1
Towards Wild Psychometrics: Linking Individual Cognitive Differences to
2
Fitness
3
Alex Thorntona,*, Jess Isdenb and Joah R. Maddenb
4
a
5
Penryn Campus, Treliever Road, Penryn, TR10 9FE, UK
Centre for Ecology and Conservation, Department of Biosciences, University of Exeter,
6
7
b
8
Exeter, Perry Road, Exeter EX4 4QG, UK
Centre for Research in Animal Behaviour, School of Psychology, University of
9
10
* Address correspondence to Alex Thornton
11
Email: alex.thornton@exeter.ac.uk
12
Tel: +44(0)1326 255081
13
14
15
16
17
18
19
20
21
Running title: Towards wild psychometrics
22
Towards Wild Psychometrics: Linking Individual Cognitive Differences to
23
Fitness
24
25
Abstract
26
Our understanding of the processes underlying animal cognition has improved
27
dramatically in recent years, but we still know little about how cognitive traits evolve.
28
Following Darwinian logic, to understand how selection acts on such traits we must
29
determine whether they vary between individuals, influence fitness and are heritable.
30
A handful of recent studies have begun to explore the relationship between variation
31
in individual cognitive performance and fitness under natural conditions. Such work
32
holds great promise, but its success is contingent on accurate characterisation and
33
quantification of the cognitive differences between individuals. Existing research has
34
typically adopted a “problem-solving” approach, assuming that individuals that
35
complete novel tasks have greater cognitive prowess than those that do not. We
36
argue that this approach is incapable of determining that individual differences are
37
due to cognitive factors. We propose the adoption of psychologically-grounded
38
psychometric testing and discuss the criteria necessary to examine the fitness
39
consequences of cognitive variation in the wild.
40
41
42
43
Key words: cognition; evolution; heritability; individual differences; natural selection;
44
psychometrics
45
The evolution of cognition remains one of the most poorly understood issues in
46
biology. Traditionally, studies of cognitive evolution have employed a comparative
47
approach, relating between-species or between-population differences in brain size
48
or cognitive performance to differences in ecology, social systems or life history (e.g.
49
Balda and Kamil 2002;. Pravosudov and Clayton 2002). A recent upsurge of interest
50
in cognition among behavioural ecologists has stimulated the emergence of a
51
powerful new approach, shifting the focus to differences between individuals. Just as
52
field studies linking individual phenotypic differences to reproductive fitness have
53
furthered our understanding of selection acting on behavioural and morphological
54
traits (Clutton-Brock and Sheldon 2010), a focus on the causes and consequences
55
of variation in individual cognitive performance can provide crucial insights into
56
cognitive evolution (Thornton & Lukas 2012). A few pioneering studies have
57
generated tantalising indications of how cognitive differences may impact on fitness
58
correlates such as mating success (Keagy et al. 2009; 2011), reproductive success
59
(Cole et al. 2012; Cauchard et al. 2013) and competitive ability (Cole and Quinn
60
2012). We support this approach wholeheartedly, but are concerned that the
61
cognitive tests used in these field studies differ substantially from established
62
psychological theory and practice.
63
reliably characterised individual cognitive differences in wild animals. Here, we
64
examine the criteria necessary to examine the fitness consequences of individual
65
cognitive variation, and suggest refinements and improvements to current methods.
Thus, it is questionable whether they have
66
67
First and most importantly, we must ensure that we are measuring cognitive
68
differences and not variation in other traits or random noise. Cognition, broadly
69
defined, encompasses the mechanisms by which animals acquire, process, store
70
and act on information from the environment (Shettleworth 2010). Whereas
71
morphology and behaviour can be measured accurately under natural conditions,
72
cognitive processes cannot be observed directly and must be inferred through
73
experimentation (Shettleworth 2010). Few studies in the wild have employed
74
experimental protocols that can explicitly quantify individual differences in specific
75
cognitive processes (but see Boogert et al. 2010; Nachev & Winter 2012; Isden et al
76
2013). Instead, research has more commonly relied on what we term the “problem-
77
solving approach”, whereby animals in the field or temporarily brought into captivity
78
are presented with novel problems, under the explicit or implied assumption that
79
individuals that solve tasks at all, or do so faster than others, possess higher levels
80
of cognitive ability (Keagy et al. 2009; Cole et al. 2012; Cole and Quinn 2012;
81
Cauchard et al. 2013). Some of these tests elicit behaviour that is part of the
82
standard repertoire of the species (e.g. male satin bowerbirds removing red objects
83
from their bowers: Keagy et al., 2009; great tits removing obstructions from their
84
nest-boxes: Cauchard et al., 2013). Others use more artificial tasks to elicit
85
apparently innovative behaviour (e.g. pulling a lever to obtain food rewards; Cole et
86
al. 2012). Such tests have been used to great effect in animal innovation literature to
87
determine the physical and behavioural factors that influence individual tendencies to
88
perform novel behaviours (e.g. Morand-Ferron et al. 2011), but we strongly doubt
89
their ability to measure cognitive variability reliably. Under Shettleworth’s broad
90
definition, all behavioural outcomes involve some cognitive processing, but this does
91
not imply that solvers are necessarily more generally cognitively adept than non-
92
solvers. Solving a new problem on a single presentation may be due to chance (e.g.
93
knocking away an obstacle or lever by luck) or variation in factors such as strength,
94
dexterity, motivation or persistence (e.g. Overington et al. 2011; Benson-Amram and
95
Holekamp 2012; Thornton and Samson 2012). The studies we critique here certainly
96
provide important controls for some of these variables, but all suffer from shared
97
problems: contrary to standard practice in experimental psychology, they do not
98
target any defined cognitive process and make no theoretically grounded,
99
quantitative predictions as to what constitutes a correct outcome (see also Healy
100
2012). Consequently, there is no psychological basis for concluding that, for
101
instance, a bird that pulls a lever is more cognitively able than one that does not. We
102
suggest that future work should adopt lessons from psychometric research, where
103
tests are designed to target specific cognitive processes and test outcomes can be
104
clearly evaluated on the basis of correctness, giving quantitative scores of individual
105
performance. One obvious improvement to the current approach is to present
106
problem-solving tasks repeatedly to derive measures of individual instrumental
107
learning ability from rates of improvement over successive trials. Tasks incorporating
108
functionally relevant and irrelevant features could be particularly useful in assessing
109
not only differences in speed of solving (which may reflect strength or dexterity) but
110
also in the ability to learn the instrumental contingencies of a given task, and/or to
111
transfer learned rules from one task to others with similar features (Thornton &
112
Samson 2012). An additional approach is to design choice tests that make specific,
113
defined cognitive demands such that once a subject is engaged in a task it must rely
114
on its cognitive abilities to choose correctly between alternative options. Such an
115
approach forms the basis for numerous robust psychometric tests (see Plomin 2001;
116
Matzel et al. 2003) and has been applied in free-living animals. For instance,
117
animals’ choices in field experiments have been used to determine whether they
118
have remembered the location of food sources (Healy and Hurley 1995; Flores-
119
Abreu et al. 2012), learned to associate rewards with specific shapes or colours
120
(Boogert et al. 2010; Isden et al. 2013) or selected tools with appropriate functional
121
properties (Visalberghi et al. 2009). The challenge is now to harness similar
122
paradigms to quantify cognitive differences across large numbers of individuals, as is
123
widely done in psychometric studies of humans and other animals (Plomin 2001;
124
Matzel et al. 2003; Herrmann et al. 2010). Fortunately, recent technological
125
advances have great promise in enabling rigorous, standardised testing. For
126
instance, Nachev & Winter (2012) used computer-controlled artificial flowers to
127
quantify the ability of individual pit-tagged wild bats to discriminate between different
128
sugar concentrations. Similarly, building on their automated problem-solving tasks
129
(Morand-Ferron 2011; Cole et al. 2012), Morand-Ferron and colleagues are
130
developing instrumental conditioning choice tasks to record individual learning
131
curves for hundreds of wild pit-tagged birds over repeated visits (Morand-Ferron,
132
pers. comm.). These innovative applications of established psychological test
133
paradigms to wild animals provide a much-needed impetus to the development of a
134
robust science of wild psychometrics.
135
136
Having obtained robust measurements of individual cognitive differences, we must
137
then determine whether these are associated with differential fitness outcomes. The
138
relationship between cognitive performance and fitness may not be straightforward.
139
Neural and cognitive phenotypes are often highly plastic, so performance measures
140
may be sensitive to the conditions under which they are taken. Some birds, for
141
example, undergo seasonal changes in hippocampal volume, which is likely to affect
142
spatial memory (Smulders et al. 1995). “Problem-solving” measures in great tits are
143
also sensitive to testing conditions: individuals’ success in captivity did not predict
144
their success in the wild (Morand-Ferron et al. 2011). Where possible, we advocate
145
testing animals repeatedly to determine the consistency and plasticity of cognitive
146
traits. It is also important to note that investment in the neural tissue necessary for
147
improved cognitive processing may generate concomitant costs for other important
148
traits (Kawecki 2010; Kotrschal et al. 2013). We thus need measurements across a
149
range of fitness associated parameters to determine whether elevated cognitive
150
ability generates net benefits (Cole et al. 2012). Finally, we must consider whether
151
elevated cognitive performance could result in benefits accruing across several
152
behavioural contexts, or only in specific situations. Natural selection is often
153
assumed to fine-tune distinct cognitive mechanisms to meet specific ecological
154
challenges (Shettleworth 2010). However, some cognitive processes such as
155
associative learning are deployed across a broad range of contexts (Heyes 2012)
156
and human psychometric tests indicate that an individual’s performance across a
157
battery of disparate tasks can be summarised by a single measure of “general
158
intelligence” (g) (Plomin 2001; Boogert et al. 2011 and Thornton and Lukas 2012
159
review the equivocal evidence for g in animals). To help resolve long-standing
160
debates regarding modularity versus domain-generality and begin to pinpoint the
161
selective advantages of elevated cognitive ability, we advocate the use of batteries
162
of tests across different cognitive domains, coupled with measures of multiple fitness
163
parameters. Following psychometric conventions, test batteries should incorporate
164
tasks covering a range of pre-defined cognitive domains (e.g. association formation,
165
instrumental learning, transitive inference, inhibitory control, spatial memory), with
166
tasks designed to minimise effects of experience in one task on performance in
167
others (c.f. Matzel et al. 2003; Herrmann et al. 2010; Isden et al. 2013).
168
169
Finally, selection can only act on cognitive traits if they are heritable, an issue which
170
has attracted little attention beyond humans and a handful of laboratory studies on
171
other species (Plomin 2001; Kawecki 2010). Measures of heritability in wild
172
populations are likely to be complicated, with non-genetic factors including
173
developmental conditions (Buchanan et al. 2013) and past learning opportunities
174
(Thornton & Lukas 2012) affecting cognitive phenotypes. Nevertheless, there are
175
grounds for optimism that future studies will begin to disentangle ontogenetic and
176
heritable traits. In particular, there is much to be learned from the animal personality
177
literature, where techniques including cross fostering, experimental manipulation of
178
individual developmental trajectories, selection experiments and quantitative genetic
179
analyses are beginning to unravel genetic contributions to complex behavioural traits
180
(van Oers et al. 2005).
181
182
Individual-based studies offer huge potential to inform our understanding of cognitive
183
evolution. Laboratory research on human and non-human animals has generated
184
overwhelming evidence that individuals differ in their cognitive abilities (Boogert et al.
185
2011; Thornton and Lukas 2012). A suite of pioneering studies have recognised that
186
to understand how selection acts on cognitive traits we must now step out of the
187
laboratory and examine this variation and its fitness consequences under natural
188
conditions. However, the challenges of such an endeavour should not be
189
underestimated. Critically, if this exciting new avenue of research is to progress, we
190
must develop rigorous, psychologically-grounded experimental methods to identify
191
and quantify explicit cognitive processes and their heritability, ideally using batteries
192
of tests that call upon cognitive decision-making processes, spanning cognitive
193
domains. Only then can we begin to examine how these underlying cognitive
194
processes, either independently or in concert with others, can generate the
195
behaviour on which selection may act.
196
197
ACKNOWLEDGEMENTS
198
We are grateful to Neeltje Boogert for stimulating discussion and comments on
199
previous drafts of the paper. A.T. was supported by a BBSRC David Phillips
200
Fellowship (BB/H021817/1); J.R.M. was funded by a BBSRC Grant and J.I. was
201
funded by an EGF studentship from the University of Exeter.
202
203
REFERENCES
204
Balda RP, Kamil AC. 2002. Spatial and social cognition in corvids: An evolutionary
205
approach. In: Bekoff M, Allen C, Burghardt GM, editors. The Cognitive Animal.
206
Cambridge, MA: MIT Press. p. 129–134.
207
Benson-Amram S, Holekamp KE. 2012. Innovative problem solving by wild spotted
208
hyenas. Proc R Soc B. 279: 4087-4095
209
Boogert NJ, Monceau K, Lefebvre L. 2010. A field test of behavioural flexibility in
210
Zenadia doves (Zenaida aurita). Behav. Proces. 85: 135-141.
211
Boogert NJ, Anderson RC, Peters S, Searcy W A., Nowicki S. 2011. Song repertoire
212
size in male song sparrows correlates with detour reaching, but not with other
213
cognitive measures. Anim. Behav. 81: 1209-1216
214
Boogert NJ, Fawcett TW, Lefebvre L. 2011. Mate choice for cognitive traits: a review
215
of the evidence in nonhuman vertebrates. Behav Ecol. 22: 447–459.
216
Buchanan KL, Grindstaff JL, Pravosudov V V. 2013. Condition dependence,
217
developmental plasticity, and cognition: implications for ecology and evolution.
218
Trends Ecol Evol. 28: 290–296.
219
Cauchard L, Boogert NJ, Lefebvre L, Dubois F, Doligez B. 2013. Problem-solving
220
performance is correlated with reproductive success in a wild bird population. Anim
221
Behav. 85: 19–26.
222
Clutton-Brock T, Sheldon BC. 2010. Individuals and populations: the role of long-
223
term, individual-based studies of animals in ecology and evolutionary biology. Trends
224
Ecol Evol. 25: 562–273.
225
Cole EF, Morand-Ferron J, Hinks AE, Quinn JL. 2012. Cognitive ability influences
226
reproductive life history variation in the wild. Curr Biol. 22: 1808–1812.
227
Cole EF, Quinn JL. 2012. Personality and problem-solving performance explain
228
competitive ability in the wild. Proc R Soc B. 279: 1168–1175.
229
Flores-Abreu IN, Hurly TA, Healy SD. 2012. One-trial spatial learning: wild
230
hummingbirds relocate a reward after a single visit. Anim Cogn. 15:631–637.
231
Healy SD, Hurly TA. 1995. Spatial memory in rufous hummingbirds (Selasphorus
232
rufus): a field test. Anim Learn Behav. 23: 63-68.
233
Healy SD. 2012. Animal cognition: the trade-off to being smart. Curr Biol 22: R840–
234
841.
235
Herrmann E, Hernández-Lloreda MV, Call J, Hare B, Tomasello M. 2010. The
236
structure of individual differences in the cognitive abilities of children and
237
chimpanzees. Psychol Sci. 21: 102–110.
238
Heyes C. 2012. Simple minds: a qualified defence of associative learning. Phil Trans
239
R Soc B. 367: 2695–2703.
240
Isden J, Panayi C, Dingle C, Madden J. 2013. Performance in cognitive and
241
problem-solving tasks in male spotted bowerbirds does not correlate with mating
242
success. Anim Behav. 86: 829-838.
243
Kawecki TJ. 2010. Evolutionary ecology of learning: insights from fruit flies. Pop
244
Ecol. 52: 15–25.
245
Keagy J, Savard J-F, Borgia G. 2009. Male satin bowerbird problem-solving ability
246
predicts mating success. Anim Behav. 78: 809–817.
247
Keagy J, Savard J-F, Borgia G. 2011. Complex relationship between multiple
248
measures of cognitive ability and male mating success in satin bowerbirds,
249
Ptilonorhynchus violaceus. Anim Behav. 81: 1063–1070.
250
Kotrschal A, Rogell B, Bundsen A, Svensson B, Zajitschek S, Brännström I, Immler
251
S, Maklakov AA, Kolm N. 2013. Artificial selection on relative brain size in the guppy
252
reveals costs and benefits of evolving a larger brain. Curr Biol. 23: 168–171.
253
Matzel LD, Han YR, Grossman H, Karnik MS, Patel D, Scott N, Specht SM, Gandhi
254
CC. 2003. Individual differences in the expression of a “general” learning ability in
255
mice. J Neurosci. 23: 6423–6433.
256
Morand-Ferron J, Cole EF, Rawles JEC, Quinn JL. 2011. Who are the innovators? A
257
field experiment with 2 passerine species. Behav Ecol. 22: 1241–1248.
258
Nachev V, Winter Y. 2012. The psychophysics of uneconomical choice: non-linear
259
reward evaluation by a nectar feeder. Anim. Cogn. 15:393–400.
260
Overington SE, Cauchard L, Côté K-A, Lefebvre L. 2011. Innovative foraging
261
behaviour in birds: what characterizes an innovator? Behav Proc. 87: 274–285.
262
Plomin R. 2001. The genetics of g in human and mouse. Nature Rev Neurosci. 2:
263
136–41.
264
Pravosudov V V., Clayton NS. 2002. A test of the adaptive specialization hypothesis:
265
Population differences in caching, memory, and the hippocampus in black-capped
266
chickadees (Poecile atricapilla). Behav Neurosci. 116: 515–522.
267
Shettleworth SJ. 2010. Cognition, evolution and behaviour, 2nd Edition. Oxford:
268
Oxford University Press.
269
Smulders T V, Sassoon AD, DeVoogd TJ. 1995. Seasonal variation in hippocampal
270
volume in a food-storing bird, the black-capped chickadee. J Neurobiol. 27:15–25.
271
Thornton A, Lukas D. 2012. Individual variation in cognitive performance:
272
developmental and evolutionary perspectives. Phil Trans R Soc B. 367: 2773–2783.
273
Thornton A, Samson J. 2012. Innovative problem solving in wild meerkats. Anim
274
Behav. 83: 1459–1468.
275
Van Oers K, de Jong G, van Noordwijk, AJ, Kempenaers B, Drent PJ 2005.
276
Contribution of genetics to the study of animal personalities: a review of case
277
studies. Behav. 142: 1185-1206
278
Visalberghi E, Addessi E, Truppa V, Spagnoletti N, Ottoni E, Izar P, Fragaszy D.
279
2009. Selection of effective stone tools by wild bearded capuchin monkeys. Curr
280
Biol. 19: 213–217.