Folkecology, Cultural Epidemiology ... - ARTIS International · operative institutions) show systematic awareness of ecological complexity involving animals, plants, and people and

Folkecology, Cultural Epidemiology, and the Spirit of the Commons: A Garden Experiment inthe Maya Lowlands, 1991–2001Author(s): Scott Atran, Douglas Medin, Norbert Ross, Elizabeth Lynch, Valentina Vapnarsky,Edilberto Ucan Ek’, John Coley, Christopher Timura, Michael BaranSource: Current Anthropology, Vol. 43, No. 3 (June 2002), pp. 421-450Published by: The University of Chicago Press on behalf of Wenner-Gren Foundation for AnthropologicalResearchStable URL: http://www.jstor.org/stable/10.1086/339528 .Accessed: 06/03/2011 15:04

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .http://www.jstor.org/action/showPublisher?publisherCode=ucpress. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

The University of Chicago Press and Wenner-Gren Foundation for Anthropological Research are collaboratingwith JSTOR to digitize, preserve and extend access to Current Anthropology.

http://www.jstor.org

http://www.jstor.org/action/showPublisher?publisherCode=ucpress

http://www.jstor.org/action/showPublisher?publisherCode=wennergren

http://www.jstor.org/action/showPublisher?publisherCode=wennergren

http://www.jstor.org/stable/10.1086/339528?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp

http://www.jstor.org/action/showPublisher?publisherCode=ucpress

421

C u r r e n t A n t h ro p o l o g y Volume 43, Number 3, June 2002� 2002 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved 0011-3204/2002/4303-0003$3.50

Folkecology, CulturalEpidemiology, and theSpirit of theCommons

A Garden Experiment in theMaya Lowlands, 1991–20011

by Scott Atran, Douglas Medin,Norbert Ross, Elizabeth Lynch,Valentina Vapnarsky,Edilberto Ucan Ek’, John Coley,Christopher Timura, andMichael Baran

Using a variation on an experimental approach from biology, wedistinguish the influence of sociocultural factors from that of ec-onomic, demographic, and ecological factors in environmentalmanagement and maintenance. This is important to issues ofglobal environmental change, where there is little empirical re-search into cultural effects on deforestation and land use. Find-ings with three groups who live in the same rain-forest habitatand manifest strikingly distinct behaviors, cognitions, and socialrelations relative to the forest indicate that rational self-interestand institutional constraints may not by themselves account forcommons behavior and cultural patternings of cognition are sig-nificant. Only the area’s last native Itza’ Maya (who have few co-operative institutions) show systematic awareness of ecologicalcomplexity involving animals, plants, and people and practicesclearly favoring forest regeneration. Spanish-speaking immigrantsprove closer to native Maya in thought, action, and social net-working than immigrant Q’eqchi’ Maya (who have highly coop-erative institutions). The role of spiritual values and the limita-tions of rational, utility-based decision theories are explored.Emergent cultural patterns derived statistically from measure-ments of individual cognitions and behaviors suggest that cul-tural transmission and formation consist not primarily of sharedrules or norms but of complex distributions of causally con-nected representations across minds.

s c o t t a t r a n is a senior research scientist at the CNRS-Insti-tut Jean Nicod, Paris, and the Institute for Social Research, Uni-versity of Michigan, Ann Arbor, Mich. 48106-1248, U.S.A. [[email protected]] and Leverhulme visiting professor ofanthropology at the University of London.

d o u g l a s m e d i n is chairman of the Department of Psychol-

ogy and director of the Cognitive Science Program at Northwest-ern University.

n o r b e r t ro s s is an assistant research professor of psychologyat Northwestern University.

e l i z a b e t h l y n c h is a postdoctoral fellow in psychology atNorthwestern University.

v a l e n t i n a v a p n a r s k y is a research scientist with CNRS-EREA in Villejuif, France.

e d i l b e r t o u c a n e k ’ is director of the Herbolaria Maya“Bo’oy Ch’iich’” in Uman, Yucatan.

j o h n c o l e y is an assistant professor of psychology at North-eastern University.

c h r i s t o p h e r t i m u r a is pursuing a J.D./Ph.D. in law andanthropology at the University of Michigan.

m i c h a e l b a r a n is a graduate student in anthropology at theUniversity of Michigan.

The present paper was submitted 23 iii 01 and accepted 11 x 01.

This article reports an attempt to establish the causalpathways that determine how cultural ideas result inbehaviors that affect the environment and commonsmanagement. In order to do so, it is important to separateselected cultural effects from those of demography, econ-omy, ecology, and so forth. A critical case for the im-portance of cultural selection versus environmental de-termination comes from a variation on the “gardenexperiment” in biology. When members of a species havedifferent phenotypes in different environments, samplesare taken from both environments and replanted in onlyone. If the differences still exist, they are probably ge-netic (two genotypes); if not, then they are probably en-vironmental (one genotype producing two phenotypes).In using a variation on this experimental approach, ouraim is not to distinguish genetic nature from environ-mental nurture but to distinguish the influence of cer-tain sociocultural factors (social networks, cognitivemodels) from that of economic (sources and level of in-come), demographic (family and population size), and ec-ological (habitat and species) factors in environmentalmanagement and maintenance. Evidence for the influ-ence of culturally transmitted factors on behavior is datashowing that groups of people with different culturalhistories and cultural ideas behave differently in thesame physical environment.

We adopt a threefold approach to understanding causal

1. We thank the people of San Jose, La Nueva, and Corozal for theirgracious reception and collaboration. Nicole Berry helped with datacollection, Ximena Lois with transcriptions, and Sergey Blok withfigures. This work was supported by the National Science Foun-dation (SBR 9422587, SES-9981762) and the Russell Sage Founda-tion (87-99-02). A report of initial findings appeared in the Pro-ceedings of the National Academy of Sciences, U.S.A. (Atran et al.1999), with some of the data reported here. Errors in tables 3 and4 (but not in data analyses) are corrected here. Residual analysesalso had errors in presentation (but not results) that are here clar-ified and amplified. [Supplementary materials appear in the elec-tronic edition of this issue on the journal’s web page (http://www.journals.uchicago.edu/CA/home.html).]

422 F current anthropology Volume 43, Number 3, June 2002

relations between individual cognitions, human behav-iors that directly affect the biotic environment, and cul-tural patterns that emerge from populationwide distri-butions of cognitions and behaviors:

1. Folkecology. We aim to furnish a cross-culturalmethodology for modeling people’s cognitions of the ec-ological relationships between plants, animals, and hu-mans. We verify the material implications of psycholog-ical measures of how people think about nature withphysical measures of accompanying actions that modifynature (biodiversity, biomass, soil composition, etc.).The goal is to understand how people use what theyknow about their environment to sustain or destroy it.

2. Cultural epidemiology. We seek to provide ways ofmapping individual variation and interinformant agree-ment in the flow of ecologically relevant informationwithin and between societies. We use social networkanalysis to trace transmission pathways in the transferof knowledge. We also ask whether cultural orientationaffects how events in nature are interpreted independentof any specific social transmission. The goal is to trackstatistically emerging patterns of cognition and behaviorin human groups to see how different cultures form, in-teract, and evolve in similar environments.

3. The spirit of the commons. We attempt to opera-tionalize the role of “noneconomic” entities and valuessuch as supernatural beings and taboo trade-offs in en-vironmental cognition and behavior. We use projectivevaluation techniques to measure people’s own prefer-ences against others’ preferences, including those ofmembers of other groups, forest spirits, and God. Thegoal is to reveal how different cognitive strategies maycausally contribute to environmental decisions that fa-vor cooperative or selfish use of common-pool resourcessuch as forest species. To foreshadow: nonhuman forestspecies may come to have intentions and act as nego-tiating partners through spirits.

We show that differences in folkecology are tied todifferences in the way people act with regard to common-pool resources among three culturally distinct groupsthat live in northern Guatemala and depend on the samehabitat. This is particularly relevant to issues of inter-national global environmental change, where empiricalresearch showing that culture matters in assessments ofdeforestation and land change is sorely lacking. Envi-ronmental management increasingly involves diversegroups with distinctive views of nature (Arizpe, Paz, andVelazquez 1996). Understanding ways in which local cul-tural boundaries are permeable to the diffusion of rele-vant knowledge may offer clues to success with moreglobal, multicultural commons.

The lowland Maya region faces environmental disasterin part because of the open access to forest resources thatis available to a host of nonnative actors (Schwartz 1995).A central problem concerns differential use of common-pool resources such as forest plants by different culturalgroups exploiting the same habitat. Research on “thetragedy of the commons” indicates that individual cal-culations of rational self-interest collectively lead to abreakdown of a society’s common-resource base unless

institutional mechanisms restrict access to cooperators(Hardin 1968, Berkes et al. 1989). The reason is clear: inthe absence of monitoring and punishment, exploitersgain the same benefits as cooperators but at reduced cost.Cooperators are driven to extinction, and exploitersflourish until the commons is destroyed. Still, economicrationality and institutional constraints on action maynot sufficiently account for differences in environmentalbehavior (Ostrom 1998). For example, people may attachnoneconomic, or “spiritual,” values to resources to pro-tect them. To make better sense of these differences, weexamine links between environmental cognitions andbehaviors.

Another concern is the relation of lowland Maya totheir tropical limestone environment, including anthro-pogenic effects on biodiversity patterning, which con-tinues to puzzle and inspire various disciplines withnotions of how humans best manage their natural sur-roundings. Study of contemporary Maya thought and be-havior informs attempts to understand how this ancientpeople endured (Lundell 1938, Flannery 1982, Rice 1993);however, operationally reliable and replicable data arerare (cf. Nations and Nigh 1980, Schwartz 1985). Previ-ous results on Itza’ focused exclusively on maize pro-duction for a better understanding of the cereal basis forancient Maya civilization (Cowgill 1962, Reina 1967).Now there is increasing consensus that tree tending,multicropping, and varied sorts of landscape modifica-tion may have been critical to the survival of lowlandMaya over the past two millennia of intermittent andcatastrophic upheaval (Turner 1978, Fedick and Ford1990). Our studies provide findings for the further de-velopment of this line of research.

Our case study principally involves three linguisticallydistinct samples from three cultural groups exploitingthe same habitat in the Municipality of San Jose in Gua-temala’s Department of El Peten: native lowland Maya(Itza’), immigrant Maya from the neighboring highlands(Q’eqchi’), and immigrant Spanish-speaking Ladinos (ofmixed European and Amerindian descent). Studies withhighland Q’eqchi’ from Aldea Paapa in the Coban regionof Alta Vera Paz, Guatemala, were designed to comple-ment and inform our main results. The choice of casestudy was motivated, in part, by a further considerationof timeliness: The Itza’ Maya represent the last lowlandMayan-speaking group with demonstrable ties to pre-Columbian civilization in the north-central Peten. Al-though they survived violent conquest and three cen-turies of forced assimilation, they are threatened withimminent demise as a people with a viable linguistic andcultural identity.2 This article, then, also elucidates theimpending loss for us all.

2. In the 1930s the Guatemalan director General Jorge Ubico in-stituted a virulent anti-Maya language policy that led the Lacandon(Lakantun) Maya to flee Peten and resulted in the loss of Itza’ as afirst language in San Jose. Thus, one might argue that any attemptto link contemporary with pre-Columbian Peten Maya culture isunwarranted. Indeed, Hofling (1996:111–12) declares that “thesmall Mayan-speaking populations in the Peten have receivedscholarly attention all out of proportion to their numbers,” reflect-

atran et al . Folkecology, Cultural Epidemiology, and the Commons Spirit F 423

Fig. 1. The Maya Biosphere Reserve, El Peten,Guatemala.

Folkecology

The most reliable research on environmental cognitionstems from ethnobiology, or folkbiology, which revealsuniversal principles that reflect the mind’s ability tobound and essentialize natural discontinuities (Atran1998) and to organize them hierarchically into taxo-nomic kinds (Berlin 1992). Although the use of taxonomyto reason about the biotic world is roughly the same indiverse cultures, this leaves aside important insights intothe behaviorally relevant ways in which people modelthe environment. There are precedents for our approach(e.g., Posey 1983), but to our knowledge this is one ofthe first attempts to show the role of cultural orientationin deforestation and land use in ways meaningful to nat-ural science.

the common setting

In Peten, topographic and microclimatic variation allowfor a dramatic range of vegetation over small areas, andsustaining both this diversity and people’s livelihoodover the past two millennia likely required correspond-ingly flexible agroforestry regimes.3 Paleolimnologicalanalysis of sediments in the central lakes region of Petenassociates the demise of Classic Maya civilization to-ward the end of the 1st millennium with geometricallyincreasing rates of deforestation (Rice 1993). There isevidence of spiraling population growth (Culbert andRice 1990), warfare (Chase and Chase 1989, Demarest1993), and nutrient deficiency (Santley, Killion, and Ly-cett 1986). Economic infrastructure supporting perhaps3 million people collapsed. Transport and communica-tion links between central Peten (e.g., Tikal) and otherproduction centers (e.g., Caracol in Belize) disintegrated.While the central lakes region may have suffered lessdrastic population loss than neighboring Tikal, resettle-ment of Peten seems never to have surpassed the 100,000or so people estimated for the immediate preconquestera.

Dense forest cover reappeared during the Late Post-classic period, which preceded a brutal Spanish conquestin 1697 (Wiseman 1978). By and large, this cover enduredthrough the mid-20th century. Since 1959, when the mil-itary government opened up Peten to “colonization anddevelopment,” more than half of its forest cover has beenrazed and converted to agriculture (Schwartz 1995). Therate of deforestation, which averaged 287 km2 yearly be-

ing in part “exaggerated claims about the uniqueness of Itza’ knowl-edge of the forest environment.” The only “scholarly work” citedis an article by Atran (1993) in this journal, but subsequent studyshows a much wider range of culture-bound knowledge and prac-tice, with greater historical depth, than suggested in previous work(Lopez et al. 1997, Lois 1998, Atran 1999, Atran et al. 1999, Atranand Ucan Ek’ 1999, Atran, Lois, and Ucan Ek’ n.d.).3. At the height of the growing season, July rainfall in Flores (siteof the pre-Columbian Itza’ capital) went from 121 mm in 1993 to335 mm in 1996 and in nearby Tikal from 58 mm to 137 mm; inMay, when crops are planted, there was no rainfall in Tikal in 1993for 23 days, then 130 mm in 3 days, and so on (Guatemala Gov-ernment Institute of Meteorology, INSIVUMEH).

tween 1962 and 1987, nearly doubled to 540 km2 in1988–92 as population rose from 21,000 to over 300,000.Population estimates for 1999–2000 range from 500,000to 700,000 (Nations 1999, Shriar n.d.).

Most of southern Peten’s rain forest has vanished. Ina project engineered by the U.S. Agency for InternationalDevelopment (USAID) and supported by a debt-for-na-ture swap, Guatemala’s government set aside remainingforests north of 17�10� latitude as a Maya Biosphere Re-serve, a designation recognized by UNESCO in 1990. Yet,even within the Biosphere, forest continues to burn(Prensa Libre, April 22, 1998). Deforestation is especiallyprevalent along migration routes into northern Peten(Sader 1999). A new European-financed paved road nowlinks Guatemala City to Flores, virtually ensuring thebreakup of Mesoamerica’s largest remaining contiguoustropical forest. Projections based on remote sensing andground measurements indicate a 14.5% increase in therate of deforestation during 1999–2000 (Grunberg 2000).The major cause of deforestation, though, continues tobe population pressure from the overcrowded and tiredlands of southern Guatemala (Schwartz 1995). There,more than 11 million people live in an area roughly twicethe size of Peten and nearly two-thirds of the land iscontrolled by about 2% of the population.

Each of the three cultural groups with which we areconcerned has founded and predominates in a distinctlocality: Itza’ in the town of San Jose, Ladinos in thenearby settlement of La Nueva San Jose, and Q’eqchi’ inthe hamlet of Corozal. Interviews were in Itza’, Spanish,and Q’eqchi’ respectively. All three groups lie within theMaya Biosphere Reserve’s official “buffer zone” betweenthat latitude and Lake Peten Itza to the south (fig. 1).Here, vegetation is quasi-rain-forest; mean annualtemperature is 25�C and mean annual precipitation

http://www.jstor.org/action/showImage?doi=10.1086/339528&iName=master.img-000.jpg&w=233&h=204


1,600–1,800 mm. In the Reserve and adjacent areas, Itza’make up a majority of the population in only one set-tlement, Q’eqchi’ are a majority in 25 settlements, La-dino immigrants are a majority in 134 settlements, andLadino “Peteneros” (in the area for at least three gen-erations) are a majority in 6 settlements (Grunberg andRamos 1998).

In 1998, San Jose had 1,789 habitants. Most identifythemselves as Itza’, although only a minority speak thenative tongue. Nearly all 625 people in neighboring LaNueva are Ladinos (people of mixed European and Am-erindian descent). Most migrated to the area in the 1970sas nuclear families stemming from various towns ofsouthern Guatemala. Corozal was settled at the sametime by Q’eqchi’-speakers, a highland Maya group. Mostof the 395 inhabitants speak only Q’eqchi’ (not mutuallyintelligible with Itza’). Q’eqchi’ filtered in as nuclearfamilies, migrating in two waves that transplanted partsof highland communities to Corozal: (1) directly fromtowns in the vicinity of Coban (capital of the Departmentof Alta Vera Paz, south of Peten) and (2) indirectly fromAlta Vera Paz via the southern Peten town of San Luis(home to a mixed community of Q’eqchi’ and MopanMaya). Q’eqchi’ immigration into Peten began as earlyas the 18th century, though massive population displace-ment into Peten is recent. The Q’eqchi’ now constitutethe largest identifiable cultural group in Peten whilemaintaining the smallest number of dialects and the larg-est percentage of monolinguals (Wilson 1995:38; cf.Stewart 1980). This reflects the suddenness, magnitude,and relative isolation of the recent Q’eqchi’ migration.4

In all three groups, people pay rent to the municipalityfor a farm plot. Itza’ and Ladinos interact often, as theirvillages are 1 km apart. Q’eqchi live 18 km away, butdaily buses connect them with the other two groups (whoalso farm regularly around Corozal). All groups practiceagriculture and horticulture, hunt game, fish, and extracttimber and nontimber forest products for sale. Eachhousehold (about five persons) has usufruct rights to 30manzanas (21.4 ha) of ejido land (municipal commons).Farmers pay yearly rent of less than a dollar for eachmanzana cleared for swidden plots, known as milpa,whose primary crop is maize. Yearly crop patterns canvary widely, partly because of microclimate and rainfallfluctuation. People can hold plots in scattered areas andcan change plots. Plots from all groups may abut. Hunt-ing is tolerated on neighbors’ plots but not access toanother’s crops or trees.

To ensure maximum social coverage from our sample,our initial informants could not be immediate blood rel-atives (children, grandchildren, parents, grandparents,siblings, first cousins, nieces, nephews, uncles, aunts),immediate affines (spouses, in-laws), or godparents (com-padres). The distributional view of culture that we adoptleads to the use of sampling techniques that are likelyto reveal cultural differences rather than estimating pop-

4. There is evidence for Q’eqchi’ migration in baptism and marriageregisters beginning in 1718 for the Peten towns of Santo Toribio,Dolores, and San Luis (Archivo Apostolica, Flores, Peten).

ulation parameters. We assumed that younger Itza’ Mayamight have notions of biology that differed from thoseof Itza’ elders and that these differences might reflectassimilation to “Western culture.” In addition, becauseItza’ is a dying language and few younger Itza’ speak it,a random sample would have tended to hide rather thanemphasize the differences we were interested in. Insteadof randomly sampling, we restricted our initial sampleto Itza’-speaking Maya as the best representatives of Itza’Maya “culture.” This is not to suggest that there wassome pure Itza’ culture in the past that is now beingdegraded. In fact, Itza’ cultural life is a rich blend of ideasand habits stemming from different inputs, with muchSpanish influence. We assume only that across time and(varying) outside influences the knowledge base differsbetween individuals.

reported agroforestry practice

Although the three groups share a reliance on land andawareness of local species for survival, analyses of athree-year period of milpa practice among 12 to 16 in-formants in each group showed striking differences inthe ways in which the groups utilize land and theirknowledge of species. The data reported here are chieflyself-reports elicited from informants, but long-term spotchecks and subsequent measurements (reported below)confirm a correspondence to actual behavior. Reports ex-hibit no evident bias (e.g., elicited maps of milpa plotsdepict land cleared more accurately and in amounts sys-tematically greater than municipal tax records show).

Analyses of variance were used to reveal group differ-ences between Itza’, Ladinos, and Q’eqchi’, with theScheffe statistic (p ! .05) used for post-hoc comparisons.The following abbreviations are used with the compar-ative statistics: I p Itza’ Maya, L p Ladino, Q p Q’eqchi’Maya; M p milpa (swidden plot), G p guamil (fallowmilpa), R p reserve (secondary forest). Analyses revealedno differences among groups in age, family size, landavailable for cultivation, or per capita income from alltraceable sources.5 Q’eqchi’ produce one set of crops peryear, Itza’ and Ladinos usually two. Q’eqchi’ cut and burnforest for new plots every year, compared with an averageof every 2.3 years for Itza’ and every 1.8 for Ladinos(F[2,41] p 12.92, p ! .001; I, L ! Q). Difference in burnfrequency produces differences in destructiveness inde-pendently of need for income.

Q’eqchi’ clear in a contiguous S-pattern that rollsthrough the forest leaving few trees within or betweenplots, including hill crowns. Ladinos intermittentlyleave trees between and within plots. Itza’ regularly ringplots with trees, clear firebreaks around valuable treesinside plots, and change plots in a noncontiguous pat-tern. This is a strategy apparently shared with somegroups of lowland Lacandon and Yukatek Maya in Mex-

5. Median family income, however, was lower for Q’eqchi’ (US$730)than for Ladinos (US$1,330) or Itza’ (US$1,460). In part, this mayreflect less dependence on public works projects for supplementarywages and greater reliance on sales of surplus maize.


table 1Analysis of Variance of Peten Swidden (Milpa) Practices

N Crops/Year Years of Land Use Hectares Cleared Years FallowSpecies/YearCultivated

Itza’ (R) 16 2 2.3 1.6 4.7 7.8Itza’ (O) 10 – – 2 – 9.7Ladino (R) 16 2 1.8 2.6 3.6 3.3Ladino (O) 10 – – 2.4 – 6.4Q’eqchi’ (R) 12 1 1 4.1 3.3 3.6Q’eqchi’ (O) 10 – – 3.6 – 6.2Other Q’eqchi’∗ – – 1.6 3.7 3.3 2.5

note: R, reported two-year average; O, observed in year 3.∗Average of five Q’eqchi’ settlements (Fagan 2000).

ico, and there is evidence of a pre-Columbian origin (Go-mez-Pompa, Flores, and Sosa 1987, Remmers and DeKoeijer 1992). Itza’ explain it in terms of forest regen-eration: birds such as the chachalaca (ix b’aach p Ortalisvetula) roost in the milpa’s outer ring (t’ool che’) but flyto inner stands (watan che’) to feed on crops and excreteundigested seeds of outlying trees. Left to fallow, areasaround inner stands begin to emulate and bridge withthe outer ring. Birds take undigested seeds of valuableinner-stand trees such as ramon (’ooxp Brosimum ali-castrum) to the outer ring, thereby increasing its valuefor people.

Itza’ differ from Ladinos and Q’eqchi’ in reportedamount of land cleared for cultivation (F[2,41] p 5.45, p! .01, I ! L, Q), fallow length (F[2,41] p 6.982, p ! .002,I ! L, Q), and number of species cultivated (F[2,34] p13.94, p ! .001, I ! L, Q) (table 1). To map these differentpatterns of use onto an overall measure of destructive-ness, we make the strong simplifying assumption thatdestructiveness (D) is an increasing function of land usedper cycle through a plot (L) and rate of cycling througha plot (R). That is, D p L # R. To determine L and Rwe use A (amount of land a farmer clears), Yc (numberof years a cleared plot is used continuously), and Yf (num-ber of years the land is left fallow). From this it isstraightforward to determine that L p A # (Yc � Yf)/Yc and that R p 1/(Yc � Yf). Multiplying these two termsyields the result that D p A/Yc, which is simply landcleared per year. By this measure, Q’eqchi’ destroy morethan five times as much forest and Ladinos less thantwice as much as Itza’.6

6. F(2,41) p 25.04, p ! .0001, D(I) p .753, D(Q) p 1.39, D5(L) p3.92. Our equation oversimplifies the consequences of different pat-terns of use, which involve a trade-off between the costs of farminga plot longer and the benefits of fallowing longer. In theory, thecosts and benefits could be quantified to assess sustainability, butwe have already seen that these groups farm differently. One po-tential limitation of our formula is that a shorter growing periodfor Q’eqchi’ could leave the land in better shape for recovery, butsoil tests (reported below) reveal that nitrogen, a limiting factor inthese calcified soils, is much more abundant in Itza’ fallow landthan in Q’eqchi’ fallow.

biodiversity, forest cover, and soilconditions

Remote sensing confirms extensive deforestation alongQ’eqchi’ migration routes into Peten (Grunberg and Ra-mos 1998, Sader 1999, Grunberg 2000). Reported patternsof crop diversity and awareness of greater ecological com-plexity and reciprocity between animals, plants, and peo-ple should favor regeneration of forest used by Itza’ ver-sus Ladinos. Despite mutual imagined similaritiesbetween Itza’ and Q’eqchi’, on nearly all reported mea-sures Ladinos are closer to Itza’ than Q’eqchi’ are to Itza’.This tendency is reliably confirmed by other measures.

To corroborate cultural behavior patterns, after a two-year lapse we measured for ten new informants fromeach group plot sizes, species diversity, tree counts (min-imum circumference 1 0.3 m at 1–1.5 m from ground),coverage (m2 foliage for each tree crown), and soil com-position (10-cm and 20-cm depths). For each informant,we sampled land held in usufruct in three locations:milpa, guamil (fallow milpa), and reserve (land unculti-vated since initial clearance at the onset of usufruct). Alllocations were sampled after burning, planting, andweeding of a first-year milpa (when maize stalks reached0.5–0.8 m before flowering). Reserve samples were onehectare, and guamil was three years old on average. Ourinitial study suggested that for all group measures rela-tive to forest health and productivity, Itza’ ≥ Ladino ≥Q’eqchi’; hence, we report both two-tailed (Scheffe’s p !

.05) and one-tailed (Fisher’s p ! .05) post-hoc compari-sons, the latter indicating marginal reliability in the pre-dicted direction.7

Again, table 1 shows that Itza’ plant more species (9.7)than Ladinos (6.4) or Q’eqchi’ (6.2) and clear less landyearly (2.0 ha) than Ladinos (2.4 ha) or Q’eqchi’ (3.6 ha);however, an analysis of variance of crop species/ha as afunction of group shows a reliable difference only be-tween Itza’ and Q’eqchi’ (F[2,27] p 3.339, p ! .05). Forall groups, the most frequent crops are maize, then beans,then squash. Overall, Itza’ cultivate 43 different species

7. We normalized highly variable distributions of raw scores witha natural log transformation.


Fig. 2. Biodiversity (number of tree species) and treecover (m2 per hectare) as a function of cultural groupand location type (error bars, 95% confidenceinterval).

in milpas, Ladinos 26, and Q’eqchi’ 23, with a greateryearly species mix for Itza’. We predicted tree diversityto parallel crop diversity as a biodiversity indicator: Itza’averaged 9.0 species/ha, Ladinos 7.2, and Q’eqchi’ 4.4.

Number of tree species was examined with an analysisof variance using group (I p Itza’, L p Ladino, Q pQ’eqchi’) and location (M p milpa, G p guamil, R preserve). Results show effects of group (F[2,81] p 10.48,p ! .0001; I, L ! Q), location (F[2,81] p 171.98, p ! .0001;R 1 M, G), group # location (F[4,81] p 4.45, p p .003;M: I 1 L, Q; G, R: I, L[marginal] 1 Q). As a relative mea-sure of biomass, average tree cover shows the same pat-tern (fig. 2), with effects of group (F[2,81] p 6.17, p p.003; I 1 Q, L[marginal]), location (F[2,81] p 75.08, p !

.0001; R 1 M, G), group # location (F[4,81] p 3.43, p p

.01; M: I[marginal] 1 Q; G: I 1 Q, L[marginal]; R: I 1 Q).For total land cleared (M � G), Itza’ differ reliably fromQ’eqchi’ and Ladinos. Group differences cannot be dueto base-rate differences in species frequency, given theadjacency of parcels across groups.

Soil analysis also suggests that Itza’ agroforestry isleast harmful and most productive. Each soil sample wasrated on a scale of 1 to 22 as a joint function of texture(sandy clay loam ! clay loam ! silty clay loam ! sandyclay ! clay ! silty clay) and structure (small grain ! me-dium grain ! large grain ! small block ! medium block! large block). The best soil ( p 1) is sandy clay loamcomposed of small granular structures that become nei-ther too hard when dry nor too compact when wet toprevent water and root penetration. The worst soil ( p22) is silty clay structured in large blocks which becomerock-hard when dry and extremely compact when wet.This scale reflects the fact that not all possible combi-nations of texture and structure were present. Physicalcharacter of soils was subjected to an analysis of vari-ance: group (I, L, Q), location (M, G, R) # level (1 p 10cm, 2 p 20 cm). Only level proved significant (F[1,162]p 11.37, p p .001; 1 ! 2). There were no reliable between-group differences for any location. Averages for eachgroup across all locations fell within the range of clayswith block structures (I p 14.1, L p 16.9, Q p 14.0).These are able to hold water and fix phosphorus but be-come unworkable and impede root growth during thevery dry and wet spells frequent in Peten. Erosion andlack of tree cover magnify the effect.

All soils are moderately alkaline, with no significantgroup differences in pH or organic matter. Differencesare most apparent for (normalized) measurements ofphosphorus and nitrates. Neither is abundant in the ge-ological materials of limestone regions, and their avail-ability represents limiting factors on life-support sys-tems (Rice 1993). Phosphorus and nitrate levels weresubjected to analyses of variance using group # location# level. Phosphorus showed effects for location (F[2,162]p 25.67, p ! .0001; M 1 G, R), level (F[1,162] p 18.86,p ! .0001; 10 cm 1 20 cm), and group # location (F[4,162]p 3.79, p p .006; M: I, L 1 Q; R: L 1 I). Itza’ differ fromQ’eqchi’ in the upper milpa level (p ! .05), where phos-phorus is most abundant and useful to new plant growth.Overall, Itza’ have the highest milpa and lowest reserve

scores, indicating greater phosphorus storage by plantsin reserve with more available for release in milpa.8

High levels of phosphorus in milpa arise from burning;however, intense heat volatilizes nitrates essential to

8. An analysis of variance was performed on a composite of stan-dardized scores for basic nutrient elements P � (K � Mg � Ca).Calcium is antagonistic to the fixing of potassium and magnesium,so the composite represents a balance of the available nutrient el-ements: phosphorus for root growth, potassium for stem strength,magnesium for photosynthesis, calcium for cell formation. Resultsparalleled those for phosphorus for location (F[2,162] p 15.15, p !

.0001; M 1 G, R), level (F[1,162] p 34.10, p ! .0001; 1 1 2), andgroup # location (F[4,162] p 4.02, p p .004; M: I[marginally] 1 Q;R: L 1 I).

http://www.jstor.org/action/showImage?doi=10.1086/339528&iName=master.img-001.jpg&w=233&h=422


table 2Peten Forest Plants and Animals

Ref. Common Name Scientific Name

PlantsFruit trees

P1∗ ramon Brosimum alicastrumP2∗ chicozapote Manilkara achrasP3∗ ciricote Cordia dodecandraP4∗ allspice Pimenta dioicaP5∗ strangler fig Ficus obtusifolia

F. aureaPalms

P6∗ guano Sabal mauritiiformeP7∗ broom palm Crysophilia

stauracanthaP8∗ corozo Orbignya cohune

Scheelea lundelliiP9 xate Chamaedorea elegans

C. erumpensC. oblongata

P10 pacaya Chamaedoreatepejilote

P11 chapay Astrocaryummexicanum

Grasses/herbsP12 herb/underbrush (various families)P13 grasses Cyperaceae/Poaceae

Other plantsP14∗ mahogany Swietenia

macrophyllaP15∗ cedar Cedrela mexicanaP16∗ ceiba Ceiba pentandraP17∗ madrial Gliricidia sepiumP18∗ chaltekok Caesalpinia velutinaP19∗ manchich Lonchocarpus

castilloiP20∗ jabin Piscidia piscipulaP21∗ Santa Maria Calophyllum

brasilenseP22∗ amapola Pseudobombax

ellipticumBernoullia flammea

P23∗ yaxnik Vitex gaumeriP24∗ kanlol Senna racemosaP25∗ pukte Bucida bucerasP26 water vine Vitis tilaefoliaP27 cordage vine Cnestidium rufescensP28 killer vines (various epiphytes)

AnimalsArboreal

A1 bat ChiropteraA2 spider monkey Ateles geoffroyiA3 howler monkey Alouatta pigra

A. palliataA4 kinkajou Potos flavusA5 coatimundi Nasua naricaA6 squirrel Sciurius deppei

S. aureogasterBirds

A7 crested guan Penelopepurpurascens

A8 great curassow Crax rubraA9 ocellated turkey Meleagris ocellataA10 tinamou Tinamou major

Crypturellus sp.A11 toucan Ramphastos

sulfuratusA12 parrot Psittacidae in partA13 scarlet macaw Ara macaoA14 chachalaca Ortalis vetulaA15 pigeon/dove Columbidae

leaf formation. Thus, higher phosphorus levels shouldbe correlated with lower nitrate levels (and perhaps lessfoliage cover in the long run), but for Itza’ and, to a lesserextent, Ladinos the reverse is true. Nitrate levels showeffects of group (F[2,162] p 11.42, p ! .0001; I[marginally]1 L, Q), location (F[2,162] p 6.44, p p .002; M 1 G), andgroup # location (F[4,162] p 2.87, p p .02; M: I, L 1 Q;G: I 1 L, Q). For total land cleared (M � G), Itza’ differmarginally from Ladinos and significantly from Q’eqchi’.Interrelated factors allow Itza’ to enjoy relatively highphosphorus and nitrate levels. Itza’ cultivate more va-rieties of nitrogen-fixing pole beans that climb maizestalks than Q’eqchi’ or Ladinos. Q’eqchi’ and Ladinosweed only once shortly after planting; Itza’ weed a secondtime before maize has flowered and leave the weeds asmulch. Intense rainfall at this time favors bacterial de-composition of mulch, which releases nitrogen (alsophosphorus, potassium, and magnesium). Finally, Itza’tend to light smaller and more dispersed fires to clearland and to protect valuable trees with firebreaks 2 m inwidth. (A side effect is that the less intense heat causesless volatilization of nitrogen.)

In sum, physical measurements generally corroboratereported behaviors and track their consequences, indi-cating that Itza’ practices encourage a better balance be-tween human productivity and forest maintenance thando immigrant practices. However, significant differencesin immigrant practices reveal that immigrant Spanish-speakers are measurably closer in behavior to nativeMaya than are immigrant Maya. Studies of milpa prac-tices in other immigrant Q’eqchi’ communities in Petenconfirm the patterns in our study (Fagan 2000).

mental models

To determine whether group differences in behavior arereflected in distinct cognitive patterns, we elicited folk-ecological models. In preliminary studies, we asked in-formants “Which kinds of plants and animals are mostnecessary for the forest to live?” From these lists wecompiled a set of 28 plants and 29 animals most fre-quently cited across informants (plant vouchers depos-ited at the University of Michigan Herbarium).9 The 28plant kinds in the study include 20 trees and 1 ligneousvine counted among the species in the preceding study(starred in table 2). Although these 21 species representonly 17% of the species enumerated, they account for44% of all trees in Itza’ parcels, 50% in Ladino parcels,

9. For plant stimuli we compiled a list of the 26 generic speciesmost frequently mentioned by informants in earlier studies. Ge-neric species (also called folkgenera) are basic building blocks offolkbiological taxonomies everywhere (Atran 1998). The genericspecies mentioned most were all trees, vines, or palms. We added2 often-mentioned life forms: herbs/underbrush and grasses. Lifeforms are superordinate folktaxa that contain many generic species.For animals, the overlap among all our populations warranted useof the same 28 names in each case. In addition to the 28 animalsgenerated by the task, we introduced the bat as a 29th animal be-cause of its special status within Maya taxonomy. Colored drawingswere taken from field guides.


RummagersA16 collared peccary Tayassu tacajuA17 white-lipped peccary T. pecariA18 paca Cuniculus pacaA19 agouti Dasyprocta punctataA20 red brocket deer Mazama americanaA21 white-tailed deer Odocoileus

virginianusA22 tapir Tapirus bairdiiA23 armadillo Dasypus

novemcinctusPredators

A24 jaguar Felis oncaA25 margay Felis wiediiA26 mountain lion Felis concolorA27 boa Boa constrictorA28 fer-de-lance Bothrops asperA29 laughing falcon Herpetotheres

cachinnans

∗Species counted in tree-frequency study.

and 54% in Q’eqchi’ parcels. This confirms the salienceof the species selected for the folkecological study.

How plants affect animals. Instructions and responseswere given in Itza’, Spanish, or Q’eqchi. Equal numbersof informants (six men and six women in each com-munity) were asked first to explain how each planthelped or hurt each animal. The task consisted of 28probes, one for each plant. On each trial, all animal pic-ture cards were laid out and the informant was asked ifany of the animals “searched for,” “went with,” or “werecompanions of” the target plant and whether the planthelped or hurt the animal. Questions were pretested forsimplicity and easy applicability across cultures. Unaf-filiated animals were set aside. For each animal, inform-ants were asked to explain how the plant affected theanimal, and their explanations were recorded. We usedprincipal-components analysis to determine if a singleunderlying model of ecological relations held for all in-formants in a population. Analysis was done on each ofthree 12 # 12 subject-by-subject matrices. Each matrixwas adjusted for guessing. Consensus was assumed if (1)the first eigenvalue was notably larger than the secondand accounted for most of the variance and (2) the firsteigenvector was all positive. Under these conditions, theagreement pattern among informants should reflect acommon model, and first-factor scores provide indicesof the degree to which individuals’ responses reflect theconsensus (Romney, Weller, and Batchelder 1986). Eachperson’s first-factor score, then, may be considered ameasure of that person’s individual cultural “compe-tence” with respect to the aggregated cultural “model.”To establish consensus, all tasks involved a minimumof 12 participants from each group.10 Finding consensusjustifies further study of groupwide patterns.11

10. With certain formal qualifications (cf. Lopez et al. 1997:263),our analysis is based on the culture consensus model of Romney,Weller, and Batchelder (1986). When the three conditions above aremet, a sample as small as 12 informants with average competenceof 0.5 is enough to signal consensus.11. Pairwise agreement was calculated between all informants sep-arately for plant-animal and animal-plant relations. All animal-

Analyses of residual agreement were used to revealdifferences among groups (Nakao and Romney 1984,Medin et al. 1997).12 To compute residual agreement, wefirst calculated the agreement predicted by the model forany pair of actors by multiplying their first-factor scores.Residual agreement was then correlated with observedagreement. Systematic residual agreement suggests dis-tinct group patterns beyond overall consensus. If residualagreement is correlated with observed agreement, thenpatterns of observed agreement among informants occurbeyond what is explained by participation in the con-sensus (cf. Boster 1986).13 If not, participation in the con-sensus basically accounts for the entire pattern of ob-served agreement. If a single model fits all individuals,there should be only chance residual agreement.

Two results are apparent (fig. 3): Itza’ and Ladinosshowed very similar patterns of relations, and Q’eqchi’perceived many fewer and those tended to be a subsetof those seen for the other two groups. The overwhelm-ing majority of interactions within each group involvedplants helping animals by providing them food.14 Plants’providing shelter to animals was also a common re-sponse. An analysis of variance for plants helping ani-mals shows Q’eqchi’ reporting on average many fewerrelations (46.8) than either Ladinos (163.2) or Itza’ (187.5),who did not differ from each other (F[2,33] p 23.10, p !

.001). Itza’ and Ladinos showed a large overlap for whichplants helped which animals (r [I, L] p .82 versus r [I,Q] p .42 and r [L, Q] p .54). This picture is supportedby consensus and residual analyses.

Using agreement adjusted for guessing as the depen-dent variable, a large cross-group consensus emerged (thefirst eigenvalue, 23.98, was 12.3 times the second andexplained 67% of the variance). Moreover, consensusscores for Q’eqchi’ (mean competence p .89) were reli-ably higher than for Itza’ (.79) or Ladinos (.85), who didnot differ from each other (F[2,33] p 8.82, p ! .001,Scheffe post-hoc p’s ! .05). Even if a majority of Ladinosand Itza’ agreed that a plant helped an animal, often all

plant pairs for which either informant did not know one or bothspecies were excluded. If both gave the same response (either pos-itive, negative, or no relation), they were counted as agreeing; ifthey reported different relations or if one reported a relation andthe other none, they were scored as disagreeing. Each pair of in-formants was given a score ranging from 0 to 1.0 representing theproportion of species pairs on which they agreed. Observed agree-ment was adjusted for guessing: adjusted agreement p ([observedagreement # number of possible responses] � 1)/(number of pos-sible responses � 1).12. Multiple regressions were performed within and between allthree groups using age, gender, and residence (for Itza’, residence pinformant age) as independent variables against first-factor consen-sus scores. There were no significant differences within or betweenpopulations (though gender together with judgments of expertisewas significant for Ladinos [see below]).13. Boster’s (1986) method of determining residual agreement mo-tivated our initial interest in this method; however, his methoddoes not work if the items are not all of equal difficulty. Accord-ingly, we use within- versus between-group residual agreement asour measure.14. For nearly all respondents, a given plant helps a given animalor does nothing for it. Only two Itza’ said of one epiphyte that itharmed animals by killing the plants those animals feed upon.


Fig. 3. Frequency of reports of effects of plants on animals for Itza’, Ladinos, and Q’eqchi’. Plant and animalnumbers refer to the ordering of species in table 2. The height of each point reflects the proportion of inform-ants reporting such interaction.

Q’eqchi’ reported no effect, making the modal answerno effect. Thus, Q’eqchi’ responses drove the overall con-sensus. Given this situation, residual analyses are moreeffective in revealing cultural models than simple mea-sures of interinformant agreement.

We analyzed a 3 # 36 residual agreement matrix. Foreach of 36 informants (12 in each group) there were threemeasures: average residual agreement of that informantwith members of the same group and that informant’saverage residual agreement with members of each of theother two groups. Within-group agreement was reliable:for each group, F[2,22] 1 23, p ! .001. Itza’ and Q’eqchi’had greater withinthan between-group residual agree-ment. Ladinos showed higher withinthan between-group residual agreement vis-a-vis Q’eqchi’ but did notshare more residual agreement with one another thanwith Itza’.15 This indicates that the Ladino model was aversion of the Itza’ model. One distinction between Itza’and Ladinos was the latter’s tendency to generalize thebeneficial effect on animals of economically and cultur-ally important plants such as mahogany (the prime woodexport) and ceiba (Guatemala’s national tree) withoutapparent justification. Overall, Ladino and Itza’ modelsconverged on how plants helped animals; the Q’eqchi’model was a severely limited subset.

How animals affect plants. Reports of how animalsaffect plants yielded larger differences (fig. 4). Q’eqchi’signaled too few interactions for consensus analysis (only13 interactions of 812 possible animal-plant pairings foreach of 12 participants). Itza’ and Ladinos showed strongcross-group consensus (ratio eigenvalue 1:2 p 18.9, var-iance p 72%) but also greater residual agreement withinthan between groups (t[11] 1 4.5, p ! .0001). Negativereports of animals hurting plants occurred with equalfrequency (8.0% of cases by Itza’, 8.2% by Ladinos), but

15. Results of t-tests were: for Itza’ t(I, L) p 6.71, t(I, Q) p 8.88;for Q’eqchi’ t(Q, L) p 16.7, t(Q, I) p 20.9; for Ladinos t(L, Q) p4.38, t(L, I) p n.s.

Itza’ were 4 times more likely to report positive inter-actions (F[2,33] p 3.74, p ! .05) and 3.4 times more likelyto report reciprocal relations (a plant and animal helpingeach other) (t[22] p 3.31, p ! .005).

Itza’ reported that classes of animals (arboreal, bird,rummager, predator) differentially affected classes ofplants (fruit, grass/herb, palm, other), whereas Ladinosreported more universal effects.16 Arboreals were re-ported to be much more likely to interact with fruit treesthan with other plant groups; birds were reported to bemost likely to interact with fruit trees but also to havemoderate levels of interactions with palms; rummagerswere reported to interact primarily with grasses/herbsand to a lesser extent with fruit trees; predators werereported to have few if any interactions with plants. Ananalysis of variance reveals a plant-by-animal interactionfor Itza’ but not for Ladinos (F[9,99] p 26.04, p ! .0001).The absolute level of interactions was much lower forLadinos, who reported that all animal groups (save pred-ators) interacted with all plant groups in roughly thesame ways. Animals reported to be most likely to affectplants were rummagers, birds, and arboreals, and theplants most likely to be affected were said to be fruittrees and “other” plants.

On a qualitative level, although both groups acknowl-edged that animals had a large impact on fruit trees, Itza’

16. Participants were given two scores for each pairing of animaland plant groups, reflecting the proportion of possible positive andnegative interactions acknowledged. A score of .25 for negativearboreal-fruit interactions indicates that the participant identifiednegative interactions between one-quarter of all possible pairingsof arboreal animals and fruiting plants. Scores were entered into 2(type of interaction: positive, negative) # 4 (animal group: bird,rummager, arboreal, predator) # 4 (plant: fruit, grass/herb, palm,other) analyses of variance. Thus, tests of plants had 3,33 degreesof freedom, as did tests of animals, and tests of plants by animalshad 9,99 degrees of freedom. Ladinos showed main effects of in-teraction type (F[1,11] p 6.95, p ! .05), plant (F[3,33] p 9.89, p !

.0001), and animal (F[3,33] p 14.40, p ! .0001) but no animal byplant interaction.

http://www.jstor.org/action/showImage?doi=10.1086/339528&iName=master.img-002.png&w=430&h=149


Fig. 4. Frequency of reports of effects of animals on plants for Itza’ and Ladinos. Animal and plant numbersrefer to the ordering of species listed in table 2. The height of each point reflects the proportion of informantsreporting such interaction.

differed from Ladinos in understanding these relations.Ladinos inferred that animals harmed plants by eatingfruit. Itza’ had a subtler view based on the properties ofthe seed and on how the animal chews and digests. Ifthe seed is soft and the animal crunches through the fruitcasing, the interaction is harmful because the animal islikely to destroy the seed; but if the seed is hard anddigestion is rapid, the interaction is likely to be helpfulif the seed passes through the animal’s body, for the an-imal assists seed dispersal and fertilization.

The picture thus far suggests common models withdistinct variations for each group. On plants helping an-imals, Itza’ and Ladinos had similar models with over80% overlap on pairwise interactions. The Q’eqchi’model was much less elaborated, being a proper subsetwith less than one-sixth of the relations reported by theother groups. An examination of how animals are re-ported to affect plants further reveals the paucity of theQ’eqchi’ folkecological model. Q’eqchi’ reached a non-zero consensus on only 10 out of 812 possible relations.These findings suggest a complex Itza’ folkecologicalmodel of the forest wherein different animals affect dif-ferent plants and relations among plants and animals are

reciprocal. Ladinos also possess a relatively elaboratemodel, but relations were more unidirectional and lessspecific. Q’eqchi’ acknowledge a much reduced role ofplants and almost no role of animals in the folkecologyof the forest.17

Bidirectional plant-animal relations. Analyses wereperformed on pairwise relations among plant and animalspecies reported by the three groups. For each plant-an-imal pair, an informant could report that the plant eitherhelped, hurt, or had no relation to the animal and viceversa. Although responses were coded as to the specificnature of the helping or hurting relation, for these anal-yses those differences were glossed over and relationswere scored only as helping or hurting. Thus, the bi-directional relation between each plant and animal couldbe classified as one of the following: “mutualist” ( � 1,� 1), “commensalist” ( � 1, 0), “parasitic” ( � 1,�1),

17. These variations represent interactions, not general differencesin response thresholds. Thus, Ladinos responded at the same rateas Itza’ for plant-animal relations and for negative animal-plantrelations but reported dramatically fewer positive animal-plant re-lations. Q’eqchi also showed an interaction. Difference in knowl-edge is therefore the most parsimonious description of our results.



“destructive” (�1, 0 or �1,�1) or “neutral” (0, 0). Itza’recognized reliably more mutualist relations than didLadinos; Q’eqchi’ recognized almost none.

Relations between people and plants. For the past fewmillennia Peten forest ecology has been intensely an-thropogenic, and so we might expect awareness of thehuman role to be critical to long-term forest mainte-nance. To explore perceived interactions among peopleand plants, we asked each informant whether people intheir community helped (�1), hurt (�1), or did not affect(0) each item on the plant list and vice versa, therebyassessing actual practice rather than ideal behavior. Wealso asked informants to explain their responses. Nooverall consensus emerged, but each population showedwithin-group consensus (ratio eigenvalue 1:2 1 3, vari-ance accounted for 1 50%). Analyses of variance confirmthat each group had a distinct pattern (F[2,33] p 5.92, p! .01). Itza’ reported more instances of humans’ affectingplants than Ladinos, and both groups reported manymore than Q’eqchi’ (F[2,33] p 157.37, p ! .0001). Onaverage, Itza’ reported helping over twice as many plants(18.7) as they hurt (7.1), Ladinos reported that they helped(10.8) and hurt (10.2) in equal numbers, and Q’eqchi’reported that they hurt (3.4) over three times as manyplants as they helped (1.0). Finally, Itza’ and Ladinos to-gether showed consensus.

Explanations of responses regarding how plants helppeople focused on use. Response patterns for Itza’ andLadinos were similar; Q’eqchi’ provided considerablyfewer uses. Even so, at least two-thirds of the Q’eqchi’informants mentioned a use for the overwhelming ma-jority of plants.18 The rank order of primary use cate-gories for each population reveals the priorities in eachpopulation’s mental model of utility. The major differ-ence between Itza’ and Ladinos concerns the relative pri-orities of artisanry and firewood. Itza’ were loath to con-sider the use of ecologically or economically importanttrees for firewood if the wood could be used for otherpurposes. Although all three populations considerednourishment among the top priorities, only the Q’eqchi’considered it primary. Even more significant, whereasQ’eqchi’ also considered cash value a high priority, Itza’and Ladinos assigned cash value relatively low priority.

To assess reported human impact, we tallied individ-ual responses to whether their community helped, hurt,or did not affect a given plant and computed each group’smean response to each plant. Each “impact signature”ranged from entirely beneficial (�1.00), through neutral(0) to entirely harmful (�1.00). Impact signatures for Itza’and Ladinos were moderately correlated (r p .65, p !

.001), suggesting somewhat similar views of how hu-mans affect plants. Signatures for Q’eqchi’ were nega-

18. Itza’ gave 577 positive and 14 negative responses, ranging from41 to 57 responses per informant (modal response of 2 uses perplant). Ladinos gave 562 positive and 2 negative responses, with arange of 40 to 58 per informant (modal response of 2). Q’eqchi’ gave307 positive and 2 negative responses, ranging from 21 to 35 perinformant (modal response of 1). Plants not given uses by at leasta third of the Q’eqchi’ were grasses, the pukte tree, and the stranglerfig.

tively correlated with those of Itza’ (r p �.28) and La-dinos (r p �.16), suggesting a very different model ofhuman effects on plants. Itza’ reported beneficial impacton all ecologically and economically important plantsand absolute commitment to protect ramon and chicle(Manilkara achras). Itza’ call ramon “the milpa of theanimals” because many bird and mammal species feedon its fruits and leaves (Atran 1993). The chicle tree isalso visited often by animals and, as does ramon, has along history of local use. Extraction of chicle latex forchewing gum has been Peten’s prime cash source for acentury. Itza’ report variable impact on herbaceous un-dergrowth, strangler figs (Ficus sp., which nourish manyanimals but kill other trees), and yaxnik (Vitex gaumeri),which they characterize as a marginally useful “forestweed.” Itza’ report harmful impact on pukte (Bucidabuceras), another “forest weed,” on kanlol (Senna ra-cemosa), a “village weed,” and on vines cut for waterand cordage.

Ladinos also reported a highly positive impact for val-uable plants (including Ceiba pentandra). For palms theyreported a positive impact only for those used for thatch(corozo palm fruits are also sold to a local nongovern-mental organization [NGO]). For most plants they re-ported variable impact. Q’eqchi’ reported a positive im-pact only for thatch palms and a negative impact forPeten’s most important cash sources: chicle, tropical ce-dar (Cedrela mexicana), mahogany (Swietania macro-phylla), and xate (decorative Chamaedorea dwarf palmscollected for export).

In sum, only Itza’ had a globally positive view of hu-mans’ impact upon the plants judged most necessary forthe forest to live. Informant justifications of responsesrevealed that Itza’ provided ecological reasons for pro-tecting economically important plants. Both immigrantgroups judged human impact on the largest number ofplants to be variable, but only the Q’eqchi’ saw the hu-man impact on the economically most important plantsas decidedly harmful and costly to the forest. Overall,Q’eqchi’ said that they had relatively little impact onplants, a striking observation given that this group ex-hibited the most destructive agroforestry. This differencedid not arise from unfamiliarity, because they mentioneduses for nearly all plants.

Human impact and ecological centrality. Regressionswere performed to clarify relations between use, ecolog-ical centrality, and reported human impact on plants. Wedefined ecological centrality for each plant as the pro-portion of plant-animal associations in a group’s con-sensual ecological model and used two measures of use:(1) wood, shelter, and cash combined and (2) cash alone.For Q’eqchi’, none of these variables predicted impactsignature, and (nonsignificant) correlations were alwaysnegative. For Ladinos, mean number of cash uses was areliable predictor of reported impact (r2 p .34, F[2,25] p6.55, p ! .01), but neither plant-animal nor animal-plantassociations were reliable. The correlation between cashuse and impact was positive, indicating that Ladinos pro-tect plants with cash value. For Itza’, the combined r2 onplant-animal associations was .44 (F[2,25] p 9.13, p !


.001) and both wood, shelter, and cash (p ! .01) and ec-ological centrality (p ! .01) predicted impact signature.For animal-plant relations the values were r2 p .36, withwood, shelter, and cash (p p .02) and centrality (p ! .01)both reliable. Ecological importance and combined util-ity—not cash value alone—predicted which plants Itza’seek to protect. Only Itza’ saw people as generally ben-efiting plants that benefit animals.

Finally, regression analysis reveals that, for Itza’, weedstatus (i.e., whether a plant is considered a weed) andratings of human impact actually predicted (normalized)frequencies of trees observed in informant parcels (r2 p.46, F[2,20] p 7.58, p p .004; both predictors, p ! .01).No such relation emerged for Ladinos or Q’eqchi’. Ra-mon, which is most common to Itza’ parcels, exemplifiesthis tendency (2.6 times more counts than for Ladinos,4.2 more than for Q’eqchi’).

Human-animal relations. Focusing on the role of hu-mans in Itza’ and Ladino folkecology, we did a follow-up study of interactions among animals and people (with12 new informants for each group). Itza’ and Ladinosshared consensus on numbers and kinds of negative an-imal-human interactions (ratio eigenvalue 1:2 p 3.3, var-iance p 45%), based mainly on animal damage to milpacrops; however, Itza’ reported more positive animal-hu-man interactions (F[1,112] p 98.38, p ! .001), based onuse of animals (e.g., in medicine) and their role in forestregeneration (e.g., optimizing seed distributions of val-uable trees). This is the same pattern as in the animal-plant interaction study. Correlations (p ! .05) betweenthe ways animals help plants and the ways humans helpanimals were also positive for Itza’ (r p .40) but negativefor Ladinos (r p �.50).

In summary, Itza’ show awareness of ecological com-plexity and reciprocity between animals, plants, and peo-ple, and Itza’ agroforestry favors forest regeneration.Q’eqchi’ acknowledge few ecological dependencies, andQ’eqchi’ agriculture is insensitive to forest survival. La-dino folkecology and practice are intermediate. Itza’agroforestry thought and practice encourage a potentiallysustainable balance between human productivity andforest maintenance.

Cultural Epidemiology

To examine how ecological models and practices werelearned, we used social network analysis to answer ques-tions about within- and between-group consensus. Webegan with the informants from the plant-animal study(12 in each group). Previous literature indicated that fourto eight intimate ties are readily elicited from people(Wellman 1979, Hammer 1983). We asked each inform-ant to name, in order of priority, the seven persons out-side the household “most important to your life” and tojustify inclusion of these names in the informant’s com-munity of social network. Later, we asked each to nameby priority seven persons “to whom you would turn tofind out something that you do not understand about theforest” and to justify inclusion of names in the inform-

ant’s expert network. For each informant we coded age,gender, occupation, relation to subject (friend, work-mate, neighbor, relative, etc.), ethnicity, and frequencyof contact with the person named (from the subject’sstandpoint). We used a “snowball” method to elicit so-cial and expert networks from the first and last personsnamed in each original informant’s social network, asdirect ties involving one intermediary often suffice toestablish the networks of close and extended ties thatregulate information flow within a community (Free-man, White, and Romney 1992).19

social and expert networks

In their social networks, Itza’ named no one outside theethnic community, Q’eqchi’ named 1 Ladino, and La-dinos named 1 Itza’. This confirms our sample’s ethnichomogeneity. Overall social network density (Dh p ratioof possible to actual names) was substantially greater forQ’eqchi’ (4.6) than for Ladinos (2.4) or Itza’ (1.94). Thiswas also true for measures of group centrality: Q’eqchi’named 18 people more than 15% of the time, Ladinos14, and Itza’ 2. The same pattern emerged from analysesof interconnectedness, or l-level (Wasserman and Faust1994). The l-level refers to the average person over thegroup and indicates the number of actual links that haveto be severed to disconnect a given person from all otherpersons in the group. Among the Q’eqchi’, actors namedin the social networks were connected at l p 4, Ladinosat l p 2, and Itza’ at l p 1. Level 5 (l p 5) includes90% of the Q’eqchi’, 21% of the Ladinos, and only 10%of the Itza’.

By contrast, Q’eqchi’ had the lowest agreement on whothe forest experts were and Itza’ the highest. This is sur-prising given that the Q’eqchi’ community is the small-est; one might expect higher agreement if only for lackof choices. Q’eqchi’ named 4 experts at least 15% of thetime, Ladinos 8, and Itza’ 12. The 2 “experts” cited mostby Q’eqchi’ (60%) were a Washington-based NGO andthe Guatemalan agency responsible for the Maya Bio-sphere. Itza’ named only Itza’. For Ladinos, 3 of the 4most cited experts were also the 3 named most by Itza’.We elicited expert networks for 6 of the top 10 Ladinoexperts, and these 6 cited Itza’ as their experts over La-dinos by a ratio of 6:1. Overlap between social and expertnetworks was also greatest for Itza’ and least for Q’eqchi’.For Itza’, 14 well-cited (chosen three or more times) so-cial partners were among the 22 well-cited forest experts.For Q’eqchi’, only 6 well-cited social partners wereamong the top 18 experts (all cited much less often thanoutside institutions). For Ladinos, 11 well-cited socialpartners were among the top 25 experts, and the 3 topLadino experts were also among the 6 most socially in-terconnected Ladinos (l p 5). Of 43 named Ladino ex-perts, 42 were men (men were also almost exclusivelythe experts for the other groups).

For Ladinos, strong overlap between socially con-

19. If the first or last person on a list could not be interviewed, wewent to the second or sixth, etc.


Fig. 5. Social networks for Itza’, Ladinos, and immigrant Q’eqchi’. Circle graphs (top) and multidimensionalscalings (bottom) are alternative representations of the same data sets.

nected individuals and Ladino experts (who themselvesnamed Itza’ as experts) intimates a network of reliablebut noninstitutionalized ties for learning about the forestfrom Itza’. Patterns of residual variance on the plant-animal task further indicate learning. As noted earlier,though Ladinos showed higher withinthan between-group agreement vis-a-vis Q’eqchi’, they did not sharemore residual variance with one another than with Itza’.Whatever residual model of plant-animal relations La-dinos shared with one another beyond the overall con-sensus model they also shared with Itza’.

Thus, the three populations markedly differ in theirsocial and expert network structures, with different con-sequences for the flow of information about the forest.The Q’eqchi’ networks suggest that information perti-nent to long-term survival of the forest comes from out-side organizations with little long-term experience in Pe-ten. What outside information there is seems unlikelyto penetrate deeply into the Q’eqchi’ community be-cause it is not conveyed by socially relevant actors (al-though we need further evidence to show that this isfact). Radical dissociation of the Q’eqchi’ expert networkfrom a densely interconnected social network suggeststhat little outside information pertaining to ecologicalcomplexity or forest expertise is being assimilated by the

Q’eqchi’ community. For Itza’, expert information aboutthe forest appears integrally bound to intimate patternsof social life as well as to an experiential history traceableover many generations. For Ladinos, expert informationis also likely to be assimilated into the community. Butthis expert information comes primarily from Itza’ ex-perts to Ladino experts, with the Ladino experts, in turn,selectively channeling information to the wider Ladinocommunity.

pathways of knowledge transmission

Visual representations of the social network analyses (fig.5) show for the Q’eqchi’ a dense, highly interconnectednetwork, with no dominant individual or subgroup. Thisredundant social structure favors communal and cere-monial institutions that organize accountability, andthese are manifestly richer among Q’eqchi’ than amongItza’ or Ladinos. Only Q’eqchi’ practice agroforestry incorporate groups: neighbors and kin clear and burn eachhousehold’s plot, kin groups seed together, and the com-munity sanctions unwarranted access of family standsof copal trees (Protium copal), whose resin is rituallyburned to ensure the harvest. This implies that insti-tutional monitoring of access to resources, cooperating



kin, commensal obligations, a vibrant indigenous lan-guage, and knowledge of the land (including recognitionof important species) may not suffice to avoid ruin ofcommon-pool resources. For the Q’eqchi’ of Corozal,continued corporate and ceremonial ties to the sacredmountain valleys of the Q’eqchi’ highlands do not implya corresponding respect for lowland ecology. A relativelyclosed corporate structure that channels information fo-cused on internal needs and distant places may functionto impede access to ecological information relevant tocommons survival.

Both images of the Itza’ network indicate that node Yis the best-connected individual. This person is also citedas the top Itza’ forest expert. His expertise is indepen-dently confirmed. For example, in a study of tropical birdclassification involving American birdwatchers andItza’, Y scored highest among Itza’ on measures of cor-respondence with scientific (classical evolutionary) tax-onomy (Bailenson et al. n.d.). The multidimensional scal-ing suggests that the Itza’ community is currentlydivided into two social factions: one dominated by Y andthe other by V and W. These factions are also political:Y is founder and chief executive of the Bio-Itza Associ-ation, and V is the current mayor of San Jose. W is V’sfather, and he is also cited as one of the top three Itza’forest experts. Y and V-W head two families that havecontinuous genealogical links to preconquest Itza’ clansof the same name (although there are also numerousintervening links with Yukatek Maya).20

One possibility consistent with this structure is thatecological knowledge is directly transmitted from so-cially well-connected forest experts such as Y. To eval-uate the latter possibility we analyzed patterns of resid-ual agreement in relation to social and expert networkstructure. We focused on nonempty cells (as determinedby our most cited expert Itza’ informant, Y), becauseknowledge transmission should primarily take the formof noting an existing relation, not the absence of rela-

20. The original mission, or reduccion, of San Joseph, founded in1702, was located between present-day Santa Elena and San Benito,opposite the former Itza’ island capital of Noj Peten (present-dayFlores). In the wake of the chaos and rebellion of the early postcon-quest years, a new Itza’ reduccion of San Joseph arose at the presentsite of San Jose sometime before 1750 (Jones 1998). The earliestmarriage record from San Jose dates from that year (5 mayo 1750,Libro de Casamientos de la Parochia de los Pueblos de San Andres,San Joseph y San Geronimo, Ano de 1751, Archivo Apostolica,Flores, Peten). Extant (but partial) records of marriages and baptismsbetween 1751 and 1788 reveal the surnames mentioned most to bepreconquest Itza’ (and allied Mopan) patronyms. In descending orderof frequency they are (in original spelling) Tun (12), Chayax (11),Canek (9), Tz’in (7), Chabin (7), Kinyocte (7), Cuouh (6), Chata (6),Tut (6), Quixoban (5), Xiquen (5), Citcan (5), Cante (4), Chan (4),Puc (4), Kanchan (4), Tzuntecun (4), and Tesucun (4). Our socialnetwork analysis of contemporary Itza’ includes 32 surnames: 8Spanish (Ramos, Lopez, Dıaz, Cortez, Lines, Garcıa, Morente, Cin-turon), 8 those of families that came from Yucatan between 1750and 1900 (Huex, Colli, Vitzil, Mex, Panti, Tz’ul, Mis, Yej), and 16preconquest names from Itza’-ruled territory (Chayax, Cohouj,Chan, Suntecun, Zacal, Tesucun, Zac, Cauich, Ek, Tut, Xiken, Ba-tab, Cante, Chata, Quixchan, Chuc). The genealogies of these fam-ilies date, in part, to the founding of San Jose and therefore, verylikely, to preconquest Peten Itza’.

tions. Analyses revealed little residual agreement, andthis agreement was inconsistent across different tasks(e.g., r2 p .02 between residuals for positive plant-animaland positive animal-plant relations, r2 p .15 for positiveplant-animal and negative animal-plant relations, r2 p.03 for positive and negative animal-plant relations). Inno case could we discern relationships between residualagreement and social or expert network proximity.

There is an alternative scenario to learning about theforest that is more consistent with independent discov-ery than direct social transmission of ecological knowl-edge. When asked how they learn, Itza’ acknowledge con-sulting experts on hard problems but mostly claim toacquire knowledge elicited in our tasks by “walkingalone” in the forest they call “the Maya House.” ForItza’, diffusely interconnected social and expert networkssuggest multiple pathways for persons to gain and for thecommunity to assimilate and store information aboutthe forest. Cultural stories, values, and the like bias theinterpretation of experience in different ways: for ex-ample, a bird or monkey eating fruit is perceived to beharmful by Q’eqchi’ and Ladinos but can be inferred byItza’ to be helpful. Although culturally channeled in thisway, Itza’ knowledge of specific plant-animal interac-tions is acquired through individual experience andexploration.

The circle graph of the Ladino network shows a cleargender division of the community: C1–R are women,A–Q are men. At the center of the graph is D1, the mayorof San Jose (i.e., the same person as V in the Itza’ net-work). Both the circle graph and the multidimensionalscaling point to I as the best-connected individual. He isalso cited most as the top Ladino forest expert. In fact itwas I who first received permission for a Ladino settle-ment in the area from Y in the Itza’ network (when Ywas mayor of San Jose).

Because Ladino experts are socially well-connected, in-formation that may come through Itza’ experts has ac-cess to the greatest number of multiple interaction path-ways. To test this learning hypothesis, we regressedgender and frequency of being cited as an expert againstLadino first-factor scores in the combined Itza’-Ladinocultural consensus model. The r2 on Ladino scores was.63 (F[2,10] p 6.97, p p .02) with gender (p p .02) andexpertise (p p .008) both reliable. One subgroup (fourmen, one woman) averaged 5.8 expert citations, 6.0 so-cial network citations, and a first-factor consensus of .73(versus .75 for Itza’). Averages for the other subgroup (fivewomen, one man) were respectively 0, 1.3, and .59. MaleLadino experts appear to be driving the Ladino popula-tion to a convergence of knowledge with Itza’.

Over time, socially well-connected male Ladinos con-verge toward the consensus of Itza’ experts. For example,we found that judgments of plant-animal associationsfor the most highly rated Ladino expert comprised aproper subset of the pairwise judgments of the mosthighly rated Itza’ expert. It is highly improbable thatLadinos who approximate Itza’ response patterns actu-ally observe and copy what Itza’ say and think about eachof the species pairs in question. Rather, individual La-


dinos, in large part, seem to project fragmentary obser-vations of Itza’ behavior to a richly textured cognitivemodel of folkecology by inference rather than imitation(Atran 2001a).

the learning landscape

In line with evolutionary models of prosocial learning,let us assume that, when in doubt or ignorance aboutsome domain of activity vital to everyday life, peoplewill seek to emulate those with knowledge (Boyd andRicherson 1985, Lansing and Kremer 1993, Henrich andBoyd 1998). Let us also assume that they have directaccess not to the deep knowledge they wish to emulatebut only to surface signs or “markers” of that knowledge(much as people who wish to be like a famous acade-mician, powerful politician, or celebrity adopt that per-son’s outward trappings in the hope that these will helpguide them to success). One reasonable strategy is to seekknowledge from those to whom deference (respect) isshown by others (Henrich and Gil-White 2001). In manysmall-scale societies, knowledge-bearers tend to be el-ders, political leaders, economically well-off, and so on.

In the Itza’ case, forest experts are experts in a varietyof relevant domains (birds, mammals, trees, soils), eldermales, and former political town leaders. Ladinos expressdoubt about forest knowledge and also express a desireto acquire knowledge from Itza’. Apparently, the mostrespected and socially well-connected Ladinos attend tothose Itza’ to whom other Itza’ defer, and these Ladinos,in turn, become subjects of emulation and sources ofknowledge for other Ladinos. But how do Ladinos obtainrelevant knowledge without initially knowing how it isrelevant? Besides patterns of deference, which carry noknowledge content, how and what do Ladinos learn?

Nearly all evolutionary models of prosocial learningassume that the most important information learned isabout norms, that is, shared rules or principles of knowl-edge, judgment, or behavior. Norms are supposed to befunctional units of cultural selection and evolution(Irons 1996, Sober and Wilson 1998, Boyd and Richerson2001, Gil-White 2001). Evidence indicates that neitherLadino experts nor the wider Itza’ or Ladino populationsare learning norms about the forest from Itza’ experts orfrom imitating one another. Itza’ express no content-spe-cific normative or prototypical attitudes to the forestother than to “care for the forest as it cares for us.” Eventhe notion of “reciprocity” that we invoke to describeItza’ patterns is only a gloss for statistically consensualpatterns of thought and behavior with considerable var-iation in content. Ladinos may be acquiring knowledgein part through different isolated examples that triggerpreexisting inferential structures to generate convergentpatterns and in part through stories and other evocativeconduits (on stories, see Atran 2001b).21 Thus, a Ladino

21. Anthropologists are typically instructed to go out into the fieldalone for some months or—in exceptional cases—a few years andsingle-handedly bring back a description of a society. In this situ-ation, there seems to be little alternative to normative description.

may observe or hear about a specific exemplar of eco-logical knowledge (perhaps embedded in a tale) from arespected Itza’, observing, for example, that Itza’ elderslook for fallen ramon fruits after spider monkeys havepassed through trees. Itza’ pick up the fruits that are notchewed through and leave the rest, knowing also thathalf-chewed fruits are even more likely than unchewedfruits to generate new ramon stands. From such Itza’behavior a Ladino observer may readily deduce that (1)ramon is desired and useful for people and (2) spider mon-keys can negatively affect ramon seeds. But Ladinosdon’t generally learn that (3) spider monkeys can alsopositively affect ramon seeds and so help both the forestand people in it. Although Ladino observers seem to lackthe Itza’ cultural bias of conceiving species relationshipsreciprocally, they still spontaneously induce much morefrom a single instance of experience than simply 1 and2.

One alternative to normative accounts of cultural for-mation, transmission, and evolution is “cultural epide-miology,” which assumes that socially learned infor-mation is acquired chiefly through inference rather thanimitation (Sperber 1996, Atran 2001a). Information is ac-quired via inference whenever it presents a “contentbias.” For example, people in every society readily gen-erate richly structured folkbiological taxonomies fromfragmentary samples of plant and animal kinds becausethey have an evolved, task-specific (i.e., “modular”) sys-tem for folkbiological induction (Atran 1998). This “liv-ing-kind module” allows anyone, anywhere, to take iso-lated plant or animal exemplars (whether observed,reported, seen in a book, on television, etc.) and auto-matically assign them to one and only one generic spe-cies that occupies one and only one position within ataxonomic structure (Berlin 1992).

To illustrate: using standard taxonomic sorting exper-iments, we elicited highly consensual mammal taxon-omies (see Lopez et al. 1997). For each population therewas a single-factor solution (I p 7.2:1, 61%; L p 5.9:1,50%; Q p 5.8:1, 48%). First-factor loadings were uni-formly positive, and mean first-factor scores reflectedhighly shared competence for each population (I p .77,L p .71, Q p .68). The aggregated Ladino taxonomycorrelated equally with Itza’ and Q’eqchi’ taxonomies (rp .85), indicating very similar structures and contents.22

Normative accounts of society are also closely bound to the doc-trine of functionalism. It is not our intention to review the im-portant critiques of functionalism (and allied theories of behavior-ism [cf. Murdock 1949]). We would simply point out that from apsychological perspective it is not clear where norms exist: in thebrain of an omniscient informant, as an emergent structure par-tially distributed among individual minds, as a prototypical rep-resentation of a statistical pattern, or as a summary account in theanalyst’s mind. There is scant detail in normative accounts of socialstructure that allows evaluation of patterns of individual variation,agreement, and disagreement within and between groups (e.g., Hu-man Relations Area Files). Without such detail, normative claimsare difficult to verify or falsify.22. All three populations grouped taxa according to general-purposesimilarity rather than special-purpose concerns (e.g., wild peccarywith domestic pig, house cat with margay, etc.). Special-purposeclusters such as domestic versus wild or edible versus nonedible


Thus, we should expect Ladinos to generalize obser-vations much as Itza’ do when Itza’ and Ladino taxon-omies coincide. All that is required is that Ladinos obtainsome ecologically relevant content from observing Itza’(Q’eqchi’ aren’t looking to Itza’ for ecologically relevantcontent and so cannot generalize). Much of the differencebetween the top Ladino and Itza’ experts owes to thisLadino’s relative lack of knowledge about small palms(P09–P11) and felines (A24–A26). These are well-delim-ited taxonomic groups for Ladinos and Itza’ (Atran 1999).The Itza’ expert readily generalizes ecological informa-tion from any member of these taxonomic groups toother members; the Ladino expert simply may not haverelevant information about plants and animals in thesetwo taxonomic groups to generalize from.

One distinctive aspect of the aggregate Itza’ folktax-onomy for animals that emerges is that only Itza’ indi-cate the effect of ecological proclivity on mammal group-ings. A notable example is the clustering of arborealprocyonids, monkeys, and tree-dwelling rodents as a tax-onomic group (including A02–A06 [Lopez et al. 1997]).Ladinos also generally lack knowledge about bidirec-tional ecological relations between plants and animals,especially with regard to positive animal-plant relation-ships (animals helping plants [fig. 3]). These two fac-tors—taxonomic awareness of an arboreal animal asso-ciation and a culturally salient propensity to appreciatebidirectional ecological relations—favor more consistentgeneralization of ecological centrality among arborealanimals and plants with which they interact, such asfruit trees (P01–P05). This also applies to ecological ap-preciation of the Itza’ taxonomic cluster of game birds(including A07–A12).

More generally, these data suggest that patterns of un-dergeneralization and overgeneralization are predictablefrom the Ladino–Itza’ correspondence of taxonomic andecological groupings. Where Ladino groupings are a sub-set of Itza’ groupings we expect undergeneralization andwhere they are a superset overgeneralization. Moreover,Itza’ may know better when there are exceptions to anexpected generalization. We also have evidence thatsome Ladinos are making plausible but unwarranted in-ferences that Itza’ do not. For example, in the absenceof direct observation of furtive, nocturnal felines, it isplausible to believe that they would hide under the coverof leafy fruit trees to prey upon other animals that feedon the fruit. Female Ladinos who seldom venture intothe forest overwhelmingly infer that felines prefer fruittrees (75%). Male Ladinos (17%) and Itza’ (16%) knowbetter because they actually witness feline stalking be-havior in the forest underbrush.

In brief, it appears that preexisting taxonomic struc-tures and lack of culturally prior conceptions of ecolog-ical bidirectionality in plant-animal relationships con-strain how Ladinos infer knowledge of ecological

can also be elicited (Lois 1998), but they do not belong to the generalconsensus of “kinds that go together by nature” (cf. the idiosyn-cratic version of “Itza’ folk taxonomy” in Hofling and Tesucun1997).

centrality. Of course, any significant body of social in-formation includes many different kinds of biases towardparticular kinds of content, but norms and imitation mayhave little to do with this learning process. Rather, sociallearning in the case described arguably involves infer-ential processes that are mobilized by several factors: (1)domain-specific cognitive devices (e.g., taxonomy for bi-ological kinds), (2) prior cultural sensitivity to certainkinds of knowledge (e.g., species reciprocity in ecologicalrelationships), (3) awareness of lack of knowledge andthe motivation to acquire it (doubt), (4) selective atten-tion (e.g., Ladino focus on the patterns of deference toand the behavior of Itza’ elders versus Itza’ deference andattention to the forest itself), and (5) preexisting values(weighted preferences) regarding a given cognitive do-main (e.g., overvaluing economic utility relative to otherdeterminers of interest such as sacredness or role in theeconomy of nature).

The particular persons observed, actual exemplars tar-geted, and specific inferences made can vary widely fromperson to person. Here the culturally specific learninglandscape further constrains the canalization process ofour specieswide evolutionary landscape. Much as rainfalling anywhere in a mountain valley converges into thesame natural mountain-valley river basin and is furtherchanneled through the gates of a dam constructed there,so each person’s knowledge will converge toward thesame cultural basin of thought and action (Sperber 1996).

This “learning landscape” shapes the way in whichinferences are generalized from particular instances (ex-periences, observations, exemplars). It channels the in-formation acquired toward convergence with a generalbody of knowledge (an emergent structural pattern thatachieves a statistical consensus in a population). It pro-duces convergence toward the emergent consensus eventhough specific inputs vary widely in richness and con-tent (just as many different people, observing many dif-ferent exemplars of dog under varying conditions of ex-posure to those exemplars, all still generate more or lessthe same general concept of dog). Other learning factorsmay be involved, including norms and narratives, butthey are not the only or even the primary ones.

In summary, Ladino knowledge is a subclass of Itza’knowledge that underrepresents its ecological complex-ity. Ladinos look to Itza’ for what is important, whereasItza’ look directly to the forest. To be sure, Ladinos usetheir own taxonomic and ecological knowledge to gen-eralize inferences from Itza’ behavior, but they do notappear to have learned quite how “to walk alone in theforest” as Itza’ do. From studies of other Ladino com-munities in Peten, however, it seems that some PeteneroLadino communities have learned to think and act muchas Itza’ do after three of four generations of the kind ofcontact described between our Itza’ and Ladino samples(Schwartz 1990).23 This may well involve assimilating“spiritual values.”

23. Some Peteneros perform Maya rain ceremonies that Itza’ nolonger do (Schwartz 1990).


establishing a knowledge baseline: thehighland q’eqchi’

Social network analysis suggests that one set of factorsmilitating against Q’eqchi’ preservation of lowland ecol-ogy involves linguistic isolation coupled with a compactsocial structure that forecloses the intercultural ex-changes that might convey appropriate lowland tech-niques. Moreover, Q’eqchi’ immigrants tend to invokecorporate and ceremonial ties with the sacred highlandmountain valleys when faced with economic and eco-logical problems (Schackt 1984). This may serve to deter(not just detour) access to ecological information rele-vant to lowland commons survival. Studies of other im-migrant Q’eqchi’ communities also indicate selectiveuse of inappropriate highland techniques (clear-cutting,cash-cropping, continuous cultivation) and failure or in-ability to transfer highland techniques favoring forestmaintenance (intercropping, terracing) (Castellon 1996).

One open issue is whether Q’eqchi’ immigrants arrivein Peten with a cognitive model that is already impov-erished with respect to knowledge of species relation-ships or whether they are simply unable to use richerhighland models because these are inappropriate to low-land ecology. Accordingly, to understand the cognitivefactors responsible for Q’eqchi’ immigrant patterns wesought to establish a cognitive baseline for highlandQ’eqchi’ in their original home area of Alta Vera Paz. Weelicited mental models from highland Q’eqchi’ in AldeaPaapa near Coban, employing the same techniques as inPeten. The kinds most frequently mentioned as impor-tant to the forest included 27 animals and 20 plants.Nearly half the animals (13) mentioned as important alsocounted among the most important Peten species. Plantsmentioned as most important to the forest included onlytwo of the most important Peten species, the pacayapalm and the allspice tree, and also included species pri-marily associated with orchard (e.g., peach tree, Prunuspersica) and milpa (e.g., chile pepper, Capsicum an-nuum). Although two animal kinds and half the plantsmust still be identified scientifically, patterns of inter-action between plants, animals, and people can be reli-ably described.

As with the Peten groups, highland Q’eqchi’ viewplants as positively affecting animals, first by providingfood and second by furnishing shelter. Consensus on pos-itive plant-animal relations is marginal (eigenvalue 1:2p 2.97, variance p 44%). Nearly 20% of all possibleplant-animal relations are positive. This figure is aboutthe same for Itza’ and Ladinos. Highland Q’eqchi’ rec-ognize more negative animal-plant relations (2.3% forall possible animal-plane relations) than immigrantQ’eqchi’ (! 1%) but less than Itza’ (7.8%) or Ladinos(8.2%). Highland Q’eqchi’ recognize fewer positive re-lations of animals affecting plants (! 1%) than Ladinos(2.1%) and far less than Itza’ (8.2%). Q’eqchi’ evince finerappreciation of local ecology in their highland homelandthan in their lowland habitat, but this appreciation issignificantly less rich than that of Itza’ or even immi-grant Ladinos. Highlanders also show good consensus on

how humans negatively affect plants (eigenvalue 1:2 p7.68, variance p 75%) but no consensus on how humanspositively affect plants. This reinforces the picture ofsimilar cultural notions of how plants affect animals butdifferent models of animals affecting plants, of positiveanimal-plant relations in particular, and of reciprocitybetween animals, plants, and humans in general.

Measures of human impact and use confirm this pat-tern in content-specific ways. For highland Q’eqchi’, re-gression analyses show that food value and ecologicalcentrality predict human impact (r2 p .58, F[2,12] p 8.20,p p .006, both predictors p p .06). Food value and impactare positively correlated (t p 4.937, p p .0001); highlandQ’eqchi’ tend to protect food plants. By contrast, eco-logical centrality and impact are negatively correlated,as are ecological centrality and food value (t p �2.379,p p .03). In other words, highland Q’eqchi’ do not con-sider food plants to be ecologically important and do notprotect plants that they consider to be ecologically im-portant. The most important predictor of ecological im-portance is use of the plant for firewood (r2 p .54, F[1,18]p 21.457, p p .0002).

Only Itza’ seem to have a positive vision of the roleof plants, animals, and humans in helping the forest tosurvive that is based on species reciprocity. For example,an analysis of variance involving all Peten groups to-gether with highland Q’eqchi’ shows that only Itza’ differreliably from each of the other groups in appreciatingpositive animal-plant relations; the other groups do notreliably differ from one another (F[3,44] p 21.24, p !

.0001). Both immigrant and highland Q’eqchi’ report thatanimals have little impact on plants. Immigrant Q’eqchi’view humans as having a markedly negative impact oneconomically important plants, especially those withcash value. Highland Q’eqchi’ tend to destroy plants thatthey consider ecologically important—especially plantsthat can be used for firewood—and to protect only plantsthat have high food value.

Cash sale and firewood are arguably the least produc-tive categories in terms of forest regeneration. Cash saleof important plants is not part of a local system of pro-duction; it is driven by an extractive economy that de-pends almost entirely on demand from outside marketsand even outside the region. For example, extraction of“nontimber forest products” such as xate has beentouted as a prime source of “sustainable development”of the Peten forest (Reining and Heinzman 1992), butxate extraction (or even latex tapped from the chicle tree)traditionally had little value in the subsistence economy.

Organization of labor and production for extraction ofsuch products has little to do with local subsistence re-quirements—including the requirement that the localpeople and the environment they live in be mutuallysustaining—but everything to do with outside markets.When those markets shift or collapse, no provisions aremade for the consequent shift or collapse of the socialorganization and ecology necessary to maintain produc-tion. Moreover, in contrast to the situation in advancedmarket economies, alternative outlets for labor and pro-duction are scarce. A return to premarket conditions is


liable to be just as difficult. In short, the more divorcedthe money item is from local subsistence needs or a sys-tem of local production, the less likely in the long runit is to sustain the local economy and the environmentsupporting that economy.

It would take us too far afield to explore here the his-torical reasons for the relatively impoverished Q’eqchi’models. Nevertheless, a few summary observations arein order. There is scant evidence that the trauma of re-cent civil war is a key factor in immigrant Q’eqchi’ at-titudes toward Peten. Highland Q’eqchi’ models of spe-cies relations were likely already impoverished, perhapsin part because the highland environment is relativelydeforested (compared with Peten), and earlier and on-going Q’eqchi’ migrations into the lowlands show littleconcern with maintaining forest biodiversity. Under theprotection of the Dominican clergy for centuries, high-land Q’eqchi’ institutionally managed their own highlycommensalist and intense forms of cultivation. Whenland was scarce, they migrated into the Peten lowlands,often for the short term.

Other Q’eqchi’ communities that immigrated into Pe-ten and adjacent areas of Belize both before and after thecivil war behave similarly to our study group (Carter1969, Fagan 2000). When environmentally related eco-nomic difficulties arise (e.g., banana blight, hurricanes,etc.), immigrant leaders may send delegations to sacredplaces in the Q’eqchi’ highlands to seek aid and redressfrom highland spirits (cf. Schackt 1984), but our immi-grant Q’eqchi’ do not concern themselves with lowlandspirits or consult Itza’. When we asked why they failedto consult Itza’ about the forest, the Q’eqchi’ often re-marked that they did not feel the need to seek out orplacate lowland spirits as long as they remained true totheir ancestral deities.

The Spirit of the Commons

Anthropologists and sociologists target norms as func-tional building blocks of cultures and societies. Econo-mists and political scientists see norms as institutionalmeans for solving public goods problems such as “thetragedy of the commons” (Hardin 1968, Fukuyama 1995).The general idea is that to solve the problems of rationalchoice inherent in balancing individual with collectiveneeds, individuals must be made to forsake a measureof self-interest and to sacrifice resources in accordancewith institutional norms that function to maintain thepublic good(s).

The tragedy of the commons and similar social dilem-mas are basically variants of a fundamental problem indecision theory and game theory known as “the pris-oner’s dilemma.” Consider a group of n persons whoshare a common territory of fixed size on which theyhunt animals. Each hunter has one of two choices: hecan cooperate with the others by not overhunting on thecommons, or he can hunt in a way that is advantageousto him but ultimately results in the overuse and destruc-tion of the common resource. The second option, it ap-

pears, is more rational in the short term. This is becausethe short-term advantage to one who overhunts (e.g., 1)always outweighs the short-term disadvantage to himwhen that disadvantage is equally distributed among theother hunters (1/n). If all cooperate, the common re-source is preserved, but if the rationale of self-interestpervades the camp, no one will have an incentive tocooperate and all will defect.

Field and laboratory studies indicate that individualcalculations of rational self-interest collectively lead toa breakdown of a society’s common resource base unlessinstitutional or other normative mechanisms are estab-lished to restrict access to cooperators (Berkes et al. 1989,Atran 1986). This is so even when people’s “basic needs”are satisfied, no matter how small the group or how well-informed of the looming tragedy (White 1994; but seeOstrom 1998 on strangers’ cooperating, at least for low-cost items). Yet, evidence from our “garden experiment”indicates neither the primacy of norms in explaining cul-tural differences with regard to the tragedy of the com-mons nor the exclusivity or primacy of institutionalmechanisms as means for preserving common resources.Immigrant Q’eqchi’ form the most socially intercon-nected and institutionalized community but are the leastlikely to preserve the resource base (perhaps because thecommunity is so culturally hermetic). The Itza’ com-munity is the most socially atomized and least institu-tionalized (at least in terms of coordinated agriculturalschedules), but its individuals most clearly act to main-tain the common environment. If neither institutional-ized learning nor institutional control mechanisms areresponsible for commons maintenance among Itza’, whatis?

values

More generally, the puzzle for decision theory is: Howdo people sustainably manage limited resources withoutapparent institutional or other obvious normative con-straints to encourage and monitor cooperation? Our ten-tative line of reasoning is that Itza’, and perhaps othernative peoples with a long history of ecological main-tenance, may not treat resources as traditional decisionand game theory suggests, that is, as objects of a payoffmatrix (extensional items substitutable along some met-ric, such as one that assigns monetary value to everyobject). Instead, they may treat resources such as speciesas intensional, relational entities that are subjectivelydefined, like friends or enemies (cf. Rappaport 1979, In-gold 1996).

Our next study explored this possibility. We asked peo-ple from each of the three Peten groups to rank-order 21plant species in terms of their importance according to(1) members of their own community, (2) and (3) mem-bers of each of the other two communities, (4) God, and(5) the forest spirits. The average age for Itza’, Ladino,and Q’eqchi’ informants, respectively, was 63.4, 59.8,and 55.9 years. An analysis of variance yielded no sig-nificant age differences across populations.

Itza’ showed consensus regarding their own individual


values (ratio eigenvalue 1:2 p 4.73, eigenvalue 1 p 59%of variance), imagined Ladino (4.97, 50%) and Q’eqchi’(4.33, 56%) values, and God’s imagined values (5.16,56%). Overall, Itza’ did not show consensus on the imag-ined values of spirits, but male Itza’ did show such con-sensus (3.49, 52%; p ! .05 given mean competence p.72 for N p 7 informants [see Romney, Weller, and Batch-elder 1986:326]). Male Itza’ stay much longer in the forestthan women and report many more encounters with for-est spirits (arux p uyumil k’aax, “masters of the forest”).To be a successful hunter (aj b’ak saj, “the meat getter”),chiclero (aj men cha’, “the chicle latex collector”), orfarmer (aj men kol, “the milpero”) requires that a manresist the constant teasing, tricks, and traps of the forestspirits without cursing the spirits for his misfortune. Hemust also show the spirits that he knows how to protectthe animals and plants under their care, and he must doso with “valor” (muk’). Only then will the forest spirits,in turn, protect him and his activity (Atran 2001b).

Ladinos also showed consensus on Ladinos (4.34,58%), Itza’ (3.38, 47%) and God (3.19, 50%). Theyshowed marginal consensus on Q’eqchi’ (2.80, 49%) butno consensus on spirits. By contrast, Q’eqchi’ showed noconsensus on Q’eqchi’, Itza’, Ladinos, God, or spirits. Afollow-up study of younger Q’eqchi’ men (average age42.5 years) and women (average age 41 years) indicatedsome consensus on Q’eqchi’ (3.27, 47%). These youngermen and women, who have spent nearly their entireadult life in Peten, may be establishing their own in-formation network about the Peten forest, althoughthey are still largely impervious to outside knowledgesources.

Itza’ overestimated the agreement between their re-sponses and those of the Q’eqchi’ but underestimatedagreement with Ladinos (Q’eqchi’ also overestimatedagreement with Itza’). For example, the correlation be-tween Itza’ beliefs about Q’eqchi’ preferences and theirown preferences (i.e., [Iq, Ii], r p .94) differed appreciablyfrom the correlation between Itza’ beliefs about Q’eqchi’preferences and actual Q’eqchi’ preferences ([Iq, Qq], r p.74). By contrast, the correlation between Itza’ beliefsabout Ladino preferences and Itza’ preferences ([Il, Ii], rp .90) did not differ appreciably from the correlationbetween Itza’ beliefs about Ladino preferences and actualLadino preferences ([Il, Ll], r p .93). Similarly, ([Qi, Qq],r p .84) � ([Qi, Ii], r p .62) p .22, whereas ([Lq, Ll], r p.92) � ([Lq, Qq], r p .84) p .08. Ladino beliefs about Itza’preferences correlated almost perfectly with Ladino be-liefs about God’s preferences (r2 p .96).

Only Itza’ saw the forest spirits as actively protectingthe forest: Itza’ rankings from the point of view of theforest spirits were significantly related to Itza’ models ofhuman impact as well as ecological centrality. For ex-ample, multiple regressions showed that male Itza’ con-sensus on spirits together with the overall Itza’ consen-sus on combined use (wood � shelter � cash) accountedfor most of the variance in human impact (r2 p .70,F[2,18] p 20.71, p p .0001, with spirits and use equallyreliable predictors [p’s ! .01]). The most reliable com-bination of predictors for what (male Itza’ believe) the

spirits think is ecological centrality and God (both p’s !

.01) (r2 p .65, F[2,18] p 17.0, p p .0001). Ladinos andQ’eqchi’ asserted belief in forest spirits and even pro-vided normative accounts of spirit life similar to thoseof Itza’, but in these two groups there was no consensusabout spirit preferences, nor was belief in spirits reliablylinked to forestry practice.24

Finally, we asked 17 members of several NGOs at aworkshop on the Maya Biosphere Reserve (November-December 1999) to rank the same trees in terms of im-portance to forest life. For the NGOs, there was marginalconsensus, with one slightly negative first-factor score(�.076) (ratio eigenvalue 1:2 p 2.73, variance p 45.6%).The most valued species for the NGOs were, in rankorder, mahogany, tropical cedar, allspice, and chicle.These are the most important trees for the extractiveeconomy and export market. NGO preferences partiallypredict the consensus on preferences expressed by La-dinos (r2 p .72) and Itza’ (r2 p .44),25 but the worst pre-dictors of NGO rankings are Itza’ male rankings of spiritpreferences (r2 p .06, n.s.) and Itza’ ratings of ecologicalcentrality (r p �.229).

spiritual games

To date, rational-decision and game-theoretic accountsinvolving human use of nonhuman resources have notconsidered the possibility of resources’ (e.g., species’)having their own measures of “utility” or resources’ andhumans’ being “players” in the same game. Prima facie,this idea is implausible, because species are assumed notto have motives, desires, beliefs, or strategies for coop-eration or deception that would be sensitive and syste-matically responsive to corresponding aspects of humanintention. Nevertheless, both in increasingly globallyoriented ecological movements in the industrial worldand in the religious practices of small-scale societiesthere are public pronouncements of respect for species.Indeed, one claim for “animistic” and “anthropomor-phic” interpretations of species in many small-scale so-cieties is that the “intention gap” between humans andspecies is thus bridged (at least to human satisfaction)with outcomes beneficial to the survival of species and

24. In one of the few studies to replicate findings on theories ofmind in a small-scale society (cf. Wimmer and Perner 1983), Knightet al. (2001) showed monolingual Yukatek children a tortilla con-tainer and told them, “Usually tortillas are inside this box, but Iate them and put these shorts inside.” Then they asked each childin random order what a person, God, the sun (k’in), the principalforest spirits (yumil k’ax’ob, “masters of the forest”), and otherminor spirits (chiichi’) would think was in the box. Children over5 attributed true beliefs according to a hierarchy with God at thetop and people at the bottom. As do Itza’, these Yukatek Mayaconsider the masters of the forest powerful and knowledgeable spir-its that punish people who try to overexploit forest species. Yukatekchildren tend to believe that the forest spirits, like God and thesun, “live” (kukuxtal) but do not “die” (kukumil) (see Atran et al.2001).25. Recall that for the Q’eqchi’ there was no consensus.


of the human groups that depend on them (cf. Bird-David1999).26

When asked, Itza’ men and women express the beliefthat they will be punished if they violate spirit prefer-ences, although women are less clear about what suchpreferences are likely to be. Especially for men, it ap-pears, the spirits are intermediaries or “spokesmen” forthe forest species, although there is wide individual var-iation in response to what the spirits say about any givenspecies. This has intriguing implications for ecologicaldecision and game theory in that individual Itza’ may bebasing cognitive and behavioral strategies for sustainingthe forest more on playing a game with spirits than onplaying a game with other people (on the wide role ofspirits in Itza’ life and religion, see Atran 2001b). Evo-lution itself provides mechanisms for interactive“games” that make commensurate the incommensura-ble (e.g., “strategies” of bacteria and their hosts), and somay human minds (semantically rather than biologi-cally) in ways consistent with maintaining respect forsacred or “taboo” values basic to long-term survival andquality of life (Fiske and Tetlock 1997, Medin et al. 1999).

Highland Q’eqchi’ also religiously taboo unbridled ex-ploitation of forest species or, more accurately, certainforested places. The taboo on exploitation of the forestedareas near certain highland caves and mountain streamsand worship of these sites (tzuultaq’a) are akin to oursociety’s protection of and reverence for wilderness ar-eas, but they imply a set of attitudes and understandingsthat differs fundamentally from Itza’ values with regardto forest species. For Itza’, the forest is a mental andecological landscape that people—the most savvy ofwhom follow spiritual guides—create and manage by liv-ing with it in cooperative exchange. For the Q’eqchi’, asfor many conservation organizations, the forest can beused and exploited or preserved and worshipped but sel-dom exploited and worshipped simultaneously.

Conclusion

Different cultural groups subject to equal pressures oncommon resources respond with strikingly different be-haviors and cognitions. These culturally distinctive ac-tions and folk models are systematically related to oneanother and to distinct prospects for maintenance or de-struction of the common environment. Native Itza’

26. There is nothing in principle to prevent rational-choice theoryfrom assigning extensional values to relational entities (e.g., peoplemay be willing to choose to save their pet over a favorite tree, theirchild over their dog, their nation over their children). Do sacredvalues form a special class of “protected choices” that are internallynegotiable but off-limits to more mundane, monetary exchanges?It is not clear how current approaches could model such choicesexcept as ad hoc “externalized contingencies.”

Maya, who have few cooperative institutions,27 showawareness of ecological complexity and reciprocityamong animals, plants, and people; Itza’ agroforestry fa-vors forest regeneration. Immigrant Q’eqchi’ Maya, whohave highly cooperative institutions, acknowledge fewecological dependencies; Q’eqchi’ agriculture is insen-sitive to forest survival. The folkecological models andagroforestry practices of Spanish-speaking immigrant La-dinos fall between. There is no overriding “local” or “in-digenous” relation to the environment.

The area that these groups reside in is not completelyopen to access by outsiders, but all of the actors arekeenly aware that their commons is highly vulnerableto uncontrolled immigration, depredation, and lawlessdeforestation. Theories of rational action predict that in-creases in the number of noncooperative players in theenvironment and their apparent disregard for the futureshould lead even native cooperators to abandon long-term interest for short-term gain unless institutional re-straints can compel individual action toward the com-mon good. In other words, it is irrational to continue anact to sustain a diminishing resource that others increas-ingly deplete. No doubt economic rationality and insti-tutional constraints are important in determining ac-tions upon common-pool resources, but they may notsuffice. There also appears to be an important cognitivedimension to the way people learn to manage environ-mental resources. Valuation studies suggest that, at leastfor some small-scale societies like the Itza’, cognition ofsupernatural agents may serve not only to guaranteetrust and foster cooperation between nonkin, as standardcommitment theories assume (Frank 1988, Irons 1996),but also to foster human interaction with nonhumanresources in relations of “indirect reciprocity” (Alexan-der 1987) to monitor and accommodate to nature’s re-quirements for obtaining its support.

It is no surprise that native Maya with centuries-olddependence on a particular habitat better resist actionsthat lead to its degradation than immigrants, althoughthe underlying models for behavior and modes of learn-ing are not predictable on a priori grounds. What is sur-prising is that Ladino immigrants who share no evidenttradition with native Maya come to resemble them inthought and action. Network analyses reveal reliablethough noninstitutionalized channels that allow sociallywell-connected Ladinos access to Itza’ forest expertise.The highest overlap, or “fidelity,” among individual pat-terns stems from subtle forms of inference based on in-dividual exposure to role models, not instruction or im-itation. No identifiable “rules, “norms,” or other discretebits or cultural information or behavior function as plau-

27. Faced with the rapid and relentless demise of their languageand forest, in 1997 an association of Itza’ successfully petitionedGuatemala’s National Assembly to grant them self-managementof a small portion of former Itza’ territory as the Bio-Itza’ reserve.Here Itza’ hope to teach the value of the forest before it disappears.Some Itza’ elders have adopted Q’eqchi’ prayer ceremonies to helpcreate this new institutional identity. The Q’eqchi’ are appreciativeand seem to be paying attention.


sible candidates for cultural transmission and selection.28

Our results call into question an assumption of agent-based norm models (Axelrod 1997, Rogers 1998), namely,that societies that do not share cultural norms (e.g., Itza’and Ladinos) are less likely to interact and converge thansocieties that do (e.g., Itza’ and Q’eqchi’).

This bears on the problem of extending the lessons oflocal commons to mobile and multicultural societies:Even in a relatively open-access environment (e.g., withuncontrolled immigration), if there is ready access torelevant information, then ecologically sound behaviorsmay be learned by relative newcomers who have no so-cial institutions, cognitive predispositions, or culturaltraditions favoring commons survival. But having timeto learn is critical. Rates of environmental and culturaldegradation in neotropical areas are awesome by anystandard because of global economic and political pro-cesses that function similarly across such areas. Themassive demographic, economic, and political upheavalsof recent years may no longer allow sufficient lag timefor immigrants to contact, learn, and implement Itza’techniques.

Although an examination of the cultural values andcausal processes that mediate the relationship betweencognition and behavior lie beyond our scope, we thinkthat the studies reported here are important to (1) estab-lish a methodology for assessing a first approximation ofecological cognition across cultures, (2) set up a frame-work for further exploration of the relationship betweenthe structure and the flow of ecological informationacross cultures, (3) provide operational criteria for as-sessing “spiritual” values that may help human societiesto resolve “the tragedy of the commons” and other ec-ologically pertinent forms of “the prisoner’s dilemma,”and (4) offer a model for empirical research into culturaleffects on deforestation and land change.

28. We used the same techniques to monitor ecological cognitionand social networks for Yukatek Maya (Xk’opchen) and Ladinos(Xkomha) in Quintana Roo (Mexico), among Lacandon Maya(Metzab’ak) and Tzeltal and Tzotzil immigrants in the Selva La-candona (Chiapas, Mexico), and among Native American Menom-inee and majority-culture rural groups along the Wolf River in Wis-consin (Medin, Ross, and Atran 2001). Preliminary studies(interrupted by civil strife) suggest that the patterns of knowledgeand behavior among Lacandon (Lakantun) Maya versus Tzeltal andTzoltil Maya born to families that immigrated into the Selva La-candona from the Chiapas highlands resembles that of Itza’ toQ’eqchi’ immigrants (Ross n.d.; cf. Nigh 2002). The fact that thesedescendants of immigrants have lived all their lives in the SelvaLacandona suggests that mere personal exposure to the local ecol-ogy is not a deciding factor. The Wisconsin studies concern fishand fishing rather than agroforestry. Results show that differencesbetween Menominee and rural majority-culture on fish-fish andhuman-fish relations resemble more the patterned differences be-tween Itza’ and Ladinos. This underscores the generalizability ofour methods.

Comments

francisco gil -whiteDepartment of Psychology, University ofPennsylvania, 3815 Walnut St., Suite 400,Philadelphia, Pa. 19104-6196, U.S.A.([email protected]). 23 i 02

Atran et al. conduct admirably careful measurements toestablish not only that Itza’, Ladinos, and Q’eqchi’ havedifferent relationships to and impacts on the forest butalso that what each community predominantly knowsand believes is causal. Their data make the interestingpoint that solutions to commons-management problemsmay have little to do with whether people play lots ofcooperative games in other domains. Despite these ob-vious strengths, the paper has an insufficiently clear andincomplete discussion of (1) knowledge-transfer mech-anisms, (2) the question of “norms,” and (3) the role ofimitation and/or inference.

The authors suggest that perhaps “ecological knowl-edge is directly transmitted from socially well-connectedforest experts” and proceed to test this. For the Itza’, thecorrelation analysis apparently did not reveal a relation-ship between knowledge and proximity to experts. Theauthors then state that “There is an alternative scenarioto learning about the forest that is more consistent withindependent discovery than direct social transmission,”one that involves “multiple pathways,” “cultural sto-ries,” etc., all of which assist inference as opposed toimitation.

But if the Itza’ are the biggest experts around, thennew knowledge needs to be produced by the occasionalItza’ insight. This ceiling effect should preclude findinga strong relationship between knowledge and proximityto experts even if knowledge indeed does flow from ex-perts to others and first through those close to experts.Cannot, then, the very expertise of the Itza’ explain whythe authors could not discern “relationships between re-sidual agreement and social or expert network proxim-ity”? This idea seems supported by the fact that Ladinosapparently did show such a relationship: “Male Ladinoexperts appear to be driving the Ladino population to aconvergence of knowledge with the Itza.’” If Ladinos arequickly absorbing many centuries’ worth of accumulatedItza’ knowledge, for them there is no ceiling effect, andso if knowledge flows from experts to others one willfind here a strong relationship between prestige andknowledge. If this argument is reasonable, the authors’conclusion that knowledge does not really flow from ex-perts and their search for an alternative “multiple-path-ways” explanation are confusing—doubly so becausesome of the stated “alternatives” are not even incom-patible with a prestige-effect (e.g., prestigious individualsmay tell stories, and these may be the stories best re-membered) and triply so because the authors later appearto endorse after all the “prestige-bias” account of thetransmission of empirically useful knowledge (defended


in Henrich and Gil-White 2001), denying merely thatwhat people learn from experts is “norms.”

Though my work is referenced (Gil-White 2001), Icould not recognize my arguments as those criticized inthe section on “norms.” With others, I am alleged tomaintain that most important information learned isabout “norms,” which are “rules or principles” and“functional units of cultural transmission and evolu-tion.” But I and others (notably Boyd and Richerson 1985)have used the term “norm” as an idea or belief with amoralistic or “ought” connotation. Norms need not be“functional”—they may be costly and counterproduc-tive. And neither I nor others (as far as I can tell) havecharacterized them as “most” of what is learned. Theyare interesting, but people also learn stories, songs, un-conscious mannerisms, skills, technologies, beliefsabout the world and about people, etc. Still on norms,the authors claim that “Itza’ express no content-specificnormative or prototypical attitudes to the forest.” Thisseems to equate “prototype” with “norm,” but the termsare hardly synonymous. Also confusing is the apparentlinkage between a defense of “norms” and a defense of“imitation.” Those who are alleged to insist that culturaltransmission is all about “norms” apparently therebyinsist on imitation, so the authors’ debunking of normsis presented as a rebuke of imitation. I admit that I ammaking inferences about a passage that is less than clear,but even if my translation reasonably got the gist of it Istill do not understand the argument.

The proposed “alternative” to imitation is an “infer-ence” triggered by observed behaviors. But isn’t the ideathat people make inferences when they copy behavior aplatitude? How can this be an alternative? Few items ofinformation can plausibly be transmitted whole throughnaked imitation because usually some (much?) of theinformation cannot be perceived. When I copy somebodyelse’s tennis serve I am in fact attempting to reproducean abstraction of what my model’s goals are, and whichI infer (in continuous updates) as I expose myself to astatistical distribution of his serves. I have never actuallyseen the thing I “copy.” But what violence does callingthis process “imitation” entail? None that I can see. Atother times the role of inferences may be more inter-esting. For example, it would be next to impossible forme to readily display adaptive behaviors for the Petensimply by plopping myself there in the absence of Itza’models (just look at the Q’eqchi’). Given Itza’ models, Imight get the behaviors, but my inferences about thebeliefs that support them in the Itza’ could be mistaken.Thus, models of cultural transmission will sometimesneed to keep track of behavior-memes copied throughimitation and belief-memes generated through inference(and which “explain” the behavior-memes to the copi-ers), especially given that, depending on which infer-ences are made, different biases for future cultural evo-lution will result. But this is more subtle than trying todecide between “inference” and “imitation” as if cul-tural transmission had to be about one or the other.

nora haennDepartment of Anthropology, Box 2402, Arizona StateUniversity, Tempe, Ariz. 85287, U.S.A. ([email protected]). 14 i 02

Harkening to reevaluations of the ways in which peopleconnect to physical settings, Atran et al. attempt “toseparate cultural effects from those of demography, econ-omy, ecology, and so forth,” while their material speaksto two questions within these inquiries. What role doesculture play in expanded human-environment research?As the factors contributing to an environmental outcomemultiply, what is the relationship between particularismand holism? The authors succeed in challenging utility-based theories and in offering a methodological model.Their transparent methods are especially welcome. Itherefore write in the spirit of moving the conversationto other considerations.

Because human-environment research is multidisci-plinary, it is important to outline disciplinary biases. Asan anthropologist, I view culture as an all-encompassingcategory, one that, as Atran et al. assert, clarifies appar-ently irrational behavior. I was, however, uncomfortablewith their separation of culture from other social cate-gories—an approach common in other disciplines. KayMilton (1996:220–21) argues that if anthropologists fol-low them in reducing culture to one of a series of spheres,anthropological perspectives lose their distinctivenessand intellectual diversity suffers. To my way of thinking,culture (in the broadest sense) should be a departurepoint from which anthropologists assess how culturalmanifestations (in a narrower sense) affect human-en-vironment relations.

This is more than a disciplinary concern, for the au-thors’ definition of culture as cognitive models weaklyaids elucidation of alternative Q’eqchi’ logics. At theend, Q’eqchi’ still appear as irrational actors, “insensi-tive toward forest survival” and lacking “respect” forPeten ecology. This irrationality is only partly softenedby assertions that Q’eqchi’ practices make sense in lightof their social world. I see two issues in conflict here.One lies within ethnoecology’s dual aims of validatingnon-Western knowledge and comparing it with Westernscience; here the attempt at validation runs counter toa comparison that casts Q’eqchi’ in a negative light.Where Nazarea (1999) suggests that the answer to thisdilemma lies in deeply situating knowledge systems,loosening ethnoecology’s rigid categories, and exploitingethnoecology’s methodological strengths, this articlesuggests that such measures may also require a recon-sideration of culture. A firmly grounded Q’eqchi’ ration-ality demands a theory of culture that examines partic-ular cases in the context of an even greater whole.

The second conflict centers on the description of theenvironment as both a constructed reality and existingoutside people’s models and symbols (an entity againstwhich beliefs can be measured as destructive or not).This dichotomy merits further consideration. Where in-dividuals locate themselves in respect to this distinctionhas to do with disagreements on whether or what kind


of an environmental crisis exists, which actors might beresponsible for that crisis, and what might be appropriatesteps for its resolution. The authors sustain the dichot-omy by connecting people and the environment throughcognitive models and land use. Recent theorizing sug-gests that other connections include emotions, identi-ties, history, social structure, and biophysical concerns.I agree that teasing out the relative contributions of suchfactors is important, and the authors’ exploration of cos-mology is appropriate in this sense. However, using adefinition of ecology as outside people’s models (the basisfor arguing Q’eqchi’ destructiveness) precludes full con-sideration of the more expansive setting.

Given Guatemala’s recent civil war, which set ethnicgroups against one another, I particularly wonder aboutthe impacts of political factors. The combination of po-litical and ecological issues is pressing for anthropolo-gists and environmentalists alike (Brosius 1999). Reflect-ing on politics, I noticed parallels between the Peten andCalakmul, just north of Guatemala on Mexico’s YucatanPeninsula, which share a setting marked by protectedareas. In Calakmul, ideas of Mayan, Chol, and Mestizoduplicate those of Itza, Q’eqchi, and Ladino, respectively.Julia Murphy and I examined how notions of Mayanadaptiveness (prevalent in local conservation circles) areconnected with the group’s close ties to the ruling party.Ideas that Chol are environmentally destructive haveroots in the group’s support for opposition parties. Stateagents describe Chol as “closed” or uninterested in othergroups. This depiction obscures the fact that the au-thorities avoid communication with largely monolin-gual Chol. In the meantime, Calakmul’s Mestizos formworking relationships with Mayans to appropriate thelegitimacy conveyed to the latter by authorities whoshare their heritage (Haenn and Murphy 1998). Mestizosview Chol in derogatory terms, but in meetings withforeign conservationists a Mestizo cleverly touched onideas of indigenous adaptiveness by asserting, “We, theindigenous people, know how to care for our forests.”How might political complexities affect notions of ra-tionality, utility, “causally connected representationsacross minds,” and the commonality of commonresources?

A deeper contextualization driven by an expansive un-derstanding of culture would strengthen the authors’conclusions. Game-theory models are incomplete be-cause not all people play by the same rules, are motivatedby the same concerns, or always view themselves as par-ticipating in the same contest. Such an approach is madeeffective by moving between parts and wholes the def-initions of which vary according to actors’ perspectivesas well as the intentions of outside agents. Given theauthors’ skills in quantitative analysis, I see them asbeing in a strong position to communicate these neweranthropological concerns to a broad social science andpolicy audience.

giyoo hatano and keiko takahashiDepartment of Psychology, University of the Air, 2–11Wakaba, Mihama-ku, Chiba, Japan 261-8586 ([email protected])/Department of Psychology,University of the Sacred Heart, Tokyo, 4-3-1 Hiroo,Shibuya-ku, Tokyo, 150-8938, Japan. 20 xii 01

No one denies the importance of preserving our preciouscommon resources, including the tropical rain forest.The study reported in this article is important in that itnot only presents a detailed description of the ways inwhich three different ethnic groups living in the samearea treat the forest but also offers what might be calledan “indigenous-practice model” for maintaining com-mon environmental resources. It shows that the nativeinhabitants practice agroforestry that favors forest re-generation even though their behavior is not constrainedby the cooperative principle because the accumulatedexperience of many generations has made them awareof a complex ecology involving animals, plants, and hu-mans. Moreover, even recent immigrants can obtain thisfolkecological knowledge and the relevant skills by ob-serving indigenous practices.

This indigenous-practice model may fit the dynamicsbetween the people and the natural environment in thenorthern Peten region very well, but we must questionwhether it can serve as the basis for measures for pre-serving the forest and other environmental resources ingeneral. We present an alternative model of environ-mental preservation developed in Japan that we call the“conscious-control model” and contrast it with the in-digenous-practice model.

The conscious-control model and the innovative en-vironmental education programs and preservation move-ments based on it are deeply rooted in the 40-year-longstruggle of the victims of mercury poisoning in Mina-mata. The Minamata trials clearly indicate that almostall environmental problems are political and economicones. Preserving a clean, safe, and comfortable environ-ment requires a conscious effort to control it and, morespecifically, deliberate resistance of the temptation tooverlook or acquiesce in the practices of economic andpolitical powers such as big business and the govern-ment. Even when people were suffering from poisoningand local medical specialists could pinpoint the cause ofthe disease, it was tempting to keep silent to avoid abattle that would be long and hard to win. The peopleof Minamata became vocal and assertive because theyrealized that there was no other way to change theirsituation. Although the arguments offered by authorizedenvironmental protection movements and educationalprograms are often apolitical, the Minamata-rootedmodel indicates that, without conscious and deliberatestruggle based on understandings of the political and ec-onomic aspects of the issue in question, our commonresources cannot be preserved.

Some Japanese schoolteachers are developing ambi-tious educational programs based on the conscious-con-trol model to foster interdisciplinary understandings ofenvironmental problems among their students. For ex-


ample, Kazuko Otsu, a high-school teacher, has con-structed a program to show how the everyday life of herstudents relates to that of other people on earth who areinvisible to them through the “window” of the price ofbananas. She first draws the students’ attention to thefact that most of the bananas they see are labeled by bigenterprises and imported from the Philippines. Theyoungsters are motivated to understand why these im-ported goods are so inexpensive. Then they view slidesof banana plantations in the Philippines to learn how theplantations damage the people, lands, and forests. Theyalso learn about the economic systems of production andcirculation of bananas and why the farmers’ share of thesales is so small. Finally, they are asked what they cando to protest this damaging economic and environmentalchain. In another example, an elementary-schoolteacher, Tamotsu Chiba, has tried to help children un-derstand “the hamburger connection”—to recognize therelationship between the increase in beef production andthe decrease in rain forest in South and Central America.

The need to refrain from exploiting common resourcesmay not be very great in the Reserve, although the forestis increasingly being burned off even there. In this highlyprotected area, the workings of economic and politicalsystems beyond the individual are not transparent.Therefore, the indigenous-practice model is individual-istic as well as cognitive, probably too much so. Thepreservation of environmental resources is “naturalized”in the sense that practices favoring forest regenerationare naturally led by the rich accumulated knowledge ofecological complexity and that this knowledge is natu-rally distributed by observing the practices of moreknowledgeable and respected old-timers. Although thisstory seems tenable, the conscious-control modelstrongly suggests that this is so only because (1) thenorthern Peten rain forest is not openly accessible to bigbusiness, (2) no powerful technology has been broughtin, and (3) the economy of the ethnic groups observed inthis study has not yet been integrated into a larger cap-italistic economy—their life is not driven by a desire tomaximize profit. Human behavior and learning, espe-cially with regard to resource consumption, have to beviewed from sociocultural as well as cognitive perspec-tives, because once selling becomes the purpose of pro-duction even the learning process changes (Greenfield1999).

ueli hostettlerInstitut fur Ethnologie, Universitat Bern, Langgasstr.49-A, CH-3000 Bern 9, Switzerland ([email protected]). 21 xii 01

Atran and associates make it clear that social actors rep-resenting three groups subsisting in the same habitat incentral Peten manifest differences in their behavior, cog-nition, and relationship relative to their immediate rain-forest environment. A first round of analysis begins toreveal these differences in these groups’ agricultural im-pact on the forest. As successive rounds of sophisticated

analysis of measurements of individual cognitions andbehaviors proceed, the differences harden into a patternwhich appears to leave no room for doubt that the an-cestrally local Itzaj Maya are the only conscious protec-tors of the forest. Spanish-speaking immigrants whohave learned their trade from the Itzaj remain close be-hind them, while the Q’eqchi’, who have only recentlymigrated to central Peten, are far behind on all counts.In addition, there is little hope for short-term change inthe Q’eqchi’s (destructive) interaction with the forest be-cause they neither interact with nor listen to those whocould teach them better.

Clearly, the research and analysis behind this paperare more innovative and sophisticated than one can ac-knowledge in as brief a comment as this. I seek ratherto highlight some issues which would likely havegreatest (probably negative) impact upon the people stud-ied—the Q’eqchi’. In particular, I wish that Atran andassociates would provide more information on the wayin which their research is related to local politics of cul-ture and development. They describe their work as de-veloping in near-laboratory conditions, characterized byboth optimal control of variables and maximum disin-terest of the analysts. Comments on Atran’s (1993) paperon Itzaj tropical agro-forestry and Hofling’s (1996) reporton the Maya Itzaj struggle for linguistic revitalizationindicate, however, that Atran and associates are not dis-interested but motivated researchers. This, of course, isnot a problem per se; it is the way anthropology works.Land tenure and land use are highly politicized issues inthe Peten (see Clark 2000, Grunberg 2000, Macz andGrunberg 1999, Schwartz 1995). Atran and associates arepart of that political situation, and they should havemade that clear or to their readers. A disclaimer in apreviously published version of the paper to the effectthat the authors “aim not to offer moral judgements onbehavior, but grounds for understanding and reconcilingconflicting behaviors” (Atran et al. 1999:7603) may notbe enough.

The data and analysis presented in this article are verystrong. However, on the basis of my own research in theMaya lowlands and the published research of others 1question whether the authors have isolated the right setof factors to explain why the practices of some cultiva-tors are more environmentally harmful than those ofothers. They isolate sociocultural factors (including so-cial networks and cognitive models) from economic,demographic, and ecological factors in local actors’ in-teraction with their environment and make them theprimary target of their research. I see a problem of cal-ibration in the way in which the key factor of “destruc-tiveness” is assessed here for agroforestry practice. Myown research in central Quintana Roo indicates that inorder to understand the dynamics of land use and henceof the “destructiveness” of agroforestry practice we needto study both the scale and the centrality of milpa ag-riculture in the (long-term) historical context of house-hold economies (Hostettler 1996, 2001, 2002). By “scale”I mean the relative economic importance of milpa ag-riculture as part of particular diversified household econ-


omies, and by “centrality” I mean the degree of identi-fication of producers with milpa agriculture in terms of“craftsmanship” (i.e., differences in situated practice). Inorder to examine “destructiveness” we need to considerthe complex interplay of social, cultural, economic, andecological factors (including particularities of market in-tegration, history of land use, and socioeconomic differ-ences among and social factors within households) andfind ways to weigh and calibrate them. Otherwise “de-structiveness” is measured, as in the case of this paper,on the group-as-a-whole level (a kind of “ethnicized de-structiveness”), which is methodologically not of thesame order as the authors’ subsequent detailed differ-ential assessment of “nurturantness” based on the mea-surement and analysis of individualized behaviors andcognitions.

Atran and associates conclude that Q’eqchi’ whomoved into central Peten around the 1970s seem unableto adapt to the lowland environment or learn from theirneighbors because of their cultural baggage, which seemsto be suited to highland requirements only (e.g., theirceremonial attachment to sacred highland mountainsdetours access to ecological information relevant to low-land commons survival). Work by Richard R. Wilk (1991)and James R. Gregory (1984) indicates, however, thatQ’eqchi’ who moved into southern Belize roughly a cen-tury earlier along with lowland-trained Mopan seem tohave maintained their highland ceremonial attachmentand survived for so long on Toledo’s poor tropical soilsnot because their agriculture could destroy unlimitedamounts of resources over an extended period of timebut because of their intimate understanding of the re-gional ecology. Is it only time that accounts for thesedifferences?

Reply

scott atran and douglas medinAnn Arbor, Mich., U.S.A. 4 ii 02

Gil-White suggests that our failure to find any evidencefor systematic residual agreement among the Itza’ maybe something of a “ceiling effect,” with knowledge andagreement being so high that the only remaining varia-bility is random variability. This potential problem isnot, we believe, a problem in practice. For example, aprincipal-components analysis of interinformant agree-ment conducted on plant-animal relations for the Itza’reveals a single-factor solution accounting for 46% of thevariance with a mean first-factor score of .67. In otherwords, the average cultural “competence” of our Itza’informants regarding the consensual “cultural model” isa long way from a ceiling effect.

Lack of reliable residual agreement among Itza’ eldersdoes not mean that Itza’ ecological models are not beingtransmitted socially. It may be that knowledge flowsfrom older Itza’ to both Ladinos and younger Itza’. In

ongoing work we are studying ecological models inyounger Itza’ and collecting data on expert and socialnetworks. We are finding that younger Itza’ see reciprocalrelations between plants and animals but that their con-sensual knowledge only partially overlaps with that oftheir elders. More detailed analysis should revealwhether the younger Itza’ are learning from their eldersand whether this learning is best described in terms ofabstract notions like reciprocity or in terms of detailedinteractions involving generic species.

Gil-White objects to our characterization of norms as“rules or principles” that are “functional units of cul-tural transmission and evolution.” Elsewhere (Gil-White2001) he has argued that norms functionally mark ethnicboundaries and, further, that this functional relationshipbetween norms and ethnicity is a direct product of bi-ological and cultural evolution. Boyd and Richerson(2001) have lately reiterated their view of norms as“shared social rules.” All of us agree that some normscan also be dysfunctional. Our argument concerningprestige-bias transmission is that Ladinos learn fromItza’ not through imitation of norms but through atten-tion to social deference patterns that point to (and donot describe) likely sources of relevant observations thatconstrain candidates for relevant inferences. To implythat cultural behaviors (consensual statistical patterns)relating to people’s views of species relations, rankingsof spirit preferences, and the like, pertain to “norms”seems to rob the notion of norm of its analytical use-fulness. This is not to suggest that there are no normsin the functional sense described above. It is only to denythem an exclusive or primary role in cultural transmis-sion, formation, and evolution.

In principle, there is nothing in Boyd and Richerson’sapproach (or Henrich and Gil-White’s) that requiresnorms to be discrete units acquired through imitation-copying, although the mathematical models they havepreferred in the past do sometimes use these simplifyingassumptions. At other times they suggest that sociallearning is indeed based on inferential capacities influ-enced by “cognitive biases” (Boyd and Richerson 1985:70–71) and, more recently, that “mental representationsare not replicated, but rather are ‘reconstructed’ throughan inferential process that is strongly affected by cog-nitive attractors” (Henrich and Boyd n.d.; cf. Sperber1996). Attention to such cognitive processes has beenscarce in the past, but more recent work is encouraging.Examples include Gil-White’s (2001) cross-cultural in-vestigation of the inferential underpinnings of essen-tialism in the formation of ethnicity (cf. Hirschfeld 1996)and Henrich and Boyd’s (n.d.) modeling of the ways inwhich cognitive attractors shape the social learning pro-cess (e.g., with variable or otherwise noisy new infor-mation “snapped” back or forward by cognitive modulesto reduce noise and facilitate mnemonic retention, socialtransmission, and cultural survival of the information).

It appears that some of our disagreements have to domore with matters of word meaning than with mattersof fact. We think that ordinary use of terms like “norm”and “imitation” (in political science, anthropology, so-


ciology, economics) is willfully blind to the causal roleof cognition in cultural formation and evolution. Theusual theories that accord these terms key roles ignorecognitive mechanisms as “proximate mechanisms” inmaking causal descriptions and predictions about the bi-ological and cultural evolution of human behavior (Den-nett 1995:379; Sober and Wilson 1998:182), much as ec-onomic theory often contentiously ignores how mindsactually process preferences and decisions (with simi-larly dubious results for understanding human life).There is a similar expression of disregard for cognitiveprocesses in Richard Dawkins’s (1976, 1999) theory of“memes.” What Dawkins means by “meme” is a unitof culture that may be passed on by imitation (this isalso the primary sense of “meme” as now defined in theOxford English Dictionary). On this view, “memes” sup-posedly replicate in minds and cultures much as genesreplicate in bodies and populations. Memes theoristscustomarily describe memes as self-encapsulated unitsthat maintain structural integrity in ways that are largelyimpervious to exogenous influences, such as the infer-ential structures of the mind that surround and harborthe memes that colonize it (for a fuller discussion, seeAtran 2001a). On this standard view, then, the notion of“belief-memes generated thru inference” comes close tobeing an oxymoron. Given his comments, Gil-Whitepretty clearly does not want to use “meme” (or “imi-tation” or “norm”) with the customary sense. If so, whypersist in using such terms at all?

Haenn is “uncomfortable” with our alleged “separa-tion of culture from other social categories,” our “defi-nition of culture as cognitive models,” and our use of “adefinition of ecology outside people’s models.” She sug-gests that an alternative conception of culture as an “all-encompassing category” that includes “emotions, iden-tities, social histories, social structure, and biophysicalconcerns” may “clarif[y] the apparently irrational be-havior” of the Q’eqchi’. It is difficult to respond to thesecomments, because we don’t recognize the views attrib-uted to us.

First, we nowhere claim that the Q’eqchi’ are or appearto be irrational. On the contrary, from the standpoint ofrational decision theory they are eminently rational (fol-lowing the classical course of the “tragedy of the com-mons”). It is the Itza’ who are apparently irrational inthat they appear to be making efforts to preserve a re-source that they know others destroy; that is, the Itza’are in effect subsidizing those who are degrading theirhabitat.

Second, we nowhere define culture as cognitive mod-els. We imply in this article and make explicit elsewherethat “cultures are causally distributed assemblages ofmental representations, their public expressions, and theresultant behaviors in given ecological contexts.” Morefiguratively (Medin and Atran 2001; cf. Atran 2001a):

Cultural landscapes include ecological contours, sothat the cultural paths chosen in life suggest ananalogy to physical paths (Sperber 1996, Atran n.d.).The existence of a physical path depends upon the

(i) nature of the path’s ecological setting, which con-strains where it can be made to go; (ii) behavioralitineraries that groove the path and determine wherein fact the path leads; and (iii) cognitive models thatgive purpose and direction to the path. Likewise, theexistence of any systematic distribution of ideas andbehaviors, or cultural “path,” results from an inte-gration of cognitive, behavioral, and ecological con-straints that neither resides wholly within mindsnor is recognizable in a world without minds. Cul-tural paths do not exist apart from the individualminds that constitute them and the environmentsthat constrain them, any more than a physical pathexists apart from the organisms that tread it and thesurrounding ecology that restricts its location andcourse.

This “cultural-epidemiological” framework is designedto address a series of methodological issues in psychol-ogy, including (1) the fallacy of conceiving culture to bea well-bounded entity, well-defined system, or indepen-dent variable and (2) the tendency to “essentialize” cul-ture and treat it as an explanation rather than as phe-nomena to be explained. It is also meant to replace (3) atotalizing and analytically opaque notion of “culture”common in anthropology, namely, “as an all-encom-passing category, one that . . . clarifies apparently irra-tional behavior.” Culture is not a thing, with some spe-cial ontological status over and above the individualminds, bodies, behaviors, artifacts, and ecological fea-tures that causally constitute the (statistically identifi-able) cultural pattern, any more than a biological speciesexists over and above the individual organisms-in-their-environments that make up the (statistically identifia-ble) species.

In our analysis of how distinct cultural groups behavedifferently in the same habitat, we have sought to isolatefactors that reliably account for those differences. By iso-lating such factors (e.g., demographic composition, ec-onomic activity) we do not imply that they are not partof the cultural landscape. We only claim to show thatother factors in this landscape (e.g., cognitive models,social patterns of deference) are reliably correlated withand perhaps predict the relevant behavioral differences(concerning physical survival).

As Nigh (2002) points out in this issue, our notion ofMayaland as a “spirited landscape” aims precisely to in-tegrate “the emotions, identities, history, social struc-ture, and biophysical concerns” that Haenn thinksshould be included (Atran 2001b). Still, nothing that wehave learned about Itza’ or Q’eqchi’ emotions, history,social structure, and so forth—and nothing about defi-nitions of culture or ecology—can change this univer-sally apparent human perception: that the local ecologyincludes trees and that if a lot of trees are cut down ina relatively short period of time, then the local ecologyis being destroyed.

When we began the present series of studies over adecade ago, we did not expect the Q’eqchi’ to be destroy-ing the forest as the Itza’ were preserving it. Rather than


attempt to redefine this destructiveness relativisticallyin terms of some as yet unfathomed (and wishfully pos-itive) set of needs or values, we have opted to presentour findings directly to the Q’eqchi’, Ladino, and Itza’communities in day-long meetings. For example, duringa recent session (April 26–27, 2001), the top Itza’ andLadino experts in our studies acknowledged the plausi-bility of the analysis showing the latter’s knowledge tobe a proper subset of the former’s. Q’eqchi’ leaders alsoconfirmed the general reliability of our results andanalyses.

Q’eqchi’ representatives asked for help from Itza’ ontwo counts. First, in light of our findings that smallerfires allow Itza’ to maximize natural fertilizers, includingboth phosphorus and nitrates in upper-level soils, theQ’eqchi’ asked for instruction on Itza’ burning tech-niques. Second, after presentations by Itza’ women onthe advantages of biodiversity for maintaining a livingpharmacopoeia, Q’eqchi’ women asked Itza’ for instruc-tion on which plants to preserve for medicinal uses. Themeeting began and ended with Itza’ and Q’eqchi’ prayers.

Q’eqchi’ did not express interest in conservation assuch but were eager to produce more crops with less landwhile maintaining a richer stock of medicinally usefulplants. They asked that other meetings be organized onsimilar lines, that contacts between the communities bemore regular, and that specimens or photographs anddescriptions of Itza’ medicinal plants be made available.We agreed to prepare a monograph of some 400 species,covering over 100 botanical families, with descriptionsof properties, uses, and accompanying photographs(Atran, Lois, and Ucan Ek’ n.d.).

Finally, Haenn intimates that by including a morethorough analysis of the political context, issues of ra-tionality, utility, culture, and the commons may becomemore transparent and understandable. She is undoubt-edly right, although the parallel that she sees betweenItza,’ Q’eqchi,’ and Ladinos in Peten and Maya, Chol,and Mestizos in Mexico’s Calakmul Reserve is dubious.Members of our research team have also worked withcommunities in the Calakmul area for a number of years,with a goal of facilitating exchange of information abouttraditional medicine between communities (Ucan Ek’,Ortega, and Came 1998). While it is true that the gov-ernments of Quintana Roo and Campeche are often morefavorably disposed to Yukatek Maya than to immigrantChol (Chol are not native to Quintana Roo, and thereare many more Yukatek than Chol voters in Campeche),there are almost no Yukatek-speakers whose immediateancestors were born in the Calakmul area. The focus onYukatek Maya culture in Calakmul is largely a politicalartifact. In Peten, local authorities have not privilegednative Itza’ Maya over other groups (Schwartz 1990). Infact, when members of our team first arrived in the areain the mid-1970s, elderly Itza’ still expressed fear of beingoverheard speaking “ugly words” (palabras feas, i.e., Itza’Maya). Today, the overwhelming bulk of developmentaid in and around the Maya Biosphere Reserve area goesto Ladino and Q’eqchi’ communities. Moreover, as oursocial network studies indicate, it is the Q’eqchi’ who

rely most on contact with outside governmental andnongovernmental agencies in dealings with the forest.The causal effects of this contact on Q’eqchi’ attitudesand behaviors should be explored.

Hostettler echoes Haenn’s concern with factoring outcertain variables (which may not be the right ones) toexplain differences in behavior and with ignoring thepolitical context. Hostettler also dislikes the supposedemphasis on “destructiveness” as a “group-as-a-whole”phenomenon. Hostettler refers to work by Wilk (1991)and Gregory (1984) that seems to suggest that Q’eqchi’who have resided for generations in southern Belize maynot behave destructively toward the local ecology. If thatis so, then what besides time alone makes for an intimateunderstanding of regional ecology? Hostettler suggeststhat a closer look at factors related to the relative scaleand centrality of milpa agriculture or to household com-position might be more telling.

In isolating factors that reliably account for group dif-ferences, we did not make a priori choice to excludefactors such as scale and centrality of milpa agriculture.They were in fact among the first things we looked at.But we found that there were no reliable differences be-tween communities. For example, we analyzed differentsources of household income, of which sale of milpacrops was one. There were no reliable differences con-cerning various aspects of Q’eqchi’, Ladino, and Itza’household composition (number of family members, age-grade, etc.). Nor were there statistically significant dif-ferences overall with respect to scale and centrality ofmilpa. For example, average annual income for maizesales from milpa was 1,806 quetzales for a Q’eqchi’household and 2,065 quetzales for a Ladino household.Although maize sales represent 27% and 41% of totalhousehold income for Q’eqchi’ and Ladinos, respectively,the differences were not reliable (because of wide vari-ation between individuals in each group).

Our measure of “destructiveness” was only one of anumber of converging measures. It was decidedly not thesort of “group-as-a-whole” reduction that Hostettler im-plies. It was computed from three separate measure-ments (land cleared, years cultivated, years fallow) foreach of 44 informants. The group patterns that emergedwere reliable statistical aggregates (as were our othermeasures, which Hostettler seems more comfortablewith), not simple averages. But Hostettler’s more generalargument is a good one: a natural step forward would beto look more deeply into within-culture variation, whichour methodology allows us to explore.

The work of Wilk and Gregory is an important startfor assessing different historical strategies amongQ’eqchi’ communities, but there is not enough quanti-tative evidence for reliable comparative analyses (and themultiplication of anecdotes is not data). Historical re-cords of Q’eqchi’ migrations (communicated by Atran toJones 1998:522) indicate that Q’eqchi’ began settling insouthern Belize shortly after the Spanish conquest (if notbefore). In some communities, highland elders instructedemigrants to the lowlands to construct mounds for wor-ship there (Q’eqchi’ deities typically reside in high


places, which are scarce in many parts of the lowlands).In this way, immigrant Q’eqchi’ could propitiate localdeities and not have to send delegations back to the high-lands to get highland deities to solve lowland problems(e.g., blights, hurricanes). In other communities, otherfactors may be at work, as Hostettler suggests. Yet it isdoubtful that such yet-to-be-discovered factors fromother communities, if included in the analysis of ourstudy sample, would render destruction of trees lessdestructive.

Finally, in line with Hostettler’s recommendation thatmore attention be given to the political context, we ap-peal to the scholarly community to replace the term “It-zaj” that Hostettler and others use with “Itza.’” The useof “Itzaj” dates to the early 1990s, when C. A. Hoflingand the national Academia de Lenguas Mayas de Gua-temala recommended it. Phonetic analysis clearly indi-cates that there is no principled reason for preferring theending “j” over the glottal stop (�); they are alternatingphonemes. More to the point, at the community meetingcalled by the local Academia de Lenguas Mayas de SanJose, Peten, on August 16, 1998, the Itza’ decided thatthey want to be referred to in print as “Itza.’”

Hatano and Takahashi pose the broader practical ques-tion of whether our findings generalize to contexts inwhich local groups must also deal with corporations andother political entities. This may be crucial to the prob-lem of “upscaling the commons.” Even in Peten thereare issues of governmental land use policy as well as theencouragement by numerous NGOs of practices that“foster sustainable economic development.” The pilotdata reported in our paper suggest that NGO mentalmodels of the forest may be relevant to issues of inter-group conflict and negotiation. A more graphic caution-ary note comes from a recent conference we attended onthe future of Peten held in Flores (to which no milperoswere invited). One government speaker touted the ni-trogen-fixing virtues of a particular variety of bean, ap-parently unaware that this bean (Cannavalia ensiformis)grows so thickly that it provides a haven for (poisonous)snakes and requires constant cutting that interferes withother seasonal activities (hunting, chicle collecting, etc.).In short, Hatano and Takahashi are correct that researchon the commons increasingly has to take into accountnonlocal players, and we are attempting to do this.

Where we differ from Hatano and Takahashi is on theiroverreliance on a “conscious-control model” that giveslittle attention to mental models of resources. In closelyrelated work in the U.S.A. we are studying aquatic andforest ecological models among Native American Me-nominee and majority-culture hunters and fishermen ofcentral Wisconsin. We find both consensual cross-cul-tural agreement and systematic disagreement, much aswe find in Peten. The differences are also correlated withintergroup conflict over resource use and pressures to actin concert to prevent sulfide mining by a large multi-national corporation that would threaten the health ofthe Wolf River, which runs through both communities.Although this work is not as far along as our Peten stud-ies, we believe that these cultural models are highly rel-

evant to understanding both local conflict and attemptsto act in concert to fight external threats. In sum, al-though we have begun in each site by assessing localknowledge, beliefs, and practices, we do recognize thatthere are institutional players in almost all cases of re-source management. This makes cultural models more,not less, pertinent.

References Citeda l e x a n d e r , r . 1987. The biology of moral systems. New

York: Aldine de Gruyter.a r i z p e , l . , f . p a z , a n d m . v e l a z q u e z . 1996. Culture

and global change: Social perceptions of deforestation in theLacandona rain forest. Ann Arbor: University of MichiganPress.

a t r a n , s . 1986. Hamula organisation and masha’a tenure inPalestine. Man 21:271–95.

———. 1993. Itza Maya tropical agro-forestry. current anthro-pology 34:633–700.

———. 1998. Folkbiology and the anthropology of science: Cog-nitive universals and cultural particulars. Behavioral and BrainSciences 21:547–609.

———. 1999. “Itzaj Maya folk-biological taxonomy,” in Folkbiol-ogy. Edited by D. Medin and S. Atran. Cambridge: MIT Press.

———. 2001a. The trouble with memes. Human Nature 12:351–81.

———. 2001b. “The vanishing landscape of the Peten Maya low-lands,” in On biocultural diversity. Edited by L. Maffi. Wash-ington, D.C.: Smithsonian Institution Press.

———. n.d. In gods we trust: The evolutionary landscape of re-ligion. New York: Oxford University Press. In press.

a t r a n , s . , x . l o i s , a n d e . u c a n e k ’ . n.d. Plantas delos Mayas Itza’ del Peten. Memoirs of the University of Michi-gan Museum 37. In press.

a t r a n , s . , d . m e d i n , n . ro s s , e . l y n c h , j . c o l e y,e . u c a n e k ’ , a n d v. v a p n a r s k y. 1999. Folkecologyand commons management in the Maya lowlands. Proceedingsof the National Academy of Sciences, U.S.A. 96:7598–7603.

a t r a n , s . , d . m e d i n , e . l y n c h , v. v a p n a r s k y, e .u c a n e k ’ , a n d p . s o u s a . 2001. Folkbiology doesn’tcome from folkpsychology: Evidence from Yukatek Maya incross-cultural perspective. Journal of Cognition and Culture 1:3–43.

a t r a n , s . , a n d e . u c a n e k ’ . 1999. “Classification of use-ful plants among Northern Peten Maya,” in Reconstructingancient Maya diet. Edited by C. White. Salt Lake City: Univer-sity of Utah Press.

a x e l ro d , r . 1997. The complexity of cooperation: Agent-based models of competition and collaboration. Princeton:Princeton University Press.

b a i l e n s o n , j . , m . s h u m , s . a t r a n , d . m e d i n , a n dj . c o l e y. n.d. A bird’s eye view: Biological categorizationwithin and across cultures. Cognition. In press.

b e r k e s , f . , d . f e e n y, b . m c c a y, a n d j . a c h e s o n .1989. The benefit of the commons. Nature 340:91–93.

b e r l i n , b . 1992. Ethnobiological classification. Princeton:Princeton University.

b i r d - d a v i d , n . 1999. “Animism” revisited: Personhood, en-vironment, and relational epistemology. current anthropol-ogy 40:S67–92.

b o s t e r , j . 1986. Exchange of varieties and information be-tween Aguaruna manioc cultivators. American Anthropologist88:428–36.

b o y d , r . , a n d p . r i c h e r s o n . 1985. Culture and the evo-lutionary process. Chicago: University of Chicago Press.

———. 2001. “Norms and bounded rationality,” in The adaptive


toolbox. Edited by G. Gigerenzer and R. Selten. Cambridge:MIT Press.

b ro s i u s , j . p e t e r . 1999. Analyses and interventions: An-thropological engagements with environmentalism. currentanthropology 40:277–309. [nh]

c a r t e r , w. 1969. New lands and old traditions: Kekchi culti-vators in the Guatemalan lowland. Gainesville: University ofFlorida Press.

c a s t e l l o n , m . 1996. Dynamics of deforestation: Q’eqchi’colonists in Guatemala’s Sierra de los Minas, 1964–1995. Ph.D.diss., University of Wisconsin, Madison, Wis.

c h a s e , a . , a n d d . c h a s e . 1989. The investigation of Clas-sic period Maya warfare at Caracol, Belize. Mayab 5:5–18.

c l a r k , c . 2000. Land tenure delegitimation and social mobil-ity in tropical Peten, Guatemala. Human Organization 59:419–27. [uh]

c o w g i l l , u . 1962. An agricultural study of the southernMaya lowlands. American Anthropologist 64:273–86.

c u l b e r t , t . p . , a n d d . r i c e . Editors. 1990. Precolumbianpopulation history in the Maya lowlands. Albuquerque: Uni-versity of New Mexico Press.

d a w k i n s , r . 1976. The selfish gene. New York: Oxford Uni-versity Press.

——— 1999. “Foreword,” in The meme machine, by S. Black-more. Oxford: Oxford University Press.

d e m a r e s t , a . 1993. The violent saga of a Maya kingdom. Na-tional Geographic 183:95–111.

d e n n e t t , d . 1995. Darwin’s dangerous idea. New York:Simon and Schuster.

f a g a n , c . 2000. Cultural and economic constraints to farm-ing in a core-zone community of the Maya Biosphere Reserve.Durham: Duke University Center for Tropical Conservation.

f e d i c k , s . , a n d a . f o r d . 1990. The prehistoric agricul-tural landscape of the central Maya lowlands. World Archae-ology 22:28–33.

f i s k e , a . , a n d p . t e t l o c k . 1997. Taboo trade-offs. Politi-cal Psychology 18:255–97.

f l a n n e ry, k . Editor. 1982. Maya subsistence. New York: Aca-demic Press.

f r a n k , r . 1988. Passions within reason. New York: W. W.Norton.

f r e e m a n , f . , d . w h i t e , a n d ro m n e y. a . k . Editors.1992. Research methods in social network analysis. London:Transaction Books.

f u k u y a m a , f . 1995. Trust. New York: Free Press.g i l - w h i t e , f . 2001. Are ethnic groups biological species to

the human brain? current anthropology 42:515–54.g o m e z - p o m p a , a . , j . f l o r e s , a n d v. s o s a . 1987. The

“pet kot”: A man-made tropical forest of the Maya. Intercien-cia 12:10–15.

g r e e n f i e l d , p . m . 1999. “Cultural change and human devel-opment,” in Development and cultural change: Reciprocal pro-cesses. Edited by E. Turiel, pp. 37–59. New Directions forChild and Adolescent Development 83. [gh, kt]

g r e g o ry, j . r . 1984. The Mopan: Culture and ethnicity in achanging Belizean community. University of Missouri Mono-graphs in Anthropology 7. [uh]

g r u n b e r g , j . , a n d v. r a m o s . 1998. Base de datos sobrepoblacion, tierras y medio ambiente en la Reserva del la Bio-sfera Maya, Peten, Guatemala. Guatemala City: CARE/CONAP.

g ru n b e r g , w. 2000. Modeling deforestation: Risks for theMaya Biosphere Reserve, Guatemala. Tucson: University ofArizona School of Renewable Natural Resources.

h a e n n , n o r a , a n d j u l i a m u r p h y. 1998. Identity, mi-gration, and regional political economy: Being “Mestizo” andbeing “Chol” in Calakmul, Campeche. Paper presented at the97th annual meeting of the American Anthropological Associa-tion, Philadelphia, Pa. [nh]

h a m m e r , m . 1983. “Core” and “extended” social networks inrelation to health and illness. Social Science Medicine 17:

405–11.h a r d i n , g . 1968. The tragedy of the commons. Science 162:

1243–48.h e n r i c h , j . , a n d r . b o y d . 1998. The evolution of con-

formist transmission and the emergency of between-group dif-ferences. Evolution and Human Behavior 19:215–41.

———. n.d. On modeling cognition and culture: Why cultural ev-olution does not require replication of representations. Journalof Cognition and Culture 2. In press.

h e n r i c h , j . , a n d g i l - w h i t e . f . 2001. The evolution ofprestige. Evolution and Human Behavior 22:165–96.

h i r s c h f e l d , l . 1996. Race in the making. Cambridge: MITPress.

h o f l i n g , c . a . 1996. Indigenous linguistic revitalization andoutsider interaction: The Itzaj Maya case. Human Organiza-tion 55:108–16.

h o f l i n g , c . a . , a n d f . t e s u c u n . 1997. Itzaj-Maya-Spanish-English dictionary. Salt Lake City: University of UtahPress.

h o s t e t t l e r , u . 1996. Milpa agriculture and economic diver-sification: Socioeconomic change in a Maya peasant society ofCentral Quintana Roo, 1900–1990s. Ann Arbor: UniversityMicrofilm International. [uh]

———. 2001. “Milpa, land, and identity: A Central Quintana RooMaya community in a historical perspective,” in Maya survi-valism. Edited by U. Hostettler and M. Restall, pp. 239–62.Markt Schwaben, Germany: Verlag A. Saurwein.

———. 2002. Labor regime and social justice: Consequences ofeconomic and social stratification among Maya peasants incentral Quintana Roo, Mexico. Anthropos 97:107–16.

i n g o l d , t . 1996. “The optimal forager and economic man,”in Nature and society: Anthropological perspectives. Edited byP. Descola and G. Palsson. London: Routledge.

i ro n s , w. 1996. “Morality, religion, and human nature,” inReligion and science: History, method, dialogue. Edited by W.M. Richardson and W. Wildman. New York: Routledge.

j o n e s , g . 1998. The conquest of the last Maya kingdom. Stan-ford: Stanford University Press.

k n i g h t , n . , s . a t r a n , j . b a r r e t t , a n d e . u c a ne k ’ . 2001. Understanding the mind of God: Evidence from Yu-katek Maya children. Paper presented at the annual meeting ofthe Society for the Scientific Study of Religion, Columbus,Ohio, October.

l a n s i n g , s . , a n d j . k r e m e r . 1993. Emergent properties ofBalinese water temple networks: Coadaptation on a rugged fit-ness landscape. American Anthropologist 95:97–114.

l o i s , x . 1998. Gender markers as “rigid determiners” of the It-zaj Maya world. International Journal of American Linguistics64:224–82.

l o p e z , a . , s . a t r a n , j . c o l e y, d . m e d i n , a n d e .s m i t h . 1997. The tree of life: Universals of folk-biologicaltaxonomies and inductions. Cognitive Psychology 32:251–95.

l u n d e l l , c . 1938. Plants probably utilized by the Old EmpireMaya of Peten and adjacent lowlands. Papers of the MichiganAcademy of Science, Arts, and Letters 24:37–56.

m a c z , n . , a n d j . g r u n b e r g . 1999. Manual de comunida-des de Peten. Santa Elena: CARE Guatemala/Cooperacion Aus-triaca para el Desarrollo. [uh]

m e d i n , d , a n d s . a t r a n . 2001. The native mind: Biologi-cal categorization, reasoning, and decision making in develop-ment and across cultures. Paper presented to the “Innatenessand Structure of the Mind Project,” Sheffield, U.K., and to theMRC Cognition and Brain Sciences Unit, Cambridge, U.K.,November.

m e d i n , d . , e . l y n c h , j . c o l e y, a n d s . a t r a n . 1997.Categorization and reasoning among tree experts: Do all roadslead to Rome? Cognitive Psychology 32:49–96.

m e d i n , d . , n . ro s s , a n d s . a t r a n . n.d. The role of cul-ture in the folkbiology of freshwater fish. MS.

m e d i n , d . , h . s c h w a r t z , s . b l o k , a n d l . b i r n -b a u m . 1999. The semantic side of decision making. Psycho-nomic Bulletin and Review 6:562–69.

m i l t o n , k a y. 1996. Environmentalism and cultural theory:


Exploring the role of anthropology in environmental discourse.New York: Routledge. [nh]

m u r d o c k , g . 1949. Social structure. New York: Macmillan.n a k a o , k . , a n d a . ro m n e y. 1984. A method for testing

alternative theories: An example from English kinship. Ameri-can Anthropologist 86:668–73.

n a t i o n s , j . 1999. “The uncertain future of Guatemala’s MayaBiosphere Reserve,” in 13 ways of looking at a tropical forest.Washington, D.C.: Conservation International.

n a t i o n s , j . , a n d r . n i g h . 1980. Evolutionary potential ofLacandon Maya sustained-yield tropical forest agriculture. Jour-nal of Anthropological Research 36:1–30.

n a z a r e a , v i r g i n i a , 1999. “A view from a point: Ethnoecol-ogy as situated knowledge,” in Ethnoecology: Situated knowl-edge/local lives. Edited by Virginia Nazarea, pp. 3–20. Tucson:University of Arizona Press. [nh]

n i g h , r . 2002. Maya medicine in the biological gaze: Biopros-pecting research as herbal fetishism. current anthropology43:51–77.

o s t ro m , e . 1998. A behavioral approach to the rational choicetheory of collective action. American Political Science Review92:289–316.

p o s e y, d . 1983. “Indigenous ecological knowledge and devel-opment of the Amazon,” in The dilemma of Amazonian de-velopment. Edited by E. Moran. Boulder: Westview Press.

r a p p a p o r t , r . 1979. “On cognized models,” in Ecology,meaning, and religion. Berkeley: North Atlantic Books.

r e i n a , r . 1967. Milpa and milperos: Implications for prehis-toric times. American Anthropologist 69:1–20.

r e i n i n g , c . , a n d r . h e i n z m a n . 1992. “Nontimber forestproducts in the Peten, Guatemala,” in Sustainable harvest andmarketing of rain forest products. Edited by M. Plotkin and L.Famolare. Washington, D.C.: Island Press.

r e m m e r s , g . , a n d h . d e k o e i j e r . 1992. The t’olche’, aMaya system of communally managed forest belts. Agrofores-try Systems 18:149–77.

r i c e , d . 1993. “Eighth-century physical geography, environ-ment, and natural resources in the Maya lowlands,” in Low-land Maya civilization in the eighth century a.d. Edited by J.Sabloff and J. Henderson. Washington, D.C.: Dumbarton Oaks.

ro g e r s , e . 1998. 4th edition. Diffusion of innovations. NewYork: Free Press.

ro m n e y, a . k . , s . w e l l e r , a n d w. b a t c h e l d e r .1986. Culture as consensus: A theory of culture and informantaccuracy. American Anthropologist 88:313–38.

ro s s , n . n.d. Cognitive aspects of intergenerational change:Mental models, cultural change, and environmental behavioramong the Lacandon Maya. Human Organization. In press.

s a d e r , s . 1999. “Deforestation trends in northern Guatemala:A view from space,” in 13 ways of looking at a tropical forest:

Guatemala’s Biosphere Reserve. Washington, D.C.: Conserva-tion International.

s a n t l e y, r . , t . k i l l i o n , a n d m . l y c e t t . 1986. On theMaya collapse. Journal of Anthropological Research 42:123–59.

s c h a c k t , j . 1984. The Tzuultak’a: Religious lore and culturalprocesses among the Kekchi. Belizean Studies 12:16–29.

s c h w a r t z , n . 1985. A note on “weights, measures,” andswidden. Culture and Agriculture 3:9–12.

———. 1990. Forest society. Philadelphia. University of Pennsyl-vania Press.

———. 1995. “Colonization, development, and deforestation inPeten, Northern Guatemala,” in The social causes of environ-mental destruction in Latin America. Edited by M. Painter andW. Durham. Ann Arbor: University of Michigan Press.

s h r i a r , a . n.d. Resource conservation and rural neglect: Anexample from Peten, Guatemala. Delaware Review of LatinAmerican Studies. In press.

s o b e r , e . , a n d d . s . w i l s o n . 1998. Unto others. Cam-bridge: Harvard University Press.

s p e r b e r , d . 1996. Explaining culture: A naturalistic approach.Oxford: Blackwell.

s t e w a r t , s . 1980. Grammatica Kekchi. Guatemala City: Edi-torial Academica Centro Americana.

t u r n e r , b . l . , i i . 1978. “The development and demise ofthe swidden thesis of Maya agriculture,” in Pre-Hispanic Mayaagriculture. Edited by P. Harrison and B. L. Turner II. Albu-querque: University of New Mexico Press.

u c a n e k ’ , e . , l . m . o r t e g a , a n d h . c a m e . 1998.Taller de medecina natural y tradicional: Memorias de ZohLaguna, Calakmul, Campeche, Mexico, 1994–1997. Merida:Herbolaria Maya y Pronatura Penınsula de Yucatan, A.C.

w a s s e r m a n , s . , a n d k . f a u s t . 1994. Social networkanalysis: Methods and applications. Cambridge: CambridgeUniversity Press.

w e l l m a n , b . 1979. The community question. American Jour-nal of Sociology 84:1201–31.

w h i t e , s . 1994. Testing an economic approach to resource di-lemmas. Organization and Human Decision Processes 58:428–56.

w i l k , r . r . 1991. Household ecology: Economic change anddomestic life among the Kekchi Maya of Belize. Tucson: Uni-versity of Arizona Press. [uh]

w i l s o n , r . 1995. Maya resurgence in Guatemala. Norman:University of Oklahoma Press.

w i m m e r , h . , a n d j . p e r n e r . 1983. Beliefs about beliefs:Representation and constraint function of wrong beliefs inchildren’s understanding of deception. Cognition 13:103–28.

w i s e m a n , f . 1978. “Agricultural and historical archaeology ofthe Maya lowlands,” in Pre-Hispanic Maya agriculture. Editedby P. Harrison and B. L. Turner II. Albuquerque: University ofNew Mexico Press.

Documents

Folkecology, Cultural Epidemiology ... - ARTIS International · operative institutions) show systematic awareness of ecological complexity involving animals, plants, and people and