Of spiders and men

I remember, as an undergraduate, being fascinated by a superb article written by James Gould (1) with Peter Marler describing the peculiar habits of nesting birds, particularly their self-defence mechanisms (2). Not being particularly large or strong, nesting birds are unable to fend off predatory birds by themselves and so have developed ‘mobbing’ techniques to scare away predators. If a nesting bird catches sight of a magpie, for example, it will start to screech and its screeching will, in turn, induce other nesting birds to also screech until the din is so loud that the magpie flees. The obvious question is: how do the birds know to screech at magpies but not at other non-predatory birds? How can they tell friend from foe?

It turns out that the answer to this question is a fairly simple but fascinating mix of nature and nurture. Nesting bird chicks do not recognise friend from foe but they do recognise the screech and will unhesitatingly join in with their elders. After a few pairings of screeching with the magpie, the young nesting chicks will begin to spontaneously screech in response to magpies and thus propagate the ‘mobbing’ of magpies. In the laboratory, experimenters have ingeniously presented magpies only to older nesting birds and, simultaneously, non-predatory birds only to the younger chicks. The younger chicks then grow up to mob the non-predatory bird. In this way, nesting chicks have also been raised to mob various banal household items, including bottles of bleach and shampoo.

What these experiments suggest to me is that nesting birds are not very bright; although born with a series of adaptive strategies for survival, those strategies do not demonstrate any thought, insight or understanding of why the behaviour is being performed. That conclusion, however, is most emphatically, and a little surprisingly, not one that James and Carol Gould want to encourage with their new book Animal Architects. In the opening pages of the book, they explain that animals have very special skills that might betray higher-level, goal-directed, conscious behaviour:

‘The engineering skill that goes into a beaver dam rivals the elegant calculations that built the pyramids and the Panama Canal. Furry rodents with paddle tails and oversized incisors gauge flow rates and stress, fell trees and cut them to size, and build roads and excavate canals to float the logs and branches to the construction site. They erect custom-designed multipurpose underwater bunkers and keep the water level just right, adjusting the flow and stopping leaks without the aid of opposable thumbs or even fingers.’ (ix)

They do allow that many animal behaviours can be considered automatic and innate and use the egg recovery behaviour of geese as a good example. When an incubating goose spots an egg-like object near the nest, it extends its neck forward, rises, and places its bill just beyond the object. The goose then rolls the egg back between its legs into the nest. This behaviour will occur in response to grapefruit, stones and a variety of egg-like shapes that are patently not eggs, and, once the recovery sequence begins, it continues to completion, even if the egg or other object is removed. Gould and Gould comment that a ‘”thoughtful” problem-solving approach would be bound to produce not only alternative solutions but also the inevitable errors that come with experimentation’. They go on, however, to propose thoughtful, insightful and goal-directed behaviour in caterpillars, wasps, spiders, bees, bower birds, ravens and beavers.

There are a number of strands to their argument but two essential ones are prominent. Firstly, Gould and Gould argue that as animals become larger and increasingly interact with members of their own species they, by necessity, become more complex. Secondly, Gould and Gould argue that certain complex behaviours, especially those that demonstrate an ability to deal with unlikely setbacks, could suggest something approaching insight, understanding and knowledge.

Their first argument is certainly true. For example, smaller creatures are able to rely upon a simple brain directly connected and interconnected with their sensory receptors; as the creature becomes larger, however, such direct connections become increasingly impossible because the brain would need to be larger than the creature’s body. Instead of direct connections, the brain is forced to extrapolate across larger regions of the body and forms a regional map of sensory receptors. Similarly, as the lives of creatures become more intertwined with one another there is increasing selective pressure for mechanisms that can predict and guide interactive behaviour, such as the linear ranking of individuals in a hierarchy.

But their second argument is certainly not true. Animal behaviour can look intelligent because the behaviour has been selected as a good solution to the problem of survival. Evolutionary selection favours animals that behave in a fashion that furthers their survival and reproduction and this can appear highly intelligent until conditions change. Burrowing provision wasps, for example, routinely drag their prey back to their burrow. Before entering the burrow with their prey they stop and enter the burrow to check for predators; otherwise they could be attacked as they backed in. This behaviour looks intelligent, but it was revealed as dumb by JH Fabre who pulled the prey back a couple of centimetres from the burrow entrance while the wasp was checking. When the wasp came back out it pulled the prey back towards the burrow and then headed back in for another check. When Fabre pulled the prey back again, the wasp repeated its behaviour, again and again and again, to the point of exhaustion on the part of Fabre and the wasp.

Gould and Gould agree that such behaviour does not indicate insight, and argue that selection favours a default plan that carries behaviour through its normal course in any event. An ability to deal with contingencies, however, especially unlikely setbacks, suggests to them that something more is happening. They propose, for example, that spiders must have goal-directed behaviour as opposed to innate pre-ordained programming in order to construct their webs, which are always produced in circumstances that have unique elements. An extreme example of the spider’s ability to cope with unique circumstances comes from outer space. Spiders have built normal webs in the absence of gravity on-board the space shuttle, and yet they usually depend on gravity to build the bottom radius of their web. Gould and Gould ask: ‘How, if the behaviour is entirely based on task-directed programming, would this be possible?’ Spiders have not evolved in space so their nonchalance in building webs in space seems to defy any evolutionarily dictated innate strategy and demand a goal-directed understanding.

It is easy to be seduced by this argument. At first glance, web-building in zero gravity appears to be compelling evidence of behavioural plasticity, which demonstrates insight to provide novel solutions. But interpreting this behaviour as intelligent is hasty. It is quite possible that Gould and Gould have just not thought imaginatively enough about the environment of spiders and the types of strategies that might benefit them. While it is true that spiders have not evolved in a zero-gravity environment, it is equally true that spiders have evolved in a gravitational environment that is not equivalent to ours. Generally speaking we do not spend our time tumbling through the air as we go about our usual behaviours. Spiders, in contrast, do spend a significant amount of their time tumbling through the air – when building webs, being buffeted by the wind and escaping more serious danger. As they tumble, their weight will pass through a number of gradations including weightlessness. In contrast to spiders, human beings typically do not move around when weightless except when exploring outer space. It is remarkable, therefore, that human beings can build things in space, but notably less remarkable that spiders can.

Half the book is spent discussing the behaviour of insects, and although it is all wonderful and fascinating none of it rises to the level of demonstrating insect insight and understanding. Perhaps that is asking too much. In the second half of the book Gould and Gould move on to birds. They discuss the famous behaviour of ground-nesting birds, especially the plover bird that can perform an elaborate broken wing display to coax predators away from the young in their nest, as Gould and Gould explain:

‘The nesting bird drags a wing, hops away clumsily (often tripping in the process), and in the climax of the performance, delivered only to the most difficult of audiences, even attempts unsuccessfully to fly…. Depending on the degree of threat, the bird may choose simply to feign nesting on a high point of ground…. The ability to choose wisely from an innate menu of distraction displays suggests an ability on the part of at least some ground-nesting birds to judge intention by behaviour, to integrate experience with innate prejudices to create expectations.’ (p159)

Again this appears to be intelligent, goal-directed, or expectation-driven behaviour, and that is clearly the conclusion that Gould and Gould draw. But that conclusion is hasty. If the plover is truly feigning behaviour it must have an understanding that the predator can be tricked, and if it can feign an inability to fly it must have a concept of flight as an alternative to hobbling along the ground. If the plover has all that knowledge then it must have a vast amount of additional knowledge as a framework for knowing what it is doing and doing that in preference to something else. It is not possible to have a single belief, held in isolation and fully formed in its own terms. Any formulation of the plover’s belief about her behaviour must involve concepts – ‘flight’, ‘ground’, ‘pretence’ – themselves rooted in general beliefs that no plover could have because it would be absurd to attribute that much knowledge to a plover (let alone a spider).

In short, the suggestion that the plover is deciding to feign injury imports too much understanding, an understanding that we can be certain the plover does not have based on the sheer implausibility and also on the plover’s own behaviour. If plovers can feign nesting in a tree (that is, they know that nesting in a tree is an option and they also know that enough other birds nest in trees for it to be a plausible pretence to encourage a ground predator to move on) and can make choices about the best defensive strategy, then why don’t they actually nest in trees, at least some of the time? Clearly this is something they have the ability to do, as demonstrated by their feigning the behaviour, but they don’t do it. The obvious interpretation is that plovers do not make choices at all but operate according to statistical probabilities that are coded into their behaviour by natural selection. Because these strategies work they have survived and propagated – and because these strategies work they also look intelligent.

Gould and Gould concede, albeit accidentally, that the behaviour of ground nesters is not as smart as it seems when they complain that: ‘[G]round nesters continue to thrive except where our agricultural practices, dune buggies, and dogs have changed the habitat, and thus the basic biological rules.’ That is precisely the problem with interpreting animal behaviour as insightful and intelligent; a stable evolutionary strategy will propagate and look intelligent until things change to reveal the strategy as automatic and dumb.

Gould and Gould also make great effort to convince us that the decorations bower birds bring to their nests have aesthetic value for the birds, that they like them and prefer some colours and shapes over others. It is not clear why Gould and Gould eschew the more plausible interpretation that these behaviours have evolved to attract mates and are part of the innate sexual selection behaviour of the bowers. They argue that the variation in artefacts, including flowers, fruits and stones, used to decorate the bower nests suggest preferences that must be driven by a mechanism beyond DNA. Undoubtedly that is true, just as the nesting birds that mob bleach bottles are being driven by more than just their DNA – but the mechanisms do not have to involve insight.

Although the behaviour of animals does not have to involve insight, does that mean we are correct to always rule it out? Gould and Gould make the observation that ‘in theory, at least, we cannot rule out the possibility that a goose reaching for a grapefruit to roll is marvelling at its crazy compulsion, much as we sometimes do at [our inappropriate behaviour]’. Could it be that spiders think about alternative ways of making webs and consider the relative virtues of one structure over another? That plovers enjoy nesting on the ground and decide to keep on doing it despite the ever-present danger from ground predators and the possible benefits of nesting elsewhere? That bowers are real artists and beavers real builders with opinions, ideas and thoughts about how best to decorate a nest or construct a dam? As already stated, such possibilities import so much knowledge and insight that the possibilities are rendered obviously implausible. Knowledge and insight of that depth would generate demonstrations of understanding beyond the immediate behaviour. Tools and general signs relating to the thoughts would be created to accompany the insight, representing the animal’s recognition of their own being. Outside of our own species, these things do not occur.

Only human beings generate obvious, evident and consistent signs of their conscious presence, but Gould and Gould are perversely reluctant to make the necessary distinctions between humans and animals. It is a reluctance that is common, diminishing the importance of humanity and talking up the importance of animals. Sadly, in this case, that reluctance means Gould and Gould diminish their otherwise excellent work exploring the fascinating, but fascinatingly limited, building behaviours of animals.

Stuart Derbyshire is a senior lecturer in the School of Psychology at the University of Birmingham, England.

Animal Architects: Building and the Evolution of Intelligence by James and Carol Gould was published by Basic Books. (Buy this book from Amazon(UK))

(1) James Gould has a variety of middle initials. On the cover of Animal Architects, he is referred to as James R. Gould but on the back sleeve he is James L. Gould. Various references and copies of his classic 1987 Scientific American article are referenced on the web as both James L. Gould and James G. Gould. James Gould, whatever his middle initial may be, is most certainly not to be confused with the sadly deceased Stephen Jay Gould

(2) ‘Learning by instinct’ by Gould JL and Marler P, Scientific American, 1987; 256: p.74-85