Thursday, February 24, 2011

A metaphor too far

I have a Muse on Nature’s online news about metaphor in science; here’s the pre-edited version. In this huge and complex topic, this piece is a drop in the ocean.

Are scientists addicted to using metaphorical imagery at the cost of misleading the public and themselves?

Metaphors influence the way we think. In a recent paper in PLoS ONE, Stanford psychologists Paul Thibodeau and Lera Boroditsky show that how people judge the appropriate response to crime differs significantly when it is presented as a ‘beast’ or a ‘virus’ ravaging society [1]. In the former case they were more likely to call for stronger law enforcement, whereas in the latter there was more openness to solutions involving reform and understanding of root causes.

Perhaps the most striking aspect of this study is that the participants were unaware of the role the metaphorical context was playing. Instead they found ways to rationalize their decision based on apparently objective information such as statistics. “Far from being mere rhetorical flourishes”, the researchers say, “metaphors have profound influences on how we conceptualize and act with respect to important societal issues.”

To have this demonstrated and quantified is valuable – but perhaps mostly because it underlines what politicians and their advisers have never doubted. If there is a spin doctor or speechwriter who does not already recognize that metaphors sway opinion, it is a mystery how they ever got the job.

It isn’t hard to see why ‘crime as wild beast of prey’ encourages people to think about how to cage or kill it, whereas ‘crime as virus’ fosters more eagerness for ‘scientific’ understanding of causes. But too rarely are such metaphors interrogated at a deeper level.

In both the cases here, crime is presented as a (malevolent) force of nature, outside human agency. Whether beast or virus, the criminal is not like us – is not in fact human. By the same token, a ‘war on drugs’ or a ‘war on terror’ not just is an emotive image but deploys a narrative that bears little relation to reality.

In literature metaphor serves poetic ends; in politics it is a (subtly manipulative) argument by analogy. But in science, metaphor is widely considered an essential tool for understanding. So where then does this latest work leave us?

While the example of crime here imputes natural agency to human actions, science generally invokes metaphors the other way around: natural processes are described as if resulting from intention. This anthropomorphizing tendency was called the ‘pathetic fallacy’ by the nineteenth-century critic John Ruskin, though it was noted two centuries earlier by Francis Bacon.

It is an ingrained and profoundly influential habit, especially in biology [2-6], where intimations of intelligent agency seem irresistible even to those who deplore them. Most famous in this respect is Richard Dawkin’s selfish gene. Given the idea Dawkins strove to convey in his 1976 book of that title, the metaphor seems apt and understandable almost to the point of inevitability. But its problems go well beyond the fact that genes are of course not selfish in the way that people are (which is to say, they are not selfish at all).

For the selfish gene props up the whole notion of a Darwinian world as uncaring to the point of being positively nasty: an image that has sometimes provoked resistance to the sciences in general and natural selection in particular. And as physiologist Denis Noble has compellingly argued, the idea that genes are ‘selfish’ is totally unnecessary for understanding how they work, and in some ways misleading [7].

But it is no better to talk instead of the ‘cooperative gene’, which is equally value-laden and misinforming. Genes are not selfish or cooperative any more than they are happy or short-tempered. The central problem here is that of scientific metaphor in general [8,9].

Books of life, junk DNA, DNA barcodes – all can and have distorted the picture, not least because sometimes scientists themselves start to forget that these are metaphors. And when the science moves on – when we discover that the genome is nothing like a book or blueprint – the metaphors tend nonetheless to stick. The more vivid they are, the more dangerously seductive and resistant to change.

Thibodeau and Boroditsky give us new cause to be wary, for they show how unconsciously metaphors colour the way we reason. This seems likely to be as true in science – especially a science as emotive as genetics – as in social and political discourse.

Most scientists would probably agree with physiologist Robert Root-Bernstein that ‘metaphors are essential to doing and teaching science’ [10]. They might sympathize with biologist Paul Hebert’s response to criticisms of his ‘DNA barcoding’ metaphor [11]: “Why want to be so scientifically proper as to make our science tedious?” [12]

But the need for metaphor in science stands at risk of becoming dogma. Maybe we are too eager to find a neat metaphor rather than just explaining what is going on as clearly and honestly as we can. We might want to recognize that some concepts are “a reality beyond metaphor”, as David Baltimore has said of DNA [13]. At the very least, we might admit metaphor into science only after strict examination, and heed the warning of cyberneticists Arturo Rosenblueth and Norbert Wiener that “the price of metaphor is eternal vigilance” [14].


1. Thibodeau, P. H. & Boroditsky, L. PLoS ONE 6, e16782 (2011).
2. D. Nelkin, Nat. Rev. Genet. 2, 555-559 (2001).
3. B. Nerlich, R. Elliott & B. Larson (eds), Communicating Biological Sciences (Ashgate, Farnham, 2009).
4. B. Nerlich, B. & Dingwall, R., in Cognitive Models in Language and Thought: Ideology, Metaphors and Meanings (eds R. Dirven, R. Frank & M. Pütz), p.395–428. (Mouton de Gruyter, Berlin, 2003).
5. Kay, L. E., Who Wrote the Book of Life? (Stanford University Press, Stanford, 2000).
6. E. F. Keller, Refiguring Life (Columbia University Press, New York, 1996).
7. D. Noble, The Music of Life (Oxford University Press, Oxford, 2006).
8. G. Lakoff & M. Johnson, Metaphors We Live By (University of Chicago Press, Chicago, 1981).
9. T. L. Brown, Making Truth: Metaphor in Science (Univeristy of Illinois Press, Urbana, 2003).
10. R. Root-Bernstein, Am. Scient. 91(6) (2003).
11. P. Hebert, Proc. R. Soc. B Biol. Sci. 270, 313-321 (2003).
12. Quoted in ref. 3, p.161.
13. Quoted in ref. 3, p.158.
14. Quoted in R. C. Lewontin, Science 291, 1263-1264 (2001).

Thursday, February 17, 2011

Fruit flies sniff out heavy hydrogen

Here’s my latest news article for Nature. It’s worth checking out the comments on the Nature site.

Insects' ability to discriminate isotopes reignites debate over a controversial theory of olfaction.

Fruit flies can smell the difference between ordinary and heavy hydrogen, according to new research published today.

Efthimios Skoulakis of the Alexander Fleming Biomedical Sciences Research Centre in Vari, Greece, and his colleagues say that fruit flies show a preference for an odorant molecule containing ordinary hydrogen over the same molecule with the hydrogen replaced by heavy hydrogen (deuterium), when presented with both odorants in the two branches of a T-shaped maze.

The flies can also be conditioned to display a selective aversion to either of the forms of the odorant by electric-shock treatment, showing that they can clearly distinguish between them. The researchers report their findings in the Proceedings of the National Academy of Sciences USA [1].

Skoulakis and colleagues say that the results offer strong support to a controversial theory of how olfaction works, which has been proposed previously by Luca Turin of the Massachusetts Institute of Technology, who is also an author of the paper. According to Turin, odorants are identified by the olfactory apparatus not according to their molecular shape but their vibrations.

“This is an important paper, and offers very strong evidence in favour of the vibrational theory of olfaction”, says materials physicist Andrew Horsfield of Imperial College in London.

But others are not convinced. Leslie Vosshall, a neuroscientist specializing in olfaction at the Rockefeller University in New York, considers it interesting that flies show such discrimination, but adds that “these findings by themselves do not provide strong support for any of the prevailing models of smell.”

Deuterium is an isotope of hydrogen: unlike ordinary hydrogen, its atoms contain a neutron in the nucleus as well as a proton. This makes the atoms roughly twice as heavy. The chemical properties of deuterium are much the same as those of ordinary hydrogen, but its greater mass means that when the atoms are bonded to others in a molecule, they vibrate more slowly.

In the predominant theory of olfaction, odorant molecules dock into cavities in receptor proteins lodged in the olfactory membranes. This docking depends on a match between the shape of the odorant and that of the cavity; if they fit together, this triggers a neural signal to the brain.

But Turin thinks that instead the receptor proteins ‘sense’ the vibrations of the odorant, an effect made possible by the quantum-mechanical behaviour of electrons in the molecules. Horsfield and others have shown that this process could work in theory [2], but there is no direct evidence for it in practice.

If Turin is right, deuterium-substituted odorants should smell different to those with ordinary hydrogen because they have different vibration frequencies.

There is not yet any good evidence that deuterated compounds smell different to humans [3], but subtle biases are hard to eliminate from such tests. That’s why Turin teamed up with Skoulakis to test fruit flies, which are less susceptible to biases and are known to have a good sense of smell.

When presented with the attractive (to flies) odorant acetophenone, the fruit flies showed an increasing aversion to it as more of its hydrogens were substituted for deuterium. The researchers could train the flies to associate either the deuterated or normal odorant with punishing electric shocks applied to their feet via the floor of the maze, and to avoid them accordingly.

If the vibrational mechanism of smell is correct, the researchers reasoned that flies trained to avoid deuterated odorants should display a similar aversion to compounds called nitriles, since the vibration of the nitrile chemical group has a very similar frequency to that of the bonds between deuterium and carbon. They found this was so.

But Bill Hansson, a specialist in insect olfaction at the Max Planck Institute for Chemical Ecology in Jena, Germany, isn’t persuaded. He points out that, although most isotopes are chemically identical, this is not always the case with hydrogen and deuterium, given their large (2:1) difference in mass. After all, heavy water is toxic, and even in these odorants the substitution of deuterium changes properties such as melting and boiling points.

“If hydrogen bonds between the odorant and corresponding receptor play a major role, insects may well be able to discriminate between deuterated and non-deuterated compounds using conformational [shape-based] sensing”, he says.

Vosshall is also sceptical. “Insects use odorant receptors that are structurally and functionally distinct from these human receptors, yet this group claims that the same vibration mechanism operates in these very distinct proteins”, she says. “This idea is difficult to reconcile with the current knowledge of how these completely divergent protein types detect odors.”

Regardless of the mechanism, might humans discriminate isotopes by smell too? “Extrapolation to humans has to be treated with care”, Horsfield warns. Turin has, however, received unpublished reports of isotopic smell discrimination in dogs. “In one case at least the dogs are said to completely ignore the deuterated version of an odorant that they are trained to detect in the undeuterated version”, he says.

“Things are unlikely to work exactly in the same way for humans”, he acknowledges. But he is convinced that something analogous applies.


1. Franco, M. I., Turin, L., Mershin, A. & Skoulakis, E. M. C. Proc. Natl Acad. Sci. USA details to come.
2. Brookes, J. C., Hartoutsiou, F., Horsfield, A. P. & Stoneham, A. M. Phys. Rev. Lett. 98, 038101 (2007).
3. Keller, A. & Vosshall, L. B. Nat. Neurosci. 7, 337-338 (2004).

Thursday, February 10, 2011

Talk about Unnatural

While they last: I discuss Unnatural on the Guardian books podcast and on BBC Radio 4’s Today programme last Tuesday. 

Monday, February 07, 2011

Fears for tears

Here’s my latest Crucible column for Chemistry world. Weird stuff, huh?

There is an early candidate for this year’s Ig Nobel prize in chemistry, one of the annual spoof awards for ‘improbable research’. The work reported in Science by Shani Gelstein of the Weizmann Institute of Science in Rehovot, Israel, and colleagues1 has precisely the degree of risqué unlikelihood that the Ig Nobel committee clearly enjoys. It is not to descend too far into tabloid sensationalism to describe the findings thus: men say they don’t feel much like having sex after sniffing women’s tears.

To judge a piece of work worthy of an Ig Nobel is not necessarily to denigrate it, and indeed I’d argue that the research by Gelstein and colleagues raises interesting and significant questions. Several Ig Nobel laureates have investigated problems of genuine value: in one of my favourites, chemical engineers Ed Cussler and Brian Gettelfinger looked at whether the theoretical viscosity scaling relationships for drag and thrust in swimming motions are borne out experimentally by having people swim in a pool filled with syrup2. Others – and I believe Gelstein et al. fall into this category – look odd, even perverse, merely because odd and perverse things happen in nature.

That is plain from the context of the work, which confronts us with the astonishing fact that we do not know why we cry. Anyone tempted to sneer at the motivation of the Israeli team should be silenced by this stark truth. It is not hard to devise stories about the adaptive value of tears – one such invokes their potential to prevent dehydration of mucous membranes while weeping3 – yet far harder to adduce any proof. The odd thing about tears as an emotional signal is that they seem to be purely symbolic, whereas Darwin imagined that such signals must have (or have had) some functional role too – the baby’s cry broadcasts its distress, say.4

Tears are surprisingly complex structures, an investment that surely must have some payoff. They are not just salty water, but contain enzymes and other proteins, lipids and metabolites. The lipids self-organize into a surface film in which two-dimensional crystal and liquid patches interact to create resilience as the droplets deform.5 Emotional tears of humans have a slightly different composition from mere ‘eye-watering’ tears; and noting that mouse tears contain a pheromone, Gelstein and colleagues wondered if this might be true of human emotional tears too.

To study that, they harvested ‘sad tears’ from women watching weepy films, and investigated whether men could smell any difference between these and saline solution. They couldn’t. But then the researchers showed the male subjects images of women’s faces while constantly exposing them to the vapours of the tears by attaching a tear-soaked pad beneath their nostrils, and asked the men to assess the sadness and the sexual attractiveness of the images.

As psychological testing goes, this seems to be heading into strange territory. But the results were surprising: while tears did not influence judgements of sadness, they lowered significantly ratings of attractiveness. In related tests, the men reported lower sexual arousal after sniffing tears – a condition supported by measurements of psychophysiological state (such as skin conductance), testosterone levels, and even brain activity monitored by functional MRI. Importantly, the men did not know that the substance to which they were being exposed was female tears, nor had they seen the women cry.

The nature of the chemical signal in the tears presumed to be triggering these effects isn’t yet clear. But its mere existence adds an unexpected new dimension to the chemical basis of sexual interaction – which, even if the metaphor is rather archly belaboured by chemists, is already undeniable.6

Other questions abound. What are the effects of same-sex tears, or children’s tears? Are other functions besides sexual arousal affected? In any event, the current results are an invitation for evolutionary psychologists to cook up explanations of why it is adaptive to experience lower sexual arousal when someone is crying. Does that allow us to hug them without wanting to make love to them? (Countless movies, notably Don’t Look Now, insist otherwise, as does the stereotype of the sexual predator who exploits emotional vulnerability.) It’s fine to speculate, but perhaps better to exercise restraint and regard this intriguing finding as still at the stage of being a chemical rather than an evolutionary problem.


1. S. Gelstein et al. Science doi: 10.1126/science.1198331 (2011).
2. B. Gettelfinger & E. L. Cussler, Am. Inst. Chem. Engin. J. 50, 2646-2647 (2004).
3. A. Montagu, Science 130, 1572-15 (1959).
4. C. Darwin, The Expression of the Emotions in Man and Animals (John Murray, London, 1872).
5. P. G. Petrov et al., Exp. Eye Res. 84, 1140-1146 (2007).
6. G. Froböse & R. Froböse, love and Lust: Is It More than Just Chemistry? (RSC, Cambridge, 2006).