Thursday, February 7, 2013

In-silico hypothesis curator

In my earlier post about the big-ticket billion euro human brain project, I drew a laborious analogy between the HBP and the large hadron collider (LHC). I suggested that to maximally take advantage of the proposed mega simulation of the brain, we must ask ourselves what are tens or hundreds of standard models, we need to look for in the properties of the simulator.

For example, a future series of experiments that might involve a little bit of everything: say pharmacological manipulation, functional imaging (fMRI), microscopic of imaging of neural circuits (EM), extracellular electrophysiology, and psychophysics, and that would be hard/impossible to run in a single lab, could be run on the simulated brain. In a potential scenario the in-silico experimentalist could download a single instance of the "live simulated brain" (or even just a single cortical region or slice, depending on the hypothesis), run programs that mimic a pharmacological intervention, and then run other programs that provide infinite-SNR electrophysiology or microscopy data.

I feel that the best way to engage with the HBP would be to curate a transparent, ranked, well-debated, and amendable list of top 100 hypotheses to test in-silico, given the features and limitations of the platform.

I propose a workshop leading to a magazine or a journal special issue, starting with an open call to neuroscientists looking to scale their hypotheses, worded as follows: 

If we could record from all neurons in the brain and know all about its structural connectivity, OR EQUIVALENTLY if we could access molecular/ celluar-scale activity with infinite SNR (from a whole-brain simulator), what kinds of questions would we be asking? How would we scale our hypothesis to match the scaling of technology?

The idea is to source the following classes of talks and articles:

1. Invite supporters of the in-silico platform to submit blue-sky ideas for future great experiments. The set of future great experiments must span a diverse range, from the lowest to the highest levels of brain organization. Questions could include, for example:

    • How do neurons acquire tuning properties and wiring over developmental time scales?
    • How is gene expression in a particular brain area influenced by learning, and how does that in-turn influence long-term memory storage and recall?
    • How are probabilistic computations achieved within and across circuits?
    • How does slow-wave sleep spread from single neurons across the whole brain?
2. Invite contrarians to argue that we can learn all we need to from classic integrative experimental neuroscience.

3. Invite historians of science and simulation-driven scientists from other disciplines—statistical physics comes to mind—to describe the nature of fundamental discoveries have been made from large-scale computer simulations.

Friday, January 25, 2013

Human Brain Project or Euro Drain Project?

Thursday, January 24th, 2013, saw the announcement of EUR €0.5bn funding for what has now been a media darling over the past five years, the Human Brain Project (HBP).

For a sense of scale of what 500 million means, let us note that the public component of the Human Genome Project cost $3bn and the Large Hadron Collider cost $7.5bn. Have a look at some other mega projects in science.

Despite the sensationalist title of this post, I'll try to stay balanced. All said and done, this humble postdoc is looking forward to the trickle-down effects of mega-funding!

Let's start with the criticisms. I see the common criticisms and voices of support, distributed across 4 broad camps.

Too many assumptions, too little detail
The first camp is of the opinion that we simply don't have enough knowledge to simulate a brain that is faithful to nature. For example, to digital neuroanatomists such as Nobel Laureate Bert Sakmann, the BBP assumes too much about the statistics of anatomical connections within a cortical column. Without detailed microscopy-driven reconstruction of columnar connectivity (a goal of connectomics), Sakmann believes that the eventual computer simulation of columnar function is merely an unverifiable candidate for how a true column behaves. To generalize a little, I would say that his is attitude is similar to that of Tony Movshon (see Movshon vs. Seung), who believes that neuroscience is not ready yet for unified theories, and must try to become comfortable about its identity as a collection of cottage industries for another half a century or more.

Too much detail, not enough abstraction
The second camp consists of astute modelers such as Gustavo Deco, who seem to believe that running experiments on a simulated brain will only tell us what we already know. They question the epistemic value of the simulate everything approach. Modelers in general, tend to believe that artful abstractions alone — abstractions which leave noisy details out — will eventually lead to testable predictions and new knowledge. Personally, I believe that this criticism comes from a failure to grasp the difference between modeling and simulation-with-the-purpose-of providing an in-silico experimental platform.

Markram is difficult and unconventional
The third camp consists of the majority of scientists who feel that Henry Markram is hard to work with and doesn't subject enough of his findings to peer-review. One recent example of his fiery temper is the debate between him and IBM's Dharmendra Modha about simulating the cat brain, which ran afoul.

Technological consequences
Finally, we have a camp of pragmatic technocentrists, who suggest that as usually is the case with ambitious projects, there are necessary technological by-products such as the patch-clamp robot (perfect postdoc), and the IBM Blue Gene. Thus, they believe that although the HBP might not achieve its stated goals, the technology left in its wake might ultimately help neuroscientists. Indeed, this could be an amazing learning experience for neuroscientists to make the transition from cottage-industry-scale to industrial-revolution-scale knowledge creation, as certain other fields have done.

All of the above views are important, but there is yet another aspect that I would like to bring up. In one sentence, I believe that the HBP is operating in an epistemological gap. Let me explain by analogy.

LHC. Yes. Standard model. No
I personally think of the HBP as the Large Hadron Collider (LHC) of neuroscience. The LHC creates states of extremely high energy which are rare on earth, but are nevertheless known to be abundant in the known universe. Now, there are two crucial points about studying these high-energy states for the purposes of our analogy:

1. The properties of matter and interaction in these states are not completely known, but we have strong hypotheses about them based on theoretical work.

2. These states are impossible to naturally access and therefore directly characterize with known technology.

Thus, we have three recourses to the two problems:
(A) Develop theories that make testable predictions as we wait for technology to improve so that we may test them.

(B) Improve the technology needed to access these hard-to-access states*, or

(C) Simulate with high fidelity, everything we know about these states, and then do controlled experiments on the simulated system.

Having said this much, the similarities between the LHC and the HBP become immediately obvious. We don't know everything about neural circuits but have a few testable predictions. We cannot simultaneously access and measure the whole brain at all levels. Further, both the HBP and the LHC adopt strategy (C) to problems 1 and 2, except that the HBP is a computer simulation.

*Incidentally, a lesser known project that aims to record from all neurons in the mouse brain, provides an exciting and ambitious solution adopting strategy (B).

However, and finally coming to my point, it is also possible to note that unlike in the case of the LHC, whose flagship project tested the standard model of particle physics, neuroscientists have (I) no agreed-upon hypothesis which is difficult to test in vivo or in vitro, but easy to test in silico. Further, since the HBP is a computer simulation, we have (II) no agreed-upon way to declare the simulation as faithful to nature. This is what I refer to as operating in an epistemological gap.

So now that the HBP has been funded, what can cottage-industry-scale neuroscientists do, other than wringing hands, pointing fingers, twiddling thumbs, and umm, writing blogs?

Downscaling the HBP on your ordinary cluster
First, the computational neuroscientists, and high performance computing fields could independently simulate much smaller-scale brains or brain phenomena, so as to work out the details that would feed into the HBP.

There are two ways to downscale the HBP: downscale the level of complexity of a single neuron while maintaining the order of magnitude of the brain size: e.g. simulate 1 million integrate-and-fire neurons, as a Canadian group has recently done, or downscale the number of neurons while preserving the complexity, as the Blue Brain Project, the precursor to the HBP, has done.

Agree on killer apps
Second, the HBP could organize regular panels or boards, where leading neuroscientists could come up with ways to bridge the epistemological gap, that are not just hand-waving: "we will test the effect of drugs on diseased brains". Specifically,

(I) Propose and vote on a battery of tests that can benchmark the fidelity of the simulation.
(II) Propose and vote on a transparent list of the top 10 hypotheses that cannot be tested in vivo but can be tested on the HBP infrastructure.

So to summarize, neuroscience now has an LHC in the making but doesn't have a Higgs. In the 10 years that it might take to get the LHC built, those of us who wish to use it, could work on formulating as exactly as possible, our favorite top 10 hypotheses to test in-silico.

Sunday, December 2, 2012

Science: a family business?

1. As I was reading S Chandrasekhar's collection of essays and family reminiscences*, I couldn't help but notice the intellectual heavyweights in his family. Many of us know that his uncle, Sir C V Raman, was another eminent physicist and Nobel laureate. But how many of us know that his mother translated Henrik Ibsen's plays to Tamil, his sister Vidya Shankar was a notable veena artist, and his brother Purasu Balakrishnan, a notable physician, writer and Sanskrit scholar?

2. Likewise, on a previous trip to Madras, I met someone who was related to the Alladi Ramakrishnan family, and he informed me later on that V S Ramachandran, the eminent neurologist, is from the same lineage.

3. Yet again, as I was reading about microsaccades, a type eye movement that remains poorly understood even today, I was amused to learn that Robert Darwin, the father of Charles Darwin, was the first to describe them. Likewise, the wealthy Francis Galton, a cousin of Charles Darwin, although remembered more for promoting eugenics, described synesthesia, and promoted and financed the Biometrika journal, a watershed in 20th century statistical thinking.

These examples led me to ponder the idea popularized by Malcolm Gladwell, that one's environment is a much stronger determinant of success (however you choose to define it) than any individual traits. Incidentally, as I was researching the Alladi family, I encountered a note about a neighborhood called Palathope in Mylapore, Chennai, which produced an extraordinary number of lawyers during pre-independence India. This neighborhood reminded me of the Italian village Roseto Valfortore, which produced extraordinarily healthy immigrants.

While the idea of environment breeding success is by no means new, the above examples provoked a journalistic curiosity in me, to learn more about the inner workings of elite intellectual families throughout history.

Among other notable examples of intellectual families, I can recall Mary and Pierre Curie, Niels and Aage Bohr, and although not a family, Ernst Rutherford and his academic descendants. Anybody has any lesser known examples?


*As I might never get around to writing a gushing review, I would like to note that Man of Science was a moving book, which gracefully captures the exalted thoughts of the great, and somewhat under-celebrated man that was Prof. Chandrasekhar. I would go so far as to confess that, reading the lectures of Chandra, his thoughts on great scientists and artists in history, his ideas about classical literature, private letters to his siblings, and the reminiscences of his family members, I felt the same emotions that Stephen Fry might have felt when he discovered Oscar Wilde, a private, incommunicable joy of having encountered a rare and kindred spirit from a bygone era.

Let me give you just one of at least a dozen examples from the book that helped me make the connection to Fry and Wilde. S Balakrishnan writes in a reminiscence after Chanrasekhar's death, about an incident shortly after Chandrasekhar returned from Cambridge with his PhD, and shortly prior to his departure for Chicago:

I recall on two evenings, we walked on the Marina of Madras. He was a recognized scientist. He had shot into the Indian sky like a meteor, or shall I say like Professor Heisenberg in the German sky. But I saw walking beside me an earnest, eager student, thinking only in terms of the pursuit of knowledge, warmed immediately by the mention of high endeavour in any sphere, persuading me, without being patronizing, to think highly of myself. Truly, here is the seed of greatness, I thought.

I was compelled to buy all of merely three copies from bookshelves in all of Chennai's bookstores, and distribute it to friends. One day in the near future, I hope to make the trip down to University of Chicago, where he did his life's work, and get access to his other books, Truth and Beauty, and Newton's Principia for the common reader.

[Link] A brief history of neuroscience

Resonance is a monthly magazine published by the Indian Academy of Sciences, targeted at high-school or undergraduate readers. I used to read it during my JEE days. I came across a very readable piece on the history of neuroscience.

Sunday, March 4, 2012

Newell's twenty questions

In an earlier post, I had brought up the contrasting attitudes towards what constitutes physics, attributed to Rutherford and Feynman, respectively.
That which is not physics is stamp collecting. ~Ernest Rutherford.
In the haughty perspective of Rutherford, the primary concern of science was to construct explanations (theories) for observed phenomena (stamps), and it was only the physicist who would fit this role. Indeed, under this view, it would seem that all other sciences were sources of observations for physics to explain.
Physicists often have the habit of taking the simplest explanation of any phenomenon and calling it physics, leaving the more complicated examples to other fields. ~Richard Feynman.
In the more humble of perspective of Feynman, it seems that the fields outside of physics patiently investigate and characterize anomalies lying outside the realm of the most common instance of a given phenomenon. Their role could thus be viewed as fulfilling the business of collecting and describing rare stamps. Rare stamps when fed back into the activity of theory building (Rutherford's physics), would enable the development of simpler theories with greater predictive power.

The blind men and the elephant

Image courtesy

To understand the state of a field in progress, it is worth considering the Indian parable of the blind men and the elephant. The story goes that six blind men decide to understand how an elephant looks by touching it. Each manages to touch only a small part of the elephant. As a result each develops strong beliefs about the nature of the elephant. They liken the tusk to a spear, the trunk to a snake, the tail to a rope, the feet to a tree, the ears to a fan, and the torso to a wall. Since they are unable to perform further experiments, they end up debating the nature of the elephant ad nauseum.

The gist of the parable is that (1) partial and noisy observations of a system result in erroneous conflicting hypotheses, and (2) the hypotheses are at a stalemate because of the lack of right tools to perform further experiments.

Enter Newell

With this background, it is perhaps useful to pause and consider what Alan Newell had to say back in 1973. Newell is known, among many things, for advocating the mind as an information processing system, along with his advisor and Nobel laureate Herbert Simon. In a valedicatory talk titled, You cannot play 20 questions with nature and win, he attempted to characterize the field of experimental psychology, and as we shall see, his ideas are broadly applicable to the state of cognitive neuroscience today. It is interesting to note that at the time of this lecture, Ed Posner, the founder of the Neural Information Processing Systems (NIPS) conference was in the audience.

To understand the essence of his ideas, let us consider the metaphor of nature as a jigsaw puzzle.

Imagine that you have the pieces of a jigsaw puzzle face down. You don't know what the puzzle looks like, and you don't know how many pieces there are. To flip each piece, you need to do very rigorous experiments and verify the results carefully, multiple times. Each piece may be likened to a phenomenon of the mind, such as auditory short term memory, or one of its properties, such as how long a certain type of information resides in auditory short term memory. The act of spotting each new piece may be likened to an observation of the phenomenon, and the act of flipping it may be likened to the careful elucidation of its properties. Each of these acts is performed by a number of scientists working together or independently, over a number of years. Sometimes new phenomena are discovered; at other times, new properties are elucidated. At still other times, a deeper understanding of the phenomenon is gained, as Feynman explains here with a chess analogy.

Newell articulated beautifully that the journey from empirical exploration to unified theory is a complex one. Consider the following paragraph:

I stand by my assertion that the two constructs that drive our current experimental style are (1) at a low level, the discovery and empirical exploration of phenomena [...] and (2) at the middle level, the formulation of questions to be put to nature that center on the resolution of binary oppositions. At a high level of grand theory, we may be driven by quite general concerns: to explore development; to discover how language is used; to show that man (sic!) is a processor of information; to show that he (sic!) is solely analysable in terms of contingencies of reinforcement responded to. But it is through the mediation of these lower two levels that we generate our actual experiments and give our actual explanations. Indeed, psychology with its penchant for being explicit about its methodology has created special terms, such as "orienting attitudes" and "pretheoretical dispositions," to convey the large distance that separates the highest levels of theory from the immediate decisions of day to day science.

Newell's attitude was as follows. One half of him was very excited about the fact that multiple new pieces are being discovered all the time. But another half of him was concerned that if the trend of upturning new jigsaw pieces was to be extrapolated into the future, the field would be nowhere closer to seeing how the pieces fit.

Consider this statement:

Science advances by playing twenty questions with nature. The proper tactic [hyperlink, mine*] is to frame a general question, hopefully binary, that can be attacked experimentally. Having settled that bits-worth, one can proceed to the next. The policy appears optimal--one never risks much, there is feedback from nature at every step, and progress is inevitable. Unfortunately, the questions never seem to be answered, the strategy does not seem to work.
*This is a reference to another outstanding opinion piece from the 1960s called Strong Inference by John Platt. He stresses the importance of alternate hypotheses and systematically ruling out one of the two, with examples from theoretical physics and molecular biology.

After commending a selection of outstanding individual studies for example, he states:

What I wanted was for these excellent pieces of the experimental mosaic to add up to the psychology that we all wished to foresee. They didn't, not because of a lack of excellence locally, but because most of them seemed part of a mosaic of psychological activity that didn't seem able to cumulate.

If you replace experimental psychology with cognitive neuroscience, and 1973 with 2012, Newell's assessment would still ring true.

In a subsequent post, I will get around to analyzing Newell's recommendations to get unstuck, and how they could apply to the state of modern cognitive neuroscience.

Monday, December 26, 2011

Science of the human condition

As 2011 draws to a close, here are two extremely broad BBC documentaries (and companion books) about what makes us fundamentally human.

The stunning Dr. Alice Robert's The Incredible Human Journey takes us through the migration of Homo Sapiens Sapiens from West Africa to every continent in the world between 70,000-10,000 years ago. It is a 5 part series (5 hours of screen time), with each episode focusing on one continent. A very broad and fascinating discussion including speciation, the hominini, various archeological sites, genetic evidence, climate models for sea levels, competing theories of our lineage, creation myths of indigenous peoples, and much more. Some thoroughly fascinating questions include: How many waves of migration from Africa eventually survived? One or several? Did we interbreed with the Neanderthals? Did East Asians evolve from the Homo Erectus? Guaranteed to stimulate. I'm currently reading Bryan Sykes's The Seven Daughters of Eve to get a richer understanding of the archeological and genetic methods involved in this richly interdisciplinary, politically charged, and data-starved field that attempts to provide a scientific alternative to epics and creation myths. Eager watchers, catch it on youtube before it gets taken down.

Stephen Fry's Planet Word takes us through the evolution, modern day use, and dysfunction of language and symbolic communication. The breadth of this 5 part series (yet again) is impeccable with a coverage of everything from Chimpanzee communication, the FOXP2 gene, Tourette's and swearing, an introduction to Ulysses, the creation of modern Chinese and much else. Unfortunately taken down from youtube.

"Exact science is not an exact science" ~Nicola Tesla according to Christopher Nolan, in The Prestige.

Here's to another year of awe and wonder!

Saturday, October 29, 2011

Change of seasons

The late October Sun smiled weakly today, like a devout caregiver who has been strong for too long, and is unable to hide his waning strength any longer. "Stay strong without me", he said, in an unsuccessful attempt to inspire fortitude during his absence. A yellow, perforated, autumn leaf fell to the matted brown floor lined with its recently deceased kin --- apologetic, for having overstayed its welcome, and in quiet acceptance of its fate. A solitary gull, now devoid of its cacophonous bravado that the summer warmth had inspired merely months ago, circled the Lehtisaari bridge in silent anticipation of the inevitable.

In Helsinki, the change of seasons is an everyday affair. Starting tomorrow, it's time to switch back the clocks and prepare for yet another winter.