Wednesday, March 01, 2017

Why Wide Reflective Equilibrium in Ethics Might Fail

"Reflective equilibrium" is sometimes treated as the method of ethics (Rawls 1971 is the classic source). In reflective equilibrium, one considers one's judgments, beliefs, or intuitions about particular individual cases (e.g., in such-and-such an emergency would it be bad to push someone in front of an oncoming trolley?). One then compares these judgments about cases with one's judgments about general principles (e.g., act to maximize total human happiness) and one's judgments about other related cases (e.g., in such-and-such a slightly different emergency, should one push the person?). Balance them all together, revising the general principles when that seems best in light of what you regard as secure judgments about the cases, and revising one's judgments about specific cases when that seems best in light of one's judgments about general principles and related cases. Repeat the process, tweaking your judgments about cases and principles until you reach an "equilibrium" in which your judgments about principles and a broad range of cases all fit together neatly. In "wide" equilibrium, you get to toss all other sources of knowledge into the process too -- scientific knowledge, reports of other people's judgments, knowledge about history, etc.

How could anything be more sensible than that?

I am inclined to agree that no approach is more sensible. It's the best way to do ethics. But, still, our best way of doing ethics might be irredeemably flawed.

The crucial problem is this: The process won't bust you out of a bad enough starting point if you're deeply enough committed to that starting point. And we might have bad starting points to which we are deeply committed.

Consider the Knobheads. This is a species of linguistic, rational, intelligent beings much like us, who live on a planet around a distant star. Babies are born without knobs on their foreheads, but knobs slowly grow starting at age five, and adults are very proud of their knobs. The knobs are simply bony structures, with no function other than what the Knobheads give them in virtue of their prizing of them. Sadly, 5% of children fail to start growing the knobs on their foreheads, despite being otherwise normal. Knobheads are so committed to the importance of the knobs, and the knobs are so central to their social structures, that they euthanize those children. Some Knobhead philosophers ask: Is it right to kill these knobless five-year-olds? They are aware of various ethical principles that suggest that they should not kill those children. And let's suppose that those ethical principles are in fact correct. The Knobheads should, really, ethically, let the knobless children live. But Knobheads are deeply enough committed to the judgment that killing those children is morally required that they are willing to tweak their judgments about general principles and other related cases. "It's just the essence of life as a Knobhead that one has a knob," some say. "It's too disruptive of our social practices to let them live. And if they live, they will consume resources and parental energy that could instead be given to proper Knobhead children." Etc.

Also consider the Hedons. The Hedons also are much like us and live on a far-away planet. When they think about "experience machine" cases or "hedonium" cases -- cases in which one sacrifices "higher goods" such as knowledge, social interaction, accomplishment, and art for the sake of maximizing pleasure -- they initially react somewhat like most Earthly humans do. That is, their first reaction is that it's better for people to live real, rich lives with risk and suffering than to zap their brains into a constant state of dumb orgasmic pleasure. But unlike most of us, the Hedons give up that judgment after engaging in reflective equilibrium. After considerable reflection, they are captured by the theoretical elegance of simple hedonistic act utilitarianism. As a society, they arrive at the consensus that the best ethical goal would be to destroy themselves as a species in order to transform their planet into a paradise of happy cows. Let's assume that, like the Knobheads, they are in fact ethically wrong to reach this conclusion. (Yes, I am assuming moral realism.)

It seems possible that wide reflective equilibrium, even ideally practiced, could fail the Knobheads and Hedons. All that needs to be the case is that they are too implacably committed to some judgments that really ought to change (the Knobheads) or that they are insufficiently committed to judgments that ought not to change (the Hedons). To succeed as a method, reflective equilibrium requires that our reflective starting points be approximately well-ordered in the sense that our stronger commitments are normally our better commitments. Otherwise, reflective tweaking might tend to move practitioners away fromrather than toward the moral truth.

Biological and cultural evolution, it seems, could easily give rise to groups of intelligent beings whose starting points are not well-ordered in that way and for whom, therefore, reflective equilibrium fails.

Of course, the crucial question is whether we are such beings. I worry that we might be.

**********************************

Related:

How Robots and Monsters Might Break Human Moral Systems (Feb 3, 2015)

How Weird Minds Might Destabilize Human Ethics (Aug 15, 2015)

[image source]

9 comments:

Justin W. said...

Brandt says something similar in A Theory of the Good and the Right. See Chapter 1. A sample quote about reflective equilibrium: "the theory claims that a more coherent system of beliefs is better justified than a less coherent one, but there is no reason to think this claim is true unless some of the beliefs are initially credible -- and not just initially believed -- for some reason other than their coherence... " (p.20)

David Duffy said...

Isn't this moral disagreement dressed up? Presumably human ethicists have achieve reflective equilibrium too eg Feldman (2002):
"I like Desert-Adjusted Intrinsic Attitudinal Hedonism. I think I know some of its implications, and I am happy to accept them. I think I am in a state approaching reflective equilibrium while believing it."

Hanno Sauer said...

I wonder whether this post is compatible with your earlier post on the essential goodness of human nature, where you sketched an argument to the effect that our firmest commitments are also likely our best. What do you think?

Eric Schwitzgebel said...

Thanks for the reference, Justin!

David: Yes, this is a version of the problem of disagreement.

Hanno: I am inclined to think that we do have a natural moral compass for the normal range of circumstances to which we are exposed in evolutionary history plus in development, plus some reasonable extensions of those. But I think that Knobheads and Hedons might not be so fortunate (though now, why should I think *we* are lucky?) and I also think that sufficiently weird circumstances will break our compass (e.g., monster cases). Not a totally satisfactory answer to your helpful question, I know!

Pilot Guy said...

Eric - this takes me back to my Engineering days when I had to solve a nonlinear partial differential equation for part of my thesis. This is a challenging task to model as you have to seed a model. If you seed it correctly with starting values, the model will converge on a solution - if you seed it incorrectly, it will diverge and fool you that there are no solutions.
This is why I like my Philo studies so much more!

chinaphil said...

It seems like this could be a problem of disagreement - which wouldn't mean that reflective equilibrium has failed, it just takes different people to different places. It could also be a problem of unfinished reflective equilibrium, where the full ramifications of equilibrium are in fact so difficult that it takes many generations of effort to work them out, and any single person or generation along the way is not yet at the equilibrium state. Again, this wouldn't mean that reflective equilibrium fails as a goal, it's just harder to reach than we've allowed for.
I wonder about another possibility: What if getting to reflective equilibrium involves cognitive action of a kind that we are just not capable of?
Three ways in which this might be true spring to mind. One is that the path to reflective equilibrium involves computation that we can't do. Perhaps full equilibrium would involve some measure of consequentialism, and we don't have the brainpower to work out the consequences for everyone in the world. This problem could be ameliorated by the increasing power of computers.
The second again involves consequentialism, but includes the consequences for future beings. We cannot predict accurately enough what the future will be, and even the fastest supercomputers won't be able to do any better, and this failure to be able to see the future could block us from reaching an effective reflective equilibrium (e.g. we might be unable to find the right equilibrium because we don't know how much weight to put on the well-being of future people, because we don't know how many there might be).
The third is that we might be neuro-psychologically incapable of arriving at a reflective equilibrium. For example, we might decide, on reflection, that all human lives are equal and all human suffering is worth the same. But suffering that happens to the people immediately in front of us tweaks our mirror neurons (I'm allowing pop neurobiology to go mad here; it's only supposed to be an example), and sends us emotively into a tailspin. All the equilibrium in the world can't persuade me that the person in pain in front of me is no more important than another I've never met on the other side of the planet. Wait... I think I've argued round to exactly the kind of example you meant with your hedons and knobheads (you know that sounds much, much ruder in British English, right?).
One last one: What if we were destined to get stuck in a sort of eternal cycle where one generation believes that human nature is bad, and so invests heavily in education; as a result, most people turn out good, and over time a more Mencian belief takes grip, leading unfortunately to reduced inputs into moral education, so a couple of generations later more people are bad, and the Xunzi viewpoint starts to dominate once again. In theory we could break out of the cycle by reading enough history, but good luck with that program!

Callan S. said...

To me it just raises the question about inert ethics - what does the ethics do to actually get it's 'correctness' enacted? Nothing, it seems - it seems a lot like the Aesop's fable where the Hen asks who wants to grind the grain to flour, who wants to bake the cake. No one. Who wants to eat the cake? Everyone. And here this 'correct' set of ethics, having done nothing, wants to judge? How did it earn that right?

Eric Schwitzgebel said...

Thanks for the continuing comments, folks!

Pilot Guy: Yes! Also related to path dependence. But I'm not sure philosophy is so different!

Chinaphil: Interesting possibilities there. I'm realist enough to resist the suggestion in your first sentence, if interpreted strongly. If reflective equilibrium takes us different places, some of those places might be morally objectionable places, despite the equilibrium. And yes, your third suggestion is pretty close to what I had in mind in the OP. I love your final suggestion -- what an odd outcome that would be!

Callan: Right. As you know, I've long been worried about the tenuous connection between moral judgment about moral behavior!

Sam Clark said...

Hi - I think this reveals something about reflective equilibrium that Rawls became clearer about over his career: that it's the right method for ethics if moral realism is false (as you say, you're assuming moral realism to get your worry). At least as used by Rawls, RE is what we end up with when we stop thinking that there's a right answer to our moral dilemmas out there waiting to be discovered, and retreat to a purely procedural notion of moral deliberation. So, if your Hedons get to the conclusion they do by some ideal (or ideal-enough) deliberative procedure, then that conclusion is right - or at least there's no other standard against which it could found wanting.