Malevolent environmental influences may themselves be hard to predict for the same reasons as indifferent ones, but they introduce an added hazard; an added opportunity for the victim to make ‘mistakes’. The mistake made by a robin in feeding a cuckoo in its nest is presumably in some sense a maladaptive blunder. This is not an isolated, unpredictable occurrence such as arises because of the statistical unpredictability of the non-malevolent part of the environment. It is a recurrent blunder, afflicting generation after generation of robins, even the same robin several times in its life. Examples of this kind always make us wonder at the compliance, in evolutionary time, of the organisms that are manipulated against their best interests. Why doesn’t selection simply eliminate the susceptibility of robins to the deception of cuckoos? This kind of problem is one of many which I believe will one day become the stock in trade of a new subdiscipline of biology—the study of manipulation, arms races and the extended phenotype. Manipulation and arms races form the subject of the next chapter, which in some ways can be regarded as an expansion of the theme of the final section of this chapter.
4 Arms Races and Manipulation
One of my purposes in this book is to question the ‘central theorem’ that it is useful to expect individual organisms to behave in such a way as to maximize their own inclusive fitness, or in other words to maximize the survival of copies of the genes inside them. The end of the previous chapter suggests one way in which the central theorem might be violated. Organisms might consistently work in the interests of other organisms rather than of themselves. That is, they might be ‘manipulated’.
The fact that animals frequently cause other animals to perform some action that is against their own best interests is, of course, well known. Obviously it happens every time an angler fish catches prey, every time a cuckoo is fed by its foster mother. I shall make use of both these examples in this chapter, but I shall also emphasize two points that have not always been stressed. Firstly, it is natural to assume that even if a manipulator gets away with it temporarily, it is only a matter of evolutionary time before the lineage of manipulated organisms comes up with a counter-adaptation. In other words, we tend to assume that manipulation only works because of the ‘time-lag’ constraint on perfection. In this chapter I shall point out that, on the contrary, there are conditions under which we should expect manipulators to succeed consistently and for indefinite lengths of evolutionary time. I shall discuss this later under the catch-phrase of ‘arms races’.
Secondly, until the last decade or so, most of us have paid insufficient attention to the likelihood of intraspecific manipulation, especially exploitative manipulation within the family. I attribute this deficiency to a residuum of group-selectionist intuition which often lurks in the depths of the biologist’s mind even after group selection has been rejected at the surface level of reason. I think that a minor revolution has taken place in the way we think about social relationships. ‘Genteel’ (Lloyd 1979) ideas of vaguely benevolent mutual cooperation are replaced by an expectation of stark, ruthless, opportunistic mutual exploitation (e.g. Hamilton 1964a,b, 1970, 1971a; Williams 1966; Trivers 1972, 1974.; Ghiselin 1974a; Alexander 1974). This revolution is popularly associated with the name ‘sociobiology’, although the association is somewhat ironic since, as I have suggested before, Wilson’s (1975) great book of that name is in many respects pre-revolutionary in attitude: not the new synthesis, but the last and greatest synthesis of the old, benevolent regime (e.g. his Chapter 5).
I can exemplify the changed view by quoting from one recent paper, Lloyd’s (1979) entertaining review of sexual dirty tricks in insects.
Selection for haste in males, and coyness in females, results in what amounts to competition between the sexes. Males may be selected to bypass any choice that the females attempt to exercise, and then females selected to maintain their options, to not be misled or to have their choices subverted. If males subdue and seduce females with true aphrodisiacs [Lloyd gives evidence elsewhere in the paper], females may be expected to escape sooner or later in evolutionary time. And after sperm has been placed in a female, she should manipulate it: store, transfer (from chamber to chamber), use, eat, or dissolve it, as she makes additional observations on males. Females may accept and store sperm from a male for insurance that they will get a mate, and then become choosy … It is possible for sperm to be manipulated in the female (e.g. sex determination in Hymenoptera). Female reproductive morphology often includes sacs, valves, and tubes that could have evolved in this context. In fact, it is possible that some reported examples of sperm competition are actually cases of sperm manipulation … Given that females, to one extent or another, subvert male interests by the internal manipulation of ejaculate, it is not inconceivable that males will have evolved little openers, snippers, levers and syringes that put sperm in the places females have evolved (‘intended’) for sperm with priority usage—collectively, a veritable Swiss Army Knife of gadgetry!
This kind of unsentimental, dog eat dog, language would not have come easily to biologists a few years ago, but nowadays I am glad to say it dominates the textbooks (e.g. Alcock 1979).
Such dirty tricks, as often as not, involve direct action, the muscles of one individual moving to molest the body of another. The manipulation which is the subject of this chapter is more indirect and more subtle. An individual induces the effectors of another individual to work against its own best interests, and in favour of the interests of the manipulator. Alexander (1974) was one of the first to emphasize the importance of such manipulation. He generalized his concept of queen domination in the evolution of social insect worker behaviour, to produce a wide-ranging theory of ‘parental manipulation’ (see also Ghiselin 1974a). He suggested that parents are in such a commanding position over their offspring that offspring may be forced to work in the interests of their parents’ genetic fitness, even where this conflicts with their own. West-Eberhard (1975) follows Alexander in dignifying parental manipulation as one of three general ways in which individual ‘altruism’ can evolve, the others being kin selection and reciprocal altruism. Ridley and I make the same point, but do not restrict ourselves to parental manipulation (Ridley & Dawkins 1981).
The argument is as follows. Biologists define behaviour as altruistic if it favours other individuals at the expense of the altruist himself. A problem arises, incidentally, over how benefit and expense are to be defined. If they are defined in terms of individual survival, acts of altruism are expected to be very common, and will be taken to include parental care. If they are defined in terms of individual reproductive success, parental care no longer counts as altruism, but altruistic acts towards other relatives are predicted by neo-Darwinian theory. If benefit and expense are defined in terms of individual inclusive fitness, neither parental care nor care for other genetic relatives counts as altruism, and indeed a naive version of the theory expects that altruism should not really exist. Any of the three definitions can be justified, although if we must talk of altruism at all I prefer the first definition, the one that allows parental care as altruistic. But my point here is that, whichever definition we favour, it is met if the ‘altruist’ is forced—manipulated—by the beneficiary into donating something to him. For instance, the definition (any of the three) allows us no choice but to regard the feeding of a cuckoo nestling by its foster parent as altruistic behaviour. Maybe this means we need a new kind of definition, but that is another issue. Krebs and I take the argument to its logical conclusion, and interpret all of animal communication as manipulation of signal-receiver by signal-sender (Dawkins & Krebs 1978).
Manipulation is, indeed, pivotal to the view of life expounded in this book, and it is a trifle ironic that I am one of those who has criticized Alexander’s concept of parental manipulation (Dawkins 1976a, pp. 145–148; Blick 1977; Parker & Macnair 1978; Krebs & Davies 1978; Stamps & Metcalf 1980), and, in turn, been criticized for doing so (Sherman 1978; Harpending 1979; Daly 1980). Notwithstanding these defende
rs, Alexander (1980, pp. 38–39) himself has conceded that his critics were right.
Clearly some clarification is necessary. Neither I, nor any of the later critics mentioned, doubted that selection would favour parents who succeeded in manipulating their offspring, over parents who did not. Nor did we doubt that parents would indeed, in many cases, ‘win the arms race’ against offspring. All we were objecting to was the logic of suggesting that parents enjoy a built-in advantage over their children, simply because all children aspire to become parents. That is no more true than that children enjoy a built-in advantage simply because all parents were once children. Alexander had suggested that tendencies to selfishness in children, tendencies to act against the interests of their parents, could not spread because, when the child grew up, its own children’s inherited selfishness against it would detract from its own reproductive success. For Alexander this sprang from his conviction that ‘the entire parent-offspring interaction has evolved because it benefited one of the two individuals—the parent. No organism can evolve parental behaviour, or extend its parental care, unless its own reproduction is thereby enhanced’ (Alexander 1974, p. 340). Alexander was, therefore, thinking firmly within the paradigm of the selfish organism. He upheld the central theorem that animals act in the interests of their own inclusive fitness, and he understood this to preclude the possibility of offspring acting against their parent’s interests. But the lesson I prefer to learn from Alexander is that of the central importance of manipulation itself, which I regard as a violation of the central theorem.
I believe animals exert strong power over other animals, and that frequently an animal’s actions are most usefully interpreted as working in the interests of another individual’s inclusive fitness, rather than its own. Later in this book we shall dispense with the use of the concept of inclusive fitness altogether, and the principles of manipulation will be subsumed under the umbrella of the extended phenotype. But for the rest of this chapter it will be convenient to discuss manipulation at the level of the individual organism.
There is an inevitable overlap here with a paper which I wrote jointly with J. R. Krebs on animal signals as manipulation (Dawkins & Krebs 1978). Before proceeding, I feel obliged to acknowledge that this paper has been severely criticized by Hinde (1981). Some of his criticisms, which do not affect the parts of the paper I wish to use here, are answered by Caryl (1982), who had also been criticized by Hinde. Hinde took us to task for unfairness in quoting apparently group-selectionist statements by Tinbergen (1964) and others whom we labelled, for want of a better title, classical ethologists. I have sympathy with this historical criticism. The ‘good of the species’ passage which we quoted from Tinbergen was a genuine quotation, but I agree that it was not typical of Tinbergen’s (e.g. 1965) thinking at that time. The following much earlier quotation from Tinbergen would, perhaps, have been a fairer one for us to have chosen: ‘… just as the functioning of hormones and the nervous system implies not only the sending out of signals but also a specific responsiveness in the reacting organ, so the releaser system involves a specific responsiveness to particular releasers in the reacting individual as well as a specific tendency to send out the signals in the initiator. The releaser system ties individuals into units of a super-individual order and renders them higher units subject to natural selection’ (Tinbergen 1954). Even if outright group-selectionism was strongly opposed by Tinbergen and most of his pupils in the early 1960s, I still think that nearly all of us thought of animal signals in terms of a vague notion of ‘mutual benefit’: if signals were not actually ‘for the good of the species’ (as in the unrepresentative quotation from Tinbergen), they were ‘for the mutual benefit of both signaller and receiver’. The evolution of ritualized signals was regarded as mutual evolution: enhanced signalling power on one side was accompanied by increased sensitivity to the signals on the other side.
Today we would recognize that if receiver sensitivity increases, signal strength does not need to increase but, instead, is more likely to decrease owing to the attendant costs of conspicuous or loud signals. This might be called the Sir Adrian Boult principle. Once, in rehearsal, Sir Adrian turned to the violas and told them to play out more. ‘But Sir Adrian’, protested the principal viola, ‘you were indicating less and less with your baton.’ ‘The idea’, retorted the maestro, ‘is that I should do less and less, and you should do more and more!’ In those cases where animal signals really are of mutual benefit, they will tend to sink to the level of a conspiratorial whisper: indeed this may often have happened, the resulting signals being too inconspicuous for us to have noticed them. If signal strength increases over the generations this suggests, on the other hand, that there has been increasing sales resistance on the side of the receiver (Williams 1966).
As mentioned earlier, an animal will not necessarily submit passively to being manipulated, and an evolutionary ‘arms race’ is expected to develop. Arms races were the subject of a second joint paper (Dawkins & Krebs 1979). We were not, of course, the first to make the points that follow in this chapter, but nevertheless it will be convenient to quote from our two joint papers. With permission from Dr Krebs, I shall do so without breaking the flow with repeated bibliographic citations and quotation marks.
An animal often needs to manipulate objects in the world around it. A pigeon carries twigs to its nest. A cuttlefish blows sand from the sea bottom to expose prey. A beaver fells trees and, by means of its dam, manipulates the entire landscape for miles around its lodge. When the object an animal seeks to manipulate is non-living, or at least when it is not self-mobile, the animal has no choice but to shift it by brute force. A dung beetle can move a ball of dung only by forcibly pushing it. But sometimes an animal may benefit by moving an ‘object’ which happens, itself, to be another living animal. The object has muscles and limbs of its own, controlled by a nervous system and sense organs. While it may still be possible to shift such an ‘object’ by brute force, the goal may often be more economically engineered by subtler means. The object’s internal chain of command—sense organs, nervous system, muscles—may be infiltrated and subverted. A male cricket does not physically roll a female along the ground and into his burrow. He sits and sings, and the female comes to him under her own power. From his point of view this communication is energetically more efficient than trying to take her by force.
A question immediately arises. Why should the female stand for it? Since she is in control of her own muscles and limbs, why would she approach the male unless it is in her genetic interests to do so? Surely the word manipulation is appropriate only if the victim is unwilling? Surely the male cricket is simply informing the female of a fact that is useful to her, that over here is a ready and willing male of her own species. Having given her this information, doesn’t he then leave it up to her to approach him or not as she pleases, or as natural selection has programmed her?
Well and good when males and females happen to have identical interests, but examine the premise of the last paragraph. What entitles us to assert that the female is ‘in control of her own muscles and limbs’? Doesn’t this beg the very question we are interested in? By advancing a manipulation hypothesis we are, in effect, suggesting that the female may not be in control of her own muscles and limbs, and that the male may be. This example could, of course, be reversed, and the female be said to manipulate the male. The point being made has no specific connection with sexuality. I could have used the example of plants, lacking muscles of their own, using insect muscles as effector organs to transport their pollen, and fuelling those muscles with nectar (Heinrich 1979). The general point is that an organism’s limbs may be manipulated to work in the interests of the genetic fitness of another organism. This statement cannot be made convincing until later in the book when we have introduced the idea of the extended phenotype. For this chapter we are still working within the paradigm of the selfish organism, albeit we are starting to stretch it so that it creaks ominously at the edges.
> The example of male and female cricket may have been poorly chosen, because, as I said above, many of us have only recently become used to the idea of sexual relations as a battle. Many of us have yet to absorb into our consciousness the fact that ‘selection can act in opposition on the two sexes. Commonly, for a given type of encounter, males will be favoured if they do mate and females if they don’t’ (Parker 1979; see also West-Eberhard 1979). I shall return to this, but for the moment let us use a starker example of manipulation. As starkly ruthless as any battle in nature is that between predator and prey. There are various techniques a predator may use to catch his prey. He may run after them and attempt to outpace, outstay, or outflank them. He may sit in one place and ambush or trap them. Or he may do as angler fish and ‘femmes fatales’ fireflies (Lloyd 1975, 1981) do, and manipulate the prey’s own nervous system so that it actively approaches its own doom. An angler fish sits on the sea bottom and is highly camouflaged except for a long rod projecting from the top of the head, on the end of which is the ‘lure’, a flexible piece of tissue which resembles some appetizing morsel such as a worm. Small fish, prey of the angler, are attracted by the lure which resembles their own prey. When they approach it the angler ‘plays’ them down into the vicinity of his mouth, then suddenly opens his jaws and the prey are engulfed in the inrush of water. Instead of using massive body and tail muscles in active pursuit of prey, the angler uses the small economical muscles controlling his rod, to titillate the prey’s nervous system via its eyes. Finally it is the prey fish’s own muscles that the angler uses to close the gap between them. Krebs and I informally characterized animal ‘communication’ as a means by which one animal makes use of another animal’s muscle power. This is roughly synonymous with manipulation.