2. The structure of the Origin

The structure of the Origin can be divided into three main parts.Footnote ³ The first part (chapters 1–5) presents the theory of natural selection as a cause of evolutionary change. In the first chapter artificial selection is discussed, leading to the conclusion that modification within a species is in principle possible; chapters 2–4 show that an analogous process can exist in nature. Chapter 2 concerns the variation that exists within natural populations and chapter 5 presents various laws of variation. Chapter 3 concerns the struggle for existence – here the ‘doctrine of Malthus’ is applied to nature. In chapter 4 Darwin draws on the discussions of the previous chapters and shows that: (i) the existence of variations within populations and the doctrine of Malthus lead to the selection of favourable variations; (ii) if these favourable variations can be inherited (what he calls the ‘strong principle of inheritance’), they remain in the population and are further modified. Ultimately, if this process is repeated over many generations, it leads to the creation of complex adaptations of organisms to the environment, the extinction of many species, as well as the continuous divergence of organismic characters (what Darwin calls the ‘principle of divergence’).

In the second part of the Origin (chapters 6–8) the main difficulties for the theory of evolution by natural selection are discussed, for example how the transition from simple to complex structures takes place, as well as the difficulties posed by instincts and hybrids.

Finally, in the third part (chapters 9–13) the evidence in favour of the theory of common descent is presented, which includes the geological succession of organisms and their geographical distribution, as well as morphological and embryological affinities. It is argued that these observations only make sense on the theory of common descent, while on the alternative of ‘special creation’, they are left unexplained.

Based on the above, two important observations can be made regarding the structure of the Origin: first, the hypothesis of common descent (and thus the hypothesis that evolution has occurred), as well as the kinds of evidence that support it, are presented in the last part of the book, while the theory of natural selection, which is the cause of evolution, is introduced at the beginning; second, in the last part of the book Darwin seems to contrast the theory of special creation to the theory of common descent, and not to the theory of evolution by natural selection. A satisfactory analysis of the Origin, i.e. an explanation of why Darwin structures his presentation of the theory in this way, has to explain why the structure of his ‘long argument’ has these seemingly odd features.

3. The Herschelian strategy

Jonathan Hodge (Reference Hodge1977, Reference Hodge and Ruse1989, Reference Hodge2013; see also White et al. Reference White, Hodge and Radick2021) has offered a well-known analysis of Darwin’s strategy in the Origin that has been widely accepted. According to Hodge, the Darwinian ‘long argument’ for the origin of species by natural selection has to be understood in terms of the establishment of natural selection as a Herschelian ‘true cause’ (vera causa). According to Herschel, to establish a vera causa one must begin with cases where the action of the cause has been observed; such cases according to Darwin are examples of artificial selection, which he presents in the first chapter of the book. From these cases Darwin proceeds to cases in which the action of selection must be inferred; this is done in chapters 2–5, where the conditions that occur in nature that permit the action of natural selection – variation and struggle for existence – are examined. Next, according to the Herschelian strategy, it must be shown that the true cause is capable of producing the phenomena. This is done in part in chapters 1–5, and also in chapters 6–8 of the book where the difficulties of the theory are discussed. Finally, it must be shown that the true cause was indeed responsible for producing the phenomena. This is done in the remaining chapters 9–13.

Hodge’s analysis explains why the theory of natural selection is presented at the beginning of the Origin. However, there remains a puzzle, as such an analysis does not explain the fact that in the last part of the book the theory of natural selection does not play a central role. This point, i.e. the fact that Darwin seems to neglect the theory of natural selection in the last part of the book and instead focuses on the contrast between the theory of special creation and common descent, has led scholars to the position that Darwin’s ‘long argument’ is not actually a single argument, but has two parts: it consists, first, of a Herschelian argument for establishing natural selection as a true cause, which is presented in the first part of the book, and second, of an argument based on Whewell’s consilience of inductions and which mainly concerns the hypothesis of common descent and is developed in the second part of the book – see Waters (Reference Waters, Hodge and Radick2003) for development of this view.

4. Sober’s analysis

Elliott Sober’s (Reference Sober2011) analysis sheds further light on the structure of Darwin’s argument. Sober argues that the Herschelian strategy alone cannot explain the structure of the Origin and that Darwin’s book is in a way written backwards. He starts with the observation that Darwin’s theory consists of two distinct parts: the hypothesis of common descent and the hypothesis of natural selection. These two hypotheses, according to Sober, are not only logically independent of each other, but also have an asymmetric relationship to the evidence that supports them. In particular, the evidence that supports the hypothesis of common descent does not depend on natural selection having actually acted. In contrast, specific claims about the action of natural selection rest on claims about the common descent of organisms. Thus, the hypothesis of common descent has, according to Sober, evidential priority.

To support his claim Sober relies on what he calls ‘Darwin’s Principle’ (Sober Reference Sober2011, section 1.3). According to this principle, adaptive similarities between organisms provide almost no evidence for common ancestry, while similarities involving traits that are adaptively useless or even harmful to the organisms that possess them provide strong evidence for common ancestry.

Sober analyses ‘Darwin’s principle’ by means of the Law of Likelihood. According to this law, for hypotheses H1 and H2 and observation O, observation O supports hypothesis H1 over hypothesis H2 when P (O | H1) > P (O | H2) (where P (A | B) is the probability of A given that B has occurred). The expression P (O | H) is called the likelihood of hypothesis H, where the likelihood of H differs from the probability of H, P (H | O). The previous inequality therefore shows which of the two hypotheses, H1 and H2, is favoured by observation O. We can have a quantitative form of the law of likelihood by taking the ratio of the likelihoods:

$$P{(\rm O | H_{1})}\over P{(\rm O | H_{2})}$$

This ratio tells us the extent to which observation O favours H1 over H2. The law of likelihood expresses the intuitive idea that given an observation O and two hypotheses that are both compatible with observation O, the observation favours that hypothesis which makes the observation less unexpected.

The law of likelihood is related to Darwin’s principle as follows. The observation in this case is that two organisms A and B are similar in some respect (they both have trait T). The two hypotheses are, first, that the two organisms share common ancestry (CA), and second, that they have separate ancestry (SC for ‘special creation’, i.e. for the hypothesis that species have been created separately, which is the hypothesis that Darwin compares to the hypothesis of common descent in the Origin). Darwin’s principle concerns two cases: first, the case where the similarity concerns an adaptive trait; second, the case where the similarity concerns a non-adaptive trait:

$${P{(A \;\& \;B \; {\rm have \; T} \, | \,CA)}\over P{(A \;\& \;B \; {\rm have \; T} \, | \,SC)}} \approx 1, {\rm if \;T \; is \; adaptive}$$

$${P{(A \;\& \;B \; {\rm have \; T} \, | \,CA)}\over P{(A \;\& \;B \; {\rm have \; T} \, | \,SC)}} \gg 1, {\rm if \;T \; is \; not \; adaptive}$$

For example, consider the hydrodynamic shape of dolphins and sharks. In this example, which involves an adaptive feature in these two organisms, the likelihood ratio is approximately 1. But now, consider the example of gill slits in human embryos and fish embryos. In this case, the similarity concerns a non-adaptive trait (since gill slits in human embryos are considered to be non-adaptive). The likelihood ratio in this example is much greater than 1.

Sober claims that Darwin’s general argument in favour of common descent rests on ‘Darwin’s principle’ and can be analysed in terms of the above likelihood ratios. The crucial point is that Darwin’s principle does not presuppose the truth of the hypothesis that natural selection is the main cause of evolution. Specifically, what the principle presupposes is that (1) if natural selection favoured the evolution of trait T in the lineages leading to species A and B, then the fact that A and B both have trait T provides little or no evidence as to whether CA or SC is true. At the same time, Darwin’s principle does not presuppose that (2) natural selection has favoured the evolution of one or more traits in the lineages leading to species A and B. Therefore, the inference from the observations to the hypothesis of common descent does not rest on the claim that natural selection has occurred. Sober also argues that the opposite is true: specific hypotheses about the action of natural selection in specific lineages rest on the assumption of common descent.Footnote ⁴

Sober arrives at the following evidential structure of Darwin’s argumentation: first, Darwin infers from various kinds of evidence the common descent of organisms; it follows that populations can change to such an extent that speciation can occur; finally, Darwin shows (through the Malthusian argument, the analogy with artificial selection, and many examples of adaptive traits) that natural selection explains adaptive traits. And yet, the structure of the Origin follows the reverse order; this is why Sober claims that in terms of the evidential priority of the two main components of the theory (common descent and natural selection) the book is written backwards. So, the order actually followed in the Origin is explained by Sober by the fact that Darwin decided to structure his argument on the basis of the causal priority between common descent and natural selection. The causal priority of the two components is the inverse of the evidential priority, since natural selection causes evolutionary change (leading to the branching structure of the tree of life and to common descent) and not the other way around.

5. A novel analysis of the structure of the Origin

I will now argue that an analysis of the structure of the Origin is possible which, while retaining the main features of Hodge’s and Sober’s analyses, does not have the somewhat paradoxical consequence that the book has been written (evidentially) backwards. At the same time, this analysis will help us understand why in the third part of the Origin natural selection does not seem to play a prominent role. The main idea of the account I propose is that the evidence supporting the theory of common descent, presented in the third part of the book, can satisfactorily support the theory only when the theory of natural selection has been considered as an in principle possibility. Thus, first, Sober’s central idea of evidential priority is retained, in the sense that the inference of common descent does not presuppose the truth of the theory of natural selection, and second, it becomes clear that the theory of natural selection is presented first not simply because it has causal priority, but because only then can the argument in favour of common descent become fully plausible.

That this is the main reason why Darwin structures his book in this way can also be seen from the following passage from his Introduction:

Darwin’s argument in the above passage can be reconstructed, on the basis of the previous discussion, as follows. We have at our disposal various kinds of evidence (morphological, embryological, biogeographical, geological) that support the hypothesis of common descent. These observations are much less unexpected under the hypothesis of common descent, compared to the hypothesis of special creation. At the same time, however, there exists a great epistemological obstacle to accepting that all kinds of organisms form a single evolutionary tree; this epistemological obstacle consists in the fact that the complex adaptations and co-adaptations of organisms cannot be explained, as it is evident that no existing theory is satisfactory in this regard (Darwin cites as examples of unsatisfactory causal hypotheses for the explanation of adaptations the action of external conditions, the habits of organisms, and the will of organisms). Thus, for the inference from the evidence to the hypothesis of common descent to be satisfactory, one has to explain how evolutionary change is possible in such a way that complex adaptations and co-adaptations can be produced.

It is crucial here that what is missing to make the inference satisfactory is not a detailed explanation of ‘how the innumerable species inhabiting this world have been modified, so as to acquire that perfection of structure and coadaptation’. Such a thing would have been in any case impossible. Rather, what is needed is to understand how adaptations are in principle possible. And this is Darwin’s main goal, in my opinion, in the first part of the Origin. More specifically, the central claim of the analysis I propose is the following: in order for Darwin to successfully support the hypothesis of common descent, he has first to propose a plausible theory about how organismal adaptations are produced. The theory of natural selection is such a plausible theory: when this theory is accepted as an in principle possibility, we can have a very high degree of confidence in the hypothesis of common descent; the degree of confidence in the theory of natural selection is an independent matter (and can in any case be lower than the degree of confidence one has in the hypothesis of common descent).

The final epistemic state to which we are led by means of Darwin’s argument can also be seen from the following quotation, which is again from the Introduction of the Origin (the difference between ‘I can entertain no doubt/fully convinced’ and ‘convinced’ is very important here):

This difference between the degrees of confidence in the two main components of Darwin’s theory (common descent and natural selection) also explains, I think, further claims made by Darwin, which show that the theory of natural selection is of secondary importance to him, such as the following: ‘personally, of course, I care much about Natural Selection, but that seems to me utterly unimportant, compared to the question of Creation or Modification’ (from a letter to Asa Grey on May 11, 1863, emphasis added). It moreover explains the seemingly odd feature of the book we noted in section 2, i.e. why in the second part of the Origin the theory of natural selection does not have a prominent role. Finally, it makes epistemological sense of a very widespread attitude during the last quarter of the nineteenth century among biologists; while most of them were ‘fully convinced’ of the hypothesis of common descent, many were not convinced of the truth of the theory of natural selection (see Bowler Reference Bowler1992).

Let us close by briefly discussing the general epistemological significance of this analysis of the evidential structure of Darwin’s argument. For someone who adopts evidentialism, changes in what we are justified in believing are always associated with changes in the total evidence. But let us assume that the evidence in favour of evolution was known before the publication of the Origin (which is of course to simplify matters considerably, since Darwin presents new evidence in the Origin, e.g. biogeographical evidence). Then, if the degree of justification of the theory of common descent before and after the publication of the Origin is different, this must be because what seems to be the same evidence supports the hypothesis more strongly after the publication of the book. This change in the degree of justification of the theory of common descent is due to the introduction of the hypothesis of natural selection.

This argument illustrates a phenomenon that is analogous to something already known in epistemology: often the simple introduction of an alternative hypothesis H2 can undermine the degree to which the evidence supports hypothesis H1. The epistemic situation in the Origin is a bit different: the introduction of the hypothesis of natural selection increases the degree to which the evidence supports the hypothesis of common ancestry. Therefore, strictly speaking, the evidence in the two instances (i.e. before and after the publication of the Origin, on the assumption that the morphological, embryological, and other kinds of evidence in favour of evolution remained the same) is different, since, given evidentialism, the justification of what one is justified in believing supervenes on the total evidence. This means that, for the evidentialist, the total evidence for the reader of the Origin must be taken to include the hypothesis of natural selection as an in principle possibility.

Furthermore, the current account of the evidential structure of Darwin’s ‘long argument’ has consequences for a simple Bayesian analysis. Let us assume hypothesis CA (common ancestry) and hypothesis SC (special creation) that have certain prior probabilities, and let us also assume the morphological, embryological, and other evidence presented by Darwin. According to my analysis, the posterior probability of CA is greater for someone who knows the theory of natural selection than for someone who does not know the theory. But I argued that the theory of natural selection is in the first place posited as an in principle possibility, and its truth (just as Sober claims) is not presupposed for the hypothesis of common descent. Thus, the difference in the degree of belief in CA in the two instances (i.e. before and after the publication of the Origin, again on the assumption that the morphological, embryological, and other kinds of evidence in favour of evolution remained the same) is not due to some newly discovered evidence, and thus at least a simple Bayesian account of this case is incomplete.Footnote ⁵