Review of Parfit, On What Matters

[My first attempt at a book review: 1300 pages distilled into 1600 words. Comments/suggestions welcome!]

Derek Parfit’s On What Matters is arguably the most important work in moral philosophy since Reasons and Persons.  Its two massive volumes offer a comprehensive and densely-argued presentation of Parfit’s views in metaethics and normative theory, with occasional forays into other areas of philosophy -- including a probing analysis of the classic metaphysical question: ‘Why anything? Why this?’

Part One introduces Parfit’s “reasons-based” framework for normative theorizing, along with a number of important distinctions: (belief-relative) rationality vs. (fact-relative) reasons, subjective (desire-based) vs. objective (desire-independent) accounts of reasons, object-given vs. state-given reasons (generalizing the familiar distinction between epistemic and practical reasons for belief), and hedonic (dis)likings vs. meta-hedonic desires.  This background should prove invaluable to anyone wishing to catch up to the state of the art in normative theory.

Parfit goes on to offer compelling thought-experiments (including his classic “Future Tuesday Indifferent” agent) to cast doubt on subjective theories of reasons, arguing instead that we have objective reasons to avoid agony, whatever our desires might be.

Parts Two and Three comprise the normative core of On What Matters, arguing that adherents of the three major traditions of Kantianism, Contractualism, and Consequentialism are “climbing the same mountain on different sides.”  Parfit argues that the best forms of the former two theories converge on rule consequentialism, thus culminating in Parfit’s Triple Theory (p.25): “An act is wrong just when such acts are disallowed by the principles that are optimific, uniquely universally willable, and not reasonably rejectable.” (Those last three clauses representing Consequentialism, Kantianism, and Contractualism, respectively.)

Highlights from these sections includes Chapter 9’s discussion of what it is to treat someone “merely as a means”, and Chapter 13’s exploration of some complexities underlying the question, “What if everyone did that?”  In the latter, Parfit concludes that we must look for principles of which it is true that it would be better for any number of people to follow the principle than for none to do so.  In the former chapter, Parfit offers compelling counterexamples to standard Kantian interpretations of what it takes to treat someone merely as a means, arguing that it is better understood as a matter of failing to give someone due moral consideration.  On this understanding, killing one to save five may not constitute treating the one “merely” as a means, if the agent is taking the one’s interests into consideration, and would not use her so to achieve just any old (less morally pressing) goal.  Parfit is, however, careful to remain neutral on the question of whether harming someone as a means (regardless of whether “merely” so) might make some acts of utilitarian sacrifice wrong.

In these central sections, Parfit offers many interesting and insightful arguments for why Kantians and Contractualists should develop their theories in a more “consequentialist” direction.  He says much less about why he thinks consequentialists should be moved towards his (rule-consequentialist) triple theory.  Parfit mentions in passing that rule consequentialism captures more of our intuitions about cases than does act consequentialism, but this observation is unlikely to move those primarily motivated by the theoretical intuition that what ultimately matters is making the world a better place.  (And it seems likely that consequentialism’s appeal most often has a theoretical basis such as this, rather than depending upon its ability to match our intuitions about cases.)

The second volume of On What Matters kicks off in Part Four with a series of commentaries on the first volume from Susan Wolf, Allen Wood, Barbara Herman, and T.M. Scanlon.  Parfit’s responses follow in Part Five.  I found the exchange between Scanlon and Parfit to be of particular interest.  Parfit argues that Scanlon’s opposition to aggregating the interests of distinct people (“the numbers don’t count”) yields implausible conclusions, such as that in a society where all have a life expectancy of merely 30 years, we ought to give one person forty more years of life, rather than a five year increase to all the millions of people in the society.  Parfit suggests that Scanlon’s motivation for disallowing aggregation -- namely, to avoid greatly burdening some in order to slightly benefit many more others -- is better accommodated by a prioritarian concern for the worst-off.

Part Six, on metaethics, comprises the bulk of the second volume.  In it, Parfit sets out and defends the non-naturalist cognitivism that he takes to be a precondition for substantive normative inquiry.  Analytical naturalism he swiftly dispenses with on the grounds that it would make the assertion of the correct normative theory a mere “concealed tautology” rather than a substantive truth.  Non-analytical naturalism provides a more serious foil, against which Parfit presents a barrage of interrelated arguments in chapters 26-27.

Parfit begins this task with a careful analysis of how it is that cognitive significance and metaphysical distinctness can come apart in standard Kripkean cases of the necessary a posteriori (an analysis that is much in the Fregean spirit of Chalmers’ epistemic two-dimensionalism, though without reliance on the technical apparatus of primary and secondary intensions) in a way that shows their inapplicability to alleged identities of natural and normative properties.  Most crucially, whenever identity claims are informative, this is because they tell us that two distinct properties are co-instantiated by a single object.  The cognitive significance of ‘water’, say, may be given by a certain complex functional property: roughly, being the clear drinkable liquid found in lakes and rivers around here.  This differs from the cognitive significance of ‘H2O’, which is instead given by a certain chemical property.  The claim ‘water is H2O’ is informative rather than trivial because it relates these two distinct properties.  This is possible because the concept water is “gappy”: it refers to whatever actually fills the associated functional role.  This functional role could, for all we know a priori, be filled by all manner of chemical substances.  Hence it is informative to learn that H2O is the particular chemical property of the watery stuff.

This analysis provides the basis of Parfit’s attack on non-analytical normative naturalism.  To begin with, our concept of a normative reason, unlike water, does not seem to be “gappy” in the way required to give rise to informative identity claims.  (There’s no reference-fixing functional description of “reasony stuff” for a Kripkean to rigidify.)  More generally, Parfit challenges the naturalist to specify just what further information is imparted when they identify natural and normative properties.  For naturalism to be true, the further information must be a natural fact (rather than an irreducibly normative fact), but then we seem left with the “Hard Naturalist” view that normative concepts are (in principle) dispensable.  Parfit takes this to be a mere terminological variant of nihilism, trivializing normative claims in much the same way that analytical naturalism does.

Space constraints prevent me from detailing Parfit’s related Fact Stating and Triviality arguments against metaethical naturalism, though they greatly reward closer attention, and will surely play a central role in future debates about metaethical naturalism’s viability.

Here and elsewhere in Part Six, Parfit presents a forceful presentation of how the metaethical terrain looks from the perspective of a well worked out non-naturalism, and what the central flaws of the rival views look, from this perspective, to be.  This may at times prove frustrating to those who do not share Parfit’s basic perspective, as defenders of the rival views will be apt to think that he has given their views short shrift.  The chapters on expressivism, especially, seem likely not to satisfy anyone antecedently sympathetic to that view.  Parfit’s criticisms are by and large “external” ones, depending on assumptions that his opponents will not share.  But they are valuable criticisms nonetheless, for they do a good job of honing in on what, if non-naturalism is true, constitute the central flaws of the rival views.  As in so much of philosophy, this may be insightful and important work, even if it happens to be dialectically ineffective.

Let me wrap up this review by noting a point of incongruity between Parfit’s metaethics and normative theorizing.  His robustly realist metaethics, along with his arguments against subjectivist views of reasons, commit him to the possibility of (procedurally) ideal disagreement: One might valorize agony or make other normative errors whilst being perfectly internally coherent, and hence impervious to rational persuasion from one with different (even objectively correct) normative views.  For normative knowledge to be possible despite this dialectical stand-off, our normative knowledge must not depend upon securing the general agreement of (procedurally rational) others.  But in his normative theorizing, Parfit is in fact highly concerned by the argument from disagreement, which provides a major motivation for his development of the convergence results leading to his “triple theory”.  This is puzzling.  If a possible pro-agony agent is no threat to our normative knowledge that pain is bad, why should actual Kantians be any threat to our knowledge that (say) we ought to lie to the murderer at the door?  We already know that there are internally coherent normative views that conflict with ours, and hence could be reached by procedurally rational agents engaging in reflective equilibrium, should their starting points happen to differ enough from ours.  What epistemic difference does it make whether those views should happen to have actual advocates?*

Still, whether well-motivated or not, the first volume of On What Matters is a treasure trove of compelling thought experiments and incisive normative analysis.  And the second volume provides what will no doubt come to be regarded as the definitive statement and defence of non-naturalist cognitivism.

* I expand upon this point in my ‘Knowing What Matters’, forthcoming in P. Singer (ed.) Does Anything Really Matter? Parfit on Objectivity. Oxford: Oxford University Press.