The roles of ‘rationality’, ‘toxicity’ and ‘partisanship’ in interpreting scientific information

by Cami Ryan
The article Why we are poles apart on climate change by Yale U law and psychology professor Dan Kahan came across my ‘desktop’ yesterday. Climate change is a topic that is hotly debated in the mainstream media and in social media as well. There are climate change proponents and there are climate change denialists. Personally, I resist resting a foot in any camp as I don’t really know enough about the whole issue of climate change. But I do know that Kahan’s points are certainly relevant when you consider them in the context of the genetically modified food debate.
There are ardent supporters of the technology at one end of the continuum and very passionate opponents on the opposite side. But why are we so deeply divided on the topic of GMOs (genetically modified organisms)? Kahan poses this (à la climate change debate). He suggests that it’s not that people are irrational. Rather, it may be that “their reasoning powers have become disabled by a polluted science-communication environment”…“…[C]itizens are …are, in fact, too rational — at filtering OUT [the] information that would drive a wedge between themselves and their peers.”
Hmmm. Now, what does he mean by ‘polluted’ and what does he mean by ‘too rational’? Well, Kahan’s following remark provide insights into that:

“People acquire their scientific knowledge by consulting others who share their values and whom they therefore trust and understand. Usually, this strategy works just fine. We live in a science-communication environment richly stocked…The trouble starts when this communication environment fills up with toxic partisan meanings — ones that effectively announce that ‘if you are one of us, believe this; otherwise, we’ll know you are one of them’. In that situation, ordinary individuals’ lives will go better if their perceptions of societal risk conform with those of their group.”

So, we are largely influenced by our closely-tied networks, our communities and our families. Makes sense. I am contemplating Kahan’s ideas further in the context of how (dare I say if?) governments acquire / interpret science based information in order to inform policy-making decisions. What gaps out there need to be addressed? What can be done?
I would welcome your comments. Kahan’s article is attached. It’s a one-pager and a quick and relatively easy read.
Kahan, D. (2012). Why we are poles apart on climate change, Nature, 488 (7411) 255. DOI: 10.1038/488255a
Originally posted at Cami Ryan’s blog.



  1. An article about climate change from a psychologist. And now you get a comment about GMO from an engineer. Despite being a political liberal and a staunch environmentalist, I am a skeptic of catastrophic climate change and deeply suspicious about GMO. As an engineer, I perform risk assessments every day. I am genuinely interested in seeing a rigorous and exhaustive, formal risk assessment of these issues, and cannot find one (that was actually part of the charter of the IPCC but they have never even come close). In a formal risk assessment, every possibility is assigned a “likelihood” and a “severity.” There are no blanks; if information is lacking you need to generate it. Can you point me to a document like this for GMO? Unless it is out there and I missed it, I would suggest that that is where the gap lies, and that it “what can be done.”

  2. In a formal risk assessment, every possibility is assigned a “likelihood” and a “severity.”
    Really? Every possibility! That’s quite remarkable…..
    Even in the highly controlled environment of traditional engineering, it would be beyond pretentious to suggest one could even list all possibilities, let alone assign likelihoods or severities to them. These last two items are also always loaded with their own definitional assumptions.
    given that one accepts those, “risk assessment” in and of itself implies nothing more than an educated guess which some will accept and others will shoot down. From your tone and background, I get the impression you would be in the later camp for any assessment provided by a biologist. Rigor will be defined by the topic at hand and biology is one messy topic. Give us a better idea what constitutes rigorous for you (sorry, listing all possibilities is a non-starter).

  3. “Really? Every possibility! That’s quite remarkable….”
    I’ve never thought of them as remarkable, just extremely tedious.
    Your response is that no one has tried to do a risk assessment because it would be hard to do? In risk assessment, the perfect need not be the enemy of the good. Just randomly picking up two topics from this website, superweeds and allergic reactions are identified as risks from GMO. How well have allergic reactions been studied in GMO terms? What is the likelihood that someone could die from a single acute reaction? What is the likelihood that a large number of people would have chronic effects? How can we model that? These are not intractable questions, but I suspect that many of them are questions that have not been addressed. Have you ever seen any of this type of work done on any of the potential risks of GMO?

  4. Your response is that no one has tried to do a risk assessment because it would be hard to do?
    No, my response is that you have defined an ill conditioned premise. Not advisable when calling for “rigorous” testing. GMOs are the most tested (aka assessed for risk) agricultural products available.
    Yes, allergen work has been done, but, again, this is not an easy piece to nail down. Allergic reactions, for example, can be expressed at various levels from very mild to fatal. Such a range can occur within an individual within a span of years, or obviously across multiple individuals. How would you propose testing/assessing such a risk? ” The likelihood that someone could die from a single acute reaction?” That’s easy, 100%. I can safely bet that someone out there would die if they ate product X (whether it’s GMO or not). There is enough range in human variability that this is virtually guaranteed. This type of assessment is useless. With this type of variability how are you supposed to define any model that “a large number of people would have chronic effects”? Currently, as far as I know (and others probably know better on this), tests are done for known allergens (such as nut allergies) in all GMO products meant for human consumption. I don’t believe they are carried forward if the tests are positive. Perhaps D. Tribe or others can provide more details here.
    Superweeds (a misnomer if there ever was one), which I take to mean herbicide resistant weeds, are a well known and documented phenomenon observed long before GMOs were being used. They exist in any large population, even if that population has never been exposed to the herbicide. Transfer of herbicide resistant traits to wild populations has also been studied in detail, however, the results and any conclusions are dependent on the crop, the gene(s) involved, the environment, etc. Yes, genes can transfer, and it is probably certain that with some crops it will occur. Risk in such cases is better dealt with through risk abatement, not risk assessment. Cultural practices are encouraged (or required) to ensure such transfers )or indigenous populations) are not allowed to proliferate.

  5. The allergenicity question is thoroughly investigated by doing 8mer (I believe this is the current state of the art) comparisons to allergen databases before even transforming plant cells. (also one notable modification that gets raised as a potential safety problem by anti-GMO folk is the brazil nut protein (in soy?) which got pulled due to the potential for allergenicity – this however is a win with regards to showing what the system does and does not catch – it illustrates that while it is possible for allergens to be shunted about the system catches them before harm is done (which should answer your question about whether or not this is investigated – one would question how it was stopped if it is not investigated)
    This however, is not why I believe Pdiff was being persnickity about things – documenting *all* possibilities is clearly an exaggeration due to its absolute impossibility (there are an infinite number of possibles) unless one assigns some cutoff after which it is silly to investigate (I would imagine as an engineer you probably aren’t going to assess the potential impact of a new resistor setup in a walkie-talkie with regards to its impact on coral reef formation in the mid-pacific – but if you truly investigate all possibilities you’d have to (and hey, it’d make for a great vacation))

  6. Matt,
    I agree with your overall point about the need for appropriate risk assessment. (Though I also agree that evaluating every possible risk is impossible.) And I won’t pretend to know much detail on what risk asessments have been done.
    But I have a different question for you. I get the sense you (like many) think GMOs should undergo more testing and risk assessment than conventionally bred crops, simply because they’re GMO (i.e. regardless of the transgenes involved). Is that true? And if so, why?

  7. Have you seen the National Academy assessment?
    Approaches to Assessing Unintended Health Effects
    Here are a couple of databases of allergenicity: and and they link to supporting references and documentation.
    But that said–why would a protein or modification of a GMO be any different than any other food or breeding strategy? One example I like to use is the lupin flour that’s becoming quite the rage in some foodie and Euro settings. It’s not modified, but it’s new to a lot of people. It undergoes no testing before it gets to anyone. Yet lo and behold:
    Lupin flour ‘poses allergy risk’
    As someone with a peanut allergy, I am aware of this. A lot of people won’t be. I also come from a family loaded with allergies to many common and unmodified foods. Allergic response is very individual–there’s no way to assess everything in every human. Even in my family it is not consistent: grandma allergic to strawberries, brother to eggs, sister to lemons, me to peanuts (and there’s more…). But still–GMO proteins are no different than any others. And many GMO modifications are *not* going to be proteins. I continue to await the peanut with the allergen removed.
    If you want to bubble-wrap every human from every food risk, ok. But that seems both unlikely to work and unlikely to really improve safety.
    Maybe we could label allergic people. It’s probable that my family has a filaggrin mutation that increases our likelihood for allergic responses. That’s a bigger risk for all allergy than any specific food.

  8. Thanks folks, I appreciate you taking time to respond. I think the respondents have misunderstood “every risk,” which is not surprising as I did not spend much time making myself clear. This is done categorically, so for example with allergies the hierarchal risk assessment would immediately reduce the question to “what additional risks are posed beyond those that already exist? I’ll admit to having not thought about this much before, but I would imagine that the specific risk would be, to make up an example, a new variety of corn with an inserted allergen that is eaten by a susceptible person who has eaten non-GMO corn their entire life. In an engineering risk assessment that includes mitigation strategies, you might just say that the risk goes from “likely” and “severe” (unacceptable) to “remote” and “severe” (perhaps acceptable by predetermined criteria) simply by labeling the new corn (you might also want to know that 95% of people with food allergies always read the label on foods, or whatever). See, its not really that hard. This is an example of a trivial mitigation strategy (label it, duh!) that can come out of a formal risk assessment; trust me that not all strategies are quite so intuitively obvious without the formal assessment. In comparison, the superweeds response is troubling. If there is truly no risk mitigation beyond abatement, you have a “likely” and “severe” with no real mitigation. You cannot build a bridge like that. Are there no empirical tests for this? I’m sure lots of clever folks have thought about it.
    Qetzal, the answer to your question would be based upon the results of the risk assessment. If the assessment shows risks that do not exist with conventional crops, then more modeling and empirical testing is required.

  9. ““what additional risks are posed beyond those that already exist?”
    This assessment is done.
    With allergens if the value here is anything I can tell you from experience that what occurs is that your shit gets shut down. You don’t mitigate. You stop. You do not pass go. You do not collect $2Bn. You simply don’t get away with inserting something which you know is an allergen (at least not in industry)
    “you have a “likely” and “severe” with no real mitigation”
    Any herbicide use has resistance as a likely outcome, severity is questionable here I guess (it is severe with regards that herbicide, but if you don’t use the herbicide the weeds may as well be resistant anyway) and there are strategies for mitigation (much as there are with resistance in bugs to Bt etc)
    Again, the risk assessments have been done (although it does feel that in the case of herbicide tolerance they weren’t particularly well thought out in the earlier days, at least from the perspective of industry) and there are measures in place.

  10. Matt,
    Thanks for the reply, but I don’t think you really addressed my question. In your first comment, you said you are “deeply suspicious about GMO.” Why?
    I can understand being concerned about introducing a new, potentially allergenic protein into a food crop. I can also understand being concerned about possible herbicide-resistant weeds. But shouldn’t we have the same concerns over a conventionally-bred crop that raises those issues?
    Suppose someone bred a Round-up resistant corn. Should it require the same testing regarding ‘superweeds’ as the GMO version?
    Suppose some managed to breed rice that make beta-carotene, with the result that the rice grains also contain new proteins that weren’t there before. Should we be just as concerned about allergies as for the GMO golden rice?
    How do you feel about breeding techniques like polyploidy or hybridization, which might result in genome-wide effects on protein expression levels, DNA rearrangements, aberrant proteins, etc. Should those be an even greater concern than GMO, since the potential scale of the changes is so much greater and less well understood?

  11. Hi Qetzal,
    You asked:
    “Suppose someone bred a Round-up resistant corn. Should it require the same testing regarding ‘superweeds’ as the GMO version?”
    I’m starting with the easy one. Yes, absolutely, an objective risk assessment is entirely independent of historical contingency. Risk is risk.
    You make an interesting point about polyploidy. However, we know a lot of instances of polyploidy, both induced and natural, and so those risks seem to me to be easier to bound. Putting animal genes in plants is more difficult to bound in terms of potential undesirable outcomes. Please take that as a general opinion of someone not well versed in the literature.
    As to your general question about my suspicion, I would cite a few reasons. One is that I am concerned that risk assessments in the GMO world may not be objective. I don’t want Monsanto to do the risk assessment on a Monsanto GMO, and that would not be how it is done in engineering. Are there independent bodies that must approve these products before they can be marketed as in engineering? Another concern has been touched on here by me and others, that risk for something like putting animal genes in plants is very difficult to bound. As pdiff put it, “biology is one messy topic.” That is not an argument to skip the risk assessment, it is an argument for being extremely careful.

  12. risk for something like putting animal genes in plants is very difficult to bound

    I don’t see why this would have a different risk profile than putting plant proteins into plants (inter or intra species), there is no magical essence of animal that gets carried along.
    Although it is rather a moot point as animal genes don’t routinely get transformed into plants for commercial transgenics (One could not, for instance, within Monsanto, be taken seriously if you suggested a transgenic plant with an animal as the source – although this isn’t for scientific reasons, but reasons of public perception)

  13. Yeah, this one always confuses me. I don’t understand the “animalness” of a string of amino acids. Maybe it’s because I look at all sorts of species all the time, but I’m really agnostic on any sort of specialness of a species or group.
    That said–I could think of some cases of this. That new human protein used to treat Gaucher disease is done in plant cells. And it’s a great advance because human cell culture can be a source of human infection, and this should help cut down on that problem.

Comments are closed.