Current Biology Vol 17 No 23

R1004

Working memory numeral, all other numerals

were replaced by white squares.
and Video clip 2). Humans were
slower than all of the three young
of numerals in The subject had to remember
which numeral appeared in
chimpanzees in the response
(Figure S1 in the Supplemental
chimpanzees which location, and then touch data; Video clips 3 and 4).
them based on the knowledge We developed a new test called
of numerical sequence. All five the ‘limited-hold memory task’
Sana Inoue and naïve chimpanzees mastered as a novel way of comparing the
Tetsuro Matsuzawa the masking task, just like Ai working memory of chimpanzee
[7]. It must be noted that the and human subjects. In this task,
Chimpanzee memory has been chance level of this task is very after the touch to the initial white
extensively studied [1,2]. The low: p= 1/24 with four numerals, circle, the numerals appeared
general assumption is that, 1/120 with five numerals, and only for a certain limited duration,
as with many other cognitive so on, down to 1/362,880 with and were then automatically
functions, it is inferior to that nine numerals. In general, the replaced by white squares.
of humans [3]; some data, performance of the three young Three different hold duration
however, suggest that, in some chimpanzees was better than that conditions were tested: 650,
circumstances, chimpanzee of the three mothers (see Table S1 430 and 210 milliseconds. The
memory may indeed be superior in the Supplemental data). Ayumu, duration of 650 milliseconds was
to human memory [4]. Here we Ai’s son, was the best performer equivalent to the average initial
report that young chimpanzees among the subjects (Figure 1 latency of five-numeral masking
have an extraordinary working
memory capability for numerical
recollection — better even than Figure 1. Chimpanzee Ayumu
that of human adults tested in performing the masking
task.
the same apparatus following the (A) Ayumu touches the first
same procedure. numeral of the sequence;
Our subjects were six (B) immediately thereafter
chimpanzees, three mother– the remaining numerals are
offspring pairs. One of the replaced by white squares.
mothers, Ai, was the first (C) Ayumu remembered which
number appeared in which
chimpanzee who learned to locations on the screen.
use Arabic numerals to label
sets of real-life objects with the
corresponding number [5]. The
other five chimpanzees had also
participated in many previous
studies [6], but they were naïve to
tasks employing numerals.
In 2004, when the three young
reached the age of four years,
we began to teach the mother–
offspring pairs the sequence
of Arabic numerals from 1 to 9,
using a touch-screen monitor
connected to a computer. In the
numerical sequence task, each
trial was unique, in which the nine
numerals appeared in different
on-screen positions. Accurate
performance with 1–2–3–4–5–6–
7–8–9 spontaneously transferred
to non-adjacent sequences
such as 2–3–5–8–9. All naïve
chimpanzees successfully learned
this numerical sequence task (See
Video clip 1 in the Supplemental
data available on-line with this
issue).
A ‘masking task’ to test
memory was introduced at
around the time when the young
became five years old. In this
task, after touching the first
Magazine
R1005

100
Figure 2. Results of the performance than adults in
limited-hold memory task. the memory task. Our study
The graph shows how Ai,
shows that young chimpanzees
Ayumu and human subjects
80 (n = 9, the bars represent have an extraordinary working
the SD) performed in the memory capability for numerical
Percentage correct (%)

limited-hold memory task. recollection better than that

60 The x-axis shows the three of human adults. The results fit
different limited-hold dura- well with what we know
tions tested; percentage of
about the eidetic imagery in
trials correctly completed
40
under each condition is humans.
Ayumu shown on the y-axis. Each
session consists of 50 trials.
20 Human Supplemental data
Ai
Each chimpanzee received
10 sessions and each of 9 Supplemental data, with video clips
humans received a single of the numerical and memory tasks of
0 test session. A two-way humans and chimpanzees, are available
650 430 210 ANOVA revealed that both at http://www.current-biology.com/cgi/
Limited-hold duration (ms) main effects were significant content/full/17/23/R1004/DC1
Current Biology (subjects: F2,29 = 29.50, p <
0.001, hold duration length: Acknowledgments
F2,29 = 121.45, p < 0.001), as was the interaction between them (F4,58 = 20.10, p < 0.001).
This study was supported by grants
Post-hoc tests revealed that Ayumu’s performance did not change as a function of hold
duration (F2,58 = 2.07, p = 0.136), whereas Ai and the human subjects’ performance de- MEXT16002001, JSPS-GCOE (A06,
creased with shorter duration lengths (F2,58 = 58.12, p < 0.001, F2,58 = 101.45, p < 0.001, Biodiversity), and JSPS-HOPE to
respectively). Pair-wise multiple comparisons by Ryan’s method showed significant dif- TM. This is a part of PhD thesis of SI.
ferences in performance between Ayumu and human subjects at the 430 milliseconds Thanks are due to Sumiharu Nagumo
and 210 milliseconds hold durations (p < 0.001, respectively). for his help with computer program-
ming, Masaki Tomonaga, Masayuki
Tanaka, Misato Hayashi, Etsuko Nogami
task. The shortest duration, the human subjects’ average, and Tomoko Takashima for their sup-
210 milliseconds, is close to and showed the same tendency port, and Dora Biro for her comments
the frequency of occurrence of to worsen with shorter duration. and revisions to the manuscript. Thanks
human saccadic eye movement From the very first session, are also due to the veterinary staffs and
caregivers of the KUPRI chimpanzees.
[8]. This means that this condition however, Ayumu’s performance
does not leave subjects enough remained at almost the same
References
time to explore the screen by level, regardless of the hold
1. Menzel, E.W. (1973). Chimpanzee spatial
eye movement. The limited-hold duration, showing no decline memory organization. Science 182,
memory task provided a means comparable to that of the other 943–945.
2. Fujita, K., and Matsuzawa, T. (1990).
of performing an objective subjects (Video clip 6, and Delayed figure reconstruction by a
comparison between the two Figure S2 in the Supplemental chimpanzee (Pan troglodytes) and
species under exactly identical data). These data showed that humans (Homo sapiens). J. Comp.
Psychol. 104, 345–351.
conditions. We compared Ai, the the chimpanzee subjects can 3. Premack, D. (2007). Human and animal
best mother performer, Ayumu, memorize at a glance the Arabic cognition: Continuity and discontinuity.
Proc. Natl. Acad. Sci. USA 104,
the best young performer, numerals scattered on the touch 13861–13867.
and human subjects (n = 9, all screen monitor and Ayumu 4. Tinklepaugh, O.L. (1932). The multiple
university students) in this task. outperformed all of the human delayed reaction with chimpanzees and
monkeys. J. Comp. Psychol. 13, 207–243.
Figure 2 shows the results of subjects both in speed and 5. Matsuzawa, T. (1985). Use of numbers by
the comparison between two accuracy. a chimpanzee. Nature 315, 57–59.
6. Matsuzawa, T., Tomonaga, M.,
chimpanzees and human subjects Our results may be reminiscent and Tanaka, M. (2006). Cognitive
in the limited-hold memory of the phenomenon known Development in Chimpanzees
task. The number of numerals as ‘eidetic imagery’ found by (Tokyo: Springer).
7. Kawai, N., and Matsuzawa, T. (2000).
was limited to five items. For Jaensch [9,10]. Eidetic imagery Numeric memory span in a chimpanzee.
example, the numerals 2, 3, 5, 8, has been defined as the memory Nature 403, 39–40.
8. Bartz, A.E. (1962). Eye-movement
9 might appear very briefly on the capability to retain an accurate, latency, duration, and response time as a
screen and then be replaced by detailed image of a complex function of angular displacement. J. Exp.
white squares. Subjects had to scene or pattern. It is known to Psychol. 64, 318–324.
9. Jaensch, E.R., trans. by Oscar Oeser
touch the squares in the correct be present in a relatively high (1930). Eidetic Imagery and Typological
order indicated by the original percentage of normal children, Methods of Investigation, Second Edition
(New York: Harcourt, Brace and Co).
numerals. and then the ability declines with 10. Conway, A., Jarrold, C., Kane, M.,
In human subjects, the age. Our present study shows Miyake, A., Towse, J., eds. (2007).
percentage of correct trials that young chimpanzees can Variation in Working Memory (New York:
Oxford University Press).
decreased as a function of the quickly grasp many numerals
hold duration: the shorter the at a glance, with no decline
Primate Research Institute, Kyoto
duration became, the worse the in performance as the hold University, Inuyama, Aichi 484-8506,
accuracy was (Video clip 5). Ai’s duration is varied. Moreover, Japan.
performance was below that of the young ones showed better E-mail: matsuzaw@pri.kyoto-u.ac.jp