If men had wings and bore black feathers, few of them would be clever enough to be crows.Maybe the truth is somewhere in between.
Copyright © 2007 Elsevier Ltd All rights reserved.The social life of corvids Nicola S. Clayton1, and Nathan J. Emery2, 1Department of Experimental Psychology, University of Cambridge, Cambridge CB2 3EB, UK 2Sub-department of Animal Behaviour, University of Cambridge, Cambridge CB3 8AA, UK
Available online 20 August 2007.
Article OutlineOf the 120 species of birds in the corvid family, which includes the crows, ravens, magpies and jays, the bare-faced rook is perhaps the most social of them all. At a rookery in Norfolk, for example, winter roosts can number up to 60,000 individuals. The name for a congregation of rooks is a ‘parliament’. In English folklore, parliament is an apt name for rook justice, as it is said that rooks form a circle around a wrongdoer producing a cacophony of calls and caws which can go on for hours until the offender is either attacked and killed or released to live another day. Although only fiction, such tales reflect their canny reputation as thieves and tricksters, as well as possessors of great wisdom. Like most birds, corvids are monogamous, and the core unit is therefore the mated pair. This pair bond is typically for life, and the pair remains together throughout the year. For example, rooks and ravens find a partner during the autumn months, taking part in impressive aerobatic displays and food sharing which may be to assess the quality of a potential mate. Once juvenile rooks and ravens pair, they engage in extensive mutual preening and bill twining (bill holding) and support one another in fights.
ReportSpontaneous Metatool Use by New Caledonian Crows Alex H. Taylor1, , , Gavin R. Hunt1, Jennifer C. Holzhaider1 and Russell D. Gray1, , 1Department of Psychology, University of Auckland, Private Bag 92019, Auckland 1142, New Zealand
Received 27 June 2007;
revised 24 July 2007;
accepted 25 July 2007.
Published online: August 16, 2007.
Available online 16 August 2007.
SummaryA crucial stage in hominin evolution was the development of metatool use—the ability to use one tool on another  and . Although the great apes can solve metatool tasks  and , monkeys have been less successful ,  and . Here we provide experimental evidence that New Caledonian crows can spontaneously solve a demanding metatool task in which a short tool is used to extract a longer tool that can then be used to obtain meat. Six out of the seven crows initially attempted to extract the long tool with the short tool. Four successfully obtained meat on the first trial. The experiments revealed that the crows did not solve the metatool task by trial-and-error learning during the task or through a previously learned rule. The sophisticated physical cognition shown appears to have been based on analogical reasoning. The ability to reason analogically may explain the exceptional tool-manufacturing skills of New Caledonian crows.
Metatool use was one of the major innovations in human evolution  and . The use of simple stone tools to make more complex tools may reflect the “cognitive leap” that initiated technological evolution in hominins . Metatool use has three distinct cognitive challenges. First, an individual must recognize that tools can be used on nonfood objects. This recognition may require analogical reasoning abilities . Second, an individual must initially inhibit a direct response toward the main goal of obtaining food, a reaction that both children and primates find difficult to suppress ,  and . Third, an individual must be capable of hierarchically organized behavior  and . That is, they must be able to flexibly integrate newly innovated behavior (tool→tool) with established behaviors as a subgoal in achieving a main goal (tool→tool→food). Such flexible, hierarchical organization of behavior has been suggested to follow a recursive pattern and to require cognitive processing similar to language production .In early hominins, the transfer of a thrusting percussion technique from breaking nuts to knapping cutting tools was likely part of longer behavioral sequences in which tool materials and food were acquired separately . Metatool use, therefore, probably involved considerable behavioral organization in space and time. Tests for metatool use in great apes and monkeys have typically followed an experimental design where a small stick can be used to retrieve a nearby longer stick that can then be used to gain otherwise inaccessible food. The close proximity of the tools and the food in these tests eliminates tool transport and facilitates assessment of the relevant requirements of the task. It also makes it relatively easy to accidentally touch the long tool with the short tool in normal exploratory behavior, and thereby chance upon the solution. Increased distance between tools and the food source has been suggested to increase the cognitive demands of a tool task  and . Striking evidence is now emerging that Corvidae have convergently evolved cognitive abilities that rival those of our primate relatives . Evidence for convergent evolution include the impressive tool-manufacturing skills of New Caledonian crows (Corvus moneduloides) , , , ,  and  and complex physical cognition in non-tool-using rooks (Corvus frugilegus) . To test whether New Caledonian crows (crows hereafter) are capable of metatool use, we used an experimental design similar to the standard design used with great apes  and . We modified the design to give a greater degree of spatial and temporal separation between the tools and the food. In our experiments, food (meat) was placed in a 15 cm deep horizontal hole 1.75 m away from two identical “toolboxes” (Figure 1). The front of each toolbox consisted of vertical bars that allowed a crow to insert its bill but not its head. We placed an 18 cm long stick tool 4 cm inside one toolbox. This tool was long enough to extract the meat but out of reach of a crow's bill. In the other toolbox, we placed a stone in a similar position. The positions of the stone and tool were randomized between the toolboxes across trials. Presenting both a relevant and an irrelevant object controlled for random probing of the toolboxes leading to a solution by trial and error. In front of the toolboxes, we placed a 5 cm long tool (Figure 1). This tool was too short to extract the meat but could be used to extract the long tool from the tool box. Successful completion of the task required a crow to use the short stick to extract the long stick from the box and then transport the long stick to the hole and extract the food. The experimental apparatus consists of a long, functional tool in one toolbox, a stone in the second toolbox, a short, nonfunctional tool in front of both toolboxes, and a 15 cm deep horizontal hole in which meat was placed. The distance between the hole and the toolboxes was 1.75 m but is reduced in the image to save space.
All seven crows developed metatool use and extracted the food (Figure 2). Icarus, Luigi, and Gypsy spontaneously produced the correct behavioral sequence in the first trial (Gypsy's and Icarus's first trial are shown in Movies S1 and S2, respectively, in the Supplemental Data available online). This was despite the requirement to transport tools and the difficulty in obtaining a tool from behind the bars. Joker also successfully solved the problem on the first trial, but made the error of taking the short tool to the hole after a first attempt at extracting the long stick (Figure 2). Colin, Lucy, and Ruby first extracted food in the 5th, 19th, and 23rd trial, respectively. Significantly, the first use of the short stick by six of the seven crows was either successful metatool use or a failed attempt to extract the long tool. This performance is comparable with that of the great apes  and . In the first trial, five out of six gorillas and three out of five orangutans used a tool as a metatool . However, only three out of five chimpanzees (Pan troglodytes) developed metatool use, and these individuals first made the error of attempting to use the small, nonfunctional stick tool to obtain the food . Monkeys have been less successful. One out of two capuchins (Cebus apella) performed at a similar level to gorillas and developed metatool use on the first trial . In another study, only one out of six capuchins used tools as metatools and this individual succeeded in less than 50% of trials . Despite receiving considerable training on tool use, Japanese macaques (Macaca fuscata) did not attempt metatool use on the first trial and required more than 50 trials to achieve a 75% success rate .
Initial use of the nonfunctional tool in an attempt to get the food frequently occurs in primate metatool-use studies  and . In our experiment, only Lucy made the error of first taking the nonfunctional stick to the hole. Four crows (Ruby, Joker, Luigi, and Colin) occasionally attempted to use the nonfunctional tool to get food in later trials, but only after unsuccessfully trying to extract the long tool with the short tool. These crows appeared to have had difficulty extracting the long tool from the barred toolbox. They may have then taken the nonfunctional short tool to the hole because of problems inhibiting tool use when no other course of action was available. The task could have been solved by trial-and-error learning if crows had initially used tool-related exploratory behavior toward the toolboxes and stumbled across the solution. However, the crows did not randomly probe the toolboxes. The first toolbox probed by all seven crows was the one with the long stick rather than the stone. In fact, only Ruby ever probed the toolbox containing the stone; she did so once, several trials after successful metatool use. This suggests that metatool use did not develop through trial-and-error learning during the experiment. The use of a previously learned behavioral rule by the crows is also unlikely. Familiarization training with the apparatus did not involve metatool use, and we have never seen this behavior in the wild in more than 3 years of observing crows on Maré. The spontaneous development of metatool use therefore required cognition more complex than simple learning mechanisms. One possibility is that the crows solved the metatool task by analogical reasoning. Successfully constructing an analogy requires that an individual maps experience from previous problems onto a structurally similar, novel problem ,  and . One language-trained chimpanzee has been reported to have solved both figural and conceptual analogy problems . The crows may have solved the metatool-use task by perceiving the shared causal relationship between the task and normal tool use, namely that a tool can access out of reach objects. Children's performance with causal analogies depends in part on knowledge of the relevant causal properties of the task ,  and . Causal understanding is indicated by the spontaneous correction of mistakes in an appropriate, goal-directed way  and . If the crows had understood the relevant causal relationship in this experiment, we would expect them to use this knowledge to avoid making errors based on tool type. To see whether crows were sensitive to the causal aspects of the food extraction task, we carried out a second experiment where the positions of the short and long tools were reversed. The long tool was now freely available so that metatool use was not required to extract the food. In the first block of five trials, all six crows tested initially inserted the long tool into the toolbox containing the short tool, but this generally occurred in the first block of five trials (Figure 3). This behavior usually lasted momentarily and there was often no contact with the short tool. In the only exception, Lucy extracted the short stick from the toolbox in her first trial but did not take it to the hole. No crow took the short stick to the hole. The insertion of the long tool appeared to be due to the difficulties in deviating from habitual behavior . The crows may have routinely probed the toolbox with the long tool because they had been rewarded in the previous ten metatool-use trials for probing the box. The crows rapidly rectified this mistake, suggesting that they were sensitive to the causal relationship between the tools and the final goal.
Our findings provide experimental evidence that New Caledonian crows can spontaneously solve a metatool task. On their first attempt to solve the problem, six out of seven crows used the short tool to probe the toolbox with the long tool. This appropriate spontaneous behavior and the quick correction of causal errors suggest that the crows used analogical reasoning to solve the metatool task. Analogical reasoning may be the crucial factor in the exceptional tool-manufacturing skills of New Caledonian crows.Figure 1. Our work was carried out under University of Auckland Animal Ethics Committee approval R375.
1 R. Byrne, The technical intelligence hypothesis: an additional evolutionary stimulus to intelligence?. In: A. Whiten and R. Byrne, Editors, Machiavellian Intelligence Vol II: Evaluations and Extensions, Cambridge University Press, Cambridge (1997), pp. 289–311.
2 S.A. de Beaune, The invention of technology: prehistory and cognition, Curr. Anthropol. 45 (2004), pp. 139–162.
3 N.J. Mulcahy, J. Call and R.I.M. Dunbar, Gorillas (Gorilla gorilla) and orangutans (Pongo pygmaeus) encode relevant problem features in a tool-using task, J. Comp. Psychol. 119 (2005), pp. 23–32.
4 W. Kohler, The Mentality of Apes (2nd ed.), Harcourt, Brace & Co., New York (1925) translated from German by E. Winter.
5 J. Anderson and M. Henneman, Solutions to a tool-use problem in a pair of Cebus Apella, Mammalia 58 (1994), pp. 351–361.
7 S.T. Parker and P. Poti, The role of innate motor patterns in ontogenetic and experiential development of intelligent use of sticks in Cebus monkeys. In: S.T. Parker and K.R. Gibson, Editors, “Language” and Intelligence in Monkeys and Apes: Comparative Development Perspectives, Cambridge University Press, New York (1990), pp. 219–243.
8 A. Diamond, Developmental time course in human infants and infant monkeys, and the neural basis of the inhibitory control of reaching, Ann. N Y Acad. Sci. 608 (1990), pp. 637–676.
9 S.T. Boysen and G.G. Berntson, Responses to quantity: perceptual versus cognitive mechanisms in chimpanzees (Pan troglodytes), J. Exp. Psychol. Anim. Behav. Process. 21 (1995), pp. 82–86. Abstract | PDF (681 K)
10 L. Santos, B.N. Ericson and M. Hauser, Constraints on problem solving and inhibition: object retrieval in cotton-top tamarins (Saguinus oedipus oedipus), J. Comp. Psychol. 113 (1999), pp. 186–193. Abstract | PDF (3382 K)
11 R. Byrne and A. Byrne, Complex leaf-gathering skills of mountain gorillas (Gorilla g. berengei): variability and standardisation, Am. J. Primatol. 31 (1993), pp. 521–546.
12 R. Byrne and A. Russon, Learning by imitation: a hierarchical approach, Behav. Brain Sci. 21 (1998), pp. 667–721.
13 T. Matsuzawa, Chimpanzee intelligence in nature and in captivity: isomorphism of symbol use and tool use. In: W.C. McGrew, L. Marchant and T. Nisida, Editors, Great Ape Societies, Cambridge University Press, Cambridge (1996), pp. 196–212.
14 E. Jalles-Filho, R. Grassetto Teixeira da Cunha and R. Aureliano Salm, Transport of tools and mental representations: is capuchin monkey tool behaviour a useful model of Plio-Pleistocene hominid technology?, J. Hum. Evol. 40 (2001), pp. 365–377. Abstract | PDF (147 K)
15 N. Emery and N. Clayton, The mentality of crows: convergent evolution of intelligence in corvids and apes, Science 306 (2004), pp. 1903–1907.
16 G.R. Hunt, Manufacture and use of hook-tools by New Caledonian crows, Nature 397 (1996), pp. 249–251.
17 G.R. Hunt and R.D. Gray, Diversification and cumulative evolution in New Caledonian crow tool manufacture, Proc. R. Soc. Lond. B. Biol. Sci. 270 (2003), pp. 867–874.
18 G.R. Hunt and R.D. Gray, The crafting of hook tools by wild New Caledonian crows, Proc. R. Soc. Lond. B. Biol. Sci. 271 (Suppl.) (2004), pp. S88–S90.
19 G.R. Hunt, M.C. Corballis and R.D. Gray, Laterality in tool manufacture by crows, Nature 414 (2001), p. 707.
20 G.R. Hunt and R.D. Gray, Species-wide manufacture of stick-type tools by New Caledonian crows, Emu 102 (2002), pp. 349–353.
21 A.A.S. Weir, J. Chappell and A. Kacelnik, Shaping of hooks in New Caledonian crows, Science 297 (2002), p. 981.
27 M.J. Rattermann and D. Gentner, More evidence for a relational shift in the development of analogy: children's performance on a causal-mapping task, Cogn. Dev. 13 (1998), pp. 453–478. Abstract | PDF (1969 K)
28 U. Goswami, Analogical reasoning and cognitive development. In: H. Reese, Editor, Advances in Child Development and Behaviour, Academic Press, San Diego (1996), pp. 92–135.
29 L.E. Richland, R.G. Morrison and K.J. Holyoak, Children's development of analogical reasoning: insights from scene analogy problems, J. Exp. Child Psychol. 94 (2006), pp. 249–273. Abstract | Article | PDF (392 K)
31 J.S. DeLoache, S. Sugarman and A.L. Brown, The development of error correction strategies in young children's manipulative play, Child Dev. 56 (1985), pp. 928–939.
32 T. Betsch, S. Haberstroh, B. Molter and A. Glockner, Oops, I did it again—relapse errors in routinized decision making, Organ. Behav. Hum. Decis. Process. 93 (2004), pp. 62–74. Abstract | Article | PDF (216 K)
CorrespondenceWhy some bird brains are larger than others Fahad Sultan Department of Cognitive Neurology, Hertie-Institute for Clinical Brain Research, University of Tuebingen, Otfried-Mueller-Strasse 27, 72076 Tuebingen,Germany.
Available online 6 September 2005.
Article Outline1], well suited for a comparative study of size and design. The cerebellum in birds, as in mammals, consists of a strongly folded, thin sheet of gray matter, located dorsally to the brainstem. In birds, it largely consists of a single narrow strip that varies in different species in the antero-posterior extension, which corresponds to the cerebellar length. The cerebellum of birds is commonly subdivided into ten groups of folds termed lobuli . Both variability and regularity are evident in the lobular pattern of the bird cerebella. To quantify these structural varieties and relate them to functional or phylogenetic differences, a principal component analysis was performed on the residuals of the lobuli length, obtained from a double-logarithmic regression of lobuli length against body size (see Supplemental Data on-line for further details).
Figure 1. Multivariate analysis of cerebellar lobuli lengths in birds obtained from line drawings in [2,10] (see Supplemental Data for further details). The graph plots scores of the individual birds on the first two PCs. The two PCs explain 66% of the variance (PC1, 44%; PC2, 22%). The largest variation along the first PC is between the woodpeckers, crows and parrots on the one hand, and the pheasants (Partridge and Wild turkey) on the other. Owls loaded high on the second PC, while swifts and hummingbirds loaded low. Generally, birds were clustered according to their family grouping as seen in the owls (#51), ducks (#14), pigeons (#60), woodpeckers (#17), crows (#123), parrots (#45), and gulls (#82). A one-way factorial ANOVA showed a statistically significant effect of family grouping as a factor on the overall variance (Fratio 4.59, p < 0.001). Two species, the barn owl and mallard, were present in both sources [2 and 10] and are plotted with interconnected dashed lines. Two individuals of the rock dove were present in  and are also plotted separately and interconnected by a dashed line. (Numbering of bird families taken from  and listed in Table S1 in Supplemental Data.)
Generally, birds within a family (Figure 1) tended to score similarly on the two principal components (PCs). The variability in the principal plane is dominated by variability between bird families (one-way ANOVA with bird family as factor: F(23, 24)= 4.59, p<0.001). The group of birds that scored highly on the first PC consisted of crows, parrots and woodpeckers. In contrast, the nocturnal owls had a different growth pattern and scored highly on the second PC. Both PCs together explained 66% of the total variance (first PC, 44%; second PC, 22%). The cerebellar growth of the first group is based on the enlargement of Larsell’s lobuli IV and VI–IX, while in the owl group lobuli I–II and X increase in size (Figure 2A,B). The difference in enlargement of these two groups of lobuli in the two groups of birds, i.e., crows, parrots and woodpeckers versus owls was statistically significant (Figure 2C).
(A) Loadings of the individual lobuli on the first two PCs. Evident are two groups: lobuli IV, VI–IX load strongly on the first PC, while lobuli I, II and X load on the second PC. (B) Two birds with roughly equal body size that correlate with the two groups of lobuli are shown (longeared owl, 276 g; green woodpecker, 195 g). In (A,B) contributions of the lobuli to the two first PCs are color coded: red codes first PC, blue second PC. (C) Comparison of the residuals of the two lobuli groups in the owl and in the crow, parrot and woodpecker group. The summed length of the lobuli that loaded highest on either the first or second PC were taken (PC1, IV, VI–IX; PC2, I, II and X) and the residuals to body weight were calculated. Residuals from birds (n = 12) from families that loaded strongly on either the first (crows, parrots and woodpeckers) or second PC (owls) were taken. The difference between these bird groups were statistically significant (t test for lobuli IV and VI–IX: p < 0.001, df = 10; lobuli I, II and X: p < 0.01, df = 10). Error bars: ±SD. Bird drawings taken from .
The following groups of cerebellar lobuli can be related to functional subdivisions through their afferents: somatosensory (tail, III–IV; leg, IV–V; wing, IV–VIa; head, V–VII and VIIIb–IXa) [3 and 4], visual (VIc–IXc) [5 and 6], auditory (VII–VIII) and vestibular (IXd–X) . My analysis indicates that, in the diurnal bird group (crows, parrots and woodpeckers), the cerebellar regions that receive visual and trigeminal inputs show the greatest growth, while in the nocturnal owl’s group the vestibular and tail somatosensory receiving regions show the greatest enlargement. The lobuli that are enlarged in crows, parrots and woodpeckers (IV and VI–IX) contribute to about 73% of the overall cerebellar length in birds. Not surprisingly, the birds that have enlarged this part of the cerebellum also have the longest cerebella — normalized for their body weight. The residuals of total cerebellar length in the crow, parrot and woodpecker families were significantly larger than those for the owl families (0.15±0.04 compared to 0.04±0.07, t test p<0.01, df=10). One unexpected observation was that in excellent flyers only the buzzard scores positively, and that several birds with excellent flying capabilities like the swift and falcon score negatively in the principal plane (Figure 1). This implies that well-developed motor skills per se do not require a large cerebellum, contradicting the common idea that cerebellar size increase in birds is mainly linked to their flying capabilities. What could be the behavioral denominator common to crows, parrots and woodpeckers that is not developed in owls? All of these birds also have large brains; however, their cerebellar designs differ arguing against a simple co-enlargement model . The enlargement of specific visual and beak-related cerebellar parts in crows, parrots and woodpeckers fits well with their marked adeptness in using their beaks and/or tongues to manipulate and explore external objects. Their skills are even comparable to those of primates in using their hands . The tight temporal coupling between motor command, expected sensory consequences and resulting afferents during visually guided hand and beak usage may be the reason why these animals need large cerebella. The comparative analysis of the birds cerebella reveals that some brains may have enlarged to solve similar problems by similar means during phylogeny. Furthermore it shows that large brains have a specific architecture with dedicated building blocks.
1 V. Braitenberg, D. Heck and F. Sultan, The detection and generation of sequences as a key to cerebellar function: experiments and theory, Behav. Brain Sci. 20 (1997), pp. 229–245.
2 O. Larsell, The development and subdivisions of the cerebellum of birds, J. Comp. Neurol. 89 (1948), pp. 123–189.
3 D.G. Whitlock, A neurohistological and neurophysiological study of afferent fiber tracts and receptive areas of the avian cerebellum, J. Comp. Neurol. 97 (1952), pp. 567–635.
5 P.G. Clarke, The organization of visual processing in the pigeon cerebellum, J. Physiol. 243 (1974), pp. 267–285.
6 J.M. Wild, Direct and indirect “cortico”-rubral and rubro-cerebellar cortical projections in the pigeon, J. Comp. Neurol. 326 (1992), pp. 623–636.
7 D.R.W. Wylie, K.L. Lau, X.H. Lu, R.G. Glover and M. Valsangkar-Smyth, Projections of Purkinje cells in the translation and rotation zones of the vestibulocerebellum in pigeon (Columba livia), J. Comp. Neurol. 413 (1999), pp. 480–493.
8 B.L. Finlay, R.B. Darlington and N. Nicastro, Developmental structure in brain evolution, Behav. Brain Sci. 24 (2001), pp. 263–278.
9 L. Lefebvre, N. Nicolakakis and D. Boire, Tools and brains in birds, Behaviour 139 (2002), pp. 939–973.
10 K Senglaub, Das Kleinhirn der Vögel in Beziehung zu phylogenetischer Stellung, Lebensweise und Körpergröße, nebst Beiträgen zum Domestikationsproblem, Z f Wiss Zool 169 (1964), pp. 2–63.
11 G.C. Sibley and B.L. Monroe, Distribution and Taxonomy of Birds of the World, Yale University Press, New Haven (1990).
12 J.F. Naumann, Naturgeschichte der Vögel Mitteleuropas, 2nd edn, Fr. Eugen Koehler, Gera-Untermhaus (1905).
Ravens have been observed taking turns sliding on their tails, feet first, down a snow bank as well as repeatedly sliding down smooth pieces of wood in their cages. Ravens have been seen playing with dogs, taking turns chasing it around a tree. One captive raven was observed tossing a rubber ball, pebbles, or snail shells into the air and catching them repeatedly. This same bird would often lay on its back and shift various playthings (toys) between its beak and its claws much like many children do with their toys. Other birds fell forward from a perch like an acrobat, in order to hang upside down by their feet, wings outstretched, then let go one foot at a time. While upside down, they would carry pieces of food, or shift items from beak to feet. One, while holding onto a branch with his feet, learned to propel himself around and around the perch by flapping his wings, like a gymnast on uneven parallel bars in a sort of 'loop-the-loop. The same captive ravens also played balancing games: carefully walking out as far as possible to the end of a tiny branch until it bent downward, turning them upside down; or trying to stand on a stick or bone held in the feet, while balancing it on top of and parallel to a perch made from a thick, wooden dowel.When given time and the resources birds will play. The corvids do. Perhaps it is the corvids extensive use of playing, allowing themselves and their young to learn and develop through playing that allows them to thrive when other bird populations are declining at an alarming rate. Sources referenced Podulka, Sandy, Ronald W. Rohrbaugh, Jr., and Rick Bonney, Editors. Handbook of Bird Biology. 2nd edition. Ithaca, NY: Cornell Lab of Ornithology, 2004.
- Podulka, Sandy, Ronald W. Rohrbaugh, Jr., and Rick Bonney, Editors. Handbook of Bird Biology. 2nd edition. Ithaca, NY: Cornell Lab of Ornithology, 2004. [↩]