There has been a bit of confusion at Evolang this afternoon. Originally the first plenary session was to be addressed by Eőrs Szathmáry from the Collegium Budapst but a family emergency forced a change in plans. His presentation was given by his colleague Chrisantha Fernando, but was delayed until this afternoon and he spoke before a workshop rather than a plenary session. Nevertheless the presentation made a contribution to the continuing theme of how did we evolve a brain that can understand language. The first two presentations favored a genetically-based language faculty. The Szathmáry/Fernando claim looks in another direction.
Fernando proposed that replicators exist in the brain, undergo natural selection, and change the way the brain functions. Instead of depending on the genome, the brain adapts to experience.
Replicators are substances that reproduce themselves. Genes are the best known example. To date, the nature of such neural replicators is purely hypothetical and the first task is to identify the neuronal units of selection, and how the selection could occur. If such replicators could be found and understood, they would probably offer valuable insights into the origin of language and human creativity.
The appeal of such a theory is that it might explain how culture and language can be passed along through the generations and used appropriately by members of a generation. Using the genetics analogy, sentences would be the equivalent of phenotypes (individual organisms) and the neural replicators would be the equivalent of genotypes. Just as genotypes produce phenotypes, so neural replicators would produce sentences. And just as natural selection produces environmentally appropriate organisms, it would also produce contextually appropriate sentences.
The difficulty of the theory is in identifying the neural replicators. In 1949 Donald Hebb described memories, learning, and knowledge in terms of neural assemblies. These are groups of neurons in the brain connected through synapses. Experience weakens or strengthens the synaptic connection. One neuron can be part of several assemblies. The assembly is the material basis of a "mental representation" in a thought.
The problem, from the Darwinian perspective, is in identifying replicators. The assemblies are groups of neurons and, in Hebb's presentation, do not reproduce themselves.
Szathmáry and Fernando trace their neural selectionist theory to Jean-Pierre Changeux who built on Hebb's work. In Changeux's idea, the "brain spontaneously generates crude, transient representations," or pre-representations, that arise from existing assemblies or neurons. These pre-representations are the functional equivalent of genetic mutations. They happen to exist and for the most part they have no adaptive value, so they fade away. But sometimes they match reality; e.g. a pre-representation of a giraffe happens to match with a giraffe that is in front of the observers. Those pre-representations are stored—i.e., they are selected—so that in the future giraffes can be recognized more quickly.
The audience was told that "Darwin helps to reconcile Fodor with Epicurus!" The Fodor in this case is Jerry Fodor who insists that you cannot learn anything without already having the idea in your head. It seems unlikely, however, that genes could have given our whole generation the inborn idea of cell phones, yet people around the world recognize them and use them. Pre-representations suggest a way out. They enable the brain to mutate cell-phone representations at any time in history, but only be selected now when a cell phone is in front of one.
Epicurus is more famous, one of the founders (along with Democritus) of atomic theory and the materialist school of philosophy. He also defended the idea of free will by saying the motions of the particles were not always pre-determined. Darwin's notion of selection is a more sophisticated version of unpredictability than Epicurus and his swerving atoms.
A difficulty of the pre-representation approach is that we might expect to hold a cell phone for a very long time before we got the right neuronal mutation that let us see what is in our hand. An analogous situation is in the body's immune system which through a process of selection can eventually produce plenty of useful cells to fight an infection, but that process is not instantaneous, too slow to solve the cell-phone-in-the-hand problem. Szathmáry and Fernando are looking for ways to speed up the pre-representation process because an organism cannot learn more than what is in its pre-representations.
It is an interesting idea and I confess to leaning more toward learning than the language faculty, but the proposal of neural replicators seems still in the earliest stages. It does not look like something anyone can yet lean on. So far here in Utrecht, even with their defensiveness, the pro-language-faculty side is kicking *ss.