MLS Theory and Language

Language is not simply an evolutionary trait, it's a trait that enables evolution. It's the "turbo button"

· 7 min read
MLS Theory and Language

Survival is tough. There are many species that have thousands of babies only to see a few of them make it to adulthood. Intelligence is also be tough, because out of all of the species we know on this planet, we're the ones with intelligence. You don't see dolphins building hot-air balloons or Directed Energy Weapons (at least so far as we know).

Why is that?

It's a great question, and our best guess so far is that it has to have something to do with operating using various-sized groups, languages, and tools. There are a lot of animals that do some of these things by themselves. We're the only ones we know that seem to do a lot of all of them, and we do these things in ways that reinforce the others.

If we want to continue surviving and growing as a species, create organizations that adapt and learn, and finally achieve Artificial General Intelligence (or manage to avoid creating it), we have to know how these three things work together to create intelligence in our own species. We can only speculate, of course, but lately scholars in different fields have been developing theories that fit together nicely. So while we might not have an answer, we have a much more detailed and consistent narrative about intelligence than we've ever had before. Based on that narrative we can make discoveries that were previously unavailable to us.

Profound things tend to sound idiotic, and vice-versa

Why Scholars Need MLS

For a long time, scholars have been puzzled by the way some groups of creatures operate. In many cases, the individual behaves in ways that might prevent their own survival and propagation yet help the species overall. This has been called the "Altruistic Gene" by some. It's also been called the "Grandmother Hypothesis"  *  

Evolution, works in one way in an overall species or in a specific individual, but sometimes it works differently in groups of individuals. Not always, but sometimes. We don't know why some social groups show this derivative behavior and some don't. We don't even understand why certain behaviors appear and last and others don't. It all appears chaotic to us.

What is Multilevel Selection Theory

Multilevel Selection (MLS) Theory can create order out of this chaos. It was developed for the study of social behaviors in non-human species but it is equally relevant to the cultural design of human groups, including but not restricted to business corporations. It is based on the following principles, which are so elementary that they are unlikely to be wrong.

  • Evolution is based on relative fitness. It doesn’t matter how well one survives and reproduces in absolute terms; only in comparison to others in the vicinity. As the economist Robert Frank puts it in his book The Darwin Economy: Liberty, Competition, and the Common Good, life is graded on a curve.
  • The social behaviors that maximize relative fitness within a group tend to undermine the welfare of the group. This is the opposite of the metaphor of the invisible hand.
  • Social behaviors that are “for the good of the group” might be selectively disadvantageous within the group, but they can be highly advantageous in between-group competition.**
Given any problem and group, various individuals and strategies work with differing subgroups of the master group. The more dynamic and challenging the environment, the smaller the groups will be that meet the selection criteria. Multivariate changing environments could easily require shifting subgroups through time given the same initial population. The only way that some form of MLS isn't applicable is in a static, unchanging environment with a very small population (since group size itself can be an environmental stressor)

Inherent Problems With MLS Theory

The problem with any evolutionary biology or social-genetic look at humans is that implicit bias creeps into the discussion without anyone realizing it and it's a hell of a thing to try to get rid of. Reverse-engineering chaotic, complex systems is a fool's game, but we all love doing it anyway.

Evolution is about adaptation under selection pressure. Conditions change in an environment. Organisms that change alongside those external changes better than others and manage to reproduce continue to exist and evolve. Those that do not change, or that change in a way that prevents thriving or reproducing are not so lucky.

In simple ecosystems, we look at simple changes. Perhaps, say, the temperature changes in a pond. Those creatures that adapt and thrive with the new temperature thrive and reproduce, thereby evolving. In this case we clearly identify a simple adjective, temperature, and then apply the intricacies of evolution as it applies to that one variable.

Real ecosystems, however, are not so simple. This is why we're constantly having to guess and test which pressures caused which evolutionary changes. We get it wrong in many cases. The universe is under no obligation to be easy for us to understand (or perhaps even possible for us to understand). Evolution is an extremely simple thing that works sometimes in extremely complex ways. We are left to go behind it and do the best we can to tease out what happened.

Our problem with trying to work these things out, and it's a huge problem that reared its head almost immediately after evolution was discovered, is that we look at some species, some attribute, then take that attribute and claim that all of evolution for that species was such that this is the "right" attribute to have.

The naive view is that large numbers of people assemble into something resembling a larger person, as in this frontispiece for the book "Leviathan", by Thomas Hobbes. Our desire for simple systems and answers, combined with our assurance that we both understand ourselves and everything in the environment around us, destroys most analyses of evolution involving our own species before it even begins.

As a simple example, which genes make people smart? Let's find the gene that makes people smart and breed people to be smart! What could go wrong with that?

Well, there is no one gene to make people smart. There may not be even a set of genes. One possibility is that there are hundreds of genes  working together in various contexts, each of which involves a trade-off with other aspects of the organism, and all of them together only working in certain configurations in certain environments. Just because we have a simple word "smart" doesn't mean that there's a simple set of things or methods in the universe that maps to that word.

We assign simple adjectives, we assume simple causes, then we beat the theory and the data until it complies with our wishes. This is not science, it's dogma.

All of this means that at best we know that selection happens at multiple social levels inside any complex society. We have no idea what sorts of selection happens at what sorts of levels or even which adaptations might be better or worse over time to that species or society. We have an engine but we have no idea what kind of vehicle it goes in or where we're headed with it. As one wag put it about another science, if economics were actually a science, all economists would be billionaires.  The normal condition about learning about the universe around us is that we know little parts of the truth but are completely blind about the overall picture.

Predictions are tough, especially about the future.

Language is not simply an evolutionary trait, it's a trait that enables evolution. It's the "turbo button"

Why Intelligent Creatures Need Language

We do know, however, that intelligence in any kind of group or social context requires language, and we're learning a lot about language. For instance, some of the latest research suggests:

  • Language is needed for tool-making
  • Tool-making is needed for survival
  • The tools needed for future survival can never completely be known (and many times even the tools needed for current survival)

Inherent Problems with Language

The core problem with language is that it's imprecise by definition and constantly-shifting. If it wasn't, it wouldn't allow us to use it for the items above. This is another way of saying you can't explicitly program a computer to make something, instead you program it to use rigid techniques in a feedback loop to reach goals. Part of my desire may be to have the robot arm move 15mm to the left. I can certainly send that command to the robot. But the robot has to both actuate and measure. My simple command maps to a hidden and perhaps very complex internal process that the robot executes. It very well may be non-deterministic.

Language can easily have the appearance of being much more precise than it actually is. This has caused all sorts of confusion and arguments among academics, philosophers, scientists, legislatures, courts, and religious folks throughout our history.

How Language and Multilevel Selection Must Work Together

All is not lost, however. As an intelligent social species, we have to survive. Survival depends on different things at different group sizes. It also depends on adapting to sometimes rapidly-changing conditions that can't be predicted and will be resolved in ways we currently don't understand. However we resolve the problem and evolve, however, will involve the our use of language.

We may not know exactly what things are, and we may not know what things are going to be, but we are absolutely certain that conditions are always changing and that we must adapt as a species to those changing conditions in order to survive and continue evolving. Since before we came out of the savannah, we do that with our tools.

We still don't know which group sizes work best for various things or how they work, so whatever conclusions we make must by necessity be about individuals. Perhaps in the future there'll be more work here.

Given that, language and social structure must work together to provide these abilities to each person, all of which revolve around tools:

  • The ability to acquire and keep the most diverse range of types and quantities of materials that might be used in future tool creation
  • The ability to create and modify the most diverse types and the most quantity of tools to survive as changing conditions merit
  • The ability to teach and learn tool making in both the abstract, generalized, and instantiated tools from the widest and most diverse set of origins to the widest and diverse set of destinations

 *  There are many, many versions of this same theory. It's something that has perplexed scholars almost since the beginning of evolutionary theory and sociology.

 **  This was lifted almost verbatim from a recent article "The Business World Needs Multilevel Selection (MLS) Theory" Here's a great paper from the Evolutionary Institute that discusses all of this (and more) in depth.

Related Articles

The Canyon
· 4 min read
· 1 min read
Info-Ops 3: Introduction
· 13 min read
Outlines Of A Supercompiler In F#
· 10 min read