Comparative cognition is an emerging interdisciplinary field with contributions from comparative psychology, cognitive/experimental and developmental psychology, animal learning, and ethology, and is poised to move toward greater understanding of animal and human information-processing, reasoning, memory, and the phylogenetic emergence of mind. This chapter highlights some current issues and discusses four areas within comparative cognition that are yielding new approaches and hypotheses for studying basic conceptual capacities in nonhuman species. These include studies of imitation, tool use, mirror self-recognition, and the potential for attribution of mental states by nonhuman animals. Though a very old question in psychology, the study of imitation continues to provide new avenues for examining the complex relationships among and between the levels of imitative behaviors exhibited by many species. Similarly, recent work in animal tool use, mirror self-recognition (with all its contentious issues), and recent attempts to empirically study the potential for attributional capacities in nonhumans, all continue to provide fresh insights and novel paradigms for addressing the defining characteristics of these complex phenomena.

KEY WORDS: comparative psychology, imitation, tool use, self-recognition, theory of mind, animal behavior

INTRODUCTION TO CURRENT TOPICS IN ANIMAL COGNITION

The field of comparative cognition continues to emerge from several intersecting and interdisciplinary fields, including animal learning and behavior, comparative psychology, ethology, primatology, and developmental psychology. These fields have provided both methodological and empirical paradigms, and, along with philosophy, have helped to frame the theoretical issues for comparative cognition. These overlapping areas continue to provoke new directions for research and offer innovative and creative approaches to addressing fundamental questions related to basic psychological processes and their underlying mechanisms. Despite rapidly changing dramatic technological options currently available for assessing complex neural processing through a variety of imaging techniques, such as functional magnetic resonance imaging (fMRI) and positron emission tomography (PET), some perennial questions in psychology remain doggedly difficult to address empirically. From the ethological, biological, and comparative psychological sciences, we know that many species, including humans, have developed complex and dynamic social strategies, likely derived from selective pressures for species survival. How these social pressures may have influenced the cognitive strategies and abilities of species and individual animals is in itself a complex and often controversial component of the field of comparative cognition, and research has barely begun to frame the critical questions with which to address these hypotheses (Boyd & Richerson 1988).

The cognitive abilities necessary for a species to integrate and exploit a fluid social environment remains a broad question with many caveats. Some specified abilities, such as imitation, have a long history of study, whereas others such as "theory of mind" (Premack & Woodruff 1978b, Leslie 1987) are relatively recent. Tangential topics, including self-recognition and tool use in animals, provide additional directions for exploring nonhuman cognitive abilities and capacities, particularly with nonhuman primates, and offer tantalizing glimpses toward precursor capabilities that likely contributed to the evolution of cognition in early hominids.

Although worthy of study in their own right, current issues in animal cognition are invariably viewed in comparison with human intelligence and the array of processes that subserve human memory, attention, and perception, among others, including more ephemeral phenomena like intentionality and consciousness. Nonetheless, there is little question that recent findings have allowed for the emergence of richer and more definitive hypotheses, with broader implications for the ontogeny of cognitive abilities across all phyla. Dealing with these hypotheses will lead to further cross-fertilization between fields, innovation in the questions addressed, and additional methodological advances.

IMITATION

The study of imitation has long been controversial, with serious and continuing debate beginning at the turn of the century over several contentious issues. Three early concepts of imitation were represented in the writings of Romanes (1884), C. Lloyd Morgan (1900), and Thorndike (1898, 1911). Romanes, as an ardent Darwinian, viewed imitation as a capacity that one would expect to find across the psychological continuum, and, indeed, he wrote extensively about imitative processes that he believed could be observed even in bees (Romanes 1884). He defined imitation as requiring the "intelligent perception of the desirability of the modification on the part of certain individuals, who modify their actions accordingly," and he believed that these imitative acts required intentionality as well as high intelligence, representing precursors (to Romanes) of human imitative abilities (Galef 1988).

Morgan (1900), on the other hand, defined imitation as (a) instinctive, (b) purposeful and intentional, which he called reflective, with reflective behaviors having a defined goal based on observation of another individual engaged in the same behavior, or, finally, (c) intelligent imitation, which introduced a conscious component into subsequent occurrences of the behavior. Thus, Morgan proposed that imitationlike behavior observed in other species could be the result of innate, instinctively driven behaviors that could give the appearance of true imitation, but were based on other processes. Reflective imitation, however, was similar in some features to imitative behaviors observed in humans, and intelligent imitation allowed for later demonstrations of the behaviors, based on conscious, intentional reproduction.

Thorndike's (1898) definitions of imitation were ultimately based on empirical efforts to demonstrate the phenomenon with a number of nonhuman species, and failures by chickens, cats, dogs, and monkeys prompted him to cast doubt on the anecdotal instances put forth by his colleagues. He proposed that most instances cited as imitation in animals were likely based more on instinctive acts that were qualitatively different from true imitative learning as observed in humans (Galef 1988).

Nearly 100 years later, Galef(1988) provided a thorough and extensive roster of imitative learning term, which could be applied to behavioral phenomena. These included, in addition to imitation, intelligent imitation, reflective imitation, instinctive imitation, pseudo-imitation, true imitation, allelomimetic behavior, mimesis, protoculture, tradition, contagious behavior, social facilitation, local enhancement, matched dependent behavior, stimulus enhancement, vicarious conditioning, observational conditioning, copying, modeling, social learning, social transmission, and observational learning (Galef 1988). Although the definitions of Galef and others for each of these terms continue to be debated, most agree that imitation is on some level affected by social influences and some aspect of social learning. How then does one separate true imitation from other forms of social learning? This again becomes a debate over definitions and methodologies.

Of all aspects of social learning, three are most commonly associated with imitation: social facilitation, local enhancement, and stimulus enhancement. Local and stimulus enhancements were first introduced by Spence (1937) and later Thorpe (1956) and refer to instances in which an animal's attention is drawn to a particular stimulus or location during (or after) observation of another animal's interest in the same stimulus. Such enhanced attention increases the likelihood that the animal will perform behaviors already present in its repertoire that may, in turn, resemble the behaviors of the observed animal, either in the same physical location or during interaction with the same or a similar object(s). These behaviors, however, would not necessarily be considered imitative of the observed animal's behavior. Social facilitation was first introduced by Clayton (1978), who defined it as the result of increasing the likelihood of an animal performing a behavior already in its repertoire while in the presence of another animal already performing the behavior.

Attempts to define tree imitation were made early in the history of comparative psychology by Thorndike (1898): "True imitation is the ability to learn an act solely from seeing the act performed by another individual." In a functional sense, his definition did not allow differentiation of imitation from other types of social learning. More recently, Visalberghi & Fragaszy (1990) elaborated further on Thorndike's original definition and provided additional clarification, as have Nagell et al (1993). The former classify a behavior as imitative if it is sufficiently similar to the behavior of the model, with sufficiency defined a priori, if observation of the model is necessary for its production, and, finally, if the behavior is novel for the observer, with novelty required in a dimension within which it is usually absent (Visalberghi & Fragaszy 1990). Though reference is often made to their definition, it is nevertheless difficult to explicitly define "sufficiently similar," although novel can be more readily defined operationally. Nagell et al (1993) argued that, in true imitation, the observer must not only perform the modeled behavior, but must do so toward the same goal. This distinction helps separate imitation from emulation, in which the observer learns something about the task (Nagell et al 1993).

From her work on imitation in rats, Heyes (1993) originally suggested that the difference between imitation and other forms of social learning was represented by what the animal learned as a result of its observations. Nonimitative social learning involves processes through which an organism learns about a given stimulus or event, whereas imitation involves processes whereby an animal learns about specific responses or behaviors. Imitation therefore implies, according to Heyes, that the animal understands only through observation that a particular behavior or action has a specific relationship with an outcome or is goal-directed. If an individual replicates a behavior without this understanding, both Heyes (1993) and Galef(1988) refer to such a behavioral response as a "copy." There are sufficient implications for these differences in definition, which continue to fuel the debate and encourage additional research on imitation in animals and humans. For example, if an animal "mirrors" another's behavior, what cognitive abilities are necessary? In social facilitation, and local and stimulus enhancement, the mirrored behavior may actually be innate and merely triggered by the presence of a stimulus or conspecific. When imitating a novel behavior, the animal must at least process novel visual input cross-modally into novel kinesthetic output (Zentall 1996). This cross-modal process may be a cognitive process unto itself (Heyes 1993), or it may require the ability to comprehend goal-directed behavior, to form representations, or to role-play and thus be aware of other's perspectives.

There have been several recent experiments designed to explore true imitation in nonhuman species. Zentall et al (1996) attempted to control for several nonimitative aspects of social learning by examining the ability of pigeons to imitate a conspecific's behavior of depressing a treadle in two possible ways. Demonstrator pigeons were trained to either peck or step on a treadle to obtain a food reward. Observer pigeons were then exposed to one type of demonstrator (peck or step). After 15 minutes in an adjacent operant chamber, the demonstrator was removed and the observer placed in the chamber with the available treadle. The behavior of the observer pigeon was recorded over the next 30 minutes. The authors argued that their design controlled for social facilitation as well as local enhancement, because any potential effects for local enhancement would be expected to impact both types of behavior (step or peck) equally. Social facilitation was unlikely because the demonstrator pigeon was removed during the test session and thus was not present to influence the experimental subject during testing. Stimulus enhancement, on the other hand, was more difficult to control. Earlier, Dawson & Foss (1965) proposed using a design in which a task could be completed more than one way. This approach is commonly referred to as the two-action method (Whiten & Ham 1992). Thus, in the Zentall et al study (1996), two different responses, pecking and stepping, could be used to obtain food. Assuming an equivalent likelihood of occurrence, if the pigeon was imitating the demonstrator, then the manner in which the pigeon depressed the treadle could reflect imitation of the demonstrator bird, just as a similar two-action task had been used to demonstrate imitation in rats previously (Heyes & Dawson 1990).

In previous studies, however, local enhancement had been controlled by the use of a duplicate-cage procedure (e.g. Warden & Jackson 1935; Zentall & Levine 1972). This procedure entailed the use of two identical test chambers, placed side by side. If local enhancement was an influence, then it could be argued that the observer's attention should be drawn to the demonstrator's treadle and away from the treadle in the observer's chamber. In the Zentall et al (1996) study, the duplicate-cage procedure was modified. While the observer subject was in an adjacent chamber, it was subsequently tested in the demonstrator's chamber. One could argue that this change in procedure did not control for local enhancement, since the treadle used in the test was, in fact, the identical treadle used by the demonstrator bird. However, any effect of local enhancement should have impacted both experimental groups similarly, and thus the authors concluded that the birds could copy or imitate the topography of the demonstrator's response (Zentall et al 1996).

It should be no surprise that, among studies addressing imitation, nonhuman primate species would be selected for study as well (e.g. for an extensive discussion, see Visalberghi & Fragaszy 1990). Recently, Bugnyar and Huber (1997) used a similar two-action design during which marmosets (Callithrix jacchus), a New World monkey species, were required to either push or pull a pendulum door to get food from inside a box. The investigators exposed an observer monkey to one of two possible behaviors for one daily session over three days. Both behaviors had been previously acquired by the demonstrator animals to obtain food. Immediately after the day-3 demonstration, the observer animal was tested once per day for five days, using the same apparatus. Thus, Bugnyar and Huber's approach was similar to the Zentall et al (1996) paradigm, such that the demonstrator was removed from the original apparatus and the observer monkey was then tested in the same chamber. In this case, however, none of the observer monkeys demonstrated use of the particular strategy exhibited by the demonstrator animals. A control group, which was not a feature of the original Zentall et al (1996) study, included 10 animals not exposed to demonstrators. This group was tested to determine whether one of the two actions (pushing or pulling) was used more often and thus showed a higher baseline occurrence. It is interesting that none of the control animals utilized a "pull" strategy to obtain the food reward. Although some observer animals initially used this demonstrated strategy, it did not persist over the five days of testing, with the animals' responses converging towards the preferred mode seen in the control animals. The authors concluded that, although the observer animals did not persist at using the observed strategies, their results indicated that marmosets were capable of acquiring simple motor skills that were in some respects facilitated by observing conspecifics (Bugnay & Huber 1997).

In addition to studies of imitation in captivity, some innovative field studies have also been reported recently. Russon & Galdikas (1993, 1995) recounted observations of imitation in free-ranging rehabilitant orangutans (Pongo pygmaeus) at Tanjung Puting, Kalimantan, Borneo, where Galdikas has conducted long-term studies of orangutan behavior for over 25 years (e.g. Galdikas 1978, 1982, 1985, 1988, Galdikas & Vasey 1992). A central distinction in this study was the modified definition of imitation due to field limitations. Russon & Galdikas's operational criteria were that imitative actions were rare, their performance constituted an abrupt change in behavior immediately after the actions were demonstrated, or the imitator had not performed them recently (in the previous 10 minutes), despite the opportunity (Russon & Galdikas 1995). The primary focus of their study was to assess the selectivity of the models and the behavioral responses exhibited by the orangutans.

In general, they noted a preference for models with whom the observer had an established relationship. For human-reared orangutans, this preference was initially for human models and gradually transferred to other orangutans once these relationships had developed. Choice of the model also seemed to follow a dominance structure, with subordinates imitating more dominant animals. They proposed that this type of model specificity might be a possible explanation for the apparent discrepancy in imitative abilities between enculturated and nonenculturated chimpanzees, noted by Tomasello et al (1993). He and his colleagues studied the abilities of mother-reared chimpanzees, enculturated chimpanzees (those raised in close contact with humans, including rich immersion and experience with human cultural artifacts), and children. Three types of imitation tasks were devised, including simple, complex, and delayed imitative learning. For the simple and complex imitative tasks, mother-reared chimpanzees performed more poorly than both enculturated chimpanzees and children whose imitative abilities were comparable. However, the most striking difference was observed in the delayed imitation paradigm, where a 48-hour delay was imposed between the demonstration and an opportunity for a response. Under these conditions, enculturated chimpanzees outperformed mother-reared chimpanzees as well as human children. The enculturated chimpanzees demonstrated imitative behavior, including both the specific means and outcome threefold to fourfold as often as the children they tested (Tomasello et al 1993). Given some of the debate as to whether true imitation has been demonstrated in nonhumans (e.g. Heyes 1993, 1995; Tomasello 1990; Byrne & Tomasello 1995), these findings are clearly of interest.

Under nonsocial conditions, Whiten et al (1996) addressed imitation in a comparative task by using an artificial fruit-processing task. Both chimpanzees and human subjects were tested to determine similarities and differences in their methods for solving the novel task. The authors reasoned that if results from known imitators (humans) were compared with those from subjects whose imitative capacity had not been precisely characterized (chimpanzees), additional insights into the differences between the two species' imitative potential might be elucidated. Previous designs (Tomasello et al 1993, Custance et al 1995) had used arbitrary behaviors to test an animal's ability to imitate. Custance et al (1995) had previously proposed using a more naturalistic behavior such as food processing, which might increase the likelihood of eliciting imitation. The Whiten et al (1996) study was designed to address the latter. Their apparatus consisted of a two-action model that required either poke or twist responses on bolts affixed to a test box, in addition to either pull or turn responses on the handle of the box. Children of ages 2-4 years readily showed imitation of the two actions when tested. The results from young chimpanzees (mean age 4.5 years) were not as impressive, but were nevertheless imitative. Thus, even though "a twist could be done by holding the bolt in a precision grip using a finger and the thumb, a power grip wrapping fingers round it, or a chuck grip using several fingertips and thumb, or both bolts might be twisted simultaneously using two hands," (Whiten et al 1996), the critical actions performed on the bolts by the chimpanzees had the same result, and the authors concluded that they must have arisen through imitation (Whiten et al 1996).

Whiten et al (1996) also re-introduced another distinction in imitation terminology, termed emulation, originally proposed by Wood (1989). Emulation refers to instances in which an animal achieves the same end goal as the demonstrator, but does not follow the exact motor patterns that it has observed. In this sense, the observer may understand the end goal, yet does not replicate the demonstrator's actions as the means to attain the goal. The similarity of the two alternatives (twist/pull) for the barrel on Whiten's apparatus may have confounded their results somewhat, since execution of one alternatives could easily facilitate occurrence of the other. However, the authors suggest that, at more primitive levels of cultural evolution, such differences might be easily corrupted by chance slippages, trial-and-error learning, or both within potential nonexclusive responses (as in the case of the barrel manipulation of Whiten's apparatus) and that such fidelity of response(s) will occur only when a species becomes as thoroughly cultural and conventional in their imitative abilities as are humans (Whiten et al 1996).

TOOL USE

Tools offer a natural advantage in imitation studies because they often facilitate the presentation of novel tasks (Whiten et al 1996); conversely, the study of imitation can also lend itself to the study of tool use in nonhuman animals. If an animal is capable of imitating a behavior by visual input only and, by observing the behavior, is able to understand the consequences of that behavior, then it should be able functionally to replicate the behavior in a goal-oriented manner. Because, as Passingham (1982) noted, monkeys are not habitual tool users and because tool use is still fairly rare among nonprimate mammals, Whiten et al (1996) suggested that the use of tools to test for imitative ability may be inherently biased against the animal. However, such tasks have been often used to examine imitation in nonhuman species, particularly with chimpanzees (Nagell et al 1993; Custance & Whiten et al 1995; Whiten et al 1996), and this approach seems quite appropriate given the wealth of observations and cultural diversity of tool use in wild-chimpanzee populations (e.g. Goodall 1964, 1970, 1986; Kortlandt 1986; Matsuzawa 1994; McGrew 1974, 1992). Tool use has nevertheless been observed in a wide range of animals (Beck 1980, Berthelet & Chavaillon 1993) and is particularly diverse and innovative among wild chimps (Boesch & Boesch 1984a,b, 1990; Inoue-Nakamura & Matsuzawa 1997; Nishida & Haraiwa 1982; Nishida 1987; Sakura & Matsuzawa 1991; Sugiyama 1993, 1994; Sugiyama et al 1988; Yamakoshi & Sugiyama 1995). In fact, it has been recently argued that tool use in chimpanzees appears to have a basis in cultural transmission by imitation, among other potential types of social learning, although this view also has its critics (e.g. see Heyes 1993).

Among other studies of tool use in nonhuman primates, Visalberghi and her colleagues (e.g. Fragaszy & Visalberghi 1989; Visalberghi & Limongelli 1994) have conducted numerous investigations of the abilities of capuchin monkeys to use tools and have also contributed evidence (with chimpanzees) demonstrating understanding of the causal effects of tool use. Their innovative paradigm was similar to the approach of Visalberghi & Trinca (1989) in which a horizontal tube was presented to the subject, and the monkey was then provided a tool (a stick) that could be inserted into the tube to push out small pieces of candy. The capuchins were also able to modify the tool (a stick with a small crosspiece inserted at the end, which had to be removed for the tool to be usable), to obtain the food reward.

In a more recent study, Visalberghi & Limongelli (1994) used the same tool task and then modified the apparatus for a second experiment. In experiment 2, a wide hole was drilled in the center of the clear tube, and a small trap was attached, creating a "t-tube". If the monkeys had understood the causal relationship between using the stick and the new trap, they should have modified their behavior so as not to push the candy into the inaccessible trap. Four female capuchins served as subjects, all of which had successfully completed the plain tube task. With the modified t-tube, only one monkey was able to complete the task without losing the food reward. The other three demonstrated increased attention to the movement of the food reward during pushing, as well as "cautious behavior" (Visalberghi & Limongelli 1994), as the candy edged closer to the location of the trap. Four additional control conditions were completed with the successful animal, in an attempt to determine whether she was solving the task based on a perceptually based learned rule or more cognitively through some form of comprehension of the cause-effect relation between the use of the stick and the trap. In one condition, the tube was inverted so that the trap was now on top, rendering it inconsequential. In a second condition, the monkey was presented with the original clear tube without a trap. During conditions 3 and 4, the subject's performances with both an opaque tube with a trap and one without a trap were compared. Under all four control conditions, the capuchin avoided pushing the reward toward the trap even though the trap either posed no danger in terms of losing the reward or was simply no longer present. These results are consistent with a rule-based solution (e.g. "Insert the stick into the end farthest from the reward") (Visalberghi & Limongelli 1994), and such over-generalization of tool use behaviors that were successfully used in the initial experiment is also consistent with previous findings and may have adaptive importance (Visalberghi 1990, Visalberghi & Limongelli 1994).

The results also suggest that capuchins, although impressive tool users, lacked an understanding of the causal effects of the tools they used; they were able to learn how to use a tool but failed to understand how the tool actually functioned (Visalberghi & Limongelli 1994). Thus Visalberghi & Limongelli (1994) suggest caution when interpreting successful completion of a task by different species, since similar results may be obtained by very different underlying processes.

In a follow-up study, Limongelli et al (1995) conducted a similar experiment with chimpanzees (Pan troglodytes). The chimpanzees were first tested with a clear tube as in the previous monkey study and then presented a t-tube containing a trap. All five animals tested were able to solve the initial plain tube task, although the youngest required one instance of modeling by the experimenter. When the t-tube was introduced and the animals allowed a maximum of 150 trials to demonstrate their understanding of the contingency between the use of the stick and the trap, two of the chimps readily solved the task at a performance level significantly above chance. These two chimpanzees continued to a third test, during which the hole to the trap (formerly centrally located in the tube) was relocated closer to one end of the tube. If the chimps were using a rule-based strategy similarly to the capuchins, they should have failed this version of the task, because inserting the stick into the end of the tube farthest from the reward would result in the candy being pushed into the trap. Both chimpanzees, however, were able to correctly solve the task despite the modifications. It is important to note that this change in the design was also repeated with the capuchin from the previous study, and she was unable to solve the task correctly.

Additional analyses were completed to determine whether the successful chimpanzees were using some type of representational or anticipatory strategy. To address these issues, the number of single stick insertions was compared with the number of multiple stick insertions, with the assumption that a greater number of single successful insertions is reflective of a representational strategy, because a correct initial insertion of the stick would suggest that the chimpanzee was covertly determining the correct side from which to insert the stick by using only the visual cues of the location of the candy and its relationship to the trap. One animal always performed single insertions during the initial t-tube task and 27 of 30 times in test 2, suggesting use of a representational strategy (Limongelli et al 1995). Regardless of the strategy used, success indicated some level of understanding of the causal nature of the tool/trap relationship, unlike the results with capuchins, which showed no comprehension of the causality of their actions. However, the performance of the chimpanzees, although sufficient to complete the task, was in contrast to findings with children during which more rapid comprehension of causality was demonstrated (Limongelli et al 1995).

MIRROR SELF-RECOGNITION IN ANIMALS

The flexibility and understanding of causality of tool use behavior in chimpanzees reflects the capacity for complex problem-solving abilities, with questions naturally arising as to the potential adaptive significance of such learning potential in the great apes. Jolly (1966) first proposed the possibility, with similar ideas elaborated on in Humphrey's (1976) classic essay, that pressures associated with complex social living have selected for a set of characteristics that are clearly operative in humans and may also be available in the social cognition domain of all or some of the extant great ape species (Byrne & Whiten 1991, Humphrey 1976, Jolly 1966). Among these characteristics are the ability to appreciate the perspective of another individual, sometimes referred to as theory of mind (Premack & Woodruff 1978b, Byrne & Whiten 1991) or attribution of mental states, as well as the ability to recognize oneself in a mirror, which has been proposed also to reflect a state of self-awareness (e.g. Gallup 1970, 1975, 1982, 1991; Mitchell 1993).

Attributional capacities, which have been suggested as potentially available to the chimpanzee, have far-reaching implications for what we have long considered to be an exclusive domain of human uniqueness. But before one can take the perspective of another, one must understand one's own self-image, as reflected in a mirror for example. Mirror self-recognition or the "mark test" (Gallup 1970), as it has come to be called, has historically been touted as the benchmark technique for assessing self-recognition and, potentially, self-awareness. (Gallup 1982, 1991; Povinelli 1987; Parker et al 1994). Gallup (1970) first demonstrated that chimpanzees were able to recognize themselves in a mirror after being marked with vibrant red dye while asleep, and afterwards, when given access to mirrors, the animals engaged in a variety of self-directed behaviors, including using the mirror to investigate visually inaccessible areas. This differs from other mirror-mediated behaviors such as object discrimination, spatial locating, and mirror-guided reaching, because some animals have been successful with these types of mirror-mediated tasks but have failed to exhibit mirror self-recognition (e.g. Pepperberg et al 1995, with a parrot; Povinelli 1989, testing an elephant). However, a wide range of primate species in particular have been tested with the mark test, and the results include successful performance by at least one member of all four great ape species (chimpanzees, bonobos, orangutans, and gorillas) and failure by the numerous monkey species tested, including rhesus, stumptail, longtail, pigtail, Japanese and Tonkean macaques, mandrills, olive and Hamadryas baboons, and tufted and white-faced capuchins but see discussion below of recent work by Hauser et al (1995). Although the debate continues as to specifically what success on the mark test really measures or reflects in terms of cognitive abilities, the data still overwhelmingly support a division between the great apes and humans versus other species for performance on this task. On one hand, only the great apes and humans have demonstrated the ability to use mirrored information, which is suggestive of self-recognition, and, in contrast, no other species, including virtually all monkeys tested under the original criteria proposed by Gallup (1970), have shown comparable behavioral responses (Anderson & Gallup 1997, Povinelli et al 1993). This includes a lack of definitive self-directed behaviors compared with control baseline measures of spontaneous behavior with or without mirrors.

The most recent debate in the mirror recognition literature revolves around the findings of Hauser et al (1995), who reported self-directed behaviors suggestive of mirror self-recognition in cotton-top tamarins. These authors compared the behavioral responses of this small New World species to their mirror images after the authors had dyed the tufts of hair on top of the animals ' heads (which are normally white) either bright pink, blue, purple, green, or some combination of colors. Six tamarins were given 3-4 weeks of mirror exposure before the dye test, two were given short mirror exposure of 20 minutes, and one subject received no mirror exposure before testing. After marking occurred, only individuals with long exposure (3-4 weeks) were observed to touch their dyed hair while looking in the mirror. Individuals with minimal or no mirror exposure were never observed to touch their heads while looking in the mirror, nor were group members ever observed touching their heads after exposure to a single individual whose hair had been dyed. Hauser et al argued that these findings suggested that cotton-top tamarins with extensive mirror exposure show evidence of self-recognition as measured by the mark test. They also proposed that previous failures to show such self-recognition in non-ape species might be a function of a lack of mark salience. In addition, for many primate species, direct eye contact and staring are highly aggressive behaviors, and thus looking into a mirror may cause aversion. Prior methods of marking monkeys also may not have been salient enough to overcome this predisposition to avoid looking in the mirror (Hauser et al 1995).

Anderson & Gallup (1997) have raised several issues with the Hauser et al study. They have argued that the crucial control period after marking the animals before the introduction of the mirror was not adequately described or quantified in the study with the cotton-tops. This period of observation, which has traditionally been seen as a necessary procedure in the mark test paradigm (Gallup 1970), provides critical observations of baseline behaviors directed toward the mark, which are later compared with the animals' behavior in the presence of the mirror. In addition, necessary qualitative responses previously reported in mirror studies with chimpanzee subjects were similarly not reported in the Hauser et al report. For example, in numerous other studies, chimpanzees have been observed sniffing their fingers or visually inspecting their fingers after touching the marked areas on their bodies that were visually inaccessible, once they have again been given access to mirrors. Hauser et al (1995) reported that the monkeys touched only their own heads, with no subsequent investigation by any of the other tamarin subjects. Anderson and Gallup (1997) also took issue with Hauser's justification of procedural changes on the basis of mark salience, citing studies by Anderson (1984) and Benhar et al (1975), that have previously investigated the significance of mark salience. For example, Anderson (1984) demonstrated that stumptail macaques responded to visible marks on various parts of their bodies but showed no responses to inaccessible marks on their heads. Therefore, Anderson & Gallup (1997) suggested that the lack of response by the tamarins in the Hauser et al (1995) study to marks on the heads of their subjects' cage mates also provided evidence for the lack of mark salience in this study. Finally, Anderson & Gallup (1997) questioned the reliability and validity of the behaviors termed "mirror-guided," as Hauser et al (1995) reported them. Although interobserver reliability was established for the observations, the tamarins' positions within the cage (relative to the mirror) when these observations were made were not reported. Without such information, questions remain as to whether the monkeys were using the mirror to monitor any responses toward the marks or serendipitously looking toward the mirror while touching their heads.

In response to these criticisms, however, Hauser & Kralik (1997) replied that they believed the controlled version of the mirror dye test that they implemented was sufficient to demonstrate mirror-mediated self-directed behavior and that, in contrast to the specific criticism of Anderson & Gallup (1997) regarding mark salience, the procedural changes they used with the tamarins (marking the prominent white hair tuft, instead of brow ridges and the top of the opposite ear, as in the Gallup (1970) procedure) were instead a direct test of salience and a primary component of their experimental design (Hauser et al 1995). They also proposed that prolonged staring, as demonstrated in their experiment, is related to self-recognition, and that it was unlikely that the observed staring was related to something else in the environment (as suggested by Anderson and Gallup). Finally, Hauser & Kralik (1997) emphasized that it is now appropriate to pursue the question of specifically what cognitive functions might actually be tested with the mirror mark test and that it remains unclear which of these abilities may be the most significant in distinguishing between species. It is possible, as suggested by Heyes (1994) that the mirror test is simply invalid or inappropriate as a metric for self-recognition altogether (Hauser & Kralik 1997). Undoubtedly, additional experiments from several laboratories testing a variety of species will be forthcoming to address these contentious but intriguing issues.

Both methodological issues and the requisite cognitive abilities that may impact on a particular species' capacity for passing the mark test will likely continue to fuel the current debate. Of the great ape species that have demonstrated evidence for mirror self-recognition, chimpanzees' abilities have been investigated in the greatest detail, with numerous studies completed over the past three decades. Until recently, most of the archival literature was based on testing of a fairly limited number of animals and provided evidence of self-recognition at a fixed point of maturity, with the youngest animals described by Gallup (1970) as juveniles.

The first developmental study that focused on the emergence of self-recognition was reported recently by Lin et al (1992), concluding that chimpanzees can, under some conditions, demonstrate self-recognition abilities by about age 2.5 years, at least as measured by the mark test. Using a larger cross-sectional population of captive animals that were housed and raised under different conditions than the population tested by Lin and associates (1992), Povinelli et al (1993) also addressed the possible developmental emergence of self-recognition in chimpanzees and came to a significantly different conclusion about both the age at which chimps were able to definitely pass the mark test and the "loss" of such capacities in some adult animals. Whereas children typically demonstrate mirror self-recognition between 18 and 24 months of age, when tested by methods which are analogous to the mark test (Amsterdam 1972, Bertenthal & Fischer 1978, Lewis & Brook-Gunn 1979, Schulman & Kaplowitz 1977), mirror-naive chimpanzees in the Povinelli study, which included 105 animals ranging in age from 10 months to 40 years, began showing self-directed behaviors within the first 10-20 minutes of mirror exposure. However, self-recognition, with the mark test as the standard, was observed only in chimpanzees between the ages of 4 1/2 and 8 years of age. Within the population tested, the greatest percentage of individuals demonstrating self-recognition was even older, with 75% of the chimps that passed the mark test being between the ages of 8 and 15 years. Interestingly, Povinelli and his colleagues (1993) also found evidence suggesting either a "decline in self-recognition ability between 16 and 20 years of age or a critical-period effect," (Povinelli et al 1993). These findings were supported even after extended exposure of 20 days of mirror presentation for some animals. It is noteworthy that the purported decline in self-recognition ability, although reported as a group or population effect, was not observed in an individual animal. It would be interesting to see whether a longitudinal study of individual chimpanzees demonstrating self-recognition would show a decrease in this ability over time, for a given chimpanzee.

The Povinelli et al (1993) study also found no differences between mother-raised and peer-raised chimpanzees, but, given the observed differences in a small number of enculturated versus mother-reared chimpanzees with other abilities such as imitation (Tomasello et al 1993), it is likely that similar differences in human-raised and chimpanzee-raised (either by mother or peer) chimpanzees would also be exhibited during mirror exposure, the mark test, or both. Similarly, the differences in environmental rearing strategies between the Lin et al (1992) and Povinelli et al (1993) studies could readily account for age-related differences in mirror-related capacities and, thus, the reported differences in the emergence of self-recognition between the two study populations.

In the Povinelli et al (1993) population, with respect to the delayed capacities for passing the mark test and particularly the failure to exhibit such behaviors among the adult population, chimpanzees had been immersed for a longer period of time under housing conditions that likely did not include environmental enrichment and access to manipulable objects, nesting materials, etc. This study also may have included a larger proportion of chimpanzees who were reared under nonsocial conditions, including possible complete social isolation. In contrast, the chimpanzees tested by Lin and his colleagues were socially reared, including constant peer interaction and a program of human social interaction and including environmental enrichment with toys, games and other activities involving joint attention and positive socialization with their human caregivers.

More recently, chimpanzees have been tested for self-recognition by using three unusual types of mirrors, including ones with convex and concave surfaces, as well as triptych mirrors (Kitchen et al 1996). First, each of the six female chimpanzee subjects was tested with a regular mirror, and all successfully passed the mark test. During subsequent tests with the distorting mirrors, the animals' behaviors in response to the different mirror types were observed, but the chimps were not marked as in the Gallup (1970) study. Kitchen et al (1996) found that the animals did not exhibit any type of social behaviors toward the images they observed when the three different types of mirrors were introduced, and these authors also observed a decrease in the chimpanzees' interactions towards the mirror images over time. However, the authors did report evidence for self-referenced behaviors by the subjects with all three mirror types. Although the experimental manipulations used in their study did not allow the animals to have a clearly reflected image of themselves such that self-recognition based on feature cues was possible (because the mirrors distorted these cues), contingent-movement relationships could still be utilized as cues (Kitchen et al 1996; also see Mitchell 1993). If an understanding of these contingencies is accepted as evidence for some level of self-recognition (Mitchell 1993), then from the Kitchen et al (1996) findings, it follows that chimpanzees are able to abstract their self-image on some level to compensate for the distortion.

Although numerous researchers have utilized several different methodologies to assess self-recognition, passing the mark test continues to be most widely accepted as evidence for self-recognition and, in turn, some facet of self-awareness. But as noted previously, precisely what the mark test measures and the resulting implications for self-awareness and, subsequently, theory of mind remain unresolved (Gallup et al 1995; Heyes 1994, 1995, 1996; Mitchell 1993, 1995; Povinelli et al 1997).

ATTRIBUTIONAL CAPACITIES IN NONHUMANS

Premack & Woodruff (1978a), mentioned previously for their studies examining problem solving in apes, were the first to experimentally attempt to address what they termed "theory of mind" (ToM) in a nonhuman species, with the chimpanzee Sarah as their subject. Since their seminal work, studies of attributional capacities have come to represent a burgeoning literature in the human developmental field (e.g. Baron-Cohen et al 1985; Chandler et al 1989; Flavell 1986, 1988; Gopnik & Graf 1988; Leslie 1987; Perner & Wimmer 1985; Wellman 1990; Wimmer et al 1988; Wimmer & Perner 1983), with fewer investigations of these phenomena in primates until more recently (Boysen et al 1997; Cheney & Seyfarth 1990a,, 1992; Povinelli 1993; Povinelli et al 1990, 1992; Povinelli et al 1991; Kummer et al 1996). As noted earlier, Premack & Woodruff (1978a) presented Sarah, an adult female chimpanzee, with a series of videotaped sequences depicting a caregiver experiencing some type of "problem." For example, one scenario showed the caregiver without a coat, shivering violently while standing next to an unplugged portable heater.

As Sarah watched the videotaped sequence, the video recorder was paused just when the actor would have likely come up with a solution. Sarah was then presented with several photographs and was required to choose the one that best depicted a viable solution. In the case described, Sarah chose the photograph that showed the heater plugged in, inferring that, subsequently, heat would be provided, and the caregiver could warm himself. Of the eight problems presented on videotape, Sarah was able to choose correctly for seven problems (Premack & Woodruff 1978a). The authors interpreted these findings as possible evidence for the ability of chimpanzees to attribute mental states to other individuals, including humans and possibly other chimpanzees. To successfully solve the task, Sarah needed to interpret the scenario from the viewpoint of the caregiver, first determining what the caregiver was attempting to accomplish and then choosing a solution that would provide a remedy for the depicted difficulty of the person.

Analogously, role playing in children has also been interpreted as an indicator of theory of mind (Povinelli 1993). Experimental attempts to address role taking in nonverbal animals were originally addressed by Mason & Hollis (1962), using rhesus monkeys. Mason & Hollis designed an apparatus that positioned two monkeys across from each other, separated by a barrier. One monkey, designated as the "operator," could pull one of two handles, which controlled two sets of food trays, each of which would extend identical food dishes towards each of the animals. However, only one set of the food trays was baited on a given trial, and the operator animal did not have visual access to the trays. Therefore, the operator monkey had no means for determining which tray actually contained the food and thus which handle to pull. The observer monkey, on the other hand, had full view of the trays, and thus it was up to the two animals to work cooperatively, through some type of gestural or indicating behaviors from the observer monkey and subsequent understanding of the communicative function of these behaviors by the operator animal. Eventually, the animals were able to work together, with the observer indicating which tray contained food and the operator then able to select the correct handle to pull, which brought a food dish within reach of each monkey. During a second phase of the experiment in which the animals' roles were reversed, their performance dropped to chance, indicating that neither monkey really understood the relative contribution of the other to their previous and successful cooperative efforts.

Povinelli et al (1992) repeated this experiment with two chimpanzees first serving as operators, with the informant role filled by an experimenter and the other two chimpanzees first tested as informants, while the experimenter performed the operator's role. This required the operator chimpanzees to indicate the correct location of the baited food trays to the experimenter by spontaneously learning to point or gesture towards the correct dishes, and it required the operator chimpanzees to pull the correct handle corresponding to the food dishes to which the experimenter was pointing. After successful cooperation between the chimps and the experimenter, the respective social roles were reversed. Three of the four chimpanzees tested showed immediate evidence of their understanding of their new social role, performing above chance levels.

Povinelli et al (1990) modified these procedures slightly to determine whether the chimpanzees were able to comprehend the relationship between the different knowledge states of seeing versus knowing. In this study, two human observers participated, with each chimpanzee tested individually. Both experimenters were initially present, and then one experimenter (the "guesser") left the room while the knowledgeable experimenter (the "knower") hid food under one of four opaque cups that had been placed behind a low barrier in front of the chimp. The subject therefore was able to see that the experimenter was hiding food somewhere among the cups, but was unable to determine, because of the barrier, under precisely which cup. Once the cup was baited, the second experimenter re-entered the room. During each trial, each of the two experimenters pointed to a different cup. Thus, the chimpanzee had to determine which person was providing a "knowledgeable" cue that would allow them to select the correct cup containing the food and which person was simply guessing. The results suggested that, within the first five days of testing, four of the five chimpanzees tested were able to reliably discriminate between which experimenter was guessing and which actually knew the correct location of the food reward, with the fifth subject showing reliable discrimination later in testing. Although one possible explanation for the results was that the animals were able to recognize or comprehend the knowledge state of each experimenter, because the person who had left the room had not witnessed the hiding event and thus did not have visual access to that information, a more parsimonious explanation was also possible. The chimpanzees could have been simply relying on behavioral cues and the subsequent rule, "pick the person who stays in the room," without their necessarily attributing a knowledge state to either of the experimenters. To address this hypothesis, the procedures were modified slightly such that the "guessing" observer remained in the room, but covered his head with a bag to prevent him from seeing where the food was being hidden. Although it can be argued that this change in methodology introduced another confound by merely changing the behavioral cue ("pick the person who does not have a bag on his head"), Povinelli et al (1990) argued that these results suggested that chimpanzees, different from rhesus macaques (Povinelli et al 1991) and long-tailed macaques (Kummer et al 1996), were able to comprehend the perception-knowledge relationship between the two experimenters and that chimpanzees are thus able to model the visual perspectives of others.

Subsequently, Povinelli et al (1993) reported that the ability to comprehend perception-knowledge relationships was a developmental milestone in human children and emerged between 3 and 4 years of age. To assess the validity of this methodology, Povinelli & deBlois (1992) tested 3- and 4-year-old children by using an apparatus similar to that used by the chimpanzees in the earlier studies. In their study with young children, Povinelli & deBlois found evidence to support the validity of the measures used with chimpanzees, with most of the 4-year-olds correct on 7 of 10 discrimination trials between the two experimenters; only one 3-year-old met the same criterion. However, several ambiguities are worth noting in light of the previous discussion on self-recognition as an indicator of self-awareness that may be among a constellation of abilities, which include theory of mind. The ability of children to successfully complete the perception-knowledge paradigm does not emerge until between 3 and 4 years of age, yet infants as young as 18-24 months are capable of demonstrating mirror self-recognition (e.g. Amsterdam 1972, Bertenthal & Fischer 1978). This discrepancy supports the hypothesis that possessing mirror self-recognition or perhaps even self-awareness does not, in mm, indicate attributional capacities. However, these abilities may be viewed as empirical markers, indicative and necessary for the eventual emergence or development of a theory of mind (Povinelli 1993).

SUMMARY

The four areas of study in comparative cognition briefly outlined in this overview represent lively areas of current debate and discussion and should readily bring to mind a variety of directions for future investigations with both humans and nonhuman species. Despite the difficulties associated with implementing more rigorous empirical approaches to the study of topics such as self-awareness, "true" imitation, and, most notably, attributional capacities of other animals, there are methods and paradigms that offer the opportunity to experimentally explore these nonobservable behaviors beyond the anecdote. Psychology as an interdisciplinary field, with greater theoretical depth and a firmer footing than even a decade ago, appears to have developed the requisite philosophical maturity and sophistication to move forward in addressing the same types of elusive phenomena that prompted psychology to emerge as the new science of behavior over a century ago.

Literature Cited

Amsterdam BK. 1972. Mirror self-image reactions before age two. Dev. Psychol. 5: 297-305

Anderson JR. 1984. The development of self-recognition: a review. Dev. Psychobiol. 17:35-49

Anderson JR, Gallup GG Jr. 1997. Self-recognition in Saguinus? A critical essay. Anim. Behav. 54:1563-67

Baron-Cohen S, Leslie AM, Frith U. 1985. Does the autistic child have a "theory of mind"? Cognition 21:37-46

Beck BB. 1980. Animal Tool Behavior. New York: Garland. 307 pp.

Benhar EE, Carlton PL, Samuel D. 1975. A search for mirror-image reinforcement and self-recognition in the baboon. In Contemporary Primatology: Proc. 5th Int. Congr. Primatol., ed. S Kondo, M Kawai, S Ehara, pp. 202-8. New York: Karger

Bertenthal BI, Fischer KW. 1978. Development of self-recognition in the infant. Dev. Psychol. 14:44-50

Berthelet A, Chavaillon J. 1993. The Use of Tools by Human and Non-Human Primates. Oxford, UK: Oxford Univ. Press. 424 pp.

Boesch C, Boesch H. 1984a. Possible causes of sex differences in the use of natural hammers by wild chimpanzees. J. Hum. Evol. 13:415-40

Boesch C, Boesch H. 1984b. Mental maps in wild chimpanzees: an analysis of hammer transports for nut cracking. Primates 25: 160-70

Boesch C, Boesch H. 1990. Tool use and tool making in wild chimpanzees. Folia Primatol. 54:86-99

Boyd R, Richerson PJ. 1988. An evolutionary model of social learning: the effects of spatial and temporal variation. In Social Learning.' Psychological and Biological Perspectives, pp. 29-48. Hillsdale, NJ: Erlbaum

Boysen ST, Kuhlmeier VA, Kuzdak NE. 1997. Attribution of ignorance by chimpanzees. Presented at Annu. Meet. Am. Psychol. Assoc., Chicago, IL

Bugnyar T, Huber L. 1997. Push or pull: an experimental study on imitation in marmosets. Anim. Behav. 54:817-31

Byrne RW, Tomasello M. 1995. Do rats ape? Anim. Behav. 50:1417-20

Byrne RW, Whiten A. 1991. Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes and Humans. Oxford, UK: Oxford Univ. Press. 413 pp.

Chandler M, Fritz AS, Hala S. 1989. Small scale deceit: deception marker of 2-, 3- and 4-year-olds' early theories of mind. Child Dev. 60:1263-77

Cheney DL, Seyfarth RM. 1990a. Attending to behaviour versus attending to knowledge: examining monkeys' attribution of mental states. Anim. Behav. 40:742-53

Cheney DL, Seyfarth RM. 1990b. How Monkeys See the World. Chicago: Univ. Chicago Press. 377 pp.

Cheney DL, Seyfarth RM. 1992. Characterizing the mind of another species. Behav. Brain Sci. 15:172-79

Clayton DA. 1978. Socially facilitated behavior. Q. Rev. Biol. 53:373-91

Custance DM, Whiten A, Bard KA. 1995. Can young chimpanzees imitate arbitrary actions? Hayes and Hayes (1952) revisited. Behavior 132:839-58

Dawson BV, Foss BM. 1965. The use of social information in the problem-solving of orangutans and human children. J. Comp. Psychol. 109:301-20

Flavell JH. 1986. The development of children's knowledge about the appearance-reality distinction. Am. Psychol. 41: 418-25

Flavell JH. 1988. The development of children's knowledge about the mind: from cognition connections to mental representations. In Developing Theories of Mind, ed. JW Astington, PL Harris, DR Olson, pp. 244-71. Cambridge, UK: Cambridge Univ. Press

Fragaszy DM, Visalberghi E. 1989. Social influences on the acquisition of tool-using behaviors in tufted capuchin monkeys (Cebus apella). J. Comp. Psychol. 103:159-70

Galdikas BMF. 1978. Orangutans and hominid evolution. In Essays Presented to Sutan Takdir Alisjahbana on his Seventieth Birthday, ed. S Udin, pp. 287-329. Jakarta: Dian Rakyat

Galdikas BMF. 1982. Orang-utan tool-use in Tanjung Puting Reserve, Central Borneo (Kalimantan Tengah). J. Hum. Evol. 10: 19-33

Galdikas BMF. 1985. Orangutan sociality at Tanjung Puting. Am. J. Primatol. 9: 101-19

Galdikas BMF. 1988. Orangutan diet, range, and activity at Tanjung Puting, Central Borneo. Int. J. Primatol. 9:1-35

Galdikas BMF, Vasey P. 1992. Why are orangutans so smart? Ecological and social hypotheses. In Social Processes and Mental Abilities in Non-human Primates, ed. FD Burton, pp. 183-224. Lewiston, NY: Edward Mellon

Galef B. 1988. Imitation in animals: history, definition, and interpretation of data from the psychological laboratory. In Social Learning, Psychological and Biological Perspectives, ed. TR Zentall, BG Galef Jr, pp. 3-28. Hillsdale, NJ: Lawrence Erlbaum

Gallup GG Jr. 1970. Chimpanzees: self-recognition. Science 167:86-87

Gallup GG Jr. 1975. Towards an operational definition of self-awareness. In Socioecology and Psychology of Primates, ed. RH Tuttle, pp. 309-41. The Hague: Mouton

Gallup GG Jr. 1982. Self-awareness and the emergence of mind in primates. Am. J. Primatol. 2:237-48

Gallup GG Jr. 1991. Toward a comparative psychology of self-awareness: species limitations and cognitive consequences. In The Self: an Interdisciplinary Approach, ed. GR Goethals, J Strauss, pp. 121-35. New York: Springer

Gallup GG Jr, Povinelli DJ, Suarez SD, Anderson JR, Lethmate J, Menzel EW Jr. 1995. Further reflections on self-recognition in primates. Anim. Behav. 50: 1525-32

Goodall J. 1964. Tool-using and aimed throwing in a community of free-living chimpanzees. Nature 201:1264-66

Goodall J. 1970. Tool using in primates and other vertebrates. In Advances in the Study of Behavior, ed. D Lehrman, R Hinde, E Shaw, pp. 195-249. New York: Academic

Goodall J. 1986. The Chimpanzees of Gombe: Patterns of Behavior. Cambridge, MA: Harvard Univ. Press. 673 pp.

Gopnik A, Graf P. 1988. Knowing how you know: young children's ability to identify and remember the sources of their beliefs. Child Dev. 59:1366-71

Hauser MD, Kralik J. 1997. Life beyond the mirror: a reply to Anderson & Gallup. Anim. Behav. 54:1568-71

Hauser MD, Kralik J, Bott-Mahan C, Garrett M, Oser J. 1995. Self-recognition in primates: phylogeny and the salience of species-typical traits. Proc. Natl. Acad. Sci. USA 92:10811-14

Heyes CM. 1993. Imitation, culture and cognition. Anim. Behav. 46:999-1010

Heyes CM. 1994. Reflections on self-recognition in primates. Anim. Behav. 47: 909-19

Heyes CM. 1995. Self-recognition in primates: further reflections create a hall of mirrors. Anim. Behav. 50:1533-42

Heyes CM. 1996. Self-recognition in primates: irreverence, irrelevance and irony. Anim. Behav. 51:470-73

Heyes CM, Dawson GR. 1990. A demonstration of observational learning using a bi-directional control. Q. J. Exp. Psychol. 42: 59-71

Humphrey NK. 1976. The social function of intellect. In Growing Points in Ethology, ed. PPG Bateson, RA Hinde, pp. 303-17. Cambridge, UK: Cambridge Univ. Press

Inoue-Nakamura N, Matsuzawa T. 1997. Development of stone tool use by wild chimpanzees (Pan troglodytes). J. Comp. Psychol. 111:159-73

Jolly A. 1966. Lemur social behavior and primate intelligence. Science 153:501-6

Kitchen A, Denton D, Brent L. 1996. Self-recognition and abstraction abilities in the common chimpanzee studied with distorting mirrors. Proc. Natl. Acad. Sci. USA 93:7405-8

Kortlandt A. 1986. The use of tools by wild-living chimpanzees and earliest hominids. J. Hum. Evol., 15:77-132

Kummer H, Anzenberger G, Hemelrijk CK. 1996. Hiding and perspective taking in long-tailed macaques (Macaca fascicularis). J. Comp. Psychol. 110:97-102

Leslie AM. 1987. Pretense and representation in infancy: the origins of theory of mind. Psychol. Rev. 94:412-26

Lewis M, Brooks-Gunn J. 1979. Social Cognition and the Acquisition of Self. New York: Plenum. 296 pp.

Limongelli L, Boysen ST, Visalberghi E. 1995. Comprehension of cause-effect relations in a tool-using task by chimpanzees (Pan troglodytes), J. Comp. Psychol. 109: 18-26

Lin AC, Bard KA, Anderson JR. 1992. Development of self-recognition and self-conscious emotions. Child Dev. 106: 120-27

Mason WA, Hollis JH. 1962. Communication between young rhesus monkeys. Anim. Behav. 10:211-21

Matsuzawa T. 1994. Field experiments on use of stone tools by chimpanzees in the wild. In Chimpanzee Cultures, ed. R Wrangham, WC McGrew, F de Waal, P Heltne, pp. 351-70. Cambridge, MA: Harvard Univ. Press

McGrew WC. 1974. Tool use by wild chimpanzees in feeding upon driver ants. J. Hum. Evol. 3:501-8

McGrew WC. 1992. Chimpanzee Material Culture. Cambridge, MA: Harvard Univ. Press. 277 pp.

Mitchell R. 1993. Mental models of mirror self-recognition: two theories. New Ideas Psychol. 11:211-17

Mitchell R. 1995. Self-recognition, methodology and explanation: a comment on Heyes (1994). Anim. Behav. 51:467-69

Morgan CL. 1900. Animal Behavior. London: Arnold. 344 pp.

Nagell K, Olguin RS, Tomasello M. 1993. Processes of social learning in the tool use of chimpanzees (Pan troglodytes) and human children (Homo sapiens). J. Comp. Psychol. 107:174-86

Nishida T. 1987. Local traditions and cultural transmission. In Primate Societies, ed. B Smuts, D Cheney, R Seyfarth, R Wrangham, T Struhsaker, pp. 462-74. Chicago: Univ. Chicago Press

Nishida T, Hiraiwa M. 1982. Natural history of tool-using behavior by wild chimpanzees in feeding upon wood-boring ants. 2. Hum. Evol. 11:73-99

Parker ST, Mitchell RW, Boccia ML. 1994. Self-awareness in Animals and Humans. New York: Cambridge Univ. Press. 442 pp.

Passingham RE. 1982. The Human Primate. New York: Freeman. 390 pp.

Pepperberg IM, Garcia SE, Jackson EC, Marconi S. 1995. Mirror use by African Grey parrots (Psittacus erithacus). J. Comp. Psychol. 109:182-95

Perner J, Wimmer H. 1985. "John thinks that Mary thinks that ...": attribution of second-order beliefs by 5- to 10-year-old children. J. Exp. Child Psychol. 39:437-71

Povinelli DJ. 1987. Monkeys, apes, mirrors and minds: the evolution of self-awareness in primates. J. Hum. Evol. 2:493-509

Povinelli DJ. 1989. Failure to find self-recognition in Asian elephants (Elephas maximus) in contrast