The Extended Evolutionary Synthesis and causality in biology

I have just spent three delightful days at the Konrad Lorenz Institute for theoretical biology in Vienna, participating to a workshop of philosophers and biologists on the question of how to think about causality, especially within the context of the so-called Extended Evolutionary Synthesis, the currently unfolding update to the standard model in evolutionary theory (for more on the EES, see here).

The workshop was organized by my colleagues Kevin Laland and Tobias Uller, and hosted by the delightful Gerd Müller, an old friend of mine. In an informal sense, this was a follow-up to a meeting that Gerd and I organized at the KLI back in 2008, to explore the very meaning and conceptual boundaries of an EES, and which resulted in the publication of this book about the effort.

Below I am presenting the abstracts of the papers presented at the workshop, to give you an idea of the range of topics and a sense of the discussion. Kevin and Tobias are collecting the papers and will publish them in book form some time next year. So stay tuned…

Tobias Uller, Introduction: hereditary concepts and evolutionary explanations

The establishment of genetics as a scientific discipline turned the concept of heredity into gene transmission, a process independent of the processes that bring characters into being. Evolutionary biology has been strongly shaped by this conceptual separation of heredity and development. With heredity instantiated by a discrete channel of transmission of genes, subject to random mutation and blind copying, the adaptive match between organism and environment is possible only through consistent fitness differences among individuals. On this account, changes in phenotype through other forms of inheritance, such as epigenetic inheritance or social learning, are proximate causes that do not qualify as causes of evolution. This logic is frequently invoked against the claim that extra-genetic inheritance motivates a rethink of evolutionary theory. However, a clean separation of proximate and ultimate causes does not necessarily hold under alternative concepts of heredity. Here we give a brief overview of how our understanding of heredity affects our interpretation of causation in evolving systems.

​Armin Moczek, The shape of things to come: evo-devo perspectives on causes and consequences in evolution

The vast majority of traits studied by evolutionary biologists has been in existence for a far shorter time than the genes, pathways, cell fates, or morphogenetic processes needed to produce them during development. Yet mainstream evolutionary biology pays comparatively little attention to how this ancient and pre-existing developmental toolbox has shaped the outcomes of developmental evolution. In this presentation I explore how contemporary evo-devo research offers opportunities to expand our understanding of cause, effect, constraint on, and facilitation of innovation, diversification, and adaptation in evolution. I conclude that if evolution were a card game, conventional approaches would allow us to describe why certain hands win or lose, and the conditions under which one or the other may occur. Integrating evo-devo perspectives on causality in evolution would allow us to also understand why we have the cards and rules we do, and not others.

​Susan Foster, Incorporating the environmentally sensitive phenotype into evolutionary thinking: phenotypic plasticity mediates the relationship between selection and genotype

The genomic era has brought biologists a previously unimaginable wealth of data describing the genetic structure of populations and species, as well as detailing the nature and scale of polymorphism along the genome. Yet there has been a departure between expectations and the answers these data have actually provided in explaining the origin and evolution of the traits that matter most to biologists. This gap is attributable to the absence of a linear, deterministic relationship between genes and traits. In this manuscript we describe how better integrating the environment into our understanding of trait production can help to close the gap. We describe the ubiquitous role of the environment in shaping traits, outline the prevailing theoretical mechanisms by which such widespread phenotypic plasticity can determine evolutionary trajectories, and discuss the evidence for these mechanisms and how better incorporating them into evolutionary thought can help resolve longstanding controversies.

Sonia Sultan, Developmental plasticity: the evolving genotype and the emergent norm of reaction

“Some genotypes… may be more sensitive than others to environmental differences. So, to some extent the environmental variance is a property of the genotype. But the source of the variation is environmental and not genetic.” –D.S. Falconer, Introduction to Quantitative Genetics (3rd ed., p. 135)

Individual organisms will develop somewhat differently in different environmental conditions. These environmentally contingent patterns of trait expression or norms of reaction can be empirically characterized. Such plasticity studies reveal how particular genotypes and environments jointly determine phenotypes, and hence the adaptive variation on which selection acts. The above quotation (from Falconer’s foundational textbook) reveals the kind of perplexity that results from the effort to ‘disentangle’ genetic and environmental causes of phenotypic variation. Yet a Modern Synthesis framework requires that we do so in order to understand ‘the genetic basis of adaptive evolution’. How deeply entangled are these two causal factors, and does that entanglement require a different approach?

When the norm of reaction is understood as a property of the genotype (or indeed as the genotype), developmental plasticity can be readily fitted into a MS framework: the genotype’s specified response pattern (rather than a single favorable trait state) evolves as an adaptation, due to selection in several environments in which different phenotypes are most successful. In this view, as Dewitt and Scheiner (2004, p.5) explain, ‘…the answer to the questions, ‘is variation plastic or genetic?’ is simple — it’s genetic,’ because the plasticity (the production of alternative phenotypes in particular conditions) is assumed to be entirely genetically controlled, and hence no different from other stably inherited, gene-based adaptations. This is how plasticity was incorporated into quantitative-genetics models in the 1980’s and 1990’s without destabilizing prevailing ideas about evolutionary cauasation. For instance, Via and Lande (1985) and a number of subsequent models built on Falconer’s 1952 approach of modeling the norm of reaction as two different, genetically correlated characters expressed in two environments. The idea of genotypic norms made sense because norms of reaction clearly evolve: they differ among genotypes and on average among populations and related taxa, and the often adaptive nature of plastic responses makes clear that they have been shaped by natural selection.

As long as norms of reaction are viewed as genetically scripted response programs, developmental plasticity can be accommodated within the MS scheme. This requires only the conceptual step of ascribing environmentally contingent gene outcomes to genes (as is done for any type of phenotypic trait under this scheme). When this step is taken, as Dawkins remarks (1982, p. 23), the causal contingency of phenotypic outcomes need not pose a challenge to ‘neo-Darwinian’ evolution, as this process simply consists of the replacement of genetic alleles by other alleles with more favorable contingent effects. The organism’s repertoire of environmental responses is simply one type of ‘extended phenotype’ emanating from its genes.

However, recent insights into inherited environmental effects challenge the genotypic norm of reaction, and in doing so raise questions about evolutionary causation. The norm of reaction expressed by a given genotype in response to specified environments is not in fact determinate: the genotype’s immediate responses to alternative environments are conditioned by inherited effects of parental, grandparental, and possible ancestral environments. The precise combination and sequence of previous environments differently shape developmental response patterns in the present generation, via directly inherited maternal and paternal factors and variably persistent epigenetic modifications to the genome. Moreover, the developmental impact and dynamics of these inherited environmental influences do not act as independent causes but are influenced by genotype– hence the notion of entanglement.

In this light, the norm of reaction can more accurately be seen as an emergent entity that results from these interacting, entangled genetic and environmental influences. In quantitative genetics terms, phenotypic outcomes do not result from internally scripted genotype by environment (GxE) interactions, but from far more complex interactions between genotype, immediate environment, maternal environment, paternal environment, and the epigenotype that results from several previous environments, depending on molecular dynamics and persistence of epigenetic marks. How can we conceptualize and empirically study this emergent norm of reaction as the basis of adaptive phenotypic variation, and how does this entity evolve?

Sonia Sultan talking at the KLI

​Kevin Laland, Understanding Niche Construction as an Evolutionary Process

Traditional analyses of evolution largely ignore the agency of organisms. Niche construction theory (NCT) aims to fill this gap by treating niche construction (NC) as a process within evolutionary biology. Most analyses of evolutionary causation restrict evolutionary processes to those directly changing gene frequencies (e.g. selection, drift, mutation, gene flow). NCT is reliant on a broader interpretation of evolutionary causation. However, the claim that NC operates as an evolutionary bias has generated contention among scholars of evolutionary biology. More empirical and theoretical work is required to clarify the roles that NC plays in evolution, and to evaluate the extent and significance of NC as bias. While some evolutionary biologists and evolutionary ecologists have accepted it as a useful perspective, NCT has primarily proven productive outside of evolutionary biology. Empirical work is underway that may consolidate the evidence that NC operates as an evolutionary bias, whilst theoretical approaches to this question are also tenable. NCT is likely to remain contentious within the evolutionary mainstream until such a time as NC is shown to operate as an evolutionary bias.

​Renee Duckworth, Integrating across multiple scales to understand evolutionary processes

In 1992, Simon Levin published a now classic paper, “The problem of pattern and scale in ecology.” In this paper, he proposes that this problem is the “fundamental conceptual problem in ecology, if not in all of science.” Here, I will take up his assertion by discussing how the problem of information transfer across scales applies to evolutionary biology. Patterns in nature either emerge from the collective behaviors of smaller scale units or are imposed by larger scale constraints. This realization has important implications for resolving the debate about proximate and ultimate mechanisms in evolution because it allows us to properly place both natural selection and genes in the hierarchical order of life. In this view, natural selection is not a causal agent of evolution but rather a higher level constraint that acts as a filter, with the potential to reduce the range of phenotypes observed from one generation to the next. The genome, on the other hand, which Ernst Mayr placed solely in the category of ultimate causation, has a double role as both a lower level mechanism and a higher level constraint. One the one hand, within an organism it is a lower level entity whose behavior scales up to influence larger scale cellular and organismal processes. On the other hand, its inheritance across generations has the potential to act as a higher level constraint by defining the range of transcriptional behaviors that are possible at the genesis of each new organism. Most importantly, behavior drives the processes that form the patterns we seek to explain at every level of organization, whether this is the behavior of molecules that determine cell type or the behavior of individuals that produce community dynamics. Thus, by studying the behavior of lower level elements across multiple timescales we can understand, not just the patterns of evolution, but the actual processes of evolutionary change. Here, I illustrate various approaches to integrating across scales in evolutionary biology using empirical examples to show how such integration may be the key to resolving some of the major recent debates in evolution.

​Heikki Helanterä, A social insect view on the causal structure of a major transition

Insect societies are perhaps the best understood example of a major transition in evolution. I complement this knowledge with causal graphs to shed light on some current issues in the major transition literature and causal structures in social evolution in general. Finally, I outline the current understanding of insect societies as carriers of heritable fitness relevant variation and as members of Darwinian populations, and the potential they hold for shedding light on developmental causes in evolution in general.

​Arlin Stoltzfus, A new evolutionary cause, and its implications

A mutational (or developmental) bias in the introduction of variation may impose a bias on the course of evolution. This kind of causation was not part of the historic Modern Synthesis, and is not taught in textbooks. Its operation was not formalized until 2001. Preliminary results suggest that it is important across the tree of life, and may have an effect size on the order of the magnitude of the underlying bias. This kind of cause provides a formal basis to recognize within evo-devo and structuralism, not merely an ”alternative narrative,” but a causal theory based on a repertoire of causes broader than that sufficient for neo-Darwinism. The theories that include this kind of causation cannot be understood with the ”forces” conception of population genetics, which contemplates mutation as a mass-action pressure, but not does support correct reasoning about a point process that introduces novelty.

Richard Watson, The necessity of developmental plasticity, niche construction and soft inheritance for Darwinian evolution

A simplified model of biological evolution begins by assuming that inheritance occurs by genetic transmission only, phenotypic variation is genetically determined, and selection is a passive filtering process. Whilst various extensions such as soft-inheritance (including epigenetics and parental effects), environmentally-sensitive phenotypic plasticity and niche construction can be important in evolution; the standard model states that they are not essential to the evolutionary process. There is some justifiability to this perspective if the components of the standard model are sufficient for evolution and/or causally prior, i.e. that adaptation provided by the standard model preceded and refined phenotypic plasticity/soft inheritance/niche construction, and not vice versa. However, from the perspective of the major transitions in evolution (e.g. from unicellular to multicellular life) this causal priority is questionable. We argue that the origination of Darwinian individuality requires phenotypic or reproductive plasticity and niche construction (and either soft inheritance or reversible plasticity) at the particle level to generate heritable fitness differences for the collective over and above those of its components. Further, we argue that these mechanisms are intrinsic to maintaining Darwinian individuality at the new level of biological organisation and cannot be eradicated by subsequent selection. We briefly discuss how this perspective bottoms-out — i.e., is a standard model on primitive units required to get everything started, or conversely, could the extensions alone be enough to initiate the first Darwinian units.

​Lynn Chiu, Niche construction and natural selection

My goal is to challenge the idea that a casual process is evolutionary only insofar as it is on par with natural selection, such that (1) by itself it directly alters gene/trait frequencies or the degree of fit between organisms and their environments, and (2) it works by going against or adding to the effects of selection. This restriction on what counts as an evolutionary process has created unnecessary obstacles for those (rightly) advocating for the evolutionary importance (and centrality) of niche construction and developmental processes.

In this paper, I develop an account that better clarifies the relation between the evolutionary causes of fitness differences and niche construction, as portrayed by Niche Construction Theory (Odling-Smee et al. 2003 and beyond) and others (Walsh 2015, Sultan 2015), to help relieve these undue burdens. I argue that niche construction and fitness differences are both necessary components of evolutionary processes. Niche construction is defined as the ability of all organisms to determine the content and relevance of their biotic and abiotic environmental elements and the effective impact of this experienced environment on their own development, survival, and reproduction as well as that of others. Fitness is the total impact of the effective environment and other factors on survival and reproductive rates. Niche construction and fitness differences (as well as other evolutionary causes such as chance, migration, mutation) are parts of the same evolutionary process that together bring about evolutionary change (e.g. alterations to gene frequency or degree of fit). The account I will develop can resolve some common objections against evolutionary theories of niche construction and show why niche construction as an evolutionary cause is as important (though in different ways) as differences in fitness.

​Johannes Jaeger, Dynamical systems modeling and the Extended Evolutionary Synthesis

Much of the controversy over the Extended Evolutionary Synthesis (EES) revolves around three fundamental questions: (1) Does current evolutionary theory need an extension? I think we can answer this unanimously with a big resounding “yes!” Evolutionary theory always has been and is constantly being extended. (2) What direction should such an extension take? Here, I will follow other authors who have argued that the main aim of the EES is to overcome Mayr’s false dichotomy between proximate and ultimate causes, to provide unified causal explanations of evolutionary dynamics. In contrast, some critics argue that detailed causal explanations of proximate processes are neither necessary nor achievable and lie outside the proper scope of evolutionary theory. This argument is less about the EES than about subjective definitions of what evolutionary theory actually is and what is technically feasible in current biology. I am not particularly interested in either of these questions.

Instead, I will focus on (3): is the EES really a theory? Some critics assert that it does not qualify, since it only consists of a loose collection of diverse concepts and approaches, which do not provide an overarching deductive theoretical framework. But neither does existing evolutionary theory! The Modern Synthesis itself consists of an aggregation of more or less local models, unified by a common epistemological goal: to explain evolutionary dynamics through genetic change at the population level. I argue that the EES also has a shared epistemic goal, different from that of traditional evolutionary theory: it aims at providing causal-mechanistic explanations of the complex processes that originate and generate phenotypes, and how those processes affect the traditional population-level mechanisms of evolutionary change. In other words, the EES is structured around the central problems of novelty and evolvability in biological lineages that not only react to but also shape their ever-changing environment.

​Samir Okasha, Causation in biology

Biology, as everybody knows, is the study of living organisms. Modern biology is divided into a large number of sub-disciplines, each with its own explanatory agenda, research protocol and specialized vocabulary. These range from molecular biology and biochemistry, which study the molecular and chemical basis of life, to cell biology, embryology and ‘whole organism’ biology, which study life at higher levels of organization, to ecology, paleontology and phylogenetic systematics, which study life over extended spatial and temporal scales. Given this heterogeneity, it might be wondered whether anything useful could be said about causation in biology that would not be so general as to apply to every other science too. Though natural, this sentiment is in fact misplaced. The concept of causation does raise distinctive problems in a biological setting, at least some of which find no parallels in the physical sciences. This is particularly true of causal concepts as they feature in evolutionary theory and genetics, which will be the focus of attention here.

Restricting our focus to evolutionary theory and genetics may seem unusual, given the wealth of other biological sub-disciplines, but it is a natural choice for at least two reasons. Firstly, Darwinism is arguably the grand unifying theory in modern biology, much as Newtonianism was the grand unifying theory in 18th and 19th century physics. (The Russian-American biologist Theodosius Dobzhansky wrote in 1967 that ‘nothing in biology makes sense except in the light of evolution’, a statement which still holds true today.) So philosophical problems that arise in evolutionary theory (of which genetics is an important part) are in a sense problems for the whole of biology; and causation is the source of many such problems. Secondly, there is an intimate historical link between evolution and genetics on the one hand, and statistics and the methodology of causal inference on the other. Figures such as Francis Galton, Karl Pearson, Ronald Fisher and Sewall Wright were intimately involved in both enterprises.

​Jun Otsuka, From population to causal thinking

Kuhn (1962) pointed out that every scientific paradigm comes with its own ontological assumption or “world view” that determines its way of understanding phenomena and of conducting research. Here I identify Mayr’s population thinking as the canonical ontology of the Modern Synthesis, which both shaped the philosophical discussions in the 20th century and also created conceptual puzzles such as tautologism, historicism, and gene-centricism. Despite its wide popularity, my view is that population thinking is seriously flawed: the full-fledged ontology of evolutionary theory must acknowledge causal structures as a “type” to ground inductive reasoning. Taking causal structures as ontological units of evolution besides genes leads to a two-story view of evolutionary theory, consisting of the study of gradual, micro-scale changes in genetic/phenotypic frequencies on the one hand and that of discrete, macro-scale changes in causal structures on the other.

​Karola Stotz, Understanding the role of biological information in development and evolution

This paper aims to elucidate how biologists use and understand biological information. To this end it follows several subgoals: 1. Relate information talk in biology to causal specificity. 2. Show through case studies that information in biological systems plays several distinct roles. 3. Use an information-theoretic approach of causation to clarify the explanation of informational phenomena in terms of specificity. 4. Distinguish between three conceptions of causal specificity fine-grained control, one-to-one affinity, and limited but precise control, which are employed by biological systems to fine-tune different biological processes. They are achieved via distinctive mechanisms that employ both informational or sequence and conformational specificity. 5. Prove that causal specificity is distributed between genetic (both coding and non-coding) and non-genetic (epi- and exogenetic) sources. 6. Argue for the relevance of proximate sources of specificity as a cause in evolution.

​Denis Walsh, Population Thinking: The Higher Order Effect Perspective

Darwin’s discovery of natural selection was inaugurated by a simple shift in perspective, that Ernst Mayr (1975) has dubbed ‘population thinking’. Darwin realised that the problem of explaining the fit and diversity of organic form should be approached as a question about the constitution of populations. Rather than ask how individual organisms come to acquire their remarkable features, and their exquisite adaptedness to their conditions of existence, we should ask how populations come to comprise such individuals. With this simple change of view, and a rudimentary grasp of the principles of population change, Darwin was able to see that the daily activities of organisms’ lives suffice to explain the array of ‘endless forms most beautiful and most wonderful’. There is no need to invoke nonsubstantival forms, entelechies, vital forces, or a provident designer. All the theoretical apparatus we need is already to hand. Though this is a simple conceptual manoeuvre, it ushered a seismic change in the understanding of the fit and diversity of organismal form, that Mayr is quite right to acclaim. It isn’t just Darwin’s theory of evolution that exploits population thinking. Its successor, the Modern Synthesis theory of evolution, has as its theoretical core a highly abstract sophisticated version of population thinking, in which evolutionary change is described in terms of processes—selection and drift — that operate over ensembles of trait or gene types. When populations vary in the growth rate of their trait types, they change in predictable ways.

Simple and powerful though it is, population thinking appears to have some unfortunate consequences. While promoting the idea that the study of evolution is properly the study of the dynamics of populations, population thinking has presided over the growth of an invidious, or at least dismissive, attitude to the role of organisms in evolution. Organisms were routinely marginalized from the study of evolution throughout the 20th century as it grew under the auspices of the Modern Synthesis. The neglect is hardly surprising. The signal virtue of population thinking, after all, is its capacity to explain the dynamics of a population without adverting to the features of the individuals of which the population is composed.

It seems obvious to many evolutionary biologists, however, that the Modern Synthesis is impoverished by its penchant for ignoring the contributions of individual organisms, and the organism-level processes of development, ecology, behavior, learning, immunity, genome regulation (to name a mere few). Since the turn of this century certainly, but prior to that too, evolutionary biology has witnessed a concerted attempt to revive the organism. Those who advocate an extension, revision, or wholesale overthrow of the Modern Synthesis are motivated by the shared conviction that what organisms do makes a difference to how populations evolve. Calls to extend the synthesis typically cite the significance of the individual-level processes that occur within development and inheritance, and between organisms and their environments, to the process of evolution. These are the very processes that whose exclusion from Modern Synthesis evolutionary thinking has evidently been licensed by population thinking.

​Arnaud Pocheville, A Darwinian dream, and a few whimsical nightmares: on time, levels, and the separation of processes in evolution

The core, new idea of Darwinism is that natural selection explains adaptation (and is the only known naturalistic explanation so far). Modern challenges to evolutionary theory are rooted in this core idea. Adaptation of development through natural selection makes appeal to an implicit time-scale separation of development and natural selection (operating on almost immutable hereditary entities determining or constraining development). Niche construction and non-genetic inheritance call this (and other) time-scale separation into question. If niche construction is not separable from natural selection, development (qua dynamics) cannot be explained by the natural selection of a hereditary program anymore. The dynamical implications are evident. The implications for adaptation are more ambiguous.

Massimo Pigliucci, How do biologists and philosophers talk about causality?

As a scientist (evolutionary biology) and a philosopher (of science) I have long been interested in the relationship between the two disciplines (Pigliucci 2008). This is an issue of importance both in terms of the two academic disciplines themselves, and in the broader context of how they are perceived by the general public — with further practical implications, obviously, in terms of funding of scholarship, faculty positions, and studentships.

In this paper I will use several examples concerning discussions on the nature of causality, particularly as it applies to biological systems, as a conduit to explore how the practices of science and philosophy of science appear to be related, and then to discuss, in a sense, how they ought to be related. Hopefully, this exploration will also result in a clarification of a number of points about the philosophy of causality as it relates to its applications to science. I will begin by briefly presenting four possible models of the relationship between science and philosophy of science. I will then examine four instances of philosophical discussions of causality with respect to their relevance — or lack thereof — to the practice of science. And I will finally attempt to draw some general conclusions in order to move forward both the debates about causality and the one about the relationship between science and philosophy of science.

(The slides for my talk are here.)

The gang relaxing on the Danube, after the workshop