Exact and approximate arithmetic in an Amazonian indigene group with a reduced number lexicon Pierre Pica1, Cathy Lemer2, Véronique Izard2, and Stanislas Dehaene2,* 1 Unité Mixte de Recherche 7023 « Structures Formelles du Langage », CNRS and Paris VIII University, Paris, France 2 Unité INSERM 562 “Cognitive Neuroimaging”, Service Hospitalier Frédéric Joliot, CEA/DSV, 91401 Orsay cedex, France. * Corresponding author Abstract Is calculation possible without language? Or is the human ability for arithmetic dependent on the language faculty? To clarify the relation between language and arithmetic, we studied numerical cognition in speakers of Mundurukú, an Amazonian language with a very small lexicon of number words. Although the Mundurukú lack words for numbers beyond five, they are able to compare and add large approximate numbers, way beyond their naming range. However, they fail in exact arithmetic with numbers larger than 4 or 5. Our results imply a distinction between a universal system of number approximation, and a language-based counting system for exact number and arithmetic. One-sentence summary Psychological experiments in speakers of Mundurukú, an Amazonian language where number words exist only up to five, indicate that they can perform approximate calculations, but are unable to calculate with exact numbers. 1 All science requires mathematics. The knowledge of mathematical things is almost innate in us. . . . This is the easiest of sciences, a fact which is obvious in that no one’s brain rejects it; for laymen and people who are utterly illiterate know how to count and reckon. Roger Bacon Where does the human competence for arithmetic arise from? Two theories have been proposed. Some authors propose that humans, like many animals share a non-verbal “number sense”, an evolutionarily ancient capacity to process approximate numbers without symbols or language (1-3). Mathematics would arise from this non-verbal foundation. Others, however, postulate that language plays a special role in arithmetic. Numerical competence would arise from a progressive linguistic construction, supported by verbal counting and the recursive character of the language faculty (4-7). In the absence of language, numerical competence would therefore be drastically limited. To elucidate the relations between language and arithmetic, numerical competence must be studied in situations in which the language of numbers is either absent or reduced. In many animal species, as well as in young infants prior to the acquisition of number words, behavioral and neurophysiological experiments have revealed rudiments of arithmetic (8-12). Infants and animals only appear to represent the first three numbers exactly. Beyond this range, they can only approximate “numerosity”, with a fuzziness that increases linearly with the size of the numbers involved. This and other neuroimaging and neuropsychological experiments have yielded a tentative reconciliation of the above two theories: exact arithmetic would require language, while approximation would not (13-16). This conclusion, however, has been challenged by a few case studies of adult brain-lesioned or autistic patients in whom the lack of language did not abolish exact arithmetic, suggesting that even complex calculation might also be independent of language (17). 2 In the final analysis, the debate cannot be settled by studying subjects who are raised in a culture teeming with spoken and written symbols for numbers. What is needed is a language deprivation experiment, in which neurologically normal adults would be raised without number words or symbols. While such an experiment is ethically impossible in our Western culture, some languages are intrinsically limited in their ability to express number, sometimes using a very narrow set of number words (“one, two, many”). These often endangered languages present a unique opportunity to establish the extent and limits of non- verbal arithmetic abilities. Here, we study numerical cognition in native speakers of Mundurukú, a language which has number words only for the numbers one through five (18, 19). Mundurukú is a language of the Tupi family, spoken by approximately 7,000 people living in an autonomous territory in the Para state of Brasil (figure 1). Following two pilot trips in 2001 and 2002, one of us (P.P.) travelled through several villages during 2003 and was able to collect data from 55 speakers of Mundurukú in a computerized battery of numerical tests. Ten native speakers of French (mean age 50) served as controls. The Mundurukú entertain contacts with non- indigenous culture and individuals, mainly through government institutions and missionaries. Thus, several of them speak some Portuguese. A few, especially the children, now receive basic instruction. Thus, we distinguished six groups of subjects based on their age, bilingualism, and level of instruction (figure 1). Using a solar-powered laptop computer, we managed to collect a large amount of trials in classical arithmetical tasks, including a chronometric comparison test. This allowed us to test whether classical effects such as Weber’s law and the distance effect, which characterize numerical cognition in Western subjects (3, 20-23), remain present in the absence of a well-developed language for number. A first task explored the verbal expressions for numbers in Mundurukú (24). Participants were presented with displays of 1-15 dots, in randomized order, and were asked 3 to say in their native language how many dots were present, thus permitting an objective analysis of the conditions of use of number words. No systematic variation across groups was identified, except for lack of use of the word for “five” in the younger children, and the results were therefore pooled across all groups (figure 2). The results confirm that Mundurukú has frozen expressions only for numbers 1-5. These expressions are long, often having as many syllables as the corresponding quantity. Given the universal correlation between word length and word frequency (25), this finding suggests that numerals are infrequently used in Mundurukú. The words for three and four are polymorphemic: ebapũg=2+1, ebadipdip=2+1+1, where “eba” means “your (two) arms”. This possibly reflects an earlier base-2 system common in Tupi languages. Above 5, there was little consistency in language use, with no word or expression representing more than 30% of productions to a given target number. Participants relied on approximate quantifiers such as “few” (adesũ), “many” (ade), or “a small quantity” (bũrũmaku). They also used a broad variety of expressions varying in attempted precision, such as “more than one hand”, “two hands”, “some toes”, all the way up to long phrases such as “all the fingers of the hands and then some more” (in response to 13 dots). Crucially, the Mundurukú did not use their numerals in a counting sequence, nor to refer to precise quantities. They usually uttered a numeral without counting, although if asked to do so they could count very slowly and non-verbally by matching their fingers to the set of dots. Our measures confirm that they selected their verbal response based on an apprehension of approximate number rather than on an exact count. With the possible exception of “one” and “two”, all numerals were used in relation to a range of approximate quantities rather than a precise number (figure 2). For instance, the word for five, which can be translated as “one hand” or “a handful”, was used for 5, but also 6, 7, 8 or 9 dots. Conversely, when 5 dots were presented, the word for “five” was uttered only on 28% of trials, while the words “four” and 4 “few” were each used on about 15% of trials. This response pattern is comparable to the use of round numbers in Western languages, for instance when we say “ten people” when there are actually 8 or 12. We also noted the occasional use of two-word constructions (e.g. “two- three seeds”) which have been analyzed as permitting reference to approximate quantities in Western languages (26). Thus, the Mundurukú are only different from us in failing to count and in allowing approximate use of number words in the range 3-5, where Western numerals usually refer to precise quantities. If the Mundurukú have a sense of approximate number, they should be able to process numbers non-verbally way beyond the range for which they have number words. If, however, concepts of numbers emerge only when number words are available, then the Mundurukú would be expected to be at chance level with large numbers. We tested this alternative using two estimation tasks. First, we examined number comparison, a task which, in Western subjects, has revealed an effect of numerical distance whether the targets are presented as sets of objects or symbolically as Arabic digits (20, 21). Participants were presented with two sets of up to 80 dots, and were asked to point to the more numerous set (figure 3a). Mundurukú participants responded way above chance level in all groups (minimum 70.5 % correct in the younger group; p<0.0001). There was no significant difference among the six groups of Mundurukú subjects, suggesting that the small level of bilingualism and instruction achieved by some of the participants did not modify performance. Although average Mundurukú performance was slightly worse than the French controls, thus creating a difference between groups (F =2.58, p<0.028), this appeared to be due to a small proportion of trials with 6,55 random responses, distributed throughout the experiment, and probably due to distraction in some Mundurukú participants (this was the first test that they took). Crucially, performance varied significantly as a function of the ratio of the two numbers, thus revealing the classical distance effect in Mundurukú subjects (F =43.2 3,138 5 p<0.0001). This effect was identical in all groups, including the French controls (group X distance interaction, F<1; see figure 3a). Response times were also faster for more distant numbers in Munduruku (F =12.9, p<0.0001, and F =4.93, p<0.008). Again, although the 3,90 3,26 French controls were globally faster, thus creating a main effect of group (F =4.59, 6,37 p<0.002), the distance effect was parallel in all groups (interaction F<1). Fitting the performance curve suggested a Weber fraction of 0.17 in Mundurukú, only marginally larger than the value of 0.12 observed in the controls. Thus, the Mundurukú clearly can represent very large numbers and understand the concept of relative magnitude (27). We then investigated whether the Mundurukú can perform approximate operations with large numbers. We used a non-symbolic version of the approximate addition task, which is thought to be independent of language in Western subjects (13-15). Participants were presented with simple animations illustrating a physical addition of two large sets of dots into a can (figure 3b). They had to approximate the result and compare it to a third set. All groups of participants, including monolingual adults and children, performed considerably above chance (minimum 80.7% correct, p<0.0001). Performance was again solely affected by distance (F =78.2, p<0.0001), without any difference between groups nor a group by 3,152 distance interaction (see figure 3) (28). If anything, performance was higher in this addition+comparison task than in the previous comparison task, perhaps because the operation was represented more concretely by object movement and occlusion. Mundurukú participants had no difficulty in adding and comparing approximate numbers, with a precision identical to that of the French controls. Finally, we investigated whether the Mundurukú can manipulate exact numbers. The number sense view predicts that outside the language system, number can only be represented approximately, with an internal uncertainty that increases with number (Weber’s law). Beyond the range of 3-4, this system cannot reliably distinguish an exact number n from its 6 successor n+1. Thus, the Mundurukú should fail with tasks that require manipulation of exact numbers such as “exactly six”. To assess this predicted limitation of Mundurukú arithmetic, we used an exact subtraction task. Participants were asked to predict the outcome of a subtraction of a set of dots from an initial set comprising from 2 to 8 items (figure 3c and d). The result was always small enough to be named, but the operands could be larger (e.g. 6-4). In the main experiment, for which we report statistics below, participants responded by pointing to the correct result amongst three alternatives (0, 1 or 2 objects left). The results were also replicated in a second version in which participants named the subtraction result aloud (figure 3d). In both tasks and in all Mundurukú groups, we observed a collapse of performance when the initial number exceeded 5, which marks the end of the naming range in Mundurukú (F =44.9, p<0.0001). This collapse contrasted sharply with the good performance observed 7,336 in the French controls, which was only slightly affected by number size (F =2.36, p<0.033). 7,63 Thus, we observed a highly significant group effect (F =9.10, p<0.0001) and a group by 6,57 size interaction (F =2.44, p<0.0001). The Mundurukú’s failure was not due to 42,399 misunderstanding of the instructions, because they performed better than chance and indeed close to ceiling when the initial number was below 4, and thus could be named relatively precisely (see figure 2). In fact, performance remained slightly above chance for all values of the initial number (e.g. 49.6% correct for 8-n problems, chance = 33.3%, p<0.0001). The entire performance curve over then range 1-8 could be fitted by a simple psychophysical equation which supposes an approximate Gaussian encoding of the initial and subtracted quantities, followed by subtraction of those internal magnitudes and classification of the fuzzy outcome into the required response categories (0, 1 or 2). Thus, the Mundurukú still deployed approximate representations, subject to Weber’s law, in a task that the French controls easily resolved by exact calculation. 7 Altogether, our results bring some light on the issue of the relation between language and arithmetic. They suggest that a basic distinction must be introduced between approximate and exact mental representations of number, as also suggested by earlier behavioral and brain- imaging evidence (13, 15). The Mundurukú have no difficulty with approximate numbers. They can mentally represent very large quantities of up to 80 dots, way beyond their naming range. They also spontaneously apply concepts of addition, subtraction and comparison to these approximate representations. This is true even for monolingual adults and young children that never learned any formal arithmetic. Those data add to previous evidence that number approximation is a basic competence, independent of language, and available even to preverbal infants and many animal species (8-12). What the Mundurukú appear to lack, however, is a fast apprehension of exact numbers. Our results thus support the hypothesis that language plays a special role in the emergence of exact arithmetic during child development (29-31). What is the mechanism for this developmental change? It is noteworthy that the Mundurukú have number names up to 5, and yet use them approximately in naming. Thus, the availability of number names, in itself, may not suffice to create a mental representation of exact number. More crucial perhaps is that the Mundurukú do not have a counting sequence of numerals. Although some possess a rudimentary ability to count on their fingers, it is rarely used. By requiring an exact one-to- one pairing of objects with the sequence of numerals, verbal counting may promote a conceptual integration of the approximate number representation with the discrete object representation system (29, 30). Around the age of 3, Western children exhibit an abrupt change in number processing as they suddenly realize that each count word refers to a precise quantity (31). This “crystallization” of discrete numbers out of an initially approximate continuum of numerical magnitudes does not seem to occur in the Mundurukú. 8 Acknowledgements This work was developed as part of a larger project on the nature of quantification and functional categories developed jointly with the linguistic section of the Departement of anthropology of the National museum of Rio and the Unité Mixte de Recherche 7023 of the CNRS, with the agreement of Funai and CNPQ. It was supported by INSERM, CNRS, the French Ministry of Foreign Affairs (P.P.), and a McDonnell Foundation centennial fellowship (S.D.). We are grateful to Manuela Piazza for her help with stimulus design, and to Celso Tawe and Venancio Poxõ for help in testing. 9 Figure legends Figure 1. Location of the main Mundurukú territory where our research was conducted. Colored dots indicate the villages where participants were tested. The legend at bottom gives the size of the six groups of subjects and their average age. Figure 2. Number naming in Mundurukú. Participants were shown sets of 1-15 dots in random order, and were asked to name the quantity. The graph shows the frequency with which a given word or locution was used in response to a given stimulus number. We only present the data for words or locutions produced on more than 2.5 % of all trials. Above 5, frequencies do not add up to 100%, because many participants produced rare or idiosyncrasic locutions or phrases such as “all of my toes” (a complete list is available from the authors). Figure 3. Performance in four tasks of elementary arithmetic. In each case, the left column illustrates a sample trial. The graphs at right show the percentage of correct trials, in each group separately (M=monolinguals, B=bilinguals, NI=no instruction, I=instruction) as well as averaged across all the Mundurukú and French participants (right graphs). The lowest level on the scale always corresponds to chance performance. For number comparison (top two graphs), the relevant variable that determines performance is the distance between the numbers, as measured by the ratio of the larger to the smaller number. For exact subtraction (bottom two graphs), the relevant variable is the size of the initial number n1. The fits are based on mathematical equations described in Supplementary Online Material. 10