The comparison between what we presently know about human genomes and the brain phenotype raises the paradox of a non-linear evolution between the complexity of the genome and that of the brain (Changeux 1985, 2012b). From a molecular neurobiologist’s perspective, the cognitive abilities and skills required for the highest functions of the human brain are built from a cascade of events driven by a “genetic envelope”, which makes the difference between Homo sapiens and the human family’s earliest ancestors, but which cannot be simply related to genome size, nor to the number of genes.
The total amount of DNA housed in the haploid genome is approximately 3.1 billion base pairs, but no more than 20,000–25,000 gene sequences (1.2% of our genome code for exons—the DNA components of genes), and this number does not significantly differ from mouse to human. Moreover, the difference in full DNA sequences are very limited: between humans and chimpanzees they comprise no more than 4% of the genome. However, the total number of neurons in the human brain is in the order of 85 billion, compared to about 70 million in the brain of the mouse (Azevedo et al. 2009). Yet, notwithstanding the increase in cell numbers, with each neuron possessing its particular connectivity and its set of genes expressed, mammalian brain anatomy has evolved dramatically from a poorly corticalized lissencephalic brain with about 10–20 identified cortical areas to a brain with a very high relative cortical surface, multiple gyri and sulci, and possibly as many as 100 identified cortical areas (Mountcastle 1998). Thus, there exists a remarkable nonlinear relationship between the evolution of brain anatomy and the evolution of the genome organisation.
Molecular and cellular explanations have been suggested to account for this nonlinear relationship. One is the combinatorial expression of spatio-temporal patterns of genes that affect development (Changeux 1985; Edelman 1987; Tsigelny et al. 2013). Another, non-exclusive explanation, is the contribution of “epigenetic mechanisms” driven by interaction with the environment in the course of the long postnatal period of brain maturation—circa 15 years in humans—during which critical and reciprocal relationships take place between the brain and its physical, social, and cultural environment. It is on these epigenetic mechanisms that I shall focus here.
The word “epigenesis” can be traced back to William Harvey (1651), who stated in contrast to contemporary preformationist views that the embryo arises by “the addition of parts budding out from one another”. It was subsequently used by Conrad Waddington (1942) to specify the relationship between the genes and their environment to produce a phenotype. This is also the meaning adopted in the theory of the epigenesis of neuronal networks by selective stabilization of synapses, according to which the environment affects the organisation of connections in an evolving neuronal network through the stabilization or elimination (pruning) of labile synapses, under the control of the state of activity of the network (Changeux et al. 1973). This meaning, which I shall use henceforth, contrasts with the more recent and biochemically distinct meaning of the word epigenetic, which refers to the status of DNA methylation and histone modification in a particular genomic region. This concerns the neuronal nucleus, but not the diversity of individual synaptic contacts (Sassone-Corsi & Christen 2012). The modulatory role of chromatin modifications in long-term memory has already been described (see e.g., Levenson & Sweatt 2005), but the informational content involved—which relies upon cell bodies—is expected to be in orders of magnitude smaller that of synaptic epigenesis, based upon the combinatorial power of individual synapses.
During embryonic and postnatal development, the million billion (1015) synapses that form the human brain network do not assemble like the parts of a computer, that is, according to a plan that precisely defines the disposition of all the individual components. If this were the case, the slightest error in the instructions for carrying out this program could have catastrophic consequences. On the contrary, the mechanism appears to rely on the progressive setting of robust interneuronal connections through trial-and-error mechanisms that formally resemble an evolutionary process by variation selection (Changeux et al. 1973; Changeux & Danchin 1976; Edelman 1987; Changeux 2012a). At sensitive periods of brain development, the phenotypic variability of nerve cell distribution and position, as well as the exuberant spreading and the multiple figures of transiently-formed connections originating from the erratic wandering of growth cone behaviour, introduce a maximal diversity of synaptic connections. This variability is then reduced by the selective stabilization of some of the labile contacts and the elimination (or retraction) of others. The crucial hypothesis of the model is that the evolution of the connective state of each synaptic contact is governed globally, and within a given time window, by the overall “message” of signals experienced by the cell on which it terminates (Changeux et al. 1973).
One consequence of this is that particular electrical and chemical spatiotemporal patterns of activity in developing neuronal networks are liable to be inscribed under the form of defined and stable topologies of connections within the frame of the genetic envelope. In humans, about half of all adult connections are formed after birth at a very fast rate. The nesting of these multiple traces directly contributes to forming and shaping the micro- and macroscopic architecture of the wiring network of the adult human brain, thus bringing an additional explanation to the above-mentioned non-linearity paradox.
Another consequence of the synapse-selection model (originally presented as a “theorem of variability”) is that the selection of networks with different connective topologies can lead to the same input-output behavioural relationship (Changeux et al. 1973). This accounts for an important feature of the human brain: the constancy or “invariance” of defined states of behaviour, despite the epigenetic “variability” between individual brains’ connectivity.
Finally, both the spontaneous and the evoked activity may contribute to synapse selection. In this framework, a suggestion has been made that reward signals received from the environment may control the developmental evolution of connectivity (Gisiger et al. 2005; Gisiger & Kerszberg 2006). In other words, reinforcement learning would modulate the epigenesis of the network. The model has been implemented in a case of the learning of a visual delayed-matching-to-sample task (see below). This process of synaptic selection by reward signals may concern the evolution of brain connectivity in single individuals, but it also concerns the exchange of information and shared emotions or rewards between individuals in the social group (Changeux 1985, 2004; Gisiger et al. 2005). This is an important part of our argument; it may thus play a critical role in social and cultural evolution.
There is an abundance of experimental studies that are consistent with, or directly support, the model of synapse selection. In humans the maximum synaptic density is reached within three years, then steadily declines until the total number stabilises around the time of puberty (Huttenlocher et al. 1997; Bourgeois 1997; Petanjek et al. 2011). Yet the process of synaptic refinement goes far beyond puberty: learning is life-long (Petanjek et al. 2011). The observed global decline in synaptic numbers during childhood plausibly reflects a rich cascade of elementary steps of learning by selection. Numerous studies have shown that when neuronal activity is experimentally modified, synaptic elimination is altered (Benoit & Changeux 1975, 1978; Stretavan et al. 1988; Purves & Lichtman 1980; Luo & O'Leary 2005; Innocenti & Price 2005; Collin & van den Heuvel 2013). At variance with the classical Lamarckist-constructivist scheme (Quartz & Sejnowski 1997), blocking the activity maintains a high number of connections: it is activity that enhances synaptic elimination (Benoit & Changeux 1975, 1978; Stretavan et al. 1988; Luo & O'Leary 2005). Thus “to learn is to eliminate” (Changeux 1985).
Among the cortical connections established in post-natal life are the long-range tracts between the frontal areas (Miller & Cohen 2001; Fuster 2008) and other brain cortical areas (including sensory ones) (Goldman-Rakic 1987; Goldman-Rakic 1999; Hagmann et al. 2008; Collin & van den Heuvel 2013). Some years ago, it was suggested, according to the “global neuronal workspace” hypothesis, that these long-range connections, by broadcasting signals to multiple brain areas, yield subjective “conscious” experience by allowing sensory inputs—seeing, hearing and so on—global access to many brain areas (Dehaene et al. 1998; Dehaene & Changeux 2011). The long-range connections would provide a structural basis for the global experience known as conscious access.
These long-range connections are particularly important in the case of the prefrontal areas which contribute to planning, decision-making, thought, and socialisation. The ontogeny and postnatal development of long-range connectivity expectedly reveal phases of exuberance and phases of selection and axonal pruning (Collin & van den Heuvel 2013). In human newborns evolution is slow, and it has been suggested that the phase of exuberant long axon removal is largely completed at the age of two years, accompanied by increasing information processing and cognitive development (Collin & van den Heuvel 2013). Evolution continues during adolescence until adulthood with decreasing segregation and increasing integration, mainly but not exclusively driven by modulation of connections strength (local synaptic elimination persists in the adult; Petanjek et al. 2011). It is expected to have major consequences on the laying down of cultural imprints including the “epigenetic rules” associated with socialisation.
The acquisition of reading and writing may be viewed as a typical example of epigenetic development of “cultural circuits”. Writing and reading are recent cultural inventions (about 5000 years old) that evolved into distinct sub-systems and put considerable demands on our cognitive system. Historically, the first evidence for specialized writing and reading circuits in the brain was the discovery by the French neurologist Dejerine (1895) of pure alexia, also known as alexia without agraphia. Individuals with pure alexia suffer from severe reading problems while other language-related skills such as naming, oral repetition, auditory comprehension or writing are typically intact. Alexia results from cerebral lesions in circumscribed brain regions including the angular and supramarginal gyri. New specialized sets of connections are present exclusively in individuals that have learned written language and have been selected and consolidated in the course of development at sensitive periods (4–6 years) as a consequence of an intensive period of education.
The human brain did not evolve to learn to read, but possesses enough epigenetic variability in the course of its development (and also—though to a lesser extent—in the adult) to incorporate a cultural invention of this kind. During the acquisition of reading and writing by Western subjects, representations for visual forms of words progressively settle into the occipito-temporal cortex, recruiting a subset of functionally-appropriate object recognition regions in the temporo-parietal junction (Dehaene et al. 2010). The group of illiterate individuals is consistently more right-lateralized than their literate controls (Petersson et al. 2007). Interestingly, alphabetic writing systems recruit circuits that differ in part from those mobilized by the Chinese ideographic systems. In French readers reading French, activations were enhanced in left-hemisphere visual area V1, with the strongest differences between French words and their controls found at the central and horizontal meridian representations. In contrast, Chinese readers reading Chinese showed enhanced activations in intermediate visual areas V3v/hV4, which was absent in French participants (Szwed et al. 2014). Also, the capacity to read sheet music is selectively altered in music-specific forms of alexia. Neuronal circuits specific to a given culture may thus become epigenetically established in the brains of social group members. Written language-learning is only one of the many cultural imprints acquired during the development of the human brain (Changeux 1985). For instance, cross-cultural differences between Asian and Western participants manifest themselves as differential increases of fMRI in the medial prefrontal cortex with reference to self-judgment (Zhu et al. 2007; Ray et al. 2010) and also to diverse brain recordings in mind reading (Kobayashi et al. 2007), holistic attention (Hedden et al. 2008), or facial photo recognition (Na & Kitayama 2011). The adult human brain thus builds up from a complex intertwining of cultural circuits progressively laid down during development within the framework of a human-specific genetic envelope.
There is no compelling evidence that culturally-acquired phenotypes will sooner or later be genetically transmitted. What the evidence does show is that they have to be learned by each generation, by children from adults, and epigenetically transmitted from generation to generation, beginning in the mother’s womb and up until the adulthood. Teaching reading and writing to circa five-year-old children requires elaborate pedagogic strategies, which in a general manner are absent in non-human primates (Premack 2007).
In short, cultural imprints have a physical reality in the human brain. Cultural imprints have also been demonstrated in non-human brains, e.g., by Peter Marler's work on birds’ song-learning (Marler 1970). Yet the importance of cultural imprints on behaviour are comparatively much more important in humans compared to non-humans, in particular due to the long postnatal period of brain maturation. They play a critical role in shaping the brain phenotype in relation with the social group, through oral and written language but also though diverse culture-specific habits, traditions, and symbolic systems, including the ethical and social norms embodied in the adult brain.
I shall now proceed to discuss issues raised by the combinatorial explosion of brain representations and the channelling of behaviour through epigenetic rules and top-down control of decision-making.
epigenetic rules =Df In neurobiological terms, these “rules” shall be viewed as acquired patterns of connections (scaffoldings), hypothetically stored in frontal cortex long-term memory. They frame the genesis of novel representations and regulate decision-making in a top-down manner.