intelligent source to respond to these questions.

"I doubt we'll ever fully understand how consciousness emerges." -Mo

In other words, I agree...and disagree at the same time, but not with your principle, more with your certainty about the principle.

The multiple drafts model that Dennett has put forth (and more recently has described in terms of "fame" or "clout" among neural trains) seems to account for pretty much everything there is to account for in regards to consciousness (once one dispenses with the sort of incoherent nonsense like ineffable qualia and zombies found in the philosophy of mind literature). I can't imagine why anyone thinks that staring at a connectome for 10 or 100 years will reveal some "secret" of consciousness or what form they think that secret will take. That involves a category mistake, much like physicists looking for consciousness in quantum phenomena inside microtubules.

the neural correlates of consciousness

Looking for that is like looking for the neural correlates of the center of mass of the brain. Or looking for the molecular signature of the surface of the ocean. But of course what's special about the water molecules that make up the surface of the ocean isn't chemical.

there really isn't a way to dive into a persons "private subjectivity" to understand how their consciousness is reacting or interpreting or understanding, or feeling, or....ect.

This is ignorant nonsense. We can make reliable inferences about such things, just as we make inferences about what goes on inside of stars, or what happened microseconds after the Big Bang, or who killed Lincoln even though none of us were there. The scientific, empirical method for making such inferences about mental phenomena is called heterophenomenology.

But even if it IS as you say, you could never know with any real certainty if it was or not.

This is an epistemological truism, certainly about empirical matters, and says nothing interesting.

"We can make reliable inferences about such things, just as we make inferences about what goes on inside of stars, or what happened microseconds after the Big Bang, or who killed Lincoln even though none of us were there."

And to me this is utterly incoherent. What is a "reliable" inference? I am assuming (another inference) that you are referring to empirical evidence based on scientific discovery that has been repeated at least "enough" times to actually cause us to live under the belief that "such and such" occurs.

Tell me what evidence we have that consciousness even exists at all, or that we each individually have a different one? People have made similar inferences based on Newtonian physics, which turned out to not be exactly right...and phlogiston theory which we NOW know to be utter nonsense but when it was popular was considered to be undeniable truth. People thought the earth was flat for hundreds of years and inferred things from that. What you are claiming is that those inferences were and are just as accurate and compelling and true as any other inference made from science throughout the history of science. That is quite simply nonsensical.

So much like phlogiston, you claim that me being careful what I say about subjective experience is utter nonsense but I say that, you can say whatever you want to about my statements, but here's the proof I submit to you that you could not make a correct inference about my subjective experience. Based on the paragraphs above, I believe that from my own subjective experience, a person who thinks that a man like Dennett has the whole of consciousness somehow explained or accounted for must then therefore mean that you are paying me a...compliment...by saying that I speak incoherent nonsense because what I say doesn't line up with what you tend to want to infer about the world.

Now the explanation here is that I (once again) infer from your tone above that you were actually trying to say that what I had previously said deserved to be slashed to bits and was so worthless it was worth only the attention given in order to bash it and send it away crying, but in fact, I take it as a compliment based on other inferences. So which of us made the correct inference? Which, if any of them, are "reliable" and who decides that they are reliable? Is there an expert who we defer to on such things? Because I have to tell you, that appears like buying into a paradigm for the sake of the paradigm and vigorously defending it as though it were a priori truth, when in fact there is plenty of past evidence to suggest that paradigms, while perfectly useful, certainly tend to not be quite right. (Theory of gravity, Newtonian Physics, Phlogiston, The earth is flat, The sun revolves around the earth ect.) Giving no credit to coherent statements such as mine above is no better than simply saying you do not understand it and so you will pay it no mind.

Quite so.
Looking at the field of neuroscience as an amateur, I am still puzzled by the slow speed of progress. Way back in the golden age of cybernetics, when Lem read McCullough-Pitts and Shannon, the pioneers thought that electronic brains would soon be everywhere. Their optimism was well founded in recent advances in mathematics, electrical engineering and biology/biochemistry. And yet, more than half a century later, more than fifty years after the Mountcastle hypothesis, we are still not able to build artificial brains.
This strikes me as odd. Uncomfortably odd.
On the one hand, so much is known about individual neurons,their behaviour, and their connections; right down to the level of biochemistry/genetics. But we do not understand the function of a single cortical column or ganglion. Not in a snail, not in a rat, and not in a human.
The blue brain project is a quite desperate project, an heroic effort the break the deadlock.

Connectivity in the large, a different timing of development, and size distinguishes the human brain from those of other mammals. I'm are more interested in small animals :-)

Thank you for the article, Mo. Really interesting project. Even a rough map of connection may help scientists unraveling the mechanisms behind neurological and psychiatric diseases, so why aim at consciousness right away..

Interesting exchange. Three thoughts:

1. How much detail is right for a map?
I am a bit more optimistic about the connectomics project (though I despise the name "connectomics"). Sure, some details will be left out. That is true of any map. My map of Durham, NC leaves out Spunky, the dog at 21 Main Street that barks when the mail carrier walks by. That doesn't mean my map is incomplete in any interesting sense. The map is as complete as the authors intended.

The synaptic connectivity map has an additional messiness factor alluded to in the original article, in that brains (and synaptic connections) have to be described statistically. For one, there is variability in the location and size of brain areas across subjects. Also, the probability that a neuron will connect to another neuron cannot be predicted with perfect accuracy, so descriptions of synaptic connections are typically probabilistic. For instance, you will commonly see claims such as 'There is a 0.10 probability that a neuron in V1 layer IV will synapse onto a layer V neuron that is 50 microns away.'

A statistical description isn't invalid, or even necessarily incomplete. A statistical description of synaptic connectivity, at a certain target level of organization, is what we should count as a "complete" map.

The initiative isn't aiming for a reconstruction of each synaptic connection between each neuron in the human brain, but a much higher-level map, something feasible using MRI and human-friendly electrical signals such as EEG. Because the organizers of the connectome project seem reasonable in the level of detail they expect to include, then they will likely succeed. In some ways, they are guaranteed to succeed as in five years they can just say that they got just the amount of detail they were aiming for :).

2. Synaptic organization is necessary, not sufficient
Let's assume they do succeed. As was already pointed out by Björn Brembs, the wiring diagram of a brain is not sufficient for understanding how the thing works.

A couple of years ago in an entry on connectomics, I wrote, "[H]aving a circuit diagram of a chunk of tissue is not sufficient to reconstruct its functional architecture. The anatomy doesn't tell you the biophysical properties of the individual neurons and synapses, and network properties can be dramatically influenced by the neuromodulatory milieu in which the circuits are bathed. However, ultimately we want to wed the detailed anatomical and functional stories, and this work is a promising step in such a direction."

In sum, understanding synapse distribution/abundance is necessary but not sufficient, for functionally decomposing a brain.

3. There is a middle stance on consciousness
The commenter who suggested that consciousness cannot be an object of scientific study is refuted by the fact that it is an object of perfectly legitimate and cool scientific studies (e.g., studies of rivalry).

On the other hand, to suggest that Dennett's 'multiple drafts' metaphor explains consciousness is a clear overstatement. The science of consciousness is a fragile green shoot, having just recently broken free from its reputation as fringe science. Consciousness is too interesting to leave in the hands of philosophers.

I can't resist mentioning that I am presently writing a series of posts on the biology/psychology of consciousness, the TOC is here.

Hardy Falk,
I'm not surprised on how slow are we going. I'm surprised on how fast the understanding of the brain is going considering the extreme dificult of the subject.
The big advantage of the Blue Brain Project is the integration of such different knowledge.
We have a lot of research in molecular bases of neuroscience, but not as much in the concections arena. So this conectoma development is a wellcome and needed effort.

By Antonio Orbe (not verified) on 31 Aug 2009 #permalink

You all are missing the point(s). Have you read the NIH proposal? They are funding a 5 year initiative with overarching objectives, NIH is not saying the "connectomes" or whatever else will be mapped. At one scale, obviously, will know more than now. Olaf Sporns, Tononi and Kotter proposed the term "connectome" for the dataset of a structural description of the network of elements and connections forming the human brain-almost everything-...And also, why is this so impossible? didnt everybody and their dogs recoiled when Craig Venter jumped unto the genome mapping? Who knows what Sporns and his clan of alienas are cooking? more power to them: it is a brilliant, conceptual frame to address the works of brainfat. By the way, consciousness, has been suggested is nothing more than warm brainfat. Heads up: what did the genome map told us? it told us that the long suspected hunch that genes are not genes holds true.