Transcript
Precursors of morality in the use of the symbols‘‘good ” and ‘‘bad ” in two bonobos ( Pan paniscus ) and a chimpanzee ( Pan troglodytes ) Heidi Lyn a,b,* , Becca Franks c , E. Sue Savage-Rumbaugh b a Sea Mammal Research Unit, Gatty Marine Lab, School of Biology, University of St.Andrews, St. Andrews, Fife KY16 8LB, United Kingdom b Great Ape Trust of Iowa, 4200 S.E. 44th Avenue, Des Moines, IA 50320, United States c Columbia University, New York, NY 10027, United States Abstract Morality is a concept that is based on value judgments of ‘‘right ” or ‘‘wrong ” and ‘‘good ” and‘‘bad ” . Three language-competent apes (two bonobos and a chimpanzee) are shown to use the sym-bols ‘‘good ” and ‘‘bad ” in appropriate contexts and to co-construct these values with the humans intheir environment, indicating that the specific expression of value judgments is cultural. Their devel-opmental use of these symbols parallels studies in children which suggest that conceptual simplicityand internal development may affect the development of moral precursors. These findings supportrecent research that has found moral precursors in several species of nonhuman primates. 2008 Elsevier Ltd. All rights reserved. Keywords: Moral precursors; Primates; Value judgments; Language-competent apes 1. Introduction Morality is a concept that is most frequently left to philosophers, as there are few directbehavioral measurements that can be made. In nonhumans, the study of morality has gen- 0271-5309/$ - see front matter 2008 Elsevier Ltd. All rights reserved.doi:10.1016/j.langcom.2008.01.006 * Corresponding author. Address: Sea Mammal Research Unit, Gatty Marine Lab, School of Biology,University of St. Andrews, St. Andrews, Fife KY16 8LB, United Kingdom. Tel.: +44 01334 467260; fax: +4401334 463443. E-mail address:
[email protected] (H. Lyn). Available online at www.sciencedirect.com Language & Communication 28 (2008) 213–224 www.elsevier.com/locate/langcom LANGUAGE&COMMUNICATION erallyfocusedonbehavioralprecursors,suchasaltruism,cooperation,andfairness(Bekoff,2004;Katz,2000),astheseallowfordirectbehavioralevaluation.Forexample,Brosnanandde Waal (2003) found that brown capuchin monkeys ( Cebus apella ) will reject rewards thattheyconsider‘‘unfair ” .Indeed,previousresearchhasdelineatedasuiteofmoralprecursorsthatarepresentinnonhumanprimates,includingempathy,sympathy,andaconceptofrec-iprocity(FlackanddeWaal,2000),buthasyettofullyexploretheconceptsofperceptionof intention or using/perceiving social norms (Call, 2000), and has not yet touched other con-cepts that are more linguistically based, such as value judgments.At its very essence, morality must rely on value judgments. For, if there is no ‘‘right ” or‘‘wrong ” , no ‘‘good ” or ‘‘bad ” , then there is nothing on which to base a morality. There issome discussion by philosophers and scientists about whether humans have an innate con-cept of morality, and therefore of value judgments, and when said concept would haveevolved (e.g., Hauser, 2006; Wilson, 1993). However, specific views and expressions of value judgments vary from culture to culture and are therefore seen to be (at least par-tially) culturally acquired (also e.g., Wilson, 1993). Certainly, the expression of value judg-ments as a specific linguistic behavior is measurable and therefore may allow for a moredirect glimpse into the development of morality.Supporting this argument, Lamb (1991) states that ‘‘the acquisition of morally relevantwords such as good, bad, naughty and nice may reflect ” the beginnings of morality. Inter-estingly, Snow (1987) found that at least one child began by using the term ‘‘bad ” morefrequently in moral utterances than the term ‘‘good ” , but then switched their preference,using ‘‘good ” more frequently than ‘‘bad ” . Snow (1987) does not discuss this switch in herchapter, but we suggest that children first begin to label inappropriate actions (which arelikely more frequently pointed out to them by their caregivers and may therefore be easierto understand) and only after, label themselves or their actions as appropriate (‘‘good ” ) todistinguish them. This interpretation is supported by work in older children in which chil-dren initially describe ‘‘good ” acts as acts absent of ‘‘bad ” behavior (e.g., ‘‘Being good isnot breaking windows ” ), and only later describe both ‘‘good ” acts in terms of prosocialbehavior (Hill and Hill, 1977).This late development of moral understanding may therefore be driven, not just by cul-tural acquisition, but by other factors such as the simplicity of the concepts themselves, ormore direct internal development. Supporting this possibility, Lamb (1991) found that inchildren, the use of internal state words, or words that represent internal psychologicalstates, such as happy, tired, and hungry, preceded the first stages of moral development,while following the development of ‘‘awareness of standards ” . Contrary to the culturalacquisition theory, however, Lamb found that the caregivers’ use of internal state wordswas rare and did not seem to drive the children’s use of these words. However, internalstate words did appear at the same time that empathy appeared, and so may be relatedto the development of empathy, an internally driven developmental path.Complex symbolic abilities in two bonobos and a chimpanzee allow for a similar lin-guistic exploration of value judgments in nonhuman apes. Savage-Rumbaugh et al.(e.g., 1986; 1993) have explored symbolic capacities in bonobos ( Pan paniscus ) and chim-panzees ( Pan troglodytes ) in studies that have indicated that, when reared in an environ-ment with English, a keyboard of visual symbols (lexigrams), and cultural emersion, apescan acquire symbolic capacities without explicit training (Brakke and Savage-Rumbaugh,1995, 1996; Greenfield and Savage-Rumbaugh, 1991; Lyn et al., submitted for publication;Lyn and Savage-Rumbaugh, 2000; Savage-Rumbaugh et al., 1986, 1993, 1980). 214 H. Lyn et al./Language & Communication 28 (2008) 213–224 Among others, two bonobos ( P. paniscus ), Kanzi – a male born in 1980, and Panban-isha – a female born in 1985, and a chimpanzee ( P. troglodytes ), Panpanzee (or Panzee) – afemale born in 1986, have demonstrated the abilities to use lexigrams to name objects indouble blind studies (Brakke and Savage-Rumbaugh, 1995, 1996; Savage-Rumbaughet al., 1986), to associate novel English names with novel objects with very few exposuresto both object and word (Lyn and Savage-Rumbaugh, 2000), to hierarchically categorizethese referents (Lyn, 2007), to utilize imitation in an intentionally communicative context(Greenfield, 1980), to make semantically-based combinations across both lexigram andgestural combinations (Greenfield and Lyn, 2007; Greenfield and Savage-Rumbaugh,1990, 1991; Lyn et al., submitted for publication), and to comprehend English sentencesat least at a similar level to a two-and-a-half year old child tested in the same manneras the ape (Savage-Rumbaugh et al., 1993). Additionally, recent publications haveexplored the development of a unique ape/human culture at the Language Research Cen-ter and now the Great Ape Trust of Iowa (Savage-Rumbaugh et al., 2004). The intensive24-hour interaction among the apes and humans, sharing daily activities and communica-tive modes have produced extended spontaneous cognitive and social abilities not other-wise documented in these species, for example, pretend play (Lyn et al., 2006) and jointattention and imitation (Savage-Rumbaugh et al., 2004).DuringinteractionswiththeapesattheLanguageResearchCenter,andnowattheGreatApeTrustofIowa,humancaregiversutilizedtheterms‘‘good ” and‘‘bad ” (inEnglishand/orlexigram form) as value judgments as they would in ordinary interactions. For example,‘‘Panzee’sbeingbad ” or,‘‘Thisisagoodapple ” .Theywouldalsoproduceinterrogativesthatrequested value judgments from the apes. For example, ‘‘Do you think that’s a good idea? ” or‘‘Doyouknowhowyou’vebeenacting? ” Accordingly,withinthehumanandapeculture,the apes have learned to use the ‘‘bad ” and ‘‘good ” lexigrams in many diverse contexts.We propose an exploration into the apes’ use of ‘‘good ” and ‘‘bad ” lexigrams and theco-construction of these value judgments within a Pan/Homo culture. 2. Methods 2.1. Subjects Two bonobos ( P. paniscus ) – Kanzi, born April 1980, and his half-sister Panbanisha,born December 1985, and one chimpanzee ( P. troglodytes ) – Panpanzee, born January1986, served as subjects. All three were reared at the Language Research Center inAtlanta, Georgia in a within- and cross-species communicative environment that includedthe use of English by human caretakers and a keyboard printed with visuo-graphic sym-bols (lexigrams) used by several of the apes (including all three subjects) as well as the care-takers. In this environment, all three apes have been shown to acquire symbolic capacitieswithout explicit training (Brakke and Savage-Rumbaugh, 1995, 1996; Greenfield andSavage-Rumbaugh, 1991; Lyn et al., submitted for publication; Lyn and Savage-Rumb-augh, 2000; Savage-Rumbaugh et al., 1986, 1993, 1980). The apes have demonstratedthe abilities to use lexigrams to name objects in double blind studies (Brakke andSavage-Rumbaugh, 1995, 1996; Savage-Rumbaugh et al., 1986), to associate novel Englishnames with novel objects with very few exposures to both object and word (Lyn andSavage-Rumbaugh, 2000), to utilize imitation in an intentionally communicative context(Greenfield, 1980), to make semantically-based combinations across both lexigram and H. Lyn et al./Language & Communication 28 (2008) 213–224 215 gestural combinations (Greenfield and Lyn, 2007; Greenfield and Savage-Rumbaugh,1990, 1991; Lyn et al., submitted for publication), and to comprehend English sentencesat least at a similar level to a two-and-a-half year old child tested in the same manneras the ape (Savage-Rumbaugh et al., 1993). 2.2. Database The utterance database includes information about the apes’ use of the keyboard fromNovember 11, 1985 to January 19, 1997. During intensive study periods (1987–1990), theutterance database included all uses of the keyboard by Kanzi, Panbanisha and Panpan-zee, as well as communicative gestures used in combination with keyboard utterances.These were recorded by hand by the caregivers and input into the computer at the endof the day. This procedure provided the researchers with an exhaustive written recordof their symbol use. At other times (1985–1987 and 1990–1997), due to caretaker short-ages, data recording was more sporadic, including all uses of the keyboard for only certainhours of the day and certain days of the week. At no time were utterances selectivelyentered into the database. As a reliability check, real-time recording was checked against4.5 h of videotape. Thirty-seven out of 46 utterances, or 80%, were noted by both the real-time and the video observer and there was 100% agreement on the lexigram that had beenused when both observers noted the utterance. Hence, we conclude that our corpus ishighly reliable, but an underestimation of quantity.Each record contained the utterance, date, record number, ape, researcher, codes as topragmatic force of the utterance, behavioral concordance notes, and a short contextualnote. The procedure was similar to observational protocols used in studies of child lan-guage which provided the comparative foundation for this study (Bowerman, 1973;Brown, 1973). Several studies have been published utilizing this database (Brakke andSavage-Rumbaugh, 1995, 1996; Greenfield and Savage-Rumbaugh, 1990, 1991; Lyn,2007; Lyn et al., submitted for publication).Contextual notes were provided by a second caregiver who functioned as an observer.This contextual description in the utterance database was our basis for coding these combi-nations as to their semantic meaning. When the ape was with only one caregiver, the datawould be excluded from our corpus for lack of contextual information. We estimate thattwocaregiver/researcherswerepresentwiththeapesaboutfourandone-halfhoursperday.The database contains exhaustive utterance information for Panbanisha and Panpanzeefrom 1 year of age (1988) through 3 years of age (1990). Kanzi was already over 5 years of age at the initiation of the utterance database. Therefore, all developmental analyses willfocus on Panbanisha and Panpanzee. 3. Results 3.1. Use of ‘‘good ” and ‘‘bad ” Kanzi, Panbanisha, and Panpanzee all use the terms ‘‘good ” and ‘‘bad ” in appropriatecontexts and showed comprehension. The database contained 119 utterances of good andbad by Kanzi, including 100 utterances containing a ‘‘good ” lexigram and 24 a ‘‘bad ” lex-igram (5 utterances were combinations that included both ‘‘good ” and ‘‘bad ” ), 860 utter-ances by Panbanisha, 697 ‘‘good ” and 174 ‘‘bad ” (11 with both), and 115 by Panpanzee, 216 H. Lyn et al./Language & Communication 28 (2008) 213–224 33 good and 83 bad (1 with both). Panpanzee’s preference for the ‘‘bad ” lexigram in con-trast to Panbanisha’s preference for the ‘‘good ” lexigram will be further analyzed below.Utterance examples follow (In these and all following examples, all caps indicates the useof a lexigram): Bad . Panpanzee – Date: 6/12/1988; Record: 14. I declined Panpanzee’s PLAY requestand continued my litany, saying next ‘‘Do you know how you’ve been acting? ” Panpanzeeturned to the large faceplate & touched BAD. I agreed that she had been bad & advisedher to be GOOD now. Bad Good . Panbanisha – Date: 10/21/88; Record: 15. Liz and Panpanzee have gone tovisit the orangutans in the colony room and Panbanisha very much wants to go also. Theexperimenter says ‘‘NO GO ” as Panbanisha is getting her physical now. Panbanisha startsto leave. The experimenter tells her not to be BAD because she said NO GO ORANU-TANS. Panbanisha imitates BAD, then comments GOOD.Some uses of good and bad are harder to interpret than others, but analyses revealed nosignificant difference in the percentage of utterances that were coded uninterpretable(‘‘unknown ” code) or babbling for ‘‘good ” and ‘‘bad ” utterances than for all utterancescombined (see Fig. 1). However, a significantly lower percentage of ‘‘bad ” or ‘‘good ” utterances were coded as requests than were coded for all utterances combined for all threeapes (Kanzi – v 2 (1833, 1) = 15.10, p < 0.001; Panbanisha – v 2 (9779, 1) > 23.9, p < 0.001;Panpanzee – v 2 (10,219, 1) = 17.39, p < 0.001). Also, a significantly higher percentage of ‘‘bad ” or ‘‘good ” utterances were coded as statements or comments than were codedfor all utterances combined for all three apes (Kanzi – v 2 (1139, 1) > 23.9, p < 0.001; Pan-banisha – v 2 (6569, 1) > 23.9, p < 0.001; Panpanzee – v 2 (6296, 1) > 23.9, p < 0.001). 3.2. Developmental analysis3.2.1. Good and bad Both Panbanisha and Panpanzee used ‘‘bad ” before ‘‘good ” and, initially, ‘‘bad ” wasused more frequently by both apes (see Fig. 2). Panbanisha showed a strong spike in Percentage of utterances by pragmatic force 0.0%10.0%20.0%30.0%40.0%50.0%60.0%70.0%80.0%90.0%100.0% r e q u e s t c o m m e n t / s t a t e m e n t a n s w e r q u e s t i o n u n k n o w n / b a b b l i n g s t r u c t u r e d / i m i t a t i o n c o m p r e h e n s i o n M i s c . Pragmatic Force Kanzi - good and badKanzi - all utterancesPanbanisha - good and badPanbanisha - all utterancesPanpanzee - good and badPanpanzee - all utterances Fig. 1. Percentage of utterances by the three apes that were coded as one of seven categories of pragmatic force. H. Lyn et al./Language & Communication 28 (2008) 213–224 217
