Earlier this week, the journal Perspectives on Psychological Science published a paper by a team of psychologists on “absolute moral opposition” to GM foods. The paper has attracted some attention on my social media networks, where it’s being interpreted as evidence that many GM opponents are evidence-insensitive, that is, that their opposition to GM crops has nothing to do with “the facts,” and that this explains why the debate is intractable.\ For example, Jesse Singal concludes from the study that

it’s going to be really, really hard to convince skeptical Americans that GMO foods are okay to eat, because people’s beliefs about them are so driven by disgust rather than by any real understanding of the issues.

However, the actual study did not investigate whether GM opponents are evidence-insensitive.

Here’s what the study actually did. Participants were first asked a series of general questions about their attitudes to GM foods:

(a) “I do not oppose this,” (b) “This should be prohibited no matter how great the benefits and minor the risks from allowing it,” (c) “It is equally wrong to allow some of this to happen as to allow twice as much to happen. The amount doesn’t matter,” and (d) “This would be wrong even in a country where everyone thought it was not wrong.” (317)

Someone was classified as an “absolutist opponent” if they answered “no” to both (a) and (b).

Next, the respondents were presented with one of eight different scenarios. In four of the scenarios, an individual knowingly consumed a GM food (tomatoes, apples, tuna, or milk); in the other four, they consumed one of the same GM foods without knowing about it. Here’s the example text in the paper:

“Mary eats tomatoes that have been genetically modified. She knows [does not know] the tomatoes have been genetically modified. Scientists have inserted genes in them so that they stay fresh longer.”

Respondents then had to describe their feelings about the scenario in two ways. First, they either picked a word or a facial expression. The only options were “anger” (either the word or an “angered face”) and “disgust” (again, either the word or the face). Second, then rated, on a scale from 1 to 9, “how disgusted and how angered they were when imagining the scenario” (317). Finally, the respondents filled out some questions about their policy views and a “25-item Disgust Scale” to measure their “overall disgust sensitivity” (319).

That’s basically it. I want to make five points about this study.

First, the conceptual framework that the researchers use conflates several categories recognized by ethicists. Consider the way absolute moral values are explained in the introduction:

[Absolute moral values’] defining characteristic is the unconditional proscription of certain actions (e.g., “Do not cause the extinction of a species” or “Do not kill another human being”). Absolute moral values are explicitly regarded as axiomatic, requiring no further justification, and are protected from trade-offs with nonmoral (secular) values — especially money. (316)

Specifically, question (b) in the study instrument is examining non-consequentialist reasoning: whether someone is opposed to GM crops for reasons that are independent of the harms and benefits. This is not the same as an unconditional proscription; for example, someone might think that GM foods should be labeled because consumers have a right to make informed decisions about what they’re eating, regardless of the risks and benefits. This reasoning is non-consequentialist, but the pro-labeling view is not unjustified (it’s justified by general principles about autonomy and bodily integrity), and it may not be an unconditional proscription (e.g., the view might be “no GM foods unless they’re labeled”). For another example, someone might be opposed to GM foods because they think the benefits go overwhelmingly to a few large multinational corporations and the risks are borne by consumers, small farmers, and the environment. For someone with this view, the key question is not how great the benefits are, or how small the risks, but instead whether those benefits and risks are distributed fairly, such as according to John Rawls’ difference principle (the greatest benefit to the least advantaged). Or, for a third alternative, someone might be opposed to GM foods because they think the process for making decisions about GM foods is controlled by industry, not democratically.

Second, the study design treats GM opponents and supporters asymmetrically. The researchers do not consider absolute, non-consequentialist, or evidence-insensitive moral support for GM foods (which, in my opinion, is pretty common in some “pro-science” circles), or the role of disgust or other strong emotional responses in GM support. In this way, the study design pathologizes GM opponents even before any data have been collected: behind the research questions is an assumption that GM opponents are deviants, and their deviance needs to be understood in order to manage and neutralize it.

Third, the narrow study design means that we do not understand how the respondents understood the prompts. When someone read (b), what risks and benefits were they thinking about? Was anyone thinking about the views I suggested in the last paragraph? Were there other reasons that neither I nor the researchers have considered? Similarly, on the responses to the scenarios, we don’t know how people were interpreting the fact that Mary didn’t know the tomatoes were GM. At one point the researchers speculate that “participants were more disgusted by inferred deception on the part of the firms selling the food” (319); but they have literally no additional data to investigate that possibility.

Fourth, the study design funnels emotional responses into two simplistic measures of just two very broad emotions, disgust and anger. If someone were worried that Mary was being deceived, they might not consider themselves either disgusted or angered; instead, they might prefer to describe themselves as “suspicious” or “mistrustful.” But maybe “disgust” is the closest option, so they pick that word, then give it a rating somewhere in the middle of the scale.

This point is closely related to the second point. But it also raises issues of hermeneutical injustice. The respondents were forced by the study design to articulate their views and attitudes in the (very limited) set of categories used by the researchers; they did not have an opportunity to represent themselves in their own terms. More broadly, the respondents didn’t have an opportunity to point out flaws, limitations, and ambiguities in the study, which means that the researchers can ignore — indeed, might never even be aware of — these flaws, limitations, and ambiguities. Consequently, it’s much easier for researchers — and other readers — to impose their own interpretations on the respondents.

Finally, the study simply didn’t ask anyone why they hold the beliefs that they do; nor did it examine how people reacted when presented with arguments from other side of the GM controversy. This is the biggest issue for the interpretation — as in the blog post I quoted above — that the study shows GM opponents are evidence-insensitive. If your study does not test whether people’s views are sensitive to evidence, you cannot conclude that their views are insensitive to evidence.

It’s one thing for some editor at New York magazine to make this mistake. But the study authors make it themselves. In the conclusion, their “first key finding” is that “absolutist” GM opponents “indicate that they would maintain their opposition for any balance of risks and benefits; that is, they profess to be evidence insensitive” (320). But, as we saw up above, the respondents professed to no such thing. All the “absolutists” indicated is that their opposition to GM foods is not based on consequentialist considerations of risks and benefits.

The researchers might be extrapolating from previous research on disgust responses. In the introduction, they refer to studies of how people respond to “disgusting but putatively harmless behaviors — such as consensual sex between siblings or a family consuming its deceased pet dog.” Unsurprisingly, in these studies, “People are extremely reluctant to abandon [their] moral condemnation even when any harmful consequences (e.g., the siblings might get pregnant, dog flesh might make you ill) are explicitly eliminated” (316). Then, for the “first key finding,” they might be inferring that, since people with “absolutist” views were evidence-insensitive in these earlier studies, then “absolutist” GM opponents are also evidence-insensitive.

However, this extrapolation assumes that disgust responses to incest and eating pets are relevantly similar to the disgust responses measured in this study. And I would like to suggest that this assumption is at least not obviously true.

In the conclusion, the authors briefly report some preliminary findings from a “pilot study.” In this study, respondents “rate[d] the persuasiveness of 10 arguments in favor of GM” either before or after they answered the four “absolutism” questions (a)-(d). Because it does not consider arguments against GM foods, all four arguments concern risks and benefits, and the study design is still limits the ways respondents can answer and doesn’t test attitudes before and after presenting the arguments, this study design doesn’t actually address any of the concerns I’ve raised in this post. But it would, at least, provide an instrument that could actually test evidence insensitivity.

::: {#footer} [ May 26th, 2016 11:52am ]{#timestamp} [gmo controversy]{.tag} [science and values]{.tag} :::