For those of you who cannot make my talk on language evolution today at UCL. http://www.ucl.ac.uk/anthropology/seminars/ To read the forthcoming paper see below: http://sociogenomics.wordpress.com/?attachment_id=69

1. The parental antagonism theory of language evolutionPreliminary evidence for the proposal William M Brown BA MSc PhD Human Evolutionary Behavioural Science Lab w.m.brown@uel.ac.uk

2. Overview General perspective on social evolution Genomic imprinting and conflict Parental antagonism theory of language Empirical evidence Summary of findings

3. ‘Out-dated’ ranked view Physics Chemistry Biochemistry Biology Neurosciences Social Sciences Humanities

4. A revised view So-called higher or more complex levels of social behaviour and organisation can tell us something about lower levels of biological organisation (e.g., genome evolution). There are social principles of evolution.

5. Royal chambers of termites

6. Division of labour societies

7. Ecological variability

8. The major transitions 1.Replicating molecules -> Populations of molecules 2. Independent replicators -> Chromosomes 3. RNA as genes & enzymes -> DNA genes / protein enzymes Prokaryotes -> Eukaryotes Asexual clones -> Sexual populations Protists -> Multicellular organisms Solitary individuals -> Colonies w/ non-reproductives Primate societies -> Human societies with language Amalgamation: e.g. Chromosomes, eukaryotes, sex multicellular colonies. Specialization: e.g. DNA & protein, organelles, anisogamy, tissues, castes Obligate Symbiosis: e.g. Organelles, tissues, castes Conflict and Mediation: Meiotic drive (selfish non-Mendelian genes), New Forms of Information Transmission: DNA-protein, cell heredity, epigenesis, cultural transmission.

9. Hamilton’s rule b r> c b:help given to recipient r:degree of genetic relatedness between altruist and recipient c:price to altruist in terms of fitness Formula valid for invasion and maintenance Applies to major transitions in evolution.

10. Language evolution Language can evolve by means of natural selection provided there were: (a) alternative linguistic information-processing mechanisms in ancestral populations; (b) differences in linguistic information-processing mechanisms were heritable; and (c) some underlying linguistic information-processing mechanisms conferred an inclusive fitness advantage while others did not.

11. Previous hypotheses Group-benefit Gossip Social bonding Grooming Hunting Pair Bonds Individual-benefit Motherese Sexual selection Song Mental toolkit Tool making Számadó & Szathmáry (2006). Selective scenarios for the emergence of natural language. Trends in Ecology and Evolution, 21 (10), 555-561.

12. The fragility of cooperation It is incorrect to assume that human language must have evolved as a system of harmonious cooperation. When cooperation is observed it must be explained rather than assumed (Brown, 2008). Brown (2008). Sociogenomics for the cognitive adaptationist. In C. Crawford & D. Krebs (Eds.), Foundations of evolutionary psychology (pp. 171-182). Psychology Press/Lawrence Erlbaum.

13. Sender-receiver conflicts “Language permits individuals to share information and its use includes both signal and receptor functions. Since other signal-receptor systems seem susceptible to super-stimulation (Ryan 1990), language may provide an exceptional opportunity for sensory exploitation…” (page 7) Rice & Holland (1997). The enemies within: intergenomic conflict, interlocus contest evolution (ICE), and the intraspecific Red Queen Behavioral Ecology and Sociobiology, 41, 1-10.

14. Social transmission Cultural systems must be susceptible to sensory exploitation, as why else would we have evolved cognitive mechanisms for filtering out socially transmitted information? Brown (2001). Genomic imprinting and the cognitive architecture mediating human culture. Journal of Cognition and Culture, 1, 251-258.

15. Intragenomic conflicts “Perhaps the most interesting thing to come out of the realization of possible conflict within the genome is a philosophical one. We see that we are not even in principle the consistent wholes that some schools of philosophy would have us be.” W.D. Hamilton FRS (1936-2000)

16. Parental investment (PI) Defined as all care by a parent for offspring that increases the likelihood that the offspring survives at the expense of that parents capacity to care for others (alive or yet to be born). Examples of parental investment include but are not limited to gamete size, lactation, feeding, protection, and teaching. TriversR. (1972). Parental investment and sexual selection. In: Campbell B, editor. Sexual selection and the descent of man 1871–1971. Chicago: Aldine Press (p. 139–179).

17. Parent-offspring conflict Offspring are selected to extract more resources than mothers are willing to give. Offspring are more closely related to themselves than future siblings sired by the same mother. In mammals with internal gestation, multiple paternity and sex-biased dispersal intragenomic conflicts within offspring are expected.

18. Parental antagonism Evolutionary theorist David Haig has hypothesised that organisms are not cohesive wholes and conflicts between parental genomes within offspring are expected. Haig, D. (2000). Genomic imprinting, sex-biased dispersal, and social behavior. Annals of the New York Academy of Sciences, 907, 149–63.

19. Relatedness asymmetries

20. Barton, Surani, & Norris, M.L. (1984). Role of paternal and maternal genomes in mouse development. Nature 311 (5984), 374–376.

21. What is genomic imprinting? Keverne, Fundele, Narasimha, Barton, Surani (1996). Genomic imprinting and the differential roles of parental genomes in brain development. Developmental Brain Research, 92 (1), 91–100.

23. Parental antagonism theory I suggest that language evolved for two functions. First, as a means for offspring to elicit resources from their parents (e.g. infant crying as requests for food) benefiting paternally-expressed genes. Secondly, to fostering cooperation between maternal kin, predicting that these cooperative functions of language (e.g. infant cooing indicating satiation), will be determined in part by maternally-expressed genes.

24. Patrigenes Matrigenes Tomasello, M. (2008). Origins of Human Communication. MIT Press.

25. Study One Questions What is the percentage of imprinted allelic variants associated language phenotypes? Are these language-associated phenotypes consistent with parental antagonism theory? Prediction: The frequency of imprinted genes involved in language phenotypes will be greater than expected by chance.

26. Study One Methods To test the hypothesis the Online Mendelian Inheritance in Man (OMIM) database (www.ncbi.nlm.nih.gov/omim ) was consulted for allelic variants that have effects on language (i.e., where the word “language” was found in the full text search of genes with and without allelic variants).

27. Study One Methods Since chromosomal size could bias the results, it was held constant in analyses by dividing the frequency of language search results by the estimated number of genes on that chromosome. For example if language loci were randomly distributed to each chromosome, then we would expect the likelihood of a ‘language’ allele to occur on the X chromosome would be approx seven percent.

28. Study One Results Since only 1-2 percent of the mammalian genome is imprinted we would not expect any of the 14 ‘language loci’ to be imprinted unless parental antagonism had played a role language evolution. Interestingly, 36 percent of language loci uncovered in the OMIM database are subject to possible parent-of-origin effects: binomial test p < 0.001.

29. Study One Results The observed frequency of so-called language loci on the X chromosome is 29 percent which is much higher than would be expected by chance (binomial test p < 0.001). This unexpected finding is consistent with the X-linked inhibitory bias hypothesis (Haig, 2006), whereby relatedness asymmetries could favour matrilineal cooperative exchanges via language.

31. Angelman andPrader-Willi Syndromes Sister genetic conditions with critical epigenetic difference (paternal or maternal mark). Angelman syndrome children (pat > mat expression) are good at making demands, but never fully develop language. Brown & Consedine (2004). Just how happy is the Happy Puppet? An emotion signalling and kinship theory perspective on the behavioral phenotype of children with Angelman syndrome. Medical Hypotheses, 63, 377-385.

32. Communication Transmission Feuk et al. (2006) found that the absence of paternal FOXP2 gene among individuals with Silver-Russell Syndrome causes developmental verbal dyspraxia in the expressive but not receptive domains. Reception Hamelin et al (2006) found that Turner’s syndrome subjects with an X chromosome of maternal origin were less likely to have sensorineural hearing loss compared to those with an X of paternal origin.

33. Human uniqueness? Overlapping genes are genes whose transcription regions are shared. Overlapping genes regulate key gene expression mechanisms in genomic imprinting. If imprinted genes played a special role in the evolution of human language, we expect to find significant differences between chimpanzees and humans’ overlapping imprinted loci.

34. Study Two Methods For investigating of evolutionary relationships I used the Evolution Visualizer for Overlapping Genes (EvOG) http://neobio.cs.pusan.ac.kr/evog/ It contains overlapping genes common across Human, Chimpanzee, Cow, Mouse, Chicken, Rat, Zebrafishetc.

35. Study Two Methods Similarity of overlapping gene proximity.

36. Study Two Results Do human chromosomes with language adaptations exhibit less similarity in their overlapping imprinted genes compared to chimpanzees? Seemingly not, as chromosome 11 has significantly lower (both p’s < 0.01) similarity to chimpanzees (M = 0.89, SD = 0.14) than chromosomes 7 (M = 0.99, SD = 0.01) and 15 (M =0.97, SD = 0.03): [F (2, 157) = 23.22, p < 0.001].

37. Study Two Results Why Chromosome 11? The human-chimp divergence between overlapping imprinted genes on chromosome 11 is notable, because there appears to have been recent natural selection (Voight, Kudaravalli, Wen, Pritchard, 2006), and chromosome 11 contains genes associated with schizophrenia (Klar, 2004).

38. Study Two Results Interestingly, on chromosome 11 there has been greater chimpanzee-human evolutionary divergence in the maternally expressed overlapping gene region H19, OSBPL5 (similarity index = 0.5392) compared to all other overlapping imprinted gene pairs (mean similarity index = 0.9471): t (174) = 57.86, p < 0.001. H19 is a growth suppressor gene.

39. t (100) = 3.41, p < 0.05 Levene’s test p < 0.001)

40. Study Two Summary Similarity between chimpanzees and humans in overlapping imprinted genes could either be due to constraint or selection maintaining gene proximity. Divergence between human and chimpanzee overlapping maternal genes (compared to paternal gene pairs) is consistent with Keverne’sendocrinological emancipation hypothesis. Keverne et al. (1996). Primate brain evolution: genetic and functional considerations. Proc. R. Soc. Lond. B Biol. Sci. 263 (1371), 689–696.

41. Cultural Hitchhiking? If fissioning of parental language demes is caused (in part) by conflicts within families, then parentally- and maternally-biased gene expression patterns could be correlated with present-day language diversity patterns. Hal Whitehead’s (1998) idea of cultural hitchhiking. Whitehead, H. (1998). Cultural selection and genetic diversity in matrilineal whales. Science, 282, 1708-1711.

42. Cultural Hitchhiking ‘Shibboleths’ Languages use arbitrary conventions to signal meaning. This arbitrariness could be a clue to origins. Signals restricted to families of relatives, such as shibboleths (i.e., signals only decodable by those who are part of the group), probably are a key characteristic of early hominin language evolution. Can parental antagonism theory tell us anything about language diversity?

43. How to test hypothesis One way to test this hypothesis is to have information on a genetic element that exhibits parent-of-origin effects, varies in frequency across human populations with differing levels of language diversity. An example of ‘markers’ that vary across human populations are Alu elements (note: one of my favourite ‘molecular parasites’)

44. Alu elements Alu elements are a family of retrotransposons specific to primates (often referred to as molecular fossils) that were integrated early during primate evolution. There are approximately 5000–7000 Alu insertions unique to humans

45. Parent-of-origin Alu elements Imprinted genes and Alu insertion sites share several characteristics One is that imprinted and Alu elements are transcriptionally regulated by CpG methylation. Another is that both imprinted and Aluelements tend to cluster in the genome. Waterland & Jirtle (2004). Early nutrition, epigenetic changes at transposons and imprinted genes, and enhanced susceptibility to adult chronic diseases. Nutrition, 20, 63-68

46. Parent-of-origin Alu’s Parent-of-origin effects have been found among 19 Alu elements in a sample of 48 three-generation families. Six Alu polymorphisms were more strongly methyated paternally relative to maternally Only one Alu element that exhibited significantly stronger maternalmethylation. Sandovici et al. (2005). Interindividualvariability and parent of origin DNA methylation differences at specific human Alu elements. Human Molecular Genetics, 14, 2135–2143

47. Study Three Methods I used the TranspoGene database, (http://transpogene.tau.ac.il/) which provides a complete Alu map of the human genome, whereby each Alu element is annotated with respect to coding region and exon/intron location. Are Alu’s inserting themselves in imprinted language loci?

48. Alu’s and language loci As predicted, there was significantly higher than expected frequency of Aluelements inserted in the protein coding machinery of imprinted language loci (472) versus the number of Alu elements inserted into non-imprinted autosomal language loci (124): χ² (1) = 203.20, p < 0.001. A similar pattern was revealed for the X-chromosomal language loci.

49. Alu’s and language diversity? Language diversity was calculated using Greenberg’s index values from www.ethnologue.com. Greenberg’s language diversity index is the probability that two randomly selected people in a country would have different first languages. Higher values (e.g., “1”) indicates that no two individuals have the same first language.

51. Study Three Methods To determine the language diversity for a culture that resides across multiple countries the mean index was used in analyses. Larger cultural groups (i.e., African, Asian, European, and Indian) were included as a covariate in analyses to help reduce to some extent the effects of shared geographic region.

52. Study Three Results Controlling for larger cultural group, it was found that the frequency of paternally-silenced, but not maternally-silenced Alu element insertions were positively associated with language diversity across cultures: partial r pat (28) = 0.54, p < 0.01; partial r mat (28) = 0.35, p > 0.057. Alu elements exhibiting strongest correlations with language diversity were on Chromosome’s 13 and 20.

53. Recap of Findings Theoretically I predict that paternal genes (within the child) extract resources via demands on mother; while maternal genes (within the child) minimise demands via satiation cues. Imprinted and putatively imprinted genes (e.g., UBE3A & FOXP2) are implicated in language phenotypes. Overlapping transcripts among maternal genes (compared to paternal genes) show greater evolutionary divergence. Brown, W. M. (in press). The Parental antagonism theory of language evolution: Preliminary evidence for the proposal. Human Biology

54. Recap of Findings So-called molecular parasites Alu elements are likely to play an important role in imprinted gene regulation of language loci. Cultural hitchhiking of parentally-derived Alu elements may track language diversity.

55. What makes us human?(Eörs Szathmáry slide for “Darwin Day” at the Collegium Budapest) Note the different time-scales involved Cultural transmission: language transmits itself as well as other things A novel inheritance system

56. Conclusions Genomic imprinting plays a role in language development and maternally-derived overlapping transcripts show greater evolutionary divergence. Cultures with more language diversity appear to have more Alupaternal silencing (relative to maternal silencing). Unlike previous approaches to language evolution, parental antagonism is inherently testable using genomic data.

57. Acknowledgments Natural Sciences and Engineering Research Council of Canada David Haig, Ben Dickins, Lee Cronk; Christopher Badcock, Bernie Crespi, Alex Mesoudi, Pete Richerson , Boris Palameta; Chris Moore; Harry Smit