- Email: [email protected]
Sequential use of rigid and pliable tools in tufted capuchin monkeys (Sapajus spp.)
Animal Behaviour 87 (2014) 213e220
Contents lists available at ScienceDirect
Animal Behaviour journal homepage: www.elsevier.com/locate/anbehav
Sequential use of rigid and pliable tools in tufted capuchin monkeys (Sapajus spp.) Gloria Sabbatini a, *, Héctor Marín Manrique b, c, Cinzia Trapanese a, d, Aurora De Bortoli Vizioli a, Josep Call b, Elisabetta Visalberghi a a
Institute of Cognitive Sciences and Technologies of CNR, Rome, Italy Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany Department of Psychology, University of the Balearic Islands, Palma, Spain d Department of Biology, University of Tor Vergata, Rome, Italy b c
a r t i c l e i n f o Article history: Received 7 August 2013 Initial acceptance 3 September 2013 Final acceptance 9 October 2013 Available online 3 December 2013 MS. number: 13-00658 Keywords: meta tool use monkeys motor planning object properties primates problem solving
Previous studies have shown that a few bird and primate species successfully use two or more tools sequentially to obtain a goal. Our study incorporated some novel features: we aimed to assess whether captive capuchins (Sapajus spp.) are able to use tools differing in rigidity sequentially to obtain a reward. A 90 angled Plexiglas tube filled with yoghurt and a platform were placed 121 cm apart in front of the test compartments. The experimenter showed the subject the functional properties of a tool (rigidity or flexibility) as well as its nonfunctional ones (i.e. shape, colour, etc.) then deposited it on a platform out of the subject’s reach. Then the experimenter handed the subject another tool (rigid or flexible). To obtain the reward, the subject had to insert the flexible tool inside the tube. Four conditions (2 2 design) were presented. Capuchins used the tools in sequence to get the reward when necessary (nine out of 10 subjects on the first trial) and not as a result of a simple heuristic of using the tool placed nearest to the food. Whenever the platform tool was unnecessary, capuchins used the flexible tool already within reach. These results demonstrate that capuchins are able to organize their actions in the hierarchically correct sequence using tools differing in their rigidity. Ó 2013 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Using an object to alter the form, position or condition of another object is widespread in the animal kingdom (Shumaker, Walkup, & Beck, 2011). However, only a few individuals of a limited number of primate and corvid species show sequential tool use in which a tool is used to obtain another tool, which subsequently will serve to obtain an out-of-reach goal. According to Taylor, Hunt, Holzhaider, and Gray (2007) sequential tool use is challenging because the subject must (1) recognize that tools can be used on a nonfood item, (2) resist the immediate motivation to use the tool to attempt to reach the food directly, and (3) be able to organize her/his behaviour hierarchically. A task commonly used in the laboratory to test sequential tool use consists of presenting subjects with a reward that is out of reach and a readily available tool that is not long enough to reach the reward but long enough to reach another tool, which can be used to reach the reward. Subjects have to use the tools sequentially: first the shorter one to retrieve the longer one, and then the latter to recover the food. Spontaneous use of up to two tools in
* Correspondence: G. Sabbatini, Istituto di Scienze e Tecnologie della Cognizione, Consiglio Nazionale delle Ricerche, Via Ulisse Aldrovandi 16/B, 00197 Rome, Italy. E-mail address: [email protected] (G. Sabbatini).
sequence has been reported in chimpanzees, Pan troglodytes (Köhler, 1925/1976), gorillas, Gorilla gorilla, and orang-utans, Pongo pygmaeus (Mulcahy, Call, & Dunbar, 2005). More recently, spontaneous use of up to five stick tools in sequence has been reported in great apes (Martin-Ordas, Schumacher, & Call, 2012). After receiving some training, Japanese macaques, Macaca fuscata (Hihara, Obayashi, Tanaka, & Iriki, 2003) and cottontop tamarins, Saguinus oedipus (Santos, Rosati, Sproul, Spaulding, & Hauser, 2005) can also use tools in sequence. Capuchin monkeys (Sapajus spp.) spontaneously use two tools in sequence (Anderson & Henneman, 1994; Klüver, 1933) but can also be trained to use up to eight rake tools in sequence (Warden, Koch, & Fjeld, 1940). Moreover, wild capuchins were observed combining the use of two tools, stones and sticks or two stones, in sequence (Mannu & Ottoni, 2009). Sequential tool use is not limited to primates. New Caledonian crows, Corvus moneduloides, spontaneously use a short stick to extract a longer stick (placed in a box), which they use to extract meat from a vertical tube (Taylor et al., 2007). More recently, Taylor, Elliffe, Hunt, and Gray (2010) presented seven New Caledonian crows with a task requiring (1) pulling up a string to obtain a short stick, (2) using the short stick to extract a longer stick from a toolbox, and finally (3) using the long stick to extract food from a hole. Crows with and without previous experience with the task
0003-3472/$38.00 Ó 2013 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.anbehav.2013.10.033
214
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220
successfully solved the problem. Wimpenny, Weir, Clayton, Rutz, and Kacelnik (2009) reported that New Caledonian crows were able to use up to three tools in sequence to get the reward. Here, crows were presented with an out-of-reach reward, two tools too short to reach the reward and four out-of-reach tools differing in length (and therefore functionality). The distance to the food and/ or which tools were required to get it varied according to the experimental condition. However, only the subjects with pretest experience of extracting tools from tubes and of using tools to extract food from another tube, solved the task. Rooks, Corvus frugilegus, a nontool-using species, have been reported to use tools sequentially by spontaneously dropping a large stone into a container to release a small stone, which was then used to acquire food (Bird & Emery, 2009). Disentangling the cognitive processes involved in sequential tool use is difficult (Wimpenny et al., 2009) because of several methodological limitations present in previous studies. Our study was meant to control some of them. In some studies (Hihara et al., 2003; Parker & Potì, 1990; Santos et al., 2005) the out-of-reach tool and the reward were in proximity so that actions directed to the reward could accidentally be directed to the out-of-reach tool. To prevent this, in our experiment the out-of-reach tool was not positioned between the subject and the reward, but in the opposite direction. In some studies the unreachable object was always the correct tool to solve the problem and there were no control conditions (Anderson & Henneman, 1994; Hihara et al., 2003; Parker & Potì, 1990; Santos et al., 2005; Taylor et al., 2010). Wimpenny et al. (2009) argued that to qualify as goal directed, tool retrieval should occur only when required. Therefore, we included control conditions to test whether tool retrieval did reflect an active choice. Finally, whereas all previous studies used tools that varied in visual properties, namely length or size, we tested capuchin monkeys with two tool tasks simultaneously and with tools differing in rigidity, a nonvisual property. In our study capuchins were required to use a rigid tool to retrieve a flexible tool from a platform and use the flexible tool to obtain the yoghurt available at the bottom of a 90 angled tube. We chose rigidity as the nonvisually detectable functional property since in previous studies it has been shown that capuchin monkeys are able to detect objects’ properties efficiently by nonvisual functional clues (friability, rigidity, weight; e.g. Manrique, Sabbatini, Call, & Visalberghi, 2011; Visalberghi et al., 2009). In particular, Manrique et al. (2011) found that subjects were able to infer an object’s property by observing an experimenter repeatedly bending the tool or by directly manipulating it. When facing an outof-reach reward capuchins efficiently used information previously gathered about tool affordances and selected the appropriate tool above chance levels (i.e. rigid or flexible depending on the task), thus demonstrating a good appreciation of the relation between the tool’s properties and the task requirements. We took advantage of Manrique et al.’s (2011) experimental setup to assess whether capuchin monkeys are able to use tools sequentially by taking into account the functional properties of the tools (i.e. rigidity or flexibility) in relation to the requirements of the task at hand, and hierarchically organize their actions. Unlike previous studies investigating sequential tool use, we had two tasks, each relying on opposite properties of the tool (rigid versus flexible). Depending on the condition, subjects were required to use a flexible tool to recover the yoghurt from a tube or to use a rigid tool to retrieve a second flexible tool from a rectangular platform and use it to obtain the yoghurt. Hence, when subjects were handed one tool, this tool was both functional and nonfunctional, the functionality being determined by the apparatus the subjects needed to reach. This brought about an additional difficulty to the sequential tool-using task that needed to be overcome. In particular, when a flexible tool was given to the subject, then the reward
inside the tube was directly obtainable. When a rigid tool was given, the subject could not reach the reward and therefore should first recover the flexible tool from the platform. We expected capuchin monkeys to use first the rigid tool to get the flexible tool, and then the flexible one to reach the reward. When the two tools available were rigid, subjects should not have attempted to use the tools because neither of them was suitable to get the reward. METHODS Subjects and Housing Ten tufted capuchin monkeys (Sapajus spp., formerly attributed to the Cebus genus; see Lynch Alfaro, Boubli, et al., 2012; Lynch Alfaro, De Souza, Silva, & Rylands, 2012) housed at the Primate Centre of the Institute of Cognitive Sciences and Technologies, CNR in Rome, Italy, participated in this experiment. They were six males and four females ranging in age from 11 to 28 years. Four subjects were mother reared and six hand reared. Group A consisted of six individuals, group B five individuals and group C 11 individuals. The groups were housed in enclosures consisting of an outdoor area (group A ¼ 106.5 m3, group B ¼ 127.4 m3, group C ¼ 374 m3) which was furnished with ropes, climbing frames, slides and wood chips, and two indoor cages (overall about 24.5 m3) which were furnished with perches and slides. Attached to one of the indoor cages, there was a tunnel made from three identical wire-mesh compartments (a, b and c; see below). Capuchins were tested individually in the indoor area, to which they had access through a sliding door from the adjacent outdoor enclosure. Each subject was separated on a voluntary basis from the group solely for the purpose of testing without causing her/him any apparent distress, just before each test session. All procedures complied with protocols approved by the Italian Health Ministry (Licence no. 12/2011-C) and were performed in full accordance with the European law on humane care and use of laboratory animals. Moreover, we adhered to the ASAB/ ABS Guidelines for the use of animals in research. Nine out of 10 capuchins had experience with Manrique et al.’s (2011) tool tasks and seven with other tool tasks as well (see Table 1). Only one capuchin was naïve to these tool-using experiments. Monkey chow (Altromin-A pellets, Rieper standard diet for primates), fresh fruits and vegetables were given every afternoon after testing. Water was freely available at all times. Apparatus and Tools The test area comprised three identical compartments (each 60 75 cm and 75 cm high) separated by sliding lockable doors (Fig. 1). We used two different apparatuses: a transparent Plexiglas
Table 1 Demographics of subjects and their previous experience with tool-using tasks Name
Group
Sex
Age (years)
Rearing
Previous experience
Carlotta Gal Paprica Pedro Pepe Roberta Robin Hood Rucola Sandokan Vispo
A A A C B B C C B B
F M F M M F M F M M
28 22 23 11 25 26 15 12 12 12
Hand Hand Hand Mother Hand Hand Mother Mother Hand Mother
1, 2, 2, 5 2, 3, 5
3, 4, 5, 6 5, 6 5, 6 5 5, 6
5, 6 5, 6
(1) Visalberghi and Trinca, 1989. (2) E. Visalberghi, unpublished data (3) Visalberghi and Limongelli, 1994. (4) Parker and Potì, 1990. (5) Manrique et al., 2011. (6) Sabbatini et al., 2012.
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220
215
Figure 1. Schematic representation of the test set-up. A capuchin monkey is shown in the central compartment (b) waiting for the tool held in the experimenter’s hand. After having received the tool the subject had to decide whether to try to insert it inside the tube in compartment a or to try to retrieve the demonstrated tool from the platform. The sliding doors separating compartments a, b, c were open at the time the subject decided what to do.
tube and a rectangular platform (see Fig. 1). The transparent Plexiglas tube (4 cm in diameter) presented an angle of approximately 90 ; it was open at one end (top) and closed at the other (bottom). The tube was attached to the wire mesh of compartment a (the furthest from the indoor cage), and filled with yoghurt. To reach the yoghurt the subject needed to choose a flexible tool that could bend at the tube’s angle. After removing the tool, the subject could lick the yoghurt off it. The rectangular platform (80 50 cm) was located in front of compartment c (the closest to the indoor cage). An out-of-reach item (a banana slice during the pretest and a tool during the test) was positioned on the platform. The item was outside the subject’s reach since it was placed beyond her/his arm length, at a distance of about 30 cm for females and 35 cm for males. To reach the item the subject needed a rigid tool. During the pretest, the presence of the tube and the platform were alternated whereas they were present at the same time during the test. Sets of tools were composed of four visually distinct tools; all were 38 cm long and differed in colour, material, diameter, weight and rigidity (Fig. 2). Four different sets of four tools each were employed in this experiment (for a total of 16 different tools). Procedure Pretest The pretest consisted of familiarization and choice trials. Each capuchin was given familiarization trials in each of the following tasks: (1) getting, or trying to get, the yoghurt inside the tube with each tool in the sets and (2) getting, or trying to get, the out-ofreach slice of banana placed on the platform with each tool in the sets. For each task the familiarization trials were carried out in blocks of four (one trial for each tool of the set). The two tasks were presented in alternating blocks. During each familiarization trial, the experimenter manipulated each tool for 5 s in front of the subject and then gave the tool to the subject to use. The subject could retrieve the food only when the tool was suited for the presented task. After each familiarization block, to assess whether the subject could choose the appropriate tool for the task, the experimenter allowed her/him to choose between one rigid and one flexible tool. Each rigid tool was paired with each flexible tool of the set. Two choice trials were given after each block. To advance to the test phase, from each of the four sets of tools the subject had to choose the correct tool in 16 consecutive choice trials. To sum up, before
the test, each subject had to reach the criterion for each tool set before moving onto the next, and reach the criterion of 16 correct tool choices for the four sets of tools. Test The experimenter placed the Plexiglas tube in front of compartment a and the platform in front of compartment c and showed the rigidity properties (bending and unbending it) of a tool for 4e5 s (see Supplementary material). Immediately after this demonstration the tool was deposited on the platform out of the subject’s reach. Then the experimenter handed another tool to the subject through the wire mesh of compartment b. At this point, the subject had to choose between using the tool handed to him/her and getting the tool from the platform. There were four different experimental conditions according to the rigidity/flexibility of the tool placed on the platform (after demonstration) and the tool given to the subject: (1) Platform: Flexible tool/Handed: Rigid tool (PlatFlex/HandRig); (2) Platform: Flexible tool/Handed: Flexible tool (PlatFlex/HandFlex); (3) Platform: Rigid tool/Handed: Flexible tool (PlatRig/HandFlex); (4) Platform: Rigid tool/Handed: Rigid tool (PlatRig/HandRig). Capuchins received four sessions of four trials each, for a total of 16 trials. Each session was presented on a different day. The order of presentation of each condition was randomly assigned within one session and each tool set was presented in the four conditions in only one session. The order in which sets were administered was counterbalanced across subjects, three subjects receiving sets in the order 1-2-3-4, three in the order 2-3-4-1, two in the order 3-4-1-2 and two in the order 4-1-23. Before the first trial of the test phase, each subject was given a 1 min period of familiarization with the experimental set-up in the absence of the second tool (i.e. the one handed to the subject). Each trial lasted 5 min. Tests were conducted between December 2011 and June 2012. Data Scoring and Analyses We videotaped all test trials and scored how the subject used the tool(s) to retrieve, or try to retrieve, the reward. In particular, we scored whether the subject: (1) inserted, or tried to insert, the tool handed to him/her or the platform tool inside the tube to retrieve the yoghurt; (2) raked in, or tried to rake in, the platform tool with the tool handed to him/her; and (3) discarded the tool inside or outside the compartments.
216
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220
Figure 2. Sets of tools employed in the experiment. Each set of tools consisted of two flexible (a, b) and two rigid tools (c, d), all of them 28.5 cm long. Set 1: (a) cream phone cord, 1.2 cm in diameter; (b) golden electric wire covered by a transparent plastic shaft, 0.5 cm in diameter; (c) black wooden stick (1.5 1.5 cm); and (d) copper pipe, 1 cm in diameter. Set 2: (a) grey plastic cable collector, 1.8 cm in diameter; (b) orange rope, 0.8 cm in diameter; (c) silver metal bar, 1.3 cm in diameter and (d) brown rounded wooden stick, 1.5 cm in diameter. Set 3: (a) black electric wire, 0.5 cm in diameter; (b) cream rope, 1.5 cm in diameter; (c) plastic-covered green metal stick, 1.5 cm in diameter; and (d) yellow dowel, 2 cm in diameter. Set 4: (a) black plastic clamp (0.5 0.2 cm); (b) rope wrapped in grey duct tape, 0.8 cm in diameter; (c) white rounded plastic stick, 2 cm in diameter; and (d) light brown U-shaped metal rail (1 0.8 cm).
Dependent measures were the percentage of trials in which subjects displayed each of these behaviours. We paid particular attention to the performance on the first trial of each condition. C.T. scored all trials and A.D.B.V. scored a random selection of 20% of the trials to calculate interobserver reliability. Interobserver reliability was perfect for all behaviours (Cohen’s kappa ¼ 1). We analysed the data using two-tailed nonparametric statistics. The KruskaleWallis test was used to analyse the effect of order of tool set presentation on the percentage of retrievals of food with the correct tool. Since in the PlatRig/HandRig condition there was no correct response, we analysed the percentage of trials in which the tool was discarded. Friedman and Wilcoxon tests were used to assess differences between conditions in the percentages of behaviours performed. A Wilcoxon test was also run to assess differences between percentages of behaviours performed within each condition. The binomial test was used to assess the subject’s performance in the first trial. Spearman correlations were run to assess the relation between subjects’ performances in the PlatFlex/HandRig condition, in which the platform tool was flexible and the one handed to the subject was rigid, with prior experience (expressed as the numbers of tool-using experiments in which subjects participated) and age (in years). Descriptive statistics include the median and the interquartile range (IQR).
RESULTS Pretest Table 2 and Fig. 3 provide descriptive information on each subject’s individual performance. We found a significant decrease in the number of trials needed by capuchins to achieve criterion between the first and the last tool sets (Wilcoxon test: T ¼ 3, N ¼ 10, P ¼ 0.04). Test Table 2 provides descriptive information on the performance of each individual. There was no significant effect of the order of tool set presentation on the capuchins’ performance (PlatFlex/HandRig: H3 ¼ 2.44, P ¼ 0.49; PlatFlex/HandFlex: H3 ¼ 2.33, P ¼ 0.51; PlatRig/ HandFlex: H3 ¼ 4.00, P ¼ 0.26; PlatRig/HandRig: H3 ¼ 2.54, P ¼ 0.47). Therefore, we pooled the data from the different groups for subsequent analyses. The percentage of raking in the tool placed on the platform using the tool handed to the subject differed between conditions (Friedman test: c2 3 ¼ 25.97, N ¼ 10, P ¼ 0.00001; Fig. 4).
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220 Table 2 Summary of capuchins’ results
Carlotta Gal Paprica Pedro Pepe Roberta Robin Hood Rucola Sandokan Vispo Mean
Pretest
Test
Mean number of trials needed to achieve criterion
Number correct out of four trials for each condition PlatFlex/ HandRig
PlatFlex/ HandFlex
PlatRig/ HandFlex
50 52 90 38 40 40 44 30 32 24 44
4 4 4 3 3 4 4 4 4 4 3.8
4 4 4 4 4 4 4 4 4 2 3.8
4 4 4 4 3 4 4 4 4 4 3.9
Capuchins displayed significantly less raking in those conditions in which they received the flexible tool compared to those conditions in which they received the rigid tool (Wilcoxon tests: PlatFlex/ HandRig versus PlatFlex/HandFlex: T ¼ 0, P ¼ 0.005; PlatFlex/HandRig versus PlatRig/HandFlex: T ¼ 0, P ¼ 0.005; PlatRig/HandRig versus PlatFlex/HandFlex: T ¼ 0, P ¼ 0.008; PlatRig/HandRig versus PlatRig/HandFlex: T ¼ 0, P ¼ 0.005; PlatFlex/HandRig versus PlatRig/HandRig: T ¼ 2, P ¼ 0.14; PlatFlex/HandFlex versus PlatRig/ HandFlex: T ¼ 5.5, P ¼ 0.29). These results were already apparent in the first trial: nine out of 10 capuchins raked in the flexible tool using the rigid tool when the flexible tool was out of reach on the platform, while all 10 capuchins used the tool handed to them to retrieve food in those
217
conditions in which they received the flexible tool (binomial tests: PlatFlex/HandRig: P ¼ 0.039; PlatFlex/HandFlex: P ¼ 0.004; PlatRig/HandFlex: P ¼ 0.004). Moreover, in the PlatFlex/HandRig condition subjects retrieved the flexible tool from the platform significantly more often than they tried to insert into the tube the rigid tool handed to them (Wilcoxon test: T ¼ 1.5, N ¼ 10, P ¼ 0.003; Fig. 5). In contrast, in the PlatFlex/HandFlex and PlatRig/HandFlex conditions (when the platform tool was unnecessary to obtain food) subjects retrieved food with the flexible tool handed to them significantly more often than they tried to retrieve the platform tool (Wilcoxon tests: T ¼ 2.5, N ¼ 10, P ¼ 0.007; T ¼ 0, N ¼ 10, P ¼ 0.003; Fig. 5). In the PlatRig/HandRig condition, in which only rigid tools were available, subjects equally often retrieved the platform tool (median ¼ 87.5, IQR ¼ 25), probed into the tube directly with the tool handed to them (median ¼ 62.5, IQR ¼ 75; Wilcoxon test: T ¼ 9, N ¼ 10, P ¼ 0.11; Fig. 5) and discarded the latter tool (median ¼ 75, IQR ¼ 50; Wilcoxon test: T ¼ 16.5, N ¼ 10, P ¼ 0.83). Subjects’ performances in the PlatFlex/HandRig condition did not correlate with prior experience in tool-using experiments (Spearman rS ¼ 0.31, N ¼ 10, P ¼ 0.38) or with age (Spearman rS ¼ 0.18, N ¼ 10, P ¼ 0.63). DISCUSSION Capuchin monkeys engaged in sequential use of tools whose functional properties are not visually apparent. All capuchins successfully performed sequential tool use and nine out of 10 did so already in the first trial of the crucial condition (i.e. PlatFlex/HandRig, when the platform tool was flexible and the tool handed to the subject was rigid). In the conditions in which the platform tool
200
180
160
Number of trials to criterion
140 Carlotta Gal Paprica Pedro Pepe Roberta Robin Hood Rucola Sandokan Vispo
120
100
80
60
40
20
0
First tool set
Second tool set
Third tool set
Fourth tool set
Figure 3. Number of trials needed to achieve criterion by each subject in the pretest from the first to the fourth tool set presented. To advance to the test phase, the subject had to choose the correct tool in 16 consecutive choice trials for each set of tools.
218
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220
** Percentage of retrieving (or trying to retrieve) platform tool
** **
** 100
80
60
40
20
0
PlatFlex/HandRig
PlatFlex/HandFlex
PlatRig/HandFlex
PlatRig/HandRig
Condition Figure 4. Median (white square) IQR (box) and range (whiskers) of the percentages of retrieving the platform tool with the tool handed to the subject (for PlatFlex/HandRig and PlatRig/HandRig conditions) and of trying to retrieve the platform tool with the tool handed to the subject (for PlatFlex/HandFlex and PlatRig/HandFlex conditions). ** P < 0.01.
**
**
**
PlatFlex/HandRig
PlatFlex/HandFlex
PlatRig/HandFlex
Percentage of behaviour
100
80
60
40
20
0 PlatRig/HandRig
Condition Figure 5. Median (white square) IQR (box) and range (whiskers) of the percentages of retrieving, or trying to retrieve, the tool from the platform (in white) versus the percentages of retrieving, or trying to retrieve, food with the tool handed to the subject (in black) for each condition. ** P < 0.01.
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220
was unnecessary to obtain food (PlatFlex/HandFlex, PlatRig/HandFlex), capuchins used the flexible tool that was already within reach. These results demonstrate that subjects were sensitive to both the tool rigidity and the task requirements since they chose which tool to use accordingly. Our tasks posed the capuchins an unprecedented cognitive challenge since they needed to deal simultaneously with conflicting information that required flexible and hierarchical behaviour. When they received a tool they had to judge its suitability to meet the tube task requirements. If the tool was unsuitable, they needed to judge whether the demonstrated tool resting on the platform was appropriate to be used in the tube task and if they could retrieve it by using the tool handed to them. Note that looking at the tool presented on the platform was not sufficient to judge whether it was suitable to obtain the reward, since the tool functional property, flexibility, was not visually detectable. Flexibility could be inferred only on the basis of the behaviour of the tool when it was manipulated by the experimenter. This ability, present in great apes (Manrique et al., 2010) as well as in capuchins (Manrique et al., 2011), can be considered in terms of affordance learning, a process in which, by watching the object while it moves, the observer detects its affordances, such as dynamic properties and temporalespatial relations. During the second year of life, human infants become able to learn object affordances when only the object movement component is presented without a demonstrator performing it (e.g. Huang & Charman, 2005; Tennie, Call, & Tomasello, 2006). Experiments with a ‘ghost’ condition in chimpanzees have produced mixed results (Hopper, Lambeth, Schapiro, & Whiten, 2007; Hopper, Spiteri, Lambeth, Schapiro, & Whiten, 2008; Tennie et al., 2006). Future studies should assess whether capuchin monkeys are able to learn object affordances through a ghost condition. Since rigidity was not directly observable, it is conceivable that capuchins had to keep active in their working memory the properties of the platform tool (its rigidity) to organize her/his behaviour efficiently, which imposes an additional demand not present when judging tool length because length is a visible property (e.g. Martin-Ordas et al., 2012; Wimpenny et al., 2009). In these studies the tools were or were not long enough to reach the food. When long enough they could be used directly. When too short, the subjects had to retrieve one or several longer tools before attempting to get the food. The task required subjects to compare tool length with distance to the food visually or mentally (when tools and reward were not visible at the same time) and judge tool suitability accordingly. In a way, one could say that changes in the critical property (length) were gradual and unidirectional. Thus, a rule such as ‘when the tool available is too short then pick a longer tool’ could satisfactorily explain subjects’ success. However, when we talk about rigidity things are not that straightforward because being flexible is just the opposite of being rigid and, hence, the changes in tool properties are not a simple question of degree of flexibility but rather a matter of quality. It would be interesting to see how other animal species (e.g. New Caledonian crows) perform in an analogous task to the one we implemented. Wimpenny et al. (2009) argued that to qualify as goal directed, tool retrieval should occur only when required (i.e. when the tool in hand was not flexible enough to get the food). Not only did capuchins retrieve the flexible tool when required, but for those conditions in which the platform tool was rigid, capuchins used the flexible tool that was already within reach to obtain food. In these conditions in which tool retrieval was unnecessary, capuchins adopted the more efficient strategy from the beginning; thus their performance resembled more that of great apes than that of New Caledonian crows (Martin-Ordas et al., 2012; Mulcahy et al., 2005; Taylor et al., 2007; Wimpenny et al., 2009). It can be argued that a
219
full understanding of the PlatRig/HandRig condition required no probing at all, as neither tool was suitable for solving the task. In this ‘impossible’ condition capuchins showed no clear preference towards retrieving the platform tool compared to probing directly into the tube with the tool handed to them or discarding the latter tool. Capuchins made different attempts to get the reward with the tool(s), as already reported for gorillas and orang-utans in an analogous condition in which the available tool was not suitable for getting the reward (Mulcahy et al., 2005). In our opinion, this lack of preference for one of the tools reflects a lack of inhibitory control in the sight of food, rather than a lack of understanding of the task requirements. This is supported by observations that capuchins display a very active and persistent foraging style as well as a tendency to manipulate and combine objects that is difficult to restrain in the presence of or in searching for food (Fragaszy, Visalberghi, & Fedigan, 2004). Moreover, capuchin monkeys typically show worse capacities than chimpanzees to delay gratification, waiting for approximately 10e20 s (Anderson, Kuroshima, & Fujita, 2010) or 30e40 s (Pelé, Micheletta, Uhlrich, Thierry, & Dufour, 2011) in comparison to 120 s for chimpanzees (Beran & Evans, 2006). Our subjects spontaneously used two tools in sequence. In fact they were trained to criterion in the pretest but not to use a tool to retrieve another tool in the test. Although an associative process such as chaining (sensu Epstein, Kirshnit, Lanza, & Rubin, 1984) could not be completely ruled out, we think that it cannot satisfactorily account for our results. There are three reasons for rejecting the idea that once subjects ‘discovered’ the action of tool retrieval from the platform, they automatically performed this action and subsequently inserted the flexible tool into the tube to retrieve food, or that they relied on random probes for tools, followed by probes for food. First, in the PlatFlex/HandRig condition, in which the tool handed to them was unsuitable for the tube task, nine out of 10 subjects showed successful and spontaneous sequential tool use, using the tool to retrieve another tool from their first trial. Second, in those conditions in which retrieving the platform tool was unnecessary all capuchins used the tool that was already within reach to obtain food from their first trial. Third, if during the pretest subjects had simply learnt to associate the rigid tool with the platform task, in the PlatRig/HandRig condition, in which the two tools available were rigid, we should have found more attempts to rake in the platform tool, as evident in the PlatFlex/HandRig condition, in which only the tool handed to them was rigid. In contrast, capuchins almost equally tried to insert the tool handed to them into the tube, discarded it and tried to retrieve the platform tool. Moreover, prior tool-using experience and age did not seem to affect subjects’ performance. The most experienced capuchin (Carlotta) did not even outperform the least experienced one (Rucola). More than two decades ago Carlotta participated in an experiment in which she had to use a shorter stick to rake in a longer one (Parker & Potì, 1990). She was then 2.5 years old and her success rate varied from 33% to 66% depending on the position of the short stick relative to the long stick. Moreover, when tested in several other tool tasks she performed well. In contrast, Rucola (who was 12 years old) had never been tested in tool tasks. Despite the different previous experience and age, both subjects were equally successful in sequential tool use. The ability to organize one’s own actions hierarchically is also supported by experiments that have assessed planning during sequential responding to computer-generated stimuli by capuchin monkeys (Beran & Parrish, 2012). In two experiments, eight capuchin monkeys selected five arbitrary stimuli in distinct locations on a computer monitor in a learned sequence. In experiment 1, shift trials occurred in which the second and third stimuli were
220
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220
transposed when the first stimulus was selected by the animal. In experiment 2, mask trials occurred in which all remaining stimuli were masked after the monkey selected the first stimulus. Capuchins made more mistakes on trials in which the locations of the second and third stimuli were interchanged than on trials in which locations were not interchanged, suggesting they had already planned to select a location that no longer contained the correct stimulus. When mask trials occurred, capuchins performed at levels significantly better than chance for the first and the second selections on a trial. In conclusion, the current study demonstrated that capuchin monkeys were able to organize their actions in the hierarchically correct sequence using tools differing in their rigidity, a nonvisually detectable property. Moreover, they were able to manage conflicting information when handed a tool that is both functional and nonfunctional depending on the apparatus. Our present findings also contribute to the growing body of research revealing that capuchins also use tools to act upon another object. One limitation of our study was that the only possible form of sequential tool use consisted of retrieving the flexible tool on the platform with the rigid tool handed to the subject. It could be interesting to present capuchins with the possibility to do the opposite (retrieve a reward placed on a platform). Future studies on capuchins should also include the possibility to use more than two tools in sequence, thus increasing the attentional and cognitive demands of the task. Acknowledgments This research was supported by the European Community’s Seventh Framework Programme FP7/2007-2013, ‘Challenge 2e Cognitive Systems, Interaction, Robotics’, under grant agreement No FP7-ICT-IP-231722, project ‘IM-CLeVeReIntrinsically Motivated Cumulative Learning Versatile Robots’ and the PNR-CNR Aging Program 2012-2014. We thank the keepers of the Primate Center of ISTC-CNR for help with animal testing, the Roma Capitale-Museo Civico di Zoologia and the Fondazione Bioparco for hosting the Unit of Cognitive Primatology and Primate Centre, and the two anonymous referees and Dr Ximena Nelson for their useful comments. Supplementary Material Supplementary material associated with this article is available, in the online version, at http://dx.doi.org/10.1016/j.anbehav.2013. 10.033. References Anderson, J. R., & Henneman, M. C. (1994). Solutions to a tool-use problem in a pair of Cebus apella. Mammalia, 58, 351e361. Anderson, J. R., Kuroshima, H., & Fujita, K. (2010). Delay of gratification in capuchin monkeys (Cebus apella) and squirrel monkeys (Saimiri sciureus). Journal of Comparative Psychology, 124, 205e210. Beran, M. J., & Evans, T. A. (2006). Maintenance of delay of gratification by four chimpanzees (Pan troglodytes): the effects of delayed reward visibility, experimenter presence, and extended delay intervals. Behavioural Processes, 73, 315e 324. Beran, M. J., & Parrish, A. E. (2012). Sequential responding and planning in capuchin monkeys (Cebus apella). Animal Cognition, 15, 1085e1094. Bird, C. D., & Emery, N. J. (2009). Insightful problem solving and creative tool modification by captive nontool-using rooks. Proceedings of the National Academy of Sciences, 106, 10370e10375.
Epstein, R., Kirshnit, C. E., Lanza, R. P., & Rubin, L. C. (1984). ‘Insight’ in the pigeon: antecedents and determinants of an intelligent performance. Nature, 308, 61e 62. Fragaszy, D. M., Visalberghi, E., & Fedigan, L. M. (2004). The complete capuchin. The biology of the genus Cebus. Cambridge: Cambridge University Press. Hihara, S., Obayashi, S., Tanaka, M., & Iriki, A. (2003). Rapid learning of sequential tool use by macaque monkeys. Physiology & Behaviour, 78, 427e434. Hopper, L. M., Lambeth, S. P., Schapiro, S. J., & Whiten, A. (2008). Observational learning in chimpanzees and children studied through ‘ghost’ conditions. Proceedings of the Royal Society B, 275, 835e840. Hopper, L. M., Spiteri, A., Lambeth, S. P., Schapiro, S. J., & Whiten, A. (2007). Experimental studies of traditions and underlying transmission processes in chimpanzees. Animal Behaviour, 73, 1021e1032. Huang, C.-T., & Charman, T. (2005). Gradations of emulation learning in infants’ imitation of actions on objects. Journal of Experimental Child Psychology, 92, 276e302. Klüver, H. (1933). Behavior mechanisms in monkeys. Chicago: University of Chicago Press. Köhler, W. (1925/1976). The mentality of apes. New York: Liveright. Lynch Alfaro, J. W., Boubli, J. P., Olson, L. E., Di Fiore, A., Wilson, B., GutiérrezEspeleta, G. A., et al. (2012). Explosive Pleistocene range expansion leads to widespread Amazonian sympatry between robust and gracile capuchin monkeys. Journal of Biogeography, 39, 272e288. Lynch Alfaro, J. W., De Souza, Silva, J., & Rylands, A. B. (2012). How different are robust and gracile capuchin monkeys? An argument for the use of Sapajus and Cebus. American Journal of Primatology, 14, 1e14. Mannu, M., & Ottoni, E. B. (2009). The enhanced tool-kit of two groups of wild bearded capuchin monkeys in the Caatinga: tool making, associative use, and secondary tools. American Journal of Primatology, 71, 242e251. Manrique, H. M., Gross, A., & Call, J. (2010). Great apes select tools on the basis of their rigidity. Journal of Experimental Psychology: Animal Behavior Processes, 36, 409e422. Manrique, H. M., Sabbatini, G., Call, J., & Visalberghi, E. (2011). Tool choice on the basis of rigidity in capuchin monkeys. Animal Cognition, 14, 775e786. Martin-Ordas, G., Schumacher, L., & Call, J. (2012). Sequential tool use in great apes. PLoS ONE, 7, e52074. Mulcahy, N. J., Call, J., & Dunbar, R. I. M. (2005). Gorillas (Gorilla gorilla) and orangutans (Pongo pygmaeus) encode relevant problem features in a tool-using task. Journal of Comparative Psychology, 119, 23e32. Parker, S. T., & Potì, P. (1990). The role of innate motor patterns in ontogenetic and experiential development of intelligent use of sticks in cebus monkeys. In S. T. Parker, & K. R. Gibson (Eds.), ‘Language’ and intelligence in monkeys and apes (pp. 219e243). New York: Cambridge University Press. Pelé, M., Micheletta, J., Uhlrich, P., Thierry, B., & Dufour, V. (2011). Delay maintenance in tonkean macaques (Macaca tonkeana) and brown capuchin monkeys (Cebus apella). International Journal of Primatology, 32, 149e166. Sabbatini, G., Truppa, V., Hribar, A., Gambetta, B., Call, J., & Visalberghi, E. (2012). Understanding the functional properties of tools: chimpanzees (Pan troglodytes) and capuchin monkeys (Cebus apella) attend to tool features differently. Animal Cognition, 15, 577e590. Santos, L. R., Rosati, A., Sproul, C., Spaulding, B., & Hauser, M. D. (2005). Meansmeans end tool-use in cotton-top tamarins (Saguinus oedipus): finding the limits on primates’ knowledge of tools. Animal Cognition, 8, 236e246. Shumaker, R., Walkup, K. R., & Beck, B. B. (2011). Animal tool behavior: the use and manufacture of tools by animals. Baltimore: The Johns Hopkins University Press. Taylor, A. H., Elliffe, D., Hunt, G. R., & Gray, R. D. (2010). Complex cognition and behavioural innovation in New Caledonian crows. Proceedings of the Royal Society B, 277, 2637e2643. Taylor, A. H., Hunt, G. R., Holzhaider, J. C., & Gray, R. D. (2007). Spontaneous metatool use by New Caledonian Crows. Current Biology, 17, 1504e1507. Tennie, C., Call, J., & Tomasello, M. (2006). Push or pull: imitation versus emulation in great apes and human children. Ethology, 112, 1159e1169. Visalberghi, E., Addessi, E., Truppa, V., Spagnoletti, N., Ottoni, E., Izar, P., et al. (2009). Selection of effective stone tools by wild bearded capuchin monkeys. Current Biology, 19, 213e217. Visalberghi, E., & Limongelli, L. (1994). Lack of comprehension of cause-effect relations in tool-using capuchin monkeys (Cebus apella). Journal of Comparative Psychology, 108, 15e22. Visalberghi, E., & Trinca, L. (1989). Tool use in capuchin monkeys: distinguishing between performing and understanding. Primates, 30, 511e521. Warden, C. J., Koch, A. M., & Fjeld, H. A. (1940). Instrumentation in Cebus and Rhesus monkeys. The Journal of Genetic Psychology, 56, 297e310. Wimpenny, J. H., Weir, A. A. S., Clayton, L., Rutz, C., & Kacelnik, A. (2009). Cognitive processes associated with sequential tool use in New Caledonian Crows. PLoS ONE, 4, e6471.
Contents lists available at ScienceDirect
Animal Behaviour journal homepage: www.elsevier.com/locate/anbehav
Sequential use of rigid and pliable tools in tufted capuchin monkeys (Sapajus spp.) Gloria Sabbatini a, *, Héctor Marín Manrique b, c, Cinzia Trapanese a, d, Aurora De Bortoli Vizioli a, Josep Call b, Elisabetta Visalberghi a a
Institute of Cognitive Sciences and Technologies of CNR, Rome, Italy Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany Department of Psychology, University of the Balearic Islands, Palma, Spain d Department of Biology, University of Tor Vergata, Rome, Italy b c
a r t i c l e i n f o Article history: Received 7 August 2013 Initial acceptance 3 September 2013 Final acceptance 9 October 2013 Available online 3 December 2013 MS. number: 13-00658 Keywords: meta tool use monkeys motor planning object properties primates problem solving
Previous studies have shown that a few bird and primate species successfully use two or more tools sequentially to obtain a goal. Our study incorporated some novel features: we aimed to assess whether captive capuchins (Sapajus spp.) are able to use tools differing in rigidity sequentially to obtain a reward. A 90 angled Plexiglas tube filled with yoghurt and a platform were placed 121 cm apart in front of the test compartments. The experimenter showed the subject the functional properties of a tool (rigidity or flexibility) as well as its nonfunctional ones (i.e. shape, colour, etc.) then deposited it on a platform out of the subject’s reach. Then the experimenter handed the subject another tool (rigid or flexible). To obtain the reward, the subject had to insert the flexible tool inside the tube. Four conditions (2 2 design) were presented. Capuchins used the tools in sequence to get the reward when necessary (nine out of 10 subjects on the first trial) and not as a result of a simple heuristic of using the tool placed nearest to the food. Whenever the platform tool was unnecessary, capuchins used the flexible tool already within reach. These results demonstrate that capuchins are able to organize their actions in the hierarchically correct sequence using tools differing in their rigidity. Ó 2013 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Using an object to alter the form, position or condition of another object is widespread in the animal kingdom (Shumaker, Walkup, & Beck, 2011). However, only a few individuals of a limited number of primate and corvid species show sequential tool use in which a tool is used to obtain another tool, which subsequently will serve to obtain an out-of-reach goal. According to Taylor, Hunt, Holzhaider, and Gray (2007) sequential tool use is challenging because the subject must (1) recognize that tools can be used on a nonfood item, (2) resist the immediate motivation to use the tool to attempt to reach the food directly, and (3) be able to organize her/his behaviour hierarchically. A task commonly used in the laboratory to test sequential tool use consists of presenting subjects with a reward that is out of reach and a readily available tool that is not long enough to reach the reward but long enough to reach another tool, which can be used to reach the reward. Subjects have to use the tools sequentially: first the shorter one to retrieve the longer one, and then the latter to recover the food. Spontaneous use of up to two tools in
* Correspondence: G. Sabbatini, Istituto di Scienze e Tecnologie della Cognizione, Consiglio Nazionale delle Ricerche, Via Ulisse Aldrovandi 16/B, 00197 Rome, Italy. E-mail address: [email protected] (G. Sabbatini).
sequence has been reported in chimpanzees, Pan troglodytes (Köhler, 1925/1976), gorillas, Gorilla gorilla, and orang-utans, Pongo pygmaeus (Mulcahy, Call, & Dunbar, 2005). More recently, spontaneous use of up to five stick tools in sequence has been reported in great apes (Martin-Ordas, Schumacher, & Call, 2012). After receiving some training, Japanese macaques, Macaca fuscata (Hihara, Obayashi, Tanaka, & Iriki, 2003) and cottontop tamarins, Saguinus oedipus (Santos, Rosati, Sproul, Spaulding, & Hauser, 2005) can also use tools in sequence. Capuchin monkeys (Sapajus spp.) spontaneously use two tools in sequence (Anderson & Henneman, 1994; Klüver, 1933) but can also be trained to use up to eight rake tools in sequence (Warden, Koch, & Fjeld, 1940). Moreover, wild capuchins were observed combining the use of two tools, stones and sticks or two stones, in sequence (Mannu & Ottoni, 2009). Sequential tool use is not limited to primates. New Caledonian crows, Corvus moneduloides, spontaneously use a short stick to extract a longer stick (placed in a box), which they use to extract meat from a vertical tube (Taylor et al., 2007). More recently, Taylor, Elliffe, Hunt, and Gray (2010) presented seven New Caledonian crows with a task requiring (1) pulling up a string to obtain a short stick, (2) using the short stick to extract a longer stick from a toolbox, and finally (3) using the long stick to extract food from a hole. Crows with and without previous experience with the task
0003-3472/$38.00 Ó 2013 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.anbehav.2013.10.033
214
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220
successfully solved the problem. Wimpenny, Weir, Clayton, Rutz, and Kacelnik (2009) reported that New Caledonian crows were able to use up to three tools in sequence to get the reward. Here, crows were presented with an out-of-reach reward, two tools too short to reach the reward and four out-of-reach tools differing in length (and therefore functionality). The distance to the food and/ or which tools were required to get it varied according to the experimental condition. However, only the subjects with pretest experience of extracting tools from tubes and of using tools to extract food from another tube, solved the task. Rooks, Corvus frugilegus, a nontool-using species, have been reported to use tools sequentially by spontaneously dropping a large stone into a container to release a small stone, which was then used to acquire food (Bird & Emery, 2009). Disentangling the cognitive processes involved in sequential tool use is difficult (Wimpenny et al., 2009) because of several methodological limitations present in previous studies. Our study was meant to control some of them. In some studies (Hihara et al., 2003; Parker & Potì, 1990; Santos et al., 2005) the out-of-reach tool and the reward were in proximity so that actions directed to the reward could accidentally be directed to the out-of-reach tool. To prevent this, in our experiment the out-of-reach tool was not positioned between the subject and the reward, but in the opposite direction. In some studies the unreachable object was always the correct tool to solve the problem and there were no control conditions (Anderson & Henneman, 1994; Hihara et al., 2003; Parker & Potì, 1990; Santos et al., 2005; Taylor et al., 2010). Wimpenny et al. (2009) argued that to qualify as goal directed, tool retrieval should occur only when required. Therefore, we included control conditions to test whether tool retrieval did reflect an active choice. Finally, whereas all previous studies used tools that varied in visual properties, namely length or size, we tested capuchin monkeys with two tool tasks simultaneously and with tools differing in rigidity, a nonvisual property. In our study capuchins were required to use a rigid tool to retrieve a flexible tool from a platform and use the flexible tool to obtain the yoghurt available at the bottom of a 90 angled tube. We chose rigidity as the nonvisually detectable functional property since in previous studies it has been shown that capuchin monkeys are able to detect objects’ properties efficiently by nonvisual functional clues (friability, rigidity, weight; e.g. Manrique, Sabbatini, Call, & Visalberghi, 2011; Visalberghi et al., 2009). In particular, Manrique et al. (2011) found that subjects were able to infer an object’s property by observing an experimenter repeatedly bending the tool or by directly manipulating it. When facing an outof-reach reward capuchins efficiently used information previously gathered about tool affordances and selected the appropriate tool above chance levels (i.e. rigid or flexible depending on the task), thus demonstrating a good appreciation of the relation between the tool’s properties and the task requirements. We took advantage of Manrique et al.’s (2011) experimental setup to assess whether capuchin monkeys are able to use tools sequentially by taking into account the functional properties of the tools (i.e. rigidity or flexibility) in relation to the requirements of the task at hand, and hierarchically organize their actions. Unlike previous studies investigating sequential tool use, we had two tasks, each relying on opposite properties of the tool (rigid versus flexible). Depending on the condition, subjects were required to use a flexible tool to recover the yoghurt from a tube or to use a rigid tool to retrieve a second flexible tool from a rectangular platform and use it to obtain the yoghurt. Hence, when subjects were handed one tool, this tool was both functional and nonfunctional, the functionality being determined by the apparatus the subjects needed to reach. This brought about an additional difficulty to the sequential tool-using task that needed to be overcome. In particular, when a flexible tool was given to the subject, then the reward
inside the tube was directly obtainable. When a rigid tool was given, the subject could not reach the reward and therefore should first recover the flexible tool from the platform. We expected capuchin monkeys to use first the rigid tool to get the flexible tool, and then the flexible one to reach the reward. When the two tools available were rigid, subjects should not have attempted to use the tools because neither of them was suitable to get the reward. METHODS Subjects and Housing Ten tufted capuchin monkeys (Sapajus spp., formerly attributed to the Cebus genus; see Lynch Alfaro, Boubli, et al., 2012; Lynch Alfaro, De Souza, Silva, & Rylands, 2012) housed at the Primate Centre of the Institute of Cognitive Sciences and Technologies, CNR in Rome, Italy, participated in this experiment. They were six males and four females ranging in age from 11 to 28 years. Four subjects were mother reared and six hand reared. Group A consisted of six individuals, group B five individuals and group C 11 individuals. The groups were housed in enclosures consisting of an outdoor area (group A ¼ 106.5 m3, group B ¼ 127.4 m3, group C ¼ 374 m3) which was furnished with ropes, climbing frames, slides and wood chips, and two indoor cages (overall about 24.5 m3) which were furnished with perches and slides. Attached to one of the indoor cages, there was a tunnel made from three identical wire-mesh compartments (a, b and c; see below). Capuchins were tested individually in the indoor area, to which they had access through a sliding door from the adjacent outdoor enclosure. Each subject was separated on a voluntary basis from the group solely for the purpose of testing without causing her/him any apparent distress, just before each test session. All procedures complied with protocols approved by the Italian Health Ministry (Licence no. 12/2011-C) and were performed in full accordance with the European law on humane care and use of laboratory animals. Moreover, we adhered to the ASAB/ ABS Guidelines for the use of animals in research. Nine out of 10 capuchins had experience with Manrique et al.’s (2011) tool tasks and seven with other tool tasks as well (see Table 1). Only one capuchin was naïve to these tool-using experiments. Monkey chow (Altromin-A pellets, Rieper standard diet for primates), fresh fruits and vegetables were given every afternoon after testing. Water was freely available at all times. Apparatus and Tools The test area comprised three identical compartments (each 60 75 cm and 75 cm high) separated by sliding lockable doors (Fig. 1). We used two different apparatuses: a transparent Plexiglas
Table 1 Demographics of subjects and their previous experience with tool-using tasks Name
Group
Sex
Age (years)
Rearing
Previous experience
Carlotta Gal Paprica Pedro Pepe Roberta Robin Hood Rucola Sandokan Vispo
A A A C B B C C B B
F M F M M F M F M M
28 22 23 11 25 26 15 12 12 12
Hand Hand Hand Mother Hand Hand Mother Mother Hand Mother
1, 2, 2, 5 2, 3, 5
3, 4, 5, 6 5, 6 5, 6 5 5, 6
5, 6 5, 6
(1) Visalberghi and Trinca, 1989. (2) E. Visalberghi, unpublished data (3) Visalberghi and Limongelli, 1994. (4) Parker and Potì, 1990. (5) Manrique et al., 2011. (6) Sabbatini et al., 2012.
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220
215
Figure 1. Schematic representation of the test set-up. A capuchin monkey is shown in the central compartment (b) waiting for the tool held in the experimenter’s hand. After having received the tool the subject had to decide whether to try to insert it inside the tube in compartment a or to try to retrieve the demonstrated tool from the platform. The sliding doors separating compartments a, b, c were open at the time the subject decided what to do.
tube and a rectangular platform (see Fig. 1). The transparent Plexiglas tube (4 cm in diameter) presented an angle of approximately 90 ; it was open at one end (top) and closed at the other (bottom). The tube was attached to the wire mesh of compartment a (the furthest from the indoor cage), and filled with yoghurt. To reach the yoghurt the subject needed to choose a flexible tool that could bend at the tube’s angle. After removing the tool, the subject could lick the yoghurt off it. The rectangular platform (80 50 cm) was located in front of compartment c (the closest to the indoor cage). An out-of-reach item (a banana slice during the pretest and a tool during the test) was positioned on the platform. The item was outside the subject’s reach since it was placed beyond her/his arm length, at a distance of about 30 cm for females and 35 cm for males. To reach the item the subject needed a rigid tool. During the pretest, the presence of the tube and the platform were alternated whereas they were present at the same time during the test. Sets of tools were composed of four visually distinct tools; all were 38 cm long and differed in colour, material, diameter, weight and rigidity (Fig. 2). Four different sets of four tools each were employed in this experiment (for a total of 16 different tools). Procedure Pretest The pretest consisted of familiarization and choice trials. Each capuchin was given familiarization trials in each of the following tasks: (1) getting, or trying to get, the yoghurt inside the tube with each tool in the sets and (2) getting, or trying to get, the out-ofreach slice of banana placed on the platform with each tool in the sets. For each task the familiarization trials were carried out in blocks of four (one trial for each tool of the set). The two tasks were presented in alternating blocks. During each familiarization trial, the experimenter manipulated each tool for 5 s in front of the subject and then gave the tool to the subject to use. The subject could retrieve the food only when the tool was suited for the presented task. After each familiarization block, to assess whether the subject could choose the appropriate tool for the task, the experimenter allowed her/him to choose between one rigid and one flexible tool. Each rigid tool was paired with each flexible tool of the set. Two choice trials were given after each block. To advance to the test phase, from each of the four sets of tools the subject had to choose the correct tool in 16 consecutive choice trials. To sum up, before
the test, each subject had to reach the criterion for each tool set before moving onto the next, and reach the criterion of 16 correct tool choices for the four sets of tools. Test The experimenter placed the Plexiglas tube in front of compartment a and the platform in front of compartment c and showed the rigidity properties (bending and unbending it) of a tool for 4e5 s (see Supplementary material). Immediately after this demonstration the tool was deposited on the platform out of the subject’s reach. Then the experimenter handed another tool to the subject through the wire mesh of compartment b. At this point, the subject had to choose between using the tool handed to him/her and getting the tool from the platform. There were four different experimental conditions according to the rigidity/flexibility of the tool placed on the platform (after demonstration) and the tool given to the subject: (1) Platform: Flexible tool/Handed: Rigid tool (PlatFlex/HandRig); (2) Platform: Flexible tool/Handed: Flexible tool (PlatFlex/HandFlex); (3) Platform: Rigid tool/Handed: Flexible tool (PlatRig/HandFlex); (4) Platform: Rigid tool/Handed: Rigid tool (PlatRig/HandRig). Capuchins received four sessions of four trials each, for a total of 16 trials. Each session was presented on a different day. The order of presentation of each condition was randomly assigned within one session and each tool set was presented in the four conditions in only one session. The order in which sets were administered was counterbalanced across subjects, three subjects receiving sets in the order 1-2-3-4, three in the order 2-3-4-1, two in the order 3-4-1-2 and two in the order 4-1-23. Before the first trial of the test phase, each subject was given a 1 min period of familiarization with the experimental set-up in the absence of the second tool (i.e. the one handed to the subject). Each trial lasted 5 min. Tests were conducted between December 2011 and June 2012. Data Scoring and Analyses We videotaped all test trials and scored how the subject used the tool(s) to retrieve, or try to retrieve, the reward. In particular, we scored whether the subject: (1) inserted, or tried to insert, the tool handed to him/her or the platform tool inside the tube to retrieve the yoghurt; (2) raked in, or tried to rake in, the platform tool with the tool handed to him/her; and (3) discarded the tool inside or outside the compartments.
216
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220
Figure 2. Sets of tools employed in the experiment. Each set of tools consisted of two flexible (a, b) and two rigid tools (c, d), all of them 28.5 cm long. Set 1: (a) cream phone cord, 1.2 cm in diameter; (b) golden electric wire covered by a transparent plastic shaft, 0.5 cm in diameter; (c) black wooden stick (1.5 1.5 cm); and (d) copper pipe, 1 cm in diameter. Set 2: (a) grey plastic cable collector, 1.8 cm in diameter; (b) orange rope, 0.8 cm in diameter; (c) silver metal bar, 1.3 cm in diameter and (d) brown rounded wooden stick, 1.5 cm in diameter. Set 3: (a) black electric wire, 0.5 cm in diameter; (b) cream rope, 1.5 cm in diameter; (c) plastic-covered green metal stick, 1.5 cm in diameter; and (d) yellow dowel, 2 cm in diameter. Set 4: (a) black plastic clamp (0.5 0.2 cm); (b) rope wrapped in grey duct tape, 0.8 cm in diameter; (c) white rounded plastic stick, 2 cm in diameter; and (d) light brown U-shaped metal rail (1 0.8 cm).
Dependent measures were the percentage of trials in which subjects displayed each of these behaviours. We paid particular attention to the performance on the first trial of each condition. C.T. scored all trials and A.D.B.V. scored a random selection of 20% of the trials to calculate interobserver reliability. Interobserver reliability was perfect for all behaviours (Cohen’s kappa ¼ 1). We analysed the data using two-tailed nonparametric statistics. The KruskaleWallis test was used to analyse the effect of order of tool set presentation on the percentage of retrievals of food with the correct tool. Since in the PlatRig/HandRig condition there was no correct response, we analysed the percentage of trials in which the tool was discarded. Friedman and Wilcoxon tests were used to assess differences between conditions in the percentages of behaviours performed. A Wilcoxon test was also run to assess differences between percentages of behaviours performed within each condition. The binomial test was used to assess the subject’s performance in the first trial. Spearman correlations were run to assess the relation between subjects’ performances in the PlatFlex/HandRig condition, in which the platform tool was flexible and the one handed to the subject was rigid, with prior experience (expressed as the numbers of tool-using experiments in which subjects participated) and age (in years). Descriptive statistics include the median and the interquartile range (IQR).
RESULTS Pretest Table 2 and Fig. 3 provide descriptive information on each subject’s individual performance. We found a significant decrease in the number of trials needed by capuchins to achieve criterion between the first and the last tool sets (Wilcoxon test: T ¼ 3, N ¼ 10, P ¼ 0.04). Test Table 2 provides descriptive information on the performance of each individual. There was no significant effect of the order of tool set presentation on the capuchins’ performance (PlatFlex/HandRig: H3 ¼ 2.44, P ¼ 0.49; PlatFlex/HandFlex: H3 ¼ 2.33, P ¼ 0.51; PlatRig/ HandFlex: H3 ¼ 4.00, P ¼ 0.26; PlatRig/HandRig: H3 ¼ 2.54, P ¼ 0.47). Therefore, we pooled the data from the different groups for subsequent analyses. The percentage of raking in the tool placed on the platform using the tool handed to the subject differed between conditions (Friedman test: c2 3 ¼ 25.97, N ¼ 10, P ¼ 0.00001; Fig. 4).
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220 Table 2 Summary of capuchins’ results
Carlotta Gal Paprica Pedro Pepe Roberta Robin Hood Rucola Sandokan Vispo Mean
Pretest
Test
Mean number of trials needed to achieve criterion
Number correct out of four trials for each condition PlatFlex/ HandRig
PlatFlex/ HandFlex
PlatRig/ HandFlex
50 52 90 38 40 40 44 30 32 24 44
4 4 4 3 3 4 4 4 4 4 3.8
4 4 4 4 4 4 4 4 4 2 3.8
4 4 4 4 3 4 4 4 4 4 3.9
Capuchins displayed significantly less raking in those conditions in which they received the flexible tool compared to those conditions in which they received the rigid tool (Wilcoxon tests: PlatFlex/ HandRig versus PlatFlex/HandFlex: T ¼ 0, P ¼ 0.005; PlatFlex/HandRig versus PlatRig/HandFlex: T ¼ 0, P ¼ 0.005; PlatRig/HandRig versus PlatFlex/HandFlex: T ¼ 0, P ¼ 0.008; PlatRig/HandRig versus PlatRig/HandFlex: T ¼ 0, P ¼ 0.005; PlatFlex/HandRig versus PlatRig/HandRig: T ¼ 2, P ¼ 0.14; PlatFlex/HandFlex versus PlatRig/ HandFlex: T ¼ 5.5, P ¼ 0.29). These results were already apparent in the first trial: nine out of 10 capuchins raked in the flexible tool using the rigid tool when the flexible tool was out of reach on the platform, while all 10 capuchins used the tool handed to them to retrieve food in those
217
conditions in which they received the flexible tool (binomial tests: PlatFlex/HandRig: P ¼ 0.039; PlatFlex/HandFlex: P ¼ 0.004; PlatRig/HandFlex: P ¼ 0.004). Moreover, in the PlatFlex/HandRig condition subjects retrieved the flexible tool from the platform significantly more often than they tried to insert into the tube the rigid tool handed to them (Wilcoxon test: T ¼ 1.5, N ¼ 10, P ¼ 0.003; Fig. 5). In contrast, in the PlatFlex/HandFlex and PlatRig/HandFlex conditions (when the platform tool was unnecessary to obtain food) subjects retrieved food with the flexible tool handed to them significantly more often than they tried to retrieve the platform tool (Wilcoxon tests: T ¼ 2.5, N ¼ 10, P ¼ 0.007; T ¼ 0, N ¼ 10, P ¼ 0.003; Fig. 5). In the PlatRig/HandRig condition, in which only rigid tools were available, subjects equally often retrieved the platform tool (median ¼ 87.5, IQR ¼ 25), probed into the tube directly with the tool handed to them (median ¼ 62.5, IQR ¼ 75; Wilcoxon test: T ¼ 9, N ¼ 10, P ¼ 0.11; Fig. 5) and discarded the latter tool (median ¼ 75, IQR ¼ 50; Wilcoxon test: T ¼ 16.5, N ¼ 10, P ¼ 0.83). Subjects’ performances in the PlatFlex/HandRig condition did not correlate with prior experience in tool-using experiments (Spearman rS ¼ 0.31, N ¼ 10, P ¼ 0.38) or with age (Spearman rS ¼ 0.18, N ¼ 10, P ¼ 0.63). DISCUSSION Capuchin monkeys engaged in sequential use of tools whose functional properties are not visually apparent. All capuchins successfully performed sequential tool use and nine out of 10 did so already in the first trial of the crucial condition (i.e. PlatFlex/HandRig, when the platform tool was flexible and the tool handed to the subject was rigid). In the conditions in which the platform tool
200
180
160
Number of trials to criterion
140 Carlotta Gal Paprica Pedro Pepe Roberta Robin Hood Rucola Sandokan Vispo
120
100
80
60
40
20
0
First tool set
Second tool set
Third tool set
Fourth tool set
Figure 3. Number of trials needed to achieve criterion by each subject in the pretest from the first to the fourth tool set presented. To advance to the test phase, the subject had to choose the correct tool in 16 consecutive choice trials for each set of tools.
218
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220
** Percentage of retrieving (or trying to retrieve) platform tool
** **
** 100
80
60
40
20
0
PlatFlex/HandRig
PlatFlex/HandFlex
PlatRig/HandFlex
PlatRig/HandRig
Condition Figure 4. Median (white square) IQR (box) and range (whiskers) of the percentages of retrieving the platform tool with the tool handed to the subject (for PlatFlex/HandRig and PlatRig/HandRig conditions) and of trying to retrieve the platform tool with the tool handed to the subject (for PlatFlex/HandFlex and PlatRig/HandFlex conditions). ** P < 0.01.
**
**
**
PlatFlex/HandRig
PlatFlex/HandFlex
PlatRig/HandFlex
Percentage of behaviour
100
80
60
40
20
0 PlatRig/HandRig
Condition Figure 5. Median (white square) IQR (box) and range (whiskers) of the percentages of retrieving, or trying to retrieve, the tool from the platform (in white) versus the percentages of retrieving, or trying to retrieve, food with the tool handed to the subject (in black) for each condition. ** P < 0.01.
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220
was unnecessary to obtain food (PlatFlex/HandFlex, PlatRig/HandFlex), capuchins used the flexible tool that was already within reach. These results demonstrate that subjects were sensitive to both the tool rigidity and the task requirements since they chose which tool to use accordingly. Our tasks posed the capuchins an unprecedented cognitive challenge since they needed to deal simultaneously with conflicting information that required flexible and hierarchical behaviour. When they received a tool they had to judge its suitability to meet the tube task requirements. If the tool was unsuitable, they needed to judge whether the demonstrated tool resting on the platform was appropriate to be used in the tube task and if they could retrieve it by using the tool handed to them. Note that looking at the tool presented on the platform was not sufficient to judge whether it was suitable to obtain the reward, since the tool functional property, flexibility, was not visually detectable. Flexibility could be inferred only on the basis of the behaviour of the tool when it was manipulated by the experimenter. This ability, present in great apes (Manrique et al., 2010) as well as in capuchins (Manrique et al., 2011), can be considered in terms of affordance learning, a process in which, by watching the object while it moves, the observer detects its affordances, such as dynamic properties and temporalespatial relations. During the second year of life, human infants become able to learn object affordances when only the object movement component is presented without a demonstrator performing it (e.g. Huang & Charman, 2005; Tennie, Call, & Tomasello, 2006). Experiments with a ‘ghost’ condition in chimpanzees have produced mixed results (Hopper, Lambeth, Schapiro, & Whiten, 2007; Hopper, Spiteri, Lambeth, Schapiro, & Whiten, 2008; Tennie et al., 2006). Future studies should assess whether capuchin monkeys are able to learn object affordances through a ghost condition. Since rigidity was not directly observable, it is conceivable that capuchins had to keep active in their working memory the properties of the platform tool (its rigidity) to organize her/his behaviour efficiently, which imposes an additional demand not present when judging tool length because length is a visible property (e.g. Martin-Ordas et al., 2012; Wimpenny et al., 2009). In these studies the tools were or were not long enough to reach the food. When long enough they could be used directly. When too short, the subjects had to retrieve one or several longer tools before attempting to get the food. The task required subjects to compare tool length with distance to the food visually or mentally (when tools and reward were not visible at the same time) and judge tool suitability accordingly. In a way, one could say that changes in the critical property (length) were gradual and unidirectional. Thus, a rule such as ‘when the tool available is too short then pick a longer tool’ could satisfactorily explain subjects’ success. However, when we talk about rigidity things are not that straightforward because being flexible is just the opposite of being rigid and, hence, the changes in tool properties are not a simple question of degree of flexibility but rather a matter of quality. It would be interesting to see how other animal species (e.g. New Caledonian crows) perform in an analogous task to the one we implemented. Wimpenny et al. (2009) argued that to qualify as goal directed, tool retrieval should occur only when required (i.e. when the tool in hand was not flexible enough to get the food). Not only did capuchins retrieve the flexible tool when required, but for those conditions in which the platform tool was rigid, capuchins used the flexible tool that was already within reach to obtain food. In these conditions in which tool retrieval was unnecessary, capuchins adopted the more efficient strategy from the beginning; thus their performance resembled more that of great apes than that of New Caledonian crows (Martin-Ordas et al., 2012; Mulcahy et al., 2005; Taylor et al., 2007; Wimpenny et al., 2009). It can be argued that a
219
full understanding of the PlatRig/HandRig condition required no probing at all, as neither tool was suitable for solving the task. In this ‘impossible’ condition capuchins showed no clear preference towards retrieving the platform tool compared to probing directly into the tube with the tool handed to them or discarding the latter tool. Capuchins made different attempts to get the reward with the tool(s), as already reported for gorillas and orang-utans in an analogous condition in which the available tool was not suitable for getting the reward (Mulcahy et al., 2005). In our opinion, this lack of preference for one of the tools reflects a lack of inhibitory control in the sight of food, rather than a lack of understanding of the task requirements. This is supported by observations that capuchins display a very active and persistent foraging style as well as a tendency to manipulate and combine objects that is difficult to restrain in the presence of or in searching for food (Fragaszy, Visalberghi, & Fedigan, 2004). Moreover, capuchin monkeys typically show worse capacities than chimpanzees to delay gratification, waiting for approximately 10e20 s (Anderson, Kuroshima, & Fujita, 2010) or 30e40 s (Pelé, Micheletta, Uhlrich, Thierry, & Dufour, 2011) in comparison to 120 s for chimpanzees (Beran & Evans, 2006). Our subjects spontaneously used two tools in sequence. In fact they were trained to criterion in the pretest but not to use a tool to retrieve another tool in the test. Although an associative process such as chaining (sensu Epstein, Kirshnit, Lanza, & Rubin, 1984) could not be completely ruled out, we think that it cannot satisfactorily account for our results. There are three reasons for rejecting the idea that once subjects ‘discovered’ the action of tool retrieval from the platform, they automatically performed this action and subsequently inserted the flexible tool into the tube to retrieve food, or that they relied on random probes for tools, followed by probes for food. First, in the PlatFlex/HandRig condition, in which the tool handed to them was unsuitable for the tube task, nine out of 10 subjects showed successful and spontaneous sequential tool use, using the tool to retrieve another tool from their first trial. Second, in those conditions in which retrieving the platform tool was unnecessary all capuchins used the tool that was already within reach to obtain food from their first trial. Third, if during the pretest subjects had simply learnt to associate the rigid tool with the platform task, in the PlatRig/HandRig condition, in which the two tools available were rigid, we should have found more attempts to rake in the platform tool, as evident in the PlatFlex/HandRig condition, in which only the tool handed to them was rigid. In contrast, capuchins almost equally tried to insert the tool handed to them into the tube, discarded it and tried to retrieve the platform tool. Moreover, prior tool-using experience and age did not seem to affect subjects’ performance. The most experienced capuchin (Carlotta) did not even outperform the least experienced one (Rucola). More than two decades ago Carlotta participated in an experiment in which she had to use a shorter stick to rake in a longer one (Parker & Potì, 1990). She was then 2.5 years old and her success rate varied from 33% to 66% depending on the position of the short stick relative to the long stick. Moreover, when tested in several other tool tasks she performed well. In contrast, Rucola (who was 12 years old) had never been tested in tool tasks. Despite the different previous experience and age, both subjects were equally successful in sequential tool use. The ability to organize one’s own actions hierarchically is also supported by experiments that have assessed planning during sequential responding to computer-generated stimuli by capuchin monkeys (Beran & Parrish, 2012). In two experiments, eight capuchin monkeys selected five arbitrary stimuli in distinct locations on a computer monitor in a learned sequence. In experiment 1, shift trials occurred in which the second and third stimuli were
220
G. Sabbatini et al. / Animal Behaviour 87 (2014) 213e220
transposed when the first stimulus was selected by the animal. In experiment 2, mask trials occurred in which all remaining stimuli were masked after the monkey selected the first stimulus. Capuchins made more mistakes on trials in which the locations of the second and third stimuli were interchanged than on trials in which locations were not interchanged, suggesting they had already planned to select a location that no longer contained the correct stimulus. When mask trials occurred, capuchins performed at levels significantly better than chance for the first and the second selections on a trial. In conclusion, the current study demonstrated that capuchin monkeys were able to organize their actions in the hierarchically correct sequence using tools differing in their rigidity, a nonvisually detectable property. Moreover, they were able to manage conflicting information when handed a tool that is both functional and nonfunctional depending on the apparatus. Our present findings also contribute to the growing body of research revealing that capuchins also use tools to act upon another object. One limitation of our study was that the only possible form of sequential tool use consisted of retrieving the flexible tool on the platform with the rigid tool handed to the subject. It could be interesting to present capuchins with the possibility to do the opposite (retrieve a reward placed on a platform). Future studies on capuchins should also include the possibility to use more than two tools in sequence, thus increasing the attentional and cognitive demands of the task. Acknowledgments This research was supported by the European Community’s Seventh Framework Programme FP7/2007-2013, ‘Challenge 2e Cognitive Systems, Interaction, Robotics’, under grant agreement No FP7-ICT-IP-231722, project ‘IM-CLeVeReIntrinsically Motivated Cumulative Learning Versatile Robots’ and the PNR-CNR Aging Program 2012-2014. We thank the keepers of the Primate Center of ISTC-CNR for help with animal testing, the Roma Capitale-Museo Civico di Zoologia and the Fondazione Bioparco for hosting the Unit of Cognitive Primatology and Primate Centre, and the two anonymous referees and Dr Ximena Nelson for their useful comments. Supplementary Material Supplementary material associated with this article is available, in the online version, at http://dx.doi.org/10.1016/j.anbehav.2013. 10.033. References Anderson, J. R., & Henneman, M. C. (1994). Solutions to a tool-use problem in a pair of Cebus apella. Mammalia, 58, 351e361. Anderson, J. R., Kuroshima, H., & Fujita, K. (2010). Delay of gratification in capuchin monkeys (Cebus apella) and squirrel monkeys (Saimiri sciureus). Journal of Comparative Psychology, 124, 205e210. Beran, M. J., & Evans, T. A. (2006). Maintenance of delay of gratification by four chimpanzees (Pan troglodytes): the effects of delayed reward visibility, experimenter presence, and extended delay intervals. Behavioural Processes, 73, 315e 324. Beran, M. J., & Parrish, A. E. (2012). Sequential responding and planning in capuchin monkeys (Cebus apella). Animal Cognition, 15, 1085e1094. Bird, C. D., & Emery, N. J. (2009). Insightful problem solving and creative tool modification by captive nontool-using rooks. Proceedings of the National Academy of Sciences, 106, 10370e10375.
Epstein, R., Kirshnit, C. E., Lanza, R. P., & Rubin, L. C. (1984). ‘Insight’ in the pigeon: antecedents and determinants of an intelligent performance. Nature, 308, 61e 62. Fragaszy, D. M., Visalberghi, E., & Fedigan, L. M. (2004). The complete capuchin. The biology of the genus Cebus. Cambridge: Cambridge University Press. Hihara, S., Obayashi, S., Tanaka, M., & Iriki, A. (2003). Rapid learning of sequential tool use by macaque monkeys. Physiology & Behaviour, 78, 427e434. Hopper, L. M., Lambeth, S. P., Schapiro, S. J., & Whiten, A. (2008). Observational learning in chimpanzees and children studied through ‘ghost’ conditions. Proceedings of the Royal Society B, 275, 835e840. Hopper, L. M., Spiteri, A., Lambeth, S. P., Schapiro, S. J., & Whiten, A. (2007). Experimental studies of traditions and underlying transmission processes in chimpanzees. Animal Behaviour, 73, 1021e1032. Huang, C.-T., & Charman, T. (2005). Gradations of emulation learning in infants’ imitation of actions on objects. Journal of Experimental Child Psychology, 92, 276e302. Klüver, H. (1933). Behavior mechanisms in monkeys. Chicago: University of Chicago Press. Köhler, W. (1925/1976). The mentality of apes. New York: Liveright. Lynch Alfaro, J. W., Boubli, J. P., Olson, L. E., Di Fiore, A., Wilson, B., GutiérrezEspeleta, G. A., et al. (2012). Explosive Pleistocene range expansion leads to widespread Amazonian sympatry between robust and gracile capuchin monkeys. Journal of Biogeography, 39, 272e288. Lynch Alfaro, J. W., De Souza, Silva, J., & Rylands, A. B. (2012). How different are robust and gracile capuchin monkeys? An argument for the use of Sapajus and Cebus. American Journal of Primatology, 14, 1e14. Mannu, M., & Ottoni, E. B. (2009). The enhanced tool-kit of two groups of wild bearded capuchin monkeys in the Caatinga: tool making, associative use, and secondary tools. American Journal of Primatology, 71, 242e251. Manrique, H. M., Gross, A., & Call, J. (2010). Great apes select tools on the basis of their rigidity. Journal of Experimental Psychology: Animal Behavior Processes, 36, 409e422. Manrique, H. M., Sabbatini, G., Call, J., & Visalberghi, E. (2011). Tool choice on the basis of rigidity in capuchin monkeys. Animal Cognition, 14, 775e786. Martin-Ordas, G., Schumacher, L., & Call, J. (2012). Sequential tool use in great apes. PLoS ONE, 7, e52074. Mulcahy, N. J., Call, J., & Dunbar, R. I. M. (2005). Gorillas (Gorilla gorilla) and orangutans (Pongo pygmaeus) encode relevant problem features in a tool-using task. Journal of Comparative Psychology, 119, 23e32. Parker, S. T., & Potì, P. (1990). The role of innate motor patterns in ontogenetic and experiential development of intelligent use of sticks in cebus monkeys. In S. T. Parker, & K. R. Gibson (Eds.), ‘Language’ and intelligence in monkeys and apes (pp. 219e243). New York: Cambridge University Press. Pelé, M., Micheletta, J., Uhlrich, P., Thierry, B., & Dufour, V. (2011). Delay maintenance in tonkean macaques (Macaca tonkeana) and brown capuchin monkeys (Cebus apella). International Journal of Primatology, 32, 149e166. Sabbatini, G., Truppa, V., Hribar, A., Gambetta, B., Call, J., & Visalberghi, E. (2012). Understanding the functional properties of tools: chimpanzees (Pan troglodytes) and capuchin monkeys (Cebus apella) attend to tool features differently. Animal Cognition, 15, 577e590. Santos, L. R., Rosati, A., Sproul, C., Spaulding, B., & Hauser, M. D. (2005). Meansmeans end tool-use in cotton-top tamarins (Saguinus oedipus): finding the limits on primates’ knowledge of tools. Animal Cognition, 8, 236e246. Shumaker, R., Walkup, K. R., & Beck, B. B. (2011). Animal tool behavior: the use and manufacture of tools by animals. Baltimore: The Johns Hopkins University Press. Taylor, A. H., Elliffe, D., Hunt, G. R., & Gray, R. D. (2010). Complex cognition and behavioural innovation in New Caledonian crows. Proceedings of the Royal Society B, 277, 2637e2643. Taylor, A. H., Hunt, G. R., Holzhaider, J. C., & Gray, R. D. (2007). Spontaneous metatool use by New Caledonian Crows. Current Biology, 17, 1504e1507. Tennie, C., Call, J., & Tomasello, M. (2006). Push or pull: imitation versus emulation in great apes and human children. Ethology, 112, 1159e1169. Visalberghi, E., Addessi, E., Truppa, V., Spagnoletti, N., Ottoni, E., Izar, P., et al. (2009). Selection of effective stone tools by wild bearded capuchin monkeys. Current Biology, 19, 213e217. Visalberghi, E., & Limongelli, L. (1994). Lack of comprehension of cause-effect relations in tool-using capuchin monkeys (Cebus apella). Journal of Comparative Psychology, 108, 15e22. Visalberghi, E., & Trinca, L. (1989). Tool use in capuchin monkeys: distinguishing between performing and understanding. Primates, 30, 511e521. Warden, C. J., Koch, A. M., & Fjeld, H. A. (1940). Instrumentation in Cebus and Rhesus monkeys. The Journal of Genetic Psychology, 56, 297e310. Wimpenny, J. H., Weir, A. A. S., Clayton, L., Rutz, C., & Kacelnik, A. (2009). Cognitive processes associated with sequential tool use in New Caledonian Crows. PLoS ONE, 4, e6471.