Tinbergen’s Questions with Mechanistic Answers

Half a century or so ago, the ethologist Niko Tinbergen published an influential paper called “On aims and methods of Ethology” (1963), which, as the name suggests, discusses and distinguishes various as-pects in the study of animal behaviour.⁷³ It distinguishes four areas of

73 This distinction has since become the orthodox way to sort the different tasks of behavioural biology; see Manning 2005.

129

causal explanation in the study of behaviour. This distinction has be-come to be known as “Tinbergen’s Four Questions” or “Tinbergen’s Four Whys”, because it distinguishes between four different senses in which a question like “why is the male sparrow singing?” can be un-derstood. These senses are, according to Tinbergen:

1) Causation, which refers to the “behavioural machine” or mechanisms underlying the behaviour; for example:

What proximate causes make the sparrow sing?

2) Survival value (almost invariably called “function” in more recent literature; see Hogan & Bolhuis 2005 & 2009), which refers to whatever it is that the behaviour does for the animal that makes it better off (directly or through indirect consequences) in its natural environment: What is the singing for?

3) Ontogeny (or development), which refers to how the be-haviour and its underlying mechanisms appear in the course of the individual development of members of the species: How did the singing behaviour come about and change during the sparrow’s lifetime?

4) Evolution, which refers to the actual history of the trait:

How did the singing behaviour come to exist on the population level in the course of the history of the species and its ancestors?

As such, these questions are not precise enough to be four different explanatory questions. Rather, they are four different perspectives from which to frame explanatory questions about animal (and human) be-haviour, mapping different explanatory dimensions (as I have been call-ing them), each of them collectcall-ing a set of explanatory questions about the same behavioural phenomena. Furthermore, each of them con-tains a set of questions that differ in more than just what kind of (com-plementary) explanatory information they seek. One could even go as far as to say that the different explanatory questions within a dimension

130

ask about different parts of the same causal network and the answers are either complementary or competing, whereas the different dimen-sions simply ask different questions (Sherman 1988). However, the questions in different dimensions are connected through the subject matter: the answers to explanatory questions in any given dimension may sometimes depend on facts about explanations given in some other dimension (Mitchell 1992 & 2002; Pigliucci & Müller 2010). This also means that the formulation of an explanatory question in one di-mension may derive its motivation from needs in another dimen-sion.⁷⁴ I will discuss the four explanatory dimensions (mostly) from the point of view of evolution of social behaviour now, taking the evo-lutionary functionalist stance explicated above to relate other dimen-sions to the evolutionary dimension. The evolutionary function is the perspective that sets the relevant explanatory questions in the other di-mensions. The answers to these questions in turn affect the evolution-ary explanation. This perspective is needed for evolutionevolution-ary pur-poses, and it may be relevant for other purposes (for the reasons dis-cussed previously), but this is not meant to be a perspective-free sys-tematisation of how these questions are related.

3.1.1. Causation

Causation is a category that includes questions about both the external and internal immediate causal factors that trigger a certain behaviour, and about the mechanisms underlying these causal dispositions and participating in the causal processes that produce the external behav-iour. Questions about external conditions only (what proximate

74 In practice, questions that cross the boundaries between different research areas within biology are seldom in focus. When attention is paid to them, a whole new research program gets initiated sometimes. This has been the case, for example, in evolutionary developmental biology (Raff 1996; Hall 2003;

Müller 2007). The boundaries within a discipline are not that different from ones between them sometimes.

131

environmental factors cause the behaviour?) leave the details of the “be-havioural machine” black-boxed, concentrating on the interaction of the animal with its environment, to what stimuli it is responding in which ways. On the other hand, one could be interested in precisely this “machine”. This interest includes both comparative-psychologi-cal (cognitive-ethologicomparative-psychologi-cal) level questions (animal cognition and mo-tivational structures that are inferred from input-output relations in natural or artificial environments) and neurophysiological questions.

These, again, differ in kind from each other, and both can be ap-proached from both structural and functional perspectives. (See Ho-gan 2005; Shettleworth 2010.) A central question for the (proximate) causal explanation of behaviour is how to define what to explain: is the explanandum a singular behavioural occurrence (usually not) or a behavioural trait? If a trait, there are different ways to define it: all structurally similar behaviour (regardless of context), all behaviour in a given context, all structurally similar behaviour in a given context, or all behaviour that is produced by a given set of proximate causal factors. There is no unambiguous way to slice behaviour into traits in biology (see Lidicker & Freund 2009). What behaviour goes together as a trait depends on the explanatory interests. This is also the case with human behaviour (Longino 2013).

A further complication in explaining human behaviour is that we categorize both the behaviour and its causes as action guided by rea-sons and intentions in our folk-psychological practices. This practice individuates behaviour individualistically and (if given a causal inter-pretation) expresses its causes as well. However, if we approach be-haviour from the evolutionary functionalist perspective (as we should for evolutionary purposes), we should instead approach behaviour along the lines that I discussed with the example of the honey buzzard.

Various causes work in an orchestrated manner, forming a mechanism that produces a systematic behavioural pattern, and this pattern is iden-tified through what adaptively significant effect it achieves. From this perspective, internal and external factors are connected: the evolution of internal dispositions (perceptual, cognitive, and motivational),

132

when guided by selection, is such that the combination of dispositions and external stimuli (in typical environments) jointly cause the behav-iour that is functional in that environment. I will return to how to un-derstand this in human behaviour and how this relates to our com-mon way to understand human action in detail later.

A special case of environmental stimuli is those caused by other individuals. Social behavioural dispositions that result in (evolution-arily) beneficial outcomes without the individual processing the ends and means can be favoured over dispositions that involve individuals actively processing the ends and means. In human context, these dis-positions would include things like emotional reactions to other indi-viduals’ behaviour. They do not need to determine the behaviour to be selected dispositions; learning and reflection can shape both the perception triggering the emotion and the range of behavioural re-sponses within the emotional state and the social context while the emotional response is still an evolved biasing factor for behaviour (see Mallon & Stich 2000). Non-reflective dispositions are likely to evolve if the social settings are robust enough to couple social triggers and a functional reaction reliably, since individual processing (contra se-lected reactive dispositions) needs to go through proxies (that is, indi-viduals need to want things that lead into fitness-increasing behav-iour, and they need to recognize these things in a situation; see Sterelny 2003), processing consequences takes time and effort, and it is prone to errors that may be unnecessary. If there are persistent or frequent features in the social environment, selection favours reactive dispositions over individual cognition – although the two may be mixed. The selection for social environment does not need to be dif-ferent from evolution guided by other features of the environment:

behaviour is a selected response to the environment as it is, guided by the cognitive and motivational mechanisms that are selected to bring about this behaviour under the (social) environmental conditions, based on the clues that individuals are able to perceive within the en-vironment and to process.

133

An individual does not need to aim at those consequences that are beneficial, but simply achieve those consequences because of what she or he is aiming to do: people may want to achieve something, but given the constraints, they achieve something else, which is what they actually “need” from the evolutionary point of view (see Sterelny 2003). For example, people do not have an evolved urge for beneficial amount of energy intake, but for things like sugar, which lead into a healthy energy intake with availability constraints of the environment of adaptation, while urges without these constraints may lead into too much energy intake. This holds in social environments too. We may want to help others, for example, because unbeknownst to us this ben-efits us (in an evolutionary sense) in the long run. An unavoidable difference from other conditions is that the selection of social behav-ioural traits is dynamic: the fitness of a trait depends on other traits in the population, making the evolution sensitive to the repertoire and frequency of other traits in the population. But this is still a competi-tion between traits.

However, the fitness-relevant effects of the trait may depend on the response from others in a way that the behaviour makes sense only as a relational trait between two or more individuals: the consequences of the behaviour depend on the form of interaction, not just the indi-vidual disposition to participate in the interaction in a particular way.

This means that the explanandum for an evolutionary explanation of such behaviour is non-individualistic. This, of course, depends on both the facts about the proximate mechanisms guiding the behaviour and the adaptive functionality. I will make a case for this later. Neverthe-less, the selection process itself may still be individualistic selection be-tween individuals, with mutually beneficial consequences emerging.

Game-theoretical models of behavioural strategies are usually inter-preted in this way. I will argue that it may be inadequate to consider the object of evolutionary explanation (even if game theory is used in modelling the behaviour) to be only individual behaviour or psychol-ogy, but sometimes the structures of social interaction may need to be taken as the object of explanation. The idea is roughly the following.

134

The behaviour has positive fitness effects for the individual because of indirect consequences that depend on the behaviour of other indi-viduals. If you are comparing the fitness of individuals in the popula-tion, it is not enough to know their individual behavioural disposi-tions, but also what kinds of social interactions they are involved in.

The object of the evolutionary explanation should be the collection of causal factors responsible for the link between the behaviour and its fitness consequences. If this collection is robust enough to evolve, and if it is a system of several individuals, the evolutionary explanation will be holistic in the proximate dimension. It is, however, another question whether the evolutionary explanation itself needs to be holis-tic (that is, the selection takes place between groups of individuals) or if it can still be individualistic (that is, the selection takes place be-tween individuals only). I will discuss this issue in the last chapter.

3.1.2. Ontogeny

Ontogeny, or development, also includes both internal and external fac-tors. In this case, however, what counts as internal is not so clear.

Sometimes, if the gene’s point of view is adopted, the distinction is made between genes and what counts as environment for their func-tioning – including the hormonal environment within the organism.

There are two possible justifications for this. First, the idea that genes are the source of intrinsic information that, in a linear causal sequence in the development, produces definite effects on phenotype, in inter-action with the external factors. This image, however, has proven to be wrong: development involves, even on the molecular level, not only coding sequences of DNA (the molecular genes), but also regu-latory sequences, the RNA and protein products of the DNA’s func-tioning, and the environmental signals that influence the regulatory machinery in the cell, which in turn is a functioning living cell from the very beginning (see Wolpert et al 2010; Griffiths & Stotz 2013). An alternative way to divide between internal and external is to start with the fact that the development of an organism is a process, in which

135

some causal factors are external to the process, and some are internal to it. In this, internal factors include both the genome and the non-genetic organismic properties in the developmental processes that build on the previous steps of the process. Even then, some of the ex-ternal factors may be causally essential to development: some of the internal factors of development depend on some external factors such that they are not “fundamentally” internal. This, however, is a more realistic image of the process. (Oyama, Griffiths & Gray 2001; Griffiths

& Stotz 2013.)

The second argument for the first way to make the distinction would be an evolutionary-theoretical argument from the gene’s point of view a la Dawkins (1976): the genes are the developmental resources that repro-duce and are there necessarily, in contrast to what depends ultimately on environment. This is an ad hoc distinction from a purely develop-mental point of view, but if the question comes from the evolutionary point of view instead, the very interest to make distinctions is founded differently. The perspective could entail different criteria for what is explanatorily adequate. This solution, however, does not work since it makes unwarranted assumptions about reproduction even from the evolutionary point of view, including modularizing the evolutionary units of reproduction and assuming only one route for the reproduc-tion of traits (see Oyama, Griffiths & Gray 2001; Jablonka & Lamb 2005; Godfrey-Smith 2009; Pigliucci & Müller 2010). Furthermore, the concept of gene used here is an abstraction of inheritance patterns from the actual developmental processes, not something that could be identified with the molecular genes that are being copied and are the ones that matter for the developmental process (see Moss 2004; Grif-fiths & Stotz 2013). I will return to this in more detail later.

The external factors include the environmental factors external to the organism that participate in the developmental processes, both during embryonic development (the mother being the most relevant part of the environment) and after that. They may include specific properties of the environment that the development is sensitive to during the special periods of sensitivity for those particular factors, as

136

well as more generally affecting behaviour-changing properties. The factors of the developmental process can also be divided, for example, into more and less fundamental factors (for example, through the depth of generative entrenchment; Wimsatt 1986 & 1999), and into per-missive and instructive factors (Woodward 2001; Griffiths & Stotz 2013;

Calcott 2017). Both may include internal and external factors. Molec-ular genetics is the study of one and only one component of this. I will return to the discussion on whether it is illuminating or not, or even seriously misleading, to differentiate between internal and external in development in the first place, especially in the development of be-haviour. But even though molecular genetics can be said to study a component of individual development of behaviour, much of behav-ioural genetics does not: it studies the association between genes and traits on the population level and should be classified as a different, although not independent, category. Evolutionary approaches work on the population level, and development used to be a black-boxed process between population genetics and population-level pheno-typic variation, but this is an oversimplification. (See Wolpert et al 2010; Oyama, Griffiths & Gray 2001; Carroll 2005; Griffiths & Stotz 2013; Dediu 2015.)

The role of the social (including cultural) environment that affects development is an important special case. As with proximate mecha-nisms, the social environment can be just a part of the (developmental) environment from the evolutionary point of view, or it could play a more specific role. Individual development always takes place in the interaction between internal and external factors, but what are im-portant in this from the evolutionary point of view are the individual differences that are transmitted to the next generation: the offspring resemble the parents in the ways that their parents are dissimilar from other individuals (Godfrey-Smith 2009). The features of the environ-ment (and of the genome as well) that are not difference-makers for this, but are instead shared, indifferent regarding the outcome (even if they contribute to development), or random, do not matter. The fac-tors that have the capacity to transmit traits, and differences in them,

137

to the next generation (or that participate in their reconstruction in the next generation; Griffiths & Gray 1994; Oyama, Griffiths & Gray 2001) can be targets of selection. This transmission may, at least in principle, take four different routes: genetic, epigenetic (that is, biological factors transmitted from mother to the offspring other than DNA that affect the developmental process), behaviour copying, and symbolic learn-ing (Jablonka & Lamb 2006).

From the evolutionary point of view, individualism and holism in the developmental dimension can be defined as follows. If the re-production or reconstruction of the traits under selection relies only on internal factors (genetic or otherwise) regarding the systematic similarities between parents and offspring and the differences be-tween individuals, these differences being preserved bebe-tween family lines (that is, the transmission is only vertical), the evolutionary pro-cess is individualistic in the developmental dimension. This leaves several options open for environmental factors in development: they may participate in the causal processes but the variation in these fac-tors does not affect the outcome; there may be a range of variation that depends partly on the environmental factors in a non-discrete way; or there may be a limited number of discrete variants where specific en-vironmental factors decide which of the possible variants gets devel-oped. The object of evolutionary explanation may be a variant of a trait, but it may be the existence and frequency of the alternatives on the population level as well.

A charitable reading of the so-called nativist evolutionary psychol-ogy (more about this soon) is that it takes the object of the evolutionary psychological explanations to be those aspects of psychology that may have a range of variation but for which the range of variation is ex-plainable as an adaptation. The alternatives may be explained by fac-tors in individual development but the range of developmental dispo-sitions by evolution alone. A charitable reading of what a trait being

“innate” means would likewise work along these lines. I will go deeper into this issue later. I will defend a limited usability of the con-cept of innateness, and I re-interpret nativism as a form of

138

methodological individualism. Nativist evolutionary psychology is narrowing its explanations to those aspect of psychology that develop because of internal factors only. I would, however, consider linear be-haviour copying from parents to offspring and even

methodological individualism. Nativist evolutionary psychology is narrowing its explanations to those aspect of psychology that develop because of internal factors only. I would, however, consider linear be-haviour copying from parents to offspring and even