Cs.arizona.edu

mals for use as tools, without extensive prior experience, is almost unknown. In experiments by Povinelli [experiments 24 to 26 in (2)], chim-panzees (Pan troglodytes) repeatedly failed to unbend piping and insert it through a hole toobtain an apple, unless they received explicitcoaching. Further experiments [exp. 27 in (2)] Alex A.S.Weir, Jackie Chappell, Alex Kacelnik* (8) have shown a similar lack of deliberate,specific tool modification in primates. There are,however, numerous suggestive field observa- Many animals use tools, but their understand- different from those previously reported and tions (9) and one report of a male capuchin ing of physical forces or causal relations is would be unlikely to be effective with natural monkey (Cebus apella) unbending a piece of unclear (1, 2). Primates are considered the materials. She had little exposure to and no prior most versatile and complex tool users, but training with pliant material, and we have never Our finding, in a species so distantly related observations of New Caledonian crows (Cor- observed her to perform similar actions with to humans and lacking symbolic language, rais- vus moneduloides) (3–5) raise the possibility either pliant or nonpliant objects. The behavior es numerous questions about the kinds of un- that these birds may rival nonhuman primates probably has a developmental history that derstanding of “folk physics” and causality in tool-related cognitive capabilities.
includes experience with objects in their envi- available to nonhumans, the conditions for ronment ( just as infant humans learn about ev- these abilities to evolve, and their associated the observation that a captive female spontane- neural adaptations. Comparisons between New ously bent a piece of straight wire into a hook Caledonian crows and their relatives, as well as and successfully used it to lift a bucket contain- between other cognitively exceptional birds and ing food from a vertical pipe (Fig. 1A). This their relatives (11), offer a unique natural ex- occurred on the fifth trial of an experiment in periment to examine hypotheses about the eco- logical and neural preconditions for complex hooked and a straight wire and only after the cognition to evolve. It is not yet known if New hooked wire had been removed by the other Caledonian crows are also exceptional in cog- subject (a male). The animals had prior experi- nitively demanding tasks not involving tools.
ence with the apparatus, but their only previousexperience with pliant material was 1 hour of References and Notes
free manipulation with flexible pipe-cleaners a 1. S. Chevalier-Skolnikoff, Behav. Brain Sci. 12, 561 (1989).
2. D. J. Povinelli, Folk Physics for Apes (Oxford Univ.
year before this experiment, and they were not 3. G. R. Hunt, Nature 379, 249 (1996).
To investigate the importance of this ob- 4. G. R. Hunt, Proc. R. Soc. Lond. Ser. B Biol. Sci. 267,
servation, we conducted several new trials in 5. J. Chappell, A. Kacelnik, Anim. Cogn. 5, 71 (2002).
which we placed a single straight piece of 6. The female subject waswild-caught asa juvenile in March 2000 and hasbeen in our laboratory ever since. The male subject was in a zoo in New long) on top of the tube and did not intervene Caledonia for over 10 yearsuntil he wasmoved to until either of the birds obtained the food (a our laboratory, also in March 2000 (it is not known valid trial) or dropped the wire irretrievably when or how he wascaught). See (5) for further details of subjects, history, and housing conditions.
7. The male rarely attempted thistask and never bent Out of 10 valid trials (interspersed with the wire. He observed the female bending the wire seven invalid ones), the female bent the wire and stole the food from her in three trials. The birds are tested together because they are highly social and used it to retrieve the food nine times, and and, when separated, are less motivated to partici- Fig. 1. Bending wire into hooksby a captive New
the male retrieved the food once with the Caledonian crow. (A) The female New Caledonian
straight wire (7). To bend the wire, she first 8. E. Visalberghi, Int. J. Primatol. 18, 811 (1997).
crow extracting the bucket containing meat using wedged one end of it in sticky tape (available 9. M. Tomasello, J. Call, Primate Cognition (Oxford Univ.
a piece of wire she had just bent. This is a photo around the bottom of the tube and the side of taken after the experiment wascompleted, but 10. J. R. Anderson, M. C. Henneman, Mammalia 58, 351
the plastic tray containing the apparatus) or held the hook and posture depicted are typical of it in her feet at a location 3 m from the food, experimental trials. (B) Outline tracingsof all the
11. I. M. Pepperberg, The Alex Studies: Cognitive and bent wires, with the end inserted into the tube Communicative Abilities of Grey Parrots (Harvard where there was no tape. She then pulled the facing right. Numbersrefer to trial number. The other end orthogonally with her beak (see Mov- 12. We thank D. Wilson for technical assistance, the wire bent in trial 8 was not successfully used to ie S1), resulting in a bend with an angle of 74 Ϯ Wissenschaftskolleg zu Berlin for a research fellow- retrieve the bucket (it wasdropped into the tube).
ship to A.K., and Joan Silk for comments on the 30° (mean Ϯ SE) (see Fig. 1B for individual Because of experimenter error, the wire in trial 10 manuscript. The work was funded by a grant from the tool shapes). She started to bend the wire 35 Ϯ was2 cm longer than the wire in the other trials.
Leverhulme Trust to A.K. and a scholarship from the 8 s after the start of each trial and used the resulting hook 6 Ϯ 2 s later. In all cases but one, Supporting Online Material
she tried with the straight wire (for 15 Ϯ 4 s) eryday physics from their manipulative experi- www.sciencemag.org/cgi/content/full/297/5583/981/ before starting to make the hook. In all valid ence), but she had no model to imitate and, to trials, the birds retrieved the food within 2 min.
our knowledge, no opportunity for hook-making Thus, at least one of our birds is capable of to emerge by chance shaping or reinforcement novel tool modification for a specific task. In the of randomly generated behavior. She had seen Department of Zoology, South ParksRoad, Oxford wild, New Caledonian crows make at least two and used supplied wire hooks before but had not sorts of hook tools using distinct techniques (3, *To whom correspondence should be addressed.
4), but the method used by our female crow is Purposeful modification of objects by ani- www.sciencemag.org SCIENCE VOL 297 9 AUGUST 2002

Source: http://www.cs.arizona.edu/projects/wonac/papers/weir-2.pdf