I was only 16 at the time, but I had already developed a strong interest in moral psychology thanks to the lingering suspicion that ethics was a fatal weakness for philosophical naturalism. And so when Marc Hauser‘s now classic book Moral Minds first came out in paperback, I rushed to buy a copy.
The book was a detailed exploration of human moral cognition through the lens of trolley problem experiments and Hauser’s (now dubious) research with primates. And despite Hauser’s indefensible academic misconduct, it remains a tour de force. In fact it is still in my possession, now twice as thick and stained by sunlight from multiple re-reads.
At the time I became convinced of Hauser’s basic approach that updated David Hume in light of Chomsky’s work on innate syntax. This view says that our moral sense is at base noncognitive, that it is a product of our “passions” or sensations built into us like a “moral organ”. While morality may often seem relative to culture and upbringing, it is constrained by a “universal grammar” common to all moral orders. That grammar, I believed, was the key to resolving the moral divergences between tribes. If we could only speak clearly about our shared inheritance there could be no lasting rational disagreements.
Joshua Greene’s “Deep Pragmatism”
Consider this a premonition of what Joshua Greene has since dubbed “deep pragmatism”. Greene is also a Harvard neuroscientist and expert on trolley problems, and his recent book Moral Tribes is also concerned about what he calls the “failure of common sense morality,” i.e. when divergent moral orders collide. While I am about to be quite critical of Greene, let me say at the outset that I am actually a massive fan, and that I tend to be most critical of the ones I love.
If it is fair to say Hauser’s theory merged Hume with Chomsky’s linguistics, then Greene’s theory merges Hume with Daniel Kahneman’s Dual Process Theory. He claims our non-cognitive passions are part of our System One, or automatic / intuitive mode. But if we study the evolutionary function of our passions, we can then use our System Two, or rational / conscious mode, to resolve impassioned disputes deliberatively. Specifically, Greene posits that if morality is fundamentally about enforcing cooperation in order to reap collective benefits, two tribes with distinct ethical systems for cooperation simply have to recognize that they are using different means but have common ends.
The only thing truly novel about Greene’s argument is its tantalizing terminology. Indeed, on a recent EconTalk episode Greene admits that “deep pragmatism” is just his word for plain vanilla utilitarianism. Despite formal utilitarianism’s many problems, Greene believes clashing cultures can settle disputes by consciously reformulating their ethics based on the greatest good for the greatest number. When pressed by the host with counter-examples, Greene contended that the problems with his proposal are either merely empirical or due to an insufficient application of utilitarianism (for thinking too short-term, say).
I believe Greene makes three fundamental mistakes and thus has not provided a compelling solution to the tragedy of common sense morality. On top of this, his scientific pretenses distract from the fact that his core moral arguments come straight from the proverbial arm-chair. Indeed, as meticulously demonstrated in Selim Berker’s The Normative Insignificance of Neuroscience, Greene has a tendency to obscure his philosophical presuppositions behind a fascinating and important, but ultimately tangential, deluge of empirical data.
The three deep mistakes Greene makes are: a) to accept the Humean starting point of moral noncognitivism, b) to reify deontological thinking and utilitarian thinking as “System One” and “System Two” respectively, and c) to leap to utilitarianism when, even accepting his premises, better alternatives exist.
a) Noncognitivism Is False
Noncognitivism rose in popularity after the Enlightenment in large part due to an incorrect Cartesian view that morality like belief required an ultimate foundation. Hume put foundationalism to the test by taking it to its logical conclusion. In lieu of an infinite regress, Hume realized that connecting ought to is was impossible. Thus noncognitivism — and thus moral skepticism. And while Hume’s argument and conclusion were valid, the premise that we need foundations in the first place was dead wrong.
Since Quine, philosophers have largely accepted coherentism for beliefs. That is, it makes most sense to think of any particular belief as inhabiting a holistic web of beliefs rather than to link beliefs in a linear chain of justifications down to some “foundational” belief. When we are persuaded to change our beliefs we thus often are required to update a large number of interdependent beliefs to ensure coherence.
It turns out the same Quinean argument works for desires, preferences and other vernaculars for Hume’s passions. It’s tempting to think of desires as following a linear chain down to some base foundational affect, implanted somewhat arbitrarily by evolution. But this is an elementary error.
While true that evolution has equipped us with certain somatic states (like hunger pangs), desire (like “I desire to eat”) contains propositional content. Like beliefs, desires are part of a holistic web that we draw from in the discursive game of giving or asking for reasons. In turn, desires like beliefs are capable of being updated based on rational argumentation and the demand for coherence.
For whatever reason ethicists have been much slower to embrace coherentism for morality, preferring to soak in tired debates like deontology vs consequentialism. Greene is no different. And his attempted foundationalist argument for utilitarianism has not closed Hume’s gap one iota.
b) Dual Process Theory is Irrelevant
Using fMRIs to conflate deontology with automatic thinking and consequentialism with deliberative rationality is neither valid nor advances the argument. To quote University of Toronto philosopher Joseph Heath in his overview of empirical approaches to ethics:
Greene offered no reason to think that the theory of value underlying the consequentialist calculus was not based on the same sort of emotional reactions. In this respect, what he was really doing was presenting an essentially sceptical challenge to moral reasoning in general, yet optimistically assuming that it undermined only the position of his opponents.
Moreover, there are good reasons for thinking of deontological modes of reasoning are essentially cognitive. As Heath argues in his book Following the Rules, social norms take the form of a web of deontic constraints that we reference just like when we reference beliefs or desires when pressed to defend certain behavior. This makes social norms — and deontology in turn — analytically cognitivist. That is, regardless of the fact that deontic violations are more likely to elicit an emotional response, deontic reasoning must still inherently make use of System Two at some point.
Greene even acknowledges the more plausible explanation for why deontological violations cause more emotional fMRI activity than utilitarian ones: namely, that they each require different kinds of construal. Utilitarian reasoning tends to be about system wide outcomes and that level of construal imposes a psychological distance between the agent and the moral dilemma. But even if there is a link between construal level and dual process theory, just because utilitarian thinking is slow does not make slow thinking utilitarian!
c) Utilitarianism is a Non-sequitur
Even accepting all of Greene’s major premises, the conclusion of utilitarianism is still unwarranted. Greene suggests that the social function of moral psychology points to a “common good” through cooperation, but utilitarianism is only one possible interpretation.
In economics there are two basic approaches to social welfare, one top down and the other bottom up. The top down approach is the closest in spirit to the utilitarianism expressed by Greene. It posits a social welfare function and conditions that must hold for its maximization, aka the greatest good for the greatest number. Adherents of this approach have spanned centuries, from Bentham up to Pigou.
The other approach begins with the process of transaction itself. It posits that two people will only exchange if they each preceive a mutual advantage in doing so — that is, if the trade will move them toward a Pareto improvement or win-win outcome. This is at the heart of bargaining theory, which would presumably make it a good candidate for solving the “tragedy of common sense morality” or any scenario where conflicting interests or value systems collide.
One of the worse “tragedies of common sense morality” in history occurred in the 1600s when Protestants and Catholics fought throughout Europe in the 30-Years War. From the ruin rose modern Liberalism and the legal basis for religious toleration and value pluralism. Liberalisms core value is thus mutual advantage in the Paretian sense, not a crude formula for util maximization.
In fact there is a substantial literature within trolley problem research analyzing the effect of Paretian considerations on moral judgement. Greene is even a contributor. Indeed, in all sorts of artificial moral dilemma subjects are consistently more likely to judge harm as permissible if it leads to a Pareto improvement.
For instance, this 2011 paper [pdf warning] co-authored by Marc Hauser suggests that “Paretian considerations should be treated as an abstract principle that is operative in folk-moral judgment across a wide variety of contexts, involving different sources of threat and different degrees of contact.” Note that this fits the criteria for Greene’s “deep pragmatism” surprisingly well, without any of the attending controversy or highly demanding prescriptions surrounding Peter Singer style utilitarianism. Indeed, the authors are correct to report that Paretian considerations “provide a reason for action even for the non-consequentialist.”
Despite my skepticism for Joshua Greene’s “deep pragmatism” I strongly commend his efforts. In fact it is mostly in line with my own approach. Yet its current manifestation suffers from philosophical naiveté.
Humean noncognitivism is tempting for any student of psychology, but it turns out to be philosophically untenable. Indeed, by their very nature the deontic statuses we assign taboos and other social norms are part of a cognitive process of giving and asking for reasons. We can even reason and debate over our desires and preferences since (in contrast to pure affect) they carry propositional content.
Furthermore, while utilitarian calculations often require over-riding our more intense “gut reactions,” that does not make them any more foundational to morality. This is especially the case when it is always possible to interpret ostensibly utilitarian outcomes as resulting from a bottom up process that respects the Pareto standard.
And from the point of view of resolving tragedies of common sense morality, liberal principles like value neutrality and free expression that implicitly endorses Pareto have never been more influential on a global scale, nor more vital for our continued peaceful coexistence. The inferiority of the utilitarian alternative is shown in the recent attacks on free expression in Paris. Who today could defend Charlie Hebdo’s provocative iconoclasm on purely utilitarian grounds in a country of perhaps 6 million Muslims?
Finally, it important to remind ourselves that free expression as such is not a “Western Value” unique to the strategy of our hemispheric “tribe”. Rather, the Pareto standard of mutual benefit transcends the tribe and individual as the only proven basis for peaceful, pluralistic civilization.