Primates (2006) 47:279–283 DOI 10.1007/s10329-006-0180-1
SH O RT CO MM U N IC A T IO N
Robert C. O’Malley Æ W. C. McGrew
Hand preferences in captive orangutans (Pongo pygmaeus)
Received: 24 June 2005 / Accepted: 13 December 2005 / Published online: 8 April 2006 Japan Monkey Centre and Springer-Verlag 2006
Abstract The strength of the evidence for populationlevel handedness in the great apes is a topic of considerable debate, yet there have been few studies of handedness in orangutans. We conducted a study of manual lateralization in a captive group of eight orangutans (Pongo pygmaeus) ranking the degrees of manual preference according to a defined framework. We analyzed five behavioral patterns: eat (one- and two-handed), make/modify tool, oral tool-use, and manual tool-use. Although some individuals showed significant manual preferences for one or more tasks, at the group-level both one-handed and two-handed eating, oral tool-use, and make/modify tool were ranked at level 1 (unlateralized). Manual tool-use was ranked at level 2, with four subjects demonstrating significant hand preferences, but no group-level bias to the right or left. Four subjects also showed hand specialization to the right or left across several tasks. These results are consistent with most previous studies of manual preference in orangutans. The emergence of manual lateralization in orangutans may relate to more complex manipulative tasks. We hypothesize that more challenging manual tasks elicit stronger hand preferences. Keywords Handedness Æ Manual laterality Æ Orangutan Æ Pongo pygmaeus Æ Tool use
R. C. O’Malley (&) Department of Anthropology, University of Southern California, Los Angeles, CA 90089, USA E-mail:
[email protected] Tel.: +1-213-7401900 Fax: +1-213-7478571 W. C. McGrew Department of Biological Anthropology, Cambridge University, Downing Street, Cambridge, CB2 3DZ, UK
Introduction Laterality of manual function, or ‘‘handedness,’’ may be a trait that non-human primates share with Homo sapiens, at least to some degree (MacNeilage et al. 1987). There is evidence that laterality is related to posture (e.g., Hopkins 1993) and expressed more strongly in physically and cognitively demanding tasks, including tool-use (e.g., Boesch 1991; McGrew and Marchant 1992, 1997a, b; McGrew et al. 1999) or bimanual tasks (Hopkins and Rabinowitz 1997; Olson et al. 1990) in both non-human primates and humans (Marchant et al. 1995; Perelle and Ehrman 1994). Researchers disagree on how to interpret the laterality data currently available. McGrew and Marchant (1997a) in a meta-analysis of primate laterality studies argued that ‘‘there is as yet no compelling evidence that non-human primates are lateralized at the population level’’ (p. 226) while Hopkins (1996) in an independent review of laterality research concluded that ‘‘handedness at a population level appears evident in chimpanzees, at least for some measures’’ (p. 453). Further research by Hopkins and colleagues (e.g., Hopkins et al. 2004) has consistently yielded right-hand preferences in multiple populations of captive chimpanzees for a variety of tasks. All investigators agree that factors such as the age, sex, and rearing history of study subjects, posture, and task complexity may influence patterns of hand-use. Authors disagree as to the relevance of certain methodological issues (such as recording data in bouts vs events), and the significance of the differences that may exist between studies of wild (naturalistic) and captive (laboratory) populations (e.g., Hopkins 1996, 1999; Hopkins et al. 2001; McGrew and Marchant 1996, 1997a, b). There have been few studies of orangutan hand preferences in either wild or captive settings, and evidence for lateralization has been mixed (Heestand 1986; Hopkins 1993; Hopkins et al. 2003; Olson et al. 1990; Rogers and Kaplan 1996; see Discussion). Here, we
280
present data on manual laterality in orangutans (Pongo pygmaeus) at the National Zoological Park in Washington, DC. We examined hand-use in eating, toolmaking, and tool-use in an extraction task similar to that seen in wild Sumatran orangutans (Fox et al. 1999) for individual and group-level hand preferences.
In order to maintain independence of data-points, we recorded data in bouts rather than events (McGrew and Marchant 1997a, pp. 205–206). Bouts were distinguished by a change in the hand, foot, or mouth in use, a substantial change in the subject’s position that required climbing (i.e., to move from the cage mesh to the floor), or another, intervening behavior pattern.
Methods
Data analysis
Subjects
McGrew and Marchant’s (1996, 1997a, b) five-tier framework of manual specialization has been applied to several meta-analyzes and was used to evaluate the present results. In level 1, most individuals in a population are unlateralized, showing no hand preference for a task or set of tasks. In level 2, most or all subjects are incompletely lateralized (show a statistically significant but not exclusive hand preference) either to the left and right for a given task or set of tasks. In level 3, most or all subjects use only one hand exclusively for a task (show complete lateralization), but the distribution to the left and right remains about 50:50. In level 4, the majority of individuals are significantly but incompletely lateralized for a given task, and their overall distribution is skewed either to left or right. Finally, in level 5, the majority of individuals are completely lateralized (always or almost always using the same hand) for a given task, and their overall distribution is significantly skewed to the left or the right. Humans are usually considered to be lateralized at level 5 (but see Marchant et al. 1995; Perelle and Ehrman 1994). We used two-tailed binomial tests to compare frequencies of bouts by the right and left hands in each behavioral category for each subject to determine the degree of hand preference (McGrew and Marchant 1997a). Results were compared both across subjects and across tasks. Behavioral bouts in which both hands were used simultaneously and in the same way were excluded from analysis. Use of the subordinate, or non-active hand was ignored for the present study except in twohanded eat. When a tool was held in the mouth, the hand holding the block was scored as the ‘‘dominant’’ hand, though it was actually in a subordinate role relative to the mouth. Hand preferences within each individual for each task were considered to be significant if p £ 0.05. Subjects for whom there were too little data for analysis are marked ‘‘n/a’’.
The subjects were four male (11–32 years old, mean 20) and four female (11–31 years old, mean 21) captive-born orangutans (Pongo pygmaeus). All subjects had prior experience with tool-use tasks, including dipping tubes and raisin blocks similar to those employed in this study (see Gilbert 1996). Further details are provided in O’Malley and McGrew (2000). Procedures Each morning and afternoon from Monday to Friday each available subject was given a 15.2·8.6·3.6 cm wooden block in which four 1.0 cm diameter holes were drilled to a depth of about 2.5 cm. Each hole was stuffed with 2–4 raisins. R.C.O. provided a large, freshly cut branch of golden bamboo (Phyllostachys aurea) as a raw material for tool manufacture. He recorded hand, foot, and mouth use in the tool-assisted removal of the raisins and noted the dominant (= more complex or precise component) and subordinate (= secondary or supportive component) roles adopted by the hands, feet, and mouth. He also noted hand-use during the modification of bamboo for tools. A trial began with the acceptance of the block, and ended when the subject (1) returned the block to the researcher, (2) lost interest in the block and left it on the ground, or (3) surrendered the block to another subject. On about 25% of observation days one or more subjects were available for only one trial or were unavailable entirely due to medical procedures, keeper schedules, public shows, and/or subject refusals. R.C.O. conducted additional observations during morning and afternoon feedings when possible. R.C.O. collected data from June to August 1998 by hand and via a Psion Workabout hand-held computer and the Noldus Observer program. Five behavioral categories are discussed in this manuscript: eat (pick up food or other object and place in mouth), two-handed eat (pick up food or other object and place in mouth while holding other food items in other hand), make/ modify tool (alter wand, branch, stick, or other object for use as a tool), mouth-fish (spear, lever, pry, or otherwise extract raisins from a wooden block using a twig, straw, splinter, or other tool held in lips or teeth), and hand-fish (spear, lever, pry, or otherwise extract raisins from a wooden block using a twig, straw, splinter, or other tool by hand).
Results Significant hand preferences at the individual level emerged in one-handed eat (three of eight subjects), make/modify tool (three of eight subjects), mouth-fish (three of seven subjects), and hand-fish (four of six subjects). When bouts were pooled across tasks for each subject, three (two males, one female) showed significant overall hand preferences to the left and three (one male, two females) to the right (Table 1). There was no evi-
281
dence of population-level hand preferences within or across tasks (Table 2). Of the 37 possible categories of 8 subjects · 5 categories (plus 3 for which there was not sufficient data for analysis), 24 (65%) showed no departure from chance (50%L:50%R). Both types of eating, oral tool-use (mouth-fish), and tool manufacture/ modification were classed at level 1 (unlateralized), with most individuals being ambilateral and a minority showing significant but incomplete lateralization to one side or the other. Only manual tool-use (hand-fish) was classified at level 2, with a majority of individuals significantly lateralized for each task but no populationlevel bias to the left or right. Two of the adult males (AZ, TU) and one adult female (BO) showed a significant left-hand preference across multiple (>1) behavior categories, while one female (IN) showed a consistent right-hand preference. One individual (KI) showed a significant left-hand preference when engaged in oral tool-use, but a right-hand preference when engaged in manual tool-use. This reversal may be related to KI’s favored mode of extraction, which was to forcefully jam a tool held in the mouth into the block (which was held in one or both hands) to rupture the raisins before retrieving them with the same tool, held manually. This method was scored as manual tool-use rather than oral tool-use. It may be that the use of the hand or hands during the oral component of this method biased the hand that was subsequently used to retrieve ruptured raisins. Two of the eight subjects (JU and IR) showed no significant hand preferences. We noted a total of eight bouts of foot use (among four subjects) in the tasks discussed here. Five were bouts of eating from one foot while holding food in both hands, two were bouts of tool modification while holding the block in one foot, and one was a bout of handfishing while holding the block in one foot.
Discussion The techniques employed by these orangutans notably resembled those seen in wild Sumatran orangutans for
extracting embedded seeds and insects (Fox et al. 1999). O’Malley and McGrew (2000) reported a strong preference for oral tool-use over manual tool-use in five of the eight subjects. Interestingly, the four individuals who showed no significant hand preferences in manual tooluse in this study (IR, JU, PE, TU) all favored oral tooluse, though only two (PE, TU) showed a significant hand preference in the latter task. For a minority of subjects (AZ, IN, TU), there is a consistent trend to the left or right in all tasks (though not all trends are significant), suggesting that larger sample sizes may have yielded additional statistically significant hand preferences at the individual level. A valid criticism of this study, as noted by one anonymous reviewer, is that our categories are quite broad and we did not examine details of posture, body orientation, off-hand use, and fine movements and manipulation of hands and digits. If laterality is highly context-dependent and easily masked, it may not be surprising that only weak hand preferences were detected. Nevertheless, our results are largely consistent with other studies of orangutan manual activities in which most non-tool use activities are ranked at level 1 or 2 (McGrew and Marchant 1997a). Heestand (1986) found no population-level bias in picking up food or other objects in captive orangutans, and those subjects were ranked at level 1. Olson et al. (1990) and Hopkins (1993) reported individual hand preferences in orangutans, but no population bias to the right or left in food retrieval tasks, ranking at level 2. However, Rogers and Kaplan (1996) found population-level left-hand preferences in rehabilitant orangutans for fine manipulation of parts of the face and head (n=21, 1RH, 14LH), ranking at level 4. Bimanual hand-use while feeding in a propped position (n=18, 5RH, 4LH) ranked at level 2. Other behaviors they examined showed hand preferences at the individual but not the population-level, ranking at level 1. Most recently, Hopkins et al. (2003) reported a population-level left-hand preference in a coordinated bimanual task in captive orangutans (n=19, 4RH, 15LH), ranking at level 4.
Table 1 Summary of hand preferences Subjects
AZ BO IN IR JU KI PE TU Total
EA
BM
MT
MF
HF
Total
L
R
p
L
R
p
L
R
p
L
R
p
L
R
p
L
R
p
81 51 44 79 46 65 61 66 493
38 48 91 97 58 77 76 24 509
