Carel P. van Schaik,Robert O. Deaner &Michelle Y. MerrillDepartment of BiologicalAnthropology and Anatomy,Duke University, Box 90383,Durham, NC 27708-0383,U.S.A. E-mail:vschaik@acpub.duke.edu

Received 2 July 1998Revision received11 December 1998 andaccepted 2 March 1999

Keywords: manipulative skills,extractive foraging,intelligence, social tolerance,nonhuman primates,interspecific variation,intraspecific variation,feeding tool use, foodsharing.

The conditions for tool use in primates:implications for the evolution of materialculture

In order to identify the conditions that favored the flourishing ofprimate tool use into hominid technology, we examine inter- andintraspecific variation in manufacture and use of tools in extantnonhuman primates, and develop a model to account for theirdistribution. We focus on tools used in acquiring food, usually byextraction. Any model for the evolution of the use of feeding toolsmust explain why tool use is found in only a small subset of primatespecies, why many of these species use tools much more readily incaptivity, why routine reliance on feeding tools is found in only twospecies of ape, and why there is strong geographic variation withinthese two species. Because ecological factors alone cannot explain thedistribution of tool use in the wild, we develop a model that focuseson social and cognitive factors affecting the invention and transmis-sion of tool-using skills. The model posits that tool use in the wilddepends on suitable ecological niches (especially extractive foraging)and the manipulative skills that go with them, a measure of intelli-gence that enables rapid acquisition of complex skills (through bothinvention and, more importantly, observational learning), and socialtolerance in a gregarious setting (which facilitates both invention andtransmission). The manipulative skills component explains the distri-bution across species of the use of feeding tools, intelligence explainswhy in the wild only apes are known to make and use feeding toolsroutinely, and social tolerance explains variation across populations ofchimpanzees and orang-utans. We conclude that strong mutualtolerance was a key factor in the explosive increase in technologyamong hominids, probably intricately tied to a lifestyle involvingfood sharing and tool-based processing or the acquisition of large,shareable food packages.

? 1999 Academic Press

Journal of Human Evolution (1999) 36, 719–741Article No. jhev.1999.0304Available online at http://www.idealibrary.com on

Introduction

Material culture is one of the hallmarks ofthe human species. The origins of ourmaterial culture are often sought in Plio-Pleistocene hominids who are known tohave used stone tools and probably also avariety of nonstone tools. However, homi-nids were probably not the first primates todo so, as some extant primates also makeand use tools. In this paper we develop amodel to explain the variation in the manu-

0047–2484/99/060719+23$30.00/0

facture and use of tools among and withinnonhuman primate species. If the modelsurvives further testing it should informspeculation on the causes of the flourishingof primate tool use into the elaboratematerial culture of hominids.

Nonhuman primates show at least sixfunctionally different modes of tool use(sensu Beck, 1980). First, many arborealspecies dislodge branches, and terrestrialspecies dislodge rocks from cliff faces, tointimidate predators or rivals. Kortlandt &

? 1999 Academic Press

720 . . ET AL.

Kooij (1963) and Hall (1963) consider thisthe basic tool use from which all other formshave been derived. It involves no manufac-ture and only crude manipulation, and isvirtually universal in primates large enoughto dislodge objects (see Table 1). Thesecond form, defensive tool use, involvingaimed clubbing and hitting of potentialpredators, is an elaboration of intimidationdisplays. It is quite rare even though thecontext is not (chimpanzees: Kortlandt &Kooij, 1963; capuchin monkeys: Boinski,1988). Third, tools are made and used ashunting weapons, but only by hominids(Klein, 1989). Fourth, wild chimpanzeesincorporate tools into social displays meantto attract rather than intimidate con-specifics, e.g., the leaf clipping display(McGrew, 1992). Although this may con-tribute to fitness, the tool use is arbitrary andderives its meaning from the universal usewithin a population; signals not involvingtool use are expected to be equally effective.Fifth, objects (often leaves) may be used toclean body parts. This use has beenobserved in most apes and two monkeyspecies (Tomasello & Call, 1997), but againis rare. Sixth, wild chimpanzees and orang-utans make and use tools to extract insectsor insect products or to smash nuts(Goodall, 1986; McGrew, 1992; Boesch,1996; van Schaik et al., 1996; Fox et al.,1998). Some nonhominoid primates also dothis occasionally in the wild, and more oftenin captivity (Beck, 1980; Candland, 1987).

The search for conditions favoring theorigins of hominid technology is probablybest focused on feeding tools, for severalreasons. First, unlike some other forms oftool use, the use of feeding tools can greatlycontribute to fitness in extant pre-agricultural people (McGrew, 1992:131), asin great apes (see below). Second, againunlike some other forms, feeding tool useshows strong phylogenetic continuity:population-wide manufacture and use offeeding tools is found only in hominids and

great apes, i.e., the hominoid clade. Third,many of the tools found in hominid sites arethought to be involved in food processing(Klein, 1989). Finally, feeding tools aremore likely to require modification thanother kinds of tools used by nonhumanprimates; most great ape tool users learnto make feeding tools through some formof social learning, comparable to thoseinvolved in human material culture. Hence,this paper focuses on feeding tools.

The distribution of tool use and manufactureExplanation of the evolution of materialculture requires an overview of the manufac-ture and use of tools among primates in thewild and in captivity. Table 1 (second col-umn) shows the taxonomic distribution ofdocumented feeding tool use in primates.Many older reports, especially of diggingwith tools recounted by Kortlandt & Kooij(1963) and Beck (1980), are not included inTable 1, for lack of sufficiently detaileddescription and confirmation in spite ofextensive subsequent field studies. Nonethe-less, cases of possibly anecdotal or idiosyn-cratic (cf. McGrew & Marchant, 1997) tooluse have been included if reliable primaryreports were available, because they doreflect the capacity of the species. For bothwild and captive information we mustaddress the problem of negative evidence.Published field data now exist for over 150nonhuman primate species (Nunn & vanSchaik, in press), and primates have beenstudied intensively in zoos and institutionsaround the world. Hence, we presume thattool use, if it is routinely present in a species,has been reported in the literature. We alsopresume, therefore, that the distributionalinformation is reasonably complete.

In the wild, only a few species have beenobserved to use feeding tools. There arevarious reports for capuchin monkeys (Cebusspp.). An adult male C. apella was observedusing a piece of oyster shell to pound onother oysters (Fernandes, 1991), a subadult

721

Table 1 The distribution of tool use among non-human primate genera: object-throwing and useof feeding tools

GenusThrowing

objectsFeedingtools*

Otolemur NGalago NGalagoides NNycticebus NLoris NPerodicticus NCheirogaleus NMirza NMicrocebus NLemur NEulemur NHapalemur NVarecia NPropithecus NIndri NDaubentonia NTarsius NCallithrix NCebuella NSaguinus NLeontopithecus NCallimico NSaimiri Y NCebus Y U, CAotus NCellicebus NCacajao NPithecia Y NChiropotes NAlouatta Y NAteles Y NBrachyteles NLagothrix Y NMacaca Y U, CCercocebus CPapio Y U, CMandrillus Y CTheropithecus NCercopithecus Y CErythrocebus Y CColobus Y NPresbytis Y NNasalis Y NHylobates Y CPongo Y U, M, CPan Y U, M, CGorilla Y C

Based on Beck (1980), Candland (1987), andTomasello & Call (1997), and various references men-tioned in the text. Y=present; N=absent; *U=use oftools in the wild; M=manufacture of tools in wild;C=tool use in captivity.

male C. capucinus was once seen to insert astick into a treehole, pull it out and putit into his mouth (Chevalier-Skolnikoff,1990), and a few young C. albifrons wereseen to use leaves to scoop up water fromtree holes (Philipps, 1998). There areseveral observations of long-tailed and lion-tailed macaques (Macaca fascicularis, M.silenus) rolling hairy caterpillars in leavesuntil the stinging hairs are removed (Chiang,1967; Hohmann, 1988). A male baboon(Papio anubis) was once seen using a twig toextract small stone fragments from a stickysoil matrix (Oyen, 1979). Beck (1980)reports on several observations of baboons(especially P. ursinus) using stones to poundon scorpions, etc.

Two features stand out in the feeding-tooluse of wild monkeys. First, none of this tooluse is habitual or customary [in McGrew &Marchant’s (1997) classification] in thepopulations concerned (i.e., shown by manydifferent individuals, regularly or predict-ably). Second, none of it involves manufac-ture: animals use objects as they find them.Thus, only chimpanzees and orang-utansare known to manufacture and use feedingtools on a regular, population-wide basis, inat least some wild populations.

Table 1 shows that in captivity feeding-tool use is more widespread. The publishedreports show remarkable overlap and clus-tering into a few taxa, the non-folivorouscatarrhines and capuchins, for both feed-ing tools and other tools (Beck, 1980;Candland, 1987; Tomasello & Call, 1997).The capacity for tool use (but not necess-arily actual tool use in the wild) thereforeprobably evolved three times: in capuchins,cercopithecines, and apes.

For the two ape species with documentedmanufacture and use of feeding tools in thewild, there is appreciable geographic vari-ation in the overall frequency of the use offeeding tools, in the size of the tool kit usedby the animals, and in the specific tasks forwhich these tools are used (Sugiyama, 1993;

722 . . ET AL.

McGrew, 1994; Boesch & Tomasello,1998). Table 2 furnishes a preliminary sum-mary of geographic variation in chimpanzeefeeding-tool use. Although systematic andquantitative comparisons are still to beundertaken and researchers have differen-tially emphasized the documentation of dif-ferent tool types, the comparison is limitedto only the best-studied sites with well-habituated populations and contains pre-dominantly tool types that are used habitu-ally or customarily. Among orang-utans,feeding tools are so far known only from alimited region in South Aceh, Sumatra,despite extensive studies elsewhere (vanSchaik et al., 1996).

Any successful model for the evolution ofthe use of feeding tools should explain thesefour patterns: (1) why tool use beyondobject-throwing can be observed in only afew primate clades; (2) why routine relianceon feeding tools and manufacture of tools inthe wild is found in only two species of ape;(3) why there is strong geographic variationwithin the two species regularly using feed-ing tools; and (4) why tools are more readilyused in captivity. The aim of this paper is todevelop a model that can account for these

patterns and thus can form the basis for anextrapolation toward the conditions thatgave rise to the extraordinary elaboration oftool use in the hominid lineage.

A model for the evolution of feeding-tool use

Table 2 Functionally different food-related tools used (mostly habitually,sensu McGrew & Marchandt, 1997) in different chimpanzee populations withlong-term studies of well-habituated animals

Tool types Gombe Mahale Kibale Tai Bossou

Leaf sponge # # # #Termite fish # #Ant dip # # #Honey dip # #Nut hammer # #Ant fish #Bee probe #Marrow pick #Pestle pound #Gum gouge #Algae scoop #Hook stick #No. of feeding-tool types 4 2 1 6 7

Based on McGrew (1994), Boesch & Tomasello (1998) and Sugiyama (1993,1997).

Suitable environmental conditions are necessarybut not sufficient. Parker & Gibson (1977)proposed that flexible (‘‘intelligent’’) tooluse was expected in species ‘‘with extractiveforaging on seasonally limited embeddedfoods and an omnivorous diet.’’ Althoughthis hypothesis may seem plausible, extrac-tive foraging alone cannot account for thewhole pattern of tool use: many extractiveforagers that are quite capable of tool useand even some tool manufacture in captiv-ity, for example, capuchins (Westergaard &Fragaszy, 1987), show virtually no sustainedtool use in the wild.

More subtle effects of ecological condi-tions can likewise be discounted as a generalexplanation for the distribution of tool use.While some of the intraspecific variation inthe incidence of feeding tool use in chim-panzees (Table 2) and orang-utans can beascribed to variation in environmental con-ditions (e.g., McGrew et al., 1979), much of

723

it cannot. For chimpanzees, Boesch et al.(1994) and McGrew et al. (1997) compareddifferent sites with and without nut-cracking. They could not find any obviousecological differences among the sites interms of abundance of nut trees orsuitable hammers and anvils. The nutsare so nutritious that optimal foragingexplanations invoking the presence ofmore profitable alternative foods areimplausible. Hence, only geographic bar-riers explained the distribution, and bothstudies therefore concluded that thevariation was cultural.

Orang-utan data similarly contradict apurely ecological hypothesis. Neesia seedsare embedded in irritant hairs that pierce theskin after the fruits have cracked open. AtSuaq Balimbing on Sumatra, adult males eatthe seeds before the fruit ripens and dehiscesby using brute force to open the fruits andthus obtain the seeds. After the fruits open,all age–sex classes use tools to extract theseeds, and thus circumvent the hairs,although adult males occasionally continueto break the woody valves off the fruit. Incontrast, in Gunung Palung on the island ofBorneo, adult males are the predominantNeesia eaters, most of their feeding is onundehisced fruits that are opened with force,and all feeding is without tools. This strongcontrast could be ascribed to the fact thatother food species in Gunung Palung aremore profitable to the local orang-utans.However, these seeds are very nutritious(46% lipids by dry weight) and may provideup to 4·5 Kcals per day to adult males beforethey dehisce (C. Knott, personal communi-cation). At Suaq Balimbing, Neesia is thepredominant fruit species for adult femalesand subadult males during several months.An unusual sequence of closely spacedNeesia crops is probably responsible for abirth peak in the population (van Schaik,unpublished data). Hence, it is implausibleto argue that the ability to extract Neesiaseeds is not energetically important for the

Gunung Palung orang-utans (Knott & vanSchaik, in prep.).

Both the chimpanzee and orang-utan dataindicate that the absence of tool use in agiven population reflects the absence ofknowledge within that population, ratherthan the absence of suitable ecologicalopportunities. Thus, we must also focus onthe process through which the tool-usingskill is acquired. Ecological explanationsimplicitly assume that these skills developthrough maturation (cf. instinct) or areinvented independently through some formof learning. However, acquisition throughsimple maturation is unlikely because atleast some of the skills are highly complex(Boesch, 1993) and the contexts are varied.Acquisition through independent discoveryof tool-using skills is unlikely for the samereasons and because of the uniformitywithin populations relative to the variabilitybetween populations (Boesch, 1996). Thepatterns therefore suggest that sociallearning is the predominant mode of skillacquisition in most individual apes in thewild.

A socioecological model for tool use. A completemodel of the evolution of tool-using skills inprimates should therefore contain the fol-lowing elements: (1) ecological opportuni-ties for feeding tool use; (2) sufficientlyprecise motor control for effective handlingof objects; (3) the appropriate mental capa-bilities to invent or rapidly acquire, throughsocial learning, appropriate tool-handlingskills; and (4) social conditions that areappropriate for the social transmission ofskills. We hypothesize that an increasingnumber of these essential elements is repre-sented in ever narrowing subsets of species(Figure 1). Thus, we expect that a subset ofextractive foragers will have the manipula-tive abilities needed to develop tool use, andthat a subset of those dexterous species willhave the required cognitive abilities andexperience the right social conditions to

724 . . ET AL.

EVOLUTION OF MATERIAL CULTURE

extractive foraging

dexterous manipulation

intelligence(insight, imitation)

tolerantgregariousness

teaching +exchange ?

many primates

cebus, apes,cercopithecines

great apes

some chimps,some orang-utans

humans

extraction

tool using(captivity)

tool making + incid.tool use (wild)

population-widetool use in wild

(material culture)

cumulativeculture

Figure 1. The nested set of conditions hypothesized to favor feeding-tool use, from ecological opportu-nities and manipulative abilities as preconditions to cognitive factors and social conditions favoring theinvention and transmission of tool-using skills in captive or wild populations. On the left side, a summaryis given of the taxa meeting the ever-narrowing requirements, on the right side the phenomena shown bythe taxa mentioned on the left.

actually invent or learn tool use and toolmaking in the wild.

The general precondition for all use offeeding tools is a foraging niche that involvesextraction (Parker & Gibson, 1977). Mostfoods eaten by primates are freely accessible,and can be picked with the hands or takendirectly with the mouth. However, somefood is encased in a matrix from which itneeds to be extracted, e.g., insects in treeholes, nutmeat from a hard shell or cormsfrom underground. There are three reasonsto expect a strong interspecific associationbetween extractive foraging, complexmanipulative abilities, and feeding-tool use.First, almost all feeding tools in primates areused in an extractive foraging context.Second, almost inevitably, extractive forag-ing requires some level of manipulative skill.Where animals engage in a variety of distinct

extractive tasks, flexible behavior is neededrather than specialized anatomical struc-tures. Hence, flexible extractive foragers aremost likely to be capable of manipulationsthat require fine neuromotor control. Third,more dexterous species are more likely thanless dexterous ones to discover tool use as afortuitous effect of their greater diversity andeffectiveness of manipulations, especially ifthe manipulations involve actions placing anobject in relation to a substrate (Menzel,1966; Fragaszy & Adams-Curtis, 1991).The fact that the invention of tool useby individuals is often accompanied byexploratory or playful actions (Fragaszy &Visalberghi, 1989) supports this idea.

Our model incorporates invention andtransmission as factors limiting the inci-dence of feeding-tool use in gregarious pri-mates that forage extractively. Specifically,

725

CONDITIONS FOR TOOL USE INDEXTEROUS EXTRACTIVE FORAGERS

invention transmission

learningthroughinsight

imitation,emulation

intelligence

tolerance gregariousness

social organization

Figure 2. The social and cognitive factors hypothesized to favor the invention and social transmission offeeding-tool use in species physically capable of tool use.

we propose that at any given level of cogni-tion, the probability that an individualinvents a new behavior, is a function of thefrequency with which the potential forexpression occurs. Likewise, at any givenlevel of cognition, the probability that anaive animal will acquire a skill throughsocial transmission is a function of the fre-quency of opportunities for transmission.This in turn depends on how often theexpression of the skill takes place in a contextenabling transmission. Thus, two indepen-dent sets of factors are involved: the socio-ecological conditions in which invention ortransmission are taking place, and the cogni-tive mechanisms underlying the invention ortransmission process (Figure 2).

Consider invention first. Invention israrely witnessed, so we know very littleabout the conditions favoring it (Kummer &Goodall, 1985), although one expects thatin most cases, an operant behavior and asuitable target object or constellation of

objects must be present simultaneously.Socioecological context and cognitionshould also affect the rate of invention. Withrespect to socioecology, the subject’s moti-vation to explore should not be hindered bysocial or ecological constraints. Factorsenhancing the freedom to explore arereduced vulnerability to predation, and flex-ible spatial organization of individuals sothat animals often feed alone, or, alterna-tively, high social tolerance. With respect tocognition, intelligent species could make theconnection between an action upon anobject and the desired outcome withoutstumbling upon the proper action throughoperant behavior alone, i.e., intelligentspecies can learn by insight (e.g., Byrne,1995). Because the conditions favoringinvention are a subset of those favoringsocial transmision of complex skills (see Fig-ure 2), we focus on the more easily studiedaspect of transmission, which is critical formaintenance of tool use in a population.

726 . . ET AL.

A new skill, once invented, is maintainedin a population only if it spreads beyond theinventor. Obviously, most social transmis-sion in primates is vertical, i.e., from motherto offspring. However, unless one is willingto assume that a new skill, say a form of tooluse, becomes established in a populationbecause the matriline of the inventor out-reproduces all other matrilines and elimi-nates them from the population, horizontaltransmission is critical for the establishmentof a population-wide technique. Thus, theconditions favoring horizontal transmissionmust be considered. The probability that anaive animal will socially acquire a skillshould be a function of the frequency ofopportunities to be present at close quarterswhen models show skilled tool use and ofthe efficiency with which this animal canlearn from these exposures (cf. Westergaard& Fragaszy, 1987).

Social learning obviously requires a cer-tain degree of gregariousness, which makesit unlikely in solitary or semi-solitary animalsthat meet infrequently (Figure 2). It alsorequires close proximity in a feeding con-text. This is not a trivial condition sincemost primates spread out during foragingfor insects and other dispersed food, thecontext for most extractive tool use (e.g.,van Schaik & van Noordwijk, 1986), andaggression toward subordinates is likely ifthey come too close to dominants. It hasbeen shown experimentally that animalsunder risk of attack by conspecifics do notlearn well (Fragaszy & Visalberghi, 1990;cf. Inoue-Nakamura & Matsuzawa, 1997).Hence, social tolerance in a potentially com-petitive foraging context is likely to stronglyfacilitate social learning (Coussi-Korbel &Fragaszy, 1995). It should do so in at leastthree ways: (1) it enhances behavioral co-ordination, and thus may bring animals inclose proximity during foraging, makingclose-range observations possible; (2) itreduces anxiety levels, making it easier foranimals to attend to the foraging tasks

(rather than to the risk of attack); and (3) itreduces the risk of losing food to scroungers(cf. Fragaszy & Visalberghi, 1990). Toler-ance is expected in a society when its mem-bers, despite competing among themselves,are in some way mutually dependent, forexample, because they must defend againsta common enemy [predators, neighboringgroups, or bullying males (van Schaik, 1989;Brereton, 1995)]. (This principle could beextended to the risk of predation, whereanimals may be able to attend to actions byothers more often or for longer where theneed for vigilance is reduced.)

Social learning could be accomplishedafter fewer exposures if the learners areintelligent (Figure 2). Intelligent animalsoften learn by emulation or imitation(Byrne, 1995), methods which require fewexposures. Other animals learn less effi-ciently through stimulus or local enhance-ment, in which the naive animal is attractedto the situation in which stimulus and thetool co-occur and then acquires the skill bytrial and error (Galef, 1990; Byrne, 1995).

An intelligent animal in a tolerant societywould require the fewest exposures to learna new skill. Natural selection should favorintelligence and social tolerance wheneverthe ability to use tools is adaptive. However,we do not know how much these twovariables had to have changed from theiroriginal state before they produced fitnessbenefits through tool use. Hence, it is moreparsimonious to argue that intelligence andsocial tolerance are more likely to act aspreconditions in whose presence thereis a highly increased probability that anextractive forager will learn to use feedingtools.

PredictionsWe can now derive the following testablepredictions from the model.( 1) Extractive foraging: species showing

dexterous manipulation should be asubset of extractive foraging species.

727

( 2) Manual dexterity: all species in whichfeeding-tool use has been observed, beit in the wild or in captivity, shouldshow dexterous foraging skills in thewild (cf. Parker, 1974; Torigoe,1985).

( 3) Intelligence: intelligent species, definedhere as those which can learn throughinsight and emulation or imitation,require fewer exposures to acquiretool-using skills, be it through inven-tion or through social learning. Hence,we predict that intelligent species (a)are the only ones to show population-wide tool use in the wild, and (b) tomanufacture tools in the wild. We alsopredict that (c) only some captive set-tings will provide enough opportuni-ties for invention or exposure to skilledusers to allow the acquisition oftool use in species lacking insight orobservational learning capabilities.

( 4) Social tolerance: social tolerance shouldbe correlated with exploratory behav-ior, and thus invention and (when ani-mals are gregarious) transmission oftool use. In particular, the intraspecificvariation among great ape populationsin the size of the tool kit should corre-late with variation in social toleranceor in gregariousness.

Although some of the predictions under(3) were known to hold approximatelybefore the study, they are examined in moredetail here. All of them will either be testedby comparative methods or evaluated byreviewing existing studies.

Methods

Measures

Extractive foraging. In order to assess speciesfor extractive foraging we examined numer-ous publications detailing primate feedingtechniques (see below). Our preliminarywork, however, showed that virtually everywell-studied species engages at least occa-

sionally in extractive foraging (cf. King,1986). In order to test our prediction, there-fore, we needed more quantitative measuresof extractive foraging, including the fre-quency, complexity and diversity of extrac-tive foraging tasks. Unfortunately, suchinformation cannot be gleaned from pub-lished foraging accounts, so we could nottest the extractive foraging prediction.

Manipulative abilities. To test the predictionsinvolving manipulative ability, we used threemeasures: manipulation pattern diversity,object-substrate combination, and bimanualasymmetric coordination (BAC).

Torigoe (1985) presented a knotted ropeand wooden cube to 74 species of captiveprimates, and recorded for each species (1)the diversity of manipulation patterns (e.g.,rubbing, rolling or rotating the rope), and(2) the diversity of object–substrate combi-nations (e.g., rubbing a rope toy on a wiremesh wall, pushing the toy into thewall). Torigoe called these primary and sec-ondary manipulation patterns, respectively.Torigoe’s study was a replication and exten-sion of a smaller earlier study by Parker(1974), which is not used here. The twodata sets are in general agreement (Parker’saction types and Torigoe’s primary manipu-lations are correlated with Pearson’sr= +0·72, n=8, P<0·05, two-tailed).

We define bimanual asymmetric coordi-nation (BAC) as using the hands to performdifferent but complementary actions on adetached object, e.g., grasping a fruit withone hand and peeling it with the other. BACindicates manipulative skill because itinvolves the independent but complemen-tary execution of motor programs (cf.Goldfield & Michel, 1986; Byrne & Byrne,1993; Fagard & Pezé, 1997). We collecteddata on BAC by reviewing published pri-mate field studies which described foragingtechniques because preliminary work sug-gested that BAC in non-foraging contexts iseither extremely rare or is not mentioned in

728 . . ET AL.

the literature. We scored a species as engag-ing in BAC if an account clearly indicatedthat an individual used two hands simul-taneously and asymmetrically in procuring adetached food. Ambiguous accounts werediscarded, as were accounts of BAC duringtool use (e.g., McGrew, 1974), in order toguarantee independence of the subsequenttests. Most commonly, BAC involved ananimal holding a partially processed foodwith one hand, e.g., a handful of herbs stillretaining thorns, and picking at it with theother. Actions such as bimanual digging andfood rubbing were not counted unless it wasclear that the two hands were used in differ-ent roles (see Table 3, first column). Weattempted to be comprehensive in reviewingthe published information on primate feed-ing techniques: (1) we systematically exam-ined four primate journals (Primates, FoliaPrimatologica, International Journal of Prima-tology, and the American Journal of Primatol-ogy), carefully reading every English articlewhich seemed even remotely likely to men-tion feeding techniques; (2) we reviewedseveral monographs and edited volumeswhich seemed likely to detail feedingtechniques; (3) we examined additionalresources as we became aware of their rel-evance; and (4) we included our personalobservations where relevant. Negative evi-dence is problematic in the absence ofreports explicitly stating that BAC does notoccur. We therefore excluded taxa whichhave been poorly studied in the wild. Thiswas quantified by searching the BiologicalAbstracts database between 1986 and 1997,entering the words ‘‘ecology’’ for eachtaxon. Only those taxa for which morethan ten records were available were scorednegatively.

Measures of social tolerance. Two classes ofmeasures of tolerance can be used. Directmeasures are social: (1) relaxed proximityand grooming, accompanied by low mutualaggression and a high conciliatory tendency

(e.g., Aureli et al., 1997); and (2) the degreeto which aggressive actions within a dyadcan go in both directions (Thierry, 1985).Other measures stress the nature of feedingcompetition: (1) frequency of food sharing;and (2) habitat productivity. Both will beused in the tests as needed and available.

AnalysesFor many of the predictions it would bevirtually impossible to conduct statisticaltests at the species level since there is littleoverlap among the measures in terms ofspecies availability. We overcame this prob-lem by pooling information at the level ofthe genus. Torigoe (1985; Figures 1, 6)presents information for species groups, i.e.,species of the same genus or of closelyrelated genera. We incorporated this infor-mation by assigning to each genus repre-sented in the species group the overall scoreof the species group, although in statisticalanalyses the original sample size wasretained. Pooling information from lowerorder taxa is obviously not ideal; however,the possible heterogeneity among specieswithin genera should make our tests moreconservative.

Because of the low resolution of the data(BAC being discrete and Torigoe’s datareferring to species groups), we combinedour three manipulation measures, which arehighly intercorrelated (see Table 3), into anoverall classification of high or low onmanipulation abilities. A genus was scoredas high if it was on or above the median forobject–substrate combination or manipula-tion pattern diversity (the two measures pro-duce identical classifications), or if it wasscored as engaging in BAC (see Table 3).The only conflict between Torigoe’s (1985)measures and BAC concerned Cercopithecus.Since the BAC evaluation is based upon aliterature search rather than direct measure-ment, it is more likely to be in error. Hence,we classified Cercopithecus as being highlymanipulative.

729

Table 3 Bimanual asymmetric coordination (the number of literature references for ecology, numberfound providing BAC, and source), and manipulation repertoire (mean number of primary andsecondary manipulations from Torigoe, 1985) for primate genera. See text for criteria to decide on highor low manipulation skills

Genus Refs BAC Primary Secondary Evaluation Source

Daubentonia 10 1 High 1Hapalemur 3Eulemur 8 15·45 0·33 LowLemur 7 15·45 0·33 LowIndri 1Propithecus 5Varecia 3 15·45 0·33 LowTarsius 1Cebuella 2Callimico 1Callithrix 13 0 9·51 0 LowSaguinus 30 0 9·51 0 LowLeontopithecus 9Callicebus 7Aotus 6Chiropotes 2Pithecia 4Saimiri 11 0 LowCebus 21 2 58·94 3·75 High 2Alouatta 18 0 LowLagothrix 2 6·7 0 LowBrachyteles 4Ateles 14 0 6·7 0 LowCercopithecus 41 0 33·25 1·9 High?Erythrocebus 3 1 High 3Cercocebus 6 1 33·33 2·16 High 4Mandrillus 5 32·6 1 HighTheropithecus 4 32·6 1 HighPapio 39 2 32·6 1 High 5Macaca 65 2 54·63 3·36 High 6Colobus 26 0 15·12 0 LowPresbytis/Trach. 31 0 15·12 0 LowNasalis 6 15·12 0 LowHylobates 13 0 31·71 0 LowPongo 10 1 68·54 7 High 7Pan 211 2 68·54 7 High 8Gorilla 32 3 68·54 7 High 9

1: Iwano & Iwakawa, 1988; 2: Izawa & Mizuno, 1977; Tomblin & Cranford, 1994; 3: Hall et al., 1965; 4: Horn,1987; 5: Rhine & Westlund, 1978; Hamilton & Tilson, 1985; 6: Kawai, 1965; Clark, 1979; 7: van Schaik, personalobservation; 8: Bermejo et al., 1994; Uehara, 1990; 9: Casimir, 1975; Goodall, 1977; Byrne & Byrne, 1993.

The patterns were analyzed using stan-dard statistical tests. However, similarity invalues among related genera may merelyreflect common ancestry, and thus generamay not be statistically independent(Harvey & Pagel, 1991). For some of theanalyses, we can couch the hypotheses in

terms of correlated trait evolution. Wetested the evolutionary correlation betweendiscrete traits using Maddison’s (1990) con-centrated changes test, which evaluates theprobability that observed evolutionary gainsand losses in tool use are concentrated onthose branches of the phylogeny with high

730 . . ET AL.

manipulation skills, given the total numberof reconstructed changes in tool use in theclade examined.

Results

Dexterous manipulation and tool useSince extractive foraging in nature could notbe quantified for this study, we could nottest the first prediction. The second predic-tion suggests that all known tool users(Table 1) are also dexterous manipulators(Table 3); or more conservatively that thetwo are strongly associated. The associationis indeed quite strong (X2

[1]=11·69,P<0·001). While only 8% of the low-dexterity genera use tools in the wild orin captivity (exception: gibbons), 83% ofthe high-dexterity genera do (exceptions:geladas and aye-ayes) (see also Figure 3).Gibbons (Hylobates spp.) are not highly dex-terous, scoring just under the median inTorigoe’s tests; their tool use is very simplein captivity (see Beck, 1980) and absent innature. Aye-ayes (Daubentonia) are highlydexterous but do not use tools; their highlyspecialized anatomical foraging adaptationsmay obviate the need to use tools. We canoffer no explanation for the absence of tooluse in geladas (Theropithecus). [Tool use isalso predicted well by the three separatemeasures for dexterity. Separate Mann–Whitney U-tests show significantly largerrepertoires of primary and secondarymanipulations (Torigoe, 1985) for toolusers than non-tool users, whereas there isalso significant association between tool useand BAC.]

The association between dexterity andtool use is not simply due to phylogeneticinertia. Figure 3 shows the reconstructedevolution of high manipulation skills. It sug-gests four origins: in the aye-aye, in capu-chins, in cercopithecines and in great apes.At the tips of the branches in Figure 3, wealso map the distribution of tool use. TheMaddison test shows that the probability of

the origins of tool use being independentof dexterous foraging skills is vanishinglysmall (P<0·001). The association remainssignificant if we use slightly different assign-ments (changing the dexterity score forCercopithecus; cf. Table 3), or slightlydifferent reconstructions of the gains andlosses in tool use. Thus, this analysis con-firms the direct test of association conductedabove.

Intelligence and tool use: great apes vs.monkeysSo far, despite all attempts to prove other-wise, only great apes are known to showunderstanding of the physical relationbetween tools and other objects in theenvironment in their tool use (Limongelliet al., 1995; Visalberghi et al., 1995; cf. Foxet al., 1999). There is also evidence that apesimitate (Russon, 1997; Whiten, 1998) andextensive evidence that they emulate (e.g.,Tomasello & Call, 1997). Tomasello &Call’s (1997) recent compilation of allstudies of social learning in primates indi-cates that the cognitively more compleximitation or emulation mechanisms forsocial learning are almost exclusively shownby great apes: 46% of 35 experiments withthe four great ape species yielded evidencefor emulation or imitation, whereas only3% of 29 studies showed these processesto operate in non-apes. The othersdemonstrated trial and error or stimulusenhancement.

Byrne (1995) and Russon & Bard (1996)suggested that the great apes have acquiredgeneral intelligence to deal with new andunexpected problems, as opposed to thedomain-specific cognitive skills displayed bymonkeys, expressed in a variety of unusualcognitive skills (e.g., self recognition,pretend play, comprehension and someproduction of language, insight, tacticaldeception, mental state attribution, etc.).Although Tomasello & Call (1997) didnot accept this dichotomy, they did note

731

absent

OtolemurGalagoGalagoidesNycticebusLorisPeridicticusArctocebusPhanerCheirogaleusMirzaMicrocebusAllocebusAvahiIndriPropithecusVareciaEulemurHapalemurLemurLepilemurDaubentoniaTarsiusCebuellaCallithrixLeontopithecusSaguinusCallimicoSaimiriCebusAotusCallicebusPitheciaChiropotesCacajaoAlouattaLagothrixAtelesBrachytelesPresbytisNasalisSimiasPygathrixRhinopithecusColobusProcolobusCercopithecusErythrocebusMiopithecusAllenopithecusMacacaTheropithecusCercocebusPapioMandrillusHylobatesPongoGorillaPan

present

Complex manipulation:

Phylogeny of complex manipulation,in relation to use of feeding tools

Use of feeding tools

Complex manipulation

Figure 3. Reconstruction of the phylogenetic changes in complex manipulation, or dexterity, in primates(based on Table 3; phylogeny based on Purvis, 1995). Also mapped onto the tips of the phylogeny is theuse of feeding tools (taken from Table 1).

(p. 376) that apes are quicker and moreflexible learners. For our purposes, thisamounts to a similar difference.

This major difference in learning abilitiesbetween apes and monkeys may explainvarious patterns in primate tool use. First,

732 . . ET AL.

extensive tool use in the wild is known onlyamong great ape species. Social learning oftool use is more easily achieved when ani-mals use imitation or emulation rather thanlocal enhancement or stimulus enhance-ment, the processes known to be used bymonkeys (Byrne, 1995; Russon, 1997).Tool use spreads remarkably slowly ingroups of monkeys (Beck, 1980; Petit &Thierry, 1993; Zuberbühler et al., 1996).Moreover, in none of the cases of feeding-tool use described for wild monkeyswas there any evidence that the tool usehad spread throughout the population(Oyen, 1979; Hohmann, 1988; Chevalier-Skolnikoff, 1990; Fernandes, 1991; Phillips,1998; see also Sinha, 1997).

Second, tool making is expected to bemore common in great apes than in mon-keys because mental representation of thetask should make tool manufacture fareasier. Learning to use an existing object toacquire food may be relatively straightfor-ward to achieve through positive reinforce-ment of some operant behavior. However,modifying an object to reach a certain goalmay require hierarchical planning and arepresentation of the goal. All great apesmake tools in captivity, and two species(chimpanzees and orang-utans) do so in thewild, whereas there is only limited evidencefor monkeys making feeding tools in cap-tivity (reviewed in Tomasello & Call, 1997)and one report of tool manufacture in thewild (Sinha, 1997). Thus, support for thisprediction is rather weak.

Third, intelligence may indirectly explainthe phenomenon that many species without(reported) sustained tool use in the wild canacquire these skills in suitable captivesettings. The model suggests that specieslacking great ape like intelligence neednumerous exposures to learn the requiredskills through trial and error (cf. Box &Fragaszy, 1986). Most situations requiringtool use in the wild are uncommon andintermittent in time (e.g., seasonal), and

animals tend to be spread out most of thetime during foraging. By contrast, the expo-sure to potential tool-use situations in cap-tivity is continuous, as is (enforced) closeproximity to any models that exist. More-over, captive animals face neither foragingpressures nor predators, and should there-fore be more predisposed toward playand exploration, facilitating invention andtransmission. Accordingly, Menzel (1966)showed that captive Japanese monkeys weremore inclined to manipulate offered objectsthan their wild counterparts. Likewise,Fragaszy & Adams-Curtis (1991) showedthat captive capuchins had more diverseobject handling techniques.

Tolerance and tool useThe hypothesis predicts that when affordedgreater social tolerance, animals are betterable to concentrate their attention on explo-ration without interference or aggression,and thus are more likely to invent or learntool-using skills. There are three mainsources of support for this hypothesis. First,tolerant species spend more time manipulat-ing novel objects in captivity. Macaquesvary widely in tolerance (Thierry, 1985;Preuschoft, 1995; Aureli et al., 1997). Com-bining data from Thierry (1985) withThierry et al. (1994), the three Macacaspecies in the sample show a perfect positiverank correlation between tolerance andtime spent manipulating novel objects. Inaddition, the most manipulative macaque,M. silenus, is also socially highly tolerant(Preuschoft 1995).

Second, experiments to induce feedingtool manufacture in monkeys have suc-ceeded primarily in tolerant species (seeTomasello & Call, 1997, Figure 3.5): Cebusapella, Macaca tonkeana and M. silenus (tol-erance data from de Waal, 1997). Oneother case of experimentally inducedtool manufacture is for a non-tolerantspecies (M. fuscata: Tokida et al., 1994), buthere the only tool maker was the daughter

733

of the alpha female, who would not besubject to harassment while concentratingon a task.

Third, differences in tolerance may leadto intraspecific differences in the inventionof novel tool-use behaviors. Althoughcapuchins are generally tolerant and readilyshare food (de Waal, 1997), Fragaszy &Visalberghi (1990) note that the high socialvigilance in larger capuchin groups inhibitsexploration of novel tool-use apparatusesand hinders the development of novel tool-use behaviors. In several species, the naiveindividuals who are first to learn tool-useor food-processing techniques usually haveaffiliative relationships with knowledgeablegroupmates, so that relaxed proximitywith a demonstrator is possible (Itani &Nishimura, 1973; Westergaard, 1988;Zuberbühler et al., 1996).

Intraspecific variation in tool use

Chimpanzees. As noted above (see Table 2),there is appreciable variation in the kinds oftools used and the numbers of differentfeeding tools used across chimpanzee popu-lations. Given an approximately equal rateof invention among sites and an approxi-mately equal number of ecological opportu-nities for feeding-tool use, the hypothesispredicts that the size of the local tool kitreflects the frequency of opportunities fortransmission, and hence gregariousness orsocial tolerance. Boesch (1996) has com-pared party sizes for most sites with long-term chimpanzee studies. Mean party size isnot correlated with the size of the toolkit (rs= "0·10, n=5, P=0·84). Tolerancemeasures can not be directly comparedacross sites. We therefore compiled varioustolerance indicators (Table 4). Their distri-bution over sites is consistent, allowing fora classification of the tolerance into threeclasses. The composite tolerance score andtool kit size show a strong correlation(rs= +0·95, n=5, P=0·057), consistent withthe hypothesis that more tolerant popu-

lations of chimpanzees have more variedtool use.

Orang-utans. The use of feeding tools inorang-utans is limited to a few sites, all insouth Aceh province in Sumatra. Becauseorang-utans are semi-solitary, both gregari-ousness and tolerance could vary. Inaddition, opportunities for tool use, i.e.,the extent of extractive foraging, couldvary. One population with tool use (SuaqBalimbing) has been studied in detail (vanSchaik et al., 1996; Fox et al., 1999), as haveseveral others without tool use.

Orang-utan densities in Sumatra areabout twice that in Borneo in similar habi-tats (van Schaik et al., 1995), suggesting thatSumatran forests are more productive thancomparable Bornean ones. As a result, gre-gariousness and tolerance also vary. Adultorang-utan females seem to be far moregregarious at the Sumatran sites than at theBornean ones (van Schaik, 1999). However,the importance of feeding aggregations inthe mean party size is much greater inKetambe, where large strangling figs areabundant, than in Suaq Balimbing, wherevirtually all parties are travel bands (sensuSugardjito et al., 1987) because large fruittrees with long patch residence time are rare.Although travel bands are more likely thanaggregations to encounter opportunities forextractive foraging, the differences betweenSuaq Balimbing and Ketambe in gregarious-ness per se are not dramatic. With respect totolerance, numerous cases of food sharingoutside the mother–infant context have beenobserved at Suaq Balimbing (in ca 12,000hours of focal observations: van Schaik,1999), both in male–female and female–female associations, and including fruit,insect nests and meat. At Ketambe, foodsharing is occasionally observed (S. Utami,personal communication, similar obser-vation time). By contrast, no food sharingoutside the mother–infant context is foundin Tanjung Puting (Galdikas & Teleki,

734 . . ET AL.

Tab

le4

Ind

icat

ors

ofso

cial

tole

ran

cein

the

chim

pan

zee

pop

ula

tion

sli

sted

inT

able

2.R

anki

ngs

are

from

low

toh

igh

tole

ran

ce.

Con

clu

sion

isb

ased

onm

ean

rela

tive

ran

ks,

wh

ich

are

hig

hly

con

sist

ent

amon

gin

dic

ator

s

Sit

eIn

terb

irth

inte

rval

*%

lone

indi

vidu

al†

Mea

tsh

arin

g‡%

hunt

sco

llabo

rati

ve§

Fem

ale

groo

min

gQM

edic

inal

plan

tus

e¶C

oncl

usio

n

Kib

ale

11

1L

owG

ombe

32

11

22·

5M

ediu

mM

ahal

e2

32

12·

5M

ediu

mT

ai5

23

Hig

hB

osso

u4

43

Hig

h

*Int

erbi

rth

inte

rval

isan

inde

xof

habi

tat

prod

ucti

vity

,w

hich

isas

sum

edto

enha

nce

tole

ranc

e.S

hort

est

inte

rbir

this

rank

edas

high

est

tole

ranc

e.D

ata

from

Tut

in(1

994)

,w

ith

Kib

ale

from

Wra

ngha

met

al.

(199

6).

†%of

part

ies

cons

isti

ngof

lone

indi

vidu

als,

also

anin

dica

tor

ofha

bita

tpr

oduc

tivi

ty.

Few

est

lone

indi

vidu

als

isra

nked

ashi

ghes

tto

lera

nce.

Dat

afr

omB

oesc

h(1

996)

,w

ith

Bos

sou

adde

dfr

omS

ugiy

ama

(198

1).

Thi

sm

easu

reis

espe

cial

lyre

leva

ntfo

rad

ult

fem

ales

;be

caus

eth

eyar

eth

ele

ast

greg

ario

usag

e–se

xcl

ass,

vari

atio

nam

ong

fem

ales

ingr

egar

ious

ness

shou

ldca

use

mos

tof

the

vari

atio

nam

ong

site

s.‡M

eat

shar

ing

isa

dire

ctin

dica

tor

ofto

lera

nce.

Rat

esw

ere

com

pare

dby

Boe

sch

&B

oesc

h(1

989)

.¶I

nco

llabo

rati

vehu

nts,

seve

ralh

unte

rsac

tto

geth

eran

dad

opt

com

plem

enta

ryro

les

(Boe

sch

&B

oesc

h,19

89),

whi

chis

only

expe

cted

whe

nsh

arin

gis

com

mon

and

volu

ntar

y.QF

emal

e–fe

mal

egr

oom

ing

isan

expr

essi

onof

fem

ale

tole

ranc

e.M

easu

reus

edis

obse

rved

/exp

ecte

dra

tios

whe

re>

100

groo

min

gbo

uts

wer

ere

cord

ed(M

uroy

ama

&S

ugiy

ama,

1994

).¶M

edic

inal

plan

tus

eis

cult

ural

beha

vior

.N

umbe

rof

spec

ies

used

med

icin

ally

isa

refle

ctio

nof

oppo

rtun

itie

sfo

rcl

ose-

rang

eob

serv

atio

n.D

ata

from

Huff

man

&W

rang

ham

(199

4).

735

30

0Kut

20

25

15

10

5

TP US Ren Ket SB

Vegetative material

Borneo Sumatra

20

0Kut

15

10

5

TP US Ren Ket SB

Cambium

Borneo Sumatra

100

0Kut

50

75

25

TP US Ren Ket SB

Fruit20

0Kut

15

10

5

TP US Ren Ket SB

Insects

Figure 4. Variation in dietary composition, as estimated by percentage of total feeding time, amongvarious wild orang-utan populations on Borneo and Sumatra. Kut=Kutai; TP=Tanjung Puting; US=UluSegama; Ren=Renun; Ket=Ketambe; SB=Suaq Balimbing. SB is the site with tool use. Data taken fromRodman (1988), supplemented by Suzuki (1989) for Kut, Sugardjito (1986) for Ket and van Schaik(unpublished) for SB.

1981), nor do any other reports on Borneanorang-utans mention it (also C. Knott,personal communication). Hence, theSumatran sites, and especially SuaqBalimbing, have higher gregariousness andtolerance in foraging situations where tooluse is likely, consistent with the hypothesis.

Opportunities for extractive tool use mayalso vary, however. Time spent foraging oninsects is the best estimate for the frequencyof opportunities for extractive foraging:much insectivory is extractive (tool use isrelatively rare and thus unlikely to cause the

time spent foraging for insects). The pub-lished data on orang-utan time budgetsshow considerable variation in the distri-bution of feeding time (Figure 4), with theSumatran populations spending more timeon insect foraging and the Suaq Balimbingpopulation spending by far the most. Thishigher tendency toward insectivory inSumatra is shared by gibbons (Palombit,1997) and long-tailed macaques [compareWheatley (1980) with van Noordwijk & vanSchaik (1987)], and reflects the productivitydifference between the islands. Thus, the

736 . . ET AL.

site with tool use is also the site with themost frequent opportunities during routineforaging, which may provide an alternativeexplanation for the observed distribution oftool use. However, the absence of Neesiatool use in Borneo, despite abundant oppor-tunities, is in favor of the hypothesisthat tolerance during foraging is the keyfactor in the distribution of tool use inorang-utans.

Discussion

Primate tool useAlthough extractive foraging and tool useare often thought to be linked (Parker &Gibson, 1977), we were unable to test thislink directly. However, other evidence doesnot support Parker & Gibson’s hypothesis.First, while all feeding-tool users in the wildor in captivity are dexterous extractive for-agers, many species that show extractiveforaging are not dexterous, and most of thedexterous extractive foragers do not usetools in the wild with any frequency.Second, there is doubt as to the theoreticalbasis for linking extractive activities to flex-ible tool use through the use of Piaget’ssensori-motor stages (cf. Fragaszy & Adams-Curtis, 1991; Byrne, 1997; Tomasello &Call, 1997). Thus, extractive foraging is notdirectly linked to tool use.

The phylogenetic test showed a strongdependence of tool use (mainly in captivity)on the presence of manual dexterity.This is not unexpected if we view tool use asthe byproduct of manipulative actions,especially those combining objects and sub-strates. Indeed, the tool use of many speciesin captivity shows many similarities tofrequently shown operant behaviors. Forinstance, capuchin monkeys often poundobjects, e.g., smashing nuts together oragainst branches (Struhsaker & Leland,1977); where they use tools, they also tendto use these pounding actions (Fernandes,1991; Fragaszy & Adams-Curtis, 1991).

Thus, extractive foraging has favored theevolution of dexterity in a subset of speciesand provides the tasks in which tools canprofitably be used. However, as illustratedin Figure 1, ecological factors alone, whilenecessary, are not sufficient to explain thedistribution patterns of primate tool use, andit is best to regard the required capacitiesas being present or expressed as skills inever-narrowing subsets of species orpopulations.

Intelligence may account for the presencein some great apes of the use of feeding toolsin the wild, the presence of tool kits and toolsets (Parker & Gibson, 1977; McGrew,1992), and perhaps a greater ease of toolmaking (Figure 1). Unlike great apes, mon-keys lack insight and efficient observationallearning techniques, and therefore are notknown to show population-wide tool use inthe wild, although they can acquire tool-using skills if opportunities to do so arenumerous and continuous. It is theoreticallypossible, therefore, that a socially tolerantmonkey population will be found in whichroutine use of feeding tools occurs,especially if the skills used are close tonaturally occurring operants, and ifopportunities for would-be learners areabundant.

The causes of the cognitive dividebetween great apes and monkeys are notclear (Byrne, 1997). The cognitive factorsfavoring invention and transmission, thoughdifferent (Figure 2), are evolutionarilylinked because copying the behavior ofothers only provides a benefit if these indi-viduals have developed skills that are worthcopying (Boyd & Richerson, 1996). Thus,logically, insight historically preceded obser-vational learning skills such as imitation andemulation. However, given the erratic distri-bution of tool use in living great apes, it isnot likely that the intellectual capacity fortool use itself provided the selective forcethat produced more generalized cognitiveskills in great apes.

737

Tolerance is a critical element of themodel (cf. Figure 1). Although the socialenvironment is often mentioned as a factoraffecting learning latencies, tolerance has sofar rarely been used to account for variationamong species or populations within aspecies (Coussi-Korbel & Fragaszy, 1995).Yet there was support for a role of socialtolerance. In monkeys social toleranceincreases the time spent exploring objectsand leads to more likely acquisition of tool-using skills. The geographic pattern amongorang-utans is also consistent with a criticalrole for social tolerance in maintaining tooluse at a site. The geographic variation inchimpanzee feeding-tool kit is correlatedstrongly with measures of social tolerance,and poorly with a gregariousness measure(mean party size). Although the robustnessof these results needs to be evaluated usingmore systematic comparisons between sites,it remains strong in a preliminary analysis ofa larger set of sites (van Schaik, in prep.).Finally, the well-known fact that immatureanimals engage in more exploration andinvention than adults may be due in part tothe greater tolerance afforded to juvenileprimates. Nonetheless, while these variouspatterns are suggestive, they cannot as yetconstitute a critical test. In particular, exper-imental tests of learning latency of tool skillsin groups of varying tolerance are needed.

The model presented here explains thedistribution of feeding-tool use among pri-mates, and should therefore bear on thematerial culture of early hominids. How-ever, the absence of extensive tool use inwild bonobos and gorillas appears to beinconsistent with the model’s main ele-ments. This can mean three things. First,the model is wrong. Second, these twospecies may rely less on extractive foragingof the sort that invites tool use. Indeed,bonobos and gorillas rely more on herbs andmonocotyledonous understory plants thando chimpanzees (Malenky & Wrangham,1994). Accordingly, the object manipula-

tions of captive bonobos are less substrate-oriented than those of chimpanzees(Takeshita & Walraven, 1996). Note, too,that gorilla tool use, even in captivity, is lessextensive than that of the other great apes(compiled in Tomasello & Call, 1997).Third, it is conceivable that more close-range observations of well-habituated indi-viduals will yield new forms of extractivetool use in bonobos, and perhaps even ingorillas. In both orang-utans and chimpan-zees between-site variation in tool use isextensive, and the number of sites withlong-term observations of well-habituatedbonobos and even gorillas is smaller thanthat for orang-utans and especially chimpan-zees. At this stage, we cannot distinguishbetween these three possibilities.

Hominid tool useThe first manufactured stone tools appear inthe archaeological record (Semaw et al.,1997) about 2·5 m.y.a. We surmise thathominid tool use was probably not moreadvanced than that of extant chimpanzeesbefore then. Among the main conditions ofthe model (extractive foraging tasks, dex-terity, intelligence, gregariousness and socialtolerance), the most likely elements differen-tiating the stone-tool-making hominids fromgreat apes would be increased intelligence ora higher degree of tolerance accompanied byincreased opportunities for strong relianceon tools.

Increased intelligence is likely to beexpressed in increased relative brain size. Itis still disputed whether the first Oldowantool makers were Australopithecus (Paran-thropus) or Homo (Wood, 1997), but sincecranial material is lacking for both for thisperiod, it is impossible to determine if therelative brain size of the earliest tool makerswas already above the great ape range(Kappelman, 1996). However, experimentssuggest that great apes cannot attain thelevel of sophistication reached by Oldowantool makers (Wynn & McGrew, 1989; Toth

738 . . ET AL.

et al., 1993). Hence, it is likely that thesenew skills do indeed reflect increased cog-nitive abilities. On the other hand, suchabilities would not be maintained in thepopulation unless accompanied by favorablesocial conditions for invention and transmis-sion. The reduction in absolute canine sizeand canine size dimorphism in the genusHomo is consistent with the suggestion ofenhanced social tolerance (Plavcan & vanSchaik, 1997). Moreover, the Oldowanlifestyle involving communal processing oflarge food items, e.g., carcasses (Klein,1989), is inevitably accompanied by exten-sive food sharing, or at least toleratedscrounging. Social tolerance is an essentialcondition for this lifestyle.

Acknowledgements

Carel van Schaik’s orang-utan research atSuaq Balimbing is generously supported bythe Wildlife Conservation Society (foundedas New York Zoological Society). We aregrateful to the Indonesian Institute ofSciences (LIPI) for granting permission todo research in Indonesia, UniversitasIndonesia for sponsoring the project, thedirectorate general of Nature Conservationand Forest Protection (PHPA), for permis-sion to do the work in Gunung LeuserNational Park, and the Leuser DevelopmentProgramme for welcome logistical support.Cheryl Knott and Suci Utami providedimportant personal communications. Wealso thank Beth Fox, Dorothy Fragaszy,Mike Griffiths, Adriaan Kortlandt, BillMcGrew, Charlie Nunn, Catherine Smith,and Maria van Noordwijk for insightfulcomments.

References

Aureli, F., Das, M. & Veenema, H. C. (1997). Differ-ential kinship effect on reconciliation in three speciesof macaques (Macaca fascicularis, M. fuscata, and M.sylvanus). J. Comp. Psych. 111, 91–99.

Beck, B. B. (1980). Animal Tool Behavior: The Use andManufacture of Tools by Animals. New York: Garland.

Bermejo, M., Ilera, G. & Sabater Pí, J. (1994). Animalsand mushrooms consumed by bonobos (Pan panis-cus): new records from Lilunga (Ikela), Zaire. Int. J.Primatol. 15, 879–898.

Boesch, C. (1993). Aspects of transmission of tool-usein wild chimpanzees. In (K. Gibson & T. Ingold,Eds) Tools, Language and Cognition in HumanEvolution, pp. 171–183. Cambridge: CambridgeUniversity Press.

Boesch, C. (1996). The emergence of cultures amongwild chimpanzees. Proc. Brit. Acad. 88, 251–268.

Boesch, C. & Tomasello, M. (1998). Chimpanzee andhuman cultures. Curr. Anthropol. 39, 591–614.

Boesch, C., Marchesi, P., Marchesi, N., Fruth, B. &Joulian, F. (1994). Is nut cracking in wildchimpanzees a cultural behaviour? J. hum. Evol. 26,325–338.

Boinski, S. (1988). Use of a club by a wild white-facedcapuchin (Cebus capucinus) to attack a venomoussnake (Bothrops asper). Am. J. Primatol. 14, 177–179.

Box, H. O. & Fragaszy, D. M. (1986). The develop-ment of social behaviour and cognitive abilities. In(J. G. Else & P. C. Lee, Eds) Primate Ontogeny,Cognition and Social Behaviour, pp. 119–128.Cambridge: Cambridge University Press.

Boyd, R. & Richerson, P. J. (1996). Why culture iscommon, but cultural evolution is rare. Proc. Brit.Acad. 88, 77–93.

Brereton, A. R. (1995). Coercion–defence hypothesis:the evolution of primate sociality. Folia primatol. 64,207–214.

Byrne, R. W. (1995). The Thinking Ape: EvolutionaryOrigins of Intelligence. Oxford: Oxford UniversityPress.

Byrne, R. W. (1997). The technical intelligencehypothesis: an additional evolutionary stimulus tointelligence? In (A. Whiten & R. W. Byrne, Eds)Machiavellian Intelligence II: Extension and Evalu-ations, pp. 289–311. Cambridge: CambridgeUniversity Press.

Byrne, R. W. & Byrne, J. M. E. (1993). Complex leafgathering skills of mountain gorillas (Gorilla g.beringei): variability and standardization. Am. J.Primatol. 31, 241–261.

Call, J. & Tomasello, M. (1994). The social learning oftool use by orang-utans (Pongo pygmaeus). Hum.Evol. 9, 297–313.

Candland, D. (1987). Tool use. In (G. Mitchell & J.Erwin, Eds) Comparative Primate Biology; Vol. 2, PartB: Behavior, Cognition, and Motivation, pp. 85–103.New York: Alan R. Liss.

Casimir, M. J. (1975). Feeding ecology and nutrition ofan eastern gorilla group in the Mt. Kahuzi region(République du Zaire). Folia primatol. 24, 81–136.

Chevalier-Skolnikoff, S. (1990). Tool use by wild cebusmonkeys at Santa Rosa National Park, Costa Rica.Primates 31, 375–383.

Chiang, M. (1967). Use of tools by wild macaques inSingapore. Nature 214, 1258–1259.

739

Clark, T. W. (1979). Food adaptations of a trans-planted Japanese macaque troop (Arashiyama West).Primates 20, 399–410.

Coussi-Korbel, S. & Fragaszy, D. M. (1995). On therelation between social dynamics and social learning.Anim. Behav. 50, 1441–1453.

Fagard, J. & Pezé, A. (1997). Age changes in interlimbcoupling and the development of bimanual coordi-nation. J. Mot. Behav. 29, 199–208.

Fernandes, M. E. B. (1991). Tool use and predationof oysters (Crassostrea rhizophorae) by the tuftedcapuchin, Cebus apella apella, in brackish water man-grove swamp. Primates 32, 529–531.

Fox, E. A., Sitompul, A. F. & van Schaik, C. P. (1999).Intelligent tool use in wild Sumatran orang-utans. In(S. T. Parker, L. Miles & R. Mitchell, Eds) TheMentality of Gorillas and Orangutans, pp. 99–116.Cambridge: Cambridge University Press.

Fragaszy, D. M. & Adams-Curtis, L. E. (1991). Gen-erative aspects of manipulation in tufted capuchinmonkeys (Cebus apella). J. Comp. Psych. 105, 387–397.

Fragaszy, D. M. & Visalberghi, E. (1989). Social influ-ences on the acquisition of tool-using behaviors incapuchin monkeys (Cebus apella). J. Comp. Psych.103, 159–170.

Fragaszy, D. M. & Visalberghi, E. (1990). Socialprocesses affecting the appearance of innovativebehaviors in capuchin monkeys. Folia primatol. 54,155–165.

Galdikas, B. M. F. & Teleki, G. (1981). Variations insubsistence activities of female and male pongids:new perspectives on the origins of hominid labordivision. Curr. Anthropol. 22, 241–256.

Galef, B. G., Jr (1990). Tradition in animals: fieldobservations and laboratory analysis. In (M. Bekoff &D. Jamieson, Eds) Interpretations and Explanationsin the Study of Behavior: Comparative Perspectives,pp. 74–95. Boulder: Westview Press.

Gibson, K. R. (1986). Cognition, brain size and theextraction of embedded food resources. In (J. G. Else& P. C. Lee, Eds) Primate Ontogeny, Cognition,and Social Behaviour, pp. 93–105. Cambridge:Cambridge University Press.

Goldfield, E. C. & Michel, G. F. (1986). Spatio-temporal linkage in infant interlimb coordination.Dev. Psychobiol. 19, 259–264.

Goodall, A. G. (1977). Feeding and ranging behaviourof a mountain gorilla group (Gorilla gorilla beringei) inthe Tshibinda-Kahuzi region (Zaire). In (T. H.Clutton-Brock, Ed.) Primate Ecology: Studies of Feed-ing and Ranging Behaviour in Lemurs, Monkeys, andApes, pp. 450–479. New York: Academic Press.

Goodall, J. (1986). The Chimpanzees of Gombe.Cambridge, MA: Harvard University Press.

Hall, K. R. L. (1963). Tool-using performances asindicators of behavioral adaptability. Curr. Anthropol.4, 479–494.

Hall, K. R., Boelkins, R. C. & Goswell, M. J. (1965).Behaviour of patas monkeys, Erythrocebus patas, incaptivity, with notes on the natural habitat. Foliaprimatol. 3, 22–49.

Hamilton, W. J. & Tilson, R. L. (1985). Fishingbaboons at desert waterholes. Am. J. Primatol. 8,255–257.

Harvey, P. H. & Pagel, M. D. (1991). The ComparativeMethod in Evolutionary Biology. Oxford: OxfordUniversity Press.

Hohmann, G. (1988). A case of simple tool use in wildliontailed macaques (Macaca silenus). Primates 29,565–567.

Horn, A. D. (1987). The socioecology of the blackmangabey (Cercocebus aterrimus) near Lake Tumba,Zaire. Am. J. Primatol. 12, 165–180.

Inoue-Nakamura, N. & Matsuzawa, T. (1997). Devel-opment of stone tool use by wild chimpanzees (Pantroglodytes). J. Comp. Psych. 111, 159–173.

Itani, J. & Nishimura, A. (1973). The study of infra-human culture in Japan. In (J. E. Menzel, Ed.)Precultural Behavior, pp. 26–50. Basel: Karger.

Iwano, T. & Iwakawa, C. (1988). Feeding behaviourof the aye-aye (Daubentonia madagascarensis). Foliaprimatol 50, 136–142.

Izawa, K. & Mizuno, A. (1977). Palm-fruit crackingbehavior of wild black-capped capuchin (Cebusapella). Primates 18, 773–792.

Kappelman, J. (1996). The evolution of body mass andrelative brain size in fossil hominids. J. hum. Evol. 30,243–276.

Kawai, M. (1965). Newly-acquired pre-cultural behav-ior of the natural troop of Japanese monkeys onKoshima Islet. Primates 6, 1–30.

King, B. J. (1986). Extractive foraging and the evol-ution of primate intelligence. Hum. Evol. 1, 361–372.

Klein, R. G. (1989). The Human Career: Human Bio-logical and Cultural Origins. Chicago: University ofChicago Press.

Kortlandt, A. & Kooij, M. (1963). Protohominidbehaviour in primates. Symp. zool. Soc. Lond. 10,61–88.

Kummer, H. & Goodall, J. (1985). Conditions ofinnovative behaviour in primates. Phil. Trans. R. Soc.Lond., Series B 308, 203–214.

Limongelli, L., Boysen, S. T. & Visalberghi, E. (1995).Comprehension of cause–effect relations in a tool-using task by chimpanzees (Pan troglodytes). J. Comp.Psych. 109, 18–26.

Maddison, W. P. (1990). A method for testing thecorrelated evolution of two binary characters: aregains or losses concentrated on certain branches of aphylogenetic tree? Evol. 44, 539–557.

Malenky, R. K. & Wrangham, R. (1994). A quantita-tive comparison of terrestrial herbaceous food con-sumption by Pan paniscus in the Lomako forest,Zaire, and Pan troglodytes in the Kibale forest,Uganda. Am. J. Primatol. 32, 1–12.

Matsuzawa, T. (1994). Field experiments on use ofstone tools in the wild. In (R. W. Wrangham, W. C.McGrew, F. B. M. de Waal, P. G. Heltne & L. A.Marquardt, Eds) Chimpanzee Cultures, pp. 351–370.Cambridge, MA: Harvard University Press.

McGrew, W. C. (1974). Tool use by wild chimpanzeesin feeding upon driver ants. J. hum. Evol. 3, 501–508.

740 . . ET AL.

McGrew, W. C. (1992). Chimpanzee Material Culture:Implications for Human Evolution. Cambridge:Cambridge University Press.

McGrew, W. C., Baldwin, J. P. & Tutin, C. E. G.(1979). Chimpanzees, tools, and termites: cross-cultural comparisons of Senegal, Tanzania, and RioMuni. Man 14, 185–214.

McGrew, W. C., Ham, R. M., White, L. T. J., Tutin,C. E. G. & Fernandez, M. (1997). Why don’tchimpanzees in Gabon crack nuts? Int. J. Primatol.18, 353–374.

McGrew, W. C. & Marchant, L. F. (1997). Using thetools at hand: manual laterality and elementary tech-nology in Cebus spp. and Pan spp. Int. J. Primatol. 18,787–810.

Menzel, E. W., Jr (1966). Responsiveness to objects infree-ranging Japanese monkeys. Behav. 26, 130–150.

Nunn, C. L. & van Schaik, C. P. (in press). Recon-structing the behavioral ecology of extinct primates:patterns among extant species. In (M. Plavcan, W.Jungers, R. Kay & C. van Schaik, Eds) The Behaviorof Extinct Primates. New York: Plenum Press.

Oyen, O. J. (1979). Tool-use in free-ranging baboonsof Nairobi National Park. Primates 20, 595–597.

Palombit, R. A. (1997). Inter- and intraspecific vari-ation in the diets of sympatric siamang (Hylobatessyndactylus) and lar gibbons (Hylobates lar). Foliaprimatol. 68, 321–337.

Parker, C. E. (1974). Behavioral density in ten speciesof nonhuman primates. J. Comp. Phys. Psych. 87,930–937.

Parker, S. T. & Gibson, K. R. (1977). Object manipu-lation, tool use and sensorimotor intelligence as feed-ing adaptations in Cebus monkeys and great apes.J. hum. Evol. 6, 623–641.

Petit, O. & Thierry, B. (1993). Use of stone tools in acaptive group of guinea baboons (Papio papio). Foliaprimatol. 61, 160–164.

Phillips, K. A. (1998). Tool use in wild capuchinmonkeys (Cebus albifrons trinitatis). Am. J. Primatol.46, 259–261.

Plavcan, J. M. & van Schaik, C. P. (1997). Interpretinghominid behavior on the basis of sexual dimorphism.J. hum. Evol. 32, 345–374.

Preuschoft, S. (1995). ‘‘Laughter’’ and ‘‘smiling’’ inmacaques: an evolutionary perspective. Ph.D. Dis-sertation, University of Utrecht

Purvis, A. (1995). A composite estimate of primatephylogeny. Phil. Trans. R. Soc. Series B 348, 405–421.

Rhine, R. J. & Westlund, B. J. (1978). The nature of aprimary feeding habit in different age–sex classes ofyellow baboons (Papio cynocephalus). Folia primatol.30, 64–79.

Rodman, P. S. (1988). Diversity and consistency inecology and behavior. In (J. H. Schwartz, Ed.)Orang-utan Biology, pp. 31–51. Oxford: OxfordUniversity Press.

Russon, A. E. (1997). Exploiting the expertise ofothers. In (A. Whiten & R. W. Byrne, Eds) Machi-avellian Intelligence II: Extensions and Evaluations, pp.174–206. Cambridge: Cambridge University Press.

Russon, A. E. & Bard, K. A. (1996). Exploring theminds of great apes: issues and controveries. In (A. E.Russon, K. A. Bard & S. Parker, Eds) Reaching intoThought: The Minds of the Great Apes, pp. 1–20.Cambridge: Cambridge University Press.

Russon, A. E., Mitchell, R. W., Lefebvre, L. &Abravanel, E. (in press). The comparative evolutionof imitation. In (J. Langer & E. Killen, Eds) Piaget,Evolution, and Development, pp. 1–20. New York:Erlbaum.

Semaw, S., Renne, P., Harris, J. W. K., Feibel, C. S.,Bernor, R. L., Fesseha, N. & Mowbray, K. (1997).2.5-million-year-old stone tools from Gona,Ethiopia. Nature 385, 333–336.

Sinha, A. (1997). Complex tool manufacture by a wildbonnet macaque, Macaca radiata. Folia primatol. 68,23–25.

Struhsaker, T. T. & Leland, L. (1977). Palm-nutsmashing by Cebus a. apella in Colombia. Biotrop. 9,124–126.

Sugardjito. (1986). Ecological constraints on thebehaviour of Sumatran orang-utans (Pongo pygmaeusabelii) in the Gunung Leuser National Park,Indonesia. Ph.D. Dissertation, University of Utrecht

Sugardjito, J., te Boekhorst, I. J. A. & van Hooff,J. A. R. A. M. (1987). Ecological constraints on thegrouping of wild orang-utans (Pongo pygmaeus) in theGunung Leuser National Park, Sumatra, Indonesia.Int. J. Primatol. 8, 17–41.

Sugiyama, Y. (1993). Local variation of tools and tooluse among wild chimpanzee populations. In (A.Berthelet & J. Chavaillon, Eds) The Use of Tools byHuman and Non-human Primates, pp. 175–190.Oxford: Clarendon Press.

Suzuki, A. (1989). Socio-ecological studies of orang-utans and primates in Kutai National Park, EastKalimantan in 1988–1989. Kyoto Univ. OverseasRes. Rep. of Studies on Asian Non-human Primates 7,1–42.

Takeshita, H. & Walraven, V. (1996). A comparativestudy of the variety and complexity of objectmanipulation in captive chimpanzees (Pan troglo-dytes) and bonobos (Pan paniscus). Primates 37, 423–441.

Thierry, B. (1985). Patterns of agonistic interactionin three species of macaque (Macaca mulatta,M. fascicularis, M. tonkeana). Aggr. Behav. 11, 223–233.

Thierry, B., Anderson, J. R., Demaria, C., Desportes,C. & Petit, O. (1994). Tonkean macaque behaviourfrom the perspective of the evolution of Sulawesimacaques. In (J. J. Roeder, B. Thierry, J. R.Anderson & N. Herrenschmidt, Eds) Current Prima-tology, Vol. II: Social Development, Learning andBehaviour, pp. 103–117. Strasbourg: UniversitéLouis Pasteur.

Tokida, E., Tanaka, I., Takefushi, H. & Hagiwara, T.(1994). Tool-using in Japanese macaques: use ofstone tools to obtain fruit from a pipe. Anim. Behav.47, 1023–1030.

Tomasello, M. & Call, J. (1997). Primate Cognition.New York: Oxford University Press.

741

Tomblin, D. C. & Cranford, J. A. (1994). Ecologicalniche differences between Alouatta palliata and Cebuscapucinus comparing feeding modes, branch use, anddiet. Primates 35, 265–274.

Torigoe, T. (1985). Comparison of object manipu-lation among 74 species of nonhuman primates.Primates 26, 182–194.

Toth, N., Schick, K. D., Savage-Rumbaugh, E. S.,Sevcik, R. A. & Rumbaugh, D. (1993). Pan thetool-maker: investigations into the stone tool-makingand tool-using capabilities of a bonobo (Panpaniscus). J. Archaeol. Sci. 20, 81–91.

Tutin, C. E. G. (1994). Reproductive success story:variability among chimpanzees and comparisonswith gorillas. In (R. W. Wrangham, W. C. McGrew,F. B. M. de Waal, P. G. Heltne and L. A. Marquardt,Eds) Chimpanzee Cultures, pp. 181–193. Cambridge:Harvard University Press.

Uehara, S. (1990). A preliminary report on age differ-ences in plant-feeding behaviors among adult andadolescent males. In (T. Nishida, Ed.) The Chimpan-zees of the Mahale Mountains, pp. 223–236. Tokyo:University of Tokyo Press.

van Noordwijk, M. A. & van Schaik, C. P. (1987).Competition among female long-tailed macaques,Macaca fascicularis. Anim. Behav. 35, 577–589.

van Schaik, C. P. (1989). The ecology of social rela-tionships amongst female primates. In (V. Standen &R. A. Foley, Eds) Comparative Socioecology, pp. 195–218. Oxford: Blackwell.

van Schaik, C. P. (1999). Fission-fusion sociality inSumatran orang-utans. Primates 40, 73–90.

van Schaik, C. P., Azwar & Priatna, D. (1995). Popu-lation estimates and habitat preferences of orang-utans based on line transects of nests. In (R. D.Nadler, B. F. M. Galdikas, L. K. Sheeran & N.Rosen, Eds) The Neglected Ape, pp. 129–147. NewYork: Plenum Press.

van Schaik, C. P., Fox, E. A. & Sitompul, A. F. (1996).Manufacture and use of tools in wild Sumatranorang-utans. Naturwiss. 83, 186–188.

van Schaik, C. P. & van Noordwijk, M. A. (1986). Thehidden costs of sociality: intra-group variation infeeding strategies in Sumatran long-tailed macaques(Macaca fascicularis). Behaviour 99, 296–315.

Visalberghi, E., Fragaszy, D. M. & Savage-Rumbaugh,S. (1995). Performance in a tool-using task by com-mon chimpanzees (Pan troglodytes), bonobos (Panpaniscus), an orang-utan (Pongo pygmaeus), andcapuchin monkeys (Cebus apella). J. Comp. Psych.109, 52–60.

Waal, F. B. M. de (1997). Food transfers through meshin brown capuchins. J. Comp. Pyschol. 111, 370–378.

Westergaard, G. C. (1988). Lion-tailed macaques(Macaca silenus) manufacture and use tools. J. Comp.Psych. 102, 152–159.

Westergaard, G. C. & Fragaszy, D. M. (1987). Themanufacture and use of tools by capuchin monkeys(Cebus apella). J. Comp. Psych. 101, 159–168.

Wheatley, B. P. (1980). Feeding and ranging of EastBornean Macaca fascicularis. In (D. G. Lindburg,Ed.) The Macaques, pp. 215–246. New York: VanNostrand Reinhold Co.

Whiten, A. (1998). Imitation of the sequential structureof actions by chimpanzees. J. Comp. Psych. 112,270–281.

Wood, B. (1997). The oldest whodunnit in the world.Nature 385, 292–293.

Wynn, T. & McGrew, W. C. (1989). An ape’s view ofthe Oldowan. Man 24, 383–398.

Zuberbühler, K., Gygax, L., Harley, N. & Kummer, H.(1996). Stimulus enhancement and spread of a spon-taneous tool use in a colony of long-tailed macaques.Primates 37, 1–12.