Validation of a temperament test for domestic cats

Abstract

Cats are popular companion animals, particularly in Europe and North America, and appear in correspondingly large numbers in animal shelters. Temperament tests are not widely used to assess cats before adoption from shelters. However, cats exhibit a wide range of temperaments as do the families adopting them and ensuring compatibility between the two could increase the rate of successful placement. Scores on a feline temperament profile (FTP), which measures a cat's responses to standardized interactions with an unfamiliar person, were compared between cats and over time and related to responses of cats to familiar and unfamiliar persons and to basal salivary cortisol levels. Cats showed significant differences in FTP scores (p<0.001). Ranking cats according to FTP scores resulted in three distinct groups of cats. Over eight months, changes in FTP scores were minor, with cats scoring somewhat more acceptably and less questionably following adoption. Acceptable scores on pre-adoption FTPs were positively correlated with 1) positive responses to familiar caretakers in housing rooms (p=0.01) and 2) average percent of time spent near either unfamiliar men or women in open field tests in novel rooms (p=0.01 in both instances). Thus, cats displaying general positive responses to humans did so in both familiar and test environments and with familiar and unfamiliar persons. No correlation was seen between FTP scores and basal salivary cortisol levels (p>0.05), though there were significant differences in cortisol levels between cats (p=0.04). The data indicate that the FTP was relatively stable over time for adult cats, and test scores correlated well with ethological observations of cats' interactions with humans. The FTP could provide an accurate, consistent assessment of cat temperament, leading to more successful placement of cats.

Full text

Validation of a Temperament Test for Domestic Cats
Published in 2003 in Anthrozoos, 16:332-351
By
Janice M. Siegford* MS, PhD
Animal Behavior and Welfare Group, Department of Animal Science,
Michigan State University, East Lansing, MI, USA
Sally O. Walshaw† MA, VMD
University Laboratory Animal Resources, College of Veterinary Medicine
Michigan State University, East Lansing, MI, USA
Petra Brunner‡ Dr.med.vet.
Animal Behavior and Welfare Group, Department of Animal Science,
Michigan State University, East Lansing, MI, USA
Adroaldo J. Zanella BVSc, PhD
Animal Behavior and Welfare Group, Department of Animal Science,
Michigan State University, East Lansing, MI, USA
*Corresponding author: J.M. Siegford, 1287C Anthony Hall, Animal Behavior and Welfare
Group, Department of Animal Science, Michigan State University, East Lansing, MI, USA.
Phone: 517-432-8212. Fax: 517- 353-1699. E-mail: siegford@msu.edu.
†S.O. Walshaw is now at the Department of Biomedical Sciences, Atlantic Veterinary College,
University of Prince Edward Island, Charlottetown, PE, Canada.
‡P. Brunner is now at Altana Pharma KG, D-78403 Konstanz, Germany.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1

Validation of a Temperament Test for Domestic Cats
Abstract
Cats are popular companion animals, particularly in Europe and North America, and appear in
correspondingly large numbers in animal shelters. Temperament tests are not widely used to
assess cats before adoption from shelters. However, cats exhibit a wide range of temperaments as
do the families adopting them and ensuring compatibility between the two could increase the rate
of successful placement. Scores on a feline temperament profile, which measures a cat’s
responses to standardized interactions with an unfamiliar person, (FTP) were compared between
cats and over time and related to responses of cats to familiar and unfamiliar persons and to basal
salivary cortisol levels. Cats showed significant differences in FTP scores (P < 0.001). Ranking
cats according to FTP scores resulted in three distinct groups of cats. Over 8 months, changes in
FTP scores were minor, with cats scoring somewhat more acceptably and less questionably
following adoption. Positive responses to familiar caretakers in housing rooms were positively
correlated to 1) acceptable scores on pre-adoption FTPs (P = 0.01) and 2) average percent of
time spent near either unfamiliar men or women in open field tests in novel rooms (P = 0.01 in
both instances). Thus, cats displaying general positive responses to humans did so in both
familiar and test environments and with familiar and unfamiliar persons. No correlation was seen
between FTP scores and basal salivary cortisol levels (P > 0.05) though there were significant
differences in cortisol levels between cats (P = 0.04). The data indicate that the FTP was
relatively stable over time for adult cats and test scores correlated well with ethological
observations of cats interacting with humans. The FTP could provide an accurate, consistent
assessment of cat temperament, leading to more successful placement of cats.
Keywords: feline, personality, assessment, open-field test, salivary cortisol
2
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
2

Introduction
There is a growing need to understand factors that contribute towards successful adaptation of
cats to the human environment. Animal shelters across the nation take in an estimated 6-8
million dogs and cats each year, and in many shelters approximately 70% of these cats are
euthanized (Human Society of the United States, National Council on Pet Population Study and
Policy). For example, in 1998 shelters in the state of Michigan euthanized more than 33,000 cats
(Michigan Department of Agriculture). Among the top reasons cats are likely to be relinquished
to shelters is because they do not match the expectations or lifestyle of the adopting family
(Salman et al. 1998, New et al. 2000).
Temperament testing has long been performed on dogs in order to assist their placement with
families, particularly at shelters placing animals for adoption. Temperament, individuality, and
personality are terms often used interchangeably in studies of animals to describe the sum total
of behavioral attributes which characterize an individual and set it apart from others (Mendl and
Harcourt 2000). While the merits of different forms of canine temperament testing may be
debated, temperament tests and physiological measures can be used successfully to predict
problem behaviors in dogs following adoption (van der Borg, Netto, and Planta 1991, Hennessy
et al. 2001). No such practice is used routinely for cats, however, despite the fact that several
checklists or easily scored ethograms have been developed that could be used for this purpose
(Lee et al. 1983, McCune 1995, Kessler and Turner 1997).
The absence of temperament assessment of cats often leads to the placement of cats on the basis
of appearance, age, and sex of the animal. While these characteristics may be important features
3
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
3

of owner expectation as they select a cat, the temperament of the cat and its suitability for its
adoptive family should be considered. Cat owners, or persons who have intensely observed the
behavior of an individual cat, develop a sense that each cat has unique behavioral characteristics
(Feaver, Mendl, and Bateson 1986, Mendl and Harcourt 2000). It becomes important, therefore,
to match cats and owners because discrepancies between an animal’s actual behavior and the
owner’s vision of ideal behavior can affect owner attachment to the animal For example, owners
that reported their cats were affectionate to them also reported themselves as more affectionate to
their cats (Turner and Stammbach-Geering 1990). Thus owner attachment based on desired
behaviors probably affects the likelihood that the owner will relinquish the animal (Serpell
1996). It would therefore appear that shelters should provide prospective owners with an
assessment of a cat’s temperament to be used with its physical characteristics in order to find a
cat that meets owner expectations on all levels. Additionally, cat temperament tests could be
used by veterinarians to establish baseline temperament profiles of feline patients which could be
reassessed on subsequent visits.
Methods have been developed to reliably assess individual differences in cats in an objective as
well as a subjective manner (Feaver, Mendl, and Bateson 1986, McCune 1995, Kessler and
Turner 1997, Turner 2000a). A variety of studies have examined the ontogeny of cat behavior
and individuality and their consistency over time (Martin 1986, Reisner et al. 1994, Bradshaw
and Cook 1996, Durr and Smith 1997, Lowe and Bradshaw 2001), response of cats to stressful
situations (Kessler and Turner 1997, 1999a, 1999b), cat-human interactions under a variety of
conditions (Mertens and Turner 1988, Mertens 1991, Podberscek, Blackshaw, and Beattie 1991,
McCune 1995, Turner 2000a), and strength of owner attachment (Karsh and Turner 1988). The
4
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
4

results of this work suggest that individual personalities of cats are one of the most significant
factors influencing cats’ behavior towards people (Mertens and Turner 1988), and that the
friendliness of cats to humans depends both on genetic factors and socialization at an appropriate
age (Karsh and Turner, 1988, Reisner et al. 1994, McCune 1995, Turner 2000b). Additionally,
these studies showed that many aspects of cat individuality remain stable over time, particularly
once the animal is 4-5 months old (e.g., Lowe and Bradshaw 2001). In sum, previous studies
have revealed salient features of cat individuality and the basis for their affinity for humans and
demonstrated that many traits relevant to cat-human interaction remain stable over time and
across situations. (For reviews of individuality in the cat and cat-human relationships, see Mendl
and Harcourt 2000 and Turner 2000b respectively.)
However, the techniques used to assess cats in many studies would not be practical for assessing
the temperaments of cats in shelters or in veterinarian’s offices. Ethological observations, while
objective, are often time consuming and require personnel trained in behavioral observation and
statistics in order to accurately record the necessary data and to analyze it for use. Subjective
assessments, on the other hand, require the assessor to have spent enough time with the cat to
form an intimate knowledge of its behavior under a variety of circumstances, which is unlikely to
occur in a shelter situation.
A standardized temperament test has the advantage of requiring less training by the user, less
time per animal and if a scoring system is used, easier analysis of the results. Additionally,
simpler behavioral criteria may be more relevant to owner-cat attachment (Serpell 1996). Thus
an effective feline temperament test could be used in shelters across the country by existing staff.
5
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
5

However, feline temperament tests generally have not been scientifically examined either by
administering the test repeatedly to the same group of cats across time and varying situations
(but see McCune 1995, Kessler and Turner 1997, 1999a, 1999b for exceptions) or by comparing
ethological observations to test scores to assess whether these tests measure relevant and stable
aspects of temperament, and, in particular, features relating to cat-human compatibility.
There were three objectives to this study. The first objective was to assess an existing feline
temperament test (Lee et al. 1983) by comparing the results of the test to ethological
observations of the same cats in the presence of familiar and unfamiliar humans in familiar and
unfamiliar locations in order to validate the accuracy of the test. Secondly, we examined the
results of the test over time and under changing circumstances, i.e. before and after adoption, to
determine if scores remained consistent under such varying circumstances. Finally, we explored
the relationship between scores on the test and salivary cortisol levels at baseline to determine if
a relationship existed between a cat’s performance on the test and its stress level.
Methods
Animals and Housing
This study was approved by the Michigan State University All-University Committee on Animal
Use and Care. Twenty female, domestic shorthair cats, ten months of age at the start of the study,
were examined over an eight month period. All cats were specific-pathogen free cats, which
were bred and raised at the same research animal production facility and shared a similar genetic
background. Additionally the breeding facility mandated that all cats receive the same amount
and type of handling (including petting and handling) each day to ensure uniform socialization
6
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
6

from infancy onward. The cats were housed in groups of 10 in two identical rooms at the
laboratory animal facility at Michigan State University (MSU). Each room contained two food
and water dishes, two litter boxes, various toys, and several carriers, tubs, and buckets of varying
sizes allowing cats to hide. Following the experiment, cats were adopted by members of the
community and returned to MSU 3 and 6 months later for follow-up FTP tests.
Experimental Procedures
Cats underwent evaluation using a standardized temperament test (Lee et al. 1983), the feline
temperament profile (FTP), twice before and 3 and 6 months after adoption (with the exception
of one cat, L069, which was not adopted). This test was originally designed to assess the
suitability of cats for placement in nursing homes by evaluating general levels of sociability,
aggressiveness, and adaptability to new situations. The test consisted of 10 different phases
during which the investigator proceeded from calling the cat at a distance of a few meters to
more proximate interactions, such as holding the cat while petting it or pulling on the cat’s tail. A
list of 5-7 possible responses by the cat was present under the description of each phase along
with space for observations. Each response in the list was described as ‘acceptable’ or
‘questionable.’ Occurrence (1) or absence (0) of each listed response by each cat was noted
during each phase of the test. For responses that could occur during multiple phases of the test,
such as eye contact, occurrence and absence scores from each phase were combined into one
score. For example, eye contact was listed as a possible response in five phases of the test;
therefore a cat could have a score of 0-5 for eye contact on the test. Some of the listed responses
never occurred during the experiment and were not used in the scoring system or examined in the
7
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
7

analysis.1 Using this scoring system, the maximum acceptable score on the FTP was 38 and the
maximum questionable score a cat could receive was 16. A list of observed responses and their
categorization as acceptable or questionable is given in Table 1. The order in which the cats were
tested was randomized on each occasion. All FTP were administered by an investigator, the
‘tester,’ and scored by a second investigator, the ‘observer.’ (Both tester and observer recorded
scores. However, as these scores had a high degree of interobserver reliability and as there were
occasional cases where the tester could not see the cat as clearly at observer, scores from the
observer were used in generating the FTP.) Overall acceptable and questionable scores for each
test were obtained by combining scores for items of each category together. To rank cats,
average acceptable and average questionable scores using all four FTP were generated for each
cat. Cats were then ranked by acceptable scores from largest to smallest scores then again by
questionable scores from smallest to largest. An average rank for each cat was then generated by
averaging the acceptable and questionable rankings.
Prior to adoption, interactions between the cats and their caretakers at MSU were monitored in
the rooms that housed the cats during three daily visits (28 minutes each) using time-lapse video
recording. Proximity to caretaker was measured every 30 seconds during the visit. Frequency of
the following behaviors was recorded during the visit: approaches person, touches person, and
retreats. Proximity scores were added to those of approach and touch frequency and these values
were averaged over the three exposures and termed ‘positive responses.’ The frequency of
1 Responses listed on the original FTP that never occurred during our study included: strikes hand, threatens to strike
hand, bites or attempts to bite, jumps up on lap, ignores toy, attends to something else in the room rather than the
toy, doesn’t hear object dropped, startles and runs to hide, startles then shows aggressive posture when item is
dropped. The majority of these unobserved responses fell into the ‘questionable’ response category. As our cats
were from a homogenous genetic and social background, these responses may be observed when a more
heterogeneous population of cats is examined and should be retained in the FTP until such testing has been
conducted.
8
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
8
9
10
11
12
13
14
15

retreats was considered a ‘negative response’ and the frequency of this behavior was averaged
over the three visits.
Prior to adoption, cats were also tested in an open field situation to determine their response to
two unfamiliar persons in a novel place (which would be similar to a visit to the veterinarian or
shelter). The open field test was conducted three times for each cat (n = 19) with one test per day
and a week between tests. A novel room (3 m x 3.7 m) the cats had never entered before was
used as the open field area. The room was divided into 30 (0.6 m x 0.6 m) areas using tape on the
floor (designated A-E across the shorter wall and 1-6 across the longer wall, giving each cell in
the grid a unique label, e.g. A1). A door was present in the middle of one of the longer walls (E3-
4). During testing, a man sat on the floor at one end of the room in the center of one of the
shorter walls, and a woman sat on floor at the other end of the room (C3 and C6). Different
unfamiliar men and women were used in each of the three open field tests for each cat. Cats were
carried into the room in a carrier, which was placed in the center of the floor and opened. Time
started when the cat left the carrier (some cats had to be removed from the carrier) and each test
lasted 5 minutes. The number of line crossings was recorded (one in and one out of a cell
equaled one total crossing) as was amount of time spent in each cell, which was then converted
into percent of total time in each cell. For analysis, cells near the door were combined into one
category. Cells in corners were similarly combined. Cells within a one-cell radius surrounding
the woman were combined into one category and cells surrounding the man were combined into
a second category to determine if cats responded differently to unfamiliar persons of different
sex.
9
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
16

Cortisol Measurements
Basal cortisol profiles for each cat were obtained in a non-invasive manner using saliva samples
collected from unrestrained cats twice a day, at 07:00 and 19:00, on four different days. Cats
were conditioned to come to the researcher by tapping on an open can of cat food with a spoon.
Cats were then presented with cotton swabs and allowed to chew on the swabs until they were
thoroughly moistened. Cats were rewarded for chewing on the swabs by giving them some of the
canned food. Swabs were centrifuged to remove saliva, which was stored at -20°C until
processing. An enzyme-linked immunoabsorbent assay (ELISA) was used to measure cortisol as
described by Cooper et al. (1989). Briefly, polystyrene microtitre plates were coated with the
secondary antibody (goat anti-rabbit IgG at 1:8000), washed, then incubated overnight with
primary antibody (rabbit anti-bovine cortisol-3- carboxymethyloxime) diluted 1:4000. Duplicate
samples of saliva (in a range of dilutions from 1:4 to 1:100), controls, reagents, and triplicate
standards (1-1000 pg) were added to the plates over ice followed by addition of horseradish
peroxidase-labelled cortisol (1:30000) to prescribed wells. Plates were incubated for 2 hours then
washed and 3,3’,5,5’-tetramethylbenzidine in dimethyl sulfoxide (10mg/ml), made fresh with
hydrogen peroxide, added to the plates. The color reaction was stopped by addition of sulfuric
acid then optical density was read in a plate reader (Bio Rad) equipped with a 450 nm filter.
Inter-assay variability was 15.0% and intra-assay variability was 4.3%.
Analyses
To ensure that the responses listed on the FTP measured only positive or negative reactions of
the cat to the tester, the relationships between acceptable and questionable scores using average
10
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
17

scores from all four FTP were assessed using a Spearman rank correlation. We hypothesized that
acceptable and questionable scores would be negatively correlated.
Comparisons of acceptable scores on the FTP between cats across all tests and over time for all
cats were performed using a one-way analysis of variance (ANOVA) with Bonferroni multiple
comparisons tests where needed. A Kruskal-Wallis one-way ANOVA was used to examine the
differences in questionable scores between cats across all tests and over time for all cats because
the data for these scores did not meet the assumptions of normality. Where needed multiple
comparisons on these data were performed using a Kruskal-Wallis Multiple-Comparison Z-
Value test with a Bonferroni correction factor. Cats were ranked 1-20 by highest to lowest
acceptable scores averaged across all four FTP tests then ranked 1-20 by lowest to highest
questionable scores averaged across all four FTP tests. These two ranks were then averaged to
create an overall rank based on FTP responses. A cluster analysis was performed (SAS Institute,
Cary, NC) by cat on average scores from all four FTPs for each response to assess the validity of
grouping cats by FTP scores.
A Spearman rank correlation was used to assess each of the following relationships: acceptable
scores on pre-adoption FTPs with positive responses to a familiar caretaker, questionable FTP
scores with negative responses to caretakers, acceptable scores on pre-adoption FTPs with
average percent of time spent in proximity to unfamiliar men or women in the open field test,
questionable scores on pre-adoption FTPs with average time spent in the corners or at the door in
the open field test, and the average number of cell crossings made in all open field tests with
acceptable or questionable pre-adoption FTP scores.
11
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
18

Circadian cortisol rhythms were square root transformed to meet the assumptions of normality
then analyzed using a mixed general linear model (proc mixed) with repeated measures (SAS
Institute, Cary, NC) with day and time of sample as independent factors and cat as a random
effect. Least squares means of cortisol levels were compared and adjusted using the Tukey-
Kramer method to control for Type-I error rate. Results following transformation were similar to
those obtained using non-transformed data in a similar analysis; therefore results from the
transformed analysis are reported in the text. Basal cortisol levels were compared between cats
using a Kruskal-Wallis one-way ANOVA to determine if significant differences existed between
cats. A Kruskal-Wallis Multiple-Comparison Z-Value test with a Bonferroni correction factor
was then used to determine which cats had differing basal cortisol levels. Correlations between
scores on the pre-adoption FTPs and basal cortisol levels were analyzed using Spearman rank
correlation. Correlations between basal cortisol levels and groups of cats that resulted from FTP
ranking were analyzed using a one-way ANOVA.
For all analyses, an alpha of less than 0.05 was considered significant. All data are given as mean
± standard error of the mean (SEM). Unless otherwise mentioned analyses were performed using
NCSS (Number Cruncher Statistical Systems, Kaysville, UT).
Results
Behavioral Measures and Correlations between Measures
Average acceptable scores on the all FTP were negatively correlated with average questionable
scores on the test (Figure 1; n = 20, rho = -0.85, P < 0.001). Comparison of acceptable scores
12
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
19

received on all FTPs revealed significant differences in scores between cats (Table 2; F19,60 =
5.95, P < 0.001). No difference in questionable scores was observed between cats, likely due to
the generally low scores received in this category (Table 2; n = 19, χ2 = 25.08, P = 0.16).
However, several cats stand out as having higher questionable scores compared to their peers;
while several others have lower scores (Table 2). Ranking of cats (Table 2) was based on an
average of ranking received for acceptable scores (higher rank for larger acceptable scores) and
ranking received for questionable scores (lower rank for larger questionable scores). Cluster
analysis revealed the lowest ranked cats were clearly separated from other cats and some
separation of cats with the highest ranks from those ranked in the middle also existed (Figure 2).
Comparison of FTP scores over time revealed a non-significant increase in acceptable scores
(Figure 3; F3,76 = 1.29, P = 0.28). Questionable scores changed significantly over time (Figure 3;
F3,76 = 6.01, P = 0.001) as 6 months after adoption, cats had lower questionable scores when
compared with scores on the second pre-adoption test or scores 3 months post-adoption.
The correlation between acceptable scores on pre-adoption FTPs and the positive responses of
cats to familiar caretakers in their housing rooms was positive and significant (Figure 4; n = 20,
rho = 0.51, P = 0.02). There was no correlation between questionable scores on pre-adoption
FTPs and negative responses of cats to caretakers (n = 20, rho = 0.20, P = 0.39). A significant
positive correlation was observed between acceptable scores on the pre-adoption FTPs and
average percent of time the cats spent near the unfamiliar male in the novel room during the open
field test (Figure 5a; n = 19, rho = 0.60, P = 0.01). A similar significant positive correlation was
observed between percent of time the cats spent near the unfamiliar female in the novel room
13
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
20

during the open field test (Figure 5b; n = 19, rho = 0.57, P = 0.01). A significant positive
correlation was observed between questionable scores on pre-adoption FTPs and percent of time
cats spent in corners of the novel room during open field tests (Figure 5c; n = 19, rho = 0.51, P =
0.03), while no significant correlation was observed between questionable scores on pre-
adoption FTPs and average percent time spent near the door in open field tests (Figure 5d; n =
19, rho = 0.19, P = 0.44). The number of cell crossings made in open field test was positively
correlated with acceptable scores on pre-adoption FTPs (Figure 6; n = 19, rho = 0.54, P = 0.02)
but was not related to questionable scores on pre-adoption FTPs (n = 19, rho = -0.32, P = 0.18).
Cortisol Measurements
Salivary cortisol measures from unrestrained cats on each of four consecutive sampling days
were not significantly different between sample times of 7:00 and 19:00 (F1,130 = 0.06, P = 0.81),
nor did the cortisol levels vary significantly between sample days (F3,130 = 0.23, P =0.87). There
was, however, a significant interaction between time and day the cortisol samples were taken
(F3,130 = 2.92, P = 0.04) owing to a significant difference in cortisol levels between samples from
the morning and evening of the first day (t130 = 2.55, P = 0.012) and a nearly significant
difference between cortisol levels in samples from the mornings of the first and third days (t130 =
1.97, P = 0.051). Overall comparisons of basal cortisol levels between cats revealed significant
individual differences (Figure 7; n = 20, χ2 = 31.07, P = 0.04). However, after correcting for
Type-I error rate, multiple comparisons revealed no individuals significantly different from one
another.
Correlations between Temperament Test Scores and Basal Cortisol Levels
14
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
21

No relationship was observed between basal cortisol levels from saliva and acceptable or
questionable scores on pre-adoption FTPs (n = 19, t = 0.70, P = 0.49 and n = 19, t = 0.11, P =
0.91 respectively). No relationship was observed between basal cortisol levels and the groups of
cats that resulted from the FTP rankings (F2,17 = 0.39, P = 0.68).
Discussion
Feaver, Mendl, and Bateson (1986) and Turner (2000a) used subjective measures of cat
individuality, which, while they were strongly correlated with ethological observations and were
quick and easy to perform, required that the evaluator know the cat well enough to rate many
aspects of the cat’s behavior and individuality. This is not practical in a shelter setting or
veterinarian’s office where staff may have limited time and opportunity to observe each cat,
though experienced personnel are likely practiced at assessing animals quickly. Additionally in
these situations, the cats themselves may have limited opportunity to perform a wide range of
behaviors and are likely to be stressed in such situations, providing staff with little information
on which to base their judgments.
Other tests, such as the ‘cat stress test’ developed by Kessler and Turner (1997), assess the
general levels of stress a cat is experiencing in a given situation rather than temperament. While
this test evaluates a cat’s reaction to situations that may be stressful, such as housing or density
(Kessler and Turner 1997, 1999a, 1999b) and perhaps could evaluate responses to handling or
presence of humans, it does not provide an indication of a cat’s affinity for humans.
15
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
22

The ideal temperament test, then, would combine ease of use, quickness, consistency, and
reliability. However, few tests of this type exist for cats (Lee et al. 1983, McCune 1995) and
these have either not been scientifically validated or not assessed for consistency and reliability
across time or correlated with other measures. The temperament test used in the current study
was developed by Lee et al. (1983) for use in evaluating cats for placement in nursing homes.
The test is very similar to those used to evaluate dogs for adoption from shelters, and it evaluates
the cat’s responses to a variety of increasingly interactive and challenging situations in a novel
environment ranging from a novel person calling the cat to this same person involving the cat in
play and the cat’s reaction to a startling noise. This type of test, which allows for an objective
tester rather than one with extensive knowledge of the cat, has the advantage of being simple to
perform and score. As we used the test, we found some of the responses listed under various
phases of the test did not occur (Table 1). However, as our cats were from a homogenous genetic
and social background, these responses may be observed when a more heterogeneous population
of cats is examined and should be retained in the FTP until such testing has been conducted.
After more extensive validation in practical settings, it may be possible to simplify the FTP to
make it more economical for shelter use.
Results from the current study indicate that the FTP performs consistently over time, even when
the cat’s circumstances change dramatically following adoption. Acceptable scores on the FTP
correlated positively with the cats’ responses to familiar caretakers and unfamiliar humans in
housing rooms and novel (i.e., the open field test) environments, suggesting that the test also
accurately measures the cats’ affinity for humans. All of the cats studied received equal amounts
of socialization as kittens, therefore it will be important to compare FTP scores from cats that
16
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
23

received varying amounts of socialization to ensure the test is capable of measuring the full
range of variability demonstrated by cats in their affinity for humans. Additionally, the FTP itself
should be performed on cats in familiar environments as well to evaluate scores under
comfortable circumstances. This would allow for comparison with scores from the novel
situations used in the current study, mimicking potential changes in scores that could occur
between shelter and home environments.
Examination of changes in scores over time shows that they remain fairly stable, providing a
consistent measure over 8 months and changing circumstances. Observed decreases in
questionable scores suggest that cats might develop more affinity to their adopters over time or
with age, though the decrease was not enough to markedly change the rankings of cats between
the first and last test, particularly of those cats that fell into the lowest ranking group.
A negative correlation between acceptable and questionable scores on the FTP indicates that the
test is capable of discriminating between cats that react favorably to humans from those that do
not. If acceptable and questionable scores had been positively correlated, the test might have
been measuring active versus passive responses of cats rather than temperament. If the scores
were uncorrelated, however, the test would not have much predictive value because a cat with a
high acceptable score could be as likely to have a high as a low questionable score. Using FTP
scores, cats could be ranked by larger acceptable scores and smaller questionable scores (Table
2). This system resulted in three natural groups of cats, with lowest ranking cats in particular
separated from the other groups by ranking (Table 2), which compared favorably with the
statistical results of cluster analysis (Figure 2). Cats in the high group had both high acceptable
17
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
24

and low questionable scores, and as such are cats that could be placed with a family, a novice cat
owner, or a person desiring a sociable pet. Cats in the middle group had mid range scores in both
the acceptable and questionable categories, suggesting that they might require a bit more
experience and understanding from an owner (and perhaps only older children in the household)
in order to form an attachment. Cats in the low group had both low acceptable scores and the
highest questionable scores. These cats would require an experienced cat owner or someone that
does not expect a sociable, attention-seeking pet. For example, in the present study, cat K285
was adopted by a single woman living alone, however, the woman did not mention that her
grandchildren visited frequently. A month following adoption the cat was returned to MSU
because she was not social with the grandchildren, as was predicted by her FTP scores.
Lowe and Bradshaw (2001) showed that features such as investigative behavior and boldness
were stable features of cats studied in the home between four months and two years of age.
However, other work also indicates that many aspects of cat behavior and temperament continue
to change as the animals mature (Reisner et al. 1994) making a temperament test of kittens
unpredictive of adult behavior, as is the case with puppies (Wilsson and Sundgren 1998).
However, knowledge of early socialization of cats towards humans during the sensitive phase in
the first few months of life or friendliness of the father may be able to suggest the affinity of a
kitten will have for humans when it matures (Karsh and Turner, 1988, McCune 1995, Turner
2000b).
Use of the feline temperament test is not intended to condemn those cats which fall into the
lowest ranking group; rather it offers a realistic assessment of their temperament in order to
18
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
25

match cats with the expectations of adopters. The goal is to use this tool to achieve successful
adoptions of cats as measured by low return rates and high owner satisfaction as satisfaction and
retention of adopted cats have been related to the individuality, compatibility, and behavior of
the cat (Salman et al. 1998, Neidhart and Boyd 2002). Similar studies have demonstrated that
dogs’ performance on temperament tests can be related to problem behavior after adoption (van
der Borg, Netto, and Planata 1991, Ledger and Baxter 1997), giving shelters the ability to use
such tests to place dogs more appropriately.
Cortisol levels reflect activity of the hypothalamic-pituitary-adrenal axis, which is a stress
responsive system, and measures of salivary cortisol have been validated in many species,
including dogs, as a non-invasive alternative to measuring cortisol in plasma (Cooper et al. 1989,
Beerda et al. 1996). In keeping with previous findings that feline adrenal function does not have
a circadian rhythm (Johnston and Mather 1979) we found no difference in cortisol levels
between samples taken at 7:00 versus 19:00.
The absence of a relationship between cortisol levels in saliva and performance on the
temperament test may suggest no correlation between intrinsic cortisol levels and aspects of cat
temperament important in cat-human interaction. A more relevant comparison, however, may be
found between individual salivary cortisol levels (as in Figure 7) and scores from a cat stress test
(Kessler and Turner 1997). Alternatively, use of saliva as a means of non-invasively sampling
cat cortisol levels may not provide accurate information on physiological stress as little cortisol
is likely excreted in saliva and what is present may be rapidly metabolized (Carlstead et al. 1992,
Carlstead, Brown, and Strawn 1993, Graham and Brown 1996, Schatz and Palme 2001).
19
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
26

However, Beerda et al. (1996) correlated cortisol levels of saliva with those of plasma in dogs,
despite the fact that these levels were 10-fold lower than those in plasma, suggesting that with
validation, use of saliva to measure cortisol levels in cats may also be possible.
Conclusion
Currently, cats are often selected for adoption on superficial characteristics, such as coat color,
while the cat’s behavior is a frequently cited reason for relinquishment of cats to shelters where
many are subsequently euthanized. While many organizations involved in cat adoption realize
the need to assess a cat’s temperament in order to place the cat in a compatible home, feline
temperament tests either have not been validated or made available for this purpose. This study
demonstrates that temperament testing in cats can provide an accurate and consistent measure of
a cat’s sociability, aggressiveness, and adaptability and could prove a useful tool for shelter staff,
veterinarians, and others needing to assess the temperament of a cat.
20
452
453
454
455
456
457
458
459
460
461
462
463
464
27

Acknowledgements
This project was funded by a grant from the Companion Animal Fund at the College of
Veterinary Medicine at Michigan State University awarded to Dr. S.O. Walshaw and Dr. P.
Brunner. Christine Heinz provided much appreciated assistance with data collection. Our thanks
to Dr. C. Bollinger and Dr. R. Walshaw for sharing their cats with us for this study.
References
Beerda, B., Schilder, M.B.H., Janssen, N.S.C.R.M., and Mol, J.A. 1996. The use of salivary
cortisol, urinary cortisol, and catecholamine measurements for a noninvasive assessment of
stress response in dogs. Hormones and Behavior 30:272-279.
Bradshaw, J.W.S. and Cook, S.E. 1996. Patterns of pet cat behaviour at feeding occasions.
Applied Animal Behaviour Science 47:61-74.
Carlstead, K., Brown, J.L., Monfort, S.L., Killens R., and Wildt, D.E. 1992. Urinary monitoring
of adrenal responses to psychological stressors in domestic and nondomestic felids. Zoo Biology
11:165-176.
Carlstead, K., Brown, J.L., and Strawn, W. 1993. Behavioral and physiological correlates of
stress in laboratory cats. Applied Animal Behaviour Science 38:143-158.
21
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
28

Cooper, T.R., Trunkfield, H.R., Zanella, A.J., and Booth, W.D. 1989. An enzyme-linked
immunoabsorbent assay for cortisol in the saliva of man and domestic farm animals. Journal of
Endocrinology 123:R13-16.
Durr, R. and Smith, C. 1997. Individual differences and their relation to social structure in
domestic cats. Journal of Comparative Psychology 4:412-418.
Feaver, J., Mendl, M., and Bateson, P. 1986. A method for rating the individual distinctiveness
of domestic cats. Animal Behaviour 34:1016-1025.
Graham, L.H. and Brown, J.L. 1996. Cortisol metabolism in the domestic cat and implications
for non-invasive monitoring of adrenocortical function in endangered felids. Zoo Biology 15:71-
82.
Hennessy, M.B., Voith, V.L., Mazzei, S.J., Buttram, J., Miller, D.D., and Linden, F. 2001.
Behavior and cortisol levels of dogs in a public animal shelter, and an exploration of the ability
of these measures to predict problem behavior after adoption. Applied Animal Behaviour
Science 73:217-233.
Johnston, S.D. and Mather, E.C. 1979. Feline plasma cortisol (hydrocortisone) measured by
radioimmunoassay. American Journal of Veterinary Research 40:190-192.
Karsh, E.B. and Turner, D.C. 1988. The human-cat relationship. In The Domestic Cat:
22
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
29

The Biology of its Behaviour, 1st edn. 159-177. eds. D.C. Turner and P. Bateson. Cambridge:
Cambridge University Press.
Kessler, M.R. and Turner, D.C. 1997. Stress and adaption of cats (Felis silvestris catus) housed
singly, in pairs and in groups in boarding catteries. Animal Welfare 6:243-254.
Kessler, M.R. and Turner, D.C. 1999a. Socialization and stress in cats (Felis silvestris catus)
housed singly and in groups in animal shelters. Animal Welfare 8:15-26.
Kessler, M.R. and Turner, D.C. 1999b. Effects of density and cage size on stress in domestic cats
(Felis silvestris catus) housed in animal shelters and boarding catteries. Animal Welfare 8:259-
267.
Ledger, R.A. and Baxter, M.R. 1997. The development of a validated test to assess the
temperament of dogs in a rescue shelter. In Proceedings of the First International Conference on
Veterinary Behavioral Medicine. 87-92. eds. D.S. Mills, S.E. Heath, L.J. Harrington.
Wheathampsted, UK: Universities Federation for Animal Welfare.
Lee, R.L., Zeglen, M.E., Ryan, T., and Hines, L.M. 1983. Guidelines: animals in nursing homes.
California Veterinarian 3:22a-26a.
Lowe, S.E. and Bradshaw, J.W.S. 2001. Ontogeny of individuality in the domestic cat in the
home environment. Animal Behaviour 61:231-237.
23
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
30

Martin, P. 1986. An experimental study of weaning in the domestic cat. Behaviour 99:221-249.
McCune, S. 1995. The impact of paternity and early socialization on the development of cats’
behaviour to people and novel objects. Applied Animal Behaviour Science 45:109-124.
Mendl, M. and Harcourt, R. 2000. Individuality in the domestic cat: origins, development and
stability. In The Domestic Cat: The Biology of its Behaviour, 2nd edn. 47-64. eds. D.C. Turner
and P. Bateson. Cambridge: Cambridge University Press.
Mertens, C. and Turner, D.C. 1988. Experimental analysis of human-cat interactions during first
encounters. Anthrozoös 2:83-97.
Mertens, C. 1991. Human-cat interactions in the home setting. Anthrozoös 4:214-231.
Neidhart, L. and Boyd, R. 2002. Companion animal adoption study. Journal of Applied Animal
Welfare Science 5:175-192.
New, J.C., Jr., Salman, M.D., King, M., Scarlett, J.M., Kass, P.H., and Hutchison, J.M. 2000.
Characterstics of shelter-relinquished animals and their owners compared with animals and their
owners in U.S. pet-owning households. Journal of Applied Animal Welfare Science 3:179-200.
24
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
31

Podberscek, A.L., Blackshaw, J.K., and Beattie, A.W. 1991. The behaviour of laboratory colony
cats and their reactions to a familiar and unfamiliar person. Applied Animal Behaviour Science
31:119-130.
Reisner, I.R., Houpt, K.A., Erb, H.N., and Quimby, F.W. 1994. Friendliness to humans and
defensive aggression in cats: the influence of handling and paternity. Physiology and Behaviour
55:1119-1124.
Salman, M.D., New, J.G., Jr., Scarlett, J.M., Kass, P.H., Ruch-Gallie, R., and Hetts, S. 1998.
Human and animal factors related to the relinquishment of dogs and cats in 12 selected animal
shelters in the United States. Journal of Applied Animal Welfare Science 1:207-226.
Schatz, S. and Palme, R. 2001. Measurement of faecal cortisol metabolites in cats and dogs: a
non-invasive method for measuring adrenocortical function. Veterinary Research
Communications 25:271-287.
Serpell, J.A. 1996. Evidence for an association between pet behavior and owner attachment
levels. Applied Animal Behaviour Science 47:49-60.
Turner, D.C. and Stammbach-Geering, K. 1990. Owner assessment and the ethology of human-
cat relationships. In Pets, Benefits and Practice. 25-30. ed. I. Burger. London: BVA Publications.
25
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
32

Turner, D.C. 2000a. Human-cat interactions: relationships with, and breed differences between,
non-pedigree, Persian and Siamese cats. In Companion Animals and Us: Exploring the
Relationships between People and Pets. 257-271. eds. A.L. Podberscek, E.S. Paul, J.A. Serpell.
Cambridge: Cambridge University Press.
Turner, D.C. 2000b. The human-cat relationship. In The Domestic Cat: The Biology of its
Behaviour. 2nd edn. 191-206. eds. D.C. Turner and P. Bateson. Cambridge: Cambridge
University Press.
van der Borg, J.A.M., Netto, W.J., Planta, D.J.U. 1991. Behavioural testing of dogs in animal
shelters to predict problem behaviour. Applied Animal Behaviour Science 32:237-251.
Wilsson, E. and Sundgren, P.-E. 1998. Behaviour test for eight-week old puppies—heritabilities
of tested beahviour traits and its correspondence to later behaviour. Applied Animal Behaviour
Science 58:151-162.
26
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
33

Table 1. Summary of Scoring of the Feline
Temperament Profile
Cat Responses Scored as:
Responses to Tester
Eye contact Acceptable
Approaches/circles around tester Acceptable
Sniffs hand Acceptable
Rubs/bumps head against tester Acceptable
Rolls Acceptable
Vocalization: meow/purr/chirrups Acceptable
Call cat, cat approaches tester Acceptable
Retreats/withdraws Questionable
Vocalization: hisses/growls Questionable
Challenge 1: Pull Tail
Turns around Acceptable
No reaction Acceptable
Struggles Questionable
Growls/strikes/hisses Questionable
Challenge 2: With Toy
Watches toy Acceptable
Chases toy Acceptable
Allows stroking Acceptable
Challenge 3: Drop Object
Turns and relaxes Acceptable
Runs to investigate Acceptable
Ignores noise Acceptable
Responses listed on the original FTP that never occurred during our
study included: strikes hand, threatens to strike hand, bites or attempts
to bite, jumps up on lap, ignores toy, attends to something else in the
room rather than the toy, doesn’t hear object dropped, startles and runs
to hide, startles then shows aggressive posture when item is dropped.
All but one of the unobserved responses fell into the ‘questionable’
response category.
27
591
592
593
594
595
596
597
598
599
600
601
34

Table 2. Average Scores of Cats from All
Temperament Tests
Cat # Acceptable Score Questionable Score Rank Grouping
L097 22.80a0.75 1 High
L099 20.80a,d 1.00 2 High
L041 21.50a,d 2.00 3 High
L089 19.50a,c,d 1.50 3 High
L307 19.00a,c,d 1.25 3 High
L101 20.50a,d 1.75 6 High
L093 16.80 1.50 7 Middle
L371 16.00 2.25 8 Middle
L007 16.00 2.75 9 Middle
K227 15.30 2.75 10 Middle
L077 14.00 2.50 10 Middle
L057 10.50b,c,d 2.00 12 Middle
L037 14.30 2.75 13 Middle
L091 12.00 2.25 13 Middle
L079 14.00 3.00 15 Middle
L065 13.50 3.50 16 Middle
L075 7.50b,c 4.50* 17 Low
L039 5.00b3.25 17 Low
K285†6.75b5.00* 19 Low
L069‡5.00b5.75* 20 Low
Different superscripted letters represent significant differences between scores in that
category. Scores with the same superscripted letters were not significantly different
from one another and scores without any superscripts were not significantly different
from any other scores in that category. Asterisks denote negative scores that appeared
higher than those of other cats despite a lack of significant difference. †Cat K285 was
returned one month after adoption. ‡Cat L069 was never adopted.
28
602
603
604
605
606
607
608
609
610
611
612
35

Figure Legends
Figure 1. Relationship between average acceptable and questionable average FTP scores.
Note different scales on the x and y axes. A trend line has been added to the data to
highlight the significant relationship of the data sets in this and subsequent scatter plots.
Figure 2. Cluster analysis of cats based on an average of scores over all four FTPs for
each response during the FTP. Heavy, horizontal lines directly above cat identification
numbers indicate clustering of cats together in groups. See Table 2 for comparison of this
grouping with overall acceptable and questionable score ranking.
Figure 3. Temporal changes in acceptable and questionable FTP scores. Pre-adopt 1 =
pre-adoption FTP in September, pre-adopt 2 = pre-adoption FTP in October, post-adopt
3m = FTP 3 months post-adoption in December-January, post-adopt 6m = FTP 6 months
post-adoption in April.
Figure 4. A positive relationship exists between acceptable FTP scores and positive
responses of cats to familiar caretakers.
Figure 5. Relationships between A) acceptable FTP scores prior to adoption and average
percent time cats spent near the unfamiliar man across all open field tests, B) acceptable
FTP scores prior to adoption and percent time cats spent near the unfamiliar woman
across all open field tests, C) Questionable pre-adoption FTP scores and percent time
29
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
36

spent in corners during the open field tests, and D) questionable pre-adoption FTP scores
and percent time spent near the door during the open field tests.
Figure 6. A positive relationship exists between the average number of cell crossings cats
made across all open field tests and acceptable FTP scores prior to adoption.
Figure 7. Mean salivary basal cortisol levels in individual cats prior to adoption displayed
from lowest to highest levels. Error bars represent ± SEM.
30
635
636
637
638
639
640
641
642
37

Figure 1
0
1
2
3
4
5
6
7
0 5 10 15 20 25 30
Acceptable FTP Score
Questionable FTP Score
n = 20
P < 0.001
31
643
644
38

Figure 2
0.00
0.25
0.50
0.75
1.00
1.25
1.50
Cat #
L093
K227
L037
L079
L091
L065
L057
L007
L077
L371
L307
L089
L097
L101
L099
L041
L075
L039
K285
L069
Average Distance between Clusters
0.00
0.25
0.50
0.75
1.00
1.25
1.50
Cat #
L093
K227
L037
L079
L091
L065
L057
L007
L077
L371
L307
L089
L097
L101
L099
L041
L075
L039
K285
L069
Average Distance between Clusters
32
645
646
39

Figure 3
0
2
4
6
8
10
12
14
16
18
20
Acceptable Ques tionable
FTP Score
pre-adopt 1
pre-adopt 2
post-adopt 3m
post-adopt 6m
aa
b
33
647
648
40

Figure 4
0
1
2
3
4
5
6
7
0 5 10 15 20 25
Acceptable FTP Score
Positive Responses to Caretaker
n = 20
P = 0.02
34
649
650
41

Figure 5
0
5
10
15
20
25
0 10 20 30 40
Perce nt Time Near Ma n
Acceptable FTP S core
0
5
10
15
20
25
0 10 20 30 40
Perce nt Time Near Woman
Acceptble FTP Score
0
1
2
3
4
5
6
0 10 20 30 40 50 60
Perce nt Time i n Corners
Questionable FTP Score
0
1
2
3
4
5
6
0 10 20 30 40 50 60
Perce nt Time Spent Nea r Door
Questionable FTP Score
n = 19
P = 0.01
n = 19
P = 0.01
n = 19
P = 0.03
n = 19
P = 0.44
35
651
652
42

Figure 6
0
5
10
15
20
25
0 50 100 150
Number of Cell Crossings
Acceptable FTP Score
n = 19
P = 0.02
36
653
654
43

Figure 7
0
1
2
3
4
5
6
7
8
L007
L041
L065
L075
L097
K227
L079
L099
L037
L101
L057
L089
L091
K285
L039
L307
L069
L077
L093
L371
Cat #
Salivary Cortisol (nmol/L)
37
655
656
44