Two kinds of arguments against GM: evidentiary, and precautionary

The news hit the headlines this morning that genetic engineers in Hertfordshire want to trial plants that have been genetically-modified with genes from fish: http://www.farmersguardian.com/home/arable/rothamsted-trialling-gm-omega-3-plants/61696.article .

There are several aspects of the arguments around this that are of philosophical interest. They relate principally to the philosophy of rhetoric, the philosophy of science and technology (epistemology, methodological issues), and the philosophy and ethics of precaution/risk. I will explore these briefly in what follows.

My closest philosophical colleague Phil Hutchinson (@phil_hutchinson) has just had a mini ‘twitter-storm’ with Mark Lynas, over this latest GM business. Phil has been making the argument that the evidence does not support the need for fish genes to be put into plants in order to produce fish oil, because the evidence does not support the claim that doing so is beneficial and necessary.
This ‘mirrors’ the argument that my other current close philosophical colleague Nassim Taleb (@nntaleb) and I (@rupertread) have recently had on twitter with Lynas (Go back to Jan.5 if you want to see this ‘twitter-storm’ from the start). Taleb and I made the argument that (e.g.) taking genes from fish and putting them into plants is reckless, because it is unprecautious: it violates the Precautionary Principle. In other words, our argument was not evidentiary but precautionary.

It seems to me that the ‘evidence’ line against GM combined with the ‘precautionary’ line against it catches GM-apologists such as Lynas in a bind. In a pincer movement.

In outline, the full (the two-pronged) case then runs roughly like this (for references to back this up, if desired, see the material on Twitter):

A GM company wants to take genes from fish and put them into a plant: specifically, in today’s furore: they want to produce Omega three GM camelina.

In brief: There is first NO conclusive evidence for heart-related benefits of Omega 3 fish oil, which demonstrate it as beneficial separate from the fish, as a supplement. There is NO evidence that we need fish oil omega 3 over and above that our bodies already convert from vegetable-based ALA Omega 3 from things like flax.

To elaborate somewhat: We’ve had over ten years of hype from food manufacturers and supplement manufacturers about the heart-benefits of fish-sourced Omega 3 oil. But the evidence for benefits is still inconclusive, at best.

Basically there are three types of Omega 3 fatty acids that humans need: ALA (found in plant oils), EPA, and DHA (found in fish oils).

ALA is in flax seeds and hemp seeds as well as other veg (brussels sprouts for example). Our bodies convert ALA in to EPA and DHA.

Over the past decade or so all sorts of wild claims have been made for the benefits of consuming a diet high in EPA and DHA fatty acids. Goldacre has some sport exposing some of the nonsense hereabouts in Bad Science.

However, there are one or two RCTs that do seem to show some benefit of a diet high in EPA and DHA Omega 3 for heart disease, but, and this is important, only when eaten as part of a fresh fish which contains it. There is simply no evidence for EPA and DHA taken as a supplement being beneficial to health. So, the real kicker is, that they cannot say for sure that it is the EPA and GLA and not just the fact that those who eat fresh fish are likely to eat healthier diets in any case and be better off, socio-economically.

So, why would anyone assume that GM camelina with EPA and DHA would work better than the _ineffective_ supplements? No reason whatsoever. Indeed, as noted, while diets high in fish oil do seem to (in a few cases) have benefits, even there it is unclear this is because of some magical properties of the oil, but rather because of other factors that might be related to a diet rich in oily fish.

So, no clear evidence at all that consumption of EPA and DHA as a supplement has health benefits.

When Phil made these points to Lynas and the GM company, they shifted ground away from talking about the alleged health benefits of omega 3 fish oil (to humans) to talking about the health benefits of feeding omega 3 fish oil to fish.

So: there really – clearly — is no clear evidence that we need EPA and DHA in any case, as our bodies convert ALA (from vegetable sources into those). Lynas et al, when pressed, concede this. They then say: this is about improving aquaculture by making fish food. But then we have the same problem: we have no reason to think that even if the GM splicing worked and they could get it into the seeds that this would work for the fish. Oily fish that are high in Omega 3 get it from the krill and shrimps they eat.

This is about salmon-farming! Not, as they tried to mislead us all this morning into thinking, about human health.

Human health would be better served by better balanced diets.

To sum up the case so far: there is no reason to see what the GM wizards are trying to put into the plant from the fish as useful for fish food if there is no evidence for the benefits of Omega 3 fish oil supplements. At this point, when forced into seeing this, the company replied that that’s allegedly why they need to do the research they are seeking to do… Which is close to a concession that there are (few or) no evidentiary grounds for thinking GM fish-omega 3 camelina will be beneficial: But of course, surprise surprise, that is not what their rep said on the Today programme this morning, nor what Lynas were arguing when Phil first responded to him.

The final phase of the argument (at the time of writing) is I think very telling. It runs thus:

Phil Hutchinson @phil_hutchinson

@Rothamsted @mark_lynas consumed as fish. Barely any conc. evidence for supplement benefits. Your version will be akin to consuming a supp.

Kate de Selincourt @Kate_de

But, @Rothamsted & @mark_lynas, since all livestock farming turns more nutrient into less, why not just eat the fish food? @phil_hutchinson

Mark Lynas @mark_lynas

@Kate_de @Rothamsted @phil_hutchinson That’s an argument for veganism. Fine by me, but hardly a realistic way to tackle overfishing.

‘Fine by me’. Lynas has essentially conceded the case. He prefers a problematic techno-fix which lacks evidential support to a behaviourial and political change that is perfectly possible (i.e. for humans to consume less (factory-farmed) fish (from which a profit can be extracted), and find their omega 3 in other ways).

That’s the evidence-based argument against GM (which has to be made in each individual case on grounds specific to that case (in other cases, the argument will be based on poor yield, or on the inputs to the GM-farming being unsustainable, or on alleged damage to human health, or on actual epidemics of superweeds, or on the desperately-problematic political economy of GM; etc etc), and can be made in each individual case I think, with the possible exception of some GM-cotton). The case benefits from a savvy understanding of the nature of evidence-based arguments, obviously, and thus from a sound philosophy of science and technology perspective. But it is essential an ‘empirical’ argument.

The precautionary argument is different. It is philosophical from the get-go. It is an argument about where the burden of proof lies.
This is in my view the deepest argument against GM: a precautionary one which shifts the burden of proof. It’s no longer about one trying to find a particular counter-argument to claims that GM-enthusiasts are making: it’s suggesting that the onus is rather on THEM to establish the safety of the technology that they are puffing.

The precautionary case against GMOs, in brief, runs thus: If we (for example) take a gene from a fish and put it in a plant, a move utterly without precedent in the whole of evolution, we are recklessly fiddling with and unavoidably changing a system we don’t fully understand and doing something novel whose consequences we cannot possibly predict. This is a reckless gamble, stupid in the short- to medium- term, unconscionably short-sighted and selfish in the long term, as we risk imposing a world of new danger on those who are yet even to be born. We are launching a vast uncontrolled natural (sic.) experiment. The consequences for superweeds, for damaging biodiversity, for creating dangerous mutations, and possibly directly for human health, are unforeseeable. There is a strong precautionary argument against GM, or at the very least in favour of keeping some parts of the world (e.g. an island-nation!) GM-free. IF GM could be properly safely researched to determine what bad ‘side effects’ it may have, then I would favour such research, in good empirical fashion. But it mostly can’t – because it can only be properly ‘researched’ in this way outside the laboratory. In this regard, it differs profoundly from most medical advances, for instance.
This is the terrible dilemma of field trials for GM: The more extensive they are, the more they resemble conditions in the real world, the longer-term they are, then the more reliable they are – BUT also, the more dangerous they are. The more likely it is that they will escape their confines, affect the broader ecosystem, produce unexpected and dangerous drift of genetic materials, etc. . One can’t get the evidence one needs to assess GM with without creating vast uncontrolled new risks.
If we in Britain as a nation contaminate our countryside with GMOs, then that can never be undone. Simple caution and commonsense enjoins – overwhelmingly – against such recklessness.
Defenders of GM sometimes say that there is an absence of evidence of harm from GM. Even if this is true, it is not good enough. What the precautionary argument shows is that we need evidence of absence of harm from GM. And that is what we don’t have. And what will be very hard ever to get without taking an unconscionable risk.
That is the point of the precautionary principle.

Until we have ultra-long-term large-scale trials which cannot contaminate the surrounding countryside, then GM must be considered unsafe. Such trials are at present impossible to carry out. They might one day be possible, though I doubt that they ever will be (this is the dilemma expressed above). If they ever were, then, rather than jumping in precipitously to make a quick buck (as is happening at Rothampsted today) we would then need to wait dozens of years for the results.
In other words, I have argued that one current impossibility is to adequately research contamination, possible damage to biodiversity, etc., without actually potentially causing limitless such damage. One somehow needs long-term (generations of ) trials, in the natural environment, but contained. Something like a huge part-permeable dome that somehow lets in what you want to let in (e.g. sun, rain) in an unaltered way without letting out the GM-crops, over an area of many square miles. Good luck with that…

I am sometimes accused of inconsistency, in making this kind of argument. For, as I’ve made clear in previous posts on this site, I am, like any reasonable person, a fan of climate science, which is vital to the survivability of our species, as we breach the limits to growth. So, why not of ‘GM science’? But this alleged parallel with manmade climate change is very weak. That is a matter of science; while GM is a technology.
Of course genetics is science, but genetic engineering, as the name suggests, is not: it is engineering, i.e. technology. GM is a technology, and so we should be very wary of GM-advocates dressing themselves up in the clothing of science. It is not ‘anti-science’ to oppose GM technology. There are strong empirical and precautionary arguments for doing so.

The parallel in relation to climate is with geo-engineering, not with climate science! And I’m no more a fan of genetic engineering than of geo-engineering, which involves perhaps the ultimate hubristic lack of precaution (or of ethics)… That is: It seems to me, as I’ve sketched, that there are profound philosophical reasons not to be a fan of either of these forms of engineering…

[[Big thanks to Phil Hutchinson for contributing very generously to the researching and writing of this piece, and for our ongoing joint work on 'evidence-based medicine'. Thanks also to Nassim N. Taleb for his influence on my thinking in this area, through the dialogue we are having on it and the arguments we are making against others over it. But responsibility for the piece is mine alone.]]

Leave a comment ?

15 Comments.

  1. Suppose I carefully examine all the plants in a corn field and choose any that have unusual and desirable properties. I then breed these plants together to get a new variety with several positive properties. I do this for several generations, and I end up with a new variety of corn which is superior in some respects to other varieties. For the sake of discussion, let’s suppose my new variety produces more folic acid than normal corn. Now, what does the precautionary principle indicate I should do if I want to start selling seeds for my new variety?

    Consider that I have no understanding of the biochemical basis for my corn’s properties – perhaps the process which produces the folic acid does so by means of a protein which increases consumers’ susceptibility to IBS. Perhaps it also increases sterility in certain butterflies. Perhaps it leaves soil traces which damage the roots of other plants. Who knows? What testing regimen should my corn be subjected to before I may bring it to market?

    What important difference is there between a plant which has some desirable property because of a deliberately introduced, specific, and well understood mutation, and a plant which has a desirable property because of one or more random, accidental, and unknown mutations? (By “important difference” I mean a difference which is significant enough to vitiate precautionary principle in the case of bred varieties, but increase its power in the case of engineered varieties.) Or do you believe that all new varieties should be subjected to the same tests?

  2. Steve, there’s a rather simple difference: sexual recombination of genetic material has had a long time to work itself out (it is highly robust). Ordinarily, genetic material is not shared between mostly unrelated taxa (like fish and plants). Although bacteria have long mastered the art of gene-swapping (which is why our human downstream version piggy-backs on things like Agrobacterium), the human intervention is the relatively untested one. You are confusing our ability to understand what is going on (or think we understand) with the stability and robustness conferred by long evolutionary and co-evolutionary processes, about which we are constrained to understand little.

  3. Chris, I don’t think I understand your point. It’s not necessary for genetic material to be shared between unrelated taxa for novel mutations to arise; plants (and animals) are perfectly capable of developing mutations on their own. What I want to know is why the author thinks it’s safer to trust unknown mutations than known mutations (a known mutation being a deliberately inserted gene). (Keep in mind that corn and other grains have spent millenia developing defenses against getting eaten; I would not be at all surprised if modern grains contain a number of currently-deactivated defense mechanisms which could easily be turned back on by the right mutation.)

  4. What Chris said.

    The principle is that something which is relevantly similar to what has occured naturally for yonks is safer than something radically new.

  5. I’m surprised that I need to remind a person writing for “Talking Philosophy” that the appeal to nature is a fallacy. If we were talking about fruit your point might have some merit – fruits tend to benefit plants more when they also benefit the animals which eat them. Grains don’t benefit from having their seeds eaten, so they are more likely to have a reservoir of defensive techniques which have been suppressed by domestication, but which could be reactivated by minor mutations.

    Predispositions aside, it’s impossible to say without a large amount of investigation what all the effects of any random mutation will be, nor is it reasonable to make a blanket claim that the effects will always be small. Retroviruses can inject their entire genome into a host’s DNA, where it can have effects ranging from nothing at all to who knows what. I also think it would be difficult to justify a claim that plant breeding has never had deleterious results – can you demonstrate that in the ~10,000 years we’ve been domesticating plants, no community has ever accidentally bred a variety which had negative health effects? Or that no community has ever bred a variety which had negative ecological effects?

    Let’s get to the heart of this distinction you want to draw between the “natural” and the “radically new”. Suppose Rothamsted Research decided to develop their omega-3 camelina through a massive breeding program. Suppose that after hundreds of generations they were successful in producing a camelina which generated omega-3 oils. Obviously, the omega-3 obviously wouldn’t be produced using the genes Rothamsted is currently engineering – no one would know how it was being produced. And without a doubt, there would be a large number of additional mutations with unknown effects. Would you be comfortable with this new cemolina, with all its unknown mutations and unpredictable quirks, being brought to market, simply because it was created using the “natural” method of generating mutant varieties?

  6. the problem with the precautionary principle is it works like this:

    (possible benefit * probability of possible benefit) – (possible harm * probability of possible harm) = do it (more than 0) / don’t do it (less than 0)

    In this case, and other cases where the principle is used, such as the approval of new drugs or chemical additives by the EU or US FDA, it is quite easy to cook up a “possible” scenario which is maximally harmful, apocalyptic, but unlikely. As harm tends towards infinity, it starts to become impossible to get an output more that 0, especially as humans seem to have an innate bias towards avoiding harm vs seeking benefit. So the precautionary principle must be used with caution.

    I don’t like the way you characterise the evidence based and precautionary aspects of your argument as a two-pronged “pincer movement”; they are more sensibly read as the terms in the equation above, since “evidence” is a measure of the probability of benefit. The two-pronged argument is sound, since if the evidence for possible benefit is low (I am no expert, and do not dispute the case you cite), it will be outweighed by even the most incautious risk assessment. However, the precautionary principle argument alone is less convincing, and one could create arguments of the same form that were manifestly unconvincing. I do not agree that yours is a “philosophical” argument “about where the burden of proof lies”. Most people would not agree to the principle that it is always up to the person proposing to do something to prove that it won’t be harmful. This is actually impossible, as we can’t see the future, and thus every action which has ever been performed would count as morally damnable, reckless. I believe this “burden of proof” argument appears convincing because it is actually just a challenge based on cost/benefit analysis: the “anti” camp are saying “we believe we can show that the equation < 0, so we won't be satisfied until you can show otherwise". Yet you are apparently claiming that your argument would go through even if there were lots of good evidence of benefit, ie. that omega-3 has myriad uses that would help the common man and not just private interests and that it was much better all things considered to extract it from plants rather than fish.

    A possible counter-argument rebutted:

    You might like to claim that it is equally possible to over-generously assess possible benefit, thus cancelling out the bias. But this doesn't seem to be the case. There are natural limitations on what a proposed action is intended to achieve. THIS is where the notion of burden of proof comes into play. Someone has to delineate a "ballpark" of possible benefits in order that we can even begin to assess their probability, this burden should indeed fall on the proposer. We should be able to prohibit someone in the "pro" camp from saying "omega-3 (for example), might end all suffering in the world", because no one ever has or could suggest this as an aim for the project.

    The burden needn't be particularly heavy, however. We avoid counter-arguments like "if Marie Curie had had to demonstrate the benefit of her research, we would never have discovered things like radiotherapy or radiosterilisation" through your science-technology distinction. "The advancement of science" is an admissible benefit to weigh in the balance in the case of scientific research, but not in the case of a proposed engineering project with technical aims.

    (That said, given the main sources of science funding, the line between "the advancement of science" and commercial R and D is quite indistinct, so putting a limit on the types of possible benefit we can consider may in practice be difficult. The science-technology distinction raises difficult questions about the definition of science. If we define science as research having no instrumental function, then we have probably defined it out of existence, except perhaps in the case of the very furthest frontiers of theoretical physics. I'm going to leave that point there, just observing that we need to sure up the theoretical basis of the science-technology distinction)

    A point of clarification:

    When I say "we can't see the future", this is best understood as a point of epistemology not of metaphysics. We can read it as "We can't, practically speaking, be in possession of all facts, and as a corollary of this, we can't be possessed of the knowledge of which facts are pertinent". So the impossibility I spoke of was technically a physical possibility and not a logical impossibility, unless we want to defend a particular metaphysics, which I don't.

  7. Steve, you are mistaken. An appeal to nature is not a fallacy. Far far from it.
    It can under certain circumstances be a fallacy, in relation (e.g.) to matters moral. But ‘the naturalistic fallacy’ is grossly over-played. It is used as an excuse for non-thinking, and in particular it is often used for recommending a Promethean absence of any respect for or serious attention toward nature. A pretence that human beings are as gods.

    Taleb and I are working on this at present – his published work (and his twitterfeed) already contains interesting material following up or detailing my thought here.

  8. Steve:

    Retroviruses: interesting example!!

    Now, can anyone think of a retrovirus that has caused trouble?…

  9. Those who can’t, get a zero on the test. Or a negative mark, if that is possible…

  10. Finally Steve, if Rothamsted did as you are suggesting, over hundreds of generations, surely it is obvious that this would be safer: because it would take hundreds of generations!

  11. Thanks John, interesting comment.

    The key point where I think you err is in always talking about probabilities. The thing is: hardly anything can actually be calculated in terms of real probabilities. CBA, RCT etc: these are toy procedures, with no direct carry-over to the real-world. Casinos and lotteries often (not always or in all respects!) yield probabilities: virtually nothing else in life does.

    On the future, I also think that you are wrong to say that this [our lack of knowledge thereof] is merely an epistemic matter. (See my previous posts on ‘time-travel’ etc for elaboration.] For: we MAKE the future (to some extent, and together). Our ignorance of it is in part due to the point above (about there being few probabilities in the real world) and in part constitutive (Because we are agents, and self-aware, and able to create or destroy ‘feedbacks’, etc. On which, see my WITTGENSTEIN AMONG THE SCIENCES and the my post on this blog about it..)

  12. This has been disappointing.

  13. “Suppose Rothamsted Research decided to develop their omega-3 camelina through a massive breeding program. Suppose that after hundreds of generations they were successful in producing a camelina which generated omega-3 oils. Obviously, the omega-3 obviously wouldn’t be produced using the genes Rothamsted is currently engineering – no one would know how it was being produced. And without a doubt, there would be a large number of additional mutations with unknown effects. Would you be comfortable with this new cemolina, with all its unknown mutations and unpredictable quirks, being brought to market, simply because it was created using the “natural” method of generating mutant varieties?”
    >>This (from Steve, above) seems to me to demonstrated what I’d like to call ‘the engineering fallacy’. The fallacy/fantasy that, because something has been engineered in a lab, we understand it better and have fuller control over it than if it has arisen through a longer process of iterated interaction with nature. The reverse is often true. (As we are seeing now, with genetics massively complexifying its understanding of how genes work, beyond the simplistic nonsense of genes for this and genes for that that used to constitute genetic engineering.)

  14. The precautionary principle
    Noddy has driven his lovely car to bit of high ground, and suddenly it starts to rain. The flood waters rise and Noddy realises he is surrounded by water. Big Ears appears on the horizon. ‘Wade over here Noddy’ says Big Ears ‘That doesn’t look like a good place to be’

    “But I must invoke the Precautionary Principle” yells Noddy “I couldn’t possibly leave my shiny red car with it’s big yellow wheels and get my blue hat wet, it’s not safe to wade across flood water:The outcome is very uncertain”

    “OK ” says Big Ears “drown then, and see if I care”

    And he does.

  15. Best VIP cheap Cowboys jerseys from china Store

Leave a Comment


NOTE - You can use these HTML tags and attributes:
<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>