Survival of the selfless

"The consequences of regarding evolution as a multilevel process, with higher-level selection often overriding lower-level selection, are profound." This under-statement is in the latest issue of New Scientist, in a must-read piece titled "Survival of the selfless", by sociobiologists David Sloan Wilson and Edward O. Wilson. (New Scientist , 3 Nov 2007).

Indeed. Since I've been presenting the case for multi-level co-evolution in my weblogs for the last 2 years, I am ecstatic to see some big names in the field take the same position.

This furor is about whether it is "genes" that evolve, or "individuals" or groups of individuals such as tribes or species. Views were and are still held by many otherwise rational scientists with religious fervor in the worst sense, and arouse equal vehemence when challenged, akin to that between creationists and evolution-supporters.

This matters because higher level groups may have a whole different "fitness" measure than individuals, and while individuals or genes might evolve faster by being "selfish", the whole society of individuals might evolve faster if everyone was cooperative and altruistic. This battle continues to rage today, and is a core issue in whether "competition" and "free markets" are a good idea or not. So it is tangled with social ramifications, just like all science ultimately is.

This is also a core question in whether a "Theory X" company, driven by internal competition between managers, can ultimately be out-performed by a "Theory Y" company, like Toyota, driven by massive internal cooperation. A lot of egos are at risk of being bruised. A lot of justification for public policy is at risk of being overturned. It's a big deal.

Well, which is it? Do individuals evolve by being better at beating each other, or do groups of individuals dominate by being better at collaboration?

Peeking ahead, of course, I usually argue that "or" is a bad concept, once feedback is involved, and the right solution to look at usually involves "and" and "all of the above, simultaneously, interacting." But, "all of the above, interacting with feedback" was way beyond anyone's ability to compute or analyze, and not an attractive model for most researchers or grant writers.

Well, back to this article. In the face of enormous opposition, and tacking a consensus in the field that group-level evolution is a dead concept, they really settle for the weak claim that "we cannot rule out group-level selection."

Hmm. What's this all about? The concept is fascinating, and the sociology of science is equally fascinating here. The Wilsons ask "Why was group selection rejected so decisively [ in the 1960's ] ?" What a great question in how Science works!

Now, I should note that I'm one of the casualties of what seems a similar disastrous and mistaken turn of a field, namely Artificial Intelligence ("AI"). I got hooked by a course at
Cornell in 1965, taught by Professor Frank Rosenblatt, titled "Learning and Self Reproducing Machines".He and his lab had developed a "perceptron", a maze of switches and wires that connected up to a 20x20 grid of 400 photocells, on which letters of the alphabet were projected.

The perceptron, a model of human vision and learning, was slowly learning to tell the letters apart and identify them. At the time, this was astounding, and many scientists confidently argued that this could never be done. Later, of course, Kurzweil and others carried this technology forward and made OCR text-scanners that are now about 99.5% accurate or better and can read license-plates at an angle from a speeding car. But, in 1965, telling "A" from "B" was a big deal, especially if the "A" wasn't always exactly straight up and down, or in the same place on the grid.

The perceptron's insides were a network of wires and "nodes", a model of our brain's neurons, where the total strength of signal coming into each node was added up, multiplied by some factor, and either triggered or didn't trigger an outgoing signal to the next layer. The system learned by changing these multiplicative factors, searching for some set of them that would ultimately trigger the highest level "A" node when an A was projected on its primitive retina, and trigger at "B" when a B was projected, etc.

Then, the field was devastated by a very authoritative and persuasive paper, ultimately retracted, by highly regarded MIT professor Marvin Minsky that this approach "could never work." Funding dried up, and researchers moved on to other projects. Labs closed.

It took over a decade until somone finally figured out that Minsky had simply proven that a two-level neural net had irrecoverable gaps in its logic, and was not "complete". What he failed to look at, or see, was that these gaps went away when you got to three-levels or more.

Wikipedia has this quote:

Its proof that perceptrons can not solve even some simple problems such as XOR caused the virtual disappearance of artificial neural networks from academic research during the 1970s, until researchers could prove that more complex networks are capable of solving these and all functions.
(source: Hassoun, Mohamad H., Fundamentals of Artificial Neural Networks, The MIT Press, 1995. pp. 35-56.)

Oopsie.

Anyway, it appears to this observer that a similar phenomenon has occurred in socio-biology. Some very persuasive people published papers "debunking" multi-level evolution, well before there was enough computer power to actually simulate it and see what happened. ( In 1978, the mainframe computer I was programming had 4,000 bytes of memory to work with. Not 4 Gig, or 4 Meg, but 4 thousand. Any cell phone today has more than that.)

The social climate at the time made this debunking seem a better idea. World Communism was the mortal enemy of all that was good and holy, threatening "our way or life", justifying huge military expenditures, and anything that suggested communal good or community was more important than individuals was instantly suspect and risked being dragged before the House UnAmerican Activities Committee, where the proponent had to renounce their views or be thrown out of their jobs or locked up as being "unAmerican." Everyone was building bomb-shelters for protection against that day's terror threat.

In addition, religions had held for thousands of years that there was really nothing of importance between man and God, and man was God's noblest creation, so the idea that something larger than humans but smaller than God mattered was suspect dogma. (These days, the evolution of the earth and global warming is in fact dominated by such a larger life-form, "corporations", which have more or less hijacked the role individuals used to play in influencing governmental decisions and policies. But that observation lives in world "A", and discussions of evolution live in world "B", and the two don't talk to each other or trade notes.)

Then, of course, some people didn't like the idea of evolution in any form, and rejected it and most of biology and science based on that view.

So, for many reasons, some good, some not so good, the idea of group evolution as a dominant or even important force was denounced, rejected with emotion, and painted as an example of wrong thinking to be avoided at all costs.

Now, by what the Wilson's say, the whole question is being raised again, this time in a climate with much more powerful computers, where cooperation and collaboration in corporations are not always dirty words, and where the old theories, frankly, didn't explain why there was just so much altruism and goodness in people.

As I say, I'm delighted.

Also, finally, as I've posted on before, finally "feedback" and dynamics are starting to be considered in models, and finally multi-level causality keeps on increasingly showing up in how individuals behave, to the point where the National Institutes of Health and the Institute of Medicine talk about the necessity of using multi-level models to understand social interactions and how the things we see around us, like poverty, are held in place by many subtle but very powerful forces at different levels.

Fascinatingly, this gets us back to what Charles Darwin himself said in The Descent of Man, published in 1871, and the lead sentence in the Wilson's article:

Although a high standard of morality gives but a slight or no advantage to each individual man and his children over the other men of the same tribe ... an advancement in the standard of morality will certainly give an immense advantage to one tribe over another.

The Wilsons say that group evolution, versus individual evolution, doesn't yet explain the observed rise in altruism, although they can computer a visible imact.

Immodestly, I'll suggest that they need, like Minsky, to look at three levels, not two, to see the effect start to dominate.

In fact, all around us, we see corporations trying to survive and be more fit than competition, by pouring resources into internal cooperation and collaboration. So while individuals may continue to follow "greedy algorithms" and seek their own advancement, the corporation is making the playing field non-level and rewarding collaboration as the method of getting ahead personally. In that sense, Corporate policy is serving one role of religion - seeing the larger picture, thinking globally, and then trying to shift the local context so that relatively less visionary individuals, acting locally, will do the right thing if they just follow the rules.

This "think globally, act locally" function is the key role that "unity" has to handle, and it works best if people stop fighting the behaviors and yield and embrace the behaviors instead.

People have to let go of their own ego, "die to themselves" as it were, only to be "reborn" where their ego now includes the other people in the larger village or familiy or corporation they now have committed to belong to. In some sense, they now are just selfish at a whole larger level, as now those villages or corporations or religions or nations start competing, and the whole cycle begins again at a higher level, as they too have to learn that collaboration beats competition hands down in the long run, even if it doesn't seem obvious locally.

This phase transition is one we should be looking for and supporting. It's built so deeply into the fabric of space and time and control loops that it is inevitable and always working in the background, at ever higher levels, simultaneously. At least, that's how I see it.

Of course, that would imply that it won't be long before earth discovers we're just one inhabited planet of millions such planets, and we have to deal with the whole unity/diversity and competition / collaboration thing all over again at an even larger scale and scope.

Which is a model that some people don't like, so this can get emotional again. Still, I think we need to get used to the idea that we are not on top of God's creation, just below Gods ourselves, but maybe quite a few levels lower than that. Earth is not in the center of our galaxy, nor in the center of the universe.

It's a very scary concept to some people, if the world is seen as a place where competition and dog-eat-dog dominates. That belief leads to an imperative to dominate the world, before someone else dominates you. On the other hand, if the world is a place where cooperation and collaboration dominate, then it is a far less scary place, and we should "get with the program."

Already our corporations, internally, are undergoing this transformation. Kicking and screaming, often, but they cannot deny that the Toyota model outshines the GM model.

We need to speed the transition on the level of nations and religions as well, and find that sweet spot where cooperation and collaboration work so much better than competition for dominance and attempts at mutual destruction.

All of those struggles are tied up in this question of the way nature, life, and/or God operate here and what the design principles are that we can rely on to work. The cells in our bodies don't triumph in "beating" each other, but in collaborating with each other. It's a good model, and it's been field tested, and it works.

We should stop fighting it and use it.

Wade


The New Scientist article says of itself:

This is an edited, abridged version of a review in the December issue of The Quarterly Review of Biology. Further reading: D. S. Wilson's book Evolution for Everyone: How Darwin's theory can change the way we think about our lives describes multilevel selection theory for a broad audience. D.O. Wilson and B. Holldobler's forthcoming book The Superorganism analyzes how insect colonies can be seen as products of colony-level selection.