You are on page 1of 10

Journal of Experimental Psychology: © 2010 American Psychological Association

Animal Behavior Processes 0097-7403/10/$12.00 DOI: 10.1037/a0020274


2011, Vol. 37, No. 1, 10 –19

Learning and Generalization of Tool Use by Tufted Capuchin Monkeys


(Cebus apella) in Tasks Involving Three Factors:
Reward, Tool, and Hindrance

Kazuo Fujita, Yoshiaki Sato, and Hika Kuroshima


Kyoto University

We tested 4 captive tufted capuchin monkeys (Cebus apella) for their understanding of physical causality
in variations of a 2-choice tool use task, 1 alternative of which allowed the monkeys easier access to food.
Our monkeys, who had been adept at this task involving 2 items, that is, tool and food, quickly learned
3-term problems involving food, tool, and 1 type of hindrance (an obstacle or a trap, which could prevent
success). All of the monkeys generalized their performance to new problems with the other type of
hindrance and those with another familiar tool. These results suggest flexibility of their abilities to
process complex physical information comprising 3 items in the environment, that is food–tool–
hindrance spatial relationships. Such flexibility also implies that capuchin monkeys may possess
rudimentary understanding of causal relationships involved in tool use.

Keywords: causal understanding, physical intelligence, tool use, tufted capuchin monkeys (Cebus apella)

Tool behavior by nonhuman animals has been widely analyzed herd, 1910), but how these species recognize causality involved in
both in captivity and in the field. Tool use involves two aspects: this complex behavior remains controversial (Fragaszy, Visal-
the technique to maneuver the tool and the causal cognition at berghi, & Fedigan, 2004; Hauser & Santos, 2007; Povinelli, 2000;
various levels to understand why the tool use leads to an otherwise Tomasello & Call, 1997; Visalberghi & Fragaszy, 2006; Visal-
unattainable goal. Abundant research has shown techniques and berghi & Limongelli, 1996; Visalberghi & Tomasello, 1998).
repertoires of animals’ tool behavior (see Anderson, 2006; Beck, Visalberghi and her coworkers (Limongelli, Boysen, & Visal-
1980, for comprehensive reviews across the animal kingdom; berghi, 1995; Visalberghi, Fragaszy, & Savage-Rumbaugh, 1995;
Emery & Clayton, 2004; Kacelnik, Chappell, Kenward, & Weir, Visalberghi & Limongelli, 1994, 1996; Visalberghi & Trinca,
2006, for reviews on corvids; Tomasello & Call, 1997, for a review 1989) investigated the causal comprehension in tool use tasks by
on primates). Among primates, most work has focused on chim- tufted capuchin monkeys (Cebus apella), the most skillful tool
panzees (e.g., Boesch & Boesch-Achermann, 2000; Goodall, 1986; users among nonape primates, comparing them with great apes.
Matsuzawa, 2001; McGrew, 1992, 1994; Ohashi, 2006; Visal- They tested capuchins under several variations of the tube task in
berghi & Fragaszy, 2002; Whiten et al., 1999, 2001; Yamakoshi, which the monkeys had to insert a stick into the tube to push the
2001, 2004) and capuchin monkeys (e.g., Fragaszy, Visalberghi, & reward out. In a variation, the tube had a trap in the middle. Thus,
Fedigan, 2004; Ottoni & Mannu, 2001; Visalberghi, 1990). Many the monkeys had to choose the side to insert the stick into so as to
experimental studies of tool behavior have been conducted on keep the reward from dropping into the trap, that is, so that the trap
primates (chimpanzees: Köhler, 1921; capuchin monkeys: Klüver, would not be on the passage of the reward pushed by the stick.
1933; rhesus macaques: Nellmann & Trendelenburg, 1926; Shep- Most monkeys failed in this “trap tube” task; only one capuchin
succeeded. Besides, even when the tube was placed upside down
(inverted trap tube) so that there was no functional trap, the sole
This article was published Online First August 16, 2010. solver kept inserting the stick across the trap as if she had avoided
Kazuo Fujita and Hika Kuroshima, Graduate School of Letters, Kyoto
the ineffective trap. Visalberghi and colleagues concluded that this
University; and Yoshiaki Sato, Faculty of Letters, Kyoto University.
Financial support was provided by the Japanese Society for the Promo- monkey did not understand the function of the trap, but used only
tion of Science (JSPS) Grants-in-Aid for Scientific Research 13410026, the distance from the reward to the end of the tube as an associa-
17300085, and 20220004 to Kazuo Fujita, by a JSPS Research Fellowship tive cue (Visalberghi & Limongelli, 1994). In contrast, two suc-
for Young Scientists to Hika Kuroshima, by the Japan Ministry of Educa- cessful chimpanzees of five tested in the trap tube tasks showed
tion, Culture, Sport, Science, and Technology (MEXT) 21st Century COE more flexible performances than capuchin monkeys, suggesting
Program D-10 to Kyoto University, and by the MEXT global COE Pro- that they understood the causal relationship between action and
gram D-07 to Kyoto University. We thank Primate Research Institute, outcome to some degree (Limongelli et al., 1995). However,
Kyoto University, for providing monkeys through its Cooperative Research
success may not be shown in all situations; Povinelli (2000)
Program. We are also grateful to James R. Anderson for editing a version
of the article. showed that chimpanzees failed in the inverted trap condition as
Correspondence concerning this article should be addressed to Kazuo capuchins in Visalberghi and Limongelli (1994). On the other
Fujita, Graduate School of Letters, Kyoto University, Yoshida-honmachi, hand, in more recent tests, two orangutans and one chimpanzee of
Sakyo, Kyoto 606-8501, Japan. E-mail: kfujita@bun.kyoto-u.ac.jp 10 apes (orangutans, gorillas, chimpanzees, and bonobos) suc-

10
TOOL USE BY CAPTIVE CAPUCHINS 11

ceeded in the modified trap tube task quickly in which the subjects (Anderson & Henneman, 1994). In the Hauser-type two-choice
were allowed to choose among two actions, namely pushing out or tasks, capuchins, unlike tamarins, converted the orientation of the
raking in, for themselves, although the apes still failed in the tool (Cummins-Sebree & Fragaszy, 2005b). Cummins-Sebree and
original trap tube task in which pushing out was the only way to Fragaszy (2005a) also reported that some capuchins mastered the
obtain the reward (Mulcahy & Call, 2006). In another type of tube correct manipulation of a hoe-like tool on the substrate with
task where subjects were required to modify tools to fit them for barriers or holes to retrieve food. Capuchins are capable of using
the opening of the tube, chimpanzees, bonobos, and an orangutan multiple tools in sequence; they can crack open walnuts with a
showed better performances than capuchins, although the apes and stone and then extract the meat with a stick (Westergaard &
the capuchins solved the problems as well (Visalberghi et al., Suomi, 1993). Furthermore, they show different grips (power grips
1995). or precision grips) when manipulating tools for different purposes
There is an argument that the inverted trap condition may not be (Westergaard & Suomi, 1997).
a critical control to assess how the subjects understand the causal- In this study, we conducted additional analyses of capuchins’
ity involved in the task because there is no extra cost caused by the causal cognition in the Hauser-type two-choice tasks, focusing on
“error” (Seed, Tebbich, Emery, & Clayton, 2006). Indeed, humans the capuchins’ learning and generalization of the task involving
also showed such unnecessary bias (Silva, Page, & Silva, 2005). spatial relationships of the three elements of the tool use, namely,
Thus, we should consider the possibility that the primates under- a reward (goal), tool (means), and environment. The third element
stood the function of the trap but still stuck to the strategy they had is often important in daily situations. In the present context, it was
learned in the preceding training. physical hindrance, which could prevent successfully obtaining
In a previous study, Fujita, Kuroshima, and Asai (2003) tested food. Although quite a few studies have analyzed the effects of
tufted capuchin monkeys for their causal understanding in another environmental objects such as a projection on the platform (capu-
tool use situation. We used the two-choice method devised by chins: Cummins-Sebree & Fragaszy, 2005a; Fujita et al., 2003)
Hauser and his coworkers with tamarins (Hauser, 1997; Santos, and a pitfall (chimpanzees: Limongelli et al., 1995; Povinelli,
Rosati, Sproul, Spaulding, & Hauser, 2005), in which the two 2000; Visalberghi & Limongelli, 1996; capuchins: Cummins-
alternatives were prearranged by the experimenter. Each alterna- Sebree & Fragaszy, 2005a; Fujita et al., 2003; Visalberghi &
tive included one piece of food and one tool, but only one alter- Limongelli, 1994, 1996; vervet and tamarin monkeys: Santos,
native led to easy access to the reward. All of the capuchins easily Pearson, Spaepen, Tsao, & Hauser, 2006), we stress here that the
solved the problems and generalized their performances to novel two-choice tasks have several advantages for analyzing causal
arrangements and tools. However, successful performances did not understanding. First, the three-term spatial relationships among a
generalize to problems that included a hindrance located on the food (target), tool (means), and hindrance (an obstacle or trap) in
path of the tools and the rewards. Therefore, capuchins appear to the two-choice experiments simulate the essential parts of the trap
have good understanding of how spatial relationships incorporat- tube problem (a food, stick, and trap tube) in a simpler and more
ing two terms, tool and reward, cause success but poor understand- obvious fashion (Povinelli, 2000). Second, the Hauser-type tasks
ing of incorporating three terms, tool, reward, and environment. facilitate the experimental control of spatial relationships among
The failure on the three-term task could explain why capuchins objects on the substrate, so that we can set conditions to conduct
failed in the trap tube task. stricter analyses of the monkeys’ cognitive ability to handle the
However, capuchin monkeys may have deeper understanding of three factors at a time. Third, these tasks require no pushing-out
causality. Like chimpanzees’ nut cracking (Sugiyama & Koman, behavior, which may interfere with learning how to avoid the trap
1979) and long-tailed macaques’ shell cracking (Malaivijitnond, in the trap tube tasks (Mulcahy & Call, 2006). Fourth, these tasks
Lekprayoon, Tandavanittj, Panha, Cheewatham, & Hamada, are more suitable for cross-species comparisons of tool use ability
2007), capuchins crack open nuts with a stone, not only in captive than other approaches such as the trap tube task because the
(Anderson, 1990; Pouydebat, Gorce, Bels, & Coppens, 2006) and two-choice tasks require participants simply to pull one of the tools
semiwild situations (Ottoni, Dogo de Resende, & Izar, 2005; presented, with no particular dexterity required.
Ottoni & Mannu, 2001, 2003) but also in the wild (Fragaszy, Izar, The failure of capuchins in the three-term tasks in Fujita et al.
Visalberghi, Ottoni, & Gomes de Oliveira, 2004; Moura & Lee, (2003) may have been due to a lack of ability. However, in fact, the
2004; Waga, Dacier, Pinha, & Tavares, 2006; see also Izawa & monkeys received only 24 unique test trials for each of the two
Mizuno, 1977). Capuchins also extract food with probing sticks in types of hindrances, trap and obstacle, which may not have been
captivity (Westergaard, Lundquist, Haynie, Kuhn, & Suomi, 1998; sufficient to move beyond an understanding of two-term spatial
Westergaard, Lundquist, Kuhn, & Suomi, 1997) and in free- relationships. To test the two possibilities, we first intensively
ranging situations (Lavallee, 1997), similar to wild chimpanzees trained capuchins that had learned the basic two-term problems on
gathering termites or ants (Goodall, 1986; Nishida, 1973). Capu- the same 24 three-term problems; then we examined how the
chins may even transport tools (Cleveland, Rocca, Wendt, & monkeys generalized their three-term performances to various new
Westergaard, 2004; Visalberghi, Spagnoletti, et al., 2009; see also problems, including ones presenting new tools and hindrances.
Jalles-Filho, da Cunha, & Salm, 2001). In fact, capuchins show Before training the monkeys on three-term problems, we tested
various similarities to apes in tool use tasks. For instance, capu- whether our capuchins remembered how to succeed in two-term
chins, as well as chimpanzees, bonobos, and orangutans, modified problems involving only food and tool after a long period with no
some of the tools potentially usable in the tube task (Visalberghi et relevant testing. Then we trained the same monkeys on three-term
al., 1995; Visalberghi & Trinca, 1989). Capuchins not only se- problems involving a reward, tool, and hindrance (Experiment 1).
lected a stick of the appropriate thickness, but they also made thick After reaching criterion, the monkeys were tested with arrange-
sticks thinner so as to dip for honey with them through a small hole ments that they never experienced in the training (Experiment 2).
12 FUJITA, SATO, AND KUROSHIMA

Next, they were given problems involving the kind of hindrance Fujita et al., 2003, for an image). The scoop had three rectangular
not presented previously (trap problems for monkeys who went corners at the top (see Figure 3 in Fujita et al., 2003, for an image).
through the obstacle problems in the training, and vice versa; It had two horizontal bars that could catch a food reward; thus, its
Experiment 3). In the final test, we replaced the tool with a familiar function is a little more complicated than the hook. Two identical
new tool used in the previous two-term problems (Experiment 4). tools of each kind were used. Their approximate width, length,
diameter, and weight were as follows: hook, 6.5 cm, 22 cm, 10
General Method mm, and 27 g, respectively; scoop, 8 cm, 23 cm, 10 mm, and 29 g,
respectively.
An obstacle and a trap were used as hindrances, the third
Subjects
element (environment) mentioned in the introduction. Obstacles
The same four tufted capuchin monkeys (Cebus apella) from the were rectangular plastic erasers wrapped in yellow-green vinyl
previous study (Fujita et al., 2003) participated: Heiji, 10-year-old tape, 2 cm wide ⫻ 4 cm long ⫻ 1 cm high, that were stuck with
male; Zilla, 10-year-old female; Kiki, 9-year-old female; and Pig- double-face tape to the substrate tray. Traps were visible pitfalls, 4
mon, 6-year-old male (ages at the start of the present study). The cm wide ⫻ 3 cm long, on the substrate tray. All of the traps were
monkeys, born and reared in a group cage at the Primate Research located within 19 to 41 cm of the edge of the substrate tray
Institute, Kyoto University, were provided by the institute by way adjacent to the experimental box.
of its Cooperative Research Program. They lived in a group of
seven individuals at the Graduate School of Letters, Kyoto Uni-
versity. The housing consisted of 10 interconnected cages, each
Problems
sized approximately 60 cm wide ⫻ 70 cm long ⫻ 80 cm high, Each problem provided two alternatives for the monkeys on the
which complied with the Guide for the Care and Use of Labora- right and the left sides of the substrate, each of which had a
tory Primates ( Primate Research Institute, Kyoto University, combination of a food reward (a 5-mm cube of sweet potato), a
2006). The monkeys were not deprived of food or water. Rewards tool, and a hindrance (obstacle or trap). The grips of the tools were
were given as a part of their daily ration, the remainder of which located within reach of the monkeys about 16 cm distant from the
was given after all scheduled experiments were completed. They sliding door of the experimental box. In one of the two alterna-
were tested individually in the test room. The experiments were tives, simply pulling the tool allowed the monkey easy access to
conducted with the approval of the Animal Experiment Committee the reward. In the other, however, the food was arranged so that it
of the Graduate School of Letters, Kyoto University. The monkeys would not get hooked by the pulled tool, the obstacle would block
had various laboratory experiences, including experiments on tool use the path of the pulled tool, or the food would fall into the trap.
(Fujita et al., 2003), operant discriminations (Anderson, Kuwahata, Although skillful or random maneuvering of the tool in the latter
Kuroshima, Leighty, & Fujita, 2005; Fujita & Giersch, 2005), social alternative could occasionally lead to obtaining a reward, only
intelligence (Anderson, Kuroshima, Hattori, & Fujita, 2005; Ander- pulling the former was defined as a correct choice. In fact, obtain-
son, Kuroshima, Kuwahata, & Fujita, 2004; Fujita, Kuroshima, & ing the reward from the wrong alternative was rare.
Masuda, 2002; Hattori, Kuroshima, & Fujita, 2005; Kuroshima, We designed several sets of problems; a set contained 12 pairs
Fujita, Adachi, Iwata, & Fuyuki, 2003; Kuroshima, Fujita, Fuyuki, of spatial arrangements of food, tool, and hindrance, each pre-
& Masuda, 2002), and mirror-image stimulation (Paukner, Ander- sented once in a random order in each session. The arrangements
son, & Fujita, 2004) before this study. were all different except that those in Problems 1⬘ through 6⬘ in
each set were right–left switches of Problems 1 through 6. See the
Apparatus Method section and figures of each experiment for the sets used in
each test.
The experimental box was made of transparent acrylic board,
measuring 46 cm wide ⫻ 46 cm long ⫻ 52 cm high. One wall of
the box had a door that could be slid upward to make an opening Procedure
25 cm wide ⫻ 3 cm high at the bottom through which the monkeys
reached their arms toward tools to collect a reward. See Fujita et The door of the box was closed before each trial. The experi-
al. (2003) for a photographic view of the apparatus. menter (E) placed the opaque panel against the box to interrupt the
A white bubbled vinyl chloride tray was used as a substrate for monkey’s view. E then arranged two combinations of food, tool,
tool use, measuring 44 cm wide ⫻ 58 cm long. Three rails on the and hindrance on the substrate tray. The panel was then removed
tray, one in the center and one about 10 cm from either edge, and, after about 6 s, E opened the sliding door by 3 cm to allow the
formed two channels and guided the manipulated tools in the monkey to choose a tool. E closed the door as soon as the monkey
appropriate direction. A black opaque panel made of vinyl chlo- pulled the appropriately arranged tool and obtained the reward
ride, measuring 47 cm wide ⫻ 60 cm long, was used to block the (defined as correct trials) or pulled the inappropriately arranged
monkey’s view when the experimenter prepared food, tool, and tool, which did not yield the reward (defined as incorrect trials).
hindrance on the substrate for a trial. Intertrial intervals were about 30 s. Each session consisted of 12
Two kinds of tool made of baked black nontoxic clay (Fimo), trials throughout the present series of experiments, as in Fujita et
also used in the previous study (hook and scoop in Fujita et al., al. (2003). One session was run each day. All sessions were
2003), were used. The tools were designed to collect an appropri- videotaped. E recorded the monkeys’ first pull of the tool as their
ately placed food reward by simply pulling them. The hook was a choice. Although the monkeys’ responses were obvious, an assis-
straight stick with a curved top like an inverted J (see Figure 1 in tant coded the monkeys’ choice in randomly chosen three sessions
TOOL USE BY CAPTIVE CAPUCHINS 13

of the recovery phase described below. The interobserver matching 1 2 3 4 5 6


of the recorded choice was 100%.
(a) obstacle set A
The subjects were tested for their long-term retention of the
previously trained performances on the 12 two-term problems
involving food and hook tool used in Experiment 1 of Fujita et al.
(2003) after 2 years 5 months with no relevant experience. We (b) obstacle set B
planned to repeat sessions until the monkeys reached the original
criterion (Fujita et al., 2003) of 10 or more correct trials of 12 in
two consecutive sessions. In fact, all four monkeys immediately (c) trap set A
performed almost perfectly and required only two sessions, the
minimum number, to reach the criterion. Heiji showed 12 and 11
correct choices of 12 in the first and the second sessions, respec- (d) trap set B
tively; Zilla showed 10 and 10; Kiki showed 11 and 12; Pigmon
showed 12 and 12. Thus, it is evident that tufted capuchin monkeys
obst obst obst trap
seem to possess robust and impressive long-term memory for this (e) 㪈㪉
type of cognitive skill. They went to Experiment 1 immediately p<0.05 (/12 trs.)
㪈㪇
after this recovery phase.

No. of Correct Choice



1st session
Experiment 1: Three-Term Training 㪍

㪋 2nd session
This experiment aimed to train the monkeys on three-term
problems involving food, a tool, and a hindrance. In Fujita et al. 㪉
(2003), no monkey was successful in generalization to these prob- 㪇
lems after being trained only on two-term problems. We hypoth- 㪟㪼㫀㫁㫀 㪱㫀㫃㫃㪸 㪢㫀㫂㫀 㪧㫀㪾㫄㫆㫅
esized two possible accounts for this result. One is that the failure Subjects
was caused by cognitive limits of tufted capuchin monkeys, that is,
because the species cannot understand the three-term spatial rela- Figure 1. (a)–(d) Arrangement of food, hook, and hindrance (obstacle or
tionship. The other is simple preseveration with the previously trap) in Experiment 1: the obstacle condition Set A (a) and Set B (b) and
the trap condition Set A (c) and Set B (d). These arrangements are the same
extensively trained strategy with the two-term spatial relationship
as those in Fujita et al. (2003). (e) The number of correct choices of 12 by
required in the original problems. If the former is true, tufted
each monkey in each session of the generalization test in Experiment 2.
capuchin monkeys would simply have difficulty learning three- Heiji, Zilla, and Kiki were tested in new arrangements of the obstacle task
term problems, or, if the monkeys manage to solve the problems, and Pigmon was tested in those of the trap task.
they would show poor generalization to variously modified novel
problems. Alternatively, if the latter is true, the monkeys would
learn three-term problems without much difficulty and, once
learned, they would show good generalization to novel problems. In Fujita et al. (2003), selection of correct options in three-term
problems was not above chance level in 24 unique test trials. In
this experiment, however, the same monkeys easily learned to
Procedure select an appropriate alternative in the three-term task. This sug-
gests that training on two-term problems did not lead to insight
The hook was used as a tool as in the recovery phase. The
into three-term relationships involved in these types of tool tasks.
problems were the four sets shown in Figures 1a–1d, two of which
This training on the three-term task was in fact faster than their
were for the obstacle condition and the other two were for the trap
first training on the two-term task, which took 15–19 sessions.
condition. All sets were those used in Fujita et al. (2003).
This finding implies that this species is indeed able to process such
The four sets were each assigned to each individual: Obstacle
higher order, more complicated, causal relationships. However, it
Set B (see Figure 1b) to Heiji, Trap Set B (see Figure 1d) to Zilla,
was necessary to test whether this learning would generalize to
Obstacle Set A (see Figure 1a) to Kiki, and Trap Set A (see Figure
novel situations before concluding that a simple learning set can-
1c) to Pigmon. Training continued until the monkeys reached the
not account for success with the three-term task. The following
same criterion as in the recovery phase, 10 or more correct of 12
experiments addressed this question using variously modified
trials in two consecutive sessions.
novel problems.
Because Zilla did not reach this criterion after 25 sessions, she
was switched to Obstacle Set B (see Figure 1b), and the experi-
ment continued. Experiment 2: Generalization to New Arrangements of
the Same Type of Hindrance
Results and Discussion Although the capuchin monkeys learned to perform well on
three-term problems, they could have learned each single problem
Heiji, Kiki, and Pigmon reached criterion in about 10 sessions without understanding a general rule leading to success. This
(10, 8, 11 sessions, respectively), as did Zilla when she was given experiment was designed to test whether the monkeys would
Obstacle Set B (11 sessions). generalize their three-term performances to untrained problems.
14 FUJITA, SATO, AND KUROSHIMA

The monkeys were tested with the other set of the same type of Test phase. The type of hindrance was switched to Trap Sets
hindrance used in Experiment 1. A and B (see Figures 1c and 1d) for Heiji, Zilla, and Kiki, who had
been tested on the obstacle condition, and to Obstacle Sets A and
Procedure B (see Figures 1a and 1b) for Pigmon, who had been tested on the
trap condition.
The same tool (hook) was used as in Experiment 1, and the We conducted two test sessions in a row for each of the two sets,
monkeys were presented with the problem sets not used during comprising 48 test trials. Two sessions per set were run succes-
training in the first experiment: Obstacle Set A (see Figure 1a) for sively. The order of sets was counterbalanced across monkeys, as
Heiji and Zilla, Obstacle Set B (see Figure 1b) for Kiki, and Trap is shown in Figure 2.
Set B (see Figure 1d) for Pigmon. We conducted two sessions of
12 trials as a generalization test.
Results and Discussion
Results and Discussion Confirmation phase. All of the monkeys immediately
reached criterion. Heiji, Zilla, Kiki, and Pigmon chose correctly in
All of the monkeys showed good generalization to untrained 22, 20, 21, and 22 trials, respectively, of 24 trials.
problems. Figure 1e shows each monkey’s number of correct Test phase. Figure 2 shows each monkey’s number of correct
choices. Two of the monkeys (Heiji, p ⫽ .039, and Kiki, p ⬍ .001, choices in each session of the test phase. Three of the monkeys
binomial tests) attained statistical significance in the first session, (Heiji, Kiki, and Pigmon) showed good scores in the test, although
and the others (Zilla, p ⫽ .039, and Pigmon, p ⫽ .006) reached it they may not have attained statistical significance in the very first
in the second session. If we combine the two sessions, all monkeys session. However, there appears to have been no improvement in
performed at above chance ( p ⬍ .001 for Heiji, p ⫽ .023 for Zilla, the second session. If we combine the two sessions for each test,
p ⬍ .001 for Kiki, and p ⫽ .002 for Pigmon). most of the monkeys’ scores were above chance (first test: p ⫽
The results suggest that the monkeys had not simply learned to .064, p ⫽ .152, p ⫽ .002, p ⫽ .007; second test: p ⬍ .001, p ⫽
perform on each arrangement in Experiment 2. That is, they are .152, p ⫽ .007, p ⫽ .023, for Heiji, Zilla, Kiki, and Pigmon,
likely to have learned some more general rule about how to solve respectively). Finally, if all of the test sessions were combined, all
the problems presented in training that was applicable to novel performances were above chance (binomial tests: p ⫽ .029 for
situations where the spatial relationship among food, tool, and Zilla, p ⬍ .001 for Heiji, Kiki, and Pigmon).
hindrance was altered. As the monkeys committed more errors than in the previous
experiment, we looked into each configuration of problems. Three
Experiment 3: Generalization to a New Type of monkeys tested in trap sets performed poorly in two types of
Hindrance problems: 3 and 3⬘ in Trap Set A (6 correct/12 trials for the data
for three monkeys summed) and 4 and 4⬘ in Trap Set B (5/12).
In this experiment, we further examined what the monkeys had
However, they performed much better in similar problems: 6 and
learned in the preceding experiments; specifically, the possibility
6⬘ in Trap Set B (10/12) and 5 and 5⬘ in Trap Set A (8/12). Thus,
of hindrance-specific learning was examined. The monkeys were
it is difficult to determine whether these poorly performed prob-
presented the same type of hindrance throughout Experiments 1
lems were due to certain arrangements or to a random error.
and 2 (obstacle for Heiji, Zilla, and Kiki, and trap for Pigmon), but
We also compared two seemingly easy problems in which only
in this experiment, obstacle sets and trap sets were switched (trap
one option had a hindrance (1, 1⬘, 2, and 2⬘ in each set) and four
for Heiji, Zilla, and Kiki, and obstacle for Pigmon).
more difficult problems having a hindrance in each option (all

Procedure
Confirmation phase. Zilla and Pigmon received a single ad- trap obst
ditional session of the same three-term problems used in Experi- 㪈㪉 * + obst-A
* * *
ment 2 to reach the same criterion, 10 or more correct choices in * + * p<0.05
No. of Correct Choice

㪈㪇 obst-B
two consecutive sessions. This additional session was unnecessary (/12 trs.)
for Heiji and Kiki, who immediately reached this criterion during 㪏 trap-A
the two test sessions in Experiment 2. trap-B

The hook was used in this confirmation phase and the following
test phase. Before the test phase, we presented one obstacle set or 㪋
one trap set to confirm the performance shown so far by each * : p<0.05 (/24 trs.)

monkey. The obstacle set was the 12 problems extracted from the +: p<0.10 (/24 trs.)
two sets, which were assigned odd numbers in Figures 1a and 1b; 㪇
1234 1234 1234 1234
the trap set was the 12 extracted problems, assigned odd numbers 㪟㪼㫀㫁㫀 㪱㫀㫃㫃㪸 㪢㫀㫂㫀 㪧㫀㪾㫄㫆㫅
Heiji Zilla Kiki Pigmon
in Figures 1c and 1d. Heiji, Zilla, and Kiki received the obstacle set
Subjects & Test Sessions
because they were given the obstacle condition in Experiments 1
and 2, and Pigmon received the trap set. In this phase, we relaxed Figure 2. The number of correct choices of 12 by each monkey in each
criterion to 18 or more correct choices in 24 trials during two session of the generalization test in Experiment 3. Heiji, Zilla, and Kiki
consecutive sessions, which was statistically significant. The mon- were tested for their generalization to the trap task and Pigmon was tested
keys proceeded to the test phase after reaching this criterion. for his generalization to the obstacle task.
TOOL USE BY CAPTIVE CAPUCHINS 15

others). Three monkeys tested in trap sets were correct in 15 1 2 3 4 5 6


(93.8%), 13 (81.3%), and 14 (87.5%) trials of 16 trials for the (a) obstacle set A
one-hindrance problems and in 23 (71.9%), 19 (59.4%), and 25
(78.1%) trials of 32 for the two-hindrance problems for Heiji,
Zilla, and Kiki, respectively. The last monkey (Pigmon) tested in (b) obstacle set B
obstacle sets was correct in 13 (81.3%) of 16 for one-hindrance
problems and 22 (68.8%) of 32 trials for two-hindrance problems.
(c) trap set A
Although the monkeys’ accuracies were in general better for
one-hindrance problems than two-hindrance problems, none of the
comparisons for each subject between one- and two-hindrance
(d) trap set B
problems reached statistical significance in Fisher’s exact tests
( ps ⫽ .079, .116, .358, and .288 for Heiji, Zilla, Kiki, and Pigmon,
respectively). Thus, the monkeys’ performances were comparable * *
(e) 㪈㪉 * * * * * *
regardless of the number of hindrances in each problem.
obst-A

No. of Correct Choice


One should note that the two kinds of hindrance, obstacle and 㪈㪇 p<0.05
(/12 trs.) obst-B
trap, are different in how to hinder the monkeys from collecting the 㪏
reward. An obstacle blocks the tool; a trap captures the reward into trap-A

itself. In other words, the same arrangements of food, tool, and trap-B
hindrance sometimes lead to different results. In particular, the 㪋

vertical location of food and hindrance is critical only for traps 㪉


*: p<0.05 (/48 trs.)
(e.g., Problems 3 and 3⬘ in Trap Set A), whereas the horizontal

location of food and hindrance inside the tool never matters (i.e., 12345678 12345678 12345678 12345678
㪟㪼㫀㫁㫀 㪱㫀㫃㫃㪸 㪢㫀㫂㫀 㪧㫀㪾㫄㫆㫅
Heiji Zilla Kiki Pigmon
always makes an incorrect option) for obstacles (e.g., Problems 2
Subjects & Test Sessions
and 2⬘ for Obstacle Set A).
To check whether and how the monkeys recognized the function
Figure 3. (a)–(d) Arrangement of food and tool in the generalization test
of the two types of hindrance, we compared performances between of Experiment 4: the obstacle condition Set A (a) and Set B (b), and the trap
the problems in which a different hindrance changes or obscures condition Set A (c) and Set B (d). (e) The number of correct choices of 12
the correct option as noted above (i.e., trial Types 3 and 3⬘ of the by each monkey in each session of the generalization test in Experiment 4.
Trap Set A and 6 and 6⬘ of the Trap Set B [see Figure 1c and d])
and those in which it does not (all others). Because the number of
trials was only eight for each monkey, the data were summed
across monkeys. The monkeys in total answered correctly in 16
Procedure
(66.7%) of 24 trials for the former and 130 (77.4%) of 168 for the Confirmation phase. To confirm the monkeys’ performances
latter. These two proportions were comparable ( p ⫽ .184, Fisher’s on the hindrance tasks, we trained the monkeys on all of the four sets
exact test). This shows that it is unlikely for the monkeys to shown in Figures 1a–1d until they scored 32 or more correct choices
confuse the two types of hindrance. It should be noted, however, on 48 trials, a statistically significant score at the 5% level, within four
that the evidence was not the strongest because 16 of 24 in these sessions in which all of the four sets were presented once.
most critical test trials was a little short of statistical significance We then examined the monkeys’ performances on the two-term
( p ⫽ .076, one-tailed), probably due to the small N. problems using the scoop. We used the two sets from Experiment
The fact that monkeys generalized their performances during 4 in Fujita et al. (2003). Each of the two sets was tested once.
training on one type of hindrance to the other suggests that they Because Trial Types 1 and 1⬘ of Set B (see Fujita et al., 2003) had
registered the necessity of taking a third environmental element no incorrect option, performance on those trials was omitted from
into account in order to collect the food inside the tool and, at least the scores, leaving a maximum of 20 correct choices.
to some degree, the different functions of different types of hin- Test phase. The three-term arrangements with the scoop
drances. However, it is still possible for the monkeys to have shown in Figures 3a–3d were presented. The two sets of the
learned some unidentified simple rules. Experiment 4 examined obstacle and trap conditions were each tested twice in sequence.
such possibilities using a different tool. Between the tests with different sets, we conducted at least one
session using the 12 arrangements extracted from those in the
confirmation phase of Experiment 3 using the hook.
Experiment 4: Generalization to a New Tool Here, we summarize the testing procedure: first, confirmation on all
of the hindrance task sets with the hook until reaching the criterion;
Experiment 4 further investigated the generality of what the second, two sessions of two-term problems with the scoop; third, at
monkeys learned in the preceding experiments. In this experiment, least one session on the extracted hindrance (obstacle or trap) set;
a new but familiar tool, the scoop, replaced the hook (see Figures fourth, two sessions of one of the hindrance (obstacle or trap) set
3a–3d). The monkeys had experienced this tool in the two-term shown in Figure 9 in Fujita et al. (2003). The third and the fourth
context with success (Fujita et al., 2003), but had never experi- phases were repeated four times, once for each hindrance set. The
enced it in the three-term context. order of testing for each monkey was as shown in Figure 3e.
16 FUJITA, SATO, AND KUROSHIMA

Results and Discussion hook. In other words, the monkeys had learned a rather general
rule that allowed them to perform correctly on three-term problems
Confirmation phase. All monkeys immediately completed involving food, tool, and hindrance, irrespective of the shape of the
the confirmation phase of the three-term problems with the hook in tool.
four sessions.
The performance in the two-term problems with the scoop
was also good for all monkeys; it was above chance for all General Discussion
monkeys: Heiji (19 correct of 20, binomial test: p ⬍ .001) and In the recovery phase, despite a break of almost 2.5 years, all
Pigmon (20 of 20, p ⬍ .001), and for Zilla and Kiki (16 of 20, four tufted capuchin monkeys maintained their good performance
p ⫽ .012 for each). The results suggest that the monkeys on the basic tool task. This result shows impressively persistent
maintained good performances on two-term problems with the long-term memory in this species of New World monkeys.
scoop. In Experiment 1, all monkeys learned to choose the correct
Test phase. Figure 3e shows each monkey’s number of cor- alternative in three-term problems incorporating food, tool, and
rect choices in each test session. The monkeys showed good hindrance. In subsequent experiments, the monkeys showed pos-
generalization to new problems, although not all monkeys reached itive transfer to novel types of spatial arrangements of the three
statistical significance in the very first session of each test. The terms (Experiment 2), to different types of hindrance (Experiment
maximum number was 48 for each condition (obstacle or trap). 3), and to a new familiar tool (Experiment 4). These results clearly
The performance in each condition for each monkey was above demonstrate that failure by the same capuchins to generalize their
chance (binomial tests: p ⫽ .006 for Zilla in the trap condition, p ⫽ performance on two-term tasks to three-term tasks in the previous
.002 for Kiki in the trap condition, and p ⬍ .001 elsewhere). To study (Fujita et al., 2003) was not due to a cognitive limitation but
estimate whether the ratio of correct choices varied across condi- to a task requirement in the original training. It may be noted that
tions (obstacle or trap), we applied Fisher’s exact tests to the data one monkey, Zilla, failed to learn the trap task at first, but after-
and failed to find a significant difference for any monkey ( p ⫽ ward she showed a positive transfer to the trap task after learning
.261 for Heiji, p ⫽ .137 for Zilla, p ⫽ .208 for Kiki, and p ⫽ .317 the obstacle task. Thus, the ability to process three elements at a
for Pigmon). time may be universal in this species.
We again looked into each configuration of problems as in Our results suggest that the capuchin monkeys learned a general
Experiment 3. For obstacle sets, monkeys in total were worst rule of how to select a correct option in this complex tool use task
for 2 and 2⬘ in Set A and 6 and 6⬘ in Set B: 10 correct of 16 involving three items, namely food, tool, and hindrance. This
trials. For trap sets, monkeys were worst for 5 and 5⬘ in Set A performance shows an excellent capacity to handle physical chal-
and for 4 and 4⬘ in Set B: 9 correct of 16 for the former and 10 lenges in this New World monkey species. The transfer of the
of 16 for the latter. For three of these problems, the distance monkeys’ performances to untrained types of hindrance in Exper-
between the reward and the functional hindrance was greater iment 3 warrants special attention in this regard; the good perfor-
than for most of the other problems. Thus, the monkeys might mance in their initial training on three-term problems was not
have had difficulty in avoiding options with such configura- limited to the hindrance the monkeys tackled. Thus, the rule that
tions. the monkeys abstracted seems to have been general enough to be
We again compared two types of seemingly easy problems in successfully applied to different types of hindrance, which may be
which only one option had a hindrance (1, 1⬘, 2, and 2⬘ in each spelled out as (a) no part of the tool is on the other side of the
set) and the remaining four types of more difficult problems obstacle, so that it would not block the tool; and (b) the food is
having a hindrance in each option (all others). For obstacle sets, neither in alignment with the trap or on the other side of it, so that
the monkeys were correct in 14 (87.5%), 11 (68.8%), 14 (87.5%), and the food would not come over the trap.
15 (93.8%) trials of 16 for the one-hindrance problems and in 29 Although our series of experiments may not completely reject a
(90.6%), 30 (93.8%), 27 (84.4%), and 30 (93.8%) of 32 for the possibility of a simple stimulus generalization, our demonstration
two-hindrance problems for Heiji, Zilla, Kiki, and Pigmon, respec- is consistent with the idea that tufted capuchin monkeys have a
tively. Fisher’s exact tests revealed that Zilla’s data were signifi- rudimentary understanding of the causality in tool use involving
cantly different between the two types ( p ⫽ .033), but her accu- the three terms, food, tool, and environment. These findings are
racy was in fact higher for supposedly more difficult two- consistent with the nut-cracking behavior reported by Fragaszy,
hindrance problems. No other monkeys’ data reached significance Izar, et al. (2004) and Moura and Lee (2004), which also involves
( ps ⫽ .546, .571, and .713 for Heiji, Kiki, and Pigmon, respec- three-term spatial relationships among a nut, a hammer, and a hard
tively). For trap sets, the monkeys were correct in 14 (87.5%), 12 substrate. Capuchins are in fact even selective in using hammers of
(75.0%), 14 (87.5%), and 14 (87.5%) trials of 16 for the one- various weights (Visalberghi, Addessi, et al., 2009), although such
hindrance problems and 24 (75.0%), 22 (68.8%), 22 (68.8%), and excellent performances still could be a result of individual trial-
27 (84.4%) trials of 32 for the two-hindrance problems for Heiji, and-error experience.
Zilla, Kiki, and Pigmon, respectively. None of these differences It is more difficult to reconcile the present findings with the
reached statistical significance: ps ⫽ .272, .462, .144, and .571 for failure by the same species in the trap tube task (Visalberghi &
Heiji, Zilla, Kiki, and Pigmon, respectively. Thus, the monkeys’ Limongelli, 1994), which also incorporates the three terms, food,
performances were mostly comparable regardless of the number of stick, and trap, in a manner more similar to our task. Here, we
hindrances in each problem. propose a few possible reasons why monkeys failed in the trap
The results suggest that the monkeys solved the three-term tasks tube task. One is that the trap tube apparatus may have looked
in Experiments 1 to 3 in a way not specific to the task with the more complicated than our two-choice situations to the monkeys
TOOL USE BY CAPTIVE CAPUCHINS 17

because the tube was supported by poles and the food was located Anderson, J. R. (2006). Tool use: A discussion of diversity. In K. Fujita &
inside the transparent tube, which made the view of the food S. Itakura (Eds.), Diversity of cognition: Evolution, development, do-
indirect. Another possibility is that the monkeys had to push the mestication, and pathology (pp. 367– 400). Kyoto, Japan: Kyoto Uni-
food away and out the other side, which may have required greater versity Press.
self-control. It should be noted that apes performed better when Anderson, J. R., & Henneman, M.-C. (1994). Solutions to a tool-use
problem in a pair of Cebus apella. Mammalia, 58, 351–361.
they raked in the reward in the tube than when they pushed it out
Anderson, J. R., Kuroshima, H., Hattori, Y., & Fujita, K. (2005). Attention
(Mulcahy & Call, 2006). Finally, in the two-choice task, the to combined attention in New World monkeys (Cebus apella, Saimiri
monkeys were able to compare the probability of success between sciureus). Journal of Comparative Psychology, 119, 461– 464.
two visible options, whereas in the trap tube task they could not. Anderson, J. R., Kuroshima, H., Kuwahata, H., & Fujita, K. (2004). Do
These factors might have concealed the capuchins’ fine sensitivity squirrel monkeys (Saimiri sciureus) and capuchin monkeys (Cebus
to spatial relationships. apella) predict that looking leads to touching? Animal Cognition, 7,
An interesting future study may ask whether capuchin monkeys 185–192.
are able to create arrangements that allow smooth collection of Anderson, J. R., Kuwahata, H., Kuroshima, H., Leighty, K. A., & Fujita, K.
food by the use of tools. The two-choice task used in this series of (2005). Are monkeys aesthetists? Rensch (1957) revisited. Journal of
experiment asked their recognition of successful and unsuccessful Experimental Psychology: Animal Behavior Processes, 31, 71–78.
arrangements. In situations that require creating arrangements, we Beck, B. B. (1980). Animal tool behavior: The use and manufacture of
tools by animals. New York: Garland STPM Press.
may be able to know their understanding of causality in much
Boesch, C., & Boesch-Achermann, H. (2000). The chimpanzees of Taı̈
more detail. Important issues may be whether they displace the
forest: Behavioural ecology and evolution. Oxford, England: Oxford
obstacles, whether they put a cap on the trap, or whether they place University Press.
a tool of a suitable shape and size in a good orientation at a good Chappell, J., & Kacelnik, A. (2002). Tool selectivity in a non-primate, the
location. New Caledonian crow (Corvus moneduloides). Animal Cognition, 5,
Another question may be whether capuchin monkeys predict the 71–78.
movement of the tool and the reward. Recent studies suggest that Chappell, J., & Kacelnik, A. (2004). Selection of tool diameter by New
humans, when they recognize the goal of the moving object, move Caledonian crows Corvus moneduloides. Animal Cognition, 7, 121–127.
their eye gaze to the goal in advance to its actual motion (Eshuis, Cleveland, A., Rocca, A. M., Wendt, E. L., & Westergaard, G. C. (2004).
Coventry, & Vulchanova, 2009). It is expected that the monkeys’ Transport of tools to food sites in tufted capuchin monkeys (Cebus
gaze may anticipate the current motion of the tool and the reward apella). Animal Cognition, 7, 193–198.
if they predict the result. This finding would provide stronger Cummins-Sebree, S. E., & Fragaszy, D. (2005a). Capuchins as stone-
knappers? An evaluation of the evidence. In V. Roux & B. Bril (Eds.),
evidence for understanding causality.
Stone knapping: The necessary conditions for a uniquely hominin be-
The two-choice method like the one conducted in this study has
haviour (pp. 171–182). Cambridge, England: McDonald Institute for
various advantages for analyzing nonhuman animals’ understand- Archaeological Research.
ing of causality. One is that this method enables systematic ma- Cummins-Sebree, S. E., & Fragaszy, D. M. (2005b). Choosing and using
nipulation of variables potentially contributing this aspect of cog- tools: Capuchins (Cebus apella) use a different metric than tamarins
nition. Our research exemplified this merit. In our case, various (Saguinus oedipus). Journal of Comparative Psychology, 119, 210 –219.
experimental modifications in numbers, types, and arrangements Emery, N. J., & Clayton, N. S. (2004, December 10). The mentality of
of tool and hindrance revealed excellent physical cognition by crows: Convergent evolution of intelligence in corvids and apes. Sci-
capuchin monkeys. Seed et al. (2006) pointed out that transfer ence, 306, 1903–1907.
tasks would form an important paradigm for investigating the Eshuis, R., Coventry, K. R., & Vulchanova, M. (2009). Predictive eye
nature of animal physical cognition, as opposed to tests for all-or- movements are driven by goals, not by the mirror neuron system.
none possession of intelligence such as causal understanding. The Psychological Science, 20, 438 – 440.
Fragaszy, D., Izar, P., Visalberghi, E., Ottoni, E. B., & Gomes de Oliveira,
two-choice task can be such a method.
M. (2004). Wild capuchin monkeys (Cebus libidinosus) use anvils and
The second advantage is that this method enables direct com-
stone pounding tools. American Journal of Primatology, 64, 359 –366.
parison of a wide range of animals because all they must do is to Fragaszy, D. M., Visalberghi, E., & Fedigan, L. M. (2004). The complete
pull one of the tools. This is an operation that even birds and fish capuchin: The biology of the genus Cebus. Cambridge, England: Cam-
can manage using their mouth. The behavior involved— bridge University Press.
pulling—is also naturalistic and this factor may make the task even Fujita, K., & Giersch, A. (2005). What perceptual rules do capuchin
easier to handle for many species. We should test not only primates monkeys (Cebus apella) follow in competing partly occluded figures?
but also birds such as corvids, which show excellent tool-using and Journal of Experimental Psychology: Animal Behavior Processes, 31,
tool-making skills sometimes almost comparable to those of apes 387–398.
(Chappell & Kacelnik, 2002, 2004; Kenward, Weir, Rutz, & Fujita, K., Kuroshima, H., & Asai, S. (2003). How do tufted capuchin
Kacelnik, 2005; Weir, Chappell, & Kacelnik, 2002; see also monkeys (Cebus apella) understand causality involved in tool use?
Journal of Experimental Psychology: Animal Behavior Processes, 29,
Emery & Clayton, 2004; Kacelnik et al., 2006). Such a compara-
233–242.
tive analysis should shed light on how this important cognitive
Fujita, K., Kuroshima, H., & Masuda, T. (2002). Do tufted capuchin
ability might have evolved. monkey (Cebus apella) spontaneously deceive opponents? A prelimi-
nary analysis of an experimental food-competition contest between
References monkeys. Animal Cognition, 5, 19 –25.
Goodall, J. (1986). The chimpanzees of Gombe: Patterns of behavior.
Anderson, J. R. (1990). Use of objects as hammers to open nuts by Cambridge, MA: Belknap Press of Harvard University Press.
capuchin monkeys (Cebus apella). Folia Primatologica, 54, 138 –145. Hattori, Y., Kuroshima, H., & Fujita, K. (2005). Cooperative problem
18 FUJITA, SATO, AND KUROSHIMA

solving by tufted capuchin monkeys (Cebus apella): Spontaneous divi- nutcrackers: What capuchin monkeys (Cebus apella) know about others’
sion of labor, communication, and reciprocal altruism. Journal of Com- tool-using skills. Animal Cognition, 8, 215–219.
parative Psychology, 119, 335–342. Ottoni, E. B., & Mannu, M. (2001). Semifree-ranging tufted capuchins
Hauser, M. D. (1997). Artifactual kinds and functional design features: (Cebus apella) spontaneously use tools to crack open nuts. International
What a primate understands without language. Cognition, 64, 285–308. Journal of Primatology, 22, 347–358.
Hauser, M. D., & Santos, L. R. (2007). The evolutionary ancestry of our Ottoni, E. B., & Mannu, M. (2003). Spontaneous use of tools by semifree-
knowledge of tools: From percepts to concepts. In E. Margolis, & S. ranging capuchin monkeys. In F. B. M. de Waal & P. L. Tyack (Eds.),
Laurence (Eds.), Creations of the mind: Theories of artifacts and their Animal social complexity: Intelligence, culture, and individualized so-
representation. (pp. 267–288). Oxford: Oxford University Press. cieties (pp. 440 – 443). Cambridge, MA: Harvard University Press.
Izawa, K., & Mizuno, A. (1977). Palm-fruit cracking behavior of wild Paukner, A., Anderson, J. R., & Fujita, K. (2004). Reactions of capuchin
black-capped capuchin (Cebus apella). Primates, 18, 773–792. monkeys (Cebus apella) to multiple mirrors. Behavioural Processes, 66,
Jalles-Filho, E., da Cunha, R. G. T., & Salm, R. A. (2001). Transport of 1– 6.
tools and mental representation: Is capuchin monkey tool behaviour a Pouydebat, E., Gorce, P., Bels, V., & Coppens, Y. (2006). Substrate
useful model of Plio-Pleistocene hominid technology? Journal of Hu- optimisation in nuts cracking by capuchin monkeys. American Journal
man Evolution, 40, 365–377. of Primatology, 68, 1017–1024.
Kacelnik, A., Chappell, J., Kenward, B., & Weir, A. A. S. (2006). Cog- Povinelli, D. J. (with Reaux, J. E., Theall, L. A., & Giambrone, S.). (2000).
nitive adaptations for tool-related behaviour in New Caledonian crows. Folk physics for apes: The chimpanzee’s theory of how the world works.
In E. A. Wasserman & T. R. Zentall (Eds.), Comparative cognition: Oxford, England: Oxford University Press.
Experimental explorations of animal intelligence (pp. 515–528). New Primate Research Institute, Kyoto University. (2002). Saru-rui no shiiku
York: Oxford University Press. kanri oyobi shiyo៮ ni kansuru shishin [Guide for the care and use of
Kenward, B., Weir, A. A. S., Rutz, C., & Kacelnik, A. (2005, January 13). laboratory primates] (2nd ed.). Inuyama, Aichi, Japan: Author.
Tool manufacture by naive juvenile crows. Nature, 433, 121. Santos, L. R., Pearson, H. M., Spaepen, G. M., Tsao, F., & Hauser, M. D.
Köhler, W. (1921). Intelligenzprüfungen an Menschenaffen [Intelligence (2006). Probing the limits of tool competence: Experiments with two
tests on apes]. Berlin: Julius Springer. non–tool-using species (Cercopithecus aethiops and Saguinus oedipus).
Klüver, H. (1933). Behavior mechanisms in monkeys, Chicago, IL: Uni- Animal Cognition, 9, 94 –109.
versity of Chicago Press. Santos, L. R., Rosati, A., Sproul, C., Spaulding, B., & Hauser, M. D.
Kuroshima, H., Fujita, K., Adachi, I., Iwata, K., & Fuyuki, A. (2003). A (2005). Means-means-end tool choice in cotton-top tamarins (Saguinus
capuchin monkey (Cebus apella) recognizes when people do and do not oedipus): Finding the limits on primates’ knowledge of tools. Animal
know location of food. Animal Cognition, 6, 283–291. Cognition, 8, 236 –246.
Kuroshima, H., Fujita, K., Fuyuki, A., & Masuda, T. (2002). Understand- Seed, A. M., Tebbich, S., Emery, N. J., & Clayton, N. S. (2006). Investi-
ing of the relationship between seeing and knowing by tufted capuchin gating physical cognition in rooks, Corvus frugilegus. Current Biology,
monkeys (Cebus apella). Animal Cognition, 5, 41– 48. 16, 697–701.
Lavallee, A. C. (1997). Capuchin (Cebus apella) tool use in a captive Shepherd, W. T. (1910). Some mental processes of the rhesus monkey.
naturalistic environment. International Journal of Primatology, 20, Psychological Monographs, 12, 1– 66.
399 – 414. Silva, F. J., Page, D. M., & Silva, K. M. (2005). Methodological–
Limongelli, L., Boysen, S. T., & Visalberghi, E. (1995). Comprehension of conceptual problems on the study of chimpanzees’ folk physics: How
cause-effect relations in a tool-using task by chimpanzees (Pan troglo- studies with adult humans can help. Learning & Behavior, 33, 47–58.
dytes). Journal of Comparative Psychology, 109, 18 –26. Sugiyama, Y., & Koman, J. (1979). Tool using and making behavior in
Malaivijitnond, S., Lekprayoon, C., Tandavanittj, N., Panha, S., Chee- wild chimpanzees at Bossou, Guinea. Primates, 20, 513–524.
watham, C., & Hamada, Y. (2007). Stone-tool usage by Thai long-tailed Tomasello, M., & Call, J. (1997). Primate cognition. Oxford, England:
macaques (Macaca fascicularis). American Journal of Primatology, 69, Oxford University Press.
227–233. Visalberghi, E. (1990). Tool use in Cebus. Folia Primatologica, 54, 146 –
Matsuzawa, T. (2001). Primate foundations of human intelligence: A view 154.
of tool use in nonhuman primates and fossil hominoids. In T. Matsuzawa Visalberghi, E., Addessi, E., Spagnoletti, N., Truppa, V., Ottoni, E., Izar,
(Ed.), Primate origins of human cognition and behavior (pp. 3–25). P., & Fragaszy, D. (2009). Selection of effective stone tools by wild
Tokyo: Springer. capuchin monkeys. Current Biology, 19, 213–217.
McGrew, W. C. (1992). Chimpanzee material culture: Implications for Visalberghi, E., & Fragaszy, D. (2002). Tool use by monkeys and apes. In
human evolution. Cambridge, England: Cambridge University Press. D. Baltimore, R. Dulbecco, F. Jacob, & R. Levi-Montalcini (Series
McGrew, W. C. (1994). The cultured chimpanzee: Reflections on cultural Eds.), Frontiers of life: Vol. IV. The living world (pp. 221–230). San
primatology. Cambridge, England: Cambridge University Press. Diego, CA: Academic Press.
Moura, A. C., de A., & Lee, P. C. (2004, December 10). Capuchin stone Visalberghi, E., & Fragaszy, D. (2006). What is challenging about tool
tool use in Caatinga dry forest. Science, 306, 1909. use? The capuchin’s perspective. In E. A. Wasserman & T. R. Zentall
Mulcahy, N. J., & Call, J. (2006). How great apes perform on a modified (Eds.), Comparative cognition: Experimental explorations of animal
trap-tube task. Animal Cognition, 9, 193–199. intelligence (pp. 529 –552). New York: Oxford University Press.
Nellmann, H., & Trendelenburg, W. (1926). Ein Beitrag zur Intelligenz- Visalberghi, E., Fragaszy, D. M., & Savage-Rumbaugh, S. (1995). Perfor-
prüfung niederer Affen [A contribution to the intelligence testing of mance in a tool-using task by common chimpanzees (Pan troglodytes),
lower monkeys]. Journal of Comparative Physiology, 4, 142–200. bonobos (Pan paniscus), an orangutan (Pongo pygmaeus), and capuchin
Nishida, T. (1973). The ant-gathering behaviour by the use of tools among monkeys (Cebus apella). Journal of Comparative Psychology, 109,
wild chimpanzees of the Mahali Mountains. Journal of Human Evolu- 52– 60.
tion, 2, 357–370. Visalberghi, E., & Limongelli, L. (1994). Lack of comprehension of
Ohashi, G. (2006). Behavioral repertoire of tool use in the wild chimpan- cause-effect relations in tool-using capuchin monkeys (Cebus apella).
zees at Bossou. In T. Matsuzawa, M. Tomonaga, & M. Tanaka (Eds.), Journal of Comparative Psychology, 108, 15–22.
Cognitive development in chimpanzees (pp. 439 – 451). Tokyo: Springer. Visalberghi, E., & Limongelli, L. (1996). Acting and understanding: Tool
Ottoni, E. B., Dogo de Resende, B., & Izar, P. (2005). Watching the best use revisited through the minds of capuchin monkeys. In A. E. Russon,
TOOL USE BY CAPTIVE CAPUCHINS 19

K. A. Bard, & S. T. Parker (Eds.), Reaching into thought: The minds of Westergaard, G. C., & Suomi, S. J. (1993). Use of a tool-set by capuchin
the great apes (pp. 57–79). Cambridge, England: Cambridge University monkeys. Primates, 34, 459 – 462.
Press. Westergaard, G. C., & Suomi, S. J. (1997). Capuchin monkey (Cebus
Visalberghi, E., Spagnoletti, N., Ramos da Silva, E. D., Andrade, F. R. D., apella) grips for the use of stone tools. American Journal of Primatol-
Ottoni, E., Izar, P., & Fragaszy, D. (2009). Distribution of potential ogy, 103, 131–135.
suitable hammers and transport of hammer tools and nuts by wild Whiten, A., Goodall, J., McGrew, W. C., Nishida, T., Reynolds, V.,
capuchin monkeys. Primates, 50, 95–104. Sugiyama, Y., . . . Boesch, C. (1999, June 17). Cultures in chimpanzees.
Visalberghi, E., & Tomasello, M. (1998). Primate causal understanding in Nature, 399, 682– 685.
the physical and psychological domains. Behavioural Processes, 42, Whiten, A., Goodall, J., McGrew, W. C., Nishida, T., Reynolds, V.,
189 –203. Sugiyama, Y., . . . Boesch, C. (2001). Charting cultural variation in
Visalberghi, E., & Trinca, L. (1989). Tool use in capuchin monkeys: chimpanzees. Behaviour, 138, 1481–1516.
Distinguishing between performing and understanding. Primates, 30, Yamakoshi, G. (2001). Ecology of tool use in wild chimpanzees: Toward
511–521. reconstruction of early hominoid evolution. In T. Matsuzawa (Ed.),
Waga, I., Dacier, A., Pinha, P. S., & Tavares, M. C. H. (2006). Spontane- Primate origins of human cognition and behavior (pp. 537–556). Tokyo:
ous tool use by wild capuchin monkeys (Cebus libidinosus) in the Springer.
Cerrado. Folia Primatologica, 77, 337–344. Yamakoshi, G. (2004). Evolution of complex feeding techniques in pri-
Weir, A. A. S., Chappell, J., & Kacelnik, A. (2002, August 9). Shaping of mates: Is this the origin of great ape intelligence? In A. E. Russon &
hooks in New Caledonian crows. Science, 297, 981. D. R. Begun (Eds.), The evolution of thought: Evolutionary origins of
Westergaard, G. C., Lundquist, A. L., Haynie, M. K., Kuhn, H. E., & great ape intelligence (pp. 140 –171). Cambridge, England: Cambridge
Suomi, S. J. (1998). Why some capuchin monkeys (Cebus apella) use University Press.
probing tools (and others do not). Journal of Comparative Psychology,
112, 207–211.
Westergaard, G. C., Lundquist, A. L., Kuhn, H. E., & Suomi, S. J. (1997). Received January 6, 2010
Ant-gathering with tools by captive tufted capuchins. International Revision received May 12, 2010
Journal of Primatology, 18, 95–103. Accepted May 12, 2010 䡲

Showcase your work in APA’s newest database.

Make your tests available to other researchers and students; get wider recognition for your work.
“PsycTESTS is going to be an outstanding resource for psychology,” said Ronald F. Levant,
PhD. “I was among the first to provide some of my tests and was happy to do so. They will be
available for others to use—and will relieve me of the administrative tasks of providing them to
individuals.”
Visit http://www.apa.org/pubs/databases/psyctests/call-for-tests.aspx
to learn more about PsycTESTS and how you can participate.
Questions? Call 1-800-374-2722 or write to tests@apa.org.
Not since PsycARTICLES has a database been so eagerly anticipated!

You might also like