Language emerges only in kin-related groups or if it used to talk to oneself

Marco Mirolli*° (mirolli2@unisi.it), Domenico Parisi°
*University of Siena	°ISTC, National Research Council

Language requires the co-evolution of both speakers and understanders. 
The ability to speak, i.e.  to produce useful signals, is useless unless
there are individuals that respond to the signals appropriately and,
vice-versa, the ability to understand signals is useless unless that are
individuals that produce useful signals that systematically co-vary with
specific aspects of experience.  This implies that if speaking is
advantageous for the receivers of signals but not for the individuals
that produce them, it is not clear how language may have evolved. 

Human language is quite unique among animal communicative systems in
that it is often used to inform conspecifics about aspects of the
environment.  It is an open problem whether this use of communication
was the principal function for which human language evolved [Bickerton,
2002] or language evolution started for more social reasons (such as
facilitating social interaction and social coordination; cf.  [Knight et
al., 2000], [Dunbar, 1996]).  In any case, the evolution of language for
informing conspecifics is something that must be explained in that when
an individual uses language to inform a conspecific about some aspect of
the environment, the advantages for the hearer are clear but those for
the speaker aren't. 

We describe a set of simulations in which a population of individuals
lives in a world containing both edible and poisonous mushrooms, with
each individual mushroom being different from all others [Cangelosi &
Parisi, 1998].  To survive and reproduce an individual must categorize
encountered mushrooms as edible or poisonous and respond by approaching
and eating the edible mushrooms and avoiding the poisonous ones.  The
behavior of the individuals is controlled by a neural network.  Input
units encode the perceptual properties of encountered mushrooms and
heard signals; output units encode movements for displacing oneself in
the environment and for producing sounds.  The networks' connection
weights are found using a genetic algorithm. 

If the sensory capacities of the individuals are spatially restricted
such that distant mushrooms can be localized but not recognized, an
individual is forced to approach all mushrooms until a mushroom is
sufficiently near to be recognized as edible or poisonous.  This of
course is not very efficient.  If a conspecific which is nearer to a
distant mushroom produces a signal which co-varies with the mushroom's
category, hearing the signal would allow the first individual to
approach only edible mushrooms - with a clear increase in efficiency. 
However, this is a situation in which the production of useful signals
benefits the receivers of signals but not their producers.  Hence,
language should not emerge in these circumstances.  Our simulations show
in fact that in our simulation scenario language does not emerge if
speakers and hearers in linguistic exchanges are randomly selected from
the population but it does emerge if the speaker and the hearer are (at
least some of the times) genetically similar individuals, i.e., if they
are kin.  This result is consistent with previous work ([Ackley &
Littman, 1994] [Oliphant, 1996]) and might suggest that language (or, at
least, language for informing other individuals about the environment)
has emerged within small groups of kin-related individuals rather than
in larger communities of unrelated individuals. 

However, even in its early evolutionary stages language may have had not
only a social function but also an individual one: to talk to oneself,
for example as an aid to memory.  In talking to oneself the same
individual is both speaker and hearer.  Hence, speaking and
understanding must necessarily co-evolve.  In another simulation we show
that language evolves even if speakers and hearers in linguistic
exchanges are genetically unrelated individuals provided that the hearer
repeats to itself heard signals in order to keep information about
mushrooms in its memory. 

References

[Ackley & Littman, 1994] Ackley, D.  H.  and Littman, M.  L.  (1994)
"Altruism in the evolution of communication".  In Brooks R.  A.  and
Maes Pattie (ed.) Artificial Life IV: Proceedings of the International
Workshop on the Synthesis and Simulation of Living Systems.  MIT Press. 

[Bickerton, 2002] Bickerton, D.  (2002) "Foraging Versus Social
Intelligence in the Evolution of Protolanguage".  In Alison Wray, (ed.)
The Transition to Language.  Oxford: Oxford University Press. 

[Cangelosi & Parisi, 1998] Cangelosi, A.  and Parisi, D.  (1998).  "The
emergence of a language in an evolving population of neural networks". 
Connection Science, 10(2):83--97. 

[Dunbar,1996] Dunbar, R.  I.  M.  (1996) Grooming, Gossip and the
Evolution of Language, London: Faber and Faber. 

[Knight et al., 2000] Knight, C., Studdert-Kennedy M.  and Hurford, J. 
(ed.) (2000) The Evolutionary Emergence of Language: Social Function and
the Origins of Linguistic Form.  Cambridge University Press. 

[Oliphant, 1996] Oliphant, M.  (1996) "The dilemma of Saussurean
communication".  Biosystems, 37(1-2):31--38.