In a complex crisis, scientists cannot avoid making value judgments.
Read the Reflection, written 28 July 2021, below the following original Transmission.
The rapidly unfolding COVID-19 pandemic has brought the interface between scientists and policymakers directly into the public eye. Examples include the role of Anthony Fauci in daily White House press briefings and the impact of a report by Neil Ferguson and his Imperial College London colleagues on decisions by several national governments. One might hope that this ongoing interaction between scientists and policymakers would respect a certain division of labor. Policy-facing scientists would provide politicians with decision-relevant facts, and in turn, politicians would make decisions that require them to assess the value to society of different possible policy outcomes.
As clean and compelling as this division of labor is, I don’t believe that it is achievable, especially when dealing with a system as complex as a pandemic. In responding to this crisis, scientists must embrace the fact that they are being called upon to make ethically-loaded decisions, including in cases where this may not be immediately obvious.
The idea that science ought to be free of value judgments has a rich history. As early as the nineteenth century, W.E.B. Du Bois (1898, cf. Bright 2018) argued that public trust in science could only be preserved if science was insulated from social and political concerns. In the twentieth century, the decision theorists Richard Jeffrey (1956) and Issac Levi (1960) put forward mathematically precise frameworks for formalizing the division of labor between scientists and policymakers. According to them, scientists should provide policymakers with an empirically-supported assignment of probabilities to different relevant outcomes under a set of policy alternatives. It is then incumbent upon policymakers to do the value-laden work of evaluating the desirability and probability of each outcome under each policy, and formulating a decision rule that outputs an optimal policy. In the context of the current crisis, this division of labor would proceed as follows. Scientists would provide policymakers with an assignment of probabilities to the various possible public health and economic consequences of policies such as extreme social distancing, gradual de-quarantining, and the isolation of vulnerable populations. Elected policymakers would then use these probabilities, along with their own normative judgments, to arrive at a decision as to the optimal policy.
However, as the Australian National University’s Katie Steele argues in a 2012 paper, scientists rarely possess an evidence base that allows them to be confident in a single assignment of probabilities to different possible outcomes. Much more often, and especially in the face of significant uncertainty about the behavior of a system, the best that scientists can offer is a range of probabilities that a given outcome will occur. Here, scientists face a clear tradeoff. Wide ranges are much more likely to be correct, but can offer limited guidance to policymakers. Narrow ranges facilitate political decision-making, but are more likely to be wrong. Thus, when scientists decide how to report results to policymakers, they have to balance the need for action-guiding advice against the risk of their advice being wrong. These are value-laden decisions that cannot be outsourced to policymakers. Thus, as politicians continue to call on the expertise of scientists in order to respond to the current pandemic, scientists must embrace the fact that they are being asked to make ethical decisions. This may not be the ideal role for a scientist, but it is one that each epidemiologist, virologist, economist, and anyone else in a position to provide scientific advice to policymakers finds themselves in, like it or not. Likewise, the public must accept that even though scientific policy advisors have not been popularly elected, we have no choice but to grant them a certain level of value-laden decision-making power, or else abandon the idea of scientifically-informed policymaking entirely.
Santa Fe Institute
Read “What the 1918 flu pandemic can teach us about coronavirus drug trials,” in The Guardian (April 5, 2020).
Read more posts in the Transmission series, dedicated to sharing SFI insights on the coronavirus pandemic.
Listen to SFI President David Krakauer discuss this Transmission on Episode 26 of our Complexity Podcast
July 28, 2021
Why We Can’t Depoliticize a Pandemic
In my 2020 Transmission, entitled “In a Complex Crisis, Scientists Cannot Avoid Making Value Judgments,” I stressed that scientists offering solutions aimed at mitigating the COVID-19 pandemic must recognize that their recommendations have an unavoidable ethical dimension. Since then, mitigation strategies for COVID-19 (e.g., masks, social distancing, or vaccines) have become deeply politicized. This is powerfully demonstrated by the fact that, as of July 2021, a map of US states indicating the percentage of adults in each state who have been vaccinated against COVID-19 very closely resembles the 2020 electoral map, with Biden-voting states tending to be more vaccinated and Trump-voting states having a lower percentage of vaccinated adults.1
In July 2021, Dr. Anthony Fauci called this political divide “unfortunate,” and declared that US citizens needed to “put politics aside” to increase vaccination rates.2 Indeed, we have good reason to believe that if Trump-voting adults were vaccinated at the rate of Biden-voting adults, then the current public health burden of COVID-19 would be less severe.
While politicization of the pandemic may be unfortunate, I will argue here that it is also inevitable. The pandemic asks each of us to arrive at a personal view of how COVID-19 works, and what we ought to do about it. Recent work in cognitive psychology suggests that we learn the causal structure of the world primarily through social interactions. Thus, each person’s understanding of the world tends to be shaped by how those around them understand the world. Moreover, foundational work in philosophy argues forcefully that our understanding of any given phenomenon is shaped by how that phenomenon fits into our broader worldview. That is, we understand the world holistically. We can’t just silo off pandemics and understand them using tools different from those that we use to understand other social phenomena, including politics. As a result, efforts to de-politicize the vaccination response to the COVID-19 pandemic amount to swimming upstream against the hard reality of human cognitive mechanisms.
In what follows, I will explain briefly how these two theses (i.e., the inherent sociality of causal learning, and the holistic nature of a person’s worldview) jointly imply that it is unlikely that we will ever be able to truly de-politicize a pandemic at scale.
As very young children, we begin understanding the world in terms of cause and effect. We learn which interventions lead to what sorts of outcomes, and we start to strategize accordingly. This capacity for causal reasoning underwrites our ability to explain why things happen in a particular way in the world around us. Importantly, we do not build this causal apparatus on our own. As Legare, Sobel, and Callanan explained in a 2017 literature review, experimental and observational evidence from cognitive science strongly suggests that causal learning in early childhood is an inherently social affair.3 Children use their parents and other adults as sounding boards for repeated “why questions” that inform their mental representation of the causal structure of the world, which in turn informs their understanding of why things happen in a particular way. As such, each person’s understanding of the world is mediated through the contingent features of their unique social environment.
This inherently social nature of our understanding of the world continues throughout adulthood. As Sloman and Fernbach argued in their 2017 book, The Knowledge Illusion, all causal cognition involves outsourcing mechanistic knowledge to experts and black-boxing the fine-grained details of how many processes work.4 This serves to reduce the cognitive load of understanding the world, but it is only possible because of our high degree of sociality; when we do need to understand something in granular detail, each of us has a cognitive social safety net that we can call on, by asking questions of those around us. Knowledge of the causal structure of the world doesn’t really live inside each of our heads; it is distributed throughout our social network.
Almost no one is an epidemiologist or virologist, and so almost all of us must call upon our social network when we attempt to understand both how the COVID-19 pandemic works, and how to mitigate its negative impacts. Social networks are inherently biased along a number of axes, including political ones. Democrats are more likely to outsource their intellectual labor to fellow Democrats, and Republicans to fellow Republicans. Thus, once a certain amount of political bias is introduced into some individuals’ understanding of how a pandemic works, these biases can quickly propagate and amplify, as more people turn to their social networks to help them understand the pandemic.
It is at this point that it becomes natural to insist on de-politicization. If each of us could see the pandemic through a strictly scientific lens, without political bias, then we would settle on a unified, and more accurate, understanding of how to mitigate its effects. However, the history of science teaches us that it is difficult to strictly separate the various aspects of our belief-forming apparatus. As Quine (1960) famously argued, it is always possible to hold on to some particular belief “come what may.”5 Consider the case of late medieval astronomers who, rather than accept a heliocentric model of the solar system, constructed increasingly complicated geocentric models. These astronomers were able to maintain one belief—that Earth was at the center of the solar system—by revising related beliefs about the laws of planetary motion. It is for this reason that philosophers of science like Carnap, Quine, and Kuhn have viewed each of us as having a fundamentally “holistic” worldview.6 That is, each of us sees the world through a richly interconnected web of particular beliefs about how the world works. In the face of data that contradict any particular part of that web, we can choose which parts to revise, in order to remain consistent with the data. Different agents may adopt different rules of revision, and it is not obvious that there is any rational basis for preferring one set of rules to another.
The history of science teaches us that it is difficult to strictly separate the various aspects of our belief-forming apparatus.
As argued above, cognitive science strongly suggests that our understanding of the world is inevitably informed by our social networks, which are themselves heavily politicized. Thus, our political and scientific views are deeply interconnected, and how we revise our understanding of the world in response to new data is invariably informed by our personal understanding of both science and politics. Short of a mass de-politicization of our social networks, it is unlikely that this close connection between science and politics will ever be eliminated.
This may seem pessimistic, but I do not believe that all hope is lost with respect to addressing wicked problems like COVID-19. Rather, I conclude only that efforts at de-politicization of the pandemic are likely to be in vain. Instead, and in keeping with the spirit of my first Transmission, I would urge scientists of all fields to embrace the inescapably political dimensions of their work. For instance, when scientists argue (with good reason) that vaccines are the most effective tool for curtailing the spread of COVID-19, they must recognize that, whether they like it or not, this message has a political valence. Thus, getting this message to the public requires scientists to embrace politics rather than shirk them. This is especially true when scientific recommendations bear so directly on the well-being of so many.
Read more thoughts on the COVID-19 pandemic from complex-systems researchers in The Complex Alternative, published by SFI Press.
1 C. Cillizza, “2 Maps that Explain How Partisanship has Poisoned our Fight against COVID-19, CNN, July 10, 2021, https://www.cnn.com/2021/07/08/politics/electoral-map-vaccine-map-covid-19/index.html
2 S. Shahrigian, “Dr. Fauci Decries Nation’s Political Divide over COVID Vaccinations,” New York Daily News, July 11, 2021, https://www.nydailynews.com/news/politics/us-elections-government/ny-fauci-delta-variant-nasty-strain-20210711-ymir45yrpresrhfdtviqxvzmuu-story.html
3 C.H. Legare, D.M. Sobel, and M. Callanan, 2017, “Causal Learning is Collaborative: Examining Explanation and Exploration in Social Contexts,” Psychonomic Bulletin & Review 24(5): 1548-1554, doi: 10.3758/s13423-017-1351-3
4 S. Sloman and P. Fernbach, 2017, The Knowledge Illusion: Why We Never Think Alone, New York, NY: Riverhead Books.
5 W.V.O. Quine, 1960, Word and Object, Cambridge, MA: MIT Press.
6 Although Quine set up Carnap as a foil for many of his arguments, and Kuhn explicitly viewed his project as an extension of Carnap’s, Becker, 2002, “Kuhn’s Vindication of Quine and Carnap,” History of Philosophy Quarterly 19(2): 217-235, convincingly argued that all three are committed to a version of the epistemic holism described herein.