Re: Evolutionary Spandrels

From: Stevan Harnad (
Date: Sun May 26 1996 - 18:21:29 BST

> Date: Fri, 19 Apr 1996 10:17:37 GMT
> From: (Judy Chatwin)
> What is an evolutionary "spandrel"?
> The concept of an evolutionary "spandrel" has come about from an
> architectural term relating to a necessary by-product of creating
> something else. The exact analogy quoted by Pinker and Bloom (1990)
> relates to Gould and Lewontin"s (1979) reference to the arches and
> spandrels in the dome of the San Marco basilica in Venice; "Spandrels -
> the tapering triangular spaces formed by the intersection of two rounded
> arches at right angles ... are necessary architectural by-products of
> mounting a dome on rounded arches." This feature has been elaborately
> decorated with mosaics. An easy mistake would be to assume that the
> triangular spaces had been put there intentionally as part of the
> detailed decoration. In fact, they were there because of something
> else. This argument is used by Gould and Lewontin to explain some
> aspects of evolution which they do not feel can be accounted for by
> natural selection. Epiphenomenal spandrels are seen as those which exist
> but are of no real interest, such as the redness of blood, the V-shaped
> space between a pair of fingers or the fact that there are a
> prime-number of digits on each limb. Such spandrels have no direct
> relevance to any behaviour or function and give no clues as to whether
> the structures that they are associated with were shaped through natural
> selection.
> The next rung up the ladder is to look at those spandrels which have
> been modified and put to better use. This cannot be done alone and
> requires the intervention of natural selection. It may be seen as the
> process responsible for the modified design of an already existing part
> in order to put it to more effective use.. This is where nature makes
> the best possible use of what already exists to suit its own needs and
> purposes;

Be careful! This does not sound like the blind fumble and find of
evolution. It sounds purposive, which evolution isn't. If anything
(whether random variation or a spandrel) is (1) genetically encoded and
(2) helps the bearer survive and reproduce then, by definition, it is
more likely to appear in the next generation. But no one is putting
anything to "more effective use"... and nature has no "needs and

> remembering that it does not have an empty page on which to
> begin its creativity.. These are usually a single, unique part or
> repeated shapes or processes involving simple physical or geometric
> laws, such as hexagonal honeycombs and spiral markings. These are
> modified and arranged alongside other more complex processes in order to
> perform a function. Ultimately, they can only be explained by natural
> selection.
> The nonselectionist aspect of evolution concerns unmodified spandrels.
> This is used to describe a structure that already has a useful purpose
> but that original function is usurped by another. Gould (1987a) (as
> cited by Pinker & Bloom (1990)) talks about a wading bird that uses its
> wings primarily as a sun visor when looking for fish in the water. It
> is difficult to argue successfully that something with such a complex
> engineering design as the wing would have evolved through natural
> selection for the primary purpose of blocking out the sun. A simpler
> structure would have done the job just as well.
> This explanation of spandrels attempts to clarify what is being
> discussed in terms of nonselectionism. It is also important when
> looking at evolution to be aware of two other possibilities which exist
> outside of selectionism.
> Firstly, Gould proposed the theory of "punctuated equilibrium". He
> disagreed with Darwinian naturalism and proposed that evolution occurred
> in bursts. These were followed by longer periods where no change
> occurred. Pinker & Bloom (1990) disagree with this and firmly believe
> that evolution is gradual from generation to generation. They see
> "adaptive complexity" as the main explanation for this theory.

Whether gradual or in bursts, the crucial thing is that there is
variation (whether random or spandrel) and then there are its adaptie
consequences (in terms of survival and reproduction).

> Secondly, "exaptation" Gould & Vrba, (1982), (as cited by Pinker &
> Bloom, 1990) refers to situations where new uses are found for parts
> which were originally adapted to a different function. It also refers
> to spandrels that had no function at all but existed because of
> architecture, history or development. This may be illustrated by
> looking back to the common origins of parts of bat wings, seal flippers,
> horse forelimbs and human arms. This theory of "exaptation" may be seen
> as a parallel to Darwin"s term "preadaptation". This is the time when
> the "part" in question is fulfilling the two functions - the old and the
> new. Gould has been seen as presenting a revolutionary theory of
> evolution, undermining much of Darwin"s original theories. This
> parallel illustrates that it is not so radical but more of a change in
> emphasis.
> Acknowledgement is made regarding the importance of natural selection
> and Gould and Lewontin are in agreement that certain structures, such as
> the vertebrate eye, must have come about through this physical process.
> There is no other possible explanation. The Darwinian view that
> selectionism is "the most important of evolutionary mechanisms" (Gould
> and Lewontin (1979); as cited by Pinker & Bloom (1990)) is accepted but
> they believe that there are nonselectionist explanations for some
> biological traits.
> Pinker & Bloom acknowledge that natural selection is not perfect and
> that some traits evolve at the cost of others, these are referred to as
> "inherent trade-offs". For example, an animal may possess an elaborate
> appearance to make it attractive to the opposite sex but not to such an
> extent as to prevent it fleeing from predators. Some compromise has to
> exist.

Evolution certainly isn't "perfect"! Perfect compared to what? And in a
changing world, how long would "perfection" last anyway?

What happens is that evolution "makes do." Nothing succeeds like
success, but success need not be perfect: It just needs to be good
enough (or better than the competition). Herb Simon called it
"satisficing" (satisfying/sufficing) as opposed to "optimising" (which
engineers mat try to do when they build a machine, but the Blind
Watchmaker would not and could not do).

> Having looked at some evolutionary theories, this essay will now
> concentrate on a particular area for discussion - language. This is not
> referring to external languages such as French, English or Japanese.
> These are only symbols used to transmit feelings and emotions.

And not to transmit concepts and meaning too??

> The area
> in question is that relating to the universals that form the basis for
> the theories of universal grammar. Debates may occur as to whether
> language exists because of our physiology or our culture. However, for
> the purposes of this essay it may be assumed that the evidence supports
> the physiological view. The issue of concern is whether the ability for
> language has evolved through natural selection or by nonselectionist
> means. This does not mean to discount any environmental involvement but
> to clarify the issue being discussed. Is language an evolutionary
> "spandrel"?
> Chomsky turned around long-standing beliefs on language acquisition in
> his review of Skinner"s Verbal Behaviour. In it he argued that language
> was not learned but dependent on "an innate, species-specific module
> that is distinct from general intelligence" (Pinker (date?)). He drew
> these conclusions from his observation that children had the ability to
> learn languages that contain very subtle and abstract principles. They
> are able to do this without teaching or imitation.

But remember he was talking specifically about GRAMMAR only... They
learn the whole language, pronunciation, vocabulary, concepts, style and
all, but Chomsky's work pertains specifically to grammar, and indeed
only the universal part of grammar.

> If it is to be believed that language came about through natural
> selection then it would be reasonable to suggest some precursor should
> exist in our closest relatives, the chimpanzees. Although they can be
> taught certain hand signals and even sounds there is no evidence that
> they have the ability for language as such. Consequently any link
> between gesture and language in chimpanzees has been rejected as having
> no connection with the universals being discussed.

But there is gesture language in people, and that IS governed by UG.

> Pinker does not see
> this as a stumbling block in his argument for natural selection.
> Instead he points out the time factors involved. Chimpanzees and
> humans shared common ancestors around 6-7 million years ago. This means
> that there are about 300,000 generations in which language could have
> evolved. This may be compared with the mathematical scenario whereby a
> mouse like animal is subject to selection pressure for increased size.
> This pressure is so small that it cannot be observed or measured for
> individual variation; it would take 12,000 generations for it to evolve
> to the size of an elephant. Such a time span is seen as geologically
> "instantaneous", making 300,000 generations sound more than sufficient
> for language to evolve. Presumably a complex cognitive function such as
> this would involve far more individual variations than simply size.

Language is not only complex, it's probably a combination of skills;
some are more likely to be inborn than others. For some, an evolutionary
story makes more sense than for others. Chomsky himself no longer speaks
about "language" in general, but about specific capacities (mostly
grammatical ones).

> Gould puts forward the argument for the nonselectionist view of language
> as a spandrel. This view is supported by Chomsky and
> Piattelli-Palmarini. He does not doubt that the evolution of the
> enlargement of the brain had an adaptive basis. He does, however,
> question the purpose of that increase in size. He suggests that
> language may be a by-product of such rather than the reason for it. He
> sees the mind as a single general- purpose computer, believing that any
> sophisticated machine will perform tasks that it was not programmed to
> do by virtue of its complexity.

Brain size is vague; UG is specific. Language in general is vague; UG is

> Pinker & Bloom suggest that the increased size of the brain may affect
> its quality but do not see it as a sufficient condition for language.
> Simply increasing the number of neurones in a circuit or putting more
> circuits into the brain does not ensure that more refined abilities will
> occur. Such an increase would require detailed design, which could only
> come about through natural selection. Pinker & Bloom feel that this is
> not explained in Chomsky"s nonselectionist theory. They say that "the
> most likely explanation for the complex structure of the language
> faculty is that it is a design imposed on neural circuitry as a response
> to evolutionary pressures."

It would be interesting to hear the evolutionary-pressure story applied
specifically to UG: How would IT evolve? Under what pressures? HOW is it
like the eye?

> Piattelli-Palmarini (1989) (as cited by Pinker & Bloom (1990)) suggests
> that "Grammar is not completely predictable as an adaptation to
> communication, therefore it lacks design and did not evolve by
> selection." He makes the point that these specific linguistic phenomena
> are not actually essential for human survival or reproduction,
> therefore, it seriously calls into question its evolution through
> natural selection. He believes that if language had been an adaptation
> then it would have been both different and better. However, he does not
> offer an alternative explanation for the existence of language only a
> criticism for the lack of adaptive support.

There are problems with an adaptive story for UG, but much fewer
problems for an adaptive story of the advantages of being able to
exchange symbolic propositions as we are now. Trouble is, that's too
general, and is probably based on a bundle of capacities.

> Pinker & Bloom suggest that just because an explanation cannot be found
> for one specific aspect of language, it does not necessarily mean that
> the entire module is a spandrel. Some human organs may contain many
> spandrels (such as the redness of blood), does this mean that the organ
> is a spandrel?

Besides, is all of language a module? Or is it several modules?

> The alternative view relating to this is that language evolved through
> natural selection. It is seen as a gradualistic, Darwinian approach
> which came about as a series of steps. In order for this theory to hold
> firm, individuals must have possessed some genetic variation in their
> grammatical ability which created tiny mutations each beneficial to its
> owner. These individual characteristics must have had some reproductive
> advantage for them to have been maintained by the species during
> evolution. Also there has to be enough time and space separating our
> species from our non-linguistic primate ancestors for this to have
> occurred. Pinker & Bloom admit that there is no conclusive data to
> support these issues but still believe that there is no other plausible
> explanation. Chomsky fails to explain the inevitable genetic variations
> that occur when putting forward his theory for language as a spandrel.

And in this respect, Chomsky seems to be describing the facts more
explicitly than Pinker and Bloom. There "must have been an advantage"
and "it must have happened gradually" are not exactly facts...

> Although communication is possible at a primitive level, it has been
> suggested that there are reproductive advantages for having better
> grammar. Those who have the ability to gain the attention of the
> listener and have strong powers of persuasion surely must be at an
> advantage over the less articulate. Good linguistic skills could make a
> Darwinian difference.

But do you need UG for that? Or would a much simpler grammar have done
just as well, perhaps even a learnable one...?

> Pinker and Bloom firmly believe that language fulfils the requirements
> for adaptation. Their argument suggests that it is a complex design
> which provides a reproductive or survival function. This is supported
> by their belief that it provides the ability necessary to communicate.
> This theory denies any possibility that language may be a spandrel of
> any general cognitive learning ability because of its dissociability.
> Disease and damage have shown that it functions independently of, for
> example, normal intelligence.

Unfortunately, "language" in this sense is too much of a mixed bag. Some
things (like vocabulary and style) no one would deny were cultural.
Others (like the vocal apparatus and humans' social/communicative
inclinations) are clearly more biological. Where does that leave UG?

> The above has shown that there are two possible schools of thought
> regarding the human, innate ability for universal grammar. Pinker &
> Bloom have attempted to argue in favour of it as a result of natural
> selection, whereas Gould and Chomsky see it as an evolutionary spandrel.
> The arguments in favour of natural selection seem greater than the
> opposing theories. Language is a complex process requiring many
> different abilities, it is hard to accept that anything so fundamental
> and intricate is nothing more than a by-product of another part.

But is what is true for some of those abilities necessarily true for all
of them?

This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:24:16 GMT