Neural Darwinism

Theories broadly based on Darwinian evolution apply at various levels in the brain. Long-term memories are copied to preserve them. Learned information is repeatedly rehearsed. Mistakes are repeatedly dwelled on. Promising ideas and plans are copied in short term memory as variations on them are generated and selectively maintained.

Also, various low level processes behave in a Darwinian fashion: most significantly, neurite tips are copied with selective retention, and nerve impulses themselves are copied as they travel down branching axons and up branching dendrites.

I think that it makes reasonable sense to refer to these low-level brain processes using the term 'Neural Darwinism'. There's a bit of a problem though - which is that the term 'neural Darwinism' has been widely used to refer to the particular theories put forth in the late Gerald Edelman's 1987 book titled neural Darwinism.

Gerald Edelman was a pioneer in applying Darwinism to the brain - and I don't want to diminish his contribution too much. Developmental selection is a reasonable idea. Experiential selection is a bit more of a dubiously-named idea. I read neural Darwinism in the 1990s and found it dry, tedious and unconvincing. However, retrospectively, Edelman's concept of neural reentry looks important. It could be the key to understanding how the brain does something functionally similar to back propagation without having a good quality bi-directional signal propagation mechanism.

Overall, though, I think things have moved on a bit since 1987. It is now clearer that there are at least three types of low-level signal copying in the brain: the conventional reproduction of cells - including neural stem cells, the splitting of axon and dentrite growth tips, and the splitting of signals travelling how axons and back-propagating up dendrites. Despite the progress, I think we can still use the term "Neural Darwinism". I would hate "Neural Darwinism" to become a reference to an out-dated and discredited theory: it deserves better than that.

Signals propagation in axons and dentrites has been effectively simulated by synthetic neural network enthusiasts. However few of them bother simulating the other main copying processes in the brain: cell splitting during development and neurite tip splitting. Maybe simulating these Darwinian brain processes would help to build better synthetic neural networks.

For references see my Keeping Darwin in mind essay.

My take on Susan Blackmore's 'big brain' hypothesis

My idea that memes led to big brains has been compared to Susan Blackmore's idea that memes led to big brains - as presented in The Meme Machine - and some earlier essays of hers. I got the idea that memes might have been responsible for the great human cranial expansion from reading Susan's book. However, it seems to me that my idea is a bit different to hers.

Susan emphasizes that true behavioural imitation is hard - so hard that only a very smart primate mastered it. It requires the complex skill of putting yourself in someone else's shoes - in order to repeat their actions. Interestingly, this is almost the same ability that leads to empathy. For Susan, the difficulty of imitation is significant. SHe writes:

The first three processes alone will produce the selection pressures required to drive a runaway increase in brain size - if one further small assumption is made. That is that being good at imitation requires a big brain.

By contrast, my idea is more that ancestral human crainia were completely filled with memes. Memes were on average beneficial - and the more of them you had space for the fitter you were.

In my book, I compare the human skull to ant domatia, figs and Alder tree root nodules - all structures produced by plants to house symbionts. The human skull houses cultural symbionts - and that is the main reason why it grew so dramatically over the last three million years.

For a while I didn't realise that I had mutated her idea. I thought I had got my idea directly from her. Later, when I reread The Meme Machine I noticed that there were some differences between her idea and mine. Some mutations had taken place in my brain.

I'm not saying that Susan's idea is wrong - but I still like my take on the idea more. One of the virtues of my idea is that it is pretty specific - which should help with testing it. The idea of the big brain as a meme nest does not require Susan's additional "small assumption" that "being good at imitation requires a big brain".

Since practically every difference between us and our nearest primate relatives is down to our cultural symbionts, it is no big surprise that our big brains are too. However - as far as I know, Susan Blackmore was the first person who really took this idea seriously and promoted it.

I find it distressing to see the extent to which her contribution has been written out of scientific history. We have many modern papers on the idea that the enlargement of the human brain was caused by cultural transmission [see references here]. Few or none even mention Susan Blackmore's contribution. It seems like a revisionist history.

Axonal and dendritic growth tip selection

I covered axonal and dendritic growth tip selection in my 2011 book on memetics. They are important forms of natural selection in the brain.

These days, animated scanned tissue images are available - which illustrate what the process looks like:

These images were originally from http://www.neuralimages.org/.

Keeping Darwin in mind

A critical comment on one of my posts raises the following issue:

In A Devil's Chaplain wrote that humans (and only humans) have the gift of foresight "something utterly foreign to the blundering short-term ways of natural selection."

Utterly foreign. In the BBC Profiles video while promoting the book he said "The human brain is the only possible engine of departure from Darwinian principles, and it really is." Dawkins and Dennett (in lecture and interview) both acknowledge that humans "are the only/first intelligent designers on the tree of life." As I have said previously the question is no longer design (which evolution is the answer to) but drive. What are the causal forces that move humans, and their thought around?

The thing is, the human brain itself works on Darwinian principles. Its components use copying ubiquitously (signals are copied whenever an axon branches), and there's variation of what is copied plus selection. Ideas evolve within minds during individual learning, just as they evolve between minds during social learning. Copying, variation and selection are the basic ingredients of Darwinian evolution, according to many formulations of it - so the brain evolves along Darwinian lines.

Within-brain Darwinism was worked out by Campbell, Skinner, Changeux, Edelman, Szathmáry and Calvin - and has been popularized by Cziko, Plotkin, Dennett and myself, among others. In short, the brain is a Darwin machine.

B. F. Skinner was one of those who pioneered this view, making the analogy between Darwinian evolution and learning explicit. He used the term "extinction" to refer to the obliteration of memories - usage that persists in psychology to this day.

This within-brain evolution is a fundamental part of universal Darwinism.

No doubt the remaining critics will say that within-brain selection is deliberative selection. However, we have deliberative selection in common-or-garden evolution as well - it's called mate choice, or sexual selection. No doubt critics will say that within-brain variation is composed of directed mutations. However both evolutionary theory and Darwinism are cool with directed mutations. The various definition(s) of "evolution" typically make no mention of the source of variation, they just say that it needs to exist. As for Darwin, he understood relatively little about how variation arose. He had some theories, but largely knew that he didn't know. Directed mutations might well be contrary to neo-Darwinism, but that's a very different ball-game.

Many critics don't have a proper appreciation of the issue of Darwinian evolution within the brain. They think that intelligent design represents a radical departure from evolution. What they don't grasp is that the brain evolves at high speed, and its productions are the result of many generations of reproduction, variation and selection of signals and ideas. Brains are a neat invention, but they still work on evolutionary principles.

Also, for many years, most of the researchers in cultural evolution didn't understand within-brain evolution either. The individual learning folk and the social learning folk came from different disciplines - and it took a while for them to find each other, swap notes and get onto the same page.

Here's an example of this misunderstanding in action among researchers generally sympathetic towards an expanded domain for Darwinism:

A number of other nonselective processes may affect the evolution of ideas. For example, people can learn an idea from others and then innovate, modifying the idea in an effort to improve it. Still other nonselective processes can arise when people synthesize their own beliefs after being exposed to a number of people who behave differently.

Here learning and critical evaluation are described as being "nonselective processes" - when it is now evident that any proper understanding of them would involve considerable quantity of selection between ideas inside minds.

Today, understanding of within-brain evolution lags behind understanding of cultural evolution - which itself remains poorly understood. Universal Darwinism is an even-less-well-understood fourth stage in the evolution revolution.

As for the idea that "humans (and only humans) have the gift of foresight", that is just daft. All animal brains forecast the future consequences of their actions, in order to allow them to choose between them. That's the basis of the memory-prediction framework.

Lastly, it is Darwinian evolution within the brain that represents the primary conceptual link between evolutionary theory and machine intelligence.

References

• Calvin, William H. (1987) The brain as a Darwin Machine.
• Plotkin, Henry (1994) Darwin Machines and the Nature of Knowledge.
• Skinner, B. F. (1953) Science and Human Behavior.
• Calvin, William H. (1998) The Cerebral Code: Thinking a Thought in the Mosaics of the Mind.
• Edelman, Gerard (1987) Neural Darwinism: The Theory Of Neuronal Group Selection.
• Changeux, Jean-Pierre (1983) L'Homme Neuronal.
• Harold P. de Vladar, Eörs Szathmáry (2015) Neuronal boost to evolutionary dynamics
• Fernando C, Szathmáry E, Husbands P. (2012) Selectionist and evolutionary approaches to brain function: a critical appraisal.
• Fernando C, Vasas V, Szathmáry E, Husbands P. (2011) Evolvable neuronal paths: a novel basis for information and search in the brain. PLoS ONE 6, e23534.
• Fernando, Chrisantha, Karishma, K.K. and Szathmáry, Eörs (2008) Copying and Evolution of Neuronal Topology. PLoS ONE 3 (11): 3775.
• Fernando, Chrisantha and Szathmáry, Eörs (2009) Chemical, Neuronal, and Linguistic Replicators.
• Fernando C, Szathmáry E. (2010) Natural selection in the brain. In Toward a theory of thinking (eds B Glatzeder, V Goel, A von Müller), pp. 291–340. Berlin, Germany: Springer.
• Fernando, Chrisantha, Goldstein, R. and Szathmáry, Eörs (2010) The Neuronal Replicator Hypothesis. Neural Computation 22 (11): 2809–2857.
• Fernando, Chrisantha (2010) Neuronal replicators solve the stability-plasticity dilemma. GECCO 2010: 153-154.
• Fernando, Chrisantha, Vasas, Vera, Szathmáry, Eörs and Husbands, Philip (2011) Natural Selection of Paths in Networks.

Memes in the driving seat

It is widely believed that cultural evolution goes much faster than the evolution of human DNA can manage.

In my book on memetics, I discuss the "upright gait hypothesis" and the hypothesis that much speciation in the hominid lineage was assisted by memes.

Looking at the changes memes have produced in us, I think it is reasonable to propose the following bold hypothesis:

Practically all the significant differences between chimpanzees and modern humans are the consequences of human cumulative cultural evolution.

We can see the normal rate of morphological and behavioural change in primate lineages - by looking at our nearest relatives. Human evolution has been like a rocket by comparison - and memes explain why.

Much modern evolution is cultural evolution - memes explain all the interesting evolutionary change, while genes plod along at a glacial rate which is hardly noticable.

Coevolution between memes and genes mostly takes the form of memes dragging genes around in the adaptive landscape. The memes lead, while the genes follow. Genetic evolution is thus the delayed consequence of cultural evolution. The practice of drinking milk led to lactase genes being active an adults; the practice of talking led to larynx changes, and the practice of walking upright led to modified knees and ankles - and so on.

This idea also explains why our brains swelled up and why humans are ultrasocial.

It is an old idea in memetics. For example, Susan Blackmore (1999 p.80) raps on the Lumsden-Wilson "leash" metaphor - of memes being held on a leash by genes - saying:

In this way the memes are, as it were, dragging the genes along. The leash has been reversed and, to mix metaphors, the dog is in the driving seat.

The hypothesis here is the polar opposite of the position of Coyne 1999 who wrote:

Similarly, the self replication of memes does not mould our biology and culture; rather, our biology and culture determine which memes are created and spread.

Of course memes and genes coevolve, but the point is that the memes lead, and the genes are dragged along in their wake. This is an example of large organisms using small symbiotes to adapt quickly.

Since this "memes-lead" hypothesis explains so much of human evolution so well, I think the challenge is to look for puzzle pieces which it doesn't explain. I've looked, and haven't found very much - thus the bold hypothesis above.

Molecular copying makes memories last

Interesting news reports on new findings on the topic of the basis of long term memory. It finds that memories are stored in synapses by a self-copying prion-like protein.

• Prion leaves lasting mark on memory
• Making memories last: Prion-like protein plays key role in storing long-term memories
• How brain keeps memories alive for decades

To quote from the second article:

A portion of the structure is self-complementary and—much like empty egg cartons—can easily stack up with other copies of itself. CPEB thus exists in neurons partly in the form of oligomers, which increase in number when neuronal synapses strengthen. These oligomers have a hardy resistance to ordinary solvents, and within neurons may be much more stable than single-copy "monomers" of CPEB. They also seem to actively sustain their population by serving as templates for the formation of new oligomers from free monomers in the vicinity.

Finding a form of molecular copying to be implicated in the basis of long-term memory is no surprise - but it is an interesting and important finding, and it looks as though it may help to illuminate the nature of some of the low-level Darwinian processes that go on in the brain.

Previous findings in the area were reported here: Prion leaves lasting mark on memory