Content uploaded by Zdenka Brzovic
Author content
All content in this area was uploaded by Zdenka Brzovic on Oct 20, 2018
Content may be subject to copyright.
Synthese
DOI 10.1007/s11229-014-0407-4
The function debate: between “cheap tricks” and
evolutionary neutrality
Predrag Šustar ·Zdenka Brzovi´c
Received: 13 August 2013 / Accepted: 24 January 2014
© Springer Science+Business Media Dordrecht 2014
Abstract We examine the use of the notion of natural selection in the philosophical
debate on functions in biology. This debate has been largely shaped by the way in
which different accounts assess various selective pressures in justifying claims about
biological functions. Cummins (Functions: new essays in the philosophy of psychol-
ogy and biology. Oxford University Press, Oxford, pp 157–172, 2002), one of the
main proponents of the causal role account of biological functions, argues that a cor-
rectly understood neo-Darwinian notion of natural selection has nothing to do with
functional talk in biology. In this paper, we counter Cummins’ account by showing
that progress in the molecular approaches to evolutionary biology—specifically sci-
entific data available in neo-functionalization research—offers valuable support to the
etiological selectionist approach to functions in biological and biologically-related
sciences. Finally, we use the presented data to build our own account of biological
functions, which tries to avoid the wrong turns taken by both major strands in the
biological function debate, namely causal role and etiological accounts. According to
our account, the function of a certain gene or a protein in the biological system that
contains it is a particular causal activity, or a group of causal activities whose man-
ifestation is in a specific way determined by corresponding mechanisms of genetic
expression. Also, we argue that in many important cases this particular expression
of genetic activity was positively selected at a certain point in evolutionary history.
Since we take selection as an important but not the only factor that grounds biological
functions, we are committed to a weak etiological account.
P. Šustar ·Z. Brzovi´c(
B)
Department of philosophy, University of Rijeka, Sveuˇcilišna Avenija 4, 51 000 Rijeka, Croatia
e-mail: zdenka@ffri.hr
P. Šustar
e-mail: psustar@ffri.hr
123
Synthese
1 Introduction
The philosophical debate on functions in biology and related scientific areas can be
divided into two large opposing groups of accounts: etiological and the causal role
accounts of function. Etiological accounts adopt an historical approach to functions:
a trait has a function if in the past the effects of that trait contributed to the selection
of organisms with that trait. On the other hand, according to the causal role accounts,
functions are not effects that explain why a trait is there, but rather they are properties
or capacities of an organism that contribute to the more complex capacities of the
system that contains them.
Cummins (1975,2002), the main proponent of the causal role account of function,
argues for a clear-cut divorce of any relevant kind of scientific function discussion
from evolutionary considerations, more precisely, from the role that the mechanism of
natural selection allegedly ought to play in producing functions. In this paper, our aim
is to examine the role of natural selection in justifying biological functional claims.
Thus, in Sect. 2, we examine Cummins’ arguments against the “strong” and “weak
variations of neo-teleology” since they rely on direct criticism of the use of selective
pressures in grounding legitimate function ascriptions.
We will specifically counter Cummins’ argument for a clear-cut divorce of any
relevant kind of scientific function talk from evolutionary considerations by examin-
ing recent contributions in the molecular approaches to evolutionary biology (Zhang
2003;Piatigorsky 2007). Accordingly, in Sect. 3, we examine the phenomena of
neo-functionalization, i.e., the processes through which novel functions are added
to a group of already existing functions of biological items, such as, at the level of
biological organization primarily under consideration here, genes and proteins. Two
main and complementary theories of neo-functionalization are introduced: the theory
based on gene duplication and the theory based on the mechanism of gene sharing.
In Sect. 4, we argue that a careful interpretation of the scientific data available in
the neo-functionalization research, such as data available in the case of human color
vision, offers valuable support for a weak etiological selectionist account of functions
in the biological and biologically-related sciences (see Buller 1998).
Our conclusion is that the aforementioned theories within molecular evolutionary
biology suggest the following general understanding of biological functions: the func-
tion of a certain gene or a protein in the biological system that contains it is a particular
causal activity, or a group of causal activities, whose manifestation is in a specific way
determined by corresponding mechanisms of genetic expression. In addition, it will
be illustrated in Sect. 3and, especially, Sect. 4, that in many cases, this particular
expression of genetic activity was positively selected at a certain point of evolutionary
history. Scientific research on human color vision—considered as a standard example
of salient biological trait—provides fairly direct support to that kind of philosophical
approach to function talk (Golding and Dean 1998).
One of the more general aims of this paper is to elaborate a less sweeping philo-
sophical account of biological functions. Apart from countering Cummins’ argument
for a clear-cut divorce from evolutionary considerations in the biological function
talk, the account put forward here also counters so-called “cheap tricks” (Cum-
mins 2002) employed within the strong etiological strand of the function debate.
123
Synthese
Our account takes selection as an important factor in grounding function attribut-
ions in biology, but not the only factor as the strong versions of etiological theories
suggest. As we will see in the next section where we examine Cummins’ criticism
of etiological accounts, the strong variations of etiological accounts generalize far
too easily over the role played by natural selection in justifying biological functional
claims.
2 Cummins’ argument for the evolutionary neutrality of functions
Cummins (2002) builds on his account of functional explanation presented in his
seminal Functional Analysis (1975) and introduces some important new elements. The
most important contribution of Cummins’ paper is his extensive critique of teleology,
the idea that we can explain the presence of a trait by appealing to its function. An
important issue concerning teleological explanations relates to the fact that there should
be a background story about how the items in question are produced and what is
the mechanism or process that produces them in virtue of their functions. This is a
requirement that Cummins calls the grounding process of teleological explanations
(Cummins, p. 159).
Traditional teleological explanations were rejected as unsatisfactory because they
used supernatural and unscientific notions for the grounding processes, such as the
intentions or plans of some superior designer or creator. However, after the formula-
tion of modern evolutionary theory, it appeared plausible to ground functional expla-
nations through the process of natural selection. Cummins refers to this type of expla-
nations as neo-teleological where the “neo” part refers to evolutionary processes:
“traits are selected because of the effects that count as their functions” (Cummins
2002, p. 162). This is the basic idea of all etiological accounts of functions and it
seems very attractive in comparison with classical teleology because the ground-
ing process that it advocates is a legitimate and important biological notion. How-
ever, Cummins argues that the evolutionary grounding processes cannot account for
neo-teleology since, as we will see shortly, evolutionary processes are insensitive to
functions.
Cummins differentiates between strong and weak variations of neo-teleology. The
strong variation of neo-teleology, which figures as the basic assumption of the most
influential selectionist etiological accounts (see Cummins 2002, p. 161), essentially
“holds that any biological trait that has a function was selected for because it per-
formed that function” (Cummins 2002, p. 164, italics added). The weak variation of
neo-teleology “holds only that some traits were selected because of their functions”
(Cummins 2002, p. 164, italics added). Now, his main point is that both versions
of neo-teleology are unsatisfactory since functional ascriptions are independent of
evolutionary processes such as natural selection.
2.1 Functions just do not track the factors driving selection
Cummins’ basic claim is that the correct understanding of selective processes occur-
ring in the biological world is detached from functional features that scientific practice
123
Synthese
ascribes to biological structures: “Selection requires variation, and there was no varia-
tion in function in the structures in question, only variation in how well their functions
were performed” (Cummins 2002, p. 165).
Consider in that respect a common example referred to in philosophical debate
about biological functions. It is often claimed that ‘the function of the eye is to see
the world out there’. At this point, we may ignore some worries about this func-
tional statement, for instance, the rather rough functional characterization of the eye
with ‘one and only’ function, and the presupposed normative standards to which
‘seeing the world out there’ should be committed—the whole example is far from
being circumscribed. In other words, the occurrence of natural selection in this, as
in every other case, depends on the relevant kind of variation in the workings of the
eye.
Consider further the following situation: other things being equal, the function of
the eye is to see that particular object out there as a mature and menacing exemplar of a
venomous snake. Let us imagine two people, one person with what we would classify
as normal vision and the other with reduced visual acuity. In the imagined case, the
first person would see the object as a venomous snake while the other might see it
as a colorful, long stick. Now, Cummins’ point here would be that the eyes of both
persons have the same biological function. Their eyes simply differ in the accuracy
of their physiological performances and only this variation can be related to natural
selection; the function of the eye, as a representative of so-called ‘complex biological
structures’, appears to be, by itself, out of reach of selective processes. Cummins
bluntly summarizes the point by emphasizing that “functions just do not track the
factors driving selection” (Cummins 2002, p. 166).
The take-home message for selectionist etiological accounts of functions is
straightforward: either you correctly apply the notion of natural selection to the
project of justifying biological functional ascriptions, in which case, however, the
application of corresponding selective processes is utterly “superfluous” (Cummins
2002)or you have a “misleading” or “distorted” picture of neo-Darwinian selec-
tive processes (Cummins 1975,2002).1Thus, according to Cummins’ argument,
the whole proposal to rejuvenate the validity of etiological accounts of functions
by relying on determinate selective processes seems to be fundamentally wrong and
hopeless.
This is especially the case with complex biological traits, as most commonly
referred to, human eyes, mammals’ heart, and birds’ wings. For something to be
selected because of its function, this function must be an adaptive variation in a popu-
lation. For it to be the case that wings, for example, were selected because they enabled
flight, we must suppose that at one point there was a subpopulation of organisms with
wings that enable flight while in the rest of the population, organisms have wings that
1One of the main consequences of the latter point is highly implausible saltationism, which the defender of
etiological account is forced to accept. We will not here venturefurther into any of Cummins’ other specific
reasons for this objection. Rather, we will concentrate on the part of Cummins’ critique, which draws out
a better proposal on how to accommodate natural selection into biological functional talk. For Cummins’
Darwinian orthodoxy on gradualism, see in particular his 2002 paper (pp. 165–166).
123
Synthese
do not enable flight.2However, this is not the case, especially with complex traits,
since as Cummins claims: “While it is plausible to suppose that there was a first flight-
enabling wing somewhere among the ancestors of today’s sparrows, those ancestors
were not sparrows, nor was the wing in question anything like a contemporary sparrow
wing.“ (Cummins 2002, p. 164) To sum up, to justify functional ascriptions to this type
of trait by calling into play a neo-Darwinian notion of natural selection is to provide a
“bad” background account for the requirement for grounding process (see Cummins
2002, p. 165).
We have just seen that Cummins’ argument refutes the strong variation of neo-
teleology, but we can still ask if that is enough to refute even the weak variation,
which holds that only some traits were selected because of their functions.
2.2 Arguments against the weak neo-teleology
Cummins admits that the weak variation of neo-teleology survives his objections
thanks to the cases where genuine functional novelties are introduced in the population
and selected because they represent an adaptive variation. However, his argument
refutes any significant philosophical potential that these scientific phenomena may
have in restoring the selectionist etiological view on biological functions: “Weak neo-
teleology comes out true only because of the rare though important cases in which the
target of selection is also the bearer of a function that accounts for the selection of that
trait”(Cummins 2002, p. 165, italics added). Cummins’ refutation of this variation of
neo-teleology is based on the following two objections: (1) an objection that relates to
the evolutionary rarity of such cases, and (2) an objection that concerns the functions
of complex biological traits.
We will examine the phenomena of neo-functionalization; the processes through
which novel functions are added to a group of already existing functions of biologi-
cal items. Our aim is to show that a careful analysis of neo-functionalization shows
that Cummins’ refutation of the weak neo-teleology is not justified since positive
Darwinian selection is, in many cases, responsible for the fixation of the traits that
perform novel functions in the processes of neo-functionalization. Objection (1) raised
by Cummins tries to emphasize the fact that the cases where the target of selection is
also the bearer of a function are exceptionally important events in the course of evolu-
tion, but “such cases must be quite rare” (Cummins 2002, p. 166). However, the rarity
objection needs to be updated. The occurrence of novel functions is quite a common
phenomenon, which is also widely spread across different biological taxa.3The traits
2This is how Cummins describes the problem, however, to be more precise it seems that it is correct to say
that in order for wings to be selected because they enable flight, there must be a subpopulation in which
organisms have wings that enable flight, while organisms in the rest of the population do not have wings.
This is because of the fact that, if they were to have wings, then we would not be able to say that wings
were selected because they enable flight, but rather a type of wings were selected that enables flight. This
might be a semantic issue, since if we define wings as organs that enable flight, it follows that the rest of
the population with wing-like organs that do not enable flight cannot have wings.
3For a more detailed update on the growth of evidence in molecular evolutionary biology that calls into
question objection (1), see, e.g., Ohno (1970), Zhang (2003), Piatigorsky (2007). With regards to a broader
scientific and philosophical context in which those data are embedded, cf. Sect. 3in this paper.
123
Synthese
performing novel functions can be fixed either by a neutral process, or by the mech-
anism of positive Darwinian selection, or even by some combination of the two. At
this point, the relative importance of the mechanism of positive selection versus ran-
dom fixation of neutral mutations cannot be precisely determined (Rodríguez-Trelles
et al. 2003), but there are a large number of cases for which there is evidence that
selection is a cause of the fixation of the trait responsible for the novel function and
that selection is a very important factor in the processes of neo-functionalization (see
Hughes 1999).4Cummins has not provided any evidence in favor of his claim about
the rarity of such cases, and it seems that this claim represents an arbitrary assessment
without evidential support. One might also wonder about the use of the term ‘rarity’
since Cummins leaves it unclear what qualifies as rare in evolutionary terms and how
often such cases should occur in order for them to be relevant for the question about
the evolutionary neutrality of functions.
Objection (2) requires a less straightforward approach. As we will see shortly,
it also represents the most challenging objection advanced in Cummins’ argument,
to which an evolution-oriented account of biological functions must respond. This
objection, which rounds up Cummins’ overall dismissal of the selectionist etiolog-
ical approach to functions, appeals to the fact that the selectionist accounts cannot
explain so-called ‘salient biological functions’ performed by complex biological traits
or structures. As seen above, Cummins argues that the overall approach of selec-
tionist etiological accounts of functions is mistaken because it tries to explain why a
trait exists or why it is the way we find it by focusing on variations that affect how
well the trait performs its function. However, the correct evolutionary explanation
should “look at the complex economy of the whole unit of selection” (Cummins 2002,
p. 168). Substantive neo-teleology, in Cummins’ view, wrongly identifies complex bio-
logical traits as targets of selection, which amounts to a “cheap trick” (Cummins 2002,
p. 170) of ignoring the historical details of trait proliferation and maintenance by sim-
ply ascribing a function to it. In other words, the etiological strand of the contemporary
function debate amounts to nothing more than bad Darwinian science.
Since the weak neo-teleology cannot explain these salient examples of biological
functions, but only a very restricted subset of what is usually considered as a biolog-
ical function, Cummins thinks that we should reject such an account of functions as
legitimate. In his words, a seemingly positive case for neo-teleology remains “insignif-
icant at best” (Cummins 2002, p. 166) in approaching the issue of complex biological
traits and their respective salient functions. Considering this, the most promising thing
to do is to divorce philosophical accounts of biological functions from evolutionary
considerations, in particular, from the notion of natural selection.
However, why should these cases be pertinent to Cummins’ refutation even of the
weakest form of an evolutionarily laden account of biological functions? As we have
seen earlier, there are at least some groups of traits or structures to which we can legiti-
4Methods for detecting positive Darwinian selection include showing that nonsynonymous nucleotide
substitutions (resulting in new amino acids) exceed synonymous nucleotide substitutions (no amino acid
change) during the early stages after duplication (Hughes 1999). However, authors like Golding and Dean
(1998) argue that in addition to this kind of evidence, information about the phenotypic level is needed. We
address this issue in Sect. 4.
123
Synthese
mately apply the basic neo-teleological claim according to which the target of selection
is also the bearer of a function that accounts for the selection of that trait. There are
cases of neo-functionalization that represent a paradigmatic case of the matching rela-
tionship between selective processes and function ascriptions to a biological trait of a
certain kind and this is something that Cummins admits. The traits and their respective
novel functions amount to a safe neo-teleological area, which cannot be called into
question even by Cummins’ evolutionarily neutral functional analysis. We believe
that these cases of matching relationships between selective processes and function
ascriptions provide support for a weak etiological account of functions according to
which a trait has a function if past tokens of that trait contributed to the fitness of
organisms bearing that trait (Buller 1998). This theory does not make selection of a
trait a necessary condition for function ascription; selection is an important factor in
many cases, but there are other factors as well, such as genetic drift or situations where
there was no variation of traits for selection to act upon. 5
We will proceed by showing that Cummins’ argument against the weak variation of
neo-teleology relies on incorrect views about the evolutionary processes that generate
functional novelties.
3 The processes of neo-functionalization
As already pointed out, neo-functionalization is a complex evolutionary process, which
brings about functional novelties in the biological system containing the function.
Firstly, we will present the theory of neo-functionalization via gene duplication, which
is a widely endorsed theory accepted by current molecular evolutionary biology. In
what follows, we will also present a new theoretical proposal of neo-functionalization
via gene sharing that complements the theory of neo-functionalization via gene dupli-
cation.
3.1 Neo-functionalization via gene duplication
The theory of neo-functionalization via gene duplication has a relatively long tradition.
Ohno’s book Evolution by Gene Duplication (1970) summarizes and points to opening
research agendas for the theory of gene duplication, specifically, with regard to the
processes of neo-functionalization (see, in particular, Zhang 2003).
In accordance with the theory, the process of gene duplication comes in two dis-
tinct phases. The first phase is the generation of gene duplicates and this phase is
followed by a second phase, which is characterized by three possible outcomes for
5The endorsement of the weak etiological theory commits us to a form of adaptationism—explanatory
adaptationism—according to which the apparent design of organisms and relations of adaptedness between
organisms and their environment represents a crucial problem in biology that is explained by invoking
selection (Godfrey-Smith 2001). It is important to stress that we accept explanatory adaptationism without
endorsing empirical adaptationism, according to which most biologically significant features of organisms
are shaped by natural selection (Sterenly and Griffiths 1999). We hold that natural selection has a special
explanatory importance even if it does not occur often and even if the majority of traits in organisms are
not adaptations.
123
Synthese
the duplicates that are generated: pseudogenization, sub-functionalization and neo-
functionalization. Generation of Gene Duplicates. This phase is concerned with dif-
ferent mechanisms by which an already existing gene in a certain genome acquires its
duplicate. So far, molecular evolutionary biology has singled out the following mech-
anisms: unequal crossing over,retroposition,chromosomal (or genome) duplication,
and segmental duplication. With regard to our species and other species members of
the hominoid group, the mechanism of segmental duplication seems to be particularly
relevant.
Evolutionary Fate of Gene Duplicates. (a) Pseudogenization. One of the main
possible outcomes of the gene-duplication mechanisms is pseudogenization, i.e., the
process by which a gene duplicate loses the function exhibited by its parental gene.
Thus, contrary to the parental gene, the gene duplicate maintains up to a certain
point in evolutionary time its so-called ‘functional redundancy’, but, then, much more
easily than its parental gene, accumulates mutations and, as a consequence, gradually
departs from the original functional performance. Now, the most interesting fact about
the process of pseudogenization, more strictly from the evolutionary point of view ,
is that a gene duplicate, contrary to its parental gene, is somehow hidden to purifying
selection at an evolutionarily early phase after the duplication event. However, given
the instability of a genome containing that type of gene, they are most frequently
removed.
At this point, let us view some recent results about our species, reached through
genomic analyses. Genomic analyses reveal that the rate of pseudogenization in
humans is about one pseudogene for every two functional genes in some com-
pletely sequenced chromosomes (see Harrison et al. 2002;Zhang 2003). There are
other significant findings about pseudogenization in humans besides its exception-
ally high rate. Consider, for instance, those physiological systems that undergo the
so-called “relaxation of functional constraints”, such as in the case of the olfac-
tory receptor gene family in our species. This gene family in humans, in hominoids
more generally, and in mice has approximately 1,000 genes. Nevertheless, the rate
of pseudogenization of the olfactory receptor gene family in humans is over 60%,
while in mice it is estimated to be only around 20% (see Rouquier et al. 2000). The
large percentage difference between the two taxonomic groups can be most plausi-
bly explained by the abovementioned relaxation of functional constraints: due to a
shift of selective pressures from olfaction to improved vision, a good deal of genes
within the olfactory receptor gene family of our species gradually underwent a massive
pseudogenization.
The above findings of human genomic analyses emphasize the importance of
pseudogenization as one of the main phases in the overall evolutionary strategy of
gene duplication. The occurrence of slightly modified or novel functions for a gene
duplicate is, however, a far more important and fascinating evolutionary outcome.
(b) Sub-functionalization. As already pointed out, the situation in which a genome
acquires, most usually, two daughter genes is rather unstable. One way out of that
instability consists in, as shown above, the processes of pseudogenization. However,
it occasionally happens that each daughter gene takes over one of the functions per-
formed by the corresponding parental gene. In other words, in some cases after the
duplication event, there is a structural and functional difference between daughter
123
Synthese
genes in which they split between themselves or sub-functionalize the former function
of their parental gene. From the perspective of evolutionary biology, it is especially
important that divergence—both structural and functional—has a stabilizing effect on
gene duplicates and genomes in general (see Nowak et al. 1997).
Let us finally see how gene duplication can lead to its most important evolutionary
outcome, that is, the process of neo-functionalization. In that way, we willcollect rele-
vant information, firstly, from the theory of neo-functionalization via gene duplication
and, then, secondly, from the theory of neo-functionalization via gene sharing, in order
to carry out an account of biological functions. (c) Neo-functionalization. In brief, neo-
functionalization, according to the theory based on gene duplication, relates to those
outcomes in which a gene duplicate diverges from its parental gene in such a way that
it acquires a new function. The new function of a corresponding protein—since it is
most frequently produced by the diversified structural gene belonging to a larger fam-
ily of similar structural genes—can hardly be entirely novel. Nevertheless, in contrast
to sub-functionalization, neo-functionalization adds a previously non-existing partic-
ular causal contribution to the individual organism bearing the favorable outcome of
gene duplication. There are many distinct cases reported in the literature that illustrate
the characteristics of neo-functionalization via gene duplication sketched above. For
example, molecular studies of evolutionary processes leading to color vision in Old
World monkeys and in primates, according to the phylogenetic analysis endorsed in
Golding and Dean (1998), are particularly successful in providing evidence for the
theory of neo-functionalization.
Human color vision in bright light is conferred by three types of visual pigment
with the maximum absorption wavelength λ(max) values of 420nm (blue), 530 nm
(green), and 560nm (red) (see Nathans 1987;Golding and Dean 1998). Genes that
contain coding sequences for opsins, i.e., the proteins that play the central role in
enabling the so-called ‘spectral tuning’, belong to the same gene family that accrued
during our evolutionary past by the gene duplication processes and, then, by functional
divergence, which in this particular case resulted in adaptive functional novelties. What
is also pertinent to the case of human color vision is the fact that gene duplicates in
distinct stages leading to the trichromatic vision in humans required just a few amino
acid replacements in the already existing opsins.
How could that elegant and relatively simple molecular event produce such a rad-
ical macro-evolutionary change? The debate in the area of molecular evolutionary
biology on the evolution of trichromacy in primates suggests that these few amino
acid replacements in the existing opsins were fixed by positive Darwinian selection:
the trichromatic vision of Old World monkeys and primates represents a functional
novelty conferring a high adaptive value on organisms bearing the trait of the gene
family containing the corresponding gene duplicates (see Golding and Dean 1998,
p. 359). The adaptive value is inferred from the ecological consequences of the color
vision acquisition: individual organisms having the functional novelty in question can
easily explore and construct completely new ecological niches, which, then, can bring
about a significant impact on the related ecosystem. We believe that this hypothesis
represents the best explanation of the described phenomena because the described case
suggests an incremental construction of fit between organisms and their environment
which leads us to conclude that a selective mechanism is at work (Sterelny 2006).
123
Synthese
However, one might wonder if we are justified in drawing this conclusion without
considering the alternative hypothesis that a neutral molecular evolution is responsible
for the acquisition of this functional novelty. This concern is further strengthened if we
take into consideration that it is often assumed that the neutral theory represents the
null hypothesis of molecular evolution (Zhang 2003). Even though there is evidence
that points to the conclusion that this is a case of positive Darwinian selection at
work (Deeb et al. 1994;Hunt et al. 1998;Mollon 1989;Melin et al. 2013;Surridge
et al. 2003;Verrelli and Tishkoff 2004) one might worry that the evidence is still not
decisive. So, why should one endorse the selection hypothesis rather than the neutral
approach that does not require positive selection?
First of all, it is important to notice that the neutral theory of molecular evolution
does not exclude the mechanism of natural selection, it just claims that the majority
of evolutionary changes at molecular level are selectively neutral (Kimura 1991).
Thus, the question is how to identify those relatively rare (according to the neutral
theory) but important cases where positive selection is at work. Even Kimura (1983),
one of the major proponents of the neutral theory of molecular evolution, argues that
the problem of detecting natural selection consists in the difficulty of assessing the
relationship between genotypes and phenotypes since this relationship is often not
straightforward. However, in the cases where this relationship is fairly direct and the
phenotypic change is important for the organisms bearing it, even the neutral theory
concedes that in such cases natural selection is at work.
We want to argue that the described case of the evolution of trichromatic color
vision represents exactly such a case where the relationship between the genotypes
and phenotypes actually is straightforward; a very simple change at the molecular
level produces an important change at the phenotypic level for which we can be fairly
certain that it brought advantage to the organisms bearing it. Golding and Dean (1998)
propose the same approach; for a detailed understanding of adaptive change it is not
enough to provide information on raw sequence and phylogeny, we need phenotypes.
Thus, in cases where we can detect a direct link between a change at a molecular level
and a change at the phenotypic level (Melin et al. 2013), and where the change at the
phenotypic level brings an incremental construction of fit between organisms and their
environment, we are justified to conclude that a positive Darwinian selection is at work.
Basically, we combine information on phylogeny, structural information, information
on physiology and ecological conditions at the time of the fixation of this trait to justify
the conclusion that positive selection is responsible for the fixation of this trait.6
These general characteristics of neo-functionalization according to the theory based
on gene duplication and, particularly, the evidential support this theory finds in the
6In the case of the evolution of trichromacy an additional reason is often provided in favor of accepting
the selection hypothesis: the same parallel amino acid replacements that generate red-sensitive opsins have
ocurred independently in fish, reptiles and mammals which makes the hypothesis that selection is responsible
for the fixation of trichromacy more likely then the hypothesis that all these independent events were the
product of the random fixation of selectively neutral mutation (Golding and Dean 1998). This is because of
the fact that if a new allele comprising duplicate genes is selectively neutral, the probability of its fixation
is small, but if there is positive selection at work, then it is much more probable that it will be fixed in the
population, especially if the same fixation ocurred independently at several times in evolutionary history
(cf. Zhang 2003).
123
Synthese
case study of color vision offer an extremely valuable source of scientific information
for making some progress within the philosophical function debate. How exactly we
can assess the new information obtained at an elevated pace in molecular evolutionary
biology will be seen after the presentation of the theory of neo-functionalization via
gene sharing.
3.2 Neo-functionalization via gene sharing
Relatively recently, evolutionary biology has been enriched by Piatigorsky’s compre-
hensive monograph, Gene Sharing and Evolution: The Diversity of Protein Functions
(2007), which stems from a long tradition of research on the molecular biology of
visual apparatus. Among the many distinct issues that this recent book addresses, a
great deal of attention has been paid to the different ways in which functional nov-
elties arise during evolution. Piatigorsky’s book contributes to the important debate
on the processes of neo-functionalization by offering to evolutionary biology a new
theoretical proposal based on the mechanism of gene sharing. In brief, depending on
the regulation of a gene’s activity, the same polypeptide can perform multiple differ-
ent functions. The following slogan faithfully summarizes the new theory: the same
polypeptide sequence, but with multiple functions. For example, to refer to Piatig-
orsky’s and his group’s main case study, the group of eye-lens crystallins, which are a
particular group of proteins, most usually exhibits the biophysical function of refrac-
tion and the biochemical function of an enzymatic, for instance, metabolic reaction.
The new theory emphasizes that a protein’s function depends directly on the differ-
ential expression of its gene. The multi-functionality of a protein in fact relies on the
same genic sequence: the same gene-coding sequence is present at distinct functions
that proteins most usually perform in the living world.
Apart from that striking feature, this theory of neo-functionalization points to a
different way through which novel, and with that, multiple molecular functions can
arise. The theory of duplication of a parental gene and the subsequent adaptive change
of the gene duplicate is now accompanied by a different view, according to which
the process of regulation of the involved structural gene is considered as the primary
source of functional novelties. Thus, by responding to environmental fluctuations,
the mechanisms of gene regulation are affecting the functional performance that the
produced polypeptide has in a biological system. For a schematic presentation of the
processes of neo-functionalization according to these two theories, see Fig. 1.Onthe
left side of the illustration we have one gene (A) encoding a protein that can perform
two functions: A1and A2. When gene A is expressed highly in a specific tissue the
protein performs both functions A1and A2, but when the gene is expressed at the low
level, the proteins perform only the function A1. On the right side of the illustration
we have the case of the functional specialization of two sibling genes: B1and B2.
Here the two functions are performed by different polypeptides encoded in duplicated
genes (see Piatigorsky 2007).
There are two main ways in which a variety of complex mechanisms of gene
regulation can influence the multi-functionality of proteins: (i) by influencing the
above-mentioned differential expression of the involved gene-coding sequence for the
123
Synthese
Fig. 1 (Reproduced from Piatigorsky (2007), Figure 1.1, p. 4.)
corresponding polypeptide, and (ii) by responding to the behavior of a polypeptide
in specific intracellular and tissue locations. The general thrust of Piatigorsky’s new
theory is effectively summarized by the figure below.
This figure puts forward the two main ways in which the process of neo-
functionalization is directly influenced by the mechanisms of gene regulation. For
instance, Function 5 of the hypothetical polypeptide in the figure above is a result
of post-translational phosphorylation at a corresponding site in the polypeptide. That
event is directly related to the gene regulation mechanisms. As seen in the case of
differential expression, the same relationship to the mechanisms of gene regulation
also holds with regard to Function 1. Functions 2–4 given in the figure illustrate the
second way in which gene regulation is deemed to be the decisive source of functional
novelties. As emphasized in Fig. 2, Functions 2–4 are most usually concerned with the
so-called ‘protein:protein’ interactions, which are determined by different conforma-
tional motifs that are present in proteins. Piatigorsky’s theory of neo-functionalization
via gene sharing clearly suggests that the functions arising from differential expression
are more directly related to the mechanisms of gene regulation than to the other subset
of functions.
So, what can be concluded from the description of these two theories of neo-
functionalization? In our opinion, they show how careful examination of the introduc-
tion of functional novelties offers support to the evolutionary grounding of functional
novelties without, as Cummins would say, cheap tricks. If that is the case, then it fol-
lows that the weak variation of neo-teleology is correct and that Cummins’ argument
against it is not sound, i.e. it is based on the false premise since the weak neo-teleology
need not ignore the details of trait proliferation and maintenance by simply ascribing
a function.7In the final section, we elaborate more specifically on these points.
7Perhaps Cummins could reply that his argument is sound since his philosophical targets are actually guilty
of engaging in cheap tricks even though this is not the only strategy available for defending an etiological
123
Synthese
Fig. 2 (Reproduced from Piatigorsky (2007), Figure 1.3, p. 9)
4 Evolutionary grounding of biological functions without “cheap tricks”
4.1 Reply to cummins’ objection (2) to the weak variation of neo-teleology (WNT)
From the examples of new functions arising through the processes of gene dupli-
cation and gene sharing, we have seen that a novel function established even at the
molecular level of biological organization can bring forth a new, salient, function
for highly complex multi-cellular individual organisms.8The next step in countering
Footnote 7 continued
account of functions. However, it seems that this is not an answer available to Cummins since he claims
that natural selection does not play a role in producing functions, so a more careful strategy without cheap
tricks seems impossible regarding his account. We thank the reviewer for pointing out this possible strategy
for Cummins.
8It could be objected that we cannot simply claim that trivariance occurring at the photoreceptor level by
itself results in trichromacy concerned with our ancestors’ feeding strategies. Firstly, by focusing on the
photoreceptor level, we are not ignoring certain adjustments that should have taken place within the human
eye in passing from the trivariance to the trichromacy in question (see e.g., MacLeod 2010). Secondly,the
occurrence of the third cone class in the case of trivariance could have been supported by already existing
higher-level mechanisms in dichromats. This is only to say that the described occurrence of trivariance in
dichromats, which, then, results in the emergent function of trichromacy as illustrated by Fig. 3, fits within
the basic explanatory scheme of positive Darwinian selection [for a clear statement of the basic scheme of
natural selection, see (Skipper and Millstein 2005)].
123
Synthese
a
a'
a''
Shared
Function
Emergent
Function
Of
Color Vision
Redundancy Class Selection
Fig. 3 An updated version of Thomas’ model for the mechanism of positive Darwinian selection of
‘emergent biological properties’
Cummins’ account of biological functions is to show that these new functions are not
evolutionarily neutral, but rather represent selected effects.
In order to demonstrate that evolutionary factors can contribute to the establishment
of new, salient functions and thereby defend the weak variation of neo-teleology, we
will analyze Thomas’ model for one of the ways in which genetic redundancy can
be positively selected in a population (see Thomas 1993). We will refer to the case
study of human color vision, which illustrates the issue raised by Cummins’ objec-
tion (2). The widespread view is that selection maintains the genetic redundancy and
Thomas discusses four possible mechanisms by which selection might act in order to
achieve that. We will focus on the mechanism concerned with the selection of emer-
gent properties, since this mechanism also applies to the case of human color vision.
However, first of all, we should examine why it is justified to accept the hypothesis
that positive selection acts to maintain genetic redundancy. In the case where two
perfectly redundant genes are present, the standard scenario would suggest that one
of these genes will be lost. The standard scenario entails neutral evolution as the null
hypothesis since under the assumption that a new allele comprising duplicate gene
is selectively neutral, it has a very small probability of being fixed in a population
(Zhang 2003). However, if we have a situation where two perfectly redundant genes
are kept for longer evolutionary times an additional hypothesis is needed to explain
how this situation could have occurred. The more likely explanation is that in these
cases selection acts to maintain both of these genes since the presence of both of them
brings advantage to the organisms bearing them.9
Figure 3represents an updated version of Thomas’ model for the mechanism
of positive Darwinian selection of ‘emergent biological properties’ (Thomas 1993,
pp. 397–398), such as, most notably, human color vision.10
9As already pointed out on p. 16 this explanation is further strengthened by evidence that these changes
in genetic traits brought about some important phenotypic changes.
10 Figure 3is updated by the data presented in Sect. 3, in particular, within the part referring to the
case of human color vision considered as an outcome of genetic redundancy, that is, as a case of neo-
functionalization via gene duplication. For other cases in which natural selection maintains genetic redun-
dancy in a population, see Thomas (1993, pp. 395–397).
123
Synthese
Redundancy class refers to a gene family containing gene duplicates with similar
function, which means that each of them alone can confer a wild-type phenotype
with that function. However, these duplicate genes together result in an emergent,
new function. To return to the color vision example, each of the duplicate genes code
for the opsin polypeptide sequence responsible for one of the three cone pigments.
The structure and function of each of these pigments is very similar, but with a slight
difference in absorbance spectrum. The gene family (a)–(a”) performs the shared
function of enabling trichromatic vision in our species and some other closely related
species as well. Hence, the explanation of the redundant genes is that they were selected
because of the emergent property of color vision.
Consider at this point a situation in which an individual organism belonging to a
certain hominoid population gains gene (a”) by one of the mechanisms of gene dupli-
cation mentioned above. When confronted with other organisms—especially taking
into account intra-specific relationships—that have dichromatic vision, in an environ-
ment favoring color vision as a feeding strategy, selection acts upon trichromatic color
vision in humans as on the so-called “emergent function” (see Fig. 3).
A more general idea behind this kind of strategy is to show that a careful analysis of
the process of neo-functionalization demonstrates that situations with salient functions
arising and spreading through populations as a result of selection pressures are actually
not uncommon. We argue that in those cases where there is a direct link between a
change at the molecular level that resulted in change at the phenotypic level that is
beneficial for the organism, we are justified in concluding that positive selection can
be a major factor in fixing the changes at the molecular level. The idea, on which
this more careful weak etiological selectionist account of functions is based, can be
now extended to traditional areas of the philosophical function debate, such as one
concerned with the physiology of human vision more generally. This extension is a
piecemeal process, far from the “cheap tricks” advanced by the strong variation of neo-
teleology, and finds in the molecular evolutionary models, such as Thomas’, the right
kind of grounding process required by Cummins’ evolutionarily neutral argument. In
sum, the weak variation of neo-teleology, especially when considered together with
the theories of neo-functionalization, offers a new option for a plausible restoration of
the etiological selectionist idea about tracking functions in the biological world.
This brings us to an interesting question: on what grounds can we claim that the
particular causal contribution made by a certain protein to the corresponding biological
system is not just a fortuitous benefit, but, rather, a kind of “programmed manifestation”
(Cummins 1975) or, moreover, to quote the etiological selectionist strand of the debate,
a“genuine adaptation”(Godfrey-Smith 1993)?
4.2 Fortuitous benefit or a programmed manifestation?
Suppose that the protein, thanks to its conformational motif, repeatedly makes the same
positive causal contribution to the ways in which the system of dynamical flux within
the cell’s nucleus responds to environmental cues. More importantly, the protein at
issue has been produced by the corresponding structural gene responsible for a certain
metabolic reaction; that particular conformational motif happens to be favorable for
123
Synthese
the system of dynamical flux, and other things being equal, it is repeatedly used.
However, the protein has been synthesized for a different particular causal contribution,
or even multiple contributions, to the same or distinct biological systems. Let us see
how we can respond to this standard issue on the basis of the two main theories of
neo-functionalization.
4.2.1 The solution based on the theory of neo-functionalization via gene sharing
The particular causal contribution made by the protein in question to the containing sys-
tem of dynamical nuclear flux is not just a fortuitous benefit, but, rather, a ‘programmed
manifestation’, because the protein’s production is regulated by the mechanism of gene
sharing. In other words, according to this theory of neo-functionalization, we consider
that causal activity as functional on the grounds that it is embedded within the reper-
toire of other causal activities of that protein. Nevertheless, all the causal activities
of the protein are determined to a certain extent by the corresponding mechanism of
gene sharing. Thus, the relationship to the mechanism of gene sharing secures that the
manifestation of the protein’s causal behavior has a distinctive programmed character,
as specifically required by Cummins’ seminal argument regarding functional analysis
(see Cummins 1975).11
Consider now the situation in which the novel protein’s function, added to the
already existing repertoire of the protein’s multiple functions, confers an incremental
advantage value to an individual organism bearing the corresponding mechanism of
gene sharing. As it is usually emphasized, the feature of incremental advantage value
points to the influence of positive Darwinian selective pressures (see, e.g., Sterelny
2006;Zhang 2003). Thus, in that particular kind of situation, which has been previously
exemplified by the case study of human color vision, we may correctly consider
the protein’s causal activity at issue as a molecular adaptation. However, as widely
recognized in recent molecular evolutionary biology (see Golding and Dean 1998;
Zhang 2003;Weinreich et al. 2006), there are many cases in which a protein shows a
distinctively ‘programmed’ character in its causal contribution to a biological system,
but it is selectively neutral. Now, on the account of biological functions defended in
this paper, which, accordingly, draws on the weak variation of neo-teleology and the
main theories of neo-functionalization, we may still consider that group of cases as
genuinely functional, even though they refer to evolutionarily neutral considerations.
The weak etiological account that we endorse accepts the claims of neutral theory of
molecular evolution that many changes at the molecular level are selectively neutral
and allows that even the products of neutral evolution can bring forward a new function.
11 As known, there are some important difficulties concerning Cummins’ notion of ’programmed mani-
festation’ in relation to his accounting for functional behavior of biological items. We will not explore its
more general difficulties, as well as what would be an adequate framework for dealing with them. At this
point in the paper, we will instead track the influence of positive Darwinian selective pressures in justify-
ing biological functional claims. In particular, we will try to ascertain in what kind of situations, we may
correctly consider causal activities of biological items as being selected for manifesting these activities.
However, as we will summarize shortly, such kinds of situations do not exclude that, according to both
theories of neo-functionalization, we may have legitimate functional ascriptions by appealing to merely
neutral evolutionary considerations.
123
Synthese
However, we maintain that there are important cases where selection is the main factor
responsible for the fixation of certain traits. Before concluding on the prospects of this
account of biological functions, we will clarify how the theory of neo-functionalization
via gene duplication responds to the problem of the distinction between fortuitous
benefits and programmed manifestations, as this issue is standardly emphasized by
the function debate.
4.2.2 The solution based on the theory of neo-functionalization via gene duplication
The particular causal contribution made by the protein in question to the system of
dynamical nuclear flux is not just a fortuitous benefit, but rather, a ‘programmed
manifestation’, because the protein is synthesized by the corresponding structural
gene. That gene was, first, duplicated by a determined mechanism from its parental
gene and, subsequently, underwent mutational pressures within the gene family to
which the gene duplicate in question belongs. Suppose, furthermore, that the gene
duplicate appears in the gene family controlling an important pathway related to the
dynamical flux of nuclear proteins, and that at a certain point in evolutionary time
the mutated gene duplicate produces the protein that specializes for one particular
function. Now, the particular causal contribution of the new protein has a distinctively
‘programmed’ character, but how should we explain the fixation of the corresponding
novel gene within a certain population of individual organisms?
There are, essentially, two distinct explanatory models for this: (1) the Dykhuizen–
Hartl effect explanatory model, and (2) the selectionist explanatory model. The
Dykhuizen–Hartl effect describes the cases where an originally neutral mutation
later becomes advantageous due to the change in the environment [this includes
the changed biochemical environment that results from other amino acid replace-
ments in the same protein (Hughes 2003)]. According to the selectionist explana-
tory model, the fixation of the gene responsible for the production of the protein
performing a novel function can be explained by the selectionist models, such as
one worked out in Thomas (1993). In other words, in that kind of situation, we
can consider the ‘programmed manifestation’ of the new protein also as a ‘gen-
uine adaptation’ or, as interchangeably stated in the function debate, an ‘effect
that was positively selected’ at a certain point in evolutionary time. The fact that
the protein’s novel function confers an incremental advantage value could point
to the fact that here positive selection was at work, since under the assumption
of neutral evolution the probability of fixation of the duplicate is very low. How-
ever, since there is no clear phenotypic effect we cannot, without further evi-
dence, conclude with certainty that positive selection is responsible for this fixa-
tion.
However, even the neutralist models, such as Dykhuizen–Hartl model, are included
in our weak etiological account since according to it a trait has a function if past
tokens of that trait contributed to the fitness of organisms bearing that trait. In the
Dykhuizen–Hartl model an originally neutral trait at a later time in a different envi-
ronment becomes advantageous which means that it contributes to the fitness of the
organisms.
123
Synthese
5 Conclusion
We have argued in this paper that the function, or even multiple functions, of a biolog-
ical item is a particular causal contribution to the containing system. Furthermore, in
Sect. 4, we argued that this kind of causal contribution can be additionally character-
ized as a ‘programmed manifestation’, given the item’s embedding into the containing
system’s repertoire for responding to determined environmental cues.
As for the role played specifically by positive Darwinian selective pressures in justi-
fying biological functional claims—in contrast to both Cummins’ thorough evolution-
ary neutrality on this issue and the panselectionist pretensions in the other major strand
of the function debate—we have tried to point out that we can have an evolutionarily
laden philosophical account of biological functions without sweepingly invoking the
mechanism of Darwinian selection (recall here the analysis of the nuclear flux case).
However, in the particular kind of situation in which there is an “incremental construc-
tion of a fit between agents and their world” (Sterelny 2006, p. 146, italics added), the
mechanism of positive Darwinian selective pressures is directly responsible for the
repertoire embedding of the novel causal contributions (recall here the case of human
color vision).
Acknowledgments We wish to thank Thomas Reydon and two anonymous referees for their careful
reading of our manuscript and their insightful comments.
References
Buller, D. J. (1998). Etiological theories of function: A geographical survey. Biology and Philosophy,13,
505–527.
Cummins, R. (1975). Functional analysis. The Journal of Philosophy,72, 741–765.
Cummins, R. (2002). Neo-teleology.In A. Ariew, R. Cummins, & M. Perlman (Eds.), Functions: New essays
in the philosophy of psychology and biology (pp. 157–172). Oxford, New York: Oxford University Press.
Deeb, S. S., Jorgensen, A. L., Battisti, L., Iwasaki, L., & Motulsky, A. G. (1994). Sequence divergence of
the red and green visual pigments in great apes and humans. Proceedings of the National Academy of
Sciences of the United States of America,91(15), 7262–6.
Godfrey-Smith, P. (1993). Functions: Consensus without Unity. Pacific Philosophical Quarterly,74, 196–
208.
Godfrey-Smith, P. (2001). Three kinds of adaptationism. In S. H. Orzack & E. Sober (Eds.), Adaptationism
and optimality (pp. 335–357). Cambridge: Cambridge University Press.
Golding, G. B., & Dean, A. M. (1998). The structural basis of molecular adaptation. Molecular Biology
and Evolution,15, 355–369.
Harrison, P. M., Hegyi, H., Balasubramanian, S., Luscombe, N. M., Bertone, P., Echols, N., et al. (2002).
Molecular fossils in the human genome: Identification and analysis of the pseudogenes in chromosomes
21 and 22. Genome Research,12, 272–280.
Hughes, A. L. (1999). Adaptive evolution of genes and genomes. Oxford: Oxford University Press.
Hughes, A. L. (2003). The origin of adaptive phenotypes. Proceedings of the National Academy of Sciences
of the United States of America,109(36), 13193–13194.
Hunt, D. M., Dulai, K. S., Cowing, J. A., Julliot, C., Mollon, J. D., Bowmaker, J. K., et al. (1998). Molecular
evolution of trichromacy in primates. Vision Research,38, 3299–3306.
Kimura, M. (1983). The neutral theory of molecular evolution.Cambridge: Cambridge University Press.
Kimura, M. (1991). The neutral theory of molecular evolution: A review of recent evidence. Japanese
Journal of Genetics, 66, 367–386.
MacLeod, D. I. A. (2010). Into the neural maze. In J. Cohen & M. Matthen (Eds.), Color ontology and
color science. Cambridge, MA: MIT Press.
123
Synthese
Melin, A. D., Hiramatsu, C. Parr, N. A., Matsushita, Y. Kawamura, S., & Fedigan, L. M. (2013). The behav-
ioral ecology of color vision: Considering fruit conspicuity, detection distance and dietary importance.
International Journal of Primatology, published online 30 November 2013.
Mollon, J. D. (1989). “Tho’ she kneel’d in that place where they grew..” The uses and origins of primate
color vision. The Journal of Experimental Biology,146, 21–38.
Nathans, J. (1987). Molecular biology of visual pigments. Annual Review of Neuroscience,10, 163–194.
Nowak, M. A., Boerlijst, M. C., Cooke, J., & Maynard Smith, J. (1997). Evolution of genetic redundancy.
Nature,388, 167–171.
Ohno, S. (1970). Evolution by gene duplication. New York: Springer.
Piatigorsky, J. (2007). Gene sharing and evolution: The diversity of protein functions. Cambridge, MA:
Harvard University Press.
Rodríguez-Trelles, F., Tarrío, R., & Ayala, F. J. (2003). Convergent neofunctionalization by positive Dar-
winian selection after ancient recurrent duplications of the xanthine dehydrogenase gene. Proceedings
of the National Academy of Sciences of the United States of America,100(23), 13413–13417.
Rouquier, S., Blancher, A., & Giorgi, D. (2000). The olfactory receptor gene repertoire in primates and
mouse: Evidence for reduction of the functional fraction in primates. Proceedings of the National Acad-
emy of Sciences of the United States of America,97, 2870–2874.
Skipper, R. A., & Millstein, R. L. (2005). Thinking about evolutionary mechanisms: Natural selection.
Studies in History and Philosophy of Biological and Biomedical Sciences,36, 327–347.
Surridge, A. K., Osorio, D., & Mundy, N. I. (2003). Evolution and selection of trichromatic vision in
primates. Trends in Ecology & Evolution,18(4), 198–205.
Sterenly, K., & Griffiths, P. (1999). Sex and death: An introduction to philosophy of biology. Chicago: The
University of Chicago Press.
Sterelny, K. (2006). Memes revisited. The British Journal for the Philosophy of Science,57, 145–165.
Thomas, J. H. (1993). Thinking about genetic redundancy. Trends in Genetics,9, 395–399.
Verrelli, B. C., & Tishkoff, S. A. (2004). Signatures of selection and gene conversion associated with human
color vision variation. The American Journal of Human Genetics,75(3), 363–375.
Weinreich, D. M., DeLaney, N. F., DePristo, M. A., & Hartl, D. L. (2006). Darwinian evolution can follow
only very few mutational paths to fitter proteins. Science,312, 111–114.
Zhang, J. (2003). Evolution by gene duplication: An update. Trends in Ecology and Evolution,18, 292–298.
123
