In Praise of Deference

It is common to hear that you should make up your own mind and not let other people make it up for you. While I wholeheartedly agree with the sentiment and believe it’s silly to be obsequious to arbitrary authority, I nonetheless think intellectual deference is both unavoidable and a virtue. This notion may seem repugnant, an affront to liberal values, and perhaps even unnecessary, so I’ll defend it in some detail. (I want to stress, however, that while I’ll be defending my considered opinion, my argument for it will be sketchy – a full, precise, treatment would require far more time and space). My argument rests on (among others) the following two premises: (1) the world is staggeringly complicated and regularly in conflict with our intuitions, and (2) intellectual responsibility requires you to know what you’re talking about before you form an opinion. (The picture, by the way, is a high-resolution map of science, from this PLoS One paper).

It’s a cliché that the more you learn the more you realize how little you know, but it’s true: things are nearly always more complicated than they seem, and thus peeling away layers of ignorance usually reveals yet more complexities, uncertainties and undreamt of intricacy. Or as Douglas Adams is supposed to have said: "the universe is a lot more complicated than you might think even if you start from a position of thinking that it’s pretty damn complicated to begin with". If you’re at all in doubt that the world really is as complex as I make out, I invite you to examine the work of those whose job it is to understand it. Have a look at contents of any respected peer-reviewed journal, Science, say, or Nature Neuroscience or Physical Review Letters. Or, if you prefer, examine an advanced-level textbook in any explanatorily successful field such as molecular biology, astrophysics, or neuroscience. (It has to be a field that is at least partially explanatorily successful because incorrect theories can be arbitrarily simple given they need not reflect reality). Explanations of empirical phenomena almost invariably require us to employ sophisticated techniques and equipment, set aside ingrained or natural ways of thinking, invoke vast bodies of previously established knowledge, and express or establish our findings with abstruse mathematics or statistics. Given that our models and theories at least partially mirror reality, the intricacy and opacity of the former reflects the complexity and opacity of the latter. (Though there are probably exceptions, explanations tend to get more complicated – in the sense of arcane or difficult to understand – as they become more successful. There thus seems to be little reason to attribute the complexity of our theories to the incompleteness of our knowledge. Also, while successful explanations may be simple in the sense of being elegant or parsimonious, they’re extremely unlikely to be simple in the sense of easy to understand or straightforward).

My second premise is that ‘making up one’s own mind’ responsibly requires, among other things, understanding what the hell is going on before settling on an opinion. It might be news to some people, but really ‘understanding what is going on’ requires much more than ‘I have a strong inner conviction’ or ‘I once read a book on this’ or ‘someone told me so’. It requires, in general and speaking roughly, proficiency in logic and critical thinking, competence in general and domain-specific scholarship and research, knowledge of the relevant facts and how sure we are of them, mastery of the relevant techniques, a familiarity with possible alternative explanations, knowledge of at least a large proportion of the relevant literature, and more. And if (1) is true – if the universe really is extremely complicated – fulfilling these requirements is awfully demanding. Carl Sagan provided an excellent example:

Imagine you seriously want to understand what quantum mechanics is about. There is a mathematical underpinning that you must first acquire, mastery of each mathematical subdiscipline leading you to the threshold of the next. In turn you must learn arithmetic, Euclidian geometry, high school algebra, differential and integral calculus, ordinary and partial differential equations, vector calculus, certain special functions of mathematical physics, matrix algebra, and group theory. For most physics students, this might occupy them from, say, third grade to early graduate school – roughly 15 years. Such a course of study does not actually involve learning any quantum mechanics, but merely establishing the mathematical framework required to approach it deeply. (The Demon-Haunted World, p. 249)

In other words, before you can hope to begin to understand quantum mechanics, you need to master a vast body of often-difficult mathematics even before learning masses of mind-bending physics. Sagan goes as far as to say there are no successful popularizations of quantum mechanics because it is so in conflict with our intuitions, dealing with it mathematically is our only option. If you think you understand quantum mechanics without having mastered the mathematics, in short, you’re confused about what the word ‘understand’ means.

But no one person could possibly understand or discover everything about a topic, I hear you protest. Exactly! That’s exactly my point. The universe is chock full of all sorts of preposterously complicated phenomena, and no one person could ever hope to understand all or even a significant proportion of it in full. No one person could single handedly maintain a modern lifestyle (grow, harvest and process all his own food, construct his own home, build his own computer from scratch, generate his own electricity, etc.) but is forced to rely on the division of labor. Similarly, no one person could hope to understand everything about even tiny bits of the universe – the causes of climate change on a particular planet, say – but is forced to rely on the intellectual division of labor. And such a cognitive division of labor means intellectual deference. Even an expert doesn't (and can’t) know everything about her field or sometimes even her own area of specialization. It is not unheard of for scientific papers to be published where none of the authors understand all of the methods and findings. Experts defer to other experts. Experts have to defer to other experts. Deference, then, is a non-optional virtue, the only alternatives to which are agnosticism and being intellectually irresponsible. Or, to be less fancy about it, there are inevitably a large number of topics about which you can either (1) form or express unjustified opinions (i.e. be irresponsible), (2) say "I don't know" or (3) defer to the experts.

But, it may be objected, experts are sometimes wrong! Experts can be bought! Experts disagree! Indeed they are, indeed they can be and indeed they do. Experts, of course, are human beings and are therefore subject to all the familiar human failings: they are as fallible, quarrelsome, susceptible to cognitive biases and illusions, prone to social climbing, self-interested, biased, driven by ideology and whatnot as the rest of us. (Well, maybe this isn’t quite true: people self-select into science and must jump through various hoops like defending a thesis, so perhaps individuals best adapted to the ideals of science are more likely to become scientists. What is clear, though, is that scientists are not immune to these human failings). Expertise – the mastery of the techniques and in-depth knowledge of the scholarship on some subject – is not itself a huge improvement over “making up your own mind”. (By the way, I don't suggest expertise requires formal education or credentials; non-PhDs who have mastered a subject certainly still count as experts). Given the complexity of the universe and the limitations of the human mind, expertise is (for many subjects) a necessary but not sufficient condition for having justified opinions. (For one thing, it is possible to be a kind of an expert in utter bollocks: there is, for example, a huge alchemical literature, complete with rival schools, arcane jargon, different techniques and so on. And don’t get me started on postmodernism). So individual expertise in some cases doesn’t seem particularly reliable (though, ceteris paribus, it’s certainly better than nothing) and deferring to an individual expert thus isn’t necessarily such a good idea. Help is at hand, however.

The scientific method, far from denying human failings like the ones I enumerated above, exists exactly because of them: it is because the human mind is so prone to error and bias that we need this vast, expensive and seemingly inefficient set of institutions, norms and practices we call “science”. Science, roughly and to first approximation, is a collaborative enterprise aimed at a unified description and explanation of natural phenomena where the ultimate arbiter of truth is empirical experimentation, the reliability and quality of which is evaluated by a community of scholars through peer-review and replication. A scientist, then, is a person who attempts to describe and explain the natural world by testing empirical hypotheses in collaboration with a group of other researchers by reviewing their work, and producing work that is in turn reviewed. Convincing your peers – who will criticize your ideas harshly and subject them to industrial-strength skepticism – by publishing in peer-reviewed journals, presenting papers at conferences, and, more informally, debating in seminars and pubs, is at the very heart of science. (The mark of a crank is not being embedded in such a system of cooperation, dismissing criticism as some conspiracy or another, and claiming the mantle of Galileo). The point of this collaborative enterprise is to minimize bias: an individual wants her ideas to be true, is limited by peculiar psychological traits and a particular background, knows only some fraction of the relevant facts, and suffers from a whole ménage of other cognitive biases and illusions. A community dedicated to collaboration (and competition) – whose members aim at rigorous explanation and consensus, and who agree on the primacy of empirical demonstration – can overcome many (though obviously not all) of these biases because, in a sense, one individual’s biases cancels out another individual’s biases. Manipulating the world in such a way to hold certain variables constant while varying others – i.e. doing controlled experiments – is the most powerful technique ever invented to discover nature’s secrets (Daniel Dennett aptly called it the “technology of truth”), and having an entire social system (complete with attending values) and a supporting set of institutions (universities, granting agencies, journals, professional organizations etc.) multiplies these powers by minimizing human failings in interpreting and conducting the experiments. Obviously, science is not perfect, but because of how it is organized – and crucially, because there is a way of falsifying hypotheses – it is a self-correcting process: explanations are tested, discarded and repeatedly refined, which then slowly ratchets our theories closer to the truth over time. (If you have doubts about the success of science, please stop reading, apply head to desk and repeat until you come to your senses).

So why is this whole story about science relevant in a post about deference? Simple: because when the relevant experts agree that some theory or explanation is correct, you can be reasonably confident that the theory is in fact correct. In other words, given the nature of the scientific method – given that claims are peer-reviewed, subject to intense scrutiny, tested, re-tested, refined, and so on – when there is a consensus among the relevant experts, it is reasonable to believe they are right. Of course and again, experts are people, so you can’t be certain the theory or proposition is true just because there is consensus, you can only be rather confident. At a minimum, individuals who disagree with the consensus have the burden of proof – they must show it is false, the majority does not need to refute the alternatives. (Though they often do). Since showing some consensus theory in science is false (or incomplete) is going to be extremely difficult, those who wish to disagree damn well better be experts themselves. (Consensus theories are of course sometimes overturned: witness plate tectonics). In general laypeople are not qualified to have independent opinions about complex topics – they lack the means to come to justified beliefs – and it is especially unreasonable for a non-expert to take a stance contrary to consensus. The upshot is that, firstly, it is reasonable to defer to the consensus opinion of the relevant experts – so I can justifiably say ‘E=mc²,' ‘DNA carries heritable information’, ‘there is a supermassive black hole at the center of the Milky Way’, etc. And, secondly, a layperson who disagrees with expert consensus – denying evolution by natural selection, anthropogenic global warming, that the Earth is about 4.5 billion years old, etc. – is unreasonable in the extreme. Experts get to have opinions on scientifically controversial questions, experts get to disagree with consensus; laypeople get to defer to consensus or reserve judgment. Doing otherwise, I think, shows what Bertrand Russell called (in another context) an “impertinent insolence toward the universe”. Scientists are often accused of arrogance and maybe I’ll be accused of this vice as well for telling laypeople who disagree to shut up. But I think the opposite is true. It is extraordinarily arrogant to have (independent) opinions on complex questions without being willing to pay your dues first – that is, without studying the question for years, reading the scholarly literature, mastering the relevant techniques and mathematics, and so on. Thinking you are entitled to an opinion without paying your dues is the very epitome of intellectual arrogance. And it is especially arrogant – mind-bogglingly so – for a non-expert to have opinions that contradict the consensus of the tens of thousands of intelligent, diligent and dedicated people who have spent decades studying, debating, doing research on and thinking deeply about their respective disciplines. The bottom line: be an expert, defer, or suspend judgment. (To be clear: I’m making an epistemic and not a political claim. People have a right of free speech and conscience, so they can form and express any opinion they like. But that doesn’t mean they have an intellectual warrant to do so).

To be sure, there are a whole bunch of complications here (how very appropriate, no?). For one, scientists are not the only people worth deferring to: if there is a consensus among plumbers, for example, that the best way to fix problem P is to do X, Y and Z, I’d be inclined to say that’s quite reliable. Nevertheless, while there are other groups (‘communities of practice’, etc.) that can reasonably be deferred to, for my purposes it is simply worth noting that scientists are one such group. Secondly, there are certainly degrees of expertise and deciding when someone has crossed the threshold to expert status is fraught with difficulty (though beware the false continuum). More problematical is the question of what topics are ‘sufficiently complicated’ that laypeople shouldn’t have independent opinions. Saying, for example, that only psychologists are qualified to determine whether Bob has a crush on Tamba, for example, is preposterous. The (partial) solution here, I think, is to invoke Richard Dawkins’ notion of the “Middle World” and to distinguish between explicit and implicit knowledge. Let's start with the former. The human mind, Dawkins convincingly argues, evolved to deal with and understand the everyday world we inhabit: of medium sized objects, operating at low-velocities, including animals and other people. “Folk biology”, “folk psychology”, and “folk physics”, for example, are regularly wrong in detail (sometimes spectacularly so), but they are often reliable when in our ‘natural environment’. The fact that we can, say, play football (which requires sophisticated ballistics), navigate a cluttered room (which requires sophisticated optics and physics), and cooperate and compete with other humans (which requires a complex theory of mind) and so on, shows we are far from cognitively incompetent. The human mind is (largely) good at solving the problems we encountered often in our evolutionary past: it is good, in other words, at Middle-World problems. But our ancient ancestors never traveled near the speed of light, never lived in large-scale complex societies, never interacted directly with the quantum world, never needed to understand the nature of stars and so. On certain topics, then, laypeople are reliably (though almost always incompletely) competent. There is a fundamental difference, in other words, between the statements “Bob is upset at Mary for cheating on him with John” and “E equals mc²”: the mind evolved to deal with the former, but not the latter. Important, also, is the difference between implicit knowledge (or behavioral competence) and explicit knowledge (i.e. justified true belief, with some modifications), While being a football quarterback, say, requires a brain capable of solving complex physics problems, this does not mean football players explicitly understand the relevant physics. When I move my arm to pick up my cup of coffee, my brain does damn complicated trigonometry, but I don't know that trigonometry explicitly - my brain's calculations are not consciously accessible to me. What this means is that behavioral competence or implicit knowledge in some domain (seeing, interacting with people and non-human animals, walking about) does not imply explicit knowledge of the underlying science. (A slam dunk argument for this, by the way, is our inability to build robots even remotely as competent as we are).

There are several more complications but I'll only mention one more. It is regularly extremely difficult to determine whether there is a scientific consensus on some topic and, if so, what it actually is, especially so when ideologically committed pseudoscientists muddy the waters. For example, the overwhelming majority of the relevant experts agree that evolution by natural selection – the fact of evolution and the theory of natural selection (etc.) – is established beyond all reasonable doubt. Creationists, however, have tried to argue there is no such consensus and have even compiled lists of scientists who 'disagree with evolution' (c.f. Project Steve). Laypeople who do not understand the scientific method might see two sides 'debating' and have real difficulty figuring out who to believe. They might not realize, for example, that scientific consensus is not about lists of people who agree or disagree, that only the relevant experts are important (engineers who 'disagree with evolution' qua engineers tell us nothing), that no paper critical of evolution has appeared in a mainstream peer-reviewed journal (save by fraud) for several decades, and so on. Even a layperson convinced it is important to defer to scientific consensus, then, will sometimes have real trouble determining whether there is a consensus and what the consensus actually is. There are two ways of dealing with this problem, I think. The first is to employ the much underused phrase "I don't know". Agnosticism  isn't particularly popular, but I think openly admitting what you do and do not know is one of the most important intellectual virtues. So when you can't figure out what the consensus is (or whether there is one), it doesn't suddenly become reasonable to form opinions in the absence of knowledge; agnosticism is then the reasonable course. The second answer to the above problem is having certain metacognitive skills - an understanding of the scientific method and the academic process, familiarity with cognitive biases, skepticism and an ability to assign onus appropriately, finely-honed critical reasoning skills, a basic understanding of statistics, and so on - that are useful for evaluating any claim. While these are not sufficient to understand the details of any area of science, it allows for a 'popular level' grasp of the field, which in turn enables one to identify what the findings are (and some of the reasons why they're established), and determine, with some (hard) work, whether there is a consensus on some question.

"The fundamental cause of the trouble," wrote Russell, "is that in the modern world the stupid are cocksure while the intelligent are full of doubt". While dividing the world into 'the stupid' and 'the intelligent' is probably going too far, I think Russell is on to something: it is those who are ignorant of science who are certain they're right - even when they're not. The Dunning-Kruger effect suggests why: the intellectually unskilled lack the intellectual skills needed to recognize that they are unskilled. They are, in other words, unskilled and unaware of it. Dunning and Kruger also showed, however, that people could be trained to become somewhat more competent, which then allows them to recognize the depth of their incompetence. What I have shown in this post, I hope, is that, in a sense, we are all cognitively incompetent relative to the stupendous complexity of the universe. It is science (or, more broadly, the project of secular reason) that holds out a candle in the dark: we have uncovered nature's secrets only because we invented this 'technology of truth' and those who wish to advance our knowledge or understand a particular phenomenon deeply must approach it humbly and pay their dues in long and intensive study. Those of us who have not paid our dues in a particular field can only defer to those who have or remain agnostic. There is no reasonable alternative. 

The last word goes to the great Bertrand Russell:

The demand for certainty is one which is natural to man, but is nevertheless an intellectual vice. So long as men are not trained to withhold judgment in the absence of evidence, they will be led astray by cocksure prophets, and it is likely that their leaders will be either ignorant fanatics or dishonest charlatans. To endure uncertainty is difficult, but so are most of the other virtues.