Learning & Behavior 2004, 32 (1), 36-52

How do apes ape?

ANDREW WHITEN, VICTORIA HORNER, CARLA A. LITCHFIELD, and SARAH MARSHALL-PESCINI University of St. Andrews, St. Andrews, Scotland

In the wake of telling critiques of the foundations on which earlier conclusions were based, the last 15years have witnessed a renaissance in the study of social learning in apes. As a result, we are able to review 31 experimental studies from this period in which social learning in chimpanzees, gorillas, and orangutans has been investigated. The principal question framed at the beginning of this era, Do apes ape? has been answered in the affirmative, at least in certain conditions. The more interesting question now is, thus, How do apes ape? Answering this question has engendered richer taxonomies of the range of social-learning processes at work and new methodologies to uncover them. Together, these studies suggest that apes ape by employing a portfolio of alternative social-learning processes in flexibly adaptive ways, in conjunction with nonsocial learning. We conclude by sketching the kind of decision tree that appears to underlie the deployment of these alternatives.

The expression to ape suggests that the three genera Africa, researchers have identified no fewer than 39 dif- most closely related to ourselves ought to offer particu- ferent behavior patterns as likely cultural variants, oc- larly rich pickings in the investigation of animal social curring commonly in at least one community, yet absent learning. There are several reasons why the generous in another, with no apparent environmental or genetic ex- space accorded to the great apes, so small an animal planations for this patterning. This procedure has more taxon by comparison with others reviewed in this issue, recently been applied to orangutans, identifying 19 cul- might be justified. tural variants, plus five more tentative candidates (van The first concerns the primary functional significance Schaik etal., 2003). Van Schaik etal. additionally tested of social learning, which for primates has long been hy- two predictions based on the hypothesis that the variants pothesized to lie in the acquisition of adaptive traditions, are transmitted by social learning. These predictions such as antipredator behavior and, particularly, skilled or were supported for both chimpanzees and orangutans: clever foraging methods (Imanishi, 1957; Nishida, 1987; Geographically closer communities share a greater pro- Passingham, 1982). The discovery of a variety of forms portion of variants, and communities in which individu- of chimpanzee tool use, together with the close observa- als spend more time in close proximity have more ex- tion of these by juveniles and the patterns of acquisition tended repertoires of variants. associated with them, was soon interpreted as indicating In these studies, a substantial contrast with studies of the existence of socially learned traditions, or cultures traditions among other species, which typically identify (Goodall, 1968, 1973). As other long-term studies were only one or a small handful of variations, is highlighted. carried out across Africa, collation of the published re- This suggests that in the evolutionary spiraling of cul- sults produced a growing list of putative cultural varia- tural complexity that has apparently occurred, apes have tions (McGrew, 1992). Most recently, recognizing that a been subject to selective pressures that have sculpted the reliance on published data might yield only a partial pic- strength and nature of mechanisms for social learning in ture of the existing variations, research site leaders have special ways: a case of gene–culture coevolution (Laland, pooled their long-term records in a two-phase procedure Odling-Smee, & Feldman, 2000). Observation of apes’ designed to establish a more definitive picture (Whiten natural behavior was essential to gain this functional per- etal., 1999, 2001). By first generating a set of hypothe- spective but, unfortunately, offers little prospect of fur- sized variations and then sieving these through a sys- ther untangling the many different learning processes tematic coding procedure at each of nine sites across that could be operative. For this purpose, experimenta- tion offers distinctive power, and given our space limits here, the results of such studies are the principal focus of For comments on earlier versions of this manuscript, we are grateful to the present review. Which social-learning processes are Christine Caldwell, Tara Stoinski, and particularly for her incisive editor- associated with the phenomena of traditions and cultures ial advice, Celia Heyes. We thank Jeff Galef for drawing our attention to remains a major, yet unresolved issue in animal learning Lloyd Morgan’s early discussion of the issue of emulation. Correspon- (Galef, 1992, 2004; Heyes, 1993). dence concerning this article should be addressed to A.Whiten, Centre for Social Learning and Cognitive Evolution and Scottish Primate Research A second reason why social learning among great apes Group, School of Psychology, University of St. Andrews, St. Andrews, begs attention derives directly from their being our closest Fife KY16 9JU, Scotland (e-mail: [email protected]). living relatives. Identifying aspects of ape social-learning

Copyright 2004 Psychonomic Society, Inc. 36 HOW DO APES APE?37 capacities shared with humans permits inferences about chimpanzees were tested and showed only limited powers these abilities in our shared ancestors of 5–6 million of social learning. Observer chimpanzees who watched years ago. Such comparisons should equally sharpen our an expert conspecific using a stick to rake food from an understanding of just where the critical changes in social out-of-reach platform were more likely to become rakers learning and culture occurred in our evolutionary past than were control subjects who did not see a model, but (Whiten, Horner, & Marshall-Pescini, 2003). they failed to copy the particular two-step technique the A third set of reasons to value ape research involves model had used to retrieve trapped food. linkages with other disciplines. For example, both theo- Such a result obtained in a chimpanzee–chimpanzee test retical and methodological aspects of research on ape so- raised the disturbing possibility that the average chim- cial learning have recently shaped related work in child panzee might not really deserve its reputation for imita- development (Want & Harris, 2002; Whiten, 2002c). tion after all. On the heels of this finding came a number Other links are developing with artificial intelligence of influential critiques of earlier social-learning studies. and robotics (Dautenhahn & Nehaniv, 2002), as well as Largely on methodological grounds, these publications with neurophysiology and neuropsychology (Hurley & questioned widely held suppositions about the propensity Chater, in press; Meltzoff & Prinz, 2002). Animal social for imitation in primates and other animals (Visalberghi learning lies at the intersection of a number of such ex- & Fragaszy, 1990; Whiten & Ham, 1992), including apes citing developments, with ape research playing a signif- (Galef, 1988, 1990; Heyes, 1993; Tomasello, 1990). icant role, for the reasons outlined above. This critical onslaught offered a challenge to received wisdoms and stimulated a raft of new studies, displaying THE RECENT HISTORY OF RESEARCH ON greater sophistication in both theory and method. Among APE SOCIAL LEARNING these, observational and field studies have played a cru- cial role, as was noted at the outset of this article. How- The last 15 years have seen an upsurge in systematic ever, it is difficult through pure observation to dissect the studies of social learning in apes, the conclusions of causal processes underlying social learning. Accordingly, which we will review below. We should begin, however, the focus of the present review is the experimental stud- by setting this recent work in its wider historical context. ies published since Tomasello etal.’s influential 1987 re- Anecdotal reports of imitation by apes stretch back for port. We will review 31 of these, a number that represents centuries, and more systematic studies of social learning a renaissance of research in this area, contrasting with were pursued all through the last century (Mitchell, the mere 4 or so such studies that Tomasello and Call 1999; Tomasello & Call, 1997; Whiten & Ham, 1992). (1997) listed for the 15 years before the “watershed” year The 20th century’s efforts included formal experiments, of 1987. There is not the space here to describe all of the ethological studies in the wild, and observations of cap- studies in detail. Instead, we shall attempt to draw out tive apes, some of the latter being reared in human fam- emerging general conclusions, illustrating them with ap- ilies, facilitating detailed comparison with the behavior propriate examples. of human children in similar contexts. By the 1970s, a common conclusion was that apes ape. Influentially, K.J. DO APES APE? and C. Hayes’s (C.Hayes, 1951; K.J. Hayes & C.Hayes, 1952) studies of the home-reared chimpanzee Vicki in- The developments summarized above meant that the cluded graphic accounts of the way she copied such urgent question in the 1990s was whether apes do or do everyday human activities as brushing one’s teeth, plus not, after all, ape—a question that most researchers, like the results of an experimental battery of tests in which Tomasello (1996), took to be synonymous with asking she was reported to respond convincingly, after training, Do they imitate? (imitation will, in turn, be defined and to the request to “do this,” even for relatively novel actions. dissected in the next section). A common conclusion Another influential strand was the discovery of what about social learning among primates was that monkeys, looked like apprenticeship to skilled tool use in the wild despite a century’s efforts, had not been shown to ape in (Goodall, 1973). Such diverse sources appeared to con- this sense, whereas the evidence for apes themselves was verge on a conclusion fairly routinely expressed in text- more ambiguous (Shettleworth, 1998; Whiten & Ham, books, that apes possess a functionally important capac- 1992). Apes had generated a curious mixture of negative ity to imitate; indeed, this was sometimes interpreted as and positive findings, some of the latter alluringly vivid, just a fuller expression of a capacity probably quite wide- even if anecdotal. Tomasello (1996) portrayed imitation spread in primates (e.g., MacFarland, 1984; Passingham, as an inherently sophisticated cognitive achievement, ap- 1982). pearing clearly only in humans, the prime question being Beginning in the late 1980s, this picture was severely whether our closest animal relatives might just achieve challenged by a string of publications. One of the first de- the same mental heights. scribed an experiment that used conspecific models, rather Largely through the use of more robust experiments, than human demonstrators (Tomasello, Davis-DaSilva, the pendulum has since swung back from such extremes Camak, & Bard, 1987). Instead of a chimpanzee such as of doubt over whether nonhuman animals imitate. In the Vicki, influenced by rich interactions with human foster Appendix, we list the accumulated evidence for apes, in- parents who were also her models, group-living captive cluding a column that summarizes the authors’ key con- 38WHITEN, HORNER, LITCHFIELD, AND MARSHALL-PESCINI clusions. As is evident, imitation figures prominently; we interpret some of the varieties of what has been called indeed, for chimpanzees and orangutans, it is the most emulation as overlapping with imitation in important common conclusion. The existence of imitation in apes ways, rather than as offering a neat dichotomy. Accord- is now accepted in relation to at least some of these stud- ingly, we will consider how apes ape in both a broad and ies by the more skeptical of the earlier critics (Galef, in a narrower sense. The narrower sense may be equated press; Heyes, 2001). Accordingly, we can move forward with imitation, although imitation may itself be defined to the more difficult, but interesting, question, How do in the literature either in relatively restricted ways (e.g., apes ape? limited to only bodily imitation or to novel actions or re- Taking this step now appears not so radical, because quiring the recognition of the model’s intent) or, instead, evidence for imitation has also recently accumulated for more liberally, not requiring these constraints. Assessing other, very different species. Of two studies in the refer- how apes ape in a yet broader sense requires us to ana- eed literature allowed to include true imitation in their lyze how imitation may operate as part of a portfolio of titles, one concerned marmosets (Voelkl & Huber, 2000) other social-learning processes and, indeed, may overlap and the other pigeons (Zentall, Sutton, & Sherburne, with some of these. In doing this, we intentionally echo 1996). Several other recent studies have demonstrated the approach taken by Want and Harris (2002), who con- imitation in small birds (see Zentall, 2004), whose brains sidered a panoply of social-learning processes, includ- are tiny, as compared with those of the great apes. Of ing stimulus enhancement, emulation, and imitation, in course, there may be important differences between what answering the question, How do children ape? we can broadly call imitation in apes and that in species The classification of social learning we shall apply so different from them, and we shall consider these fur- here is depicted in Figure1. It represents an updated ver- ther below. sion of part of the scheme presented by Whiten and Ham (1992), considerably revised in the light of theoretical How Apes Ape: Imitation and Alternative distinctions made in the intervening period. In asking Social-Learning Processes how apes ape , we appraise the growing corpus of empir- Essential preliminaries to our analysis of how apes ical evidence that suggests how apes deploy the options ape include clarifying what we intend the verb ape to suggested by Figure1. We are not proposing to answer connote and what, in the present state of knowledge, we our question at the level of what psychological or neuro- can hope to include in the how. biological processes constitute the underlying mecha- The terminology and underlying conceptual framework nisms of imitation. The discovery of mirror neurons in of social-learning research has become infamous for its monkeys has fueled recent theorizing about such mech- complexities and confusions. Disputes about whether anisms (Byrne, 2000; Heyes, 2002; Rizzolatti, Fadiga, apes (and other species) imitate have often focused on Fogassi, & Gallese, 2002; Whiten, 2000), but since no what is deemed an acceptable definition of imitation, as such work has been done on great apes, we think the level much as on what reliable data a study has gathered. How- at which we here tackle our overarching question is the ever, we believe that there will be a net benefit if, from appropriate and productive one at the present time. within these debates, we can now distill a more refined Given this approach, it is important to recognize that dissection of alternative processes of social learning. although our intention is that the categories of learning Perhaps the most influential new distinction arose from discriminated in Figure1 be conceptually and experi- the 1987 article by Tomasello etal., already highlighted. mentally separable, it is important to recognize that more Finding that chimpanzees learned something about tool than one might be in operation in any single episode of use that nonobserving control subjects did not, the authors social learning; they are not all mutually exclusive in this concluded that social learning was involved. However this respect. Through the main sections to follow, we will was said not to involve imitation, because the subjects did work from the top of Figure1 down, explaining the dis- not copy the particular two-stage technique the model tinctions between our categories with a focus on differ- had used to gain food trapped at the edge of the platform. ent senses of emulation that we think beg particular at- Even so, the authors rejected the hypothesis that the learn- tention. We will begin by comparing one of these with ing involved stimulus enhancement —attention being imitation. Key findings for apes are noted in relation to drawn to relevant objects—because the control subjects each distinction in turn. manipulated the stick as much as the observers and were motivated to attempt to obtain the food. Instead, Tomasello How Do Apes Ape? etal. (1987) concluded that the experimental subjects 1a. Imitation and object movement reenactment: had learned by observation something about the stick— Distinctions and overlaps . Following Whiten and Ham specifically, its function as a tool . Tomasello (1990) later (1992), we define imitation as a process whereby one in- labeled this emulation. dividual copies some part of the form of an action from Since 1990, the meaning of emulation has diversified, another. By contrast, according to Tomasello (1998), in in sometimes confusing ways. Nevertheless, we suggest emulation learning, animals are “learning about the en- that within this varied usage, some important distinc- vironment, not about behaviour” (p. 704). Where exactly tions are worth making more explicit and operationaliz- is the critical boundary between the two? The essential ing, a task we will pursue in the next section. Moreover, distinction implied is that an imitator is learning about HOW DO APES APE?39

Figure1. A taxonomy of social-learning processes. This scheme represents a revision of that of- fered in Whiten and Ham (1992) that takes into account theoretical and empirical progress in the intervening period. Note that the categories presented here omit those Whiten and Ham classed as only social influence , such as social facilitation. In the present scheme, different kinds of social- learning processes are discriminated, labeled, and partially defined (for further specification, see the text). Categories are arranged so that there is generally most potential for high-fidelity copying at the top (imitation), descending to enhancement at the bottom. Affordance learning may, in prin- ciple, be used to generate actions that vary in their degree of match to actions of the individual(s) observed. To the right of the “Imitation” and “Object Movement Reenactment” (OMR) categories are listed a number of different aspects in relation to which matching to a model may occur—for example, hierarchical structure might be copied in relation to either imitation of bodily movements or OMR. The same is true for the spatiotemporal shape of actions, their sequential patterning, and causal and intentional aspects of the model’s actions. Note that Whiten and Ham labeled all emu- lation as goal emulation , whereas here, goal emulation is only a subcategory of emulation, for which a tighter specification is explained in the text. events bounded within the skin of the model, whereas an Here, the observer is essentially copying what the object emulator is learning about events outside the models’ does (or to put it another way, what the model does with skin (where the environment begins). Such a distinction the object), in a way that appears analogous to copying may reflect significant differences in the underlying what body parts do . Accordingly, in Figure1, imitation information-processing mechanisms, although this re- and OMR are classified together as copying processes. mains to be established. One way to investigate the dis- To illustrate the point with an example, if A copies the tinction is the ghost control (Heyes, Jaldow, Nokes, & trajectory of a hammer swung by B, this may draw on Dawson, 1994), in which, through the experimenter’s in- imitative processes not so different from those involved genuity, the observer animal sees the environmental when A copies the way B hammers with a fist (Whiten changes normally caused by the model, but without ac- & Custance, 1996). Given that a tool, such as a hammer, tually seeing what the model does to make them happen. can come to be effectively an extension of the holder’s However, emulation defined in this way could well in- limb, sharing characteristics of the way in which the volve information processing that shares fundamental hand functions as a natural tool of the arm (Whiten, similarities with the imitation of bodily movements. 2000), the body–environment boundary may turn out not Consider cases in which an observer reproduces the to be such a critical one at which to expect a fundamen- movements of objects caused by a model, a category that tal difference in social-learning mechanism. For exam- Custance, Whiten, and Fredman (1999) have described as ple, the results of the study of Tomasello etal. (1987) “object movement re-enactment” (OMR; see Figure1). may be explicable as a case of low-fidelity OMR in 40WHITEN, HORNER, LITCHFIELD, AND MARSHALL-PESCINI which observers copied the actions of the stick at a rela- model did, and then gradually “home in” on the model’s tively crude level (describable as raking), rather than at action. (4) Touching body parts out of sight (an example the detailed level of the two-step method the model had of Heyes’s, 2002, perceptually opaque actions) may be used in certain contexts. achieved as accurately as those in sight. (5)As compared One respect in which bodily imitation might differ with children, who may show recognizable matching on fundamentally from OMR concerns cases in which the all of the actions in the battery used, fidelity is typically copier cannot see its own actions, such as picking ones low overall. (6) Successful imitation can be deferred teeth with one’s fingers. Copying such perceptually (48h in the study of Tomasello, Savage-Rumbaugh, & opaque actions can be expected to impose distinctive Kruger, 1993). Applying the same approach to OMR, information-processing requirements (Heyes, 2002). copying the movements of one object in relation to an- However, object movements may also be opaque, as in other was found to be easier than copying the movements using a brush to clean one’s teeth or applying lipstick. of a single object, including applying it to oneself One could even extend the above logic to include OMR (Myowa-Yamakoshi & Matsuzawa, 1999). within a category labeled imitation (which would then The alternative question—Do apes imitate? —is more include bodily imitation and object movement imitation ; often appropriate in experiments in which the observer is Stoinski, Wrate, Ure, & Whiten, 2001; Whiten, 2000). allowed to watch a model and investigation focuses on However, it remains an open empirical question as to what the observer subsequently does freely. The question how similar to bodily imitation copying of the various becomes What does the observer learn through observa- kinds of object movements that an animal can cause is. tion? with imitation being merely one of the possibilities. As was indicated above, some may be opaque, and oth- This approach commonly uses functional tasks, such as ers not. Some may be intimately interwoven with bodily gaining food through tool use or some other kind of ma- actions, such as hammering, but others may be much nipulation. These tasks are often designed to be relevant more distal, such as objects being moved by the end of a to the question of whether apes are likely to acquire their tool. Accordingly, in Figure1, we refrain from simply traditions in the wild through imitation or other learning assimilating OMR to imitation but indicate the possibil- processes, an example of which is our use of artificial ity that cases of OMR may vary in the extent to which the fruits (Whiten, 2002a). Such studies are grouped, for information processing involved shares characteristics each genus, in the lower parts of the Appendix. with imitation. This is a more messy approach but ex- There has been much disagreement about how apes presses potential differences and similarities that beg ape—what information they are processing—when faced empirical investigation, in contrast to a too easily ac- with observation of such tasks. As we have seen, Tomasello cepted imitation/emulation dichotomy. etal. (1987) interpreted their results as showing that 1b. Imitation and OMR in apes . Beginning with im- chimpanzees emulate, rather than imitate. Nagell, Olguin, itation, we note two importantly different questions: Can and Tomasello (1993) and Call and Tomasello (1995) apes imitate? versus Do apes imitate? Different method- concurred in this conclusion. In contrast, using the open- ological approaches have differential power in tackling ing of a complex artificial fruit as the task, instead of these. In relation to the first, the technique described as tool use, Whiten, Custance, Gomez, Teixidor, and Bard do-as-I-do offers the opportunity, once a subject has (1996; Whiten, 1998b) and Stoinski etal. (2001) ob- been trained to the basic idea of attempting to copy what tained results for chimpanzees and gorillas, respectively, a model does on command, of exploring quite fully just that they interpreted as evidence for both imitation and what they can and cannot copy. It has been successfully emulation. In these studies, chimpanzees copied by either implemented with both chimpanzees (Custance, Whiten, poking bolts out of holes or, instead, pulling and twist- & Bard, 1995; K.J. Hayes & C.Hayes, 1952; Myowa- ing them out, according to whichever of two alternative Yamakoshi & Matsuzawa, 1999; Tomasello, Savage- models they had seen. Similarly, gorillas matched their Rumbaugh, &Kruger, 1993) and orangutans (Call, 2001; models in either pulling a pin straight out of its hole or Miles, Mitchell, & Harper, 1996) with quite consistent removing it by twisting. findings that cover not only bodily imitation, but also As Tomasello (1996) noted, these results might still OMR (in the Appendix, for each genus, do-as-I-do stud- reflect emulation (in the sense of OMR) if subjects were ies are the first listed). not replicating bodily movements but, instead, had A number of significant conclusions have flowed from learned only about the objects themselves. These alter- this work. (1) The do-as-I-do method indicates that some natives may often be difficult to disambiguate in object apes, at least, can imitate, because coders blind to what manipulation tasks. However, Sto inski etal. (2001) subjects saw can identify matches at statistically signif- found that unlike the chimpanzees studied by Whiten icant levels. (2) Imitation applies to a range of actions etal. (1996), gorillas did not copy poking versus twist- that includes facial expressions, unimanual and biman- ing of the bolts when faced with a similar “fruit.” They ual gestures, and bodily postures (e.g., Custance etal., did, however, replicate the way the bolts moved, either 1995), as well as the copying of object manipulations away from the manipulator or (in the alternative experi- (OMR; e.g., Tomasello, Savage-Rumbaugh, &Kruger, mental condition) toward them. This is consistent with 1993). (3) Sometimes, subjects respond initially with ac- the gorillas’ focusin g on what the objects themselves tions already known, which only approximate what the “do” and evidencing OMR, rather than bodily imitation. HOW DO APES APE?41

However, we interpret the twisting actions observed as egory of copying (Figure1). In this we follow Miklosi bodily imitation, because the hands of the model ob- (1999), who suggested that the term copying might be an scured the object manipulated (Custance, 1998). More- appropriate one for both imitation and some of the phe- over, in more recent unpublished studies with wild-born nomena referred to as emulation. We can also appeal chimpanzees in an African sanctuary, we have recorded here to the great pioneer of social-learning research, multiple poking of the bolts back and forth by those sub- Lloyd-Morgan (1896). Anticipating discussion of emu- jects who saw the model poke, consistent with imitation, lation by a century, Lloyd-Morgan drew out a distinction but not with OMR. between reproducing another’s actions versus their re- The artificial fruit studies have thus been interpreted sults (i.e., emulation), noting, however, that “both are as evidencing imitation in two different senses. One, commonly called imitation,” yet also “both are com- adopting the more restricted definition of imitation as monly called copying” (pp.172– 173). Agreeing with bodily rather than object centered, would lead to the con- Lloyd Morgan, we suggest that it will help to minimize clusion that there is some evidence of this (although rel- confusion if we keep scientific usage sensibly in line atively little in comparison with children, which will be with the terms we borrow from everyday usage, even as discussed further below). Alternatively, by assimilating we formalize and define them for clarity’s sake. OMR to a broader sense of imitation as outlined earlier Under end-state emulation, we include emulation of (Stoinski etal., 2001), more of the copying recorded would results and goals. This reflects Wood’s (1989) reference, be seen as indicating imitative learning. quoted above, to emulation of “ends,” “ objectives,” and The overall picture of imitation from our own studies “goals.” Similarly , Tomasello (1990) referred to an em- is thus one in which both bodily and object movement ulator’s “focus on the demonstrator’s goal” (p. 283) and, copying occurs. However, the former typically plays a elsewhere, to “attending to the end result” (1998, p.704). relatively minor role in functional tasks that involve ob- Both Wood and Tomasello tended to use these terms (re- ject manipulation. This may sometimes reflect limita- sult and goal) almost interchangeably to delineate emu- tions in imitative capacity, but on other occasions, it may lation, but there is surely a significant distinction be- reflect adaptive choices, where it can be more efficient tween an emulator who seeks on ly to re-create an to replicate object movement through actions that do not observable result of the model’s actions and an emulator match those of the model. That apes’ primary focus of who regards the model as an organism with goals and attention is on what the objects or environmental fea- seeks to reproduce what it infers to be the model’s goal tures manipulated are doing is a conclusion about which on a particular occasion. Call and Carpenter (2002) like- we begin to see some real consensus between different wise emphasized this distinction in proposing a three- research groups. As has been noted, this was apparent in way taxonomy of the information acquired in social our own work with artificial fruits, but it has also been learning, referring to actions, results, and goals. emphasized with regard to a finding in Myowa-Yamakoshi Of course, a result of a model’s action can also be its and Matsuzawa (1999) that chimpanzees found it easiest goal; the question is whether the emulator is interpreting to imitate object–object actions and, of course, in the the end-state as a goal or, otherwise, just as an observ- work of Tomasello and colleagues, reviewed above (and able outcome of interest. Also, where an observer infers see the Appendix), which led them to highlight emula- a model’s goal and later attempts to achieve it, even tion as another high-level mode of social learning, con- though the model had failed, we must be dealing with the trasting with imitation. copying of goals, rather than of results. Interestingly, 2a. Emulation of results and goals: The essential Meltzoff (1995) described such a result for human in- distinctions . A formulation of emulation different from fants. He described this as “re-enactment of intentional OMR is that “the observer is attending to the end result” actions,” but this apparent allusion to an imitative pro- (Tomasello, 1998, p.704). This was clear in Wood’s (1989) cess has been challenged through a study by Huang, original introduction of the term, stating that “children Heyes, and Charman (2002) that implicated emulation. attempt to impersonate others by imitating their actions 2b. Result emulation and goal emulation in apes . but also try to emulate them by achieving similar ends or Earlier, we suggested that the results in Tomasello etal. objectives” (p. 71); more specifically, emulation in- (1987) could be a case of OMR. They could alternatively cludes “instances where children achieve common goals represent result emulation, if all that the observers were to those modeled, but do so by using idiosyncratic means learning was that an end-state was possible—“getting that were never observed” (p. 72). This contrasts starkly the food raked in.” In principle, this could be discrimi- with imitation, for in emulation as defined by Wood, pre- nated by allowing observers to see only the end-result— cisely what the observer is not doing is copying the mod- perhaps an image of only the final configuration, with el’s approach to the problem. This is why we do not in- the food and the rake drawn back in reach. Since OMR, clude our corresponding category of end-state emulation by contrast, could involve copying a whole sequence of under either imitation or OMR. stick movements, such movements would all need to be Nevertheless, the implication is that the observer is seen. Huang etal. (2002) came close to the required ex- copying something—achieving the end result they see periment with young children, in preventing them from the model achieve. Accordingly, we class end-state em- seeing the central actions normally observable in a mod- ulation, imitation, and OMR under the superordinate cat- el’s performance. However, in this study, the subjects 42WHITEN, HORNER, LITCHFIELD, AND MARSHALL-PESCINI were allowed to see the starting state as well as the end- strator in the failure phase” (p. 388). Their logic is that state. Such experiences were as powerful as having seen copying in the first attempt-but-fail case implied recog- a model’s behavioral attempts in causing children to nition of what was intended, rather than a simpler pro- achieve the same ends. cess such as stimulus enhancement, because if the latter We should also note that Wood (1989) described em- had been operative the subjects would have copied ulation as achieving “common goals to those modeled attempt-but-fail if this occurred in the later trial. How- ... by using idiosyncratic means that were never ob- ever, an alternative explanation is that in the first trial, served” (p. 71). If we include the latter in defining end- the subject used the best information available, a failed state emulation, the results in Tomasello etal. (1987) do attempt in this case; when the failed attempt instead not, in fact, fit very well, insofar as the chimpanzees did came in the second trial, this was not copied, because the not use alternative means to replicate the result achieved subject now knew about a successful technique. Myowa- by the model when food was in the trapped position; in- Yamakoshi and Matsuzawa’s (2000) approach is, never- stead, they failed at the task. Crude OMR thus seems a theless, an ingenious one that merits further refinement more apt interpretation. to tackle this critical issue. In our studies with artificial fruits, we have repeatedly 3a. Affordance learning: The essential distinctions . found that great apes can, and often do, attain the end- Emulation has also been described as learning the affor- states achieved by the model, preferentially using their dances of objects, a term borrowed from perception the- own methods where these are effective. Thus, in chim- ory and first used in delineating emulation by Tomasello, panzee experiments (e.g., Whiten etal., 1996), subjects Kruger, and Ratner (1993). Following Byrne (1998) and that had witnessed a model turn a handle to release a lid Custance etal. (1999), we use the term affordance learn- quickly pulled the handle out instead (the method, in fact, ing to cover this idea in Figure1. As Byrne (1998) noted, witnessed by the other experimental group). Clearly this affordances could include properties of objects (e.g., is not OMR, by definition; nor does it suggest imitation, stone-hammer hardness), relations between objects (ker- for the actions are different. That strategies were not just nels are enmeshed in nut-cases), and functions (a stick what subjects might do without the benefit of a model is can act as a rake; Figure1). Unlike the senses of emula- indicated by our study of gorillas, which included nonob- tion described in earlier sections, affordance learning serving subjects. These control animals simply did not in- does not necessarily invoke a copying process. To the teract with the handle part of the fruit (Stoinski etal., contrary, having learned about the affordances of vari- 2001). The directness of experimental subjects’ actions ous objects, an individual may be able to exploit this on the handle leads us to favor end-result emulation over knowledge not only in the task observed, but in others stimulus enhancement as an explanation, although fur- too. Want and Harris (2002) suggested that the latter ther control studies will be necessary to clarify this. possibility could be one way to discriminate affordance Perhaps the best finding illustrating end-result emula- learning from the rest; its consequences ought to be re- tion comes from an experiment that incorporated social flected in measurable abilities to apply the factual learning, even though this was not its prime focus (Toth, knowledge obtained to comparable but different tasks. Schick, Savage-Rumbaugh, Sevcik, & Rumbaugh, 1993). An alternative approach was adopted in a different ar- The bonobo Kanzi, having observed a human striking ticle by Want and Harris (2001), showing that young stone flakes and using them to cut a cord to release food, children avoided copying a particular segment of a mod- began to make flakes in a similar way, hitting one stone on el’s actions that was unproductive in arriving at a final another. Having made only relatively inefficient flakes, he desired outcome. The authors interpreted this as the chil- then switched to a new techniq ue that had not been dren’s learning, instead, about critical affordances in the demonstrated: smashing a large stone on the ground. task (that poking a stick into the wrong end of a tube Later, he began to throw one stone onto another. Both of caused the contents to fall into a trap) and using this in- these creative variants appear to be just the kind of re- formation in their task solution. Thus, learning from a sponse originally defined as emulation by Wood (1989). model what to ignore might be another useful focus in Goal emulation is more difficult than result emulation the investigation of affordance learning. to identify in practice. Attempting to do so, Myowa- 3b. Affordance learning in apes . Horner and Whiten Yamakoshi and Matsuzawa (2000) studied what chim- (2004) have recently completed a study that echoes the panzees learned from watching a human model trying logic of Want and Harris (2001) in pursuing the hypoth- (and sometimes failing) to open a puzzle box. The model esis that, in certain contexts, chimpanzees will make either was successful with one technique (using a tool or adaptive choices either to imitate what a model does or to only the hand) and then unsuccessful with the alterna- more selectively ignore components that can be judged tive technique or was first unsuccessful with one and unproductive in their affordances. In one condition in this then successful with the other. After the first demon- study, subjects witnessed a human model who, among stration, the subjects tended to copy whichever of the ap- other actions, poked a stick-tool into the top of an opaque proaches they had seen. However, they copied the second box and then removed it, finally inserting it into another demonstration only if it was successful. The authors in- hole at the front to retrieve a food reward. In an alterna- terpret their results as showing that “the chimpanzees tive condition, subjects saw exactly the same thing, but understood something about the intention of the demon- the box was transparent, so that when the tool was in- HOW DO APES APE?43 serted into the top, an observer could see that it unpro- Similarly neglected in apes is observational condi- ductively struck a false floor. As was predicted, the young tioning, a phenomenon principally known from the way chimpanzees who saw the latter condition were signifi- macaques have been shown to acquire fear responses to cantly less likely than those working with the opaque box specific objects through observation (Mineka & Cook, to begin the task by inserting the tool into the top hole. 1988). Although described as social referencing , a term These results appear to mirror those obtained by Want used by those studying human infants, the findings in a and Harris (2001), suggesting that they represent a kind single study on this topic by Russell, Bard, and Adamson of affordance learning. It certainly does not involve im- (1997; see the Appendix) appear to reflect precisely the itation or OMR, for instead, a component of what the effect at stake. model did is explicitly omitted from the subject’s task so- lution. Could we instead have a case of end-state emula- ARE APES SPECIAL? tion? After all, in a sense, the subject attained the result it saw the model achieve, retrieving food from the front The above review suggests that apes do indeed ape, in hole, but by a technique that omits earlier parts of what both the narrow and the broad senses. A summary an- the model did. This would correspond to our nesting of swer to How do apes ape? is that they do so by exploit- end-state emulation under copying, insofar as subjects ing a portfolio of quite varied, complementary social- copy only the final part of the action sequence they wit- learning processes at their disposal. Selection within this ness. However, we do not favor this account, because we portfolio will be discussed below in the final section. suspect that if the irrelevant part of the task had come in However, as was noted earlier, there is now good exper- the middle or later, the same tendency to ignore it would imental evidence of imitation in other primates and in have been found. Accordingly, these results seem better birds, and the comparable evidence in apes cited so far may interpreted as implying learning about affordances— appear to be no stronger (Caldwell & Whiten, 2002; specifically, learning that poking the stick into the top Heyes, 2001). For example, both starlings and budgerigars hole does not afford any productive causal linkage with have copied the technique of gaining food by either stab- achieving the critical outcome of food retrieval. bing down a stopper that was in the way or, alternatively, A study by Bard, Fragaszy, and Visalberghi (1995) picking it out, according to which model they observed represents the alternative approach to identifying affor- (Campbell, Heyes, & Goldsmith, 1999; Heyes & Sagger- dance learning, outlined earlier—that of investigating son, 2002). Likewise, chimpanzees copied either poking whether information acquired can be used in relation to bolts out of the way to open an artificial fruit or, alterna- tasks that are different from that observed. Bard etal. tively, pulling them out with a twisting action (Whiten showed that if young chimpanzees had mastered the etal., 1996). Given that apes have long had a reputation for probing of food out of a tube following observation of a being special in their social-learning capacities, we might model, they were less likely than nonobservers to make ask how well this reputation is still deserved. Many issues errors when faced with more complex problems of this are raised by this question. Here, we will discuss five. kind that required modification of the tool before it could be used satisfactorily. This finding, however, con- 1. Variety, Novelty, and Manual Dexterity cerned only two pairs of chimpanzees and, therefore, A starling can peck or grip things with its bill but can could not be evaluated statistically; it remains possible manipulate objects in limited ways, as compared with an that the apparent effects of observation were not affor- ape, who has the use of two, many-jointed, five-fingered dance learning but merely reflected individual differ- hands that offer great flexibility in action repertoires. ences. This study is, accordingly, most valuable as a Each hand is capable of several different grips—on top unique demonstration of the potential in this approach. of which, the hands can be used in complementary ways. 4. Other categories of social learning . Space does The repertoire this may generate is indicated by the cat- not permit here an equivalent in-depth analysis of the re- alogue of 72 functionally distinct manipulative actions maining categories of social learning defined in Figure1, identified by Byrne, Corp, and Byrne (2001) for moun- enhancement and observational conditioning; we per- tain gorilla foraging. Adding the use of a variety of tools force mention them briefly for completeness. Stimulus affords a further combinatorial explosion in the reper- enhancement is recognized as a cognitively low-level pro- toire of potential action patterns and in the complexity of cess, quite widespread among animals (Whiten & Ham, their combinations that might be learnable only by suf- 1992). As was noted above, when chimpanzees in a two- ficiently sophisticated copying mechanisms. action experiment show no evidence of imitation but have These contrasts are reflected in the actions that have clearly learned something nonobserving controls have been shown to be imitated by different species. For exam- not (Stoinski & Whiten, 2003; Stoinski etal., 2001), ple, among birds, the alternatives copied in two-action stimulus enhancement may be involved. However, re- tests of imitation have included pecking on a treadle ver- searchers’ preoccupation with the presumably more com- sus stepping on it (pigeons, Zentall etal., 1996; quail, plex processes of imitation and emulation means that Akins & Zentall, 1998) and stabbing versus picking up stimulus enhancement is underresearched as a category in a stopper (e.g., starlings, Campbell etal., 1999). The dif- its own right. Too often, it is merely the default explana- ferences between these pairs of alternatives appear to be tion when more complex processes remain unconfirmed. relatively simple. 44WHITEN, HORNER, LITCHFIELD, AND MARSHALL-PESCINI

The variety available to apes may be contrasted for Together, these findings suggest that apes may have a ca- both bodily and object-centered patternings of behavior. pacity for copying a much greater variety of actions that With respect to bodily imitation, the finding that mar- are relatively novel to them that is not shared by many of mosets would copy mouth versus hand use (Voelkl & the other species studied. Huber, 2000) and pigeons likewise would copy beak ver- Of course, the ape complexities outlined in this sec- sus foot use (Zentall etal., 1996) was important in class- tion need to be subjected to more direct comparative re- ing these as true imitation . Use of different body parts to search (Caldwell & Whiten, 2002). If they are borne out, deal with the same task cannot be explained as emulation this would suggest a functional fit with the variety of pu- but, rather, as imitation that relies on visuomotor mapping tative traditions that studies in the wild suggest are adap- from seen parts of the model’s body to equivalent parts tive for apes to acquire (van Schaik etal., 2003; Whiten of the self. In apes, this mapping has been most graphically etal., 1999, 2001). In the case of chimpanzees, for exam- charted in the do-as-I-do studies. In Custance etal. (1995), ple, the range of such behaviors spans different methods for example, coders blind to what each chimpanzee had for dispatching ectoparasites (stab with finger, squash on actually watched identified matching in relation to touch- leaf, or inspect on leaf and then eat or discard) and use ing several parts of the body in sight (e.g., the stomach), of a variety of tools, ranging from those for powerful as well as out of sight (e.g., the back of the head), sym- pounding to delicate probing, and grooming techniques metric and asymmetric conjunctions of hands (e.g., clap (Whiten etal., 2001). It could thus be informative to ad- back of other hand), digit movements (e.g., wiggle fin- dress further comparisons with species of birds that ma- gers), and head or whole-body actions (e.g., look up, hug nipulate tools or other objects in complex ways (Huber, self): 34 varieties in all (Custance etal., 1995). The ex- Rechberger, & Taborsky, 2001; Hunt & Gray, 2003). tent of such variation that can be copied at the bodily level appears to far exceed that possible for species that do not 2. Selective Copying: Sensitivity to Causal possess comparable flexibility. The scope of all that is Relevance copied in do-as-I-do studies implies corresponding de- Studies with birds have established that social learn- mands on the underlying action representation systems. ing can become more likely when the model is seen to Manipulation of objects and combinations of objects gain a reward (pigeons, Palameta & Lefebvre, 1985; (as in tool use) raises the prospect of additional variety to quail, Akins & Zentall, 1998). This suggests that the im- challenge copying processes. Thus, for example, Myowa- itation identified in Akins and Zentall’s study was not Yamakoshi and Matsuzawa (1999), having modeled var- mere mimicking (done blindly, irrespective of the end ious arbitrary actions on pairs of objects, reported a gained) but, instead, involved decision-making processes chimpanzee copying such distinctions as (1) putting a sensitive to what actions by the model were associated hose into a hole in a stool, (2) hanging the hose around with desirable outcomes. the back of the stool, and (3) pushing the hose against the The experiment by Horner and Whiten (2004), de- stool. With respect to functional actions involving ob- scribed earlier, took this issue further. This study inves- jects, Bering, Bjorklund, and Ragan (2000; see the Ap- tigated whether chimpanzees are sensitive to aspects of pendix) reported both orangutans and chimpanzees copy- tool use that, to a human at least, would appear to have ing such actions as picking an object up with tongs, which no causal connection with desirable outcomes observed. they had not done in baseline trials. Recall that young chimpanzees faced with the transpar- The variations in actions thus engendered have impli- ent version of the task, in which they could see a stick- cations for the related issue of novelty. Authors vary in tool ineffectually beating on a false floor, subsequently whether they insist on including novelty of the action tended to omit this action from their efforts. By contrast, pattern as a criterion for identifying imitation (Huber, when the task was opaque, the action was included sig- 1998), but many would concur that where there is evi- nificantly more often. dence of novelty, the ascription of imitation is clearer. From this, it was concluded that chimpanzees do not The generativity in ape behavior outlined above means merely mimic what they see but perform a cognitive ap- that the potential for novel action patterns is correspond- praisal that identifies plausible causal connections before ingly increased. Although no behavior can ever be ab- including an action in any imitation performed (Horner solutely novel—just one of the reasons novelty is diffi- & Whiten, 2004). This does not necessarily imply a deep cult to measure (Whiten & Custance, 1996)—the studies grasp of causality but, rather, the appreciation of such cited above to illustrate variety in what apes may imitate rules as that physical connection is necessary for a tool all made efforts to establish how improbable the act in to make desirable outcomes possible. Whether this will question would naturally be for the experimental sub- turn out to be a phenomenon special to apes we do not jects. Thus, Custance etal. (1995) tested do-as-I-do with know, for we believe that no similar tests have yet been a set of actions not used in the training phase and cate- made with other species. gorized the known novelty of these to the subjects; Myowa-Yamakoshi and Matsuzawa (1999) and Bering 3. Structural Complexity of Actions etal. (2000) tested for imitation only if a target action Byrne (1995) has argued that certain methods of for- had not been spontaneously produced in baseline trials. aging employed by apes are especially structurally com- HOW DO APES APE?45 plex and that imitation may occur at a matching level, panzees showed a significant tendency to copy the hier- which he dubbed the program level . Byrne and Russon archically organized alternative they had seen, the first (1998) defined the latter as “copying the structural organ- evidence of imitation of hierarchical structure, we be- isation of a complex process including the sequence of lieve, in a nonhuman species. However, it does not seem stages, subroutine structure and bimanual organisation ... to conform to Byrne and Russon’s (1998) definition of while furnishing the exact details of actions by individual program-level imitation, insofar as the subjects did copy learning” (p. 676). They offered the foraging techniques of the form of particular aspects of component actions but wild gorillas and orangutans’ manipulation of human ar- failed to show copying of sequential order (e.g., left–right, tifacts, such as hammocks, as possible examples, although vs. right–left across a row). as the peer discussion accompanying the article empha- We think Horner and Whiten’s (2004) study impor- sized, it was not yet possible to test these hypotheses. tantly complements these experiments. In the HAF study Whiten (1998a) pointed out that imitation might occur and in those by Whiten (1998b, 2002b, 2002c), the se- in relation to separable parts of Byrne and Russon’s quential and hierarchical structures modeled were arbi- (1998) complex definition—most interestingly, with re- trary alternatives, whereas in Horner and Whiten’s study, spect to either sequential or hierarchical organization. they were logically necessary, as they would be naturally. This surmise was borne out in an experiment designed Recall that in the opaque box condition, there was a sig- explicitly to test for copying of these kinds (Whiten, nificant tendency to begin with the top hole and then pro- 1999, 2002b), in which young children copied the hier- ceed to the lower one. In the transparent condition, by archical structure of a task, but not the specific sequence contrast, the action on the top hole was more likely to be of actions models had used to implement it. omitted. Thus, we have evidence of sequence copying, With apes, imitation of sequential structure has been manifested systematically in the opaque condition. But tested by allowing chimpanzees to observe one of two al- there is also copying of what the chimpanzee appears to ternative models that differed in the sequential order in perceive as hierarchical structure, for the two conditions which they dismantled the defenses on an artificial fruit. together show that subjects have a capacity to parse Over the course of three trials, incorporating five sepa- (Byrne, 2000) the component actions, omitting one in the rate demonstrations, subjects converged (without being transparent condition, yet assembling both in the ob- rewarded for doing so) almost perfectly on the particular served way in the opaque condition. They thus copied a sequencing they had witnessed (Whiten, 1998b). sequential order that was perceivable as nonarbitrary. No such effect was found in replications with gorillas Are apes special in showing these complexities in (Stoinski etal., 2001) or orangutans (Stoinski & Whiten, copying? Chimpanzees are the only animals to have shown 2003). However, these negative results should be treated these effects, but to our knowledge, other species remain with caution. The two alternative sequential orders of to be tested in comparable ways. dismantling are, of course, entirely arbitrary, and the ab- sence of copying may reflect not an inability to copy a 4. Recognition of the Copying Process sequence, but a judgment by the subject that the se- Whiten (2000) has argued that the very fact that chim- quence is irrelevant (young children have also ignored panzees and orangutans have been able to learn the do- the sequential order of actions, while copying higher as-I-do routine means that they have (or can acquire) a level hierarchical structuring; Whiten, 2002c). An alter- concept of imitation. This has been achieved only in native kind of experiment is thus one in which the se- great apes (see the Appendix) and dolphins (Herman, quential order is, instead, perceivable as logically neces- 2002). It has not proved possible to train monkeys to per- sary. This is, in fact, what appeared to be the case in the form the task (macaques, Mitchell & Anderson, 1993; study by Horner and Whiten (2004) with opaque and capuchins, Fragaszy, Deputte, Hemery, & Johnston, un- transparent boxes, described above. However, before re- published, cited in Visalberghi & Fragaszy, 2002). It visiting this, we will describe a recent study in which the seems unlikely to be a coincidence that great apes copying of hierarchical structure was explicitly addressed, (Tomasello & Call, 1997) and dolphins (Reiss & Marino, for we can then draw some more integrative conclusions. 2001) are also the only species to recognize themselves The logic of the hierarchically structured artificial in mirrors. These taxa appear to share with humans a fruit (HAF) experiment followed that in experiments special capacity for self-representation. In the case of performed earlier with children (Whiten, 2002b, 2002c). imitation, this constitutes a second-order cognitive pro- The HAF used by Marshall-Pescini and Whiten (2004) cess; they recognize the act of imitation itself. The im- required 12 different actions to be completed—made up plication of this apparent sophistication for the nature of of three sets of the actions remove bolts , peel back lid , imitation in apes remains to be understood. One possi- slide trap door , and turn knurled wheel —before food be- bility is that it affords a greater sophistication of execu- came accessible. In the columns method, a model per- tive control over when imitation is deployed, in prefer- formed each 4-action set (dealing with bolts–lid –door– ence to other options distinguished in Figure1. wheel) as a whole, then proceeded to the next. In the rows method, by contrast, they removed all three bolts in 5. Enculturation a row, then opened all three doors, then slid all three As was noted earlier, the evidence for imitation of the doors, and finally turned all three wheels. Juvenile chim- highest fidelity has typically come from apes that have 46WHITEN, HORNER, LITCHFIELD, AND MARSHALL-PESCINI been raised with much interaction with humans. Conse- points in which the cognitive system of the learner is de- quently, it has been suggested that their imitativeness re- ciding whether suitable means are already known that sults from an enculturation process, in which their at- will achieve desirable ends observed or, alternatively, tention is channeled in ways like those experienced by whether aspects of the observed means-to-ends need to children (Tomasello, Kruger, & Ratner, 1993). Given be adopted. The following sketch of this process is es- that this preoccupation is a feature of the ape literature, sentially a series of working hypotheses. Here, we will whereas enculturation is not mentioned by those research- use the shorthand of talking of the learner’s making de- ing imitation in birds and other taxa, it seems warranted cisions, but of course it is more accurate to say that the to raise this issue under the heading, Are apes special? brain or some part of the cognitive system is making the For some, the question of enculturation is a major fac- decisions, and there is no implication (at least, no neces- tor that deserves to be addressed at the start of any re- sary one) of conscious decision making. view (Tomasello & Call, 1997). We have not done this 1. The most basic initial step in the decision tree lies for two related reasons. One is a concern that some of at the level of stimulus or local enhancement, where the the evidence relied on for this distinction has its own learner’s attention is drawn to an object or location by problems. Thus, in the foundational study of Tomasello, another individual but the decision is to tackle the task Savage-Rumbaugh, and Kruger (1993) human-reared and by using behavioral resources already available to itself, mother-reared chimpanzees were compared, using a task rather than to learn more from the other individual. in which the apes were instructed to do what I do . How- 2. Alternatively, the learner may recognize a new out- ever, the mother-reared subjects were not trained to cri- come or result connected with the actions of the other in- terion in the manner of Custance etal. (1995), to check dividual (who now begins, in a minimal way, to deserve whether they were responding to this cue in the same the title model) and may recognize that the behavioral way as the human-reared subjects. In addition, the latter resources are at its disposal to recreate this result, rather were, of course, more familiar with their human testers. than utilizing any other information available. This A trio of studies by Bjorklund and colleagues (see the would be end-state emulation. If the learner perceives Appendix), interpreted as supporting the enculturation the outcomes (or in this case, what may be only potential hypothesis, in fact involved only human-reared subjects. outcomes) as goals, this is goal emulation; if it seeks The second concern is that the enculturation experi- only to recreate the results it saw, this is result emulation. ence may actually be replicating experiences more typi- 3. If, on the other hand, the decision is that the learner cal for wild apes, in contrast to those of captive apes that does not have the behavioral resources available for em- have been deprived of the ontogenetic inputs necessary ulation, imitation comes into play. However, the imita- for the more sophisticated kinds of social learning to de- tion process itself incorporates a set of decisions con- velop (Boesch, 1993; Whiten, 1993). Field experiments cerned with selective copying. As is indicated in Figure1 may be necessary before such matters can be settled. and the studies reviewed earlier, this copying may focus However, it should be noted that the new findings in on aspects of either bodily or object actions or, simulta- Horner and Whiten (2004), and Marshall-Pescini and neously, on patterning in both. However, several of the Whiten (2004), described above, were obtained in work studies reviewed above have suggested that the search with wild-born juveniles living in the Ngamba Island process initially focuses on what, in relation to desirable Chimpanzee Sanctuary in Uganda. Just what the neces- ends, are the closest plausible causal actions of the model. sary developmental inputs to social-learning compe- In the case of tool use, this will involve a focus on the ac- tence are (Heyes, 2002; Whiten, 2000) and whether apes tions of the tool itself in relation to its target, and only are at all special in requiring them, as compared with this aspect of the form of the actions seen may be copied, other taxa, are questions that remain to be resolved. neglecting the particular bodily movements the model used to make the tool do what it did. The tendency of CONCLUSIONS: HOW DO APES APE? chimpanzees and orangutans to focus on what manipu- lated objects are doing, at the expense of bodily fidelity, The evidence that has now accumulated shows that might reflect this bias. apes do ape, but how they do so cannot be given any sim- In suggesting that the learner is identifying plausible ple unitary answer, such as that apes are imitators or that causal actions , we are not implying that the learner must they are emulators. Rather, the picture emerging is that have a deep appreciation of the concept of causality. There apes may develop a portfolio of social-learning pro- is much evidence that apes do not closely approach this cesses that are differentially activated in different con- (Povinelli, Reaux, Theall, & Giambrone, 2000). What is texts and in relation to different functional requirements. important is that the learner has some way of function- We can now begin to sketch the latter in terms of decision ally recognizing which acts of the model are causing the trees through which the ape brain works in the course of outcome of interest. This could, in principle, be achieved social learning. Taking the case of learning about a task in a number of ways, such as by identifying strong sta- like the foraging problems documented in the wild and tistical associations where repeated viewing is possible woven into our experiments, we can begin to build a the- (Byrne, 2000) or by applying perceptual rules of thumb , ory of the underlying processes. These processes appear such as contingency, contact, and so on, which we al- to be enmeshed in hierarchical decision trees, at many luded to in discussing the study by Horner and Whiten HOW DO APES APE?47

(2004), and which could even be used by subjects to Morimura (2000), which involved chimpanzees probing learn from one-off demonstrations. through a narrow hole for honey. Learners observed those 4. Parts of the research reviewed above support Byrne’s already expert before they made their first attempts, but (1995; Byrne & Russon, 1998) suggestion that imitation not after they had been successful. Most interesting, they may occur at different levels within an action’s organi- also observed after making failed attempts at solving the zation—for example, at the top levels of overall sequen- task. Thus, they appeared to selectively attend to a model tial or hierarchical structure or, alternatively, down at the in order to gain the specific information they needed. level of details of certain components. This raises the Conversely, there is some limited evidence for confor- possibility that apes might have a systematic bias favor- mity effects, whereby the learner, having achieved some ing learning at one level rather than at another. Whether competence, converges on the actions of others even this is the case remains unclear as yet, but two different though it has used other approaches with success (Whiten, aspects of the question deserve brief discussion. 1998b). First, Byrne and Russon (1998) suggested that natu- 6. Instead of trying to copy the model, the learner may ralistic data for gorilla feeding and orangutan object ma- favor learning the affordances of tools or other objects nipulation suggest that the top level ( program level ) involved and may use this knowledge ( learning that , might be favored. However, in the first experimental rather than learning how ) to construct its own attempt at tests of this hypothesis, evidence of imitation was found the task. Such a process may often be difficult to dis- primarily in relation to details, and not overall sequenc- criminate from imitation, but as we discussed earlier, it ing (gorillas, Stoinski etal., 2001; orangutans, Stoinski may be manifested in selective copying that omits elements & Whiten, 2003). Stoinski etal. accordingly suggested that do not have productive affordances or in transfer of that we need to entertain the hypothesis converse to that of learning to other kinds of tasks. Byrne and Russon—namely, that these apes might learn In conclusion, we believe that the emerging answer to critical details by observation and might assemble the how apes ape is, accordingly, not a unitary one but, in- overall structure of a complex act themselves, a process stead, is best expressed in terms of biases and choices Russon (1999) called hierarchization . As Stoinski etal. among the varied range of options we have summarized noted, many of the key findings reviewed by Byrne and above. Some of these reflect inherent limitations, but Russon, for both wild and tame apes, are quite consis- others represent adaptive flexibility. The last 15 years tent with this. Clearly, more research is needed to differ- have yielded a rich cluster of new insights into ape social entiate these alternatives. learning, and along the way, the research involved has A second possibility is that apes initially focus prefer- highlighted theoretical and methodological implications entially on one level (high or low) and then, as they mas- important for the wider field of animal social learning. ter this and build up a scheme of action, focus more social learning on other, complementary levels. Some of our re- REFERENCES sults converge on the conclusion that this may indeed be Akins, C. K., & Zentall, T. R. (1998). Imitation in Japanese quail: the case and that details are learned first. Thus, Whiten The role of reinforcement of demonstrator responding. Psychonomic (1998b) found evidence of copying actions on individual Bulletin & Review , 5, 694-697. parts of an artificial fruit in the first trials, whereas high Bard, K. A., Fragaszy, D., & Visalberghi, E. (1995). Acquisition and fidelity of overall sequential copying emerged only with comprehension of a tool-using behavior by young chimpanzees ( Pan troglodytes ): Effects of age and modeling. International Journal of repeated viewing and performing. Marshall-Pescini and Comparative Psychology , 8, 47-68. Whiten (2004) found that evidence of hierarchical copy- Bering, J. M., Bjorklund, D. F., & Ragan, P. (2000). Deferred imita- ing emerged only in older juveniles, whereas imitation of tion of object-related actions in human-reared juvenile chimpanzees individual elements occurred already in younger infants. and orangutans. Developmental Psychobiology , 36, 218-232. Such results suggest that chimpanzees first focus on Bjorklund, D. F., Bering, J. M., & Ragan, P. (2000). A two-year lon- gitudinal study of deferred imitation of object manipulation in a ju- learning what they perceive to be critical elements of the venile chimpanzee ( Pan troglodytes ) and orangutan ( Pongo pyg- task and, as they master these, are progressively more able maeus). Developmental Psychobiology , 37, 229-237. to attend to higher level patterning. However, in the study Bjorklund, D. F., Yunger, J. L., Bering, J. M., & Ragan, P. (2002). by Horner and Whiten (2004),evidence of sequential The generalization of deferred imitation in enculturated chimpanzees (Pan troglodytes ). Animal Cognition , 5, 49-58. matching was found from the start, together with some im- Boesch, C. (1993). Towards a new image of culture in wild chimpanzees? itation of elements closest to the end-result of food being Behavioral & Brain Sciences , 16, 514-515. fished out. One explanation for this difference in find- Byrne, R. W. (1995). The thinking ape: Evolutionary origins of intel- ings may be that the latter study incorporated a task eas- ligence. New York: Oxford University Press. ier for chimpanzees to parse than the others. Again, more Byrne, R. W. (1998). Commentary on Boesch, C. and Tomasello, M. Chimpanzee and human culture. Current Anthropology , 39, 604-605. research on this question is clearly needed. Byrne, R. W. (2000). Imitation without intentionality: Using string- 5. A point inherent in the foregoing discussion de- parsing to copy the organization of behaviour. Animal Cognition , 2, serves highlighting in its own right: How apes ape may 63-72. change in the course of learning a task. Once learning is Byrne, R. W., Corp, N., & Byrne, J. M. E. (2001). Estimating the com- plexity of animal behaviour: How mountain gorillas eat thistles. Be- underway, social learning may become focused on those haviour, 138, 525-557. aspects of the task the learner continues to struggle with. Byrne, R. W., & Russon, A. E. (1998). Learning by imitation: Ahier- An illustration of this comes from a study by Hirata and archical approach. Behavioral & Brain Sciences , 21, 667-721. 48WHITEN, HORNER, LITCHFIELD, AND MARSHALL-PESCINI

Caldwell, C. A., & Whiten, A. (2002). Evolutionary perspectives on tion in rats: The role of demonstrator action. Behavioral Processes , imitation: Is a comparative psychology of social learning possible? 32, 173-182. Animal Cognition , 5, 193-208. Heyes, C. M., & Saggerson, A. (2002). Testing for imitative and non- Call, J. (2001). Body imitation in an enculturated orangutan ( Pongo imitative social learning in the budgerigar using a two-object/two- pygmaeus). Cybernetics & Systems , 32, 97-119. action test. Animal Behaviour , 64, 851-859. Call, J., & Carpenter, M. (2002). Three sources of information in so- Hirata, S., & Morimura, N. (2000). Naive chimpanzee s’ ( Pan cial learning. In K. Dautenhahn & C. L. Nehaniv (Eds.), Imitation in troglodytes ) observation of experienced conspecifics in a tool-using animals and artifacts (pp.211-228). Cambridge, MA: MIT Press. task. Journal of Comparative Psychology , 114, 291-296. Call, J., & Tomasello, M. (1994). The social learning of tool use by Horner, V., & Whiten, A. (2004). Causal knowledge and imitation/ orangutans. Human Evolution , 9, 297-313. emulation switching in chimpanzees (Pan troglodytes) and children . Call, J., & Tomasello, M. (1995). The use of social information in the Manuscript under review. problem-solving of orangutans ( Pongo pygmaeus ) and human children Huang, C.-T., Heyes, C., & Charman, T. (2002). Infants’ behavioral (Homo sapiens ). Journal of Comparative Psychology , 109, 308-320. reenactment of “failed attempts”: Exploring the roles of emulation Campbell, F. M., Heyes, C. M., & Goldsmith, A. R. (1999). Stimulus learning, stimulus enhancement, and understanding of intentions. learning and response learning by observation in the European star- Developmental Psychology , 38, 840-855. ling, in a two-object/two-action test. Animal Behaviour , 58, 151-158. Huber, L. (1998). Movement imitation as faithful copying in the ab- Celli, L. M., Tomonaga, M., Udono, T., Teramoto, M., & Nagano,K. sence of insight. Behavioural & Brain Sciences , 21, 694. (2001). Learning processes in the acquisition of a tool using task by Huber, L., Rechberger, S., & Taborsky, M. (2001). Social learning captive chimpanzees. Psychologia , 44, 70-81. affects object exploration and manipulation in keas, Nestor notabilis . Custance, D. M. (1998). Apes ape! Behavioral & Brain Sciences , 21, Animal Behaviour , 62, 945-954. 118-119. Hunt, G. R., & Gray, R. D. (2003). Diversification and cumulative Custance, D. M., Whiten, A., & Bard, K. A. (1995). Can young evolution in New Caledonian crow tool manufacture. Proceedings of chimpanzees ( Pan troglodytes ) imitate arbitrary actions? Hayes and the Royal Society of London : Series B, 270, 867-874. Hayes (1952) revisited. Behaviour, 132, 837-859. Hurley, S., & Chater, N. (Eds.) (in press). Perspectives on imitation: Custance, D. M., Whiten, A., & Fredman, T. (1999). Social learning From cognitive neuroscience to social science . Cambridge, MA: MIT of artificial fruit processing in capuchin monkeys ( Cebus apella ). Press. Journal of Comparative Psychology , 113, 13-23. Imanishi, K. (1957). Identification: A process of enculturation in the Custance, D. M., Whiten, A., Sambrook, T., & Galdikas, B. (2001). subhuman society of Macaca fuscata . Primates, 1, 1-29. Testing for social learning in the ‘artificial fruit’ processing of wild- Laland, K., Odling-Smee, J., & Feldman, M. W. (2000). Niche con- born orangutans ( Pongo pygmaeus ), Tanjung Puting, Indonesia. An- struction, biological evolution and cultural change. Behavioral & imal Cognition , 4, 305-313. Brain Sciences , 23, 131-175. Dautenhahn, K., & Nehaniv, C. L. (Eds.) (2002). Imitation in ani- Lloyd-Morgan, C. L. (1896). Habit and instinct . London: Edward mals and artifacts . Cambridge, MA: MIT Press. Arnold. Galef, B. G., Jr. (1988). Imitation in animals: History, definition, and MacFarland, D. J. (1984). Animal behaviour . London: Pitman. interpretation of data from the psychological laboratory. In T. R. Zen- Marshall-Pescini, S., & Whiten, A. (2004). Social learning of a hier- tall & B. G. Galef, Jr. (Eds.), Social learning: Psychological and bi- archically organised artificial foraging task by infant and juvenile ological perspectives (pp.3-28). Hillsdale, NJ: Erlbaum. chimpanzees . Manuscript under review. Galef, B. G., Jr. (1990). Traditions in animals: Field observations and McGrew, W. C. (1992). Chimpanzee material culture: Implications for laboratory analyses. In M. Bekoff & D. Jamieson (Eds.), Interpreta- human evolution . Cambridge: Cambridge University Press. tion and explanation in the study of animal behaviour (pp. 74-95). Meltzoff, A. N. (1995). Understanding the intentions of others: Re- Boulder: Westview Press. enactment of intended acts by 18-month-old children. Developmen- Galef, B. G., Jr. (1992). The question of animal culture. Human Na- tal Psychology , 31, 838-850. ture, 3, 157-178. Meltzoff, A. N., & Prinz, W. (2002). The imitative mind: Development, Galef, B. G., Jr. (2004). Approaches to the study of traditional behav- evolution and brain bases . Cambridge: Cambridge University Press. iors of free-living animals. Learning & Cognition , 32, 53-61. Miklosi, A. (1999). The ethological analysis of imitation. Biological Galef, B. G., Jr. (in press). Breathing new life into the study of animal Review, 74, 347-374. imitation: What and when do chimpanzees imitate? In S. Hurley & Miles, H. L., Mitchell, R. W., & Harper, S. E. (1996). Simon says: The N. Chater (Eds.), Perspectives on imitation: From cognitive neuro- development of imitation in an enculturated orangutan. In A.Russon, science to social science . Boston, MA: MIT Press. K. A. Bard, & S. T. Parker (Eds.), Reaching into thought: The minds of Goodall, J. (1968). The behaviour of free-living chimpanzees in the the great apes (pp. 278-299). Cambridge: Cambridge University Press. Gombe Stream Reserve. Animal Behaviour Monographs , 1, 161-311. Mineka, S., & Cook, M. (1988). Social learning and the acquisition of Goodall, J. (1973). Cultural elements in a chimpanzee community. In snake fear in monkeys. In T. R. Zentall & B. J. Galef, Jr. (Eds.), So- E. W. Menzel (Ed.), Precultural primate behaviour (Vol. 1, pp. 144- cial learning: Psychological and biological perspectives (pp. 51-73). 184). Basel: Karger. Hillsdale, NJ: Erlbaum. Hannah, A., & McGrew, W. (1987). Chimpanzees using stones to Mitchell, R. W. (1999). Scientific and popular conceptions of the psy- crack open oil palm nuts in Liberia. Primates, 28, 31-46. chology of great apes from the 1790s to the 1970s: Déjà vu all over Hayes, C. (1951). The ape in our house . New York: Harper. again. Primate Report , 53, 1-118. Hayes, K. J., & Hayes, C. (1952). Imitation in a home-raised chim- Mitchell, R. W., & Anderson, J. R. (1993). Discrimination learning panzee. Journal of Comparative & Physiological Psychology , 45, of scratching, but failure to obtain imitation and self-recognition in a 450-459. long-tailed macaque. Primates, 34, 301-309. Herman, L. M. (2002). Vocal, social and self-imitation by bottlenosed Myowa, M. (1996). Imitation of facial gestures by an infant chim- dolphins. In K. Dautenhahn & C. L. Nehaniv (Eds.), Imitation in an- panzee. Primates, 37, 207-213. imals and artifacts (pp. 63-108). Cambridge, MA: MIT Press. Myowa-Yamakoshi, M., & Matsuzawa, T. (1999). Factors influencing Heyes, C. M. (1993). Imitation, culture and cognition. Animal Behav- imitation of manipulatory actions in chimpanzees ( Pan troglodytes ). iour, 46, 999-1010. Journal of Comparative Psychology , 113, 128-136. Heyes, C. M. (2001). Causes and consequences of imitation. Trends in Myowa-Yamakoshi, M., & Matsuzawa, T. (2000). Imitation of in- Cognitive Sciences , 5, 253-261. tentional manipulatory actions in chimpanzees ( Pan troglodytes ). Heyes, C. M. (2002). Transformational and associative theories of im- Journal of Comparative Psychology , 114, 381-391. itation. In K. Dautenhahn & C. L. Nehaniv (Eds.), Imitation in ani- Nagell, K., Olguin, R. S., & Tomasello, M. (1993). Processes of so- mals and artifacts (pp.501-523). Cambridge, MA: MIT Press. cial learning in the tool use of chimpanzees ( Pan troglodytes ) and Heyes, C. M., Jaldow, E., Nokes, T., & Dawson, G. R. (1994). Imita- human children ( Homo sapiens ). Journal of Comparative Psychol- ogy, 107, 174-186. HOW DO APES APE?49

Nishida, T. (1987). Local traditions and cultural transmission. In B. B. Visalberghi, E., & Fragaszy, D. (1990). Do monkeys ape? In S.Parker Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham, & T. T. & K. Gibson (Eds.), Language and intelligence in monkeys and apes: Struhsaker (Eds.), Primate societies (pp. 462-474). Chicago: Uni- Comparative developmental perspectives (pp. 247-273). Cambridge: versity of Chicago Press. Cambridge University Press. Palameta, B., & Lefebvre, L. (1985). The social transmission of a Visalberghi, E., & Fragaszy, D. M. (2002). Do monkeys ape? Ten years food finding technique in pigeons: What is learned? Animal Behav- after. In K. Dautenhahn & C. L. Nehaniv (Eds.), Imitation in animals iour, 33, 892-896. and artifacts (pp.471-500). Cambridge, MA: MIT Press. Paquette, D. (1992). Discovering and learning tool-use for fishing Voelkl, B., & Huber, L. (2000). True imitation in marmosets. Animal honey by captive chimpanzees. Human Evolution , 7, 17-30. Behaviour, 60, 195-202. Passingham, R. E. (1982). The human primate . New York: Freeman. Want, S. C., & Harris, P. L. (2001). Learning from other people’s mis- Povinelli, D. J., Reaux, J. E., Theall, L. A., & Giambrone, S. (2000). takes: Causal understanding in learning to use a tool. Child Devel- The question of tool modification. In D. J. Povinelli (Ed.), Folk opment, 72, 431-443. physics for apes: A chimpanzee’s theory of how the world works Want, S. C., & Harris, P. L. (2002). How do children ape? Applying (pp.271-296). Oxford: Oxford University Press. concepts from the study of non-human primates to the developmen- Reiss, D., & Marino, L. (2001). Mirror self-recognition in the bottlenose tal study of“ imitation” in children. Developmental Science , 5, 1-13. dolphin: Acase of cognitive convergence. PNAS, 98, 5937-5942. Whiten, A. (1993). Human enculturation, chimpanzee enculturation Rizzolatti, G., Fadiga, L., Fogassi, L., & Gallese, V. (2002). From and the nature of imitation:Commentary on Cultural learning, by mirror neurons to imitation: Facts and speculations. In A. N. Meltzoff M.Tomasello et al. Behavioral & Brain Sciences , 16, 538-539. & W. Prinz (Eds.), The imitative mind (pp. 247-266). Cambridge: Whiten, A. (1998a). How imitators represent the imitated: The vital Cambridge University Press. experiments. Behavioral & Brain Sciences , 21, 707-708. Russell, C. L., Bard, K. A., & Adamson, L. B. (1997). Social refer- Whiten, A. (1998b). Imitation of the sequential structure of actions by encing by young chimpanzees ( Pan troglodytes ). Journal of Com- chimpanzees ( Pan troglodytes ). Journal of Comparative Psychology , parative Psychology , 111, 185-193. 112, 270-281. Russon, A. E. (1999). Orangutans’ imitation of tool use: Acognitive in- Whiten, A. (1999). Imitation of sequential and hierarchical structure in terpretation. In S. T. Parker, R. W. Mitchell, & H. L. Miles (Eds.), action: Experimental studies with children and chimpanzees. In Pro- Mentalities of gorillas and orangutans (pp. 119-145). Cambridge: ceedings of the Artificial Intelligence and Simulation of Behaviour Cambridge University Press. Convention (AISB): Symposium on Imitation in Animals and Arti- Shettleworth, S. J. (1998). Cognition, evolution and behaviour . Ox- facts (pp. 38-46). Edinburgh: Society for the Study of Artificial In- ford: Oxford University Press. telligence and Simulation of Behaviour. Stoinski, T. S., & Whiten, A. (2003). Social learning by orangutans Whiten, A. (2000). Primate culture and social learning. Cognitive Sci- (Pongo pygmaeus ) in a simulated food processing task. Journal of ence, 24, 477-508. Comparative Psychology , 117, 272-282. Whiten, A. (2002a). From the field to the laboratory and back again: Stoinski, T. S., Wrate, J. L., Ure, N., & Whiten, A. (2001). Imitative Culture and “social mind” in primates. In M. Bekoff, C. Allen, & G.M. learning by captive western lowland gorillas ( Gorilla gorilla gorilla ) Burghardt (Eds.), The cognitive animal (pp. 385-392). Cambridge, in a simulated food-processing task. Journal of Comparative Psy- MA: MIT Press. chology, 115, 272-281. Whiten, A. (2002b). Imitation of sequential and hierarchical structure Tomasello, M. (1990). Cultural transmission in the tool use and com- in action: Experimental studies with children and chimpanzees. In municatory signalling of chimpanzees? In S. Parker & K. Gibson K.Dautenhahn & C. L. Nehaniv (Eds.), Imitation in animals and ar- (Eds.), Language and intelligence in monkeys and apes: Comparative tifacts (pp.191-209). Cambridge, MA: MIT Press. developmental perspectives (pp. 274-311). Cambridge: Cambridge Whiten, A. (2002c). The imitator’s representations of the imitated. In University Press. A. Meltzoff & W. Prinz (Eds.), The imitative mind (pp. 98-121). Tomasello, M. (1996). Do apes ape? In C. M. Heyes & B. G. Galef, Jr. Cambridge: Cambridge University Press. (Eds.), Social learning in animals: The roots of culture (pp. 319-346). Whiten, A., & Custance, D. M. (1996). Studies of imitation in chim- London: Academic Press. panzees and children. In C. M. Heyes & B. G. Galef, Jr. (Eds.), So- Tomasello, M. (1998). Emulation learning and cultural learning. Be- cial learning in animals: Roots of culture (pp.291-318). London: havioral & Brain Sciences , 21, 703-704. Academic Press. Tomasello, M., & Call, J. (1997). Primate cognition . New York: Ox- Whiten, A., Custance, D. M., Gomez, J.-C., Teixidor, P., & Bard, ford University Press. K. A. (1996). Imitative learning of artificial fruit processing in chil- Tomasello, M., Call, J., & Gluckman, A. (1997). Comprehension of dren (Homo sapiens ) and chimpanzees ( Pan troglodytes ). Journal of novel communicative signs by apes and human children. Child De- Comparative Psychology , 110, 3-14. velopment, 68, 1067-1080. Whiten, A., Goodall, J., McGrew, W. C., Nishida, T., Reynolds,V ., Tomasello, M., Davis-DaSilva, M., Camak, L., & Bard, K. (1987). Sugiyama, Y., Tutin, C.E.G., Wrangham, R.W., & Boesch, C. Observational learning of tool-use by young chimpanzees. Human (1999). Cultures in chimpanzees. Nature, 399, 682-685. Evolution, 2, 175-183. Whiten, A., Goodall, J., McGrew, W. C., Nishida, T., Reynolds,V ., Tomasello, M., Kruger, A. C., & Ratner, H. H. (1993). Cultural Sugiyama, Y., Tutin, C.E.G., Wrangham, R.W., & Boesch, C. learning. Behavioral & Brain Sciences , 16, 495-552. (2001). Charting cultural variation in chimpanzees. Behaviour, 138, Tomasello, M., Savage-Rumbaugh, S., & Kruger, A. C. (1993). Im- 1481-1516. itative learning of actions on objects by children, chimpanzees, and Whiten, A., & Ham, R. (1992). On the nature and evolution of imitation enculturated chimpanzees. Child Development , 64, 1688-1705. in the animal kingdom: Reappraisal of a century of research. In P. J. B. Tonooka, R., Tomonaga, M., & Matsuzawa, T. (1997). Acquisition and Slater, J. S. Rosenblatt, C. Beer, & M. Milinski (Eds.), Advances in the transmission of tool making and use for drinking juice in a group of study of behavior (pp. 239-283). New York: Academic Press. captive chimpanzees. Japanese Psychological Research , 39, 253-265. Whiten, A., Horner, V., & Marshall-Pescini, S. R. J. (2003). Cul- Toth, N., Schick, K. D., Savage-Rumbaugh, E. S., Sevcik, R. A., & tural panthropology. Evolutionary Anthropology , 12, 92-105. Rumbaugh, D. M. (1993). Pan the tool-maker: Investigations into Wood, D. (1989). Social interaction as tutoring. In M. H. Bornstein & the stone tool-making and tool-using capabilities of a bonobo ( Pan J. S. Bruner (Eds.), Interaction in human development (pp. 59-80). paniscus). Journal of Archaelogical Science , 20, 81-91. Hillsdale, NJ: Erlbaum. van Schaik, C. P., Ancrenaz, M., Borgen, G., Galdikas, B., Knott, Zentall, T. R. (2004). Action imitation in birds. Learning & Cogni- C. D., Singleton, I., Suzuki, A., Utami, S.S., & Merrill, M. (2003). tion, 32, 15-23. Orangutan cultures and the evolution of material culture. Science, Zentall, T. R., Sutton, J. E., & Sherburne, L. M. (1996). True im- 299, 102-105. itative learning in pigeons. Psychological Science , 7, 343-346. 50WHITEN, HORNER, LITCHFIELD, AND MARSHALL-PESCINI

, a r 7 & 5

w o 9 , 9 , a f 9 h

9 z n , , ) 1 g & 1 a u

n n

, u d , s , , a a t a m n g e d z g g a l b k o r n e a a s i c e a r m M u m R R 6 e d B m e l u

o c 9 a o & & B G R n & & 9

d & G

- i i ,

e , , 1 , r , e r M h h A & d g ,

e e n

e s s g 9 , n n d s f c e g a o o l i 9 r & t e u l n r n k i v k k l 9 a b , a a a e u s h a a R 1 k t 6 8 d B 3

r S C s a r B Y 2 , 9 9

m 9 m W o , , a u , , a 0 9 9 T j a a 9 & , o o

d d 0 B 1 C 1 w l l Y 1 Y , e B

r l l n n 2 r , - - , a , , , c , l e e u u , o a a a o z r n n l n l l s s g e n d u e e e s a w w a w a k k n t i t s a g s t 0 0 0 7 i r r i i o o o t x n s s r g u m m 0 0 0 9 o o i h h y y a y r u u e j j a o o o 0 0 0 9 e i R M C M M T K B 2 B 2 B R M 2 T 1 W T W t c ? A g

L L E f n T i T T T T T T T T T I o U U N I I I I I I I I I n

r s M O M M R a M M M M M M M M M e I e S I I I I I I I N E I E I p L y T

n r , e

o l o n

i

n s i n h

e ’ t e d y

t n l o v o d .

d l c v

i i p . n e

e t a e

s d t t n a o n . o o a n h 7 c d - h n a e c a n - t n t o w s h e t l , a 1 e o o t i a o r i o i t o r h s t i i x i e h b t e t g a c t t d h d w e t a m e v m f l t u c t h w e s a c i e e n i n o i t i t j t t t e l o i g , a z z a i e j s s b h o l b c u e d

n a m n f h d b n c o e m . i e o 3 e b d I m t

n g a d

f f a o r i o l r d

h 1 l o . l f o e m . e f p r

o o — c o d l l s r e e . e e . 8 m f e l e t o s m s e o % a c s g m v o f d n o 4 u e s o d a

m e i n e s c 3 h f

c e o s n n e o t n h l d u d f d l i h h t o n c g 6 m o r y d n g r r t e o i o s o a c a s i A e g y h i n a i , t b c a

s

o i c e n l g . y t s u c n b s p s n m p y t p d i r e l i l

y w d n s o o n l p r u u u e o l f o e i

f i ” p r l k o e p e y r e t o m p o t b c a n t t d a o t t % s b c d a m p t o g a e p i c f a t d i t e n t s e r 8 e . m u n i r e c s o n x c a t t n o t s d i m c s y a y s i o i u 2 o d o c d e n s a c l t t a s b t o e t l

d a

u

l r . d i r l n n o e c c 4 n C . s p e f e n o o g s i a a e l j s w d n m e u u i o i g e a t t o

t i d t d e a ’ n t e i j e u o e b i f a P t c f o h m r n c l t i c e s n l t t n i h n i n i r g y b d u h l n o n i r t t e a h a u y t e a c s y f o w e c e a s ) s , j o l o t e e e a

o n t r c c a p

h h s v a r t n f n h d , y b u d e l f s m e h u e o p n l o c e s m n f o e i . e t e ) o s p t t o t e e r - t o o d w e a c r o i r f e r s i a , m o c b s d i o o u u i v e a t t 1 f n l t “ r e , 2 e g r f a i d e h r m h o c u n g r r o a b e u p n s j t o 3 y t u p A t n s o d c e l e e n r n e F e n s a d g i n n b c o t l s o o x a d a o e g s e t d p o r o g t t i v n n e e = f n c o a u e r s e

e r r v m i d n a e . x d o t a c t d s o o h o o t B i i d n t i l d d v i a o i s s t e e o t y i e t n s d u

r a t a e e e N n v w e d l g v f , c c s s g d t b e m e r s b t e c e w ( h

a i d d d c s e o a

X e e e n n n t s s d h a e o s a , o i i r t h d i d i h r a n n e y i I n t r s p t t t y o n a c s b l m d l a h r f e n e y f i w e e d r a t s a v i r l a e e r e e n

c t t o f s i e e s n t a s t D l p o b , a e e u e d a u . h e r d f i

n a a o l i r d d u a u s s s s s t t , o - d

c n s e i n d u

t n t t t t t e c o r N e m e e h ’ r o s c e t s s e m r l c i % l e o r q e c d d c c c c o e l i i n e t t v t r i f

r n b Y n r o i e r 3 E u m e e e e e e e e p e t h c a h a h n u e l j j j o j j a a r d e s o t h 6 t * d p t t e c t s r d e i i l a i P f t 5 i - e b m m b b b b ( j d o i s r r o t o e f d r n

u t o o e C e o y 1 p u o u u u u b s e a a P n c n h s n n m o u i h n m h M a I l I a t I a i S m M i c S f D m S o f S u S g S v B B t v e t n A e s e z a s s s h s n r r s s r k a L a a r r s s s s T a

e s p e e

a a s r r r r e r e e y y e e e a a a a d a y m s

h w y y e e e e g i r e 9 9 n t 5 y y y y a h y 5 1 1 . 1 6 A

a e

C 3 1 – 2 5 9 – 9 8 2 5 y s – – - 2 – – – 2 – – – 1 5 4 1 4 2 4 6 1 1 4 9 u n ) e s

s t f

o c G o

3 e b 7 2 5 3 1 3 5 0 7 4 . j 6 y * o 1 2 o b ( b n u N o S d b e t

s

i g o o o 3 t t t n

L i o s

y y y ( 3 t

n t ; t t t

; i n

i i i l c s g ; d a n n n s o e n n n t e r i i i t i e n e j s d t t i u u u t c

t t t t t p b e l s c m m m o e s r r r t

o e o j s e e a d e t e s A n o o o 0 0 0 - n m b d u e t p r r n 4 d p p p o 1 1 1 o t ; s

c o o q d e u a n e o p p p u + r r r n a n e e h l w i d r p o

l l t h m f r o o o t

b g l o o o r o e e i t o x e - e i i f f f ; ; ; ; ; ; ; ; ; e g s f g v e e t ” s t l l l l l l l l l l l d o

d

w a e d t o c s d d d ; e n e e e e e e e e e e e o o n o - t i b b o M a n e e e e e i d d d d d d d d d d d d d

s t e r r r , e m n M o o o o o o o o o o o n e r r r m e c g I h e

n o o z

t t l r r e e e a o n ; t h i o o m m m d m d m m m m m n m m i f f f b f i t t i a a d - - r

a + t s e e e a n 8 w I I s r e n n n n n e n n n n n n h e

i c e p l v - - d d d n t r 4 a a a a a a a a a a a a n t

a d r i a s s e r l w L g r r r o i m - y y y e e a a i e d t m m m m m m m m m m m n i i t l e c - - s o o f p p p s t t e o s u u u u u u u u u u u m h a o 5 o e b e c d f o o o i o e w h i H H a H d 1 H d H “ t d a H c H c H c t s H C o H d H t t m c

o

) )

d t t S s d i i

t o

n ; l t f n c u u d n a f

k o g y y a r r l e o s a o l o f f r j

o d e n e m m d ” t , , , i a o a s s o s b l l o m a f u s s s d r e o d u r a a h s s u e o t t t l t r b o u i i o n h i i n e t u e e f c , c , c , n s t p t h c c l a . . . i 2 s t e s s a s l b h e e ) e i i d e / t l c t r a h j g j g j g c 1 a n n m k f f n n i d . . . t c p o u m s n a i i i i e b b b s e e n m f e i s t t e i e e e j e b t s u h f f u w : l a o o o j r ( ( ( r r o s o - n

b “ r o

d e e s s ) i o g S e a a a T a b s w t u b n n n o e x x x e d d n n ( ( / h d n d b l 1 o , o ) l , d

o , i , o e o e o e s l c t p t - i o o s t o s , . l l l e e ; s i d n x x a s i i e i a c n a o t s s s s ) t e t m n e e t g p p p a i n p o t o o f u e e v l r t . c r r M c c e n n n n 3 - o j l a h h b b s o , n o i c e ( e u a a r m m m h h n i d t o o o o d r s b j u e t ( : s x t t e e i i i i n s

m i o o o o e d d s n a e s j s t t t t n s o b p 8 8 w o f s ” c c c e o d a e o o g g e h c c c c o e b n o o t o 4 4 b m m e r t t ) x i d r u h o o a a f a a l n n g a i r o d d d D r d t t l / f f i u i i e f f o n - n p s u d l l t r p l c s n n n s o l a y y y y o o v v m f a

e x o o s a l o r

g g d x a a a e r p r r r

o u t s o o s e o u A e e a a u l o y e y s - a a a e a n n n t t s n p e e e n t f f h i i l r r r m r r r b g b n e m m a l a n l l l e a t t t t s c x o a e e n n o a i i i i e e a g c n o p p p n r i t t e p e g i e r e r c p i m m w t t m b b b b c j n e a c o t g E o s c m n “ r r r r e p p a a o m m m o u e u a e i b r p r y y c a i i i e a r r o F r F n B f b B a o A 1 f A s t A s b A s d S o o N f c O b O b HOW DO APES APE?51

,

, h 6

n g , 9

i e , u t 9 h i a o 1

l g

h b 3 l r , , 1 & r

9 e e a 0 m , 0 7 W s e , 9 v b

0 u l o 8 0 l l a p 1 l i 2 n r 9 a 0 e & R & a , m S s , a - o 1 2 c 7 , e , i h

a o e o d , , 8 n e a h H g V T r g n a e D 9 g w U u T r r a a -

a 1 e U a & , , e u s v 7 0 g r & & n i , f , a b , x a a 0 9 , , l e o m v d e G g l u S y 0 t 9 i 2 r n m a N c a i e r a R z a , m 1 2 a 9 u n r s e u h o D k o , , 9

M B & a o c R g ,

c a e R T 1 t l i , W g M

o i 1 , n

m , , & w l & h a o , i & O 0 l o a t i r e o l a

c , M & , t l r , e 0 h k o z k F t T l k S , e s b k a 2 l s a o e u , , i a s , a t n m n i e , s n o 5 u 3 1 c i d m e l l t a i h a n g l r l m l t n m 9 9 0 v q r v a r a i o a i i a a e a a e t o a o o 9 e o 9 0 H T S B 1 P C T T M H P G T C N 1 S 2 M C ? g L L L L ? E n i T T T T U U U U N E I I I n I r O M & M M M L L L L a M M M M e S E S I T S S N E E E I I I L

.

.

d l n r o h d d e .

o t t o m r d o e a . , s n i g n - i . g r o g t o t l s

3 t i t h d a t

n n d d e n o n u f p o y c 3 s i , p e f B i t i f r , e o

r t e a o b a a 1 i l t n n . , , c o m f n p r m t c - e a m e r l e a r s s s a o e a w t o e p w s y c o e r r l a f e e c a o t s h i n n c h o i a t s m o f i t b t t . l r t a a e t t s e l t n e e o w o o l

e l f r l e L l o n a , c t t g o h i i f a u p o r

s b

p c e d t t t u l e . i f

1 S n f e o e t a e e h t n j c a a i c p a s m n o o l m o t d . s o t t m i n t s i a i t h b y u i e e e x n i w r t i i s w e t t a c u d o i v s a s t r n i r i u e e k b s , y

o s w r l r f e f o o s c a l m m t a c u o a a n n f t o o r D . i i a e s r i , l s s a e e o t n p e e t h u

o f s o l 2 o r d e d l l s w d s f t r k l e n w t i h a s l l k l s i b e l i t n o e h n , t n t e o l r o f o . n u d , t l o s t a t u u

a l a t t r o g g a f i f n s i o o f f w a o a i g i a n t s d s ” a i b r e h e s i t n r v r o n c t , d c m g t c a n r t i t t s s n i n e e a n d e i p r e b i i e r o c e t t o l c n g d s o n e a n u l u a e o i b b e e s t i k e o s l s a t a r e y t e e u p c u v p o b j c t t f h l u p a e r c m a a c o o f m p i b c e d t w t r c

i t b n m t t

l l u t m e d n e m a e e o i o e o o u m o e i i

u l e u i M s e y o m e l q n o r l r c s d f 5 e d d l p n i s l r i q l d e t m c . s u m t o a e e s i e e o a 1 r i x C t o t c a c n d o a y n o n . r u s l n h 3 u i g g v e e t i a l a w r t s F e ’ h n p f e

g o t n e n d a y i

c e w o d d , e i t s l y c l f e l . ; o o d o y n d o o a h w t l y w r s e o l u u c i d s e e a e t t b e r i d l c c e g o j j o o s t

t u r a i o k n n e e e

h d . t d e u d p n i n r t n f v j a h n t e a f o h v e e t s o i n d m a r k t m o s o o e g q e , t o s o t i b t r r e e i o n ) i o r i l n s n o a e t n

e n n h i r l d n n t u t u e e t y n s . . r o m m a i n a s t t e a e d w c e , o l l

s n i s o r a i t e 1 p a i e a c e b t h m d t A v l w w e n a a s s p e w g e g r m r p l r e i e o 6 r r N c f i i e c t i o p 1 n u e l

s x s s s o n n t t n o b o u c , d e i o e . f u o n m e u 2 i i r r r r e l e e t r f a d r y y t s m n , . s y t m n d o o u t u q v a l l v s a s a h m a l t e e e e e i g b n o e t n d o e r s r a e c c e e t t t n q s e n p n p

d n i t e n p e e n r l o e e i t e e n s t w a n s h e o i i u v d o o t i r n a o r y n i x m s s a n w i h a o s s r o o l i i e c

a c y , e s a , n p p t d l e n d % % y l y t t t o l e b t b i

t h n o u i i l s f s s c y d b a c s s s c u e b a f 6 1 s e a n l i s n s o o w s i a d

h d a d e e - e o f i e a l s r h t a v t ; e a 3 3

n m h C s l r r e d l n f i d s d s m s s t n t e i i d i r h i a s s p l r u ( f t u t t c t o g e r r f v r o l w f f e t i t e o e u a , d E b e i t e g c r c c c h w q o r e e d n t d

c i - i o f s t o o m r e n t b t i e . w n e e e e e n n r i e r , e o h h X e z n e z a j r o b 8 j j k j i a r r s t s 9 t t y o j s a e t c e i e i I n e t n m s n n l e t b b a e o 2 b b b i r t l G % % n b p o o r b b l e e t k k v f a a a n x o o h e u u u u u r f u n 6 8 a o o f o n n D u t u o r n a a h n e m O A f s w K t i s K S c T “ S f O s p L a o I b S r t S u r S v i c 5 a 5 a N s s s E s r r r s r s a a a P r r s a s s s s s e r e e a a s r r r r r e s P y a y y e e e a a a a a r

y

e y y e e e e e a g 0 A 8 2 y 9 y y y y y e 4 0 2 3 0 A 2 y 4 2 9 9 – - – 1 9 8 – 8 – – – – 2 2 4 – – – 0 – 8 5 7 2 7 2 1 1 9 4 4 1 6 1

) ) i s

k k t z * f e e ) c n o t t r d

e a 6 6 4 6 9 6 7 4 5 6 1 1 . n n j e n t 1 1 1 a a o b K a s ( h h i u

N s C C S 2 ( ( -

f n n w s

; t e e o o k h p d

l l c t

e e h h s l l e s k i n l l t t f e o m t a ) m r a k s u j o o o t , , i o a d c s f w e s

a i l l s b

r s s - ( h h t h t t a a r h t t e e d I s u d g g e t c a t p l e - c c

w s w s s d d o o n e o n c

s o e t – n e e e e 5 d i o r ; ; ; o o i

m h

z a j j ; a t r l l l l d l t r p e e f 9 n n

p e d - d n l b b a d e e e e t e p z z o m m o a 9 o n d e o a o a t e m e u u f d d d d i e g n n i i l s 1 l e a h h h p s s m d r m t d y o o o o g n a a

t s s e h o , i c c r i ; ; ; ; ; l 3 e . d e p p n t t m o d c l l l l l d p v a l m m m m n e

i b u r e a a e s e e e e e p 8 o o r a x p t d m m n r h r M e e e e o h r d d d 9 d d a x i e i s i w k w k t t c o e f e e e e e h p M e e s s h h o o o 9 o o a e n

, h t l z z z 9 n z o c o o s a a c c m t ) p 1 s t r t t l g

t t e

n n n n m n m m m m d l f w e t t , 4 , a a e - s s n c e a a a a t t a o s ( r o r d d I i i n n e n n n n n i a i d a n p p p l p p p w e e l e e - d

r c e a a a a a a

p p o a e s i t p p d w d e t s o w w m m m m m m i 6 u u t f a d m m m m m m e x x e t i i i i s p m e o e o e - h a t t o o o r o e e l t l t u u u u u u e h h h h u n r r r o c s c d l t l t t o i M p a H s C C g 3 a C g 6 a f H e m C a a H a a t H W H H d C

) d d t e

i

o n b s n n u g g a h o y a r u t t o

p n n

w t l f i d e i i

t e , d u s o a d t l i o t d d e s d r d n d u m a s s n e e b

u u u n u i n a o o n e e l l o

k g n p c o a c r s , e a s s o c c m i t p a l e l e d t f a t n r k f s n n k i t o d a c e o m i i i - p p s r e i u s i c f t i e h t f f h s - e , , l a u i c c y r a n u c o n

e - s s n j e r y o n e a a T a w a d

u s o m s d f a ( n n / c e d o n t e t e ,

w o r r e , t s , g n o o r t t o s t o a e p x s e o s e i i n - y

n d n o t t t e n i e h f f i p o r n p b g i t e k r r M o i c c y h

a s o r o b o o o s n l r c u h a a

b o i r t - i r s r f o t t o t s t u i t p u m s t m s d e

f s o n s p t

r 4 8 u n

l d s e u

o f u o ) g s e n e h t o d s o 2 4 o m o e o t e u o t i

o e s n g i k s r o e e i D t

d ) n i f a f o v f t l l d h m a i e l t t n p c o e y , t o v o c o l t i w o o a f o a i g f n x

f o a g , c o m d A h h o c h u a e y y e t n g i f g e e i n i e f n i

t r r e y c m n i l s f t e e e n n f e l c l e e n e e n f i i a f e a a a i i a h b b b t t n b e r w i e e k k e t t f p e s e p d

o o o m o a l f c m i n p h r a a a a n t u r r r y r o i n o e e i a m S C i r P P P D P U a R s w R ( ( O b B d b B f b 52WHITEN, HORNER, LITCHFIELD, AND MARSHALL-PESCINI - - a e l s

u e & t

m , a e k d ,

o t ) r o T r a I t 3 b s 5 4 0 M e m 9 9 e 0 I c 0 a ( 9 9 h 2

n 0 t S 1 1 , n e 0 f , , , r n 0 o 2 o n i o o e e

0 t l l t , f e l l . i n 0 t a e l i t e 1 e h o 2 i a s s i h R 0 , t t a a . 0 W m a l e i W

2

m m a n

, , d , & a o o t )

e l s n i e T T c R a p

u k l n g k S x s i & & ( a k e n

n t i r d l l i s l r l l g n o o a e j a u a n a t

i r B B C C G S C c o n F e ?

r . g ) e L L L n f s i T T T e U U U g I I n r I ? n

r l M M M L L a a M M M a e I I E S S E I E i m c L a o i s s

, g s

n s n

o i a t o d n

a

a b n w t o

s c l t b a s u i

s

i n l a l i a e s b f a g o u n p h i i k ( n o w e t e d

c a T o i m s s c h r i e i o . v e t n a r % s i

w s d g c e e p 0 f d f f a e a ’ n a h 5

o n i o e s t

o f i s r r s p d t ; d e e o l e o % d p - d s c e s f i u n I e 0 n s e l t n n - n s i f i k e 5 n o s e

e o o l i r p

f i o r a t d w . o a

i o . - i s e n o a t t r d c s t e o v i u h o a m i r e l t n

t c e e d i e

i a % c d ‘ o . h r n r h t m p

e t i 0 t n i e e e e o t n i m

o o y s 4 f o i c u a t a

d o c w

w u r t n t c 6 q

C e t d t

d i , s m e o s e s s s e t ’ f o e s e s s o u n r n i t n s d c

n r g n r a r q , c n o i e e o f n l a i e j s s e e o a g e t h c f . j s e e i a r b m t r h o c e t s s t b g t t s p e o

) s

a w d r y , i o u n u a

d e f y 4 i d r f o p s k h l r t n d t o c A t e r o e h o e e r a

o e n u t t n a c s i t e e g u . y o n t t n w b u e e r a i t a d p d l o a a a n n t , u l c g e . o i t i e c o

i e a h t d r q r ) l o t l t n n r m c

a e

i e e e e 3 o C e p m o t o . n n s s i d h i h d o n f i r e f i r t m t 9 u o m f f . o i T v s o s e i d m o y –

s , o s

d l i . o a e

h C 1 m d t n s e a d g e n . g c ( t g e t n w o o m e u s w n o 9 r n t c n w i l s n e i r

i t n r i e ) t i e e e e e s X y d e j r d a e n o e y c u d u e i I i r e p f i l o e k b m n l p r t t q o q f s a e o a h h e i u o b a c l o f D e e t e r s

a C a D m S a r T l s v S t c R e i l s T N ’

b e , s l E u s r 7 s s c r p a r P r i 9

a t e a a 9 s n e r P e y e e e 1

a y s y ( y e g r

A

+ e 5 l b a 2 . 0 4 l

A h e 4 t a e 5 2 3

y v – – – – C n a 2 6 3 9 8 i

h d

d n s e t a s

r

t f n a o c o e l e

e l 3 1 4 5 6 . j p m e 1 1 1 o b i p s r u a a N e

S p m m x o r e E T

t . o o e t t ) e

e E h y y s

t

t t f f , i i & h f s n n o o n n k i c

i i T

e t s u u ( i d d e

t t a a m m ; d r r r t o o

p l

w d u o o o t h h 0 0 y e t t t r t n ; d p p 1 1 p d o s l r e e

a t o p p e

r r e o e

r m

l h o o m m c o o d e d e t e d m i f f a ; ; t ; ; e o l e b b f l l t l l l d n r n d d e e a e e e w 8 8 d o a m a

a l M e e o d d d 9 d 9 d l t e r r n l o n M o o o 9 o 9 o a l r r u e f i b a a e e 1 1

g r h o

m m m m m f f t t , , n m n s

e e i t , g n n r n n n n n a u ( l d d c r e e a a a a a o n

e

t t i e H s o , i i y y t j

d m m m m m e ) ) s ) p p h h t b o i u u u u u L a l o o u A B

c H c H c ( ( s m H W H W H S a i (

d r

g ) ) n o t t i n

i i F i

n n u u n . n

f t r r r o p w o f f i m x g a

t f u o e l l i u n e e s s d t o l i l d a a s s s t t e

n t i i n o e e l c c o e s s k o c c a s s c i s a e e m e i i p h a

i j j t n n r k g f f c o d e w c i i b b p s e e r t t n n t t f e e f f f o i l a o o u a r r e e o

s e e s - d e a a T s u d n n s m d d s r u ( ( / n d e i ,

q e o o s e a g t ) o

e e x e x x p l n n h , n s s 7 s e e i t i o o o p u

t g r r M o i n n 8

a b b b o

e l n r h h n o o 9 s f o , t t e i i i v p

t m d d d

i ) 1

t t n

r h p t ( o o o o g g ) c c h o s L o a t t i e o o o a a n n e u i D t

n i i ) n f f f U r h

t c r p e y y v v w e

i g g g T e n r r , o o M A t h c e a a n n n g l g i i i i i r r E m m — a n f d n t t ( c n n n i i i i e e f i e e r l e e r e r t t k e f b b n l

u f r r p p p o c a n o u y y o e i i e A A O f p O b O b R ( ( N t l