Gould & Lewontin's Spandrels – a critical analysis.

In 1979 S. J. Gould and R. C Lewontin published The Spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist programme. It is usually considered quite an important paper, both from the theoretical and the historical point of view – a milestone of evolutionary biology. As we can infer by the title, it is a critical paper. Gould and Lewontin aimed at criticizing what they considered to be the prevailing interpretation of evolution of that time and its consequent scientific practice, especially in the Anglo-Saxon world: adaptationism. Gould and Lewontin attack adaptationism both from the scientific and the philosophical point of view. Cases and habits of bad scientific practice are addressed as well as problems regarding both the epistemology and the ontology of biology. Adaptationism, or rather, the adaptationist programme is at the same time a way of doing science and a set of theoretical assumptions and theses.

The paper is subdivided in six chapters: an introduction; an overview of the theoretical features of adaptationism and a focus on some cases of bad science; a defense of Darwin's pluralist approach; an overview of cases better explained with a non-adaptationist approach; the proposal for a different, more « European » approach.

The foundamental thesis of adaptationism is that the direct production of adaptation by means of natural selection is the primary (if not the only) cause of nearly all organic forms, functions and behaviours (see. pp. 150-151). Gould and Lewontin call adaptationism « the Panglossian paradigm ». The name comes from Dr. Pangloss, a character from Voltaire's Candide, who, to put it shortly, believed that nature gave us our noses so that we could wear glasses. In the same way, adaptationists believe that natural selection gave the Tyrannosaurus tiny front legs to titillate its female partner. The first hypothesis sounds absurd, but the second not so much. The latter is however neither self-evident nor necessarily the best kind of answer to the question “why does the Tyrannosaurus have tiny front legs?”. Adaptationist biologists think like those who might think that the spandrels of San Marco in Venice were put there to host the religious drawings, while they are actually bi-products of mounting a dome on rounded archs, only subsequently exapted for aesthetic reasons.ⁱ In other words, they fail to distinguish between the current utility of a trait and the reasons of its origin.

Before we proceed with the critical analysis of the contents of the paper, we should notice one thing. As Massimo Pigliucci and Jonathan Kaplan said:

« These assumptions, when applied to human behaviors, quickly yield conclusions with dramatic political and social implications. The attack on adaptationism expressed in Spandrels, and which Lewontin and Gould each pursued in many other works, would probably have been far less aggressive if the adaptive significance of the variation in the color of snail-shells were the only thing at stake »ⁱⁱ.

It's not just about biology as a science. Gould and Lewontin notice that we « do not impose biological biases upon » architectural features. In saying so, they admit that biological biases do exist. Anyone who is familiar with the history of scientific racism will immediately notice how biology is not just a branch of science. It is one of the ways we interact with and give meaning to the world. (Think of the classical definition of the human being as animal rationale: we're not just rational, we're animals too, and this makes all the difference.) In the best of cases, bad biology can have huge negative impact on public opinion.

Let's now look at the problematic features of the adaptationist programme. Instead of following the paper, I'll try to analyze it by topic. As I said, Gould and Lewontin attack adaptationists from many fronts (albeit in a non-systematical manner). I will try to go from more “down to earth” practical problems to more abstract theoretical ones.

According to Gould and Lewontin, the adaptationists' styles of argument are not completely honest. In the second chapter, they give some examples.

• « If one argument fails, try another. »

• « If one argument fails, assume another must exist. »

• « In the absence of a good adaptive argument in the first place, attribute failure to imperfect understanding of where an organism lives and what it does. »

• « Emphasize immediate utility and exclude other attributes of form. »

You'll surely have noticed how the main problem resides in the adaptationists' obstinacy about the kind of answer to be sought, that is to say, (quite tautologically) an adaptationist answer (“trait x exists because the organism needs it to survive”). Adaptationists, say Gould and Lewontin, have the bad habit of « telling stories », reverse-engeneering them to fit the available data. If one adaptive story is rejected, they will replace it with another one, without giving any chance to different kinds of approaches. Since, as they say, « the range of adaptive stories is as wide as our minds are fertile, new stories can always be postulated »ⁱⁱⁱ.

If we look more closely, however, this is not just a problem of stubbornness and intellectual dishonesty. Such a bad habit implies a specific epistemological position about biology itself. For the adaptationist way of doing science, the only criterion to assess a biological truth seems to be consistency with natural selection. « [T]he criteria for acceptance of a story are so loose that many pass without proper confirmation. »^iv Consistency and plausibility are necessary, but not sufficient. If we reduce biology to this, it's hard to see what its epistemic value is.^v

The adaptationist way of « telling stories » implies also that, in order to be compatible with natural selection, they won't accept any other hypothesis – and this is where the second problem resides. The adaptationist program offers a mystified and ossified view of darwinian evolution – and of Darwin himself!

In the fourth chapter Gould and Lewontin refute the depiction of Darwin as a « radical selectionist at heart who invoked other mechanisms only in retreat, and only as a result of his age's own lamented ignorance about the mechanisms of heredity »^vi. On the contrary, they cherish Darwin's « attitude of pluralism »^vii. (As proofs, they quote two passages, one from the last edition of the Origin of Species, one from a letter he wrote to Nature in 1880, in which he defended himself from accusations of being a « panselectionist ».) According to Darwin, natural selection is indeed the main means of modification of a species, but not the only on one. The number of external conditions affecting an organism and a population is high, and subsidiary mechanisms (like the « use and disuse of parts » Darwin talks about in his Variation of Animals and Plants under Domestication) can be crucial to understand the origin of the traits of an organism.

In their “panselectionism”, it's not just Darwin what adaptationists make a parody of – it's the whole theory of evolution – the whole darwinian project. Not only they're determined not to resort to any kind of non-adaptive story to explain the presence of a trait in a population – they will submit every trait to its own adaptive story, reducing the very relationship between traits in an organism (which should be considered in an “organic” manner) to a matter of competition and selection. Everything is selected.

One of the main theoretical targets of Gould and Lewontin's criticism is the atomization of the organism into traits, and the consequent explanation of every single trait as an indipendent « structure[...] optimally designed by natural selection for [its] function[...] »^viii. According to Gould and Lewontin, this schematization of the organism is not just a problem of method, nor a mere semantic problem of defining “what a trait is”. As their very vocabulary makes it evident, it is what we would call an ontological problem: « Organisms are integrated entities, not collections of discrete objects »^ix. (We will later look into their proposal in detail.)

What does it mean that an organism is “atomized into traits”? Every feature of an organism is considered as an indipendent thing, which, in a sense, just “happens” to be part of the same organism along with other traits. Every trait arrived in the present organism more or less indipendently via its own evolutionary history. (To be fair, apart from their favourite example, that of the chin, Gould and Lewontin don't provide many examples of fallacious atomization.)

When the adaptationist fails to show an explicit « part-by-part optimization », they will resort to the notion of trade-off. The idea is that an organism cannot optimize every trait, but it has to favour some at the expense of others. Adaptation is a matter of compromise between traits – the suboptimalilty of a part is explained by the optimality of another one. (Of course, the very notion of optimality stems from the habit of reverse-engeneering the origin from the current use, which is always assumed to be adaptive.)^x This (assumed) “limited capacity of optimization” of an organism is considered to be the only real constraint upon perfection of each trait.

Why is this atomization so problematic? First of all, in reducing an organism to a bundle of discrete properties, we may fail to see the nature of the relationship between those properties – especially if we end up identifying the property with its present adaptive utility. Moreover, there's no guarantee that the identification of a trait (its distinction from other traits) isn't arbitrary. What if, even evolutionistically speaking, a certain group of seemengly unrelated traits evolved together, for reasons which are hard to see because of a priori atomization? (We should note that, against the Spandrels, the process of atomization has been defended as crucial for biology!^xi) Second, atomizing an organism into traits in order to explain its current form by means of providing an adaptive story for each one of its traits implies that the real subject of selection and evolution is neither the organism (on the one hand) nor (on the other hand) the population or even the species. What is selected, and what ultimately evolves, is the trait itself. The organism is just a temporary “ontological host” for the trait, which is there just to do its job and make the organism survive and reproduce. It isn't difficult to see how this ontology of atomization is at the basis of the the selfish gene hypothesis by Richard Dawkins, which is one of the most influential adaptationists (recall that Dawkin's famous book was published in 1976, just three years before the Spandrels!). As Pigliucci and Kaplan noted, Dawkins himself was probably (one of) the implicit target(s) of the paper.

These two problems naturally imply another one, concerning the role of biology itself. Rasmus Nielsen noticed, biology is an essentially historical science too.^xii Could we explain the current state of Europe if we atomized the evolutionary histories of each nation? Probably not. Then, it's hard to see how biology can provide a deeper understanding of its essential subject – life. If all biology deals with is “traits” and “qualities” whose “organical organization” is accidental at best, a mere epiphenomenon at worst, then, what about life? (Abou that – we could argue that Richard Dawkins is especially responsible for expunging life out of biology.)

Now that we have covered some of the central issues with adaptationism, we can look at Gould and Lewontin's proposal. By contrast, our understanding of the limits and the flaws of adaptationism will be even more clear. In the fifth chapter Gould and Lewontin give an overview or cases in which it's possible to give a better account of traits and features of organisms in non-adaptive terms.

No adaptation and no selection at all. The phenomenon of genetic drift is probably the most notorious non-selectionist principle affecting the transformation of a species. It consists of « a kind of random genetic sampling error », which can lead to the fixation of some characters even for no particular reason. While acknowledging its existence, adaptationists usually downplay its role in evolution (mostly because its effects are most pronounced in smaller populations). According to Gould and Lewontin, on the contrary, genetic drift can heavily influence evolution in three non-adaptive ways.

• A trait is fixated for no particular reason at all.

• A (deleterious) trait is fixated in spite of natural selection.

• Selectively favoured traits can be “washed away”. Genetic drift, infact, « causes the immediate loss of most new mutations after their introduction »^xiii. It takes an absurdly long time for a new trait to be fixated.

No adaptation and no selection on the part at issue. This case is particularly important for Gould and Lewontin, as it includes pleiotropy (one gene → different phenotypes / same phenotype ← different genes), allometry^xiv (mathematical ratio between various parametres like weight, volume, size, anatomy, physiology etc. – see insects vs vertebrates), material compensation and mechanically forced correlations. Moreover, G&L provide an interesting example of how some morphological traits can be just the by-products of the past evolutionary & developmental history of the organism. It's the case of paedomorphism (reproduction happens in what was ancestrally a juvenille morphologic stage), which can occur in arthropod with a shorter life cycle selected for environment with abundant but ephemeral resources (es. a carcass). Since paedomorphism is not adaptive per se, it can be thought of as a by-product of a selective process regarding a different feature.

Decoupling of selection and adaptation. In particular

• Selection without adaptation (or even with dis-adaptation). A mutation which doubles the fecundity of individuals will easily get fixated, when resources remain finite, doesn't make the organism more adapted to the environment – on the contrary, it may exacerbate competition between individuals, or even attract more predators!

• Adaptation without selection (phenotypic plasticity). Many species of sponges, corals and other marine organisms have different phenotypical adaptations to their particular environment, while retaining the same genes. The environment itself shapes the phenotypical expression of an organism (← and that's why development is so important).

  Gould and Lewontin stress the importance of development in adaptation. In particular, they cite phenotypic plasticity during the ontogenesis and cultural adaptation (which is often reduced to genetic adaptation by sociobiologists).

Adaptation and selection but no selective basis for differences among adaptations. The same “selective question” can give rise to different “adaptive answers”, without one being selectively preferred to another.

Adaptation and selection, but the adaptation is a secondary utilization of parts present for reason of architecture, development or history. It's the case of the spandrels: the current utility alone doesn't say anything about the origin of a trait. Traits can arise because of architectural constraints and be exapted for newer uses.

We should notice how some of these cases don't actually provide examples of stories which are non-adaptive tout cour. Instead, the notion and the role of adaptation is complicated and interpreted to fit a more organic and integrated ontology of biology. This is the last important topic of our analysis. Gould and Lewontin propose what we could call an “architectural model” of life.

The organism, they say, has got a Bauplan. « [T]he adaptationist programme (atomization plus optimizing selection of parts) can['t] do much to explain Baupläne and the transition between them »^xv. The body plan of an organism is

« so integrated and so replate with constraints upon adaptation […] that conventional styles of selective arguments can explain little of interest about them »^xvi.

Gould and Lewontin's proposal does not deny that change is mediated by natural selection, but it holds that constraints determine the evolutionary pathways so strongly that they can be considered as the most interesting aspect of evolution. It's a bold thesis. Not only it shifts the attention from the selective process(*) of single traits (**) to the very evolutionary history (*) of a whole body-plan (**), but it almost relegates natural selection to the function of mediator.

According to Gould and Lewontin, life is essentially organization. The living world is crowded with patterns, schemes and architectures, which exist mostly because of the conditions of organization itself. Selection is a mediator, which can favour or discriminate certain schemes, but ultimately the real mover of evolution – its principle, its archè – is the scheme itself, which acts as a constraint for every possible change. The organizational schemes determine how the species can and cannot evolve, what trait an organism can and cannot have.

Constraints can be of phyletic nature. For example, our human body-plan is not optimal for upright posture, despite innumberable millennia of selection shaping our form. Our Bauplan evolved for quadrupedal life, and while today we walk on two legs, it has exerted its weight and influence on our whole evolutionary history.

Developmental constraints are even more powerful constraints. The ontogenesis of complex organisms is extremely refractory to evolutionary change, especially in its earlier stages. Embryos of different species of vertebrates, like humans, chickens, elephants and dolphins, all look quite similar – we have a common, as general as powerfully constraining, Bauplan. The process of differentiation of organs systems is so delicate that any error affecting its first stages can be fatal. What this means is that, not only it takes a lot of time for change in such basic structures to happen, but structures themselves are “integrated wholes”. They can't be theoretically atomized and pulled apart into traits because natural selection itself cannot do it. Every change of the Bauplan is an alteration of the developmental programme – in other words, of the ontogenesis itself.

In the end, Gould and Lewontin appeal to Darwin's pluralist approach as the only one who can really guide biology and let us appreciate the complexity of the history of evolution. The ontology underlying their proposal looks at Baupläne as ontological wholes. The body-plan of an organism, which summarizes both the evolutionary history and the ontogenetical development behind it, is its substantial form. (It sustains the organism, and it makes it behave like any other representative of its species.) Consequently, an account of the evolutionary changes in the substantial forms of life, can not brutally reduce them to abstract individual traits. It must be careful. In this perspective, such individual traits do not even exist.

Gould and Lewontin's aim was to reintroduce organisms into biology. According to Pigliucci and Kaplan on one hand, and Rasmus Nielsen on the other, they succedeed in it. Thanks to the developments in genetics, in the 40 years after the publishing of the Spandrels paper, biology achieved a fruitful synthesis of Gould and Lewontin's approach and the genetics-based adaptationism. As all constraints can be thought either in terms of genetics or epigenetics, constraints and selection can be considered the two foundamental deterministic principles guiding evolution.

« It is this synthesis of constraints (spandrelism) and selection (panglossianism) that is the key to a more sober and realistic understanding of phenotypic evolution. »^xvii

As « evidence of selection, and knowledge of the function of a gene, does not constitute evidence of adaptation »^xviii, “spandrelism” can mitigate adaptationism. At the same time, being strongly based on genetics, “panglossianism” can offer its rigour to the architectural view of life – not because parts and genes are the real protagonist of evolution, but exactly because of their mediating function.

Sialia sialis. Apparently, a very jelous bird.

i By the way, this architectural metaphor has been criticized, as a dome can be mounted on rounded archs by a number of different solutions. I think that, instead of being an effective counterevidence, this actually reinforces Gould's and Lewontin's pluralist approach!

ii See M. Pigliucci, J. Kaplan, The fall and rise of Dr Pangloss:adaptationism and the Spandrelspaper 20 years later.

iii Gould, Lewontin, Spandrels, p. 153.

iv Ivi, p. 154.

v One example of this way of doing science is Barash's studies (1976) on a supposed adaptive “anti-cuckoldry” system of the males of bluebirds (Sialia sialis). Barash's results are indeed « consistent with an evolutionary interpretation » (Barash, quoted in Spandrels, p. 154), but obtained by means of arbitrary experiments. After Gould's criticism (1978), Morton et al. (1978) repeated the experiments with some variations. Since they weren't able to « confirm […] that Barash's conclusion represent a widespread evolutionary reality », they just made up a different adaptive story, involving the female showing her loyalty instead of the male trying to defend his honour.

vi Spandrels, p. 155.

vii Ivi, p. 156.

viii Ivi, p. 151.

ix Ibidem.

x To be honest, adaptationists do recognize the existence of non-adaptive forces, such as genetic drift or allometry. They're however dismissed as soon as they are recognized.

xi See oikosjournal.org, WHY "THE SPANDRELS OF SAN MARCO" ISN'T A GOOD PAPER.

xii Rasmus Nielsen, ADAPTIONISM—30 YEARS AFTER GOULDAND LEWONTIN.

xiii Spandrels, p. 157.

xiv In particular, while allometric patterns are subjected to natural selection, according to Gould and Lewontine, some regularities among different species don't really need an adaptive story to be explained.

xv Spandrels, p. 160.

xvi Ibidem.

xvii Pigliucci, Kaplan.

xviii Nielsen.