News & Views | Published:

Cognitive science

Rank inferred by reason

Nature volume 430, pages 732733 (12 August 2004) | Download Citation


Pinyon jays seem to work out how to behave towards an unfamiliar jay by watching it in encounters with members of their own flock. The findings provide clues about how cognition evolved in social animals.

Susan is taller than Billy. Peter is taller than Susan. Who is taller, Billy or Peter? Knowledge about pairs of objects linked by relationships such as ‘taller’ or ‘stronger’ permits conclusions to be drawn about novel pairs (here, Billy and Peter) — a process known as transitive inference. Monkeys, rats and some birds can solve transitive-inference tasks in the laboratory1, but why would this ability evolve? A plausible answer is that transitive inference is an evolutionary adaptation in certain kinds of social group. For example, suppose I know from bitter experience that Bob always beats me in contests (that is, he dominates me). I now observe some new individual, Andy, dominating Bob. If I reason, “Andy dominates Bob, and Bob dominates me, therefore Andy will dominate me”, I can avoid fights by deferring to Andy when we meet. But there has been no well-controlled evidence that animals actually use transitive inference in social situations. In the study reported on page 778 of this issue, Paz-y-Miño and colleagues2 provide this.

In effect, the authors staged the Andy-and-Bob scenario using pinyon jays (Gymnorhinus cyanocephalus; Fig. 1), a highly social member of the crow family. These birds live in large, permanent flocks with clear pecking orders. Paz-y-Miño and colleagues created groups of captive pinyon jays that were previously unknown to each other, and allowed stable dominance relationships to develop in each group. Then jays from each group were allowed to observe individuals from other groups interacting over a peanut, and later interacted with some of those same birds. In the experiment, the observer saw a relatively dominant bird from its own group (‘Bob’ in our scenario) losing encounters with a stranger from another group (‘Andy’). To ensure that the stranger would not be seen only to dominate others, the observer also watched the same stranger losing contests with another stranger. As a control, other observer birds watched a stranger both winning and losing to members of the stranger's own group, an experience that should give observers no information on whether the stranger will be able to dominate them.

Figure 1: Pinyon jays — watching and learning.
Figure 1

If pinyon jays infer social status transitively, observers should behave more submissively in their first encounter with the stranger in the experimental condition than in the control — and this is what Paz-y-Miño and colleagues2 observed. To act in this way, the observers must first have identified both the individuals they watched and their roles in the observed encounter, and then retained this information for later use. Thus, their behaviour implies a more complex set of cognitive skills than that implied by another recent report of birds' sensitivity to a generic social relation (whether mated or not) between unfamiliar individuals present3.

It is noteworthy that the animals in this study are birds. The idea that the demands of a complex social life might drive the evolution of cognition was originally proposed to explain the apparently high intelligence of monkeys and apes4. However, other mammals, such as hyenas and elephants, and some birds also form long-lasting groups of identifiable individuals with differentiated social roles. Long-term field studies indicate that social complexity has shaped cognition in a similar way across species5,6, but in field work it can be difficult to know all the animals' relevant experiences. The experimental approach of Paz-y-Miño and colleagues2 will serve as a model of how manipulating the experiences of captive animals can provide firm conclusions.

If transitive inference is used in social life, species with more complex societies should be better at it. Laboratory studies using an abstract transitive-inference task provide some support for this idea. First, animals are taught the relative reward value of five or more pairs of colours7 — red is better than green, green is better than blue, and so on. Then they are tested with novel pairings. Monkeys behave in such tests as though they are reasoning by transitive inference, whereas pigeons do not1. But this finding need not reflect a difference in sociality between monkeys and pigeons because they differ in so many other ways. More relevant is a recent comparison of pinyon jays with the closely related, but less social, western scrub jay8. Pinyon jays perform more like monkeys in the abstract task than do scrub jays.

Members of the crow family, such as New Caledonian crows9, western scrub jays10 and ravens, feature in recent reports of remarkable cognitive abilities including making and using tools, episodic-like memory and forms of social learning. Each of these abilities is thought to reflect a species-specific adaptation, for example to extract prey or retrieve stored food. Comparisons of different species in similar tasks are necessary to test whether such abilities are in fact associated with the ecological factors supposed to select for them. Nevertheless, such findings underline the importance of viewing animal intelligence as consisting partly of species-specific adaptations, rather than being an entirely undifferentiated ‘general intelligence’. So, comparative psychology can provide experimental support for the idea that cognitive modules have evolved for solving different kinds of task — an idea so popular, but often untestable, in human evolutionary psychology5.


  1. 1.

    Cognition, Evolution, and Behavior (Oxford Univ. Press, 1998).

  2. 2.

    , , & Nature 430, 778–781 (2004).

  3. 3.

    , & Nature 430, 448–451 (2004).

  4. 4.

    Science 153, 501–506 (1966).

  5. 5.

    Trends Cogn. Sci. 8, 195–197 (2004).

  6. 6.

    de Waal, F. B. M. & Tyack, P. L. (eds) Animal Social Complexity (Harvard Univ. Press, Cambridge, 2003).

  7. 7.

    & Nature 267, 694–696 (1977).

  8. 8.

    , & Anim. Behav. 65, 479–487 (2003).

  9. 9.

    , & Science 297, 981 (2002).

  10. 10.

    & Nature 395, 272–274 (1998).

Download references

Author information


  1. Sara J. Shettleworth is in the Departments of Psychology and Zoology, University of Toronto, Toronto, Ontario M5S 3G3, Canada.

    • Sara J. Shettleworth


  1. Search for Sara J. Shettleworth in:

About this article

Publication history




By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing