Corvids Outperform Pigeons and Primates in Learning a Basic Concept

Abstract

Corvids (birds of the family Corvidae) display intelligent behavior previously ascribed only to primates, but such feats are not directly comparable across species. To make direct species comparisons, we used a same/different task in the laboratory to assess abstract-concept learning in black-billed magpies ( Pica hudsonia). Concept learning was tested with novel pictures after training. Concept learning improved with training-set size, and test accuracy eventually matched training accuracy-full concept learning-with a 128-picture set; this magpie performance was equivalent to that of Clark's nutcrackers (a species of corvid) and monkeys (rhesus, capuchin) and better than that of pigeons. Even with an initial 8-item picture set, both corvid species showed partial concept learning, outperforming both monkeys and pigeons. Similar corvid performance refutes the hypothesis that nutcrackers' prolific cache-location memory accounts for their superior concept learning, because magpies rely less on caching. That corvids with "primitive" neural architectures evolved to equal primates in full concept learning and even to outperform them on the initial 8-item picture test is a testament to the shared (convergent) survival importance of abstract-concept learning.

Full text

https://doi.org/10.1177/0956797616685871
Psychological Science
2017, Vol. 28(4) 437 –444
© The Author(s) 2017
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0956797616685871
www.psychologicalscience.org/PS
Research Article
Birds are much maligned when it comes to intelligence,
as shown by such phrases as “bird brain” or “for the
birds.” Birds’ intelligence may get a bad rap from their
primitive appearance (e.g., beaks), apparent hard-wired
behaviors (e.g., songs), and small primitive brains (no
neocortex) thought to be capable of only lower-order
cognition. Because birds and mammals evolved along
different lineages beginning about 145 million years ago
(late Jurassic), they would be expected to look different.
But looks can be deceiving when it comes to intelligence.
And, as we argue here, neither their small brain size, nor
their hard-wired behaviors, nor their lack of neocortex
has prevented birds from evolving highly intelligent
behavior.
Among the many intelligent bird behaviors docu-
mented in the past 20 or so years include remarkable
feats, particularly by members of the corvid family: tool
making and tool use by New Caledonia crows (Corvus
moneduloides) in the wild and the laboratory; identifying
“self” in a mirror (Pica pica and Nucifraga columbiana);
recaching to prevent pilfering by an observing raven
(Corvus corax), but only if the caching raven had itself
been a pilferer; and nutcrackers storing thousands of pine
seeds in hundreds of caches that are recovered months
later in the alpine forest (e.g., Bugnyar & Heinrich, 2006;
Clary & Kelly, 2016; Hunt, 1996; Prior, Schwarz, &
Güntürkün, 1998; Vander Wall, 1982; Weir, Chappell, &
Kacelnik, 2002). Because tool making and tool use were
critical in developing modern civilization, it has been
assumed that such inventive behavior was evidence of supe-
rior human intelligence (compared with other animals).
685871PSSXXX10.1177/0956797616685871Wright et al.Abstract-Concept Learning and Brain Evolution
research-article2017
Corresponding Author:
Anthony A. Wright, McGovern Medical School, University of Texas
Health Science Center at Houston, 6431 Fannin St., Houston, TX 77030
E-mail: anthony.a.wright@uth.tmc.edu
Corvids Outperform Pigeons and
Primates in Learning a Basic Concept
Anthony A. Wright1, John F. Magnotti2, Jeffrey S. Katz3,
Kevin Leonard4, Alizée Vernouillet4, and Debbie M. Kelly4
1Department of Neurobiology and Anatomy, McGovern Medical School, University of Texas
Health Science Center at Houston; 2Department of Neurosurgery, Baylor College of Medicine;
3Department of Psychology, Auburn University; and 4Department of Psychology, University of Manitoba
Abstract
Corvids (birds of the family Corvidae) display intelligent behavior previously ascribed only to primates, but such
feats are not directly comparable across species. To make direct species comparisons, we used a same/different task
in the laboratory to assess abstract-concept learning in black-billed magpies (Pica hudsonia). Concept learning was
tested with novel pictures after training. Concept learning improved with training-set size, and test accuracy eventually
matched training accuracy—full concept learning—with a 128-picture set; this magpie performance was equivalent to
that of Clark’s nutcrackers (a species of corvid) and monkeys (rhesus, capuchin) and better than that of pigeons. Even
with an initial 8-item picture set, both corvid species showed partial concept learning, outperforming both monkeys
and pigeons. Similar corvid performance refutes the hypothesis that nutcrackers’ prolific cache-location memory
accounts for their superior concept learning, because magpies rely less on caching. That corvids with “primitive” neural
architectures evolved to equal primates in full concept learning and even to outperform them on the initial 8-item
picture test is a testament to the shared (convergent) survival importance of abstract-concept learning.
Keywords
comparative intelligence, abstract-concept learning, same/different learning, evolution, magpies, corvids, nutcrackers,
primates
Received 9/8/16; Revision accepted 12/3/16

438 Wright et al.
The same can be said for birds identifying their own
reflection as “self,” attributing intentions to a conspecific
on the basis of their own pilfering behavior (theory of
mind), and superior long-term memory for hundreds of
caches, months later when covered by snow that obscures
local landmarks.
Even from these few examples, it should be clear that
at least some birds have considerable intelligence, as well
as higher-order cognition. But as more examples accu-
mulate, the comparison of intelligent behavior across
bird species and with other animals (e.g., nonhuman pri-
mates) becomes more pressing. Many of these remark-
able feats of intelligence are niche-specific and likely
evolved over many thousands of years before becoming
what we observe today, making comparisons of intelli-
gence across ecological niches difficult, if not impossible.
For example, how would the feat of fashioning a twig
into a hooked tool to fish grubs out of tree holes be com-
pared with retrieving caches of pine seeds made months
before that are now covered by snow?
Such individual feats of intelligence are not, for the
most part, directly comparable, nor can they be assigned a
level of cognitive processing for meaningful comparisons.
Instead, what is needed is a task that all species to be com-
pared can readily perform and that requires higher-order
abstraction that can be manipulated in a functional man-
ner to test for an optimal solution. The most basic of such
abstract tasks is the same/different task, proposed in 1890
by William James as “the very keel and backbone of our
thinking” and which has been shown to be functionally
important in human development for equivalence opera-
tions involved in novel sentence construction and mathe-
matical operations (James, 1890/1950, p. 459; see also
Christie, Gentner, Call, & Haun, 2016; Gentner, 1988;
Siegler, 1996; Smith, Langston, & Nisbett, 1992).
We developed a same/different task that different bird
species including pigeons (Columba livia) and Clark’s
nutcrackers (Nucifraga columbiana), a corvid species, as
well as different nonhuman primate species including
rhesus monkeys (Macaca mulatta) and capuchin mon-
keys (Cebus apella) could all learn readily and perform to
similar levels of high accuracy (Katz & Wright, 2006; Katz,
Wright, & Bachevalier, 2002; Magnotti, Katz, Wright, &
Kelly, 2015; Wright & Katz, 2006; Wright, Magnotti, Katz,
Leonard, & Kelly, 2016; Wright, Rivera, Katz, & Bachevalier,
2003). Critically, we used the same pairs of pictures (scenes,
objects, animals, buildings, people, etc.) in the same order
of testing, the same sequences, and the same relative
picture size (visual angle) for training and abstract-
concept testing across all these species. When trained
with a small set of 8 pictures, all of these species learned
to accurately (> 80% correct) identify pairs as the same or
different. After learning, we assessed abstract-concept
learning by testing for transfer on occasional trials con-
taining novel pictures; these trials were intermixed with
training trials containing familiar pictures. The training
set was then progressively doubled, and learning was fol-
lowed by transfer testing. The progressive increase in
training-set size was the critical manipulation that
increased abstract-concept learning. Doubling the train-
ing set accelerated expansion of examples of the con-
cept, which in turn promoted learning of the abstract
concept. All species and all individuals tested attained
full abstract-concept learning as shown by equivalence
of transfer performance (with pairs of novel pictures) to
their respective baseline performance (with pairs of pic-
tures used in training)—that is, they achieved the optimal
solution referred to earlier.
Other evidence directly related to our same/different
experiments contributed to understanding the nature of
our subjects’ relational processing and substantiated that
this task assessed bona fide same/different abstract-
concept learning: First, our task used only two stimuli per
trial to avoid low-level perceptual entropy cues common
with tasks using multiple-stimulus displays (e.g., Young,
Wasserman, & Garner, 1997). Second, our modeling of
faster learning with increasing set size ruled out stimulus
generalization as a contributor to abstract-concept learn-
ing (Wright & Katz, 2007). Third, our subjects’ rapid tran-
sition to delays (1–2 s) ruled out possible effects of
low-level translational-symmetry cues on same trials, tri-
als with two identical pictures, and entropy cues on dif-
ferent trials, trials with two different pictures (Katz, Sturz,
& Wright, 2010; Wright et al., 2003). Fourth, our pro-
active-interference tests using the delayed (1 s, 10 s, or 20
s) same/different task verified that our subjects were
making bona fide relational stimulus comparisons across
as many as eight trials and 5 min before testing (Devkar
& Wright, 2016; Wright, Katz, & Ma, 2012).1
In addition to the finding of full concept learning com-
mon to all species, there were, however, some fundamen-
tal species’ differences in same/different abstract learning:
Corvids, including nutcrackers, and in a more recent study
black-billed magpies (Pica hudsonia; Magnotti, Wright,
Leonard, Katz, & Kelly, 2016), revealed better transfer
(66–67% correct) with the initial eight-picture set than did
pigeons, rhesus monkeys, or capuchin monkeys, all of
which performed at chance level (50% correct). In addi-
tion, nutcrackers required fewer exemplars to attain full-
concept learning (128-picture training set) than pigeons
(256-picture training set), and their performance was
equivalent to that of rhesus monkeys and capuchin mon-
keys in this regard.
We think it is remarkable that nutcrackers (or any bird
species) could attain full concept learning with the same
number of training exemplars as two monkey species in

Abstract-Concept Learning and Brain Evolution 439
a basic relational same/different task. This advanced pro-
cessing may be facilitated by the nutcracker’s predisposi-
tion for relational processing and memory of the many
cache locations over several months of retention time.
Cache location memory depends on relational process-
ing of the so-called what (pine seeds) and where (loca-
tion in forest) for literally thousands of cache locations
that change from year to year. Changing cache locations
yearly means that nutcrackers must not confuse locations
made in prior years with those of the current year, which
therefore adds a “when” component to the relational pro-
cessing. The what, where, and when are closely related
to episodic memory (cf. Gould, Ort, & Kamil, 2012), and
primates have been thought to be best equipped in terms
of brain development for having episodic memory (e.g.,
Beran et al., 2016; Clark & Squire, 2013; Davachi, 2006;
Ranganath, 2010; Squire, Wixted, & Clark, 2007). More-
over, nutcrackers do not have the luxury of gradually
learning these relationships to retrieve cache locations;
they have to rapidly acquire this skill to a high accuracy
level because they are totally dependent on cached pine
nuts during long alpine winters (Tomback, 1998).
To test whether nutcrackers are unique among corvids
in rapid full concept learning because of their evolved
dependence on extensive caching and retrieval, we tested
another corvid species, black-billed magpies, using the
same expanding training sets in the same task previously
used with nutcrackers (and monkeys and pigeons). Mag-
pies rely much less on cached foods than nutcrackers do
(e.g., Trost, 1999). They are more omnivorous and have
better access to their preferred foods (e.g., insects, car-
rion, eggs, berries, seeds, nuts) throughout the year. If the
nutcrackers’ extensive caching and retrieval was instru-
mental in their rapid full concept learning, then the mag-
pies’ full concept learning might turn out to be similar
to that of pigeons. Alternatively, some other relational-
processing skill (e.g., a social skill) highly developed in
magpies might be instrumental in magpies outperform-
ing nutcrackers (and monkeys). Finally, if skill in rela-
tional processing was built into the evolved neural
architecture of these two corvid species, then the mag-
pies’ full concept learning should be equivalent to that
of nutcrackers (and monkeys).
Method
The subjects were 10 wild-caught black-billed magpies (4
females) that were hand-raised from the prefledgling
stage and maintained at approximately 85% to 90% of
their ad libitum weight by supplemental feeding with a
mixture of Pedigree wet dog food, Kirkland Signature dry
dog food, mixed fruits and vegetables, and a vitamin sup-
plement on completion of daily sessions. They were
housed individually in cages (86.4 cm high × 101 cm
wide × 76.5 cm deep) in a colony room maintained at a
constant temperature of 22 °C. The room was on a 12-hr
light-dark cycle with light onset at 7:00 a.m. The magpies
were tested in wooden chambers (61 cm wide × 31 cm
deep × 56 cm high), similar to those used to test other
avian species in previous research (Katz & Wright, 2006;
Magnotti, Katz, Wright, & Kelly, 2015; Wright & Katz,
2006; Wright, Magnotti, Katz, Leonard, & Kelly, 2016).
Stimuli were displayed on an LCD monitor and were vis-
ible through cutouts in a clear acrylic template that was
33 cm wide by 26 cm high. A house light (24-V, 0.04-A
bulb; Eiko, Shawnee, KS) was centered in the subject’s
portion of the chamber ceiling.
Stimuli were distinctive color “travel-slide” pictures,
the same training and transfer pictures that have been
used to test monkeys, pigeons, and Clark’s nutcrackers
(for full color displays of these training and testing stimuli
for set sizes 8, 16, 32, 64, 128, and 256, see Wright & Katz,
2006). The stimuli were sized so that the total display
(sample and comparison pictures and white rectangle)
was matched in visual angle to those used in our previ-
ous work with pigeons and nonhuman primates, approx-
imately 69° vertically and 73° horizontally as viewed from
a perch (14.5 cm from the screen, 15.9 cm from floor);
the center of the display set at the mean height of the
magpies’ eyes, 30.5 cm from the chamber floor.
All procedures were approved by the University of
Manitoba’s Animal Care Committee in accordance with
the Canadian Council on Animal Care.
Training and testing sessions were conducted 5 to 7
days a week during the light phase of the light-dark
cycle. Trials began with the presentation of a sample pic-
ture. After 20 responses to the sample picture (trained by
systematically increasing the number of responses from 1
to 20), the comparison picture was presented beneath
the sample picture, along with a white rectangle to the
right of the comparison picture (see Fig. 1a). A peck to
the comparison picture was correct when it matched the
sample picture. A peck to the white rectangle to the right
of the comparison picture was correct when the compari-
son picture did not match the sample picture. Correct
choice responses were reinforced with mealworms deliv-
ered below the monitor via a rotating wheel. Trials were
separated by a 15-s intertrial interval with the house light
on.
The magpies were trained in 100-trial sessions (50
same trials and 50 different trials) until they achieved
a performance criterion of 85% (or greater) accuracy,
with a minimum of three training sessions. Immediate-
ly following acquisition, abstract-concept learning was
as-sessed in six consecutive transfer sessions, each con-
taining 90 baseline (training) trials composed of famil-
iar training stimuli and intermixed with 10 transfer
trials with novel stimuli (5 same trials, 5 different trials).

440 Wright et al.
Same Different
40
50
60
70
80
90
100
83264 128 256 512 1,024
40
50
60
70
80
90
100
83264128 256512 1,024
40
50
60
70
80
90
100
83264128 256512 1,024
Trial Types
Magpie Set-Size Function
Avian Comparison
Percentage Correct
Percentage Correct
Percentage Correct
Magpie-Primate Comparison
Training-Set Size
Training-Set SizeTraining-Set Size
ab
dc
Transfer
Baseline
Magpie Transfer
Magpie Baseline
Nutcracker Transfer
Nutcracker Baseline
Pigeon Transfer
Pigeon Baseline
Magpie Transfer
Magpie Baseline
Rhesus Transfer
Rhesus Baseline
Capuchin Transfer
Capuchin Baseline
Fig. 1. Examples of the stimuli used in the present study and comparison of results with those of previous studies discussed in the text. In
both same trials and different trials (a), two images were presented on a black background next to a white rectangle. The birds were trained
to peck the lower picture on same trials and the white rectangle on different trials. The graph in (b) shows the mean percentage of correct
responses from 10 magpies as a function of training-set size, separately for training (baseline) trials and novel (transfer) trials. The graphs in (c)
and (d) show the results for the magpies along with the results for other bird species and primates, respectively, for purposes of comparison.
Dashed colored lines indicate mean baseline, and solid colored lines indicate transfer. Dashed gray lines indicate chance-level performance.
Error bars represent ±1 SEM.

Abstract-Concept Learning and Brain Evolution 441
Performance on the 60 transfer trials (10 per session)
was the measure of abstract-concept learning. These
transfer trials appeared within the context of ongoing
baseline trials, which were included to maintain accu-
rate performance in the task while transfer and concept
learning were being assessed. Transfer stimuli were
never repeated within or across transfer sessions to
ensure novelty. Choice responses were reinforced iden-
tically on transfer and baseline trials to prevent nonrein-
forcement (i.e., extinction) from becoming associated
with the appearance of novel stimuli.
The first set-size expansion consisted of adding 8 new
training stimuli to the original 8-item picture training set
and training the magpies until a performance accuracy of
at least 85% correct was obtained (at least two sessions of
this first set-size expansion had a correction procedure;
for details, see Wright et al., 2016). The progressive cycle
of expanding (doubling) the set size, training for three
(or more) sessions, obtaining 85% correct or better, and
novel-stimulus transfer testing was repeated for six addi-
tional doublings of the training set (32, 64, 128, 256, 512,
and 1,024 set sizes).
Results
Figure 1b shows the mean results for the 10 magpies for
the same task used to test nutcrackers, pigeons, and
monkeys. Baseline performance on training trials was
maintained at a high level (≥ 85% correct) throughout
set-size expansion. Abstract-concept learning, as mea-
sured by transfer of learning to novel picture pairs (67%
correct), was above chance performance (50% correct)
after initial learning with the initial 8-item training set and
increased regularly and monotonically with the training-
set-size expansions until transfer performance was indis-
tinguishable from baseline. To assess group-level transfer
performance, we performed a two-way repeated mea-
sures analysis of variance (ANOVA) on accuracy across
trial type (baseline or transfer) and set size (8, 32, 64, 128,
256, 512, or 1,024), which yielded a significant Trial Type ×
Set Size interaction, F(6, 54) = 19.4, p = 10−11, generalized
η2 = .30, as well as main effects of trial type, F(1, 9) =
42.9, p = .0001, generalized η2 = .15, and set size, F(6, 54) =
19.5, p = 10−11, generalized η2 = .38. The interaction was
caused by the difference between baseline and transfer
performance at early training-set sizes, but not at later
training-set sizes. Comparing baseline and transfer per-
formance within each set size using paired t tests showed
significant differences for set sizes of 8, 32, and 64 items,
all t(9)s > 2.9, ps < .02, but not at set sizes of 128, 256,
512, and 1,024 items, all t(9)s < 1.21, ps > .25.
Figure 1c shows performance of the magpies com-
pared with pigeons and nutcrackers previously tested in
this same task and discussed previously. The magpies
and nutcrackers were identical in their substantial trans-
fer after training with the initial 8-item picture set. Both of
these corvid species clearly outperformed pigeons. Com-
paring transfer accuracy of just the corvids, a two-way
repeated measures ANOVA with factors of set size and
species (nutcracker or magpie) yielded a significant effect
of set size, F(6, 90) = 37.3, p = 10−22, generalized η2 =
.58, but no main effect of species, F(1, 15) = 2.49, p = .14,
generalized η2 = .07, and no significant interaction, F(6,
90) = 0.6, p = .75, generalized η2 = .02. A similar ANOVA
comparing the performance of magpies and pigeons,
however, yielded significant main effects of set size, F(6,
72) = 46.0, p = 10−22, generalized η2 = .67, and species,
F(1, 12) = 22.5, p = .0005, generalized η2 = .47, and a
significant Set Size × Species interaction, F(6, 72) = 3.1,
p = .01, generalized η2 = .12. The interaction was caused
by an initial difference in transfer accuracy between mag-
pies and pigeons that gradually diminished as training-
set size increased.
Figure 1d shows performance of the magpies com-
pared with the primates’ performance discussed previ-
ously. The magpies outperformed the monkeys in their
transfer (partial concept learning) immediately following
training on the initial 8-item picture set. Because of the
small sample sizes for both monkey species, we com-
bined data from the rhesus and capuchin monkeys to
compare with the data from magpies. A two-way repeated
measures ANOVA on transfer accuracy with factors of set
size (8, 32, 64, or 128) and species (monkey or magpie)
yielded a significant Set Size × Species interaction, F(3,
42) = 3.8, p = .017, generalized η2 = .11, and a main effect
of set size, F(3, 42) = 71.0, p = 10−15, generalized η2 = .69,
but no main effect of species, F(1, 14) = 3.0, p = .11, gen-
eralized η2 = .11.
For direct comparisons of corvids and monkeys, we
combined the magpies and nutcrackers into one group
(n = 17) and compared their performance with the per-
formance of the two monkey species combined. A two-
way repeated measures ANOVA on transfer accuracy with
group (corvid or monkey) and set size (8, 32, 64, or 128)
as factors yielded a significant interaction between group
and set size, F(3, 63) = 5.0, p = .004, generalized η2 =
.09, and a significant main effect of set size, F(3, 63) =
85.2, p = 10−21, generalized η2 = .62, but no main effect of
group, F(1, 21) = 1.5, p = .24, generalized η2 = .04.
Other comparisons across these species included
learning rates, to determine whether they might be cor-
related with (and influence) transfer and concept learn-
ing. It is noteworthy that there were no substantial
differences in learning rate on the initial acquisition:
Magpies, on average, needed 3,500 trials (thirty-five 100-
trial sessions) to reach the performance criterion with the
initial set of 8 items; this learning rate was similar to the
average for the other species (nutcrackers: 3,300; rhesus

442 Wright et al.
monkeys: 4,000; capuchin monkeys: 3,500; pigeons:
3,000). As with the other four species tested, the magpies’
learning rates declined with increasing set sizes (the min-
imum number of trials was 300): 8 items: 3,500 trials; 16
items: 570 trials; 32 items: 480 trials; 64 items: 340 trials;
128 items: 320 trials; 256 items: 330 trials; 512 items: 340
trials; and 1,024 items: 300 trials (for the set size of 256
items, the mean rate was calculated for 9 birds because
of a computer-hardware glitch with 1 bird).
Taken together, these comparisons substantiate that
(a) mean training-set size required for the magpies to
attain full concept learning was equivalent to that
required by monkeys (and nutcrackers) but was smaller
than that required by pigeons, (b) the magpies (and nut-
crackers) outperformed the monkeys and pigeons on ini-
tial transfer (partial concept learning) with the small
8-item picture training set, and (c) the lack of learning-
rate differences suggest that learning rates were not a
factor in determining the abstract-concept learning differ-
ences noted earlier in the paragraph.
Discussion
It appears that the magpies’ neural apparatus and predis-
position for relational and abstract-concept learning were
not hampered by their lesser predilection for caching
food compared with nutcrackers. Moreover, the results
from these two corvid species point to the possibility that
corvids generally might be able to fully learn a higher-
order abstract concept following exposure to a similar
number of concept exemplars (128-picture-set training)
as either Old World (rhesus monkeys) or New World
(capuchin monkeys) nonhuman primates. Such a conclu-
sion has startling implications. The modern lineages of
birds and mammals evolved from survivors of a cata-
strophic asteroid event (Cretaceous-Paleogene extinction
event) that wiped out all of the world’s big land animals
(e.g., big land-living dinosaurs) some 66 million years
ago (e.g., Alvarez, Alvarez, Asaro, & Michel, 1980; Schulte,
2010). Some small burrowing land animals survived, such
as small feathered dinosaurs and small furry mammals
(e.g., monotremes, marsupials, and placentals). Body
architectures, including brains, were and are very differ-
ent for birds and mammals. Mammals, particularly pri-
mates, evolved large brains (compared with body weight)
with folded neocortex, including the prefrontal cortex,
and elaborate temporal-lobe structures (e.g., hippocam-
pus plus adjoining parahippocampal cortex), key struc-
tures for primate relational processing, abstract-concept
learning, and episodic memory.
The “bird brain,” by contrast, has until recently been
thought to be primitive. Because most birds fly, light
weight is required (e.g., they have hollow and trussed
bones and no bladder). Nevertheless, many bird brains
(particularly corvid brains) are substantial in size and
weight compared with the birds’ body weight. Birds do
have a well-developed hippocampus (e.g., Gould et al.,
2013), but they have no six-layer neocortex, as do pri-
mates; birds have nodal or nuclear structures that may
have some advantages in shorter connectivity and speed
(e.g., Clayton & Emery, 2015). Many functions of the
mammalian prefrontal cortex have been found in birds’
brain structure called the caudolateral nidopallium (e.g.,
Emery, 2006; Güntürkün, 2005; Kirsch, Güntürkün, &
Rose, 2008), a brain structure with tightly packed, high-
density neurons (Olkowicz et al., 2016).
So, how did the apparently primitive bird brain that
evolved from dinosaurs become competitive with, and
even initially outperform, the abilities of what has been
considered a more elaborate primate brain to perform
abstract-concept learning, which involves thoughts and
processes considered to be of the highest cognitive order?
The answer most certainly lies in evolution itself, a multi-
million-year process. Environmental pressures (social and
otherwise) undoubtedly selected for and shaped these dif-
ferent neural architectures to successfully accomplish
many of the same essential and intelligent behaviors for
survival, an example of convergent evolution in which
organisms not closely related (i.e., not monophyletic)
independently evolved similar traits or functions as a result
of having to adapt to similar environments or ecological
niches. But the example of convergent evolution presented
in the current study is comparatively novel and unique
because its identification required special tests of the cog-
nitive ability (trait) for the cognitive function of fully learn-
ing a same/different abstract concept to be revealed. Other
examples of convergent evolution have been based on
some obvious physical trait, such as wings, which typically
can be identified from fossil records and have an obvious
function of flying (some insects, birds, and bats).
Action Editor
Kathleen McDermott served as action editor for this article.
Author Contributions
A. A. Wright, J. F. Magnotti, J. S. Katz, and D. M. Kelly contrib-
uted to the study concept and design. D. M. Kelly, A. Vernouillet,
and K. Leonard collected the data. J. F. Magnotti analyzed the
data and prepared the figures in collaboration with A. A. Wright.
A. A. Wright drafted the manuscript. All the authors have read
the manuscript, provided revisions of the manuscript, and
approved the final version for submission. All the authors are
responsible for the integrity of the research.
Declaration of Conflicting Interests
The authors declared that they had no conflicts of interest with
respect to their authorship or the publication of this article.

Abstract-Concept Learning and Brain Evolution 443
Funding
This research was supported by Natural Sciences and Engineer-
ing Research Council of Canada Grant RGPIN/312379-2009 (to
D. M. Kelly).
Note
1. Although imprinted same/different behavior in newly hatched
ducklings is intriguing (Martinho & Kacelnik, 2016), it is unclear
to us how it might relate to the same/different abstract-concept
learning demonstrated and discussed here and supported by
converging evidence that our subjects were learning a bona
fide abstract concept.
References
Alvarez, L. W., Alvarez, W., Asaro, F., & Michel, H. V. (1980).
Extraterrestrial cause for the Cretaceous–Tertiary extinction.
Science, 208, 1095–1108. doi:10.1126/science.208.4448.1095
Beran, M. J., Menzel, C. R., Parrish, A. E., Perdue, B. M., Sayers,
K., Smith, J. D., & Washburn, D. A. (2016). Primate cogni-
tion: Attention, episodic memory, prospective memory, self-
control, and metacognition as examples of cognitive control
in nonhuman primates. Wiley Interdisciplinary Reviews:
Cognitive Science, 7, 294–316. doi:10.1002/wcs.1397
Bugnyar, T., & Heinrich, B. (2006). Pilfering ravens, Corvus
corax, adjust their behaviour to social context and identity
of competitors. Animal Cognition, 9, 369–376.
Christie, S., Gentner, D., Call, J., & Haun, D. B. (2016). Sensitivity to
relational similarity and object similarity in apes and children.
Current Biology, 26, 531–535. doi:10.1016/j.cub.2015.12.054
Clark, R. E., & Squire, L. R. (2013). Similarity in form and func-
tion of the hippocampus in rodents, monkeys, and humans.
Proceedings of the National Academy of Sciences, USA,
110(Suppl. 2), 10365–10370. doi:10.1073/pnas.1301225110
Clary, D., & Kelly, D. M. (2016). Graded mirror self-recognition
by Clark’s nutcrackers. Scientific Reports, 6, Article 36459.
doi:10.1038/srep36459
Clayton, N. S., & Emery, N. J. (2015). Avian models for human
cognitive neuroscience: A proposal. Neuron, 86, 1330–
1342. doi:10.1016/j.neuron.2015.04.024
Davachi, L. (2006). Item, context and relational encoding in
humans. Current Opinion in Neurobiology, 16, 693–700.
Devkar, D. T., & Wright, A. A. (2016). Event-based proactive
interference by rhesus monkeys. Psychonomic Bulletin &
Review, 23, 1474–1482. doi:10.3758/s13423-016-1005-x
Emery, N. J. (2006). Cognitive ornithology: The evolution of
avian intelligence. Philosophical Transactions of the Royal
Society B: Biological Sciences, 361, 23–43. doi:10.1098/
rstb.2005.1736
Gentner, D. (1988). Metaphor as structure mapping: The
relational shift. Child Development, 59, 47–59. doi:10.2307/
1130388
Gould, K. L., Gilbertson, K. E., Hrvol, A. J., Nelson, J. C., Seyfer,
A. L., Brantner, R. M., & Kamil, A. C. (2013). Differences in
relative hippocampus volume and number of hippocampus
neurons among five corvid species. Brain, Behavior, and
Evolution, 81, 56–70. doi:10.1159/000345560
Gould, K. L., Ort, A. J., & Kamil, A. C. (2012). Do Clark’s nut-
crackers demonstrate what-where-when memory on a
cache-recovery task? Animal Cognition, 15, 37–44.
Güntürkün, O. (2005). The avian ‘prefrontal cortex’ and cogni-
tion. Current Opinion in Neurobiology, 15, 686–693.
Hunt, G. R. (1996). The crafting of hook tools by wild New
Caledonian crows. Nature, 379, 249–251.
James, W. (1950). The principles of psychology. New York, NY:
Dover Publications. (Original work published 1890)
Katz, J. S., Sturz, B. R., & Wright, A. A. (2010). Testing the trans-
lational-symmetry hypothesis of abstract-concept learning
in pigeons. Learning & Behavior, 38, 35–41.
Katz, J. S., & Wright, A. A. (2006). Same/different concept
learning by pigeons. Journal of Experimental Psychology:
Animal Behavior Processes, 32, 80–86.
Katz, J. S., Wright, A. A., & Bachevalier, J. (2002). Mechanisms
of same/different abstract-concept learning by rhesus
monkeys (Macaca mulatta). Journal of Experimental
Psychology: Animal Behavior Processes, 28, 358–368.
Kirsch, J. A., Güntürkün, O., & Rose, J. (2008). Insight without
cortex: Lessons from the avian brain. Consciousness and
Cognition, 17, 475–483.
Magnotti, J. F., Katz, J. S., Wright, A. A., & Kelly, D. M. (2015).
Superior abstract-concept learning by Clark’s nutcrackers
(Nucifraga columbiana). Biology Letters, 11(5), Article
20150148. doi:10.1098/rsbl.2015.0148
Magnotti, J. F., Wright, A. A., Leonard, K., Katz, J. S., & Kelly,
D. M. (2016). Abstract-concept learning in Black-billed
magpies (Pica hudsonia). Psychonomic Bulletin & Review.
Advance online publication. doi:10.3758/s13423-016-1132-4
Martinho, A., & Kacelnik, A. (2016). Ducklings imprint on the rela-
tional concept of “same or different.” Science, 353, 286–288.
Olkowicz, S., Kocourek, M., Lucˇan, R. K., Porteš, M., Fitch,
W. T., Herculano-Houzel, S., & Neˇ mec, P. (2016). Birds
have primate-like numbers of neurons in the forebrain.
Proceedings of the National Academy of Sciences, USA, 113,
7255–7260.
Prior, H., Schwarz, A., & Güntürkün, O. (2008). Mirror-induced
behavior in the magpie (Pica pica): Evidence of self-recog-
nition. PLoS Biology, 6(8), Article e202. doi:10.1371/journal
.pbio.0060202
Ranganath, C. (2010). Binding items and contexts: The cogni-
tive neuroscience of episodic memory. Current Directions
in Psychological Science, 19, 131–137.
Schulte, P. (2010). The Chicxulub asteroid impact and mass
extinction at the Cretaceous-Paleogene boundary. Science,
327, 1214–1218.
Siegler, R. S. (1996). Emerging minds: The process of change in
children’s thinking. New York, NY: Oxford University Press.
Smith, E. E., Langston, C., & Nisbett, R. E. (1992). The case for
rules in reasoning. Cognitive Science, 16, 1–40.
Squire, L. R., Wixted, J. T., & Clark, R. E. (2007). Recognition
memory and the medial temporal lobe: A new perspective.
Nature Reviews Neuroscience, 8, 872–883.
Tomback, D. F. (1998). Clark’s nutcracker (Nucifraga columbi-
ana). In The birds of North America online. Retrieved from
https://birdsna.org/Species-Account/bna/species/clanut

444 Wright et al.
Trost, C. H. (1999). Black-billed magpie (Pica hudsonia). In
The birds of North America online. Retrieved from https://
birdsna.org/Species-Account/bna/species/bkbmag1
Vander Wall, S. B. (1982). An experimental analysis of cache
recovery in Clark’s nutcracker. Animal Behaviour, 30, 84–94.
Weir, A. A. S., Chappell, J., & Kacelnik, A. (2002). Shaping
of hooks in New Caledonian crows. Science, 297, 981.
doi:10.1126/science.1073433
Wright, A. A., & Katz, J. S. (2006). Mechanisms of same/differ-
ent concept learning in primates and avians. Behavioural
Processes, 72, 234–254.
Wright, A. A., & Katz, J. S. (2007). The generalization hypothesis
of abstract-concept learning: Learning strategies and related
issues in Macaca mulatta, Cebus apella, and Columba livia.
Journal of Comparative Psychology, 121, 387–397.
Wright, A. A., Katz, J. S., & Ma, W. (2012). How to be proactive
about interference: Lessons from animal memory. Psycholog-
ical Science, 23, 453–458. doi:10.1177/0956797611430096
Wright, A. A., Magnotti, J. F., Katz, J. S., Leonard, K., & Kelly,
D. M. (2016). Concept learning set-size functions for
Clark’s nutcrackers. Journal of the Experimental Analysis of
Behavior, 105, 76–84. doi:10.1002/jeab.174
Wright, A. A., Rivera, J. J., Katz, J. S., & Bachevalier, J. (2003).
Abstract-concept learning and list-memory processing by
capuchin and rhesus monkeys. Journal of Experimental
Psychology: Animal Behavior Processes, 29, 184–198.
Young, M. E., Wasserman, E. A., & Garner, K. L. (1997). Effects
of number of items on the pigeon’s discrimination of same
from different visual displays. Journal of Experimental
Psychology: Animal Behavior Processes, 23, 491–501.