Content uploaded by Christian Agrillo
Author content
All content in this area was uploaded by Christian Agrillo
Content may be subject to copyright.
SHORT COMMUNICATION
Quantity discrimination in felines: a preliminary investigation
of the domestic cat (Felis silvestris catus)
Paola Etel Pisa ÆChristian Agrillo
Received: 21 July 2008 / Accepted: 8 September 2008 / Published online: 2 October 2008
ÓJapan Ethological Society and Springer 2008
Abstract A large body of studies has investigated the
capacity of non-human primates, dogs, birds and lower
vertebrates to estimate different quantities of objects or
events. Little attention, however, has been devoted to
felines, and no study has specifically concentrated on cats’
numerical cognition. The present study aims to investigate
the capacity of domestic cats to distinguish between two
and three dots in order to obtain food; results showed that
cats can be trained to discriminate between the two quan-
tities. Furthermore our research suggests that cats do not
spontaneously use numerical information, but rather seem
to make use of visual cues that co-vary with numerosity in
order to solve the task.
Keywords Cat Numerical competence
Quantity discrimination Counting Animal cognition
Introduction
Numerical skills appear to be fairly widespread among
species: a large number of experiments conducted in the
laboratory and in the field have provided compelling evi-
dence that numerical abilities are not uniquely human
(Hauser et al. 2000). Lyon (2003), for instance, reported a
spontaneous use of numerical information in a natural
context as a strategy to reduce the costs of conspecific
brood parasitism in American coots. Previously, Wilson
et al. (2001) demonstrated how, in wild chimpanzees, the
decision to enter an intergroup contest depends on
favourable numerical asymmetries between the groups
rather than range location or other factors known to affect
response in other territorial species.
The ability to count seems to involve complex cognitive
skills, as outlined by Gelman and Gallistel (1978).
According to the authors, a robust definition of counting
should include five different principles: ‘one-to-one cor-
respondence’ (each component of a counted set must
correspond to one single numeron), ‘stable order’ (nume-
rons must be ordered in a sequence that is reproducible
every time), ‘cardinality’ (the last numeron in a sequence
also represents the total numerosity of the set), ‘abstrac-
tion’ (counting applies to homogeneous and heterogeneous
groups of objects of both physical and mental construction)
and lastly ‘order irrelevance’ (the number in which the
numerons correspond to each item is not important in the
counting process).
If, however, counting can be considered the highest
level of mathematical process, the simplest form of
numerical knowledge may be represented by a discrimi-
nation between two quantities, usually called ‘judgement of
relative numerosity’ (Anderson et al. 2005) or ‘quantity
discrimination’ (Agrillo et al. 2007; Stevens et al. 2007).
The ability to distinguish among different quantities may
have evolved to enhance survival of organisms in different
ecological contexts, such as foraging, group conflicts,
parental care and predator avoidance (Lyon 2003;
McComb et al. 1994; Wilson et al. 2001). It is possible that
quantity information may be more relevant in nature: for
example, in foraging situations animals often attempt to
maximise the amount of food acquired per unit time spent
foraging (Stephens and Krebs 1986). According to this,
P. E. Pisa
Psychology Department, Goethe-Universita
¨t,
Frankfurt am Main, Germany
C. Agrillo (&)
Department of General Psychology, University of Padova,
via Venezia 8, 35131 Padova, Italy
e-mail: christian.agrillo@unipd.it
123
J Ethol (2009) 27:289–293
DOI 10.1007/s10164-008-0121-0
even though number often predicts total amount, some-
times this is not the case, in particular when the size of food
items differs greatly (e.g. four very small grapes can be less
advantageous than three larger ones for a monkey). Ani-
mals may therefore use non-numerical quantitative
variables such as surface area as the basis for discrimina-
tion, especially when the goal is to maximise amount (and
not number), and it has been well demonstrated that the
overall area of the stimulus is one of the main non-
numerical cues used in quantity discrimination tasks when
visual stimuli are presented (Agrillo et al. 2008; Davis and
Perusse 1988; Feigenson et al. 2002).
To date, little attention has been focused on quantity
discrimination in felines. McComb et al. (1994) reported a
rare example of the use of quantity information in nature.
In playback experiments, recordings of single female lions
roaring and groups of three females roaring together were
played back, in order to simulate the presence of unfamiliar
intruders in the Serengeti National Park. Results showed
that defending adult females were less likely to approach
playbacks of three intruders than a single intruder; fur-
thermore, when the subjects approached three intruders,
they did so more cautiously. However, the exact nature of
such a skill is still unclear: since several features of the
stimuli tend to co-vary with numerosity, control experi-
ments are necessary to confirm whether felines can count
or, conversely, use other visual or auditory information to
compare two quantities. It is worth noting that, surpris-
ingly, we must record a lack of further studies in felines,
with the exception of a subsequent study on African lions
(Heinsohn 1997) in which, using a playback technique that
followed that of McComb et al. (1994), it was confirmed
that lionesses can distinguish among different quantities of
conspecifics roaring.
The present study aims to fill this gap. Four domestic
cats were initially trained to discriminate between groups
composed of two and three dots. Then, subjects were
observed in a control test, where the overall areas of the
stimuli were exactly matched within each pair of stimuli
presented, in order to see whether cats have previously
learned the task by counting the elements or, in contrast, by
comparing the amount of area (or other non-numerical cues
that co-vary with area and numerosity) between the two
groups.
Materials and methods
Subjects and stimuli
Four domestic pet cats (Felis silvestris catus) were used for
this experiment. Subjects were mature females (Nerina,
Wieso, Wilde and Suesse) between the ages of 4 and
5 years (mean age: 4.25). They were fed with meat and dry
cat food before and after the experiment. During the
training and test phases, however, no food was provided
outside the experimental session. The whole experiment
was set up inside a comfortable room in the private house
of the first author and videorecorded by a camera placed
behind the subjects and opposite the experimental setting.
Two green plastic bowls (11 cm diameter) were placed
40 cm apart, adjacent to a white wall. In the middle, an
opaque barrier (50 950 cm) divided the two bowls in
order to force the cats to choose between the two alterna-
tives soon after being released by the experimenter, thereby
reducing the potential influence of olfactory cues on the
cats’ choices. The cats were released 145 cm from the wall
where the two bowls were presented (Fig. 1). A sheet of A4
paper (21 929.7 cm) on which the stimuli were presented
was placed 10 cm above each bowl. The stimuli consisted
of a group of three black dots or a group of two black dots;
during the training phase all dots were the same size
(diameter: 3.07 cm). A total of 30 different pairs of stimuli
were used during the training; the position of the elements
was changed to avoid the cats learning how to distinguish
on the basis of the overall configuration of the stimuli
rather than on the quantity/numerosity of the sets.
In the test phase, the stimuli were figures that differed in
numerosity (three elements and two) but that were matched
for the overall area, i.e. elements in the two-dot groups
were enlarged and/or the elements in the three-dot groups
were reduced. Thirty different pairs (with different object
Fig. 1 Experimental setting used. Two bowls were placed the same
distance from the cat, and the bowl chosen by the subject was
recorded after each trial
290 J Ethol (2009) 27:289–293
123
sizes and positions) were used in the test phase. Also
during the test phase, the position of the elements was
varied.
Procedure
During the training phase, cats were individually observed
in a binary choice between two bowls: only the one asso-
ciated with the reinforced numerosity presented food
(commercial wet cat-food as normally eaten by the subjects
before the experiment). The empty bowl was scented at the
beginning and in the middle of each session with food to
prevent olfactory cues being used to find the reward. Fur-
thermore, a small amount of food was provided them
during each trial in order to reduce any olfactory cue and to
motivate cats to reach the bowls for the whole session. Two
cats (Suesse and Wieso) were reinforced toward the
smaller quantity (2), whereas the two other subjects
(Nerina and Wilde) were reinforced toward the larger
quantity (3).
Two 10-trial sessions were performed daily, one in the
morning and the second one in the afternoon, for 5 days,
and for a total of 100 trials. The location of food (and hence
the position of the two numerosities) was swapped in half
of the tests during each session: half of the trials presented
the reinforced numerosity on the left and the second half
presented it on the right. The position of the reinforced
quantity was randomly distributed within each session.
When cats found the reinforced bowl, they were allowed to
feed for a maximum of 10 s; a total of 10 s was also
allowed for the time between when they selected the empty
bowl and started the new trial. Only one experimenter was
inside the testing room during each trial. The experimenter
released the cat and stayed 80 cm behind the starting point
of the test. After the cat selected one bowl, the experi-
menter returned her to the starting point, while the other
experimenter entered the room and changed the stimuli.
However, the subject could not see any manipulation of the
stimuli since an opaque barrier was placed in front of her
by the experimenter who was near the cat.
In the test phase, we adopted an extinction procedure
used in other species with cognitive tasks that require a
training phase (Chiandetti and Vallortigara 2008; Sovrano
et al. 2007): neither bowl held food and no further rein-
forcement was therefore provided. Such a procedure was
similar to the previous training, except that a new set of
stimuli (with paired areas) was used. A total of 60 trials had
been planned (six overall sessions, two per day, ten trials
per session). However, since the cats’ motivation to select a
bowl in the absence of reinforcement was likely to decrease
with an increasing number of trials, a criterion was agreed
upon whereby the trials in which subjects spent longer than
5 min reaching any bowl should be discarded. The number
of trials that fell within this category differed for each cat
(17 Nerina, 4 Wilde, 24 Wieso, and 27 Suesse).
The cats’ ability to discriminate between the two
quantities was initially analysed by v
2
tests on the
frequency of choices between the groups; subsequently,
one-sample ttests on the proportion of accuracy were
performed to see whether cats were globally able to solve
both training and test phases. Statistical tests were carried
out with SPSS 15.0.
Results
Training: can cats learn how to distinguish between two
quantities?
Subjects easily learned to associate food and stimuli
(Fig. 2), preferentially selecting the bowl placed below the
reinforced quantity (Nerina, v
(1)
=17.640, P\0.001;
Wilde, v
(1)
=33.640, P\0.001; Wieso, v
(1)
=27.040,
P\0.001; Suesse, v
(1)
=19.160, P\0.001). Thereby,
we observed an overall ability to distinguish between the
two quantities when the proportion of correct choices was
analysed (mean ±SD: 0.758 ±0.034; one sample ttest,
t
(3)
=15.132, P\0.001). A subsequent analysis showed
that the overall proportion of correct choices in the first half
of the training (1–5 sessions) statistically differed from the
proportion of correct choices in the following trials (6–10
sessions; paired ttest, t
(3)
=-6.794, P=0.007).
Test: do cats use non-numerical cues to solve the task?
Two cats successfully discriminated the reinforced quantity
when the overall area between the two stimulus sets was
equated (Nerina, v
(1)
=10.256, P\0.001; Suesse,
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
1
Nerina
Wilde
Wieso
Suesse
Training Sessions
Proportion of choices toward the reinforced quantity
2345678910
Fig. 2 Proportion of correct choices of the four subjects during the
10 training sessions
J Ethol (2009) 27:289–293 291
123
v
(1)
=9.308, P=0.002) whereas no significant choice
was found for the other two subjects (Wilde, v
(1)
=0.286,
P=0.593; Wieso, v
(1)
=1.778, P=0.182, Table 1). An
overall analysis indicated that the cats did not correctly
choose the reinforced quantity when the area was con-
trolled for (0.633 ±0.126; t
(3)
=2.111, P=0.125).
A subsequent analysis on the proportion of correct
choices of the first session (10 trials) revealed that subjects’
performance at the beginning of the extinction procedure
(Nerina, 71% correct choices, v
(1)
=1.286, P=0.257;
Wilde, 50% correct choices, v
(1)
=0.000, P=1.000;
Suesse, 70% correct choices, v
(1)
=1.600, P=0.206;
Wieso, 60% correct choices, v
(1)
=0.400, P=0.527) was
positively correlated with the proportion of correct choices
of the following trials (r=0.994, P=0.006).
Discussion
The present work is intended to be a preliminary study of
cats’ numerical competence. We demonstrated that cats
can easily discriminate between different quantities of dots.
The ability to select the largest quantity may be used in
different ecological contexts, and feeding behaviour, in
which animals try to maximise the amount of food, rep-
resents one of the most important situations. To date, apart
from the McComb et al. study (1994) and a subsequent
investigation on African lions (Heinsohn 1997), this work
represents the only experimental evidence on rudimentary
quantity discrimination abilities in felines.
Results of the training phase clearly showed that cats
can learn how to distinguish between two groups of ele-
ments differing in numerosity. The fact that cats’
performance was more accurate in the second half of the
training phase demonstrates how quantity discrimination in
this task was learned over the trials and was not sponta-
neously achieved by the subjects, as has instead been found
in other species (Hauser et al. 2000; Uller et al. 2003).
On the other hand, results of the test phase provided
evidence that subjects did not strictly use numerical
information but rather seemed to analyse the quantity of
area of the dots, as has also been observed in non-human
primates (Tomonaga 2008). The fact that cats’ performance
during the first trials of the test was very similar to what
was exhibited in the following trials strongly supports the
idea that the reduced motivation resulting from the
extinction procedure cannot be the basis for the minor
accuracy of subjects’ response when the area was con-
trolled for.
It is not possible at present to exclude the possibility that
cats can use other non-numerical variables (such as the sum
of their contours and the density of the elements) that
cannot be fully matched when we control for the area.
Regardless of the exact perceptual cue involved, non-
numerical information seems to be spontaneously preferred
to numbers in a quantity discrimination task, as previously
observed in other vertebrates such as human infants
(Feigenson et al. 2002), apes (Beran et al. 2008), monkeys
(Stevens et al. 2007) and fish (Agrillo et al. 2008).
The trend exhibited by the subjects during the training
demonstrates that our procedure may be successfully
used in further studies on cats’ cognition, particularly to
extend research into felines’ numerical competence. For
instance, the next step of the project will involve a
training procedure where the range of the possible non-
numerical cues (such as area, contour, brightness and
density of the elements) will be fully investigated. Once
we understand the exact mechanism employed by cats to
distinguish between two quantities, continuous variables
will be matched from the start of the initial training
phase. Further research employing the methodology
presented here will therefore address whether or not cats
possess the ability to discriminate between two quantities
by counting each element.
However, we have now provided clear evidence on
quantity discrimination in felines. The literature on the
capacity for discriminating among sets containing different
numbers of objects, previously reported in human babies,
several non-human mammals, birds and fish is therefore
extended to include the domestic cat as well.
Acknowledgments The authors would like to thank the two anon-
ymous referees for their comments and useful suggestions. The
reported experiments comply with all laws of the country (Italy) in
which they were performed.
References
Agrillo C, Dadda M, Bisazza A (2007) Quantity discrimination in
female mosquitofish. Anim Cogn 10:63–70
Agrillo C, Dadda M, Serena G, Bisazza A (2008) Do fish count?
Spontaneous discrimination of quantity in female mosquitofish.
Anim Cogn 11(3):495–503
Anderson US, Stoinski TS, Bloomsmith MA, Marr MJ, Smith AD,
Maple TL (2005) Relative numerousness judgment and summa-
tion in young and old western lowland gorillas. J Comp Psychol
119(3):285–295
Table 1 Overall percentage of correct choices during training and
test phase for each cat
Subject Reinforced toward Percentage of correct choices during
Training (%) Test (%)
Nerina 3 71 74
Wilde 3 79 46
Suesse 2 77 82
Wieso 2 76 61
292 J Ethol (2009) 27:289–293
123
Beran MJ, Evans TA, Harris EH (2008) Perception of food amounts
by chimpanzees based on the number, size, contour length and
visibility of items. Anim Behav 75:1793–1802
Chiandetti C, Vallortigara G (2008) Is there an innate geometric
module? Effects of experience with angular geometric cues on
spatial re-orientation based on the shape of the environment.
Anim Cogn 11:139–146
Davis H, Perusse R (1988) Numerical competence in animals:
definitional issues, current evidence and a new research agenda.
Behav Brain Sci 11:561–579
Feigenson L, Carey S, Spelke ES (2002) Infants’ discrimination of
number vs. continuous extent. Cogn Psychol 44:33–66
Gelman R, Gallistel CR (1978) The child’s understanding of numbers.
Harvard University Press, Cambridge
Hauser MD, Carey S, Hauser LB (2000) Spontaneous number
representation in semi-free-ranging rhesus monkeys. Proc R Soc
Lond B Biol Sci 267:829–833
Heinsohn R (1997) Group territoriality in two populations of African
lions. Anim Behav 53:1143–1147
Lyon BE (2003) Egg recognition and counting reduce costs of avian
conspecific brood parasitism. Nature 422:495–499
McComb K, Packer C, Pusey A (1994) Roaring and numerical
assessment in the contests between groups of female lions,
Panther leo. Anim Behav 47:379–387
Sovrano VA, Bisazza A, Vallortigara G (2007) How fish do geometry
in large and in small spaces. Anim Cogn 10:47–54
Stephens DW, Krebs JR (1986) Foraging theory. Princeton University
Press, Princeton
Stevens JR, Wood JN, Hauser MD (2007) When quantity trumps
number: discrimination experiments in cotton-top tamarins
(Saguinus oedipus) and common marmosets (Callithrix jacchus).
Anim Cogn 10:429–437
Tomonaga M (2008) Relative numerosity discrimination by chim-
panzees (Pan troglodytes): evidence for approximate numerical
representations. Anim Cogn 11(1):43–57
Uller C, Jaeger R, Guidry G, Martin C (2003) Salamanders
(Plethodon cinereus) go for more: rudiments of number in a
species of basal vertebrate. Anim Cogn 6:105–112
Wilson ML, Hauser MD, Wrangham RW (2001) Does participation in
intergroup conflict depend on numerical assessment, range
location, or rank for wild chimpanzees? Anim Behav 61:1203–
1216
J Ethol (2009) 27:289–293 293
123