Monday, October 08, 2018

Rational Irrationality and Satter's Hierarchy of Food Needs

In HIERARCHY, DISAGREEMENT, AND FOOD POLITICS food economist Jayson Lusk discusses Maslow's Hierarchy of Needs and a modification or application by Ellyn Satter. Satter conceptualizes a hierarchy of food needs. Basically the idea is that as society reaches more advanced levels of economic development and incomes rise, our preferences related to food change.

"Satter called the top of this pyramid "instrumental food" and she said such foods were consumed to "achieve a desired physical, cognitive, or spiritual outcome."  If we're talking about food satisfying a particular view of what I think of myself (I eat what I am) or food satisfying a "spiritual outcome", why would we expect you and I to agree on what is "best"?  In this sense, we might expect food consumption to be more politicized"

Another way of thinking about this is that high end food fads marketed by the likes of Chipotle and Whole Foods are 'normal goods' i.e. as incomes rise their consumption should increase. There is a reason why you only find certain food chains and grocery stores in areas where incomes are higher.

Satter notes: "These instrumental reasons may or may not be rational or supported by scientific inquiry."

This is not so different from the concept of 'rational irrationality' discussed in Brian Caplan's 'The Myth of the Rational Voter':

"...people have preferences over beliefs. Letting emotions or ideology corrupt our thinking is an easy way to satisfy such preferences...Worldviews are more a mental security blanket than a serious effort to understand the world."

One of the main themes in the book is that this leads to systematic biases in voting behavior and policy. Particularly, these kinds of preferences create a gap between economic principles and policies supported by most economists and the general public. The authors note that this division probably is not unique to economics and they are correct. As Jayson notes, a few years ago research from the Pew Foundation identified a number of scientific issues for which there is a gap between views held by scientists and the public. 

So how does this play out? A colleague brought up a very important point. Rational irrationality implies that there are costs associated with irrational beliefs, and people are willing to hold on to certain world views given the costs are low. Since the costs associated with voting are much lower, and voters don't necessarily bear the full costs of their actions we would expect to see 'rational irrational' behavior more often in voting than we see with regard to food purchasing behavior. However, when we think about Satter's hierarchy the idea is that with increased incomes preferences for food become more abstract (related to politics, ideology, social status etc.). Consumers are willing to pay for that. However, for higher income consumers the share of food in the household budget is relatively small. Hence the costs of 'irrationality' are minimal compared to what those costs would be for lower income consumers or at the bottom of Satter's hierarchy. This means that wealthier demographics and wealthier societies can afford to be 'rationally irrational.'

So costs of irrationality for those consumers is relatively low. Add to that the opportunity this presents to food marketers in addition to special interest agendas like the non-GMO Project and US Right to Know combined with low cost voting and it’s a perfect storm.  As this influences food manufacturers and the regulatory environment, we begin to see limited food choices on the shelves. More limited options may come at the expense of more affordable options produced with more efficient and sustainable technologies. This can exasperate issues related to food waste and food insecurity. Examples include Vermont's GMO labeling law, push back against new food technologies like the arctic apple, and the attempt in Brazil to ban glyphosate.

Sunday, September 16, 2018

Don't Throw Good Science Out With the Dirty COI Bathwater

There has been a lot of conversation lately regarding conflicts of interest in research. This recent tweet by Andrew Kniss resonated a lot with me:
No doubt conflicts of interest are important to be aware of and understand. We definitely want to maintain the highest integrity with regard to science communication and research. However, it is important that COI 'labels' don't become the new 'free-from' label i.e. we don't want a conflict of interest to necessarily become conflated with bad science any more than we would want a GMO label to be conflated with unhealthy or unsustainable. We don't want COI to become the red herring for smear campaigns, political correctness, or a disincentive to doing good science.

About a year ago during one of his excellent talking biotech podcasts (specifically regarding the movie Food Evolution) Kevin Folta made an excellent point about COI and bias in research:

"I've trained for 30 years to be able to understand statistics and experimental design and interpretation...I'll decide based on the quality of the data and the experimental design....that's what we do."

COI involve a delicate balance. I think the recent controversy around COI highlights the importance for producers and communicators of science to get this right. There's also a challenge for consumers of science to know how to recognize good science without throwing it out with the COI bathwater.

Saturday, May 12, 2018

CRISPR and Gene Editing Blueprint Analogy

There was a nice guest post on AgWeb recently titled:

Gene Editing: Building Better Blueprints, One Gene at a Time. 

I really like the blueprint analogy as a means to help people understand how gene editing is similar and different from other technologies. Many consumers that are skeptical of advances in food technology are comfortable with older technologies or don't realize the differences.

As Bob Reiter stated in the article:

"Now consider this: what if there was a defect in the blueprint for the house? If we followed those instructions anyway, the defect would be built into the house – which could later lead to structural problems, ranging from minor to catastrophic, depending on which part the defect involved."

Messing up the blueprint is what a lot of consumers are hesitant about when it comes to traditional recombinant DNA technologies (a.k.a. 'GMOs'). They are worried about unknown downstream structural problems and the impact that could have on human health. To put this in other terms, genomic disruptions. In response they advocate for more regulations, testing, and labeling of 'GMO' foods and many are calling for a similar framework for gene edited foods. And food manufacturers take advantage of the marketing opportunities created by these concerns. Ever heard of the non-GMO Project?

But the blueprint analogy is actually helpful here. As Bob explains, in college he was careful about designing the blueprint. If we think of gene editing like making careful targeted changes to the blueprint, USDA organic approved technologies (methods using radiation or chemical mutagens) are more like his intoxicated fraternity brothers sneaking in making random changes to his plans without him knowing. Perhaps they introduce really cool innovations! On the other hand, the roof might leak, the plumbing could drain backwards, or worse. To put it differently, the number of genomic disruptions are far greater and unknown.

You would think, if customers and regulators were concerned about Bob's targeted changes (maybe they would insist that someone from the county does an inspection before proceeding with the construction) they would really be worried about the changes brought about by his inebriated counterparts. However, if we analogize back to mutagenic conventional and organic food, they don't seem concerned at all. They have accepted a build it and see what happens later attitude. No testing. No labeling. (other than maybe that Butterfly food marketers like to stamp on everything from rock salt to water). Of course, from a scientific risk based perspective, there probably is not a reason for testing or labeling these foods....if consumers already understand and accept this that should be a step further down a path toward newer 'safer' technologies that promise so much more.

You can only take an analogy so far but I like Bob's.

See also: Organic Activists Realize Hypocrisy On Gene Editing and Biotech


Batista R and others (2008). Microarray analyses reveal that plant mutagenesis may induce more transcriptomic changes than transgene insertion. Proceedings of the National Academy of Sciences of the United States of America 105(9): 3640–3645

Baudo MM, Lyons R, Powers S, Pastori GM, Edwards KJ, Holdsworth MJ, Shewry PR. (2006). Transgenesis has less impact on the transcriptome of wheat grain than conventional breeding. Plant Biotechnol J. 2006 Jul;4(4):369-80

Thursday, March 22, 2018

The convergence of data science, genomics, and technology

This is a couple years old but just as relevant and inspiring:

"I believe the future of agriculture will be shaped by continued improvements in efficiency. By more effectively applying the advancements that have come from the green and biotech revolutions, we can continue to see yield increases across the world. We can use data science to drive agronomic practice improvements that enhance existing technologies, like advanced seed genetics. Data science technologies can build upon the yield potential already present in each seed by providing farmers with actionable insights they can use to drive efficiency improvements on their operations, getting the most out of every single plant." - Erik Andrejko, Director of Science, Head of Data Science at The Climate Corporation (2015)


See also: Monoculture vs. the Convergence of Big Data and Genomics

Sunday, February 11, 2018

The 'free-from' Nash equilibrium

Recently I was reading an article, "The big Washington food fight" in Politico discussing challenges of bringing diverse interests and perspectives on food issues under one roof through the Grocery Manufacturers Association. 

There are a couple things influencing my thinking about this. One is the idea that voters and consumers may have systemic biases in their knowledge and preferences in general and specifically about food and technology. The other thing is related to recent research showing a divergence between public perceptions and science driven by political leaning....a divergence that widens *with* more education and science knowledge (see ).

So in this context, what does it mean to say 'the customer is always right' and how do you give the customer what they want, but do it with integrity? There seem to be two dominant approaches or paradigms followed by food companies for dealing with this. 
One approach is going all in with 'negative' advertising or 'free from' labeling regardless of scientific justification. This paradigm feigns or fakes transparency in the sense it acknowledges consumer preferences related to knowing 'what is in their food' but adds lots of confusion about substantial differences related to food safety and sustainability. The other paradigm takes a 'less is more' approach in terms of honest disclosure about these technologies.

A more generous interpretation of the first behavior is that these food retailers and manufacturers are cognizant of consumer preferences, and assume great deal of consumer ignorance. Take for instance consumers attempting to avoid gluten (reasons why merit a separate discussion). Retailers may assume consumers are extremely ignorant of the fact that gluten originates from wheat based ingredients. A number of food products (like apples, raisins, ground beef, potatoes, carrots etc.) generally would not possibly contain gluten unless they were highly processed or prepared using some wheat based ingredient. As an extreme example for illustrative purposes, we might imagine an uninformed consumer choosing between a bag of carrots and a loaf of bread. A manufacturer would feel they are helping the consumer simplify the decision by adding a 'gluten free' label to the bag of carrots. Additionally, if competitor brands don't have the label, they may risk losing a sale to the 'gluten free' labeled product if a large enough number of uninformed consumers are trying to avoid gluten. We end up with a Nash equilibrium strategy to employ 'free from'labels that really make no sense from a nutritional or scientific standpoint. Gluten is just one example, the logic easily carries over to a number of food products and ingredients (i.e. free-from labels related to added hormones in pork and poultry, GMO free tomatoes etc.)

Helping consumers avoid what they *think* or *perceive* to be harmful to themselves or the environment using this strategy may help increase or defend sales, but it does very little to truly educate consumers about food choices. It likely perpetuates ignorance and myths about food, nutrition, health, and sustainability. Worse, in some cases it may directly or indirectly lead to consumers actually choosing marginally less healthy or less sustainable products or technologies either directly or indirectly because of decreased viability of marketing alternatively sustainable foods (i.e. the loss of rBST from the milk supply).

Whichever paradigm becomes the most dominant (both in the marketplace and the ballot box) may ultimately influence the types of products we see on the shelves and reduce the potential for healthier and more environmentally sustainable solutions to challenging worldwide problems. Efforts to escape this less than optimal Nash equilibrium position could include restrictions limiting free from labels to only non-obvious food products but I am not sure this is a viable solution.

Sometimes people devise cooperative ways to escape from a Nash equilibrium without resorting to taxes or regulation. Nobel prize winning economist Elinor Ostrom's work speaks to this:

"Predictions that individuals will not devise, precommit to, and monitor their own rules to change the structure of interdependent situations so as to obtain joint benefits are not consistent with evidence that some individuals have overcome these problems, although others have not." - Governing the Commons: The Evolution of Institutions for Collective Action. By Elinor Ostrom

But how would this work? The recent backlash by the scientific community regarding StonyField Organic's portrayal of young girls in a Facebook video and the Peel Back the Label movement are two examples of social harassment costs or other monitoring behaviors within the industry that may give some hope. Will this continue or become a large enough force for change?

Sunday, January 14, 2018

Why Study Plant Breeding and Genetics?

A really nice piece in the Huffington Post by Kevin Folta evangelizing about future careers in plant breeding (link).

I like the way he describes genetic markers and marker assisted selection:

"The trick is to use DNA-based landmarks that physically travel through generations in parallel with the trait of interest. A simple DNA test, much like those used in forensic science, tells a breeder if a seedling has a high likelihood of presenting a coveted trait upon maturation. These tests allow breeders new ways to identify the most relevant parent plants, and eliminate non-candidate offspring at the seedling stage, long before valuable time, labor, acreage, fertilizers, and water are invested."

When I was an undergraduate senior, we had a speaker working on their PhD from UC Davis visit WKU to discuss their work in quantitative genetics. That was the first time I realized that there was lots of opportunity and very exciting work that could be done outside the lab, as genetics and crop improvement can be as much a statistical and quantitative science as a lab science. In other words, not all of the cool jobs were being taken by PhDs in molecular biology. As Kevin states:

"While mastery of molecular biology techniques is not required, modern breeders need to be conversant in the technologies, and excited to integrate with collaborators. A modern breeder will have strong skills in statistics, and even computational prowess."

This sounds a lot like a data scientist with domain expertise in plant breeding and genetics. From this angle there seems to be tremendous opportunities with the convergence of big data and genomics. While I did not pursue a career in plant breeding, I have leveraged my quantitative training in agronomy including experimental design and statistical genetics blended with econometrics and programming to advance my career in healthcare data science.

Think about the combination of genomic markers, plant 'wearable' sensors, remote sensing, and other precision ag data sources. Very exciting work for all involved.

Dr. Folta is an excellent evangelist for crop genetic improvement and science communication.  I highly recommend checking out his podcast Talking Biotech where each week he or Dr. Paul Vincelli from the University of Kentucky discuss agricultural and medical biotechnology.

Related: QTL Analysis and Marker Assisted Selection 

Sunday, December 31, 2017

Herbicide Resistance and GE Crops - Thinking like an economist

In a recent issue of Weed Science, Andrew Kniss investigates the relationship between herbicide resistance in weeds and adoption of genetically engineered crops.

There is a popular story with an anti-gmo theme that holds that genetically modified crops tolerant to roundup (glyphosate) herbicide have encouraged excessive levels of use. This has lead to a build up of roundup resistant weeds. In some sense, this paper may provide some evidence in favor of that story. However, I think there is a tendency among critics to extrapolate further that GMOs lead to higher levels of weed resistance (in general). The storyline does not make a distinction between weed resistance in general and specific resistance to roundup.

What this paper does indicate is that there are to some degree externality mitigating aspects of glyphosate tolerant crops. (not exactly but somewhat like the positive externalities we have seen with Bt crops). As stated in the abstract, "Increased glyphosate use in cotton and soybean largely displaced herbicides that are more likely to select for herbicide-resistant weeds, which at least partially mitigated the impact of reduced herbicide diversity....the evolution of new glyphosate-resistant weed species as a function of area sprayed has remained relatively low compared with several other commonly used herbicide SOAs."

This paper definitely provides data and evidence contrary to some of the popular stories condemning biotechnology. Stories that don't look at crop production from a more comprehensive systems based viewpoint miss these nuances. appears that thinking like an economist (systematically considering intended and unintended consequences) has a lot to lend to the nuances of herbicide resistance.

More from the Abstract:

 "adoption of GE corn varieties did not reduce herbicide diversity, and therefore likely did not increase selection pressure for herbicide-resistant weeds in that crop. Adoption of GE herbicide-resistant varieties substantially reduced herbicide diversity in cotton and soybean. Increased glyphosate use in cotton and soybean largely displaced herbicides that are more likely to select for herbicide-resistant weeds, which at least partially mitigated the impact of reduced herbicide diversity. The overall rate of newly confirmed herbicide-resistant weed species to all herbicide sites of action (SOAs) has slowed in the United States since 2005. Although the number of glyphosate-resistant weeds has increased since 1998, the evolution of new glyphosate-resistant weed species as a function of area sprayed has remained relatively low compared with several other commonly used herbicide SOAs."

Link and Citation:

Kniss, A. (2017). Genetically Engineered Herbicide-Resistant Crops and Herbicide-Resistant Weed Evolution in the United States. Weed Science, 1-14. doi:10.1017/wsc.2017.70

See also: 
Game Theoretic Analysis of Bt Resistance
Positive Externalities of Biotech Bt Traits on Non-Biotech Crops and Non Target Insects
Environmental and Health Effects of Bt Cotton
Choices Magazine - Herbicide Resistance