Reverse engineering the brain, IEEE Spectrum article
Janelia Farm is a research campus in northern Virgina where scientists are reverse engineering the brain of a fruit fly. At the facility, part of the Howard Hughes Medical Institute, a transdisciplinary team of biologists, neuroscientists, physicists, and engineers hope their studies of Drosophilia could one day enable them to generate a complete circuit diagram of the human brain. As part of IEEE Spectrum magazine's "Special Report: The Singularity," writer Sally Adee visited Janella Farm to see if the work there might even point toward a giant leap in artificial intelligence. From IEEE Spectrum:
Let's say all the engineering problems can be solved in the next five or 10 years. Could researchers then actually reverse engineer the human brain, creating its functional duplicate in silicon? Would consciousness and all its attendant joy, pain, insanity, and genius be freed from biological containment? Adler sees no reason why not. “The brain is the ultimate micromachine,” (researcher David Adler) insists. “The fact that it's made out of meat is a red herring.”Link
His vision is a Google map of the human brain that incorporates not just Janelia's circuit diagrams but also other work in neuroscience. Adler cites the work of Stanford neuroscientist Stephen Smith as “the first steps to finding the soul.” At Harvard's Center for Brain Science, neuro-scientist Jeff Lichtman mapped mouse neurons by “painting” them with fluorescent proteins. Rubin believes he'll live long enough to see an MRI-like device that measures function with such high-resolution output that neurons in fruit flies, mice, or even humans can be observed taking in and processing information in real time.
How would all these different systems work together to show us how the brain does what it does? With his 10- billion- pixel-per-second microscope, Adler is confident he'll be able to produce brain-topography images like Google's satellite views, resolving fine details in sharp focus. Smith's cartography, on the other hand, he compares with Google's map views, including street names. Rubin's fMRI data would be like real-time traffic data. Layering these different maps atop each other, says Adler, could lead to a hybrid comparable to a Google map.
Previously on BB:
• Coloring the brain's wiring Link
the latest
latest episodes
-> Would consciousness [...] be freed from biological containment? Adler sees no reason why not. “The brain is the ultimate micromachine,” (researcher David Adler) insists. “The fact that it's made out of meat is a red herring.
If Adler really said that, he completely misunderstands consciousness.
Anything made in a lab will have no more consciousness than a toaster.
I don't see then asking if this is something they should be doing or not. What is one's responsibility to a silicon human brain endowed with consciousness "and all its attendant joy, pain, insanity, and genius"? Harvard has a bioethics dept. do they have an opinion on this?
These people are babies. The brain is not "electrical", it's not even mostly electrical. THere's neurotransmitters, hormones, proteins, all sorts of soups and goos doing "computation".
This is more Extropian nonsense. (see http://en.wikipedia.org/wiki/Extropian)
This year it's computation; last year it was electrical; before that, steam, clockwork, etc. The model for the brain always seems to be the latest technology, it's embarrassing and arrogant.
Personally, I can't wait for Extropians to download themselves into giant computers; then we can get some work out of them opening and closing spaceship doors.
The proper name for the fly is Drosophila. "Drosophilia" refers to a sexual arousal obtained by sexual contact with or fantasies about drosophila flies.
Cheers.
consciousness will prove to exist on a scale. We will eventually create consciousness of a kind. We may even create consciousness of our own kind, but I think our integral stupidity function will head that off.
When we do create it, and it realizes the utter horror of whatever poor vessel we have come up with, will we then be jailers, torturers and finally murderers? I rather doubt it is going to say "mommy".
I love it when people archair quarterback neuroscientists.
@#3
Absolutely, TOMIC, but they keep throwing themselves on the wire, wave after wave. It's akin to the teams of primatologists and psychologists who set up shop every ten years with some of our hairy cousins, vowing to get them to talk; a few years later, funding gone, they walk away muttering, 'stubborn mthrfckrs,' and go back to their classrooms to dazzle the undergraduates with their bullshit.
@#6
Agreed. I'd like to see them armchair QB Newtonian physics. Why is it that Neuroscience is always seen as this squishy soft science where things are full of magic and powered by fairy dust?
And wait. the Fruit Fly? I thought it was supposed to be a bunch of lobsters, eh?
Geeze, the world is getting scary. Life used to be so simple a while ago. Now everything is such a gray area. Even the definition of "life" is put into question.
@#3:
Neurotransmitters are just a mechanism for signal propogation. That kind of propogation can be modeled - although I have doubts about how well that model scales. Hormones do affect human mental states in important ways (from an evolutionary perspective) but probably aren't crucial for the core of what we would consider "consciousness". To compare our current understand of neurobiology to the clockwork model is folly. Models are constantly improving and increasing in complexity. Just because the model has kept pace with advances in other fields doesn't mean it is invalid.
With all that said, I still think we're far away from "artificial" consciousness - just for different reasons. There are three crucial features that must be achieved to create a viable consciousness: plasticity, ultralow latency, and massive parallel computation.
Plasticity can not currently be reproduced using hardware, even with say EEPROM or similar technologies. An important feature of plasticity is that inputs to a node are weighted and these weights can be altered dynamically based on feedback.
That can be modeled well in software, but that violates the second feature: ultralow latency. In order for a consciousness to learn and interact with its environment in a meaningful way it must be able to react in something approaching "real time". There's probably a fair amount of leeway there, but even the simplest neural networks (think the complexity of a worm's "brain") take on the order of seconds to complete one processing cycle. Humans complete the same cycle on the order of milliseconds and with a much more complicated network.
That brings me to the final criterion: massive parallel computation. Worms can respond to stimuli but I think most would agree that they don't have a consciousness per se. Consciousness demands the ability to respond to a wide variety of stimuli, process the stimuli and perform a number of higher order functions (which are poorly understood) in order to understand and internalize those stimuli. All these computations must be constantly occuring in parallel in order to create an appropriate feedback network.
The solution simply has to be hardware based. The current transistor-based technology simply can't hope to provide any satisfactory intelligence without an increase in efficiency of several orders of magnitude. If we want artificial intelligence in the foreseeable future we have to rethink how integrated circuits are made and find ways to create dynamic interconnections.
@#10
...Worms can respond to stimuli but I think most would agree that they don't have a consciousness per se.
Everything has consciousness, from subatomic particles, atoms, stones, worms, on up the evolutionary ladder. Otherwise, they would not have a presence in the physical universe. Granted, the consciousness may be primitive.
Still, what is created in a lab is (loosely) "robotics" -- and a robot, for its usefulness and speed, is not conscious in the way neuroscientists or AI people promote. It does however have about the same consciousness as a pile of rocks.
I don't see why you all are moaning and griping over this...I think this is pretty awesome.
Sure, it might completely change the way we perceive "life", but hey, it's gonna happen anyway.
yes,MCPFOL @4, we covered that
http://www.boingboing.net/2008/05/06/isabella-rossellinis-1.html
Hey there Not This God at #9, please leave you by-line, website,link and/or sign-off at your profile page instead of appending it to every post. We don't do that here because it eats up space etc.
Mike at #10;
"In order for a consciousness to learn and interact with its environment in a meaningful way it must be able to react in something approaching "real time"."
aw, no thinking trees? jeez...
IEEE Spectrum can thank boingboing for its most recent subscriber. I can't wait until the first issue arrives.
@ # 11 doug117:
"...Everything has consciousness, from subatomic particles, atoms, stones, worms, on up the evolutionary ladder. Otherwise, they would not have a presence in the physical universe."
This does not make any sense at all, as far as I can see. Since we have consciousness, everything that makes us up must have consciousness? How about "Since we have life, everything that makes us up must have life"? Or "Since we have eyes, everything that makes us up must have eyes"?
"Otherwise, they would not have a presence in the physical universe."?
How about "...Everything must have cars, from subatomic particles, atoms, stones, worms, on up the evolutionary ladder. Otherwise, they [cars] would not have a presence in the physical universe."
You'll have to explain your logic....
@ #10 Mike:
I would say you are right about the three requirements, but I would say that they imply that an AI with "...the ability to respond to a wide variety of stimuli, process the stimuli and perform a number of higher order functions" (I don't know if that equates exactly with "consciousness" or not) is nearer that you suspect.
Plasticity, as you say, is quite difficult with hardware, but not particularly difficult with software. As you say, this requires extremely low lag time and extremely large amount of parallel computations. At the current roughly-doubling-every-12-18-months rate, it would not be at all absurd to think of these two requirements being very easily achievable within the next decade or so.
I disagree with the carbon fascists contributing to this thread, since I realise that, in time, our weakly-godlike AI will read all these posts...
Define 'consciousness.' Anybody? Careful....
@#17 SAMSAM
... That a being has *consciousness" means that the being experiences events. It experiences itself in the universe.
Further, a being is conscious because it is God, comes from God. A worm is God experiencing itself as a worm. [My own point of view, shared by Vedantists, Hindus, etc.]
Fortunately, the study of consciousness does not require a belief in God or anything like that. Just a little sense.
"conscious": wanting to stay that way
see: " buddhist "desire" "
Anything made in a lab will have no more consciousness than a toaster.
If you really think that, then you completely misunderstand consciousness.
Why must consciousness be restricted to naturally-evolved systems? Why must it be impossible to figure out? When (ok, If) we do figure it out, what prevents us from building a conscious machine? If we understand when and how neurons fire as a consequence of their surroundings (and such a "Game of Life"-like deterministic/causal model already exists and is not that complicated), and if we had enough processing power, what prevents us from emulating a working human brain, neuron by neuron?
The more I think about what it means to think, to be, to feel, to suffer, to want, to speak, etc, the more I agree with the Hofstadter/Dennett take on all this: We're computers. And if you find it unsatisfactory to say 'We're "just" machines', then you don't have an appreciation for the amazing things that could be done by a machine with billions of interacting parts.
Well, we've got bundle theorists (Buddhists), and someone who claims that consciousness can't exist except as a gift from the supernatural. Any other viewpoints?
@#23 Why must consciousness be restricted to naturally-evolved systems?
Must? Are there rules to follow?
IMHO, that is the way it IS.
You could very well ask "Why *must* the earth revolve around the sun?" It does, there's a reason, I don't know the reason.
IMHO, that is the way it IS.
And once we figure out the causes why systems are the way they are, we tend to be pretty good at harnessing them to our own ends.
"Why are we stuck to the ground while birds get to fly?" "I dunno, that's just the way it is, anything made in a human's machine shop seems to drop like a stone, I wish those fools talking about artificial flight would go throw themselves off cliffs. Why can't they just see it's hopeless".
Pessimism... Research... Pessimism... Research... Hmmm... I'll go with research, thanks. You can keep your pessimism.
NotThisGod, we don't put .sig lines and URLs on comments. Just put your URL on your personal profile.
Since we have consciousness, everything that makes us up must have consciousness?
You are, funnily enough, reverse engineering the original statement. Some of us believe that consciousness is the fundamental principle from which space, time, matter and energy devolve.
@#26
[FWIW, I am not pessismistic, quite the opposite.]
While airplanes do fly, they are not birds.
Robots (and simulated neuronets) do function. They are just not conscious beings of themselves.
@#23
IMHO Dennett got it wrong.
A perfect computational model of your brain might behave exacly like you but there's nothing that tells us why it should be accompanied by consciousness. And there can't be any theory about that (yet). As long as we didn't solve the 'easy problem' we cannot even remotely touch the 'hard problem' of consciousness. That's the reason why I stick with research
@rest
Read Chalmers 'The Conscious Mind' it's the best book in this area.
@#28 ...consciousness is the fundamental principle from which space, time, matter and energy devolve.
Well said!
#3: The thing that drives me nuts about you dualists is that you never seem to develop your argument beyond "you just can't do it". Short of the existence of some magical substance, like a soul or something akin to John Searle's intentionality, that makes us somehow fundamentally different from the rest of the world, I simply don't see why we can't build an artificial thinking machine.
You may think it's arrogant for us to apply our current technologies to our understanding of the mind, but I think that it's even more so to think that humans are somehow unique in our ability to reason. It's a very anthropocentric idea, and one I simply can't lend any credence to.
A perfect computational model of your brain might behave exacly like you but there's nothing that tells us why it should be accompanied by consciousness.
Ah, so you agree we COULD build a conscious machine, you just doubt we're smart enough to understand consciousness! I guess I can't argue with that.
Read Chalmers 'The Conscious Mind' it's the best book in this area.
If you say it's good, and you've obviously put thought into this area, then I'd be curious to learn about Chalmer's points. Let's see what Google turns up... Hmmm, not good...
@#33
The perfect computational model might exhibit fruit-fly-like [or human-like behavior] behavior, but in that there is no iota of consciousness.
Milarepa (#30) did *not* say we could build a conscious machine.
Do you get the difference between behavior and consciousness?
(And, no, we humans are not at all smart enough to understand consciousness.)
There already are some "Google Maps" type applications for the brain...see eg Brain maps online .
They now have an application that allows you to place zoomable images on web pages (it says here).
if you find it unsatisfactory to say 'We're "just" machines', then you don't have an appreciation for the amazing things that could be done by a machine with billions of interacting parts.
It may be anthropocentric, but this kind of talk gives me the creeps. Dissecting human consciousness like this gives me the impression that the speaker objectifies people. I don't want to fall down a slippery slope of accusation, but I can't get the vision of sociopathic eugenicists out of my head now.
@#33
Ah, so you agree we COULD build a conscious machine, you just doubt we're smart enough to understand consciousness!
I didn't say that, although (now) I really doubt that we're smart enough to understand consciousness.
The point is that it might be possible to build a machine that behaves like you do but doesn't have consciousness. (That's Chalmers' point when talking about zombies.)
Let's see what Google turns up... Hmmm, not good...
I was aware of this paper, read it, don't agree on ALL points, but one should shurely read it! But to dismiss Chalmers argument that a zombie is logically possible, just because there are none is simply ignorant, because we CAN NOT KNOW wether there are none!
But if you read more than one science article you have to agree that no single researcher has all the answers.
Chalmers main achievement is that he put the debate in an intelligent framework. To put forward a theory of consciousness like Dennet did is just immoderate at our point of knowledge and Chalmers tells you why
Do you get the difference between behavior and consciousness?
The point is that it might be possible to build a machine that behaves like you do but doesn't have consciousness.
Ah, so you think zombies are not a fallacy. Okay.
So your standard of consciousness is so high, you only know YOU have it. You THINK other people have it but you can't know for sure. Fair enough.
Then if I build a machine that talks just like a person, when you ask it about how it feels and who it is and whether it is conscious... Under your definitions, the machine might not have consciousness (since you cannot verify it feels/thinks/is) just like you. But you have to admit that it has as much consciousness as another person since, after all, you can't know that other people are not zombies. If there truly is a difference between behavior and consciousness, you are the only conscious being in the universe, because all you can see is my behavior.
So no, actually, I do NOT get the difference between behavior and consciousness. I used to think there was one, but then I realized that "I" is an invention, an illusion, a shorthand for a myriad of processes that are not as synchronized or centered as we assume they are.
I stayed away from saying what I think consciousness fundamentally is, which (as Keneke observed) is tied to why killing and torture and unjust suffering are wrong. I'm not going to try to define what consciousness is, but I will say that a conscious being has (among other characteristics) the ability to suffer. The things that allow for consciousness (the sense of "my body", the preference for some states over other states, the ability to invent and learn actions that fulfill a purpose rather than relying solely on the actions coded for in our genes) are the same that allow for what could meaningfully be called "suffering". Which is why I sympathize with vegetarians and am gradually joining them: Animal suffering is not EXACTLY like human suffering, but it's the same general kind of thing, so it's wrong. Because (or for the same reasons that) while animal consciousness is not EXACTLY like human consciousness, it's the same general kind of thing.
Reducing consciousness to a certain kind of interplay between certain kinds of systems does not make it less wrong to hurt people, it makes it MORE wrong to hurt non-human being that have anything like human consciousness.
So yes, there are ethical implications of trying to emulate a brain in a lab.
And so that I don't sound too stubborn: Yes, the Dennett/Hofstadter approach might be incorrect and/or hopeless. Maybe we have a supernatural soul. Or maybe we're naturalistic and even deterministic but too complicated to understand or emulate. Maybe the sense of "I" is not something we could possibly build in a lab. I can't really know. Unlike Dennett and Hofstadter, I do not claim that the way I think about the mind will necessarily be the foundation of all fruitful progress on this front.
But even if my model is less accurate, I have found it to be very useful (it has allowed me to see my wants as forces I can overcome, rather than as parts of my very self which I owed to myself to fulfill; This has made it much easier to be disciplined about all kinds of things), and I insist that yours by comparison is less optimistic and more defeatist since at least I can think of some avenues of research that might allow us to understand the mind and break the problem up into attackable chunks. Your model, on the other hand, basically claims that there is a Gordian knot in the center that cannot be unwoven. It's kinda like evolution: Even if creationism is right (which it might conceivable be, who knows), evolution offers more avenues for research, learning, and applications, while creationism has "We can't understand this" at its very core.
The windup toys of my childhood had behavior but not consciousness (other than as an aggregate of atoms).
No matter how complex a manufactured system gets (whether mechanical, electrical, hydraulic) it is no more than a windup toy.
So we are no more than wind-up toys. Again, we agree!
;)
I behave like a conscious being, why should anyone think I'm truly conscious?
#38
On the one side I agree with you to some degree:
1. The awareness of yourself as an entity is an illusion.
2. The sense of I is something we MAYBE cannot create in the lab.
3. We cannot know.
On the other side I don't agree on some other points.
For example I really think that the Gedankenexperiment of zombies holds for consciousness and my standard for consciousness is that I only know of myself to have. That doesn't make me a solipsist, it's just the way it's defined.
And AGAIN. I don't say that a 'machine' can not have consciousness. I don't say a machine can have consciousness. I say, we YET cannot say wether a machine can have consciousness, because there's no REAL theory of consciousness.
I don't really know what creationism has to do with all this. It's not worthy of discussion in this context. Speaking of 'things not scientific':
To come to see 'wants' as something you can overcome, you wouldn't have to consult Dennett, that's something Freud already said... ;-)
We really are just windup toys in the sense that there is a level on which both, physics and consciousness supervene.
--> An exact replica of the brain is a brain. A silicon implementation of the brain would be sth else exhibiting similar functionality.
--> It is also possible that neuronal level is not enough. See for example the 'quantum neurodynamics' approach.
I'm just saying: I chose to become a Neuroscientist because I'm aware of the fact that I will be able to find sth work to for MANY MANY more decades without getting bored.
Seems Adler is too young to have learned anything from Marvin Minsky's embarrassment.
AI is another way of saying "safe, clean, too cheap to meter."
Anyone else think that the picture looks a bit like the Mandlebrot fractal turned on it's side?
http://en.wikipedia.org/wiki/Mandelbrot_fractal
@#28 ...consciousness is the fundamental principle from which space, time, matter and energy devolve.
Well, you're free to believe that, but as far as I can tell this is not a scientific statement and so there is not much that can be debated here.
I also find that people with a modern New Age/Buddhist bent have co-opted the word "consciousness" to mean whatever they want it to mean. People used to say "soul" or "spirit" or something, but it because more modern and scientific-sounding to say "consciousness."
Then they have trouble with the dual meaning of consciousness that they've just created.
Subatomic particles now have "consciousness", but, they make great pains to argue, they don't have the SAME consciousness that we do. Nothing made up of software and wires has OUR consciousness. All matter and energy and space "devolve" from consciousness, yet it's not OUR consciousness.
The problem arises from trying to co-opt the word to make it mean some kind of all-encompassing spirit, some kind of Buddha mind, but also to keep the original meaning.
I think consciousness should return to meaning the "first-person perspective" that at least some subset of animals with brains have (even if a set of just one animal, us, but I find that unlikely) and that rocks don't have, and stop being used to mean "energy", "spirit", "Buddha mind", or anything else.
Once the nomenclature is resolved, one can actually make rational arguments about zombies and AI...
MIKE, quarterbacking neuroscience, etc: RTFM. There's lots of actual work demonstrating the actual complexity of human cognition; to summarize, your body is pretty much one with the cognition, it's not so much "distributed" (there's that networking paradigm) as "continuous". In short, there's no cognition, intelligence, etc, WITHOUT a body.
The aerie faerie mind/soul vs. brain/body split is a Cartesian idea the was a fit with 500 year old observations and pervailing ideologies. It's pretty much bust.
Has no one noticed that 40+ years of AI research has not once made a function system that could extrapolate outside it's crib? Chess, locomotive repair, not much else.
Sense-map-plan-act is folly. Nothing in the world works like that, not even unsegmented worms. It presupposes that 'sense data is expensive; computation is cheap' when every living thing cont.radicts that; they're all sensory and little explicit "computation" (I use that word cheerfully loosely) in that computer sense. 'Sense data expensive' is hardware leading research. Silly!
And no, "neurotransmitters" (named in the computation/electrical model age) are not just signal paths equiv. to electrical paths. Umm, they ARE different, notice?!. Contemporize, man, contemporize!
SamSam,
Your comment boils down to a demand for all of us to use your semantic paradigm. That's not going to happen.
Antinous:
That's a misreading of my post. My point is that unless we all use words with some approximation of the same meaning, we can't have a meaningful discussion.
If we have a discussion on whether or not consciousness can be created, and you say it can't, that's fine. But if all of a sudden it turns out that you're talking about some kind of life force that subatomic particles have and gave rise to all the energy in the universe... well then we're talking about two different things.
I'm with samsam on this, in that half the time when I'm following an argument on this here site, I have to double check exact meanings of words, then double check exact meanings of words in the original definitions, because posters are using non-standard, incorrect, and/or slang definitions of words.
PS:If we have a soul, why can that not be emulated as well?