Never laugh at fishes: some species and members of species are more intelligence than others

The eminent American neurophysiologist Donald Hebb was not lucky with his puppy Henry. Henry was part of Hebb’s project aimed to compare intellectual abilities in animals raised in enriched environment and in deprivation. Hebb (1949) previously reported that rats raised as pets performed better in mazes than normally reared laboratory rats. So, a litter of thoroughbred scotch terriers was divided into two parts where a half of puppies were deprived in cages and the others were adopted by collaborators as family pats. In general, Hebb’s prediction was true, that is, those dogs that were raised as pets in attentive families appeared to be faster learners than poor dogs which, when infants, saw nothing more than the wires of their cages. The one exception was Henry. He was extremely dull and even was not able to remember the way to home, so the family members had to release him from the animal shelter where he was caught as a clabber. It was of no surprise that he proved himself as the most retrograde individual when all siblings were tested in labyrinths.

This example supports a hypothesis that individuals show a great deal of variability in their capacity of learning and memory. In this book we have already met many examples of how levels of individual variability exceed levels of species-specific ability to learn and solve complex problems. For example, some individual monkeys outperformed apes in solving complex instrumental problems that require tool using, whereas one woodpecker finch and one New Caledonian crow outperformed them all (see Chapter 18 for details). Many papers devoted to experimental studies of animal intelligence end with a conclusion that at least one member of specimens being confronted with a definite problem coped with it and this gives a possibility to judge about intellectual potential of the species. For instance, in Gillan’s (1981) study three female chimpanzees were requested to solve complex transitive inference problem (see Chapter 16), and only one chimpanzee, Sadie, performed perfectly on the test on all 12 test trials.

Examining members of species in their performance on a series of learning problem, we obtain a possibility, at least illusory, to find a level of complexity of a task until which all animals are equal and no one is “more equal than others”. We also have a possibility to reveal the most “advanced” individuals and thus estimate limits of cognitive abilities within species. For instance, in the study of Sappington and Goldman (1994) cited in 15.3, four horses were presented with series of problems, from simple to more and more complex, in correspondence to the accepted scale of different levels of learning abilities from simple discrimination, to concept formation (Thomas, 1986). All horses in the study easily coped with the first problem, a simple black vs. white discrimination. Three of the four subjects went on to learn at least one of pattern discriminations, and only one horse completed both suggested problems involving the concept of triangularity. As it was already mentioned in the same Chapter, very similar results were obtained in the Honey Bee studies of Mazokhin-Porshnyakov (1989). All members of the hive successfully learned simple problems of discrimination. A task that required generalization on the basis of relative size, namely, to choose the biggest, or in other case the smallest, from a set of figures that varied in shape and colour, was solved by about a half of bees in the hive. Finally, the tasks that required concept formations, i.e., concepts of “triangulatity”, “twoness” and “symmetry” were solved by single “gifted” bees only.

Confronting members of closely related species with a similar experimental task, we obtain a chance to reveal how intellectual abilities of animals are tuned to the circumstances in which they live. Of particular interest are situations in which intellectual potential of species can be estimated by comparing problem solving in different specimens.

For example, field studies of ants’ ability to navigate “round mazes” of different levels of complexity revealed a close connection between percentage of individuals that solve problems and species specific foraging systems in studied ants (Reznikova, 1975, 1982, 2005). Ants were presented with mazes containing food pellets. The maze of the first level of complexity contained one circle with one entrance. This was hardly to be named “a maze” because all ants’ work was to enter the round box and to get food. The maze of the second level included two concentric circles placed one into another in such a way that their entrances were placed in two opposite sides. The most complex maze consisted of four circles (Fig. VII-1). In different series of field experiments, 100 mazes of the same level of complexity were placed on ants’ feeding territory. Two groups of ant species were compared, one group with a sole-foraging system (subgenus Serviformica) and other with group-retrieving (subgenus Formica s.str.). In sole-foraging species each forager searches for food by itself and thus undertakes long trips and makes its own decision whether to catch a prey and to take it home or not. In group-retrieving species professional specialisation takes place where a scout searches for food and a group of foragers share efforts to transport it. Feeding territories of these species are penetrated with foraging routs; systems of information transfer and food transportations allow them to exploit large food sources in optimal ways (Reznikova and Novgorodova, 1998; Reznikova and Ryabko, 1994; Robson and Traniello, 2002). In field experiments with mazes it turned out that from sole-foraging species nearly all active foragers successfully could obtain food pellets from the most complex mazes. In group foraging species not all of ants coped with the maze problem. Instead, there were “top ten” individuals which successfully navigated mazes consisting of two circles, that is, about 10% of foragers.

Similar results have been obtained by researchers comparing searching tactics and learning achievements of different rat species on radial mazes (Timberlake, Hoffman, 2002). Norway rats and laboratory reared desert kangaroo rats displayed clearly distinguishable tactics of searching for food, closely related with their different foraging patterns in nature. In the wild kangaroo rats use solitary foraging in patchy environment; these animals do not use odour trails, but the use of landmarks for minimizing travel distances is important for their foraging success. In experiments they applied a specific tactic on a maze dividing food sources into “patches”. Norway rats are highly social animals that spend the majority of their time above ground using trails established by the colony. They employed a tactic of trail following or, being deprived from the use of smell, strongly preferred the tactic of “central-place foraging”.

Besides these two studies on ants and rats, there are many other investigations which have shown that results achieved by members of different species in different laboratory environments depend on differences in field environments such as distribution of food, sociality of the life style, predator pressure, and many others.

Thorny path of searching for a common metric to measure intelligence across a broad range of species has attracted comparative psychologists. Researchers tried to classify the intellectual abilities of different animals and rank them within a universal intelligence scale. For example, Thomas (1986) complied data on concurrent discrimination learning for various fishes, reptiles, birds, and mammals, including mice, rats, zebras, donkeys, horses, and elephants. Of the species tested, only elephants and horses were able to discriminate the correct stimulus in 20 pairs of visual patterns concurrently. These were only two species that were able to successfully complete so many concurrent discriminations. Krushinskii (1973) compared species by their performance on detour tasks that required extrapolation of trajectories of moving bowls with food (see Chapter 12 for details). He subdivided species into six groups basing on their coping with tests. Among them primates, dolphins, brown bears and crows composed the top group, foxes, wolves, dogs and birds of prey belonged to the second group, whereas the “lowest” group included voles. Some results matched intuitive concepts about “animal intelligence” whereas others were difficult to explain. For instance, reptiles displayed outstanding capacities of extrapolation comparable with those of kites and falcons.

While a variety of laboratory tasks have been used, that of learning set formation has been widely explored since Harlow (1949) concluded that the results reflected evolutionary relationships. As it was considered in Chapter 17, the formation of learning set involves the inter-problem improvement in performance seen in subjects which are given a series of discriminations involving different pairs of stimuli. Rumbaugh (1968) revealed that in squirrel monkeys individuals differ in their abilities to pass the test to a great extent. Some monkeys exceeded apes in their marks whereas others were not able to pass the test at all. Subsequent work showed that closely related species may have widely divergent performances, and that some " lower" species may equal or excel " higher" species. Further, the ordering of species does not agree with that predicted from relative brain size. Macphail (1982) compared data from various resources concerning learning set formation in different species such as langurs, rhesus monkeys, bottlenose dolphins, cats, rats, squirrels, ferrets, and minks and concluded that it is not clear that any of the differences in performance in learning set formation (or any of the other types of behavioural studies considered) observed are due to differences in intellectual capacity. He cites a number of studies which demonstrate, as might be expected, that relative species performance is very dependent on details of experimental technique.

The growing set of data evidence that cognitive abilities in different species are hardly comparable. Members of species reach different results in dependence on a context of the experimental situation. For instance, primates master visual discriminations faster when the task is presented in the context of a foraging situation compared to a traditional testing situation such as the WGTA (Menzel, 1996). The second example concerns dolphins: when they were tested on auditory rather than visual discrimination, their performance improved significantly (Hermann, 1986).

Call (2002) starts his comments to Bshary’s and co-authors article (2002) on fish cognition from a note that when he told a colleague that he was reading an article on fish cognition, her first reaction was to laugh. However, research on fish promises to broaden our understanding of animal cognition in various ways: as fish are highly variable in their ecology, they can be used to determine the specific ecological factors that direct the evolution of specific cognitive abilities. Call (2002) concludes that standardization does not have much room for the special cognitive abilities of some species, precisely because they are special and therefore not suitable for comparison across a number of species. This raises a crucial question about principal possibility of cognitive specialization that will be considered further in this Part.