Life Forming

I’ve written the catalogue notes for an exhibition that has just opened at the Pangolin Gallery in the King’s Place, King’s Cross, London. It is well worth a visit (and if any Nature folks are reading this, you have no excuse – it’s next door).

________________________________________________________________

Artists have drawn on themes and concepts in science at least since Leonardo da Vinci blended both art and science in a unified representation of the natural world. In the twentieth century an interest in form, space, growth and motion was shared by artists such as Picasso, Duchamp and Richard Hamilton and scientists including Einstein, D’Arcy Thompson and Roger Penrose. Today’s collaborations and interactions of artists and scientists are often motivated by the emergence of neuroscience and the concomitant exploration of how we perceive the world.

It’s rare to find at this fecund interface any engagement with the chemical and molecular sciences, which are commonly regarded as simultaneously too abstruse (all those ball-and-stick molecules) and too applied (all those stinking flasks and vats) to have much to say about aesthetic experience. Chemistry might seem to offer no grand ideas about the universe and our place within it, but just atoms, crystals, solvents: inert materiality.

Yet that’s not how insiders see it. Chemistry is unique among the fundamental sciences in its reliance on creativity – literally, it is about making stuff. That demands of the creator both something worth saying and the skill and imagination needed to convey it. “Chemistry, like much of art, and arguably unlike the other sciences, is oriented towards transformation, process and the dichotomy between the synthetic and the natural”, says American chemist Tami Spector. “It is also, like art, highly tactile in its experimental and laboratory aspects, yet subtly conceptual in its representations.”

Briony Marshall is such an insider: an exceedingly rare example of (bio)chemist turned artist. Marshall studied biochemistry at Oxford University before following an intuition that took her to the Art Academy in Borough, south London. In 2011 she became Sculptor in Residence at the King’s Place, which hosts the Pangolin London gallery in association with the sculpture foundry Pangolin Editions, at which her works were cast during the residency.

The visual language that Marshall deploys in much of her sculpture is one deeply embedded in the tradition of chemical science, and of biochemistry in particular, in which wood, metal, plaster and resin have long been deployed to conjure up representations of the molecular fabrics of the world (including our own). These renditions are necessarily schematic, even symbolic. They can also, like the double-helical framework model of DNA at which James Watson and Francis Crick gaze in that iconic 1953 photograph, be rather beautiful.

2013 is not only the 60th anniversary of Watson and Crick’s discovery of the structure of our genetic material but also the centenary of the invention of the means by which it was disclosed: the technique of X-ray crystallography, devised by William Bragg at the University of Leeds and his son Lawrence at Cambridge. By showing that X-rays deflected from the layers of atoms in crystals can reveal the arrangement of those atoms, the Braggs opened a window onto the molecular world that enabled scientists to figure out the shapes not only of simple crystals such as diamond (one of the first crystal structures deduced by the Braggs) but of complex biological molecules such as protein enzymes. In this way they revealed how, at the smallest scales, we are put together.



So there it is: the patterns of atoms and molecules do after all connect to the human condition. This is perhaps the guiding metaphor of Marshall’s work. Chemists frequently anthropomorphize their molecules, talking of how they have attractions and aversions, how they ‘prefer’ particular shapes, conformations and environments. In Marshall’s reinvention of the Watson-Crick model DNA: Helix of Life this tendency is made explicit. Each atom is a human form, arms and legs reaching out to join with others. The assembly took meticulous planning, not least because atoms can scarcely be expected to respect human anatomy. The components are atomized in Marshall’s Individual Atoms (Kit to Make DNA), where they become dancers and stylized primitive figurines, some poised as if for flight or crucifixion. Life’s fundamental form, perilously fragile, emerges only when these hands link, just as each atomic figure must find its proper place in Marshall’s 2008 work A Dream of Society as Flawless as Diamond, which reproduces the diamond network of carbon atoms first discovered by the Braggs.



Marshall’s DNA double helix represents the culmination of a period of exploration of molecular form, which included also depictions of the more abstract shapes of enzymes. Her latest work displays a shift to a different scale in the architecture of life: the evolving shape of an embryo. In Carnegie Stages, she shows five successive stages of embryonic growth, beginning with the appearance of a central fold called the primitive streak that defines the axis of bilateral symmetry along which the embryo’s limbs and organs will later be arranged. This, then, is in a sense the ‘human form’ at its most basic: now more than just a ball of undifferentiated cells, with the characteristic symmetry of the human body established. The formation of the primitive streak, about 14 days after fertilization, has in fact been made an indicator of personhood: marking the point at which an embryo can no longer develop into twins, its appearance denotes the time after which embryos made by IVF may no longer be legally sustained in vitro.



Yet these embryonic forms show nothing recognizably human – for after all, the same shapes are seen also in the development not only of other mammals but of birds and reptiles too. And Marshall has idealized them: exaggerated the symmetry and made the surfaces smooth and edges sharp, so that the nominally biological forms elide into abstraction. If the result evokes resonances with the abstract sculptures of Henry Moore and Barbara Hepworth, that is wholly appropriate, for Moore and Hepworth were themselves inspired by the emergent, lithely contoured forms of organic nature. They were both associated with the style characterized in 1935 by the critic Gregory Grigson as Biomorphism. In part this drew on the vocabulary of macroscopic form evident in natural objects such as bones and weathered stones, but it was also enriched by the visions of amoebae and single-celled organisms that biologists were seeing under the microscope. “When I look at [Moore’s] carvings”, Grigson wrote, “I sometimes have to reflect that so much of our visual experience of the anatomical detail and microscopical forms of life comes to us, not direct, but through the biologist.” The virtuosic exploration of such forms in Scottish zoologist D’Arcy Thompson’s 1917 book On Growth And Form was an explicit influence on Moore (who read it as a student in Leeds) and other representatives of the avant-garde, and it was the inspiration for one of the first major collaborations of artists and scientists in the postwar years, the 1951 exhibition “Growth and Form” at the Institute of Contemporary Arts in London. Moore himself cultivated friendships with scientists, most notably the biologist Julian Huxley, brother of Aldous and grandson of Charles Darwin’s staunch supporter Thomas Henry Huxley.

Marshall made Embryo Spiral while unaware of D’Arcy Thompson’s work, which itself seems to testify to Thompson’s underlying implication in On Growth and Form that there is a fundamental language of natural form which we already intuit. For the so-called logarithmic spiral that she constructs in Embryo Spiral from copper and brass wire, with a tiny but now recognizably human embryo at its focus, has become Thompson’s emblem, shown on the cover of modern editions of his book and engraved on the slate placard that marks his former residence in St Andrews, Scotland. It is the shape of a snail’s shell, of a ram’s curving horn, of an insect’s flight path towards a light. It encodes mathematical relationships that have been awarded (although not by Thompson) almost mystical significance, and Thompson showed how it was a natural consequence of simple laws of growth. In Embryo Spiral it suggests the unfolding of life as a kind of hidden regularity beneath the complexities of embryogenesis and biochemistry: a reassurance of rule and order underpinning the dizzying details of biology.

Stuck in the middle again

I recently protested at criticisms published in the New Yorker by Gary Marcus and Ernie Davis of a paper published in Physical Review Letters that claimed to extract a kind of ‘intelligence’ from a simple rule governing the dynamics of particles. I’d written an unpublished account of this work, which I found interesting.

Well, I may have spoken too hastily. The paper, by Alex Wissner-Gross of Harvard and Cameron Freer of the University of Hawaii, does seem to me to be interesting and quite soberly presented. But it seems that the main target of Gary and Ernie’s criticisms was the extra-curricular claims that Alex was making for the work, especially in a video presentation for his new start-up Entropica. I’ve now taken a look at this, and I do think it seems rather over the top.

One criticism Gary and Ernie make is that the physics of the paper is ‘made up’. The idea is that, if one imposes a constraint on the particle’s dynamics that it maximizes the rate of entropy production over the entire course of its history – which means giving it an ability to look ahead – then one finds it doing all sorts of interesting things, such as cooperating with other particles or using them like ‘tools’. The objection is that real particles don’t obviously behave this way. But I’d maintain that there is a long and healthy tradition in physics of applying this sort of ‘what if’ thinking: what if the system were governed by this rule rather than that one? That’s interesting for exploring the range of possibilities that a system has access to. It’s a particularly common habit in cosmology and the outer reaches of fundamental physics such as string theory, but happens throughout physics – among other things, it’s a way of exploring what is essential and what is not for the phenomena you’re interested in. I can see that this might look a little odd to other scientists – what’s the point of inventing laws that might not be real? – but it’s a useful way of helping physicists develop intuitions.

Besides, this particular choice of ‘what if’ is well motivated. For one thing, we’re familiar with the idea that the trajectories of photons in quantum electrodynamics are determined by a kind of integration over all possible paths. What’s more, the principle of maximum entropy production – albeit in the moment, not in the future – has been invoked (e.g. by Jaynes) as a criterion for the behaviour of non-equilibrium systems. So this seems an interesting parameter space to explore, and I don’t agree with Ernie that the paper hardly seems to belong in a physics journal.

Ernie says, I think rightly, that “To some extent, I think the difference between your viewpoint and Wissner-Gross' on the one side and Gary's and mine on the other reflects the difference in disciplines. Physicists may be taken with this theory as a parsimonious equation that gives rise to behavior that looks like an elementary approximation of intelligence. As a psychologist and AI researcher, we look for theories at the state of the art in terms of explanatory or computational power, and we care very little about parsimony, which neither useful psychological theories nor useful AI programs generally manifest to any marked degree.”

But then there’s the question of whether this toy system has anything to tell us about ‘real’ intelligence of the sort one sees in the living world. And even if it doesn’t, might the approach be useful in other ways, for example in artificial intelligence?

On both of these issues, the paper itself is modest and largely silent (as it should be) and that is why I felt Gary and Ernie were being harsh. But that Entropica video seems to want to make the analogies direct – comparing the cooperating particles to cooperating animals, say, and claiming that “Entropica is a powerful new kind of artificial intelligence that can reproduce complex human behaviors”. It says that Entropica can “earn money trading stocks, without being told to do so,” and shows it commanding a fleet of ships (though it’s not too clear what they are supposed to be doing). There is an awful lot of “just as… so…” talk here, and once you start showing real animals using tools and cooperating in a task, you’re starting to imply that this is the kind of thing your model explains.

Now, maybe Alex has more concrete results than he is disclosing. But I’m not convinced on the basis of what I’ve seen so far. For example, did Entropica actually “make money”, or just perform OK in some simple simulation of a stock market? Will the ‘tool use’ results really have applications for “agriculture”, and if so, what on earth would those be?

It’s not clear from this video that Alex thinks his ‘causal entropic law’ tells us anything about actual human intelligence or animal behaviour, rather than producing behaviours that just look a bit like it. Gary and Ernie have interpreted some of his comments as making such claims, but I’m not so sure – it seems possible that he is just suggesting he has a simple framework that offers a different way of thinking about the issue, just as some simple biomechanical models can produce something that looks like bipedal walking. But I admit that a comment like “Our hypothesis is that causal entropic forces provide a useful—and remarkably simple—new biophysical model for explaining sophisticated intelligent behavior in human and nonhuman animals” could be interpreted either way, and Alex might need to be careful how he phrases things to avoid a misleading impression.

I don’t know that this is such a big deal. If I were an investor seeing the Entropica video, I’d be unimpressed by the lack of evidence to support grand claims, and indeed the lack of any indication of how Entropica works. It’s a long way from the hype that accompanies some big science projects. Yet I do now understand Gary and Ernie’s scepticism: it does rather look as if Alex is trying to jump much too far ahead too quickly. And perhaps that is part of a broader problem in science, which can’t any longer be allowed to advertise its own merits but instead has to spawn a start-up right away. In the end, Entropica will of course stand or fall on its ability to address real problems. I’ll be curious to see if it does so. In the meantime, it remains no more and no less than an interesting bit of exploratory physics.

Umbrella physics

Here is my latest article for BBC Future.

____________________________________________________________

Tents, as is well known, are designed to fold up into a coil just slightly larger than the bag they came out of. Or so it always seems when the time comes to strike camp. Such frustrations might be eased by a study published in the Proceedings of the Royal Society A [1], which describes a mathematical method for deducing particularly efficient ways of pleating large sheets into compact forms.

It is no surprise that the authors, Nicolas Lee and his doctoral supervisor Sigrid Close of Stanford University in California, work in the department of aeronautics and astronautics, for these are disciplines that have a long history of interest in folding sheets. Packing away parachutes in a form that is compact yet guaranteed to unfold easily and reliably had obvious utility for aeronautics; but there is also a growing demand for sheet-like structures on spacecraft, such as solar panels, telescope mirrors, thermal shields and solar sails. With space at a premium on rocket launches, any way to package these systems more effectively can save money.

To find strategies for collapsing sheets compactly, nature is a good place to look. Leaves and flowers are folded inside buds, and insect wings in the cocoon, in a way that not only minimizes space but enables easy unfurling. The accordion-like pleat is a common solution. In some leaves. such as hornbeam and beech, the pleats are not simply parallel ridges but radiate in V-shaped arrays from a central stem or focus, like the folds of a fan.

Lee and Close first consider the case where a sheet of paper is to be pleated into a strip-like form that can be rolled up – for example, so that a rectangular map might be not just rolled into a long tube but concertinaed into a very short tube. A seemingly simple solution would be to fold it into parallel pleats and then roll up the resulting strip. But that doesn’t work at all well, because the sheet has a finite thickness. Even though this might seem very small, it adds up in a roll until the outer layers try to stretch while the inner ones buckle.

The answer, the researchers say, is to make the pleats themselves slightly curved, in such a way as to precisely balance out the gradually increasing radius of the rolled-up pleats. They calculate what the ideal curvature should be for a rectangular sheet of given dimensions and thickness. One consequence is that the folded-up pleats themselves don’t lie flat: the inner pleats in the roll are slightly shorter than the outer ones. By marking out the curved pleats and then folding by hand, Lee and Close show experimentally that a standard piece of A4 printer paper can be wrapped into a tight spiral just 5 mm thick with a central hole of 1 cm.

Some folded structures might need to be opened up while fixed to a central hub, rather like an umbrella. A simple array of corrugated pleats, while efficient for a free-standing sheet like a map, won’t work here, and so a different approach is needed. Leaves attached to a stem are a little like this, and other researchers have shown that one effective design is based on the V-shaped pleats of the hornbeam leaf: these can collapse the sheet into a strip, which might then be rolled or folded up. One drawback of that design, however, is that unfolding involves two distinct processes: unrolling and then opening up the pleats.

An alternative approach, first explored in 2002 by Simon Guest and Davide De Focatiis at the University of Cambridge [2], is to divide up the sheet – a square one, say – into separate sectors (in their case, into four square quarters) and to use the hornbeam fold for each sector separately, resulting in a compact little star-shaped origami form. Lee and Close have now found that, by staggering the corners of the sectors so that they spiral around the central hub, and by slightly curving each V-shaped pleat, such a design can produce a coil that wraps tightly around the hub but can be pulled open in a single gesture. Again, they tried it out – this time with parcel paper, the greater thickness of which poses a more daunting challenge for efficient folding. They found that a square sheet one metre wide could be collapsed into a coil just 8 cm across when wrapped around a central 4-cm-diameter rod. The authors say that, although the curving of the pleats is hardly perceptible, if the pleats are instead perfectly straight then buckling again disrupts the packing.

One particular virtue of the mathematical scheme that the authors have developed to study these folds is that it can be adapted to even more challenging situations, for example where the sheets are made from more than one type of material with different thickness or stiffness, or even for sheets that aren’t perfectly flat, such as bowl-like telescope mirrors made from flexible reflectors – or indeed the curving domes of umbrellas.

References
1. N. Lee & S. Close, Proceedings of the Royal Society A 469, 2155 (2013).
2. D. S. A. De Focatiis & S. D. Guest, Philosophical Transactions of the Royal Society of London A 360, 227 (2002).

We need to talk about DNA

My article on DNA, genomics and evolution in Nature has predictably ruffled some feathers, and I’ve been waiting for the dust to settle a bit before broaching it here. The original piece follows below; I have a considerably longer version that I will be putting on my website soon. As for the fallout, I don’t particularly want to get into it, except to say a few things:

1. This was not intended to be a call for some kind of “paradigm shift” in biology; good god, if such a thing is needed then I’m hardly going to be the one to spot it. I’m simply suggesting that there would be some value in making a little more public how complex the picture has become (and how it is not all about DNA), rather than falling back on the obsolete tropes of ‘books of life’.
2. I have always accepted the possibility that epigenetics and ENCODE might be overblown. Or they might not. The debate is ongoing. But they are only a part of the picture anyway (and the full article explains a lot more of it).
3. It would be absurd to suggest that the Central Dogma is “dead” – I don’t even know what that could possibly mean. Of course it expresses something central to how proteins are made. But calling it the ‘central dogma’ was always a bit silly, and even more so now.
4. I accept it is wrong to imply that the Central Dogma is “DNA makes RNA makes protein.” Crick was more careful than that. My error was to rely instead on James Watson and Marshall Nirenberg. Ha, as if they would know!
5. It seems likely that there is going to be a considered, thoughtful response published in Nature soon, which pleases me.

________________________________________________________________________

DNA: Celebrate the unknowns

On the 60th anniversary of the double helix, we should admit that we don't fully understand how evolution works at the molecular level, suggests Philip Ball.

This week's diamond jubilee of the discovery of DNA's molecular structure rightly celebrates how Francis Crick, James Watson and their collaborators launched the 'genomic age' by revealing how hereditary information is encoded in the double helix. Yet the conventional narrative — in which their 1953 Nature paper led inexorably to the Human Genome Project and the dawn of personalized medicine — is as misleading as the popular narrative of gene function itself, in which the DNA sequence is translated into proteins and ultimately into an organism's observable characteristics, or phenotype.

Sixty years on, the very definition of 'gene' is hotly debated. We do not know what most of our DNA does, nor how, or to what extent it governs traits. In other words, we do not fully understand how evolution works at the molecular level.

That sounds to me like an extraordinarily exciting state of affairs, comparable perhaps to the disruptive discovery in cosmology in 1998 that the expansion of the Universe is accelerating rather than decelerating, as astronomers had believed since the late 1920s. Yet, while specialists debate what the latest findings mean, the rhetoric of popular discussions of DNA, genomics and evolution remains largely unchanged, and the public continues to be fed assurances that DNA is as solipsistic a blueprint as ever.

The more complex picture now emerging raises difficult questions that this outsider knows he can barely discern. But I can tell that the usual tidy tale of how 'DNA makes RNA makes protein' is sanitized to the point of distortion. Instead of occasional, muted confessions from genomics boosters and popularizers of evolution that the story has turned out to be a little more complex, there should be a bolder admission — indeed a celebration — of the known unknowns.

DNA dispute

A student referring to textbook discussions of genetics and evolution could be forgiven for thinking that the 'central dogma' devised by Crick and others in the 1960s — in which information flows in a linear, traceable fashion from DNA sequence to messenger RNA to protein, to manifest finally as phenotype — remains the solid foundation of the genomic revolution. In fact, it is beginning to look more like a casualty of it.

Although it remains beyond serious doubt that Darwinian natural selection drives much, perhaps most, evolutionary change, it is often unclear at which phenotypic level selection operates, and particularly how it plays out at the molecular level.

Take the Encyclopedia of DNA Elements (ENCODE) project, a public research consortium launched by the US National Human Genome Research Institute in Bethesda, Maryland. Starting in 2003, ENCODE researchers set out to map which parts of human chromosomes are transcribed, how transcription is regulated and how the process is affected by the way the DNA is packaged in the cell nucleus. Last year, the group revealed [1] that there is much more to genome function than is encompassed in the roughly 1% of our DNA that contains some 20,000 protein-coding genes — challenging the old idea that much of the genome is junk. At least 80% of the genome is transcribed into RNA.

Some geneticists and evolutionary biologists say that all this extra transcription may simply be noise, irrelevant to function and evolution [2]. But, drawing on the fact that regulatory roles have been pinned to some of the non-coding RNA transcripts discovered in pilot projects, the ENCODE team argues that at least some of this transcription could provide a reservoir of molecules with regulatory functions — in other words, a pool of potentially 'useful' variation. ENCODE researchers even propose, to the consternation of some, that the transcript should be considered the basic unit of inheritance, with 'gene' denoting not a piece of DNA but a higher-order concept pertaining to all the transcripts that contribute to a given phenotypic trait [3].

According to evolutionary biologist Patrick Phillips at the University of Oregon in Eugene, projects such as ENCODE are showing scientists that they don't really understand how genotypes map to phenotypes, or how exactly evolutionary forces shape any given genome.

Complex code

The ENCODE findings join several other discoveries in unsettling old assumptions. For example, epigenetic molecular alterations to DNA, such as the addition of a methyl group, can affect the activity of genes without altering their nucleotide sequences. Many of these regulatory chemical markers are inherited, including some that govern susceptibility to diabetes and cardiovascular disease [4]. Genes can also be regulated by the spatial organization of the chromosomes, in turn affected by epigenetic markers. Although such effects have long been known, their prevalence may be much greater than previously thought [5].

Another source of ambiguity in the genotype–phenotype relationship comes from the way in which many genes operate in complex networks. For example, many differently structured gene networks might result in the same trait or phenotype [6]. Also, new phenotypes that are viable and potentially superior may be more likely to emerge through tweaks to regulatory networks than through more risky alterations to protein-coding sequences [7]. In a sense this is still natural selection pulling out the best from a bunch of random mutations, but not at the level of the DNA sequence itself.

One consequence of this complex genotype–phenotype relationship is that it may impose constraints on natural selection. If the same phenotypes can result from many similarly structured gene networks, it might take a long time for a 'fitter' phenotype to arise [8]. Alternatively, mutations may accumulate, free from selective 'weeding', thanks to the robustness of networks in maintaining a particular phenotype. Such hidden variation might be unmasked by some new environmental stress, enabling fresh adaptations to emerge [9]. These sorts of constraints and opportunities are poorly understood; evolutionary theory does not help biologists to predict what kinds of genetic network they should expect to see in any one context.

Researchers are also still not agreed on whether natural selection is the dominant driver of genetic change at the molecular level. Evolutionary geneticist Michael Lynch of Indiana University Bloomington has shown through modelling that random genetic drift can play a major part in the evolution of genomic features, for example the scattering of non-coding sections, called introns, through protein-coding sequences. He has also shown that rather than enhancing fitness, natural selection can generate a redundant accumulation of molecular 'defences', such as systems that detect folding problems in proteins [10]. At best, this is burdensome. At worst, it can be catastrophic.

In short, the current picture of how and where evolution operates, and how this shapes genomes, is something of a mess. That should not be a criticism, but rather a vote of confidence in the healthy, dynamic state of molecular and evolutionary biology.

A problem shared

Barely a whisper of this vibrant debate reaches the public. Take evolutionary biologist Richard Dawkins' description in Prospect magazine last year of the gene as a replicator with “its own unique status as a unit of Darwinian selection”. It conjures up the decades-old picture of a little, autonomous stretch of DNA intent on getting itself copied, with no hint that selection operates at all levels of the biological hierarchy, including at the supraorganismal level [2], or that the very idea of 'gene' has become problematic.

Why this apparent reluctance to acknowledge the complexity? One roadblock may be sentimentality. Biology is so complicated that it may be deeply painful for some to relinquish the promise of an elegant core mechanism. In cosmology, a single, shattering fact (the Universe's accelerating expansion) cleanly rewrote the narrative. But in molecular evolution, old arguments, for instance about the importance of natural selection and random drift in driving genetic change, are now colliding with questions about non-coding RNA, epigenetics and genomic network theory. It is not yet clear which new story to tell.

Then there is the discomfort of all this uncertainty following the rhetoric surrounding the Human Genome Project, which seemed to promise, among other things, 'the instructions to make a human'. It is one thing to revise our ideas about the cosmos, another to admit that we are not as close to understanding ourselves as we thought.

There may also be anxiety that admitting any uncertainty about the mechanisms of evolution will be exploited by those who seek to undermine it. Certainly, popular accounts of epigenetics and the ENCODE results have been much more coy about the evolutionary implications than the developmental ones. But we are grown-up enough to be told about the doubts, debates and discussions that are leaving the putative 'age of the genome' with more questions than answers. Tidying up the story bowdlerizes the science and creates straw men for its detractors. Simplistic portrayals of evolution encourage equally simplistic demolitions.

When the structure of DNA was first deduced, it seemed to supply the final part of a beautiful puzzle, the solution for which began with Charles Darwin and Gregor Mendel. The simplicity of that picture has proved too alluring. For the jubilee, we should do DNA a favour and lift some of the awesome responsibility for life's complexity from its shoulders.

References
1. The ENCODE Project Consortium Nature 489, 57–74 (2012).
2. Doolittle, W. F. Proc. Natl Acad. Sci. USA 110, 5294–5300 (2013).
3. Djebali, S. et al. Nature 489, 101–108 (2012).
4. Jablonka, E. & Raz, G. Q. Rev. Biol. 84, 131–176 (2009).
5. Mattick, J. S. Proc. Natl Acad. Sci. USA 109, 16400–16401 (2012).
6. Wagner, A. Trends Genet. 27, 397–410 (2011).
7. Mattick, J. S. FEBS Lett. 585, 1600–1616 (2011).
8. Wagner, A. Trends Ecol. Evol. 26, 577–584 (2011).
9. Jarosz, D. F. & Lindquist, S. Science 330, 1820–1824 (2010).
10. Lynch, M. Proc. Natl Acad. Sci. USA 109, 18851–18856 (2012).

Entropy strikes at the New Yorker

Well, here is a curious thing. On this blog I wrote recently about a paper in Physical Review Letters – my piece was originally written for BBC Future, but had to be dropped when the main BBC news team picked up on the same work.

Now psychologist Gary Marcus and computer scientist Ernest Davis have commented on the work in the New Yorker, criticizing it for making overblown and unsupported claims about AI and intelligence. And they cite my piece as evidence of media hype.

I’m flattered, of course, that my humble blog should be awarded such status, as I have always assumed that it is read solely by its 43 faithful followers. I’m even more flattered that Marcus and Davis generously call me ‘well respected’. And I generally enjoy this sort of piece, which punctures the habitual hype of scientific PR and the media’s parroting of it.

But I think they are utterly mistaken in their criticisms. They seem to have misunderstood totally what the paper is saying. They are apparently under the impression that the authors think they have discovered a new law which makes inanimate particles do amazing things in the real world. But “the physics is make-believe”, they complain – “inanimate objects simply do not behave in the way that the theory of causal entropic forces asserts”. So this ‘causal entropic force’ makes a particle stay in the middle of a box – but hey, real gas particles don’t do that, they move randomly! They can’t all go to the centre, because then the gas would condense spontaneously (and incidentally, the second law of thermodynamics would crumble)! So what makes this one particle so special?

Oh lord, where to begin? Wissner-Gross and Freer are not saying that this is something that real particles do, and that no one noticed before. They are saying that if one were to assume this kind of physics, what emerges are weirdly ‘intelligent-looking’ behaviours, which even seem to have something instrumental about them. A genuinely valid complaint would be not “But that’s not how things are!”, but rather, “What’s the point in invoking a law like this, if there’s no good reason to think it is ever manifested?” But that’s to totally miss the interest here, which is that a constraint that seems very dry and abstract (the capacity to integrate over all possible futures, so as to maximize the rate of entropy production over an entire trajectory) produces behaviour that has some very striking characteristics. The point is that one would not guess those outcomes by looking purely at the law that produces them – it is an emergence-like phenomenon. When Marcus and Davis say that “There is no evidence yet that that causal entropic processes play a role in the dynamics of individual neurons or muscular motions”, they seem to be under the impression that the authors have claimed otherwise.

They build another straw man in what they say about AI: “Wissner-Gross’s work promises to single-handedly smite problems that have stymied researchers for decades.” No, it really doesn’t. There’s nothing in the paper about AI, aside from some introductory remarks about how maximum-entropy methods have been used in some approaches.

Where I might have some sympathy with Marcus and Davis is in regard to a fairly loopy piece about the work on the scifi website io9 (“We come from the future”), which says “the theory offers novel prescriptions for how to build an AI — but it also explains how a world-dominating superintelligence might come about.” Here Wissner-Gross does expand on what he has in mind about AI. He is mostly reasonably reserved about that, implying only that their approach might suggest a new angle. But then we get into Terminator territory: “one of the key implications of Wissner-Gross’s paper is that this long-held assumption [that intelligent machines will decide to take over the world] may be completely backwards — that the process of trying to take over the world may actually be a more fundamental precursor to intelligence, and not vice versa.” Huh? Well, you see, Wissner-Gross talks about particles “trying to take control of the world”. By this, I would assume he means that the causal entropic force directs the particle’s behaviour along particular trajectories that may involve a tendency to arrange the immediate environment. But for the io9’ers, “the world” becomes our planet, and “take control” becomes “impose its remorseless robotic mind”. Now, I can’t tell how much of this came from a degree of injudiciousness in Wissner-Gross’s comments to the reporter, and how much was a post hoc arranging of quotes to fit a narrative. But it seems harsh to criticize a scientific paper on the basis of what a sensationalist news account says about it.

The basic problem here seems to be that Marcus and Davis assume that, when Wissner-Gross and Freer talk about “intelligence”, they must be talking about the same thing that psychologists see day to day in humans. So it’s “intelligent” always to maximize your future options, huh? Well, then, what about this? – “Apes prefer grapes to cucumbers. When given a choice between the two, they grab the grapes; they don’t wait in perpetuity in order to hold their options open. Similarly, when you get married, you are deliberately restricting the options available to you; but that does not mean that it is irrational or unintelligent to get married.” It’s a little bit like saying that bacteria don’t show rudimentary cognition in climbing up chemical gradients because, hey, we sometimes decide not to move towards smells that we really like. The authors were not claiming that all “intelligent” behaviour must be governed by the causal entropy principle, but simply that this remarkably simple rule can produce what look like intelligent behaviours.

“What Wissner-Gross has supplied is, at best, a set of mathematical tools, with no real results beyond a handful of toy problems”, they say. And yes, that is really all the paper claims to do. “Toy problems” is here meant to be dismissive – the authors don’t seem to know that physicists talk about “toy models” all the time, meaning mnimal, obviously too-simple ones that have illustrative, heuristic and suggestive value, rather than ones that are pointless and silly. “There is no reason to take it seriously as a contribution, let alone a revolution, in artificial intelligence”, Marcus and Davis continue, “unless and until there is evidence that it is genuinely competitive with the state of the art in A.I. applications.” Can they really believe the authors think they have a way of doing AI that will beat the state of the art (but that they forgot to mention it in their paper)?

Sure, “it would be grand, indeed, to unify all of physics, intelligence, and A.I. in a single set of equations”, they jeer. To unify all of physics (let alone the rest of it)??! Come on chaps, now you’re really just making it up.

Counting fish

Here’s my most recent piece for BBC Future (although a new one appears today).

_______________________________________________________________

Newborn fish not only can count, but can be taught to count better. This discovery by a team of psychologists at the University of Padova in Italy [1], might seem bizarre, even frivolous at first glance, but in fact it bears on a deep and interesting question: how do we make quantitative estimates based on what we see?

There’s a long tradition of psychological and anthropological research on the question of innate counting systems in humans, which underpins familiar popular notions such as the existence of cultures whose number system goes “one, two three, many.” These simplistic ideas can obscure the fact that even some non-human animals recognize that there are different degrees of “many”: that ten objects are not equivalent to a hundred.

Such distinctions matter in the wild. Animals need, for example, to be able to tell which of two food sources is the largest, or to selectively join the largest group of peers so as to maximize their chance of evading predators.

However, there is a difference between seeing that one group is bigger than another, and actually counting the numbers in each group. One suggestion is that animals use a numerical system for very small numbers – up to about 4, say – but a cruder “this is more than that” ratio-based system for larger numbers. The two methods aren’t easy to tell apart – there’s only so much that a fish will tell you about its reasoning – but it can be done. For example, if an animal makes an assessment based on just the ratio of two quantities, discrimination should get less accurate as the ratio approaches 1: better for 1 vs 4 (ratio 0.25) than for 3 vs 4 (ratio 0.75). But if they’re ‘counting’, the performance should be much the same for all of these pairs.

On this basis, Laura Piffer and her colleagues at Padova have previously shown that some fish can use purely numerical information, and not just ratios, to distinguish numbers of objects more than 4 [2]. Nevertheless, fish seem to prefer a number system only up to about 4, and the ratio method thereafter. An ability to make quantitative distinctions based on ratios is something that develops gradually in humans. Six-month-olds can tell apart a 1:2 ratio (not just one and two objects, but, say, 8 and 16), but not until about 10 months do they distinguish 2:3. Pre-schoolers can handle 3:4 ratios, and 6-year-olds 5:6 [3]. On this basis, fish are approaching a preschooler’s numerical literacy, and it’s not unreasonable to suspect that studying fish might shed some light on the mathematical competencies of humans too.

For one thing, fish, like children, learn numeracy as they grow. Newborn guppies can discriminate 1 from 2, and even 3 from 4, but the ability to distinguish 4 from 8 or 12 emerges only between 20 and 40 days later. That’s a bit surprising, given that the adult guppies use ratios to judge the larger numbers: you’d think that telling 3 from 4 would be harder than telling 4 from 12. So the newborns seem innately blessed with the capacity to count to 4, but have to learn the ratio method as they grow.

The latest set of experiments by Piffer and colleagues then asks: do newborn guppies already have the mental capacity to learn ratios, or does that neural hardware develop later? Can they be taught?

But how do you teach maths to a fish? Just offer them food as a reward. The researchers put the fish in a rectangular tank and displayed images of dots at each end, delivering food near the largest quantity. The fishes’ ability to distinguish the two numbers can then be inferred from the length of time they spend near each end: they learn to wait for food by what they think is the larger number of dots.

While newborn guppies couldn’t distinguish 7 dots from 14, after about 20 trials they had learnt to do so. They were then about two weeks old, but still well below the age at which the ratio discrimination system seemed to kick in for untrained guppies. So they seem indeed to be born with this potential ability, and just need to exercise it. Might human babies have such hidden talents too?

Does this mean fish are smarter than we give them credit for? You could choose to see it that way, but in fact it supports a growing recognition that cognitive processes like counting, which we might imagine to be quite complex, can in fact be achieved with a surprisingly small number of neurons [4]. Counting and comparing numbers might not be as hard as we think.

References
1. L. Piffer, M. E. Miletto Petrazzini & C. Agrillo, PLoS ONE 8, e62466 (2013).
2. C. Agrillo, L. Piffer & A. Bisazza, PLoS ONE 5, e15232 (2010).
3. J. Halberda & L. Feigenson, Developmental Psychology 44, 1457 (2008).
4. T. Hope, I. Stoianov & M. Zorzi, Cognitive Science 34, 406 (2010).

Just allow me this little rant, then normal service will resume

I strongly suspect that bête noirs area sign that one needs to get out more. All the same I have mine, among which one of the chief ones is the idea that Giordano Bruno was a martyr to science, being burnt at the stake for his Copernican views. This is a myth. Rather, he was condemned for all manner of religious heresies. Mentioned among them at his trial was the idea of a plurality of worlds, which is of course not explicitly Copernican. Evidently Bruno did have Copernican sympathies, although it isn’t clear how well he understood Copernicus’s arguments. But there is no reason to think that the Church would have burnt him for those.

The myth has doubtless arisen because of the proximity of Bruno’s execution to Galileo’s persecution. But there is no real relation between them, and after all Galileo himself was initially granted considerable tolerance for his own Copernicanism. Now, none of this excuses the Church one whit for a barbaric and dogmatic act. But it is frustrating to see his canard trotted out as some kind of evidence in the “battle” between science and religion, especially when it comes from such an otherwise erudite individual as A. C. Grayling, who, in the April issue of Prospect, castigates Frans de Waal for being so tolerant of religion. (One could doubtless make merry play with Grayling’s remark that the Copernican cosmology was ‘geocentric’, but no one can seriously doubt that Grayling knows very well that Copernicus put the sun, not the earth, at the centre of the universe – such a blunder simply reminds me of how horribly easy it is to commit howlers to print with a slip of the pen.)

Grayling’s attack rehearses all the familiar ‘new atheist’ condemnations of religion, including (I should really have thought this beneath him) references to fairies. Much of what he says is entirely fair, such as how deplorable are the religious fundamentalist attacks on science, women, homosexuality and civil freedoms in general. What I find so endlessly frustrating is the childish conviction among the new atheists that such things will evaporate if religion is ‘vanquished’ – a refusal, in other words, to see these things as expressions of power and prejudice for which various religions provide convenient justification. (I have just seen Richard Dawkins flogging on his website a T-shirt saying “Religion: Together We Can Find the Cure”, and I’m afraid my opinion of him fell several notches.)

The idea that they are putting the cart before the horse is not simply one they reject; it seems to infuriate them. It certainly infuriated Sam Harris, who said this in response to my own remarks to that effect:

“Who does Ball imagine the Taliban would be if they weren’t “Muslim extremists”? They are, after all, Homo sapiens like the rest of us. Let’s change them by one increment: wave a magic wand and make them all Muslim moderates… Now how does the world look? Do members of the Taliban still kill people for adultery? Do they still throw acid in the faces of little girls for attempting to go to school? No. The specific character of their religious ideology—and its direct and unambiguous link to their behavior—is the most salient thing about the Taliban. In fact, it is the most salient thing about them from their own point of view. All they talk about is their religion and what it obliges them to do.”

This is so characteristic of the new atheists in its implication that if one could (hey, literally!) wave a magic wand to wish away the ills of religion, all would be well. (And you know what the magic wand is? Reason! Because reason and religion cannot coexist in a single mind!) It’s a tragically naïve tautology: “if we could make these awful people nicer, they would be nicer.” What is particularly astonishing here is that Sam seems not to have realised that his nice Taliban are… still Muslims! In other words, presumably it was not religion per se that was making them this way, but something that was inducing them to interpret their religion in a punitive, intolerant and murderous way. Whatever that thing was, it was presumably not prescribed by the Quran, since as Sam admits, it is possible to interpret the Quran in a far more moderate way.

Now, I share what I perceive to be Sam’s frustration that religious texts, notoriously the Bible, are so contradictory that one can find in them justification for whatever views one prefers, whether as a Quaker pacifist or a member of the National Rifle Association. But this is the whole point: that one’s interpretation is therefore surely shaped by other factors, related to culture and history and doubtless also individual personality and upbringing. If religion is magicked away (and oh, it will have to be powerful magic), those factors are not going to vanish – as countless secular oppressive regimes show. But addressing predilections instilled by culture and history (for which, I freely admit, religion often functions as a brainwashing tool) is hard. Arguing that religion is a tissue of unsupportable beliefs about the nature of the physical universe is so, so much easier.

Many reviews of Grayling’s latest book The God Argument have implied that this debate has got very tired, and that the argument has moved on. I agree.

Beneath the surface

Next up, a review published in Nature of the latest book by Douglas Hofstadter. Interesting, but God it’s long.

_______________________________________________________________

Surfaces and Essences
Douglas Hofstadter & Emmanuel Sander
Basic Books, 2013
608 pages
$35.00
ISBN 978-0-465-01847-5

I finished this review and stored the file in the 'Nature' folder on my desktop, then emailed it to the editor. Or did I? A file, after all, was once a sheaf of papers, and a folder a cardboard sleeve for holding them. A desktop was wooden, and mail needed a stamp (no, it needed a little piece of adhesive paper). But all I did was use an interfacing device (named for the most superficial resemblance to a rodent) to rearrange the settings of some microprocessor circuits. To see that almost everything we say and do refers by analogy to other things we or others have once said or done – which is the main point of Surfaces and Essences – there is no better illustration than the way we have constructed our computer software as a conceptual and visual simulacrum of the offices our grandparents knew.

On the one hand this is kind of obvious. Why (science fiction writers take note) would we invent new categories and labels for things when we can aid comprehension by borrowing old ones, even if the physical resemblance is negligible? What cognitive scientists Douglas Hofstadter Hofstadter and Emmanuel Sander set out to show, however, is that this sort of elision is not merely a convenience: all our thinking depends on it, from the half-truths of everyday speech ("that always happens to me too!") to the most abstruse of mathematical reasoning. I was convinced, and the ramifications are often thought-provoking. But when you have had authors telling you the same thing again and again for 500 pages, perhaps you’ll believe it whether it's true or not. I’ll come back to that.

Hofstadter is famous for his earlier, Pulitzer-prize-winning treatise on how we think, Gödel, Escher, Bach (1979). Fans of that dazzling performance might find this book surprisingly sober, but it is also lucid and, page for page, a delight to read. Whether there is any conceptual continuity between that and this vision of how we think is debatable, except perhaps that GEB’s delight in puns here becomes an assertion that pretty much all our cognition depends on punning elevated to analogy.

The claim that drawing parallels between one thing and another central in our thinking seems obvious in art: analogies are the bread and butter (there we go again) of the visual, literary and theatrical arts. (Of these, the authors seem curiously unconcerned about anything except poetry.) Yet Hofstadter and Sander are really inverting that usual picture: it is precisely because the brain seems to be an analogy machine that art is possible and meaningful.

They focus most on the use of analogy in language. Moving steadily from words to phrases and narratives, they show just how deeply embedded is our tendency to generalize, compare, categorize, and forge links. Individual examples seem trivial until you realise their ubiquity: tables have legs, melodies are hauting, time is discussed in spatial terms, and idioms are invariably analogical, if you get my drift. Thus the lexical precision on which dictionaries seem to insist is illusory – words are always standing in for other words, their boundaries malleable. This flexibility extends to our actions: we see that a spoon can serve as a knife when no knife is available. (Indeed, the spoon then becomes a knife – objects may be fixed, but their labels aren’t.)

These arguments can be carried too far. Is to extrapolate to make an analogy (I expect the future to be like the past)? Is a Freudian slip an analogy, or mere crosstalk of neural circuits? Is convention an analogy (why don’t we write mc2=E?). Can we, in fact, turn any mental process into an analogy, by that very process of analogy? These are not rhetorical questions, for one might at least examine whether the same neural circuitry is involved in each case. But a lack of interest in neuroscientific examination of their idea is another of the book’s odd lacunae.

In fact this intriguing, frustrating book seems to exist almost in an intellectual vacuum. Unless one combs through the bibliography, one could mistakenly imagine that it is the first attempt to explore the notion of analogy and metaphor in linguistics, overlooking the work of Raymond Gibbs, Andrew Ortony, Sam Glucksberg, Esa Itkonen and many others. And one is forced to take an awful lot on trust. Hofstadter and Sander describe, for example, the evolution of the concept of ‘mother’ in the mind of a child as he learns to generalize from experience. It all sounds plausible, but the authors offer no empirical evidence for the developmental pathway they describe.

Neither is there any real explanation of why we think this way. Isn't it perhaps, in part, a way of minimizing the mental resources we need to engage in a situation, to avoid having to start from scratch with every unfamiliar encounter, object or perspective? Is it an adaptive technique for making predictions? Are mirror neurons part of a built-in cognitive apparatus for analogizing ourselves into others’ shoes?

The lack of historical perspective is also a problem – it is as if people always thought like they did now. Analogy was arguably all we once had for navigating experience, for example in the Neoplatonic idea of correspondences: “as above, so below.” This “just as… so…” thinking remains at the root of pseudoscience: the Moon influences the tides, so why not our body fluids? So how do we distinguish between good and bad analogies?

There are gems of insight in here, but again flawed by the authors’ relaxed attitude towards evidence. An analysis of Einstein's thought is splendid, explaining what is missing from conventional accounts of the discoveries of light quanta, relativity and mass-energy equivalence, namely what qualities distinguish Einstein from his peers. These qualities are convincingly shown to be analogical: Einstein was able to take leaps of faith and make connections that postpone rigour and are certainly not self-evidently true. One would usually call this intuition - Einstein's friend and biographer Banesh Hoffmann did just that. But it is shown here to be intuition based on a conviction that different areas of physics were comparable. In other words, his intuition is not left ineffable but is taken apart so that the inner workings – some of them – can be seen. As a result, we see that Einstein's insights were very subtle and not self-evidently true. Analogies, the authors say, left Einstein like J. S. Bach on hearing a theme: "very quickly able to imagine all of its possible consequences." All very fine – but such a detailed account must surely be supported by Einstein’s own words. Almost none are offered; we get only fragments of Hoffmann’s commentary.

Maybe at least some of these questions are merely evidence of the fecundity of the authors’ thesis. But they’d have more excuse for not answering them if they did not fill so much space with endless examples to ram the point home: they never give one when 60 will do, and I’m not exaggerating.

Such things make me wonder whom the book is for. Academic linguists will be irritated by the absence of references to other work. Physical scientists aren’t indulged until page 450. General readers could have been given the basic ideas, with equal conviction, in half the length, and will occasionally get the feeling they have been led along and then dumped. The thesis suggests no obvious mode of further development, no manner of testing and probing. It remains stimulating, but less would certainly have been more.

Happy holidays

Here’s the previous piece for my Under the Radar column on BBC Future – there will be another column up very shortly. Peter Dodds, tells me that he and his colleagues have now created a “hedometer” site at http://www.hedonometer.org that will “provide a real-time measure of happiness that will be useful for many entities including governments at all scales, journalists, analysts, and citizens.” Peter adds that “initially, we'll be showing an interactive happiness time-series for Twitter but we'll be expanding to geography, social networks, etc., as well as other languages and other emotions.” It sounds rather fabulous, and will be free and open to all users when it goes live tomorrow.

________________________________________

Feeling low? Over-worked, anxious, bored with life? A holiday will do your mood the world of good. Really it will: there’s now scientific proof. A team of researchers at the University of Vermont in the United States has found that tweets contain significantly more happy words the further from home they are sent [1].

This is the latest dispatch from an emerging discipline in which social-networking media are mined to gauge people’s moods and opinions. Twitter is one of the most fertile sources of information for this kind of study, partly because the comments are less guarded and self-conscious than responses to questionnaires (the social scientist’s traditional means of sampling opinion) but also because huge amounts of data are available, with automatically searchable content. What’s more, Twitter feeds sometimes come accompanied with useful information such as the tweeter’s profile and location.

Previous studies in “twitteromics” have, for example, monitored the spread of news, the demographics of different languages, and the correlations between obesity and expressions of hunger in particular populations. Since public mood changes such as brewing social unrest will show up on Twitter and other social media, governments, police forces and security organizations are showing an increasing interest in twitteromics, raising questions about the right balance between privacy and security. Meanwhile, potential insights into the emergence and propagation of trends are a gift to company marketing departments.

The new study of the link between happiness and geographical location by Christopher Danforth and colleagues at Vermont takes advantage of the “garden hose” public-access feed for Twitter, which makes freely available a random 10 percent of all messages posted. This provided the researchers with four billion tweets for the year 2011 to analyse.

Since Danforth and colleagues were interested in how the mood expressed in the messages correlated with the location from which they were sent, they sifted through this immense data set to pick out those tweets that were accompanied by the precise latitude and longitude of the sender’s mobile phone – a facility optionally available for tweets, which uses the Global Positioning System (GPS) to locate the message’s origin within a 10m radius. About 1% of the messages included this information, giving a data set of 37 million messages sent by more than 180,000 individuals from all over the planet.

But identifying where the sender is situated doesn’t in itself reveal what the researchers wanted to know. They were interested in how the message content varied with distance from home. How could they know where ‘home’ was?

It turns out that positional information disclosed by our mobile phones reveals this pretty clearly. In 2008 a team of researchers in the US used the locations of mobile phones – recorded by phone companies whenever calls are made – to track the trajectories of 100,000 (anonymized) individuals [2]. They found that, as we might imagine, we tend to return over and over again to certain places, especially our homes and workplaces, and only rarely venture very far from these locations.

In much the same way, Danforth and colleagues could figure out the most common locations for each individual in their survey, along with an associated number describing how widely the person tended to roam from those places. They found that people generally have two such preferred locations, just a short distance apart, which they attributed to the home and workplace.

How, then, do the messages differ when individuals are at home, at work, or further away? To assess the ‘happiness’ of a tweet, the Vermont team has developed what they call a ‘hedonometer’ [3]: an algorithm that searches the text for words implying a positive or enjoyable context (such as ‘new’, ‘great’, ‘coffee’ and ‘lunch’) or a negative one (‘no’, ‘not’, ‘hate’, ‘damn’, ‘bored’). On this basis the hedonometer assigns each message a happiness score.

The authors report that “we see a general decline in the use of negative words as individuals travel further from their expected [home] location”. More precisely, the average happiness score first declines slightly for distances of around 1 km – the kind of distance expected for a short commute to work – and then rises steadily with increasing distances of up to several thousand kilometres. What’s more, individuals with a larger typical ‘roaming radius’ use happy words more often – a result that probably reflects the higher socioeconomic status of such jet-setting types.

So it seems we’re least happy at work and most happy when we are farthest from home. At least, that’s the case for the roughly 15% of American adults who use Twitter, or to be even more cautious, for the English-speaking subset of those who chose to ‘geotag’ their tweets. One key question is whether this sample is representative of the population as a whole – Twitter is less used among older people, for example. It’s also an open question whether ‘happy words’ are a true indicator of one’s state of mind – are you less likely to tweet about your holiday when the weather is awful and the family is fractious? But such quibbles aside, you might want to consider that costly flight to Bermuda or Kathmandu after all.

References
1. M. R. Frank, L. Mitchell, P. S. Dodds & C. M. Danforth, preprint http://www.arxiv.org/abs/1304.1296 (2013).
2. M. C. Gonzalez, C. A. Hidalgo & A. L. Barabasi, Nature 453, 779-782 (2008).
3. P. S. Dodds, K. D. Harris, I. M. Kloumann, C. A. Bliss & C. M. Danforth, PLoS ONE 6(12), e26752 (2011).

Can Google predict the markets?

Here’s another Nature news story. I’ll be interested to see what other media outlets make of it.

____________________________________________________________

Traders reveal their mood in the search terms they use.

Suppose you had a direct line into the minds of stock market traders. Would you be able to predict which investment decisions they will take, and thus anticipate the markets?

A team of researchers in the UK and US now suggests that such a crystal ball might exist, in the form of the search terms recorded and made publicly available by Google Trends. Tobias Preis of the University of Warwick Business School and his colleagues say that their analyses of Google Trends data show “early warning signs” of how the markets will shift – including the financial crash of 2008 [1].

Don’t, however, imagine that this is the way to make a fast buck. It’s one thing to offer a retrospective account of why markets behave as they do – which is what Preis and colleagues have done – and quite another to provide a genuinely predictive tool.

That’s why the work is “interesting but not earth-shattering”, in the view of British economist Paul Ormerod of the consultancy Volterra Partners in London.

Mathematical physicist Didier Sornette of the Swiss Federal Institute of Technology (ETH) in Zürich agrees, pointing out that the predictive power of the strategies the authors deduced from Google Trends data are only slightly better than predictions which assume traders make random decisions. “No investor or hedge-fund would be interested in such a strategy”, he says.

The predictive value of Google Trends has been demonstrated in other areas of social science. Most famously, outbreaks of influenza have been seen emerging in real time by monitoring the numbers of Google searches for terms related to flu prevention and cure [2].

The potential of using such information to study economic behaviour has already been spotted. Preis and coauthor Gene Stanley of Boston University have themselves shown that certain search terms reflect the volume of stock market transactions [3]. Sornette, in collaboration with Japanese economists, has found that the volatility (fluctuations) of financial markets can be correlated with the prevalence of particular topics in business news [4].

But what traders and investors really want is a method not just to assess the current state of markets but to anticipate their future course. In particular, episodes of instability, such as the financial crisis of 2008, are often preceded by periods of concern during which investors avidly seek information to decide whether to buy or sell.

Preis and colleagues figured that such anxieties and moods might be signaled by Google search terms. Just before the onset of the latest crisis, for instance, “debt” might be expected to feature prominently. That’s just what the researchers found.

To test if such correlations could be made predictive, they devised trading strategies in which a decision to buy or sell is linked to the recent prevalence of particular search terms. They simulated how these strategies would have performed between 2004 and 2011 based on real data from the financial markets.

Of the 98 ‘Google Trends’ strategies the researchers explored, that based on “debt” performed best. By 2011 it would have increased the value of a portfolio by more than 300 percent, compared with just 16 percent for a common conventional investment strategy.

Although this sounds impressive, the relevance of a predictive Google search term isn’t always clear. The second-best strategy, for example, was linked to “color”, and the fourth best to “restaurant”.

Even the use of “debt” is not obvious, since its role in the financial crash was apparent only as it happened. “How would they know in advance that they should use ‘debt’?” asks Sornette.

“In retrospect it is always possible to derive what appear to be highly successful trading strategies”, says Ormerod. “But what we want is to be able to do that before the event, not after.”

What’s more, economists acknowledge that any transparently profitable strategy for playing the markets will quickly lead to a change of trader behaviour that cancels it – a principle called Goodhart’s Law, after the British economist Charles Goodhart. “Social systems have the complication that the system may directly react to predictions being made about its behaviour”, coauthor Susannah Moat of University College London agrees.

The researchers suggest, however, that a key outcome of their approach might be to elucidate the psychological mechanisms that guide traders to their decisions, which could be encoded in their information-gathering. “Stock market data themselves tell us little about how traders make decisions”, says Preis.

“We think that the overall pattern we observe may reflect loss aversion”, he adds – the fact that humans are more concerned about losing money than they are about missing an opportunity to gain the same amount.

References
1. Preis, T., Moat, H. S. & Stanley, H. E. Nat. Sci. Rep. 3, 1684 (2013).
2. Ginsberg, J. et al. Nature 457, 1012–1014 (2009).
3. Preis, T., Reith, D. & Stanley, H. E. Phil. Trans. R. Soc. A 368, 5707–5719 (2010).
4. Hisano, R. et al., preprint http://www.arxiv.org/abs/1210.6321 (2012).

Crowdsourcing in manhunts can work

So much to post right now… but I will start with the easy stuff. Here is a news story for Nature on a preprint that seemed almost too topical to be true.

________________________________________________________________

Despite mistakes over the Boston bombers, social media can help to find people quickly.

The social news website Reddit was left red-faced after mis-identifying the suspects for the Boston marathon bombings last week, raising questions about whether crowd-sourcing to gather information might do more harm than good in such situations.

But work by a team of scientists from the United Arab Emirates and coworkers in the US and UK offers a more upbeat message about the potential of social media to assist in crime investigations and societal searches. Last year they enlisted communities on networks such as Twitter and Facebook to look for five people in different cities around the world, and were able to find three of them within the 12-hour deadline imposed [1].

In a new preprint, the researchers now analyse the behaviour that made this possible. They say that participants responded to the urgency of the search not by sending out messages to their contacts in an indiscriminate, blind panic, but by becoming even more focused and directed about whom they contacted [2].

The experiment, by computer scientist Iyad Rahwan at the Masdar Institute of Science and Technology in Abu Dhabi and his colleagues, constituted the team’s entry in the Tag Challenge staged by the US State Department in March 2012. The Tag Challenge required teams to find individuals (posing as jewel thieves) in New York City, Washington DC, London, Stockholm and Bratislava within 12 hours. Participants were given only a ‘mug shot’ of each target wearing a T-shirt bearing the competition logo, released on the morning when the competition started.

Rahwan’s team used crowd-sourcing to find the targets, offering cash incentives to individuals for uploading photos of suspects to web and cell phone apps and for recruiting more searchers. Although they failed to locate the targets in London and Stockholm, the team out-performed all others and won the competition [1].

The results showed that “among this noisy stream of tweeting and retweeting, of news articles and messages being fired off to acquaintances around the globe, people are able to efficiently guide a message towards a target in a particular city”, says Rahwan’s colleague Alex Rutherford at the Masdar Institute.

The new analysis of the information provided by participants shows that communications such as tweets became more specific and targeted as the day of the competition approached, being increasingly directed towards other users in the target cities. “Despite increasing time pressure, and its associated cognitive load, people actually became more selective in their recruitment of others, making sure information is directed in an intelligent manner”, says Rahwan.

“This makes good sense to me, and it's what I would have expected”, says Peter Sheridan Dodds of the University of Vermont. In 2003 Dodds and coworkers conducted a social-search experiment [3] to route emails to a few target people worldwide – an electronic version of the famous ‘small-world’ experiment by psychologist Stanley Milgram in 1967, in which he asked random people to forward letters to addressees identified only by name, profession and city [4]. “In our small-world experiment we found that successful searches were much more focused than unsuccessful ones and less likely to involve scattershot, connect-to-everyone attempts,” Dodds says.

Defence and security organizations have a growing interest in these outcomes. In 2009 the US defence research agency DARPA staged the Red Balloon Challenge, in which competitors were challenged to locate ten red weather balloons tethered at random locations all over the US. That challenge was won by a team at the Massachusetts Institute of Technology led by Manuel Cebrian, who collaborated with Rahwan and colleagues for the Tag Challenge. The MIT team found the red balloons in 9 hours by harnessing social-networking media [5,6].

How does all this reflect on the search for the Boston bombers? Last week, Reddit users, acting on photographs of suspects posted by the FBI, collectively pointed the finger at several individuals who had nothing to do with the bombings, including an innocent student from Brown University. The eventual arrest of Dzhokhar Tsarnaev, who previously escaped with wounds after a shootout with police, came after a neighbour spotted blood on the tarpaulin covering the boat in which he was hiding.

“The Boston manhunt is an example of how things can go wrong”, says Rahwan. “It appears that information was very much misdirected. This may be, in part, due to the high profile of the event, which led everyone to want to help even if they were incapable or misinformed.”

“There may be a tradeoff between mass mobilization and effective mobilization of a more specialized group of reliable and well-informed individuals”, he adds. “Having too many people involved might actually make things worse.” He and his colleagues have begun to explore schemes for supporting the checking and verification of crowd-sourced reporting [7].

Even President Barack Obama has commented on the hazards of search efforts like those on Reddit. “Crowd-sourcing via social media can be incredibly powerful in mobilizing people”, says Rahwan, “but it is not a silver bullet.”

“I think the web-enhanced ‘collective detective’ is potentially very powerful and is here to stay”, agrees Dodds. “But we have to ensure that the distributed social search is always used for good, meaning for example that ‘bad’ actors cannot corrupt the search, and that good, well-intentioned actors are prevented from collectively generating errors leading to witch hunts.”

“There's a lot of wisdom in the crowd when people are actually aggregating independent pieces of information”, says David Liben-Nowell of Carleton College in Northfield, Minnesota, a specialist on the searching of social networks. “But when purported information is amplified and echoed by people without truly independent information being collected, as seemed to happen in the Reddit case, then we may end up with the folly of the mob instead.”

In situations like that, says Dodds, “unofficial efforts are very important, but the onus is now on governments to create and maintain distributed social search sites that allow the public to aid in finding people. A system should already be in place that is transparent and sophisticated, and that allows for the public to provide analysis, not just photos.”

“It's not just for finding bad guys”, he adds. “Missing children are an obvious example.”

“With any kind of task like this one, we have to accept that there's a tradeoff between the risk of a false negative and a false positive”, says Liben-Nowell. “As a society, we have to think carefully about where we want to be in that spectrum.”

References
1. Rahwan, I. et al., IEEE Computer April, 68-75 (2013).
2. Rutherford, A. et al., preprint http://www.arxiv.org/abs/1304.5097.
3. Dodds, P. S., Muhamad, R. & Watts, D. J. Science 301, 827-829 (2003).
4. Milgram, S. Psychology Today 61, 60-67 (1967).
5. Tang, J. et al., Commun. ACM 54, 78-85 (2011).
6. Rutherford, A. et al., Proc. Natl Acad. Sci. USA 110, 6281-6286 (2013).
7. Naroditskiy, V., Rahwan, I., Cebrian, M. & Jennings, N. R. PLoS ONE 7, e45924 (2012).

Scooped

Here’s a piece I wrote for BBC Future this week, before discovering that the blighters on their science news desk were covering the work already. So there will be something else from me on Under the Radar later this week…

______________________________________________________________

Attempts to measure and define intelligence are always controversial and open to interpretation. But none, perhaps, is quite as recondite as that now proposed by two mathematical physicists. They say that there’s a kind of rudimentary intelligence that comes from acting in a way that maximizes your future options.

Alex Wissner-Gross of Harvard University in Cambridge, Massachusetts, and Cameron Freer of the University of Hawaii at Manoa have figured out a ‘law’ that enables inanimate objects to behave this way, in effect allowing them to glimpse their own future. If they follow this law, they can show behaviour reminiscent of some of the things humans do: for example, cooperating or using ‘tools’ to conduct a task.

The researchers think that their mathematical principle might help to provide a “physics of intelligence”: an explanation of smart actions rooted in the laws of thermodynamics.

Central to their claim is the concept of entropy. Popularly described as a measure of disorder, entropy more properly describes the number of different equivalent states a system can adopt. Think of a box full of gas molecules. There are lots more ways that they can disperse uniformly throughout the available space than there are ways they can all congregate in one corner. The former situation has greater entropy.

In principle, either arrangement could arise purely from the random motions of the molecules. But there are so many more configurations of the uniformly spread gas that it is much more likely, and in practice we never see all the gas shift into one corner. This illustrates the second law of thermodynamics, which states that the total entropy of the universe always increases – simply because that’s more probable than the alternatives.

Some scientists have generalized this idea to propose that all processes of change happen in a way that has the greatest rate of entropy production. Not only do things head for the highest-entropy state, but they do so along a route that produces entropy at the greatest rate. There’s no rigorous proof that all things must happen this way, but the hypothesis of maximum entropy production has been used to account for processes such as the appearance of life, and also to design artificial-intelligence strategies that allow computers to become adept at complex games such as Go.

Wissner-Gross and Freer wondered if this hint at a link between maximal entropy production and intelligence could be made more concrete. They hit on the idea that ‘true’ intelligence is not, as they put it, “just greedily maximizing instantaneous entropy production”, but involves foresight: looking for a path that maximizes its production between now and some distant time horizon. For example, a good computer algorithm for playing Go might seek a strategy that offers the player the greatest number of options at all points into the future, rather than playing itself into a corner.

But how would an inanimate particle find that strategy? The researchers show that it can be defined via a mathematical expression for what they call the ‘causal path entropy’: the entropy production for all possible paths the particle might take. How would a particle behave if governed by the law that it must, at every instant, maximize this casual path entropy – which means, in effect, planning ahead?

Objects whose motions are guided solely by the conventional laws of motion are doomed to a blind, dumb presentism – they just go where the prevailing forces take them. Think again of those gas molecules in a box: each particle wanders aimlessly in a random walk, exploring the confining space without prejudice.

Yet when Wissner-Gross and Freer impose on such a meandering particle the demand that it move in a way that maximizes the casual path entropy, its behaviour is quite different: it tends to hover around in the centre of the box, where it suffers the least constraints on its future motion. They then explored the consequences of their new law for a so-called ‘cart and pole’ – a pendulum attached to a mobile cart, which can be stabilized in an inverted, head-up position by moving the cart back and forth, like balancing a stick on your palm. Early hominids are thought to have mastered such a delicate balancing act when they learnt to stand upright – and it’s a trick achieved by a cart-and-pole obeying the ‘maximum causal entropy’ law.

Weirder things become possible too. Wissner-Gross and Freer looked at a system composed of three disks in a box: a large one (I), a small one (II), and another small one (III) trapped inside a tube too large for I to enter. Suppose now that the movements of disk I are dictated by causal entropic ‘forcing’. In this case, the disk conspires to collide with II so that II can bounce into the tube and eject disk III. Liberating III means that the disks now have more ways to arrange themselves than when it was confined – they have more entropy. But to gain access to that entropy, disk I essentially uses II as a tool.

Similarly, two small disks governed by the causal entropic force showed a kind of social collaboration to collectively drag down a large disk into a space where they could ‘play’ with it, offering more possible states in total – another behaviour that looks strangely ‘intelligent’.

In these cases there is no real reason why the particles should be controlled by the causal entropic force – the researchers just imposed that property. But they suggest that, in a Darwinian evolving system, objects that are able this way to ‘capture’ a greater slice of the future might gain an adaptive advantage, so that such a force might be naturally selected. Not only could this offer clues about the emergence of intelligent, adaptive behaviour in the living world, but the general principle might also be useful for designing artificial intelligent systems and perhaps even for understanding problems in economics and cosmology.

Reference
A. D. Wissner-Gross & C. E. Freer, Physical Review Letters 110, 168702 (2013).