Brains, Genes, and Language Evolution Morten H. Christiansen - - PowerPoint PPT Presentation

Brains, Genes, and Language Evolution

Morten H. Christiansen

Cornell University Santa Fe Institute

Brains, Genes, and Language

• We need genetic constraints to explain
• the close match between language and

underlying neural mechanisms

• the complex and intricate structure of

language

• the existence of cross-linguistic patterns of

similarity

• the uniqueness of human language

“It’s not a question of Nature vs. Nurture; the question is about the Nature of Nature.” Liz Bates

• The role of language evolution modeling:
• Evaluation of existing theories
• Exploration of theoretical constructs
• Exemplification of how a new theory may work
• Predictions for new experimental research

Outline

• Language shaped by the brain
• Case study: Sequential learning and language
• Modeling the emergence of word order
• Prediction: Structure from iterated sequential

learning

• Prediction: Genetic link between sequential

learning and language

Language Shaped by the Brain

Language Learning and Evolution

• Why is the brain so well-suited for learning

language?

• Why is language so well-suited to being

learned by the brain?

• Cultural transmission has shaped language

to be as learnable as possible by human learning mechanisms

E.g., Christiansen (1994), Deacon (1997), Kirby (2000)

“The formation of different languages and

• f distinct species, and the proofs that both

have been developed through a gradual process, are curiously parallel . . . A struggle for life is constantly going on among the words and grammatical forms in each language. The better, the shorter, the easier forms are constantly gaining the upper hand . . . The survival and preservation of certain favored words in the struggle for existence is natural selection.”

Darwin (1874: 106)

Language thought cognition sensori- motor socio- pragmatic

Language from Constraints

Source: Christiansen & Chater, BBS, 2008

Language Shaped by the Brain

Language Language

Case Study: Sequential Learning and Language

Might word order derive from constraints on sequential learning amplified through cultural evolution?

Modeling Goals

• Explore the role of pre-adaptations for

complex sequential learning

• Evaluate the effect of retention of pre-

language sequential learning abilities

• Exemplify interactions between cultural and

biological evolution

• Make predictions regarding the relationship

between sequential learning and language

Constraints on Sequential Learning

• Sequential Learning: The ability to encode and

represent the order of discrete elements

• ccurring in a sequence
• Non-human primates not good at learning

hierarchically ordered sequences (Conway &

Christiansen, 2001)

Sequential learning

Biological Adaptation 500 generations

Simulating the Role of Sequential Learning in Language Evolution

Time Language + Sequential learning

Biological + Linguistic Adaptation

The Learners: SRNs

(Simple Recurrent Network – Elman, 1990)

Context copy-back Output Hidden Input

• Trained on a serial-reaction time (SRT) task (Lee, 1997)

current location next location

Source: Reali & Christiansen, Interaction Studies, 2009

previous internal state

1 2 3 4 5

3 1 4 5 2

3 2 4 5 1

4 3 2 5 1

4 3 2 1 5

Scoring SL Performance

5 2 3... 4 1

Full-conditional probability vector for possible next location Probability vector for possible next location

5 2 3 ...

Mean Cosine

Context copy-back Output Hidden Input

Biological Evolution in SRNs

Generation n Generation n + 1

p < .001

Results after 500 Generations

Mean Cosine

0.5 0.6 0.7 0.8 0.9 1.0 Initial Final

Source: Reali & Christiansen, Interaction Studies, 2009

Introducing Language

Time Sequential learning

Biological Adaptation 500 generations

Language + Sequential learning

Biological + Linguistic Adaptation

Language Learning SRN

Context copy-back current word previous internal state next grammatical role Output Hidden Input

Source: Reali & Christiansen, Interaction Studies, 2009

Grammar Skeleton

S! ! !{NP VP}! (1) NP! ! !{N (PP)}! (2) PP! ! !{adp NP}! (3) VP! ! !{V (NP) (PP)}! (4) NP! ! !{N PossP}! (5) PossP! ! !{Poss NP}! (6)

Grammar Example

S! ! ! VP NP! ! (Head Final) NP! ! ! N (PP)! ! (Head First) PP! ! ! adp NP | NP adp! (Flexible) VP! ! ! V (NP) (PP)! ! (Head First) NP! ! ! PossP N ! (Head Final) PossP!! ! Poss NP | NP Poss ! (Flexible)

Scoring Language Performance

V Prep ...

Mean Cosine

EOS Poss O S

Full-conditional probability vector for possible next grammatical roles Probability vector for possible next grammatical roles

V Prep ...

Context copy-back Output Hidden Input

S! ! ! {NP VP}! (1) NP! ! ! {N (PP)}! (2) PP! ! ! {adp NP}! (3) VP! ! ! {V (NP) (PP)}! (4) NP! ! ! {N PossP}! (5) PossP!! ! {Poss NP}! (6)

Biological Evolution

Language 3’ Language 2’ Language 4’ Language 1’ Language P Language 2 Language 1 Language 3 Language 4 Language P’

Linguistic Evolution

0.25 0.50 0.75 1.00 1 20 40 60 80 100 120

Consistency Flexibility Generations

Source: Reali & Christiansen, Interaction Studies, 2009

Evolving Head-Order Consistency

Biological vs. Linguistic Adaptation

p < .001 ns

Biological Evolution (L constant) Linguistic Evolution (N constant)

Initial Final

Source: Reali & Christiansen, Interaction Studies, 2009

Mean Cosine

0.5 0.6 0.7 0.8 0.9 1.0

The Role of Sequential Learning Constraints

ns ns

Original Simulations

• Seq. Learning

Constraint (No L change)

Mean Cosine

0.5 0.6 0.7 0.8 0.9 1.0

Initial SRNs Final SRNs

Source: Reali & Christiansen, Interaction Studies, 2009

• If language and learners evolve simultaneously,

cultural evolution constrained by sequential learning overpowers biological adaptation

• Sequential learning constraints become

embedded in the structure of language

• Linguistic forms that fit these biases are more

readily learned, and hence propagated more effectively from speaker to speaker

Modeling Recap: Word Order from Sequential Learning Constraints

Prediction 1: Sequential learning constraints should drive language-like cultural evolution in humans

Iterated Artificial Language Learning

• Can sequential learning

biases lead to the cultural evolution of structure, independent

• f any language-like

task?

Iterated Sequential Learning

• Diffusion chains
• Training on 15 consonant strings
• Recall of all 15 strings
• Output recoded and used as input for the

next participant

• 10 participants in each chain

• Language-like distributional regularities

emerge, facilitating learning

• Sequential learning constraints, amplified by

cultural transmission, could have shaped language

Prediction1Recap: Structure from Iterated Sequential Learning

Prediction 2: There should be a genetic link between sequential learning and language

FOXP2 and Sequential Learning

• Recent selection for FOXP2 in humans (Enard

et al., 2002)

• FOXP2 important for the development of

cortico-striatal system (Watkins et al., 2002)

• Cortico-striatal system implicated in

sequential learning (Packard & Knowlton, 2002)

• Could sequential learning be an

intermediate phenotype (endophenotype) for FOXP2 and language?

Molecular Genetic Study

• Participants: 159 8th-graders
• 100 typical language learners
• 59 children with language impairment (LI)
• Both groups have equivalent non-verbal IQ
• Blood or saliva samples obtained for

recovery of DNA

• Visual serial-reaction time (SRT) task

Random Pattern Random 100 trials 100 trials 100 trials 100 trials 2, 4, 1, 3, 4, 2, 1, 4, 3, 1

• DNA base difference between individuals:

Single Nucleotide Polymorphism (SNP)

T A C G C G T A

SNP

Genetics 101

• DNA base difference between individuals:

Single Nucleotide Polymorphism (SNP)

• Sets of nearby SNPs inherited in blocks
• Pattern of adjacent SNPs in a block form a

Haplotype

• Tag SNP: An indicator SNP for the

composition of a haplotype block

Genetics 101

Prediction 2 Recap: FOXP2 Links Sequential Learning and Language

• FOXP2 genotypic variance is associated with

individual differences in SRT learning and language status

• Fits recent molecular genetic results:
• Humanized Foxp2 affects the

striatum in mice (Enard et al., 2009)

Case Study Summary

• Constraints on sequential learning, amplified by

cultural transmission, may help explain word

• rder patterns
• Similar neural and genetic bases for sequential

learning and language

• Sequential learning provides an important

constraint on the cultural evolution of language

Lessons from Language Evolution

• The cultural evolution of language simplifies

the problem of acquisition

• Language acquisition involves learning how

to coordinate linguistic behavior with

• thers, not grammar induction
• The learner’s biases will be the right biases

because language has been optimized by past generations of learners

Source: Chater & Christiansen, Cognitive Science, in press

Conclusion

• The fit between language and the brain

arises because language has been shaped to fit pre-existing domain-general constraints

• Languages have evolved to rely on multiple-

cue integration for their acquisition

• We need to uncover the constraints that

shape the cultural evolution of language

Acknowledgments

Nick Chater Florencia Reali Bruce Tomblin Hannah Cornish Simon Kirby

Thanks!