Ophelia Benson summarized Michael Shermer’s latest foot-in-mouth in his row with Massimo Pigliucci over whether and to what degree moral philosophy should become a moral science instead. Reading their exchange, I find Shermer is more inclined toward ideological biases and superficial worldview declarations than actual, sound, self-critical, well-thought analyses in this matter.
As a result, Shermer is doing a really awful job of defending what I actually agree with: that it’s high time moral philosophy began to be folded into the sciences (the same way philosophy of mind became psychology and cognitive science, for example). And that’s annoying. It’s like when awful Jesus myth theorists make it harder for me to argue that Jesus might not have existed after all, by their constantly using terrible arguments that then get falsely imputed to me. My case then gets judged by their failures. I now worry the same will happen here. So let me try to nip that in the bud.
The General Point
Pigliucci already exposes Shermer’s lack of understanding in this latest matter generally, so I won’t rehash all that. You can follow along with that exchange on your own. Shermer opened with this (“The Is-Ought Fallacy of Science and Morality”) at The Edge, which was in spirit correct but somewhat inept in execution, and Pigliucci took him to task for its obvious omissions in a reply at Rationally Speaking. To which Shermer responded there. To which Pigliucci then replied again (possibly it’s still going on at this point, but that’s as much as I’ve so far read).
Shermer is getting owned in this debate. He is really not coming across well. In particular, not only is he out of his depth, once again he doesn’t seem capable of learning. He pretty much just circles the wagons around his original claims. He isn’t really listening to Pigliucci or answering his concerns. And the actual goal (to articulate an actionable scientific research program in moral science) is getting lost, even though it’s supposed to be what Shermer is arguing for. Badly.
Disclosure is warranted. Like I said, I agree with Shermer’s overall point: morality should become a science. And I have responded harshly to Pigliucci on this issue before (in a letter Pigliucci read but that has not yet been published). In a previous review, Pigliucci had horribly straw manned the case for a moral science made by Sam Harris (in his book The Moral Landscape), not only misrepresenting what Harris argued, but falsely equating Harris’ insufficient defense of the case with the indefensibility of the case altogether. Pigliucci carries that flawed theme forward in his critique of Shermer, but otherwise Pigliucci is spot on. Everything he says about the ineptitude of Shermer’s arguments is pretty much correct, and I agree with a lot else Pigliucci says in this exchange.
On the matter of making morality a science, however, both Shermer and Pigliucci are wrong, for reasons I already spelled out in formal detail in “Moral Facts Naturally Exist (and Science Could Find Them)” a couple years ago, which is chapter 14 of The End of Christianity (pp. 333-64, 420-29). I had previously elaborated on that idea less formally in Part V of Sense and Goodness without God (and further illumination can be gained from my past debate over desire utilitarianism: see Goal Theory Update; and likewise from my discussion of Moral Ontology). Shermer is wrong for all the reasons Pigliucci documents. But Pigliucci is wrong for all the reasons I document. (As a philosopher, Pigliucci would be most concerned to read my chapter in TEC rather than the material in SaG or my blog, which should be viewed as an informal appendix to the formal demonstration of that more recent work in print.)
Pigliucci is thus really just tearing down a straw man here, since Shermer’s awful defense of a science of morality is easy pickings for any well-informed thinker trained to spot a fallacy. I do hope someday Pigliucci will constructively critique my chapter instead, if we want to make progress in discussing a possible research program for a real science of prescriptive (and not merely descriptive) morality. Of course, I say this knowing that in his reply to Shermer, Pigliucci actually agrees with a lot of what I say there. Indeed, my paper uses careful and peer reviewed philosophy to demonstrate what Shermer is trying to argue ineptly with a lot of sloppy inferences and hand waving.
The Specific Point
But what really gets my goat is the way Shermer uses evidence, and one particular example is emblematic. I mean the way Shermer uses a particular research paper, a product of a private think tank (and not something you can find yet in a peer reviewed academic journal as far as I know): Daniel Sacks, Betsey Stevenson, and Justin Wolfers, “Subjective Well-Being, Income, Economic Development and Growth” [PDF here], which (contrary to what Shermer says) is policy paper No. 5230 (October 2010) for the German Institute for the Study of Labor (aka the IZA).
On this Shermer says (my emphasis in bold):
…in a study…entitled “Subjective Well-Being, Income, Economic Development and Growth” by the University of Pennsylvania economists Daniel Sacks, Betsey Stevenson, and Justin Wolfers…they compared survey data on subjective well-being (“happiness”) with income and economic growth rates in 140 countries. The economists found a positive correlation between income and happiness within individual countries, in which richer people are happier than poorer people; and they also found a between-country difference in which people in richer countries are happier than people in poorer countries. As well, they found that an increase in economic growth was associated with an increase in subjective well being: “These results together suggest that measured subjective well-being grows hand in hand with material living standards.” How much difference? “A 20 percent increase in income has the same impact on well-being, regardless of the initial level of income: going from $500 to $600 of income per year yields the same impact on well-being as going from $50,000 to $60,000 per year.” Contrary to previous studies, the economists found no upper limit in which more money does not correlate with more happiness. As well, on a 0-10 scale measuring “life satisfaction,” people in poor countries averaged a 3, people in middle-income countries averaged a 5-6, and people in rich countries averaged a 7-8 (Americans rate their life satisfaction as a 7.4). The economists’ conclusion confirms my moral science theory that the survival and flourishing of individuals is what counts.
Now, let’s set aside that last line, which is a non sequitur–and for many more reasons than Pigliucci already points out (I mean, honestly, the complete disconnect in terminology from that sentence, the conclusion, with the preceding sentences, the premises, would get Shermer an F in almost any college course I can think of that he could write such a paper for). Likewise, let’s set aside the overall assumption that subjective measures of life satisfaction are sufficient in themselves to reach moral conclusions about the behaviors that increase them (Pigliucci already rightly takes Shermer to task for that sloppy inference).
I am here concerned about his use as “evidence” of the statement I put in bold, particularly the “contrary to previous studies” remark, but also his extending it to, as he quotes from that paper, the “conclusion [that] economists’ traditional interest in economic growth has not been misplaced.” This put up a skeptical red flag in my mind, because every previous study I know found otherwise, and so they have a big burden to carry here if they are going to overturn all previous science on this subject. Shermer should certainly agree. He should also certainly acknowledge that he agrees. Instead, he just assumes this is the good science, and all the rest is the bad science. All because it says what he wants. Bad skeptic.
Two key claims are implied by the study he cites (plus a few other claims of lesser importance here): that economic growth helps everyone and is therefore a desirable goal in itself, and that even the ultra-rich are happier than the merely rich, therefore gaining wealth is always a desirable goal. These conclusions are not explicitly stated in the study, but the authors hint at them repeatedly and have gerrymandered their presentation to favor that conclusion. And Shermer is explicitly using their paper as if it argues these things (as his remarks I put in bold confirm).
But there are two problems with these claims, which expose problems with the way Shermer uses this study, which further confirms Pigliucci’s criticisms of Shermer’s entire abuse of evidence generally, an abuse that more resembles pseudoscience than the actual science Shermer wants done in the moral sphere:
(1) The authors of this study did not control for the one factor that actually matters with respect to GDP: percentage of GDP spent on social services and infrastructure rather than hoarded by top earners. For example, just contrast Norway with Saudi Arabia, and right away a claim like “national economic growth is always a desirable goal for everyone” is exposed as somewhat absurd. Obviously national happiness will improve with GDP insofar as higher GDP results in improved social services: more and better police and fire departments, more and better government institutions aiding the public, more reliable social safety nets, more and better roads and sanitation and pollution abatement, more and better workplace safety regulation and building codes, cheaper and cleaner public water, and so on.
If you want to claim that this byproduct of GDP-increase is not the thing responsible for the measured increase in general happiness of a population, you would have to actually control for that factor. But they didn’t do that. Why? Possibly because if they did, they wouldn’t find what they appear to want. They’d probably find that public services investment is the actual factor, and GDP increase is not the real cause–that it often (but not always) correlates with happiness only because it often (but not always) correlates with public services investment, and it’s public services investment that causes the observed boost in happiness. Hence that old warning: correlation is not always causation. Skeptics should have that tattooed on their brain. Shermer, evidently, does not.
(2) They also conceal reality by using a logarithmic function for income. If you know anything about math, you might immediately be getting suspicious. The usual finding, of many previous studies, which they imply they are contradicting (and which Shermer went out of his way to mention they are contradicting), is that as wealth increases, gains in happiness decline. But that’s exactly what this study also shows. By mapping gains in happiness against the logarithm of gains in wealth, all they prove is that more wealth produces smaller gains in happiness. Precisely what all previous studies found.
The key to unraveling their entire argument is the single table on [PDF] page 41, figure 1 (or rather, I should say, their implied argument; all the paper actually argues is that they found this logarithmic relationship, which by itself is a somewhat trivial finding, since it does not disagree with any prior study and doesn’t tell us anything new–but they do not write their commentary as if their improvement on past studies was thus trivial, as in fact it is, but as if it discovered something revolutionary, which it didn’t). Here is the graph in question:
Notice the logarithmic scale of income along the bottom, against their non-logarithmic scale of subjective happiness (“life satisfaction”) along the left side (measured on a scale of 1 to 10). Note that past studies found that after annual income reaches around $70,000 (in adjusted dollars), gains in happiness become too small for continued acquisition of wealth to be a significant factor in making you happy. Now look at what their table says: a bump from $64,000 annual income to $128,000 annual income produces a gain in “life satisfaction” of a mere fraction of a unit–from about 7.5 average to about 7.8 average, a difference so small I would expect it to be consumed by any subject’s margin of error in reporting their own degree of happiness (see my marked up version of this graph below).
Hmmm. That’s basically what past studies said: such an income bump has only an insignificant effect on overall life satisfaction. There are very few things we should be willing to sacrifice in order to merely increase our life satisfaction from a 7.5 to a 7.8 out of 10. Thus, once we reach a certain income level, pursuing more income starts to lose any significant value for the purposes of moral decision-making (even if polled subjective happiness is the ultimate goal of moral action as Shermer claims). Indeed I suspect that more income beyond that only adds significant life satisfaction insofar as we do worldwide good with it (rather than consuming it selfishly), but that’s at least something we can study.
Meanwhile, a gain from $16,000 to $64,000 makes a significant difference in happiness: from an average rating of 6.5 to 7.5. And of course, these are averages. Tease out the effect of going from $16,000 to $64,000 for someone with medical problems (or who has a large family or is living in an expensive city or in a terrible crime zone that they can’t afford presently to escape) and I’m certain the gain in life satisfaction at that level would be even greater. But even with that speculation aside, from the data already available–the very same data Shermer is using–it’s clear that helping the poor is more important than doubling the incomes of the rich.
I’ve illustrated all this by marking up that same table:
Also notice that their table ends at $128,000. In this paper’s commentary they talk about moving from the bottom 5% to the top 5% as yielding great gains in happiness. Yet their table doesn’t even include the latter. In the U.S. the top 5% begins at around $166,000, which on their logarithmic scale would begin a bit off the edge of their current table. Yet we should want to know if their correlation line actually continues with the same slope–as their paper implies–or if in fact the slope begins to level off at some income level–and if so, at what level of life-satisfaction.
Certainly, it must. Their claim of a consistent logarithmic relationship is impossible, since the scale stops at 10, yet their correlation line would predict someone making a billion dollars a year will report a life satisfaction in the 100s [or actually, about 14.5]. Obviously not. In fact, I doubt it even goes to 10, as if there was some income level at which the average earner at that level reports a life satisfaction of 10. (When there can be no number above 10, averaging at 10 is really damn hard to do.) In fact I doubt the reported average even goes to 9. So where does it level off? These authors coyly don’t tell you. I suspect it levels off around or just under an average of 8, judging from where they decided to stop telling you. Which means once you get to around $120,000 in annual income, you are unlikely to get any happier on average than just under an 8. Essentially what everyone has been saying in this field for years.
And that’s joint household income, BTW, i.e. the total income of a couple (when a household consists of a couple, or even in some cases an extended family), not their income individually, which is significant because it refutes what Shermer implies (repeatedly), that it is individual success that is being measured here, when in fact it includes the cooperative success of pair-bonded couples. Each of whom can earn $32,000 and be just as happy as if they were alone and making $64,000. Think about that. It means that increasing the income of low wage earners is far more important than increasing the incomes of the already-rich. It means fair wage laws might be a good idea. Indeed, they might be better than all the government boons and largesse we throw at banks and corporations. Protecting million dollar bonuses for a few thousand people is morally indefensible when doubling the minimum wage would far more greatly improve the lives of tens of millions.
Sure, I can already anticipate the usual objections, like that increasing the minimum wage would hurt more people by increasing unemployment or the cost of goods and services, and so on. But once we start that conversation we’re having a debate over the facts that is far more nuanced than Shermer pretends. Shermer is right on one thing: science has to answer these questions (like, what actually are the effects of raising the minimum wage?). They cannot be answered from the armchair by pithy ideological assumptions about how markets are supposed to behave. Markets don’t obey ideologies. They obey reality. And that’s science’s domain. And the thing about science is, it often doesn’t get the results you want. You have to be okay with that. Otherwise you’re just a creationist obsessing over something other than creationism.
Anyway, back to this paper’s non-revolutionary finding that increasing income after a certain point has only negligible effects on personally reported happiness: the same mistake plagues their cross-country comparisons by GDP and GDP growth. Once again, even there, their own data show that continued gains on this logarithmic scale are of rapidly diminishing importance. Perhaps we should not, therefore, pursue national economic growth at the expense of things like ghettoizing our inner city neighborhoods, or protecting banks “too big to fail” from the actual consequences of their misbehavior, or allowing the wealthy to take corporate welfare and other “perks” from the government (while simultaneously denigrating benefits in aid of the poor).
Continued economic growth should not therefore be a major moral goal even by Shermer’s own reasoning. I doubt it even makes the list of top moral goals, once your nation has reached a certain baseline level of stable economic success. Showing that gains still continue after that is moot when those gains are so small that they simply aren’t worth the actual social costs of pursuing them.
Shermer’s ideologically obsessed failure to recognize nuances and complexities like this (which Pigliucci also points out), and his misuse of this paper in particular to push his own armchair moral agenda under the false veneer of it being scientifically proven, is just sloppy and insulting. It would be one thing if at least he said there may be nuances to this and that all he is really calling for is that we actually study these facts further to see what shakes out, and if he at least acknowledged that this does not yet connect to any actual prescription for moral action (Shermer never actually gets around to explaining what moral rules or behavior these facts are supposed to support, beyond vague slogans about individualism and market economics, which are meaningless shibboleths, words with emotive force but largely devoid of any clear content), and if he also actually responded to Pigliucci’s point that even the question of what’s best for us (what we as individuals most want for ourselves) is a matter of scientific inquiry (inquiry that might yet turn out a result Shermer doesn’t expect), and if he would concede Pigliucci’s point that right now, in the absence of the scientific research we actually need, a more scientifically informed philosophy is the only stopgap we have left for answering moral questions, maybe if he did all that then what he’s been saying on this subject would be salvageable. But he just doesn’t get any of that. He doesn’t think. He doesn’t listen. He just plays the fiddle of verification bias.
And I’m getting tired of that.
A Better Research Proposal
Shermer has one thing sort of vaguely right: science is already really good at teasing out cause-effect relationships, and if well-honed and ingeniously-directed, it could do a really good job at working out what the total lifetime consequences are for certain behavior choices (consequences psychologically, socially, reciprocally, and with respect to interpersonal relations). For example, we could determine if cultivating and abiding by the virtues of compassion, reasonableness, and integrity lead to ceteris paribus gains in personal life satisfaction, how large, for whom, and when. We already have the methodologies and research tools to do this. We just have to actually do it. And doing it would render moot much of moral philosophical pondering about such things.
But that’s just one side of the equation. Pigliucci keeps telling Shermer this, but he keeps not getting it. A moral imperative consists of two components: an implied set of consequences to a behavior, and an implied desire to have or avoid those consequences. I explain that in my cited work. But what it comes down to here is that the first component (consequences to a behavior) is what Shermer keeps talking about, while the second component (what people actually want) is what Pigliucci is talking about. Both are scientifically discoverable facts (and both are real facts of the natural world, and not mere opinions or phantoms), but they are not discovered or researched in the same way. Yet a proper moral science has to investigate both. And obviously, the latter is where a real science of morality should start–since consequences only matter once you know which consequences people want to enjoy or avoid.
This leads to a complexity that both Shermer and Pigliucci appear to overlook. The question of importance to any true moral science is not to study what people happen to want, but what they would want if they were suitably informed. All moral imperatives assume this (as I have demonstrated in my cited work). You might want designer shoes, but when informed of the child slavery producing them, your desire for designer shoes might evaporate. Knowledge makes all the difference between whether a moral imperative is true or false. The material facts of the world have to be understood as they are, in order to understand the moral facts of the world as they are.
This is a lot harder for science to suss out. But it’s not beyond our abilities. Science is incredibly creative and innovative in finding methods and ways of getting at facts. You just have to get to work on it. The task before us is not only to do what Shermer is talking about (and certainly not in the pseudoscientific way Shermer is doing it, but in properly sound and critically robust ways, in line with all good science), but also to start studying what people really want most from their lives, and more particularly, what they want when significantly informed of their options and of the contents of the world and of the possible consequences of their actions and choices–especially those they might not presently be aware of.
And after doing a lot of that, science needs to survey its results and look for cross-cultural and inter-societal universals. If it finds any, a universal morality emerges (and hence an objective and real one)…along with a way to explain it and persuade people to it that no longer hinges on speculation or hypothesis or holy books or tradition or anything other than facts and reason. If, on the other hand, science finds there is no universal set of ultimate core values (from which ultimate moral facts will necessarily derive, in the way I explain in my cited work), it will then have found two or more population subgroups who share ultimate core values, and will then be able to discover prescriptive moralities that are real and hold true for those respective subgroups.
I doubt the latter will occur, for reasons I have explained in my cited work (I doubt humans differ enough from each other biologically or indeed even environmentally for different subgroups of us to have radically incommensurate ultimate core values…and yes, I even consider the phenomenon of psychopathy). But that’s at least a verifiable (or falsifiable) empirical claim. Which science could answer…if it could be persuaded to bother. It’s high time it was.