Thursday, September 06, 2012

Has Civilization Made Moral Progress? Sketch of an Empirical Test

From a certain perspective, current liberal Western civilization seems to be a moral pinnacle. We have rejected slavery. We have substantially de-legitimized aggressive warfare. We have made huge progress in advancing the welfare of children. We have made huge progress toward gender and racial equality. In his 2011 book The Better Angels of Our Nature, Steven Pinker says he is prepared to call our recent ancestors "morally retarded" (p. 658). Imagine how we would react if a Westerner today were seriously to endorse a set of views that would not have been radical in 1800: denying women the vote (or maybe even advocating a return to monarchy), viewing slavery and twelve-hour days of child labor in coal mines as legitimate business enterprises, advocating military conquest for the sake of glory, etc. "Morally retarded" might seem a fair assessment!

From another perspective, though, it appears almost inevitable that the average person in any culture will see his or her own culture's values as morally superior. Suppose that the average person in Culture 1 endorses values A, B, C, and D, the average person in Culture 2 endorses values A, B, -C, and E, and the average person in Culture 3 endorses values A, -B, D, and F. The average person from Culture 1 will think, "Well the average person in my culture has A, B, C, and D right! The people in culture 2 have C wrong and really should pay attention to D rather than E, while the people in Culture 3 have B wrong and should attend to C rather than F. So my culture's values are superior." If Culture 1 is temporally more recent than Cultures 2 and 3, then our average person from Culture 1 can laud the change in values as "progress".

The problem is, of course, that it might not be progress at all, but rather only a preference for local values. Call this the Local Pinnacle Illusion. Someone from the past might suffer the same illusion looking forward at us, condemning (say) our liberal Western neglect of proper class and gender role distinctions, our relative irreligiosity, and our relative tolerance of homosexuality, masturbation, divorce, and moneylending -- calling the changes "decadence" rather than progress.

The question then is whether there's a good way to tell whether those of us with a Pinkeresque preference for contemporary liberal values are merely victims of the Local Pinnacle Illusion, or whether we really have made huge moral progress in the last few centuries or millennia. I've been thinking about whether there might be a way to explore this empirically, using the history of philosophy.

Here's my thought: The temporal picture of progress and the temporal picture of a random walk look very different. If, say, rational reflection over the very long haul tends to guide us ever closer to right moral principles, as Pinker thinks, then issue-by-issue we ought to see opinion changes over the very long haul that look like progressive trends of moral philosophical discovery. If, in contrast, all that's going on is the Local Pinnacle Illusion, trends should be relatively short-lived, due to local cultural pressures, and not consistently directional over the very long haul.

Suppose we chose twelve issues that have been broadly discussed since ancient times and that in at least some eras -- not necessarily our own -- are regarded as morally important issues. If we could chart philosophers' opinions on these issues quantitatively (e.g., from strong support for democracy to strong support for monarchy with moderate views in the middle, from strong support for gender-neutral role expectations to strong support for gender-specific role expectations, etc.), using -1 and +1 to mark the average position in the historically most extreme eras for each pole of each issue, then on a random-walk picture, we ought to see something like this pattern among those issues:

I have assumed here a random starting spread from -1 to 1 and a random fluctuation from -0.3 to +0.3 in each successive period; then I renormalized the maximum value of each run to 1 and the minimum to -1. The result is basically a noisy walk, with extreme values as likely in the middle as at the ends.

On the other hand, if there's moral progress through philosophical reflection, the chart ought to look more like this:
For this chart, I assumed a progress factor, negative or positive, of one-third of the spread of the random fluctuation (i.e., adding or subtracting 0.1 to each step, in addition to the -0.3 to +0.3 random fluctuation). For the "toward 0" positions, I assumed positive incremental pressure when the previous period value was (absolutely) negative and negative incremental pressure when the previous period was positive. As with the previous graph, at the end I normalized the maximum and minimum values to +1 and -1.

As you can see, many more moral issues show consistent trend directions over time and the most extreme positions tend to be held at the beginning and the end of the analysis period. The latter pattern follows from the fact that even if the initial historical position at time zero is by some objective standard moderate, as long as it is not spot on the target number, the trend over time should be roughly unidirectional toward the target number; and then the normalizing will make that initial position look extreme.

Two caveats and a final note:

Caveat 1: It will of course be difficult to code this objectively, and it's quite possible that the final outcome will vary depending on hard-to-justify coding decisions.

Caveat 2: My guess is that if we were to chart moral and political positions from the 1600s to the present we would see a chart somewhat like the second (progressive) chart, whereas if we were to chart equal time intervals back to the ancient West, we would see a more random-looking chart like the first. So which time period should be examined? A case for choosing the shorter period might be that it is only after the printing press and widespread communication of philosophical ideas that we should expect to see rationally-driven moral progress. However a consideration against choosing the shorter period is this: There might be cultural factors, such as industrialization and capitalization, that have created consistent unidirectional pressures on moral and political norms, independently of the rational case for adopting those norms; and for that reason the broader the temporal span the better.

Final note: We might be able to construct similar charts to evaluate whether there has been progress in metaphysics.

(By the way, see here for another post of mine on Pinker's intellectualist liberalism.)


Ben said...

Is this an empirical test for progress, or for unidirectional change in moral opinion? Suppose we see something like the second chart. That chart doesn't tell between progress and regress, does it?

Eric Schwitzgebel said...

Ben: Yes, that's right. I would regard a Chart 2 pattern as at best presenting evidence for a *necessary* condition of progress, not conclusive evidence of it. However, among antecedent hypotheses to test, I'm inclined to think that "random walk" and "rational progress" are more attractive that "systematic decline" -- I suppose partly in light of my own attraction to the types of Pinkeresque examples of seeming moral progress.

monesy said...

Intriguing, but presupposing that long term change correlates to moral progress in order to conclude that what we are seeing is moral progress (rather than the LPI) seems a little circular. A change in norm, followed by the entrenchment of this new norm, really says nothing about the norm even being moral, let alone the change being morally progressive.

Eric Schwitzgebel said...

Thanks for the comment, monesy. My thought is that *if* there is rational pressure toward one moral-political result, that should be reflected in long term shifts in opinion that reflect that pressure. If that prediction bears out, that creates some support for the hypothesis. There might be other explanations for long-term directional trends in moral-political opinion, too, though. Offer me a plausible one and we can think about whether there would be a way to separate the hypotheses empirically.

On your last point: Of course! Hence the "final note" in my post.

Daniel said...

One possible explanation for a consistent, unidirectional change in moral-political opinion is that moral-political opinion is tracking some other variable. For example, if it has been the case in the last 500 years that there has been a consistent trend towards urbanization, free and accessible markets, and rapid communication, then it might be the case that moral opinion over that time has shifted to best accommodate that new reality.

Put another way, it seems to me that economic determinism - the claim that moral-political beliefs are superstructure, reflecting nothing more than changes in the economic framework - presents a compelling alternative to the explanation you present.

Of course, many would argue that we have made economic progress. But that does not necessarily mean that the accompanying moral change is progress.

(Unless economic progress really is the chief good, in which case the fact that our moral systems are changing to accommodate that progress is probably also praiseworthy - by extension)

monesy said...

Even if your conditional assumption is correct, other amoral, or even potentially immoral, factors could easily create long term shifts as well. A purely political force could do this. A religious force--the same. A long-term repressive totalitarism could easily result in a shift in paradigm that could be otherwise indistinguishable from a different shift in paradigm that can be argued to be morally progressive. A single example could be women's rights, in Iran, beginning from the 1960s and looking over the next 200 years. We could easily observe a continuation of the existing trend, yielding a long-term directional shift, and incorrectly judge, based on trend alone, that repressing women is a morally progressive action. Of course, by examining the paradigm, rather than the trend of the paradigm setting-in, we'd see that all we really have on our hands is a preference for a local tradition (a morally repugnant one, I'd forward) that has become ingrained, and therefore, 'long-term'.

And this, I feel, is the crux of the issue. One must inevitably judge the paradigm itself and decide if it can be regarded as moral. Looking at the trend alone isn't reliable if a single counterexample of something morally regressive could conceivably result in the same sort of trend. But if we are having to place our own value judgements on these norms, by looking at the norms themselves, then we are really right back to square-one: having to fight against relativistic biases and LPIs.

Eric Schwitzgebel said...

Daniel & monesy: I agree with your concerns. I feel like I raised the same concern myself when I mentioned the question of over what duration to examine the trends. The most obvious types of factors seem to operate on the scale of centuries rather than over thousands of years, though we could probably think of some on that larger scale too. But there might be other empirical ways to evaluate the plausibility of those other hypotheses. And any alternative trend will probably generate its own risky empirical predictions. What I would encourage, though, is that we consider empirical ways to examine the pressures as a check against the Local Pinnacle Illusion that is likely to arise from a priori normative reflection.