About this Author
DBL%20Hendrix%20small.png College chemistry, 1983

Derek Lowe The 2002 Model

Dbl%20new%20portrait%20B%26W.png After 10 years of blogging. . .

Derek Lowe, an Arkansan by birth, got his BA from Hendrix College and his PhD in organic chemistry from Duke before spending time in Germany on a Humboldt Fellowship on his post-doc. He's worked for several major pharmaceutical companies since 1989 on drug discovery projects against schizophrenia, Alzheimer's, diabetes, osteoporosis and other diseases. To contact Derek email him directly: Twitter: Dereklowe

Chemistry and Drug Data: Drugbank
Chempedia Lab
Synthetic Pages
Organic Chemistry Portal
Not Voodoo

Chemistry and Pharma Blogs:
Org Prep Daily
The Haystack
A New Merck, Reviewed
Liberal Arts Chemistry
Electron Pusher
All Things Metathesis
C&E News Blogs
Chemiotics II
Chemical Space
Noel O'Blog
In Vivo Blog
Terra Sigilatta
BBSRC/Douglas Kell
Realizations in Biostatistics
ChemSpider Blog
Organic Chem - Education & Industry
Pharma Strategy Blog
No Name No Slogan
Practical Fragments
The Curious Wavefunction
Natural Product Man
Fragment Literature
Chemistry World Blog
Synthetic Nature
Chemistry Blog
Synthesizing Ideas
Eye on FDA
Chemical Forums
Symyx Blog
Sceptical Chymist
Lamentations on Chemistry
Computational Organic Chemistry
Mining Drugs
Henry Rzepa

Science Blogs and News:
Bad Science
The Loom
Uncertain Principles
Fierce Biotech
Blogs for Industry
Omics! Omics!
Young Female Scientist
Notional Slurry
Nobel Intent
SciTech Daily
Science Blog
Gene Expression (I)
Gene Expression (II)
Adventures in Ethics and Science
Transterrestrial Musings
Slashdot Science
Cosmic Variance
Biology News Net

Medical Blogs
DB's Medical Rants
Science-Based Medicine
Respectful Insolence
Diabetes Mine

Economics and Business
Marginal Revolution
The Volokh Conspiracy
Knowledge Problem

Politics / Current Events
Virginia Postrel
Belmont Club
Mickey Kaus

Belles Lettres
Uncouth Reflections
Arts and Letters Daily
In the Pipeline: Don't miss Derek Lowe's excellent commentary on drug discovery and the pharma industry in general at In the Pipeline

In the Pipeline

« Lilly's Gamma Secretase Inhibitor for Alzheimer's: Worse Than Nothing | Main | Not The End. Not At All »

August 18, 2010

Reverse-Engineering the Human Brain? Really?

Email This Entry

Posted by Derek

News like today's gamma-secretase failure makes me want to come down even harder on stuff like this. Ray Kurzweil, whom I've written about before, seems to be making ever-more-optimistic predictions with ever-more-shortened timelines. This time, he's saying that reverse-engineering the human brain may be about a decade away.

I hope he's been misquoted, or that I'm not understanding him correctly. But some of his other statements from this same talk make me wonder:

Here's how that math works, Kurzweil explains: The design of the brain is in the genome. The human genome has three billion base pairs or six billion bits, which is about 800 million bytes before compression, he says. Eliminating redundancies and applying loss-less compression, that information can be compressed into about 50 million bytes, according to Kurzweil.

About half of that is the brain, which comes down to 25 million bytes, or a million lines of code.

This is hand-waving, and at a speed compatible with powered flight. It would be much less of a leap to say that the Oxford English Dictionary and a grammar textbook are sufficient to write the plays that Shakespeare didn't get around to. And while I don't believe that the brain is a designed artifact like The Tempest (or Tempest II: The Revenge of Caliban), I do most certainly believe that it is an object whose details will keep us busy for more than ten years.

Saying that its entire design is in the genome is deeply silly, mistaken, and misleading. The information in the genome takes advantage of so much downstream processing and complexity in a way that no computer program ever has, and that makes comparing it to lines of code laughable. I mean, lines of code have basically one level of reality to them: they're instructions to deal with data. But the genomic code is a set of instructions to make another set of instructions (RNA), which tells how to make another even more complex pile of multifunctional tools (proteins), which go on to do a bewildering variety of other things. And each of these can feed back on themselves, co-operate with and modulate the others in real time, and so on. Billions of years of relentless pressure (work well, or die) have shaped every intricate detail. The result makes the most complex human designs look like toys.

So here I am, absolutely stunned and delighted when I can make tiny bits of this machinery alter their course in a way that doesn't make the rest of it fall to pieces - a feat that takes years of unrelenting labor and hundreds of millions of dollars. And Ray Kurzweil is telling me that it's all just code. And not that much code, either. Have it broken down soon we will, no sweat. Sure.

I see that PZ Myers has come to the same conclusion. I don't see how anyone who's ever worked in molecular biology, physiology, cell biology, or medicinal chemistry could fail to, honestly. . .

Comments (37) + TrackBacks (0) | Category: Biological News | The Central Nervous System


1. gr on August 18, 2010 12:49 PM writes...

Kurtzweil just gets worse and worse. It was disturbing enough when he limited it to flirting onstage at TED with a chat-bot duct taped to a cheezy 3D rendering of a woman and babbling about the Singularity.

Now it seems like he's figured out that he can be even lazier and still get the press to talk to him.

Permalink to Comment

2. Sili on August 18, 2010 1:02 PM writes...

Kurtzweil is getting older and more scared of death. (As nicely summarised by SMBC recently.)

Prepare to sprung upon by the believers in the Geek Rapture, though.

Permalink to Comment

3. Zsolt Zsoldos on August 18, 2010 1:03 PM writes...

Since you mention the response article by PZ Myers, it would be also prudent to pay attention to the 3rd comment on that article, which says in short (for detailed explanation of this statement look at the PZ Myers article linked and scroll down to comments):

"Yes, the whole thing is a ridiculous idea. No, it's not Kurzweil's ridiculous idea."


Permalink to Comment

4. Zsolt Zsoldos on August 18, 2010 1:49 PM writes...

A correction to my comment above, I thought you were linking to the response article of PZ Myers in gizmodo, but you linked his blog instead. So the response I was referring to is under the gizmodo response article:

The comment is by "Namarrgon", and the point raised is that the whole subject article of gizmodo is a "very creative reporting", aka miss-quoting and miss-representing.


Permalink to Comment

5. Phil on August 18, 2010 2:19 PM writes...

Unfortunately, both Ray Kurzweil and Steven Wolfram are suffering from a bad case of "When you have a really, really shiny hammer, everything looks like a nail."

Permalink to Comment

6. Hap on August 18, 2010 2:46 PM writes...

I'm wondering if he's suffering from the "i'm getting old, so you guys had better figure out how to make me immortal real soon, kthxbai." disease. It also plays into the "really shiny hammer and nail" disease angle, because if he hasn't offloaded his stack/become immortal by the time he dies, the hammer must not be all that he had hoped.

I guess Sili already got that, though. Oops.

Permalink to Comment

7. Henning Makholm on August 18, 2010 2:57 PM writes...

I don't see how anyone who've ever worked with software could buy that analysis.

It might be possible to funderstand of a million-line program well enough to fix bugs without introducing others, if the program is well designed and documented. But a million lines of unstructured spaghetti code, without comments, in a poorly understood undocumented language without any pretense of consistency, without any overarching design, full of magic numbers and willy-nilly self-modifying code, written over several million years by a series of inept contractors, none of whom tried to understand anything their predecessors did but just fiddled randomly with the code until it seemed not to crash too badly? No chance in hell.

(His numbers are wrong too. A line of code does not equal 25 COMPRESSED bytes. A few quick experiments here indicate that an average line of code compresses to about 4-5 bytes. Not that this matters much in the larger picture, mind you).

Permalink to Comment

8. HelicalZz on August 18, 2010 3:28 PM writes...

First, I see we have another Zz in town. Oh well, always room, so hello Zsolt.

Second, I find laugh out loud irony in Kurzweil wrting about the 'compression' or 'limiting redundancy' of anything. The man has never been known to convey an opinion in a sentence, when an entire essay can be substituted in its place.


Permalink to Comment

9. RKN on August 18, 2010 4:28 PM writes...

At least Kurzweil appears to understand the real objective is understanding the principle of operation (PoO)of the human brain, not merely constructing a simulation. Unfortunately, one is no more likely to get at the PoO of the human brain by assembling its parts list, than you are to understand the PoO of a high performance car merely through the assessment of its parts list.

Permalink to Comment

10. Peter Griffin on August 18, 2010 5:42 PM writes...


Permalink to Comment

11. drug_hunter on August 18, 2010 6:27 PM writes...

While I agree that Kurzweil is (once again) getting ahead of himself, the field of computer simulation of brain function is fascinating. There are respectable scientists attempting to build models of insect brains, for example. One could imagine this work being successful in the next decade. Still a long way from a human consciousness simulacrum, but then again, the singularity, according to Kurzweil, isn't until 2045. Lots and lots is gonna happen during the next three and a half decades...if only our ability to design drugs would similarly advance during that era! One can only hope.

Permalink to Comment

12. rby on August 18, 2010 6:41 PM writes...

While I agree with you~ N.B. Many programming languages do support a style of writing in which instructions generate another set of instructions (e.g. Lisp, Template Haskell). This is called meta-programming.

Permalink to Comment

13. Morten G on August 18, 2010 7:08 PM writes...

I can't believe Sili mentions SMBC without linking to the relevant comic so here goes:

Also I should warn that SMBC is hysterically funny and a bit offensive. It's a bit xkcd with a healthy preoccupation with sex and very expressive drawing.

Permalink to Comment

14. Skeptic on August 18, 2010 7:12 PM writes...

Before writing off Kurzweil as a Kook, lets examine the Medicinal Chemist WorldView:

The Central Dogma of Medicinal Chemistry is that structurally similar molecules have similar biological activities. However there are many EXCEPTIONS. Imagine that.

So Kurzweil says its just code. Guess what, the med chems say the exact same thing, only they use the word 'structure'.

Permalink to Comment

15. Anonymous on August 18, 2010 7:54 PM writes...

It may be profitable for Kurzweil to look into changing careers...perhaps making pianos/keyboards. At least the name is already established...ah, but that would be reinventing the wheel...

Permalink to Comment

16. AKS on August 18, 2010 8:30 PM writes...

Yes, Kurzweil is kind of a kook, hard to take serious. But if you want a fascinating, plausible and well thought out theory on how the brain works from a computer scientist then look at Jeff Hawkin's work:

He's the guy who invented the original Palm Pilot, but his passion was always figuring out how the brain works.

Permalink to Comment

17. MedInformaticsMD on August 18, 2010 8:38 PM writes...

Kurzweil suffers from what I described as the 'Syndrome of Inappropriate Overconfidence in Computers.'

Permalink to Comment

18. J. Peterson on August 18, 2010 11:44 PM writes...

I think Kurzweil is subject to the Law of Futurology...

Permalink to Comment

19. ale on August 19, 2010 5:37 AM writes...

The idea isn't anywhere as ridiculous as you are making it out to be. I think people are mostly confusing information theoretical bits with computer bits (eg the mistaken analogy with lines of code)
Given physics as the decoding machine, the brain design must mostly be stored in the genome. You might quibble with the estimate, and with the possibility of decoding it, but the concept shouldn't be that laughable (or even original). Processing doesn't add bits of information (eg the information content of pi is just a couple of bits, even if the digit expansion has every finite sequence of digits contained in it).

Permalink to Comment

20. cynical1 on August 19, 2010 9:17 AM writes...

@15 anonymous - Ray Kurzweil IS the name behind the keyboards with his surname. And what he did in the synthesizer world was pretty amazing if you are old enough to remember anything about the advent of synthesizers. Bob Moog and Kurzweil should be put in the same category.

Permalink to Comment

21. Cellbio on August 19, 2010 9:24 AM writes...

The brain design in the genome? Really? Such thought drives me nuts. How about turning off all your gadgets, stop texting and simply think. If that fails, try some scientific literature, or even text books. Funny how information yields more complexity that limits ideological utopia, read fantasy.

Simple question for ale and others: what about the environment as a primary force in driving brain 'design'? Seems the design of the human brain derives in some part (large part?) from observation and learning. Maybe none of these futurist have kids, or have watched their children develop, or bother to bring into these ridiculous arguments a vast field of scientific knowledge that lays out human/primate intelligence and that comes with it biologically and socially, like societies that care for young that do not emerge from an egg ready to hunt for food. Really people, display some of our species ability to think. Hey, maybe for some, the design does stop at the genome.....

Permalink to Comment

22. ale on August 19, 2010 10:17 AM writes...

@cellbio I was not referring to the adult brain, but to the brain. Does an 8 month old foetus not have a brain? And it has spent all of its life in, what, an environment which was also has most of its properties encoded in the genome.

(Why not try to understand the arguments being presented under the most generous light you can imagine instead of assuming idiocy? Do you know anything about information theory?)

Permalink to Comment

23. Cellbio on August 19, 2010 10:51 AM writes...

ale, are we talking about information theory, or an idiotic claim that the brain is encoded in the genome. And no, the environment of an 8 month old, in terms of the developing brain, is not encoded in the genome.

Further, I believe that seeing things in the most generous light one can imagine leads to great flights of fancy, fun yes, but in this case and others, these generous evaluations require one to ignore whole disciplines of science, something that I will never do. Convince me with logic and experimental fact, not requests for generosity.

Permalink to Comment

24. Craig on August 19, 2010 11:44 AM writes...

Ray Kurzweil is a fraud and essentially a science fiction writer who is attempting to claim as many future scientific discoveries as he can before he hopefully dies-- can't say for sure since he predicts he'll be able to download his mind into a robot soon---).

He is a charlatan who presents himself to the newsmedia as a visionary rather than the science-lite peddler of hyper-fantastical whimsy that he is.

A Typical Kurzweil scenario -

"Technology will one day allow us to grow small nano-bot powered jet engines on our legs. The need for automobiles will be greatly reduced and greenhouse emissions will plummet.

I have thus proposed how we can solve the transportation crisis, global warming and also the unemployment crisis as people will be able to commute vast distances powered only by a bag of Cheetos."


The above passage is not an exaggeration.

Permalink to Comment

25. RKN on August 19, 2010 11:52 AM writes...

I'm quite optimistic about information theory approaches in biology, but much of the information relevant to the principle of operation of a human brain will not be found, per se, in DNA. Quite a lot of information is bound up in protein folds, for instance. And, as Cellbio indicated, there is quite a lot of relevant information in the environment as well. It's not that there's no information in DNA, there is, but it's not the Holy Grail that Kurzweil and far too many biologists evidently think it is.

Permalink to Comment

26. Brooks Moses on August 19, 2010 1:36 PM writes...

ale @19: Yes, but if you want to make that argument, then you have to contend with the fact that a million bits of real information-theoretic information is an implausibly incredibly huge amount to understand if you have to understand all of it. It's not like a million bits of computer data where most of it is trivial or meaningless and can be just ignored.

I am reminded of the realization I had, upon completing my doctoral dissertation, that virtually all of that dissertation was defining of terms and proving my points; the actual contribution to human knowledge could be summed up in a handful of sentences. Probably, in some appropriate information-theoretic-ideal encoding as part of the larger mass of knowledge, a hundred bits or so.

So, by this measure, understanding the human brain from the genome is comparable to understanding the genuine contributions of ten thousand doctoral dissertations.

Which, actually, seems like an appropriate order of magnitude. Maybe there is something to this after all! :)

Permalink to Comment

27. Ross on August 19, 2010 2:16 PM writes...

I do tire of Kurzweil's critics attacking straw man caricatures of his ideas.

Permalink to Comment

28. Matt on August 19, 2010 10:08 PM writes...

"The design of the brain is in the genome"
This is feasible for manufacturing "blank", "near-perfect" and "near-identical" human brains, but they would amount to nothing more than computers. It would never be "alive" nor have any identity beyond the contents of its memory and its current "mental state", that is, its register dump.

And using DNA or cell-growth at all to predict bitspace does not parallel well, Kurzwell is confusing hardware and software. Since the brain itself is an infinite machine state, every input to the developing brain has a chance of altering future brain states.

You can closely predict how many "transistors" a child's brain will have, but countless unavoidable chaos factors will always lead to unpredictability in how many of those transistors "wire together" into actual bitspace.

Even given the strictest experimental controls, you'll never grow two brains with the same amount of memory space, probably not even within a single Mbit of each other. I forget the technical terms to categorize what sort of system the human brain is, but what we're dealing with is a state machine that can reconfigure itself - and at some point - at many points - some protein at location X is going to unfold more slowly in brain A than in brain B, (i.e. there was a femtogauss differece in the magnetic field between the locations of the test samples) and so-called identical brains are going to inevitably be in different states, shortly after development initiates.

Permalink to Comment

29. Anonymous on August 19, 2010 10:30 PM writes...

Er, short version with the correct quote:
"The design of the brain is perhaps 90% in the genome, and 10% in pure, chaotic circumstance"

I bet all this hack needs to do to prove himself wrong is run his brain dev simulations on two "identical" processors, phase lock the development states, compare the outputs and he'll find his comp/error lines heating up in msecs hot enough to set his pants alight

Permalink to Comment

30. Osaka on August 20, 2010 3:06 PM writes...

There is, unfortunately, a misconception in these comments; a lot of you seem to think that the concept of a blue print for the brain cannot be separated from the development of the brain.

This philosophy is untenable in terms of computer science and neurology. The fact of the matter is, the computational power of the brain is in its neurons, and the interconnections between them. They form subnetworks connected to subnetworks, domains connected to domains, lobes connected to lobes, but it is not the individual elements that hold computational power, but their aggregates and their connections. This basic understanding is well characterized; neurology can confirm it on all levels, and computer scientists have used neural networks for literally decades.

This model allows us to cleanly separate neurons, the computational base unit, and their aggregates and connections, the mechanism of learning. Because of this, it is easy to see that, yes, the genome is MORE than sufficient to build a brain, independent of environmental cues (barring necessary input from the mother, and in some cases, bacteria), as the ability to produce neurons must be encoded in it. In that sense, we already have sufficient information; we can emulate neurons to arbitrary degrees in computer programs.

The key to this understanding, however, is that learning will not occur. We do not have adequate understanding of how learning modifies these neurons connections. It is true, a large degree of these connections are provably genetic; some neuron clusters are encoded in DNA, and many patterns of clusters are inherited. But without environmental input, learning cannot occur.

However, this in no way discounts anything this fellow has said! He has not said that this will be a thinking brain, nor one capable of learning, nor one that is just like you or I. He has merely indicated we will understand how to construct one, how these neurons are put into their base positions by the genetic code, and that, while ambitious, is FAR less ambitious than any of you have made it sound.

You might want to actually understand what he is trying to say before you rip into a nature vs nurture debate, or even talk about environment at all, because it simply was not mentioned nor is it relevant to the discussion. I'm not going to say he is right; computer programs, even ones with as little as a thousand lines of code, can quite often be as convoluted and as ridiculously hard to understand as anything else (provably so, in fact, under Church-Turing thesis, so I'd direct Derek to perhaps view some of my peers "work" before talking about complexity), so proposing that a million lines of code is somehow making this process easier seems suspect. But the base concept, that the genome can tell us how to assume neurons into the correct orders to get a proto-brain, that is a powerful concept with powerful implications, and it may indeed be within our grasp. He may be overzealous in his time frame or in its implications, but it is not totally out of left field.

Permalink to Comment

31. Anonymous on August 20, 2010 8:10 PM writes...

I was just trying out some unsuccessful satire. LOL! Thanks for pointing out Bob Moog though. I almost forgot about him. I agree, he was a pioneer.

Permalink to Comment

32. Anonymous on August 21, 2010 12:59 AM writes...

Perhaps it is prudent to note that Kurzweil has responded to PZ Myers' post at his blog ( Importantly, he write, "I mentioned the genome in a completely different context. I presented a number of arguments as to why the design of the brain is not as complex as some theorists have advocated. This is to respond to the notion that it would require trillions of lines of code to create a comparable system. The argument from the amount of information in the genome is one of several such arguments. It is not a proposed strategy for accomplishing reverse-engineering."

I'm not familiar enough with Kurzweil's work to say whether his ideas are in fact wacky, but Derek's original hypothesis that Kurzweil was being misquoted in the text Derek cites (which is apparently unsourced because Derek does not provide a link) may be the more correct explanation here.

Permalink to Comment

33. Sili on August 21, 2010 11:00 AM writes...

And PeeZed has responded to the response.

Executive summary: Kurzweil still doesn't know his arse from a hole in the ground.

Permalink to Comment

34. Ross on August 22, 2010 8:56 AM writes...


Kurzweil's point, one seemingly beyond the grasp of Myers and his following of window lickers, is that the human brain is a product of a blueprint much lesser than the end result. IN essence what Kurzweil is saying is that we don't need to know how to build a brain per se, we simply need to learn how to build a system that can grow and evolve intelligent behaviour. Myers commentary is piffling and intellectually inadequate.

Permalink to Comment

35. Anonymous on August 23, 2010 2:03 AM writes...

Blue Brain?

Permalink to Comment

36. Kismet on August 23, 2010 11:49 AM writes...

Well one can certainly concede that the genome is a (sort of, kind of) blueprint of the body, incl. the brain, if you know (or can simulate) all of the physical and cellular environment. Such simulations are however way out of our reach, even assuming continued exponential growth in computing power Kurzweil's timelines would be wrong by several decades. Folding at home can do what? Msec folding of small, single polypeptides?

Hence I don't really see the appeal of the whole "genome is a blueprint" thesis.

While finding the necessary simplifications to simulate (or reproduce, understand) the brain will likely be limited by the progress of neurobiology and not computing power...

His whole line of thought about genome size and derived brain complexity strikes me as not wrong per se, but perhaps a non sequitur; irrelevant?

Permalink to Comment

37. Windows 8 Theme on December 10, 2013 12:00 PM writes...

Moѕt discοunt stoгеs only have small or limited volume
or quantitіes of discontinued or overstocked brands.
UK mail оrԁеr cpmpany Bromlеighs sells a huge variety of switches and faceplates
too suit all tasteѕ. Don't go along with too many vibrant shades, or else you danger
mind-boggling your feelingѕ. But things might change when the respective
operating system reach retaіl stores. Put together-on
your own bookcases are an economical selection, neverthеless they are typically unexciting anԁ appear somewhat cheap.

ӏn the menu, click thе "Set as Desktop Background" option.

Permalink to Comment


Remember Me?


Email this entry to:

Your email address:

Message (optional):

The Last Post
The GSK Layoffs Continue, By Proxy
The Move is Nigh
Another Alzheimer's IPO
Cutbacks at C&E News
Sanofi Pays to Get Back Into Oncology
An Irresponsible Statement About Curing Cancer
Oliver Sacks on Turning Back to Chemistry