Evolutionary Theory in Cognitive Neuroscience: A 20-Year Quantitative Review of Publication Trends

Evolutionary Psychology
www.epjournal.net – 2007. 5(3): 520-530
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Original Article
Evolutionary Theory in Cognitive Neuroscience: A 20-Year Quantitative
Review of Publication Trends

Gregory D. Webster, Department of Psychology, University of Illinois at Urbana-Champaign, Champaign, IL
61820, USA. Email: webster3@uiuc.edu (Corresponding author)
Abstract: Evolutionary cognitive neuroscience is an emerging and promising new
scientific field that combines the meta-theoretical strengths of an evolutionary perspective
with the methodological rigor of neuroscience. The purpose of the present research was to
quantify and test evolution’s influence in neuroscience and cognitive neuroscience journals
over time (1987-2006). In Study 1, analyses from a convenience sample of 10 neuroscience
journals revealed that the proportion of neuroscience articles mentioning evolution grew
significantly over the last 20 years. Moreover, beginning as early as 1990, the average
proportion of neuroscience articles mentioning evolution was significantly different from
zero. These effects were not moderated by between-journals differences in impact factor (a
citation rate index), suggesting that the observed growth was fairly consistent across
journals. In Study 2, analyses from a convenience sample of 4 cognitive neuroscience
journals revealed that the proportion of cognitive neuroscience articles mentioning
evolution neither differed from zero nor grew significantly over time (1987-2006);
however, the change-over-time effect size was large. Compared to other research areas,
evolution’s penetration into cognitive neuroscience articles grew faster than anthropology,
economics, and sociology, but not psychology. The implications of evolutionary
psychology’s increasing role in science in general, and in cognitive neuroscience in
particular, are discussed.

Keywords: evolutionary cognitive neuroscience, evolutionary psychology, history of
science, multilevel modeling, publication trends.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Introduction

Whether it is viewed as emerging interdisciplinary science, and integrative
theoretical perspective, or a broad sub-discipline of psychology (Cornwell, Palmer,
Guinther, and Davis, 2005), one thing is fairly certain about evolutionary psychology: Its
influence is growing. In loose evolutionary terms, one could argue that it is currently
undergoing the scientific equivalent of adaptive radiation. Evolutionary psychology is not
only strengthening its roots in some of its more traditional territory (e.g., social

---

Evolutionary Theory in Cognitive Neuroscience
psychology; Webster, 2007b), but it is also branching out and adapting to new scientific
terrain such as cognitive neuroscience (Krill, Platek, Goetz, and Shackelford, 2007; Platek,
Keenan, and Shackelford, 2007). For example, in their recent article, Krill et al. reviewed
how evolutionary theory has been successfully applied to previous studies in cognitive
neuroscience and then highlighted some exciting new frontiers for future research. Krill et
al. also mentioned that the emerging symbiosis between evolutionary psychology and
cognitive neuroscience might not be entirely seamless. Specifically, they noted that
evolutionary perspectives have had some difficulty permeating several cognitive
neuroscience journals (e.g., Cognitive Brain Research, Journal of Cognitive Neuroscience).

The purpose of the present research was to quantify the extent to which
evolutionary theory has influenced neuroscience (Study 1) and cognitive neuroscience
(Study 2) publications over the last 20 years and to assess evolutionary theory’s influence
compared to other human science areas (Study 2). Five specific questions were developed:

1. Study 1: Is the proportion of neuroscience articles that mention evolution
significantly different from zero as of 2006?
2. Study 1: Has the proportion of neuroscience articles that mention evolution grown
significantly for the average neuroscience journal in the past 20 years (1987-2006)?
3. Study 1: Do between-journal differences in impact factor (i.e., frequency of
citation) moderate the within-journal change-over-time effects for neuroscience
articles that mention evolution?
4. Study 2: Is the proportion of articles that mention evolution in journals that focus on
human cognitive neuroscience significantly different from zero as of 2006?
5. Study 2: How does the growth in the proportion of articles that mention evolution in
journals that focus on human cognitive neuroscience over the last 20 years compare
to the proportion of articles that mention other areas of human science (i.e.,
anthropology, economics, psychology, and sociology)?
Study 1
Method
Sample

A convenience sample of ten neuroscience journals that were indexed in PsycINFO
was obtained (Table 1). This sample of journals was fairly diverse, with 2005 impact
factors ranging from 0.68 to 15.46 with a median of 3.05 (M = 4.45, SD = 4.29). Journal
impact factors are a measure of how often the average article in a journal is cited in a given
year, and were obtained from the Institute for Scientific Information’s Journal Citation
Reports: Science Edition website (http://portal.isiknowledge.com).

Procedure
PsycINFO keyword searches were performed using the term “evol*” for articles
published in each of the ten neuroscience journals from 1987 to 2006. Thus, these searches
found articles that mentioned “evolve,” “evolved,” “evolves,” “evolving,” “evolution,”
“evolutionary,” or “evolutionarily” in their titles, abstracts, or keywords. The number of
articles that mentioned at least one of these terms in a given year was then divided by the
Evolutionary Psychology – ISSN 1474-7049 – Volume 5(3). 2007. - 521-

---

Evolutionary Theory in Cognitive Neuroscience
total number of articles published that year to obtain a proportion, and this was done for
each journal. Because proportion data often violate the homogeneity-of-variance
assumption of regression (Judd and McClelland, 1989, pp. 525-526; cf. Webster, 2007b),
an arcsine transformation was applied to the proportions: sin–1(proportion1/2).
Results
Two types of analyses were performed. First, simple ordinary-least-squares (OLS)
regressions were run for each journal with year of publication predicting either raw
percentages or arcsine-transformed proportions (Table 1). Second, a series of multilevel
random coefficient models (MRCMs) were run using restricted maximum likelihood
(RML) estimation. RML methods provide more stable and accurate parameter estimates
than weighted OLS methods (Nezlek, 2001; Raudenbush and Bryk, 2002). Detailed
examples of the MRCMs used in the present study are discussed below.

Simple Ordinary-Least-Square Regressions

Table 1 shows the results for a series of simple, within-journal regression analyses
of publication year predicting either the raw percentages (left columns) or arcsine-
transformed proportions (right columns) of articles that mention evolution. Three journals
(viz., Brain, Brain and Cognition, and NeuroReport) showed significant increases over
time in the percentage of articles mentioning evolution. Also, three-fifths of the journals
published in 2006 contained 1% or more articles referring to evolution in one form or other.
The results for arcsine-transformed proportions were similar, with two additional journals
(viz., Cognitive Brain Research and European Journal of Neuroscience) showing
marginally significant (ps < .10) increases in proportion of articles mentioning evolution
over time. Notably, in none of these analyses did mentions of evolution significantly (over
even marginally) decrease over time.

Multilevel Random Coefficient Models
The present data structure was inherently hierarchical, with time (years) nested
within journals. To this end, a two-level longitudinal growth model was employed (cf.
Raudenbush and Bryk, 2002; Singer and Willett, 2003) using the HLM 6 program
(Raudenbush, Bryk, Cheong, and Congdon, 2004). In multilevel modeling terms, within-
journal (temporal) variance was modeled at level 1, and between-journal variance was
modeled at level 2. The level-1 model was:

sin–1(proportion1/2)tj = ?0j + ?1j(Year – 1996.5)j + etj.

In this model, arsine-transformed proportions of articles mentioning evolution are modeled
as a function of publication year (mean-centered at 1996.5). The intercept, ?0j, represents
the predicted mean proportion at time t (i.e., the midpoint of 1996) within journal j. The
raw, unstandardized slope, ?1j, represents the simple change-over-time (publication year) in
proportions across t times within journal j. The random error term, etj, represents the
residual within-journal variance.


Evolutionary Psychology – ISSN 1474-7049 – Volume 5(3). 2007. - 522-

---

Evolutionary Theory in Cognitive Neuroscience
Table 1. Simple, ordinary-least-squares regression results: Change over time in the raw percentages and
arcsine-transformed proportions of neuroscience articles that mention evolution.


Raw percentages
Arcsine proportions
No. 2006
2006
Journal
Slope
t
Slope
t
yrs.
mean
mean
Behavioral Neuroscience
20
0.76
0.015
0.52
0.08 0.0026 1.20
Brain
13
3.74
0.316
2.80*
0.20 0.0158
2.95*
Brain and Cognition
20
1.85
0.092
2.38*
0.13 0.0062
2.44*
Cognitive Brain Research
14
0.94
0.061
1.47
0.09 0.0067
2.01†
Cognitive Neuropsychology
20
1.03
–0.012
–0.15
0.06 0.0005 0.13
European J. of Neuroscience
7
1.38
0.195
1.73
0.13 0.0178
2.21†
International J. of Neuroscience
17
1.60
0.043
0.48
0.09 0.0015 0.34
J. of Cognitive Neuroscience
18
0.92
–0.132
–1.55
0.10 –0.0024 –0.60
Nature Neuroscience
9
1.27
–0.089
–0.61
0.10 –0.0031 –0.34
NeuroReport
17
0.95
0.065
3.03*
0.10 0.0067
3.71*

Note. Arcsine proportions = sin–1(proportion1/2). No. yrs. = Number of years with available data (1987-2006).
2006 mean = predicted mean for 2006. J. = Journal.
†p < .10. *p < .05.

The level-1 intercepts (means) and slopes were then modeled at level 2 as a function
of their means:
?0j = ?00 + r0j,
?1j = ?10 + r1j.

In this model, both the within-journal intercepts (?0js) and slopes (?1js) are modeled as
functions of their respective intercepts (?00, ?10). For example, the unstandardized
coefficients of interest here are ?00 and ?10. First, ?00 describes the extent to which the
arcsine-transformed proportion for the average journal is different from zero at the
midpoint of the year distribution (i.e., 1996.5). The effect of ?00 can be tested at different
years simply by re-centering the year variable. For example, testing whether the arcsine-
transformed proportion for the average journal is different from zero during the 2006
Evolutionary Psychology – ISSN 1474-7049 – Volume 5(3). 2007. - 523-

---

Evolutionary Theory in Cognitive Neuroscience
publication year simply involves re-centering the year variable by subtracting 2006 from
each value and re-evaluating the model. Second, ?10 describes the extent to which the
within-journal change-over-time slopes in proportions are different from zero for the
average journal (i.e., “Has the proportion of articles mentioning evolution increased over
time for the average neuroscience journal?”). The error terms, r0j and r1j, represent the
between-journal residual variances for their respective intercepts (?0js) and slopes (?1js).

The multilevel model described above revealed that both the intercept (?00 = 0.067,
t9 = 7.04, p < .001) and the change-over-time slope (?10 = 0.0039, t9 = 2.60, p < .03, r =
.65) were significantly different from zero (Figure 1). Simple effects tests revealed that the
average proportion of neuroscience articles mentioning evolution was (a) not significantly
different from zero in 1987 (?00 = 0.030, t9 = 1.39, p = .20), (b) was significantly different
from zero by 1990 (?00 = 0.042, t9 = 2.40, p < .05) and every year thereafter, and (c) was
substantially different from zero for the most recent data in 2006 (?00 = 0.10, t9 = 9.46, p <
.001). In addition, the change-over-time slopes were fairly consistent across journals: A
random effects test indicated there was a non-significant amount of variance in the
journals’ change-over-tim

e slopes (?29 = 14.38, p = .11).

Figure 1. Predicted mean percent of neuroscience articles mentioning evolution.
1.2

)


(
%

e
c
n 0.9
i
e

c
os

ur 0.6
ne

n
0.3
i
on i
utolv
E 0.0
1986
1991
1996
2001
2006

Year of publication
To test whether differences in journals’ impact factors moderated the change-over-
time slopes, the level-1 intercepts (means) and slopes were modeled at level 2 as a function
of their means and their between-journal differences in mean-centered impact factor:

?0j = ?00 + ?01(Journal impact factor) + r0j,
?1j = ?10 + ?11(Journal impact factor) + r1j.

In this model, both the within-journal intercepts (?0js) and slopes (?1js) are modeled as
functions of their respective intercepts (?00, ?10) and between-journal differences in impact
factor (?01 and ?11, respectively). For example, the unstandardized coefficients of interest
here are ?01 and ?11. First, ?10 describes the extent to which the arcsine-transformed
Evolutionary Psychology – ISSN 1474-7049 – Volume 5(3). 2007. - 524-

---

Evolutionary Theory in Cognitive Neuroscience
proportions for the average journal at the midpoint of the year distribution (i.e., 1996.5) are
moderated by between-journal differences in impact factor. Second, ?11 describes the extent
to which within-journal change-over-time slopes in proportions are moderated by between-
journal differences in impact factor (i.e., “Do journals of different impact have different
change-over-time slopes?”). The error terms, r0j and r1j, represent the between-journal
residual variances for their respective intercepts (?0js) and slopes (?1js).
When impact factor was added as a between-journal predictor (as shown in the
model above), it moderated neither the within-journal intercepts (?01 = 0.0038, t8 = 1.06, p
= .32) nor the change-over-time slopes (?11 = –0.00014, t8 = –0.22, p = .83). Thus, growth
in mentions of evolution in neuroscience articles over time was fairly constant across
journals, at least insofar as differences in their impact factors were concerned.

Discussion

The simple, OLS, within-journal regressions suggested that at least half of the ten
neuroscience journals sampled experienced at least marginally significant growth in
mentions of evolution over time (ps < .10), but only three showed significant growth (ps <
.05). More importantly, the MRCMs revealed that mentions of evolution in the average
neuroscience journal had been significantly different from zero from 1990 onward and had
grown significantly over time between 1987 and 2006. Further, this growth was fairly
uniform across journals, and was not differentiated by journal impact factor. Thus, unlike
previous studies in which impact factors moderated temporal publication trends in article
length in psychology journals (Webster, 2007a), the present study showed no appreciable
moderating effect of impact factors on the change-over-time slopes: Evolution’s influence
on neuroscience appears to be growing equally among both high- and low-impact journals.

Study 2

Method

Sample
Four of the ten neuroscience journals surveyed in Study 1 that specialize in human
cognitive neuroscience were selected for Study 2 (Table 2). This sub-sample of journals
remained fairly diverse, with 2005 impact factors ranging from 1.84 to 4.53 with a median
of 2.74 (M = 2.96, SD = 1.15).

Procedure
As in Study 1, PsycINFO keyword searches were performed for evolution using the
term “evol*” for articles published in each of the four journals from 1987 to 2006. In
addition, other keyword searches were preformed for four other areas of human science:
anthropology (“anthropolog*”), economics (“econom*”), psychology (“pyscholog*”), and
sociology (“sociolog*”). As in Study 1, arcsine-transformed proportions were calculated.




Evolutionary Psychology – ISSN 1474-7049 – Volume 5(3). 2007. - 525-

---

Evolutionary Theory in Cognitive Neuroscience
Results

As in Study 1, a series of MRCMs were run using RML estimation. In contrast to
Study 1, a three-level model was required that accounted for the dependency of years (level
1) nested within journals (level 2) nested within topic areas (level 3). The level-1 model,

sin–1(proportion1/2)tja = ?0jt + ?1ja(Year – 1996.5)ja + etja,

was essentially the same as in Study 1, with the exception of an additional subscript, a,
which represents the area of interest (e.g., evolution, psychology, etc.). The level-2 model,

?0ja = ?00a + r0ja,
?1ja = ?10a + r1ja,

was also similar to Study 1, but with the additional subscript, a. The level-3 model,

?00a = ?000 + ?001(Evolution)a + u00a,
?10a = ?100 + ?101(Evolution)a + u10a,

took the journal-level coefficients and modeled them as a function of different topic areas.
In this model, the coefficient ?100 is of primary interest; it represents the simple change-
over-time slope for articles that mention evolution averaged across the four cognitive
neuroscience journals. Simple slopes for the other four topic areas were estimated in a
similar manner by replacing ?101(Evolution)a with the area of interest, such as
?101(Psychology)a. Each of these dummy-coded variables was created such that the area of
interest was coded as 0 and the other four areas were coded as 1. As in Study 1, testing
whether the arcsine-transformed proportion for a given area was different from zero in
2006 was done by re-centering the publication year variable by subtracting 2006 from each
value and re-evaluating the model. The level-3 error terms, u00a, and u10a, represent the
between-area residual variances for their respective intercepts (?00as) and slopes (?10as). As
a final illustration, this three-level model is analogous to examining 20 change-over-time
slopes (4 cognitive neuroscience journals x 5 topic areas), where sets of four journal slopes
are averaged (“braided together,” i.e., “ropes of slopes”) within each of their five respective
topic areas, yielding five “grand,” area-level slopes for comparison (Figure 2).
An initial, random effects test indicated there was a significant amount of variance
among the five topic areas’ change-over-time slopes (?24 = 9.73, p = .04), which suggested
that examining differences in simple slopes could be informative. The 2006 estimated
means and simple change-over-time slopes from the multilevel model described above are
shown in Table 2. For the 2006 estimated means, only mentions of psychology differed
significantly from zero (t3 = 4.50, p < .03). Psychology was also the only area to show a
trend of increased penetration over time, albeit not a significant one (p < .07). The simple
change-over-time slopes are shown as raw percentages (back-transformed from the arcsine
metric) in Figure 2. The top panel shows all five topic areas, whereas the bottom panel
shows a magnified version of the top panel to highlight differences among anthropology,
economics, and sociology.
Evolutionary Psychology – ISSN 1474-7049 – Volume 5(3). 2007. - 526-

---

Evolutionary Theory in Cognitive Neuroscience
The significance tests shown in Table 2, however, were based on only 3 degrees of
freedom because the sample consisted of only five topic areas; thus, effect sizes might be
more informative. Change-over-time effect sizes for psychology (r = .88) and evolution (r
= .49) indicated that both were large in magnitude. Evolution’s penetration into four
cognitive neuroscience journals (r = .49) was comparable to the effect size obtained across
ten neuroscience journals from Study 1 (r = .65). In contrast, change-over-time effect sizes
were comparatively small for anthropology (r = –.04), economics (r = .14), and sociology
(r = .00). Note that sociology was mentioned in none of the articles surveyed in this study,
which explains its slope of zero.

Table 2. Change over time in arcsine-transformed proportions of cognitive neuroscience articles from four
journals that mention one of five topic areas.

Area Predicted
2006
mean
Slope
t3
Anthropology 0.00021
–0.00017
–0.07
Economics
0.01411
0.00058
0.25
Evolution
0.09335
0.00231
0.98
Psychology
0.16001
0.00524
3.18†
Sociology
0.00000
0.00000
0.00

†p < .07.

Discussion


Recall that the two main questions for Study 2 were (a) whether the proportion of
articles that mention evolution in journals that specialize in human cognitive neuroscience
was different from zero by 2006 and (b) how the growth in evolution’s penetration into
cognitive neuroscience compared to that of other human sciences. Results indicated that,
although mentions of evolution were indistinguishable from zero in cognitive neuroscience
journals in 2006, the change over time in mentions of evolution was more positive than that
of other disciplines such as anthropology, economics, and sociology, but evolution still
lagged far behind the growth of psychology. In fact, according to the model, the proportion
of mentions of evolution increased by more than 2.5 times between 1987 and 2006. In sum,
Study 2 offers a mixed bag of findings: Evolution’s influence is cognitive neuroscience
continues to be small; however, it is growing at a respectable rate that outstrips at least
three other relevant human sciences.

Conclusion


In concert, these quantitative findings imply that the outlook for evolutionary
psychology’s influence on neuroscience in general, and cognitive neuroscience in
particular, is generally positive. Although mentions of evolution permeated only about 1%
of neuroscience and cognitive neuroscience articles in 2006, its growth over time is
certainly encouraging. Perhaps even more encouraging is the fact that mentions of
evolution are growing faster than mentions of other human sciences such as anthropology,
Evolutionary Psychology – ISSN 1474-7049 – Volume 5(3). 2007. - 527-

---

Evolutionary Theory in Cognitive Neuroscience
economics, and sociology, although mentions of psychology still outpace all of these areas.
Given these promising trends, the emergence of evolutionary cognitive neuroscience as a
viable interdisciplinary field is becoming less of a possibility and more of an inevitability.

Figure 2. Predicted mean percent of mentions by area in cognitive neuroscience articles. Top panel shows all
five areas; bottom panel shows magnified version of top panel to highlight differences among three areas.


es
2.5
Anthropology
t
i
cl
Economics
2.0
Evolution
ce ar
Psychology
en
1.5
Sociology
sci
o
ur
1.0
e
0.5
i
t
i
ve n
n
0.0
cog
1986
1991
1996
2001
2006
n
i
Year of publication
%


s 0.020
l
e
Anthropology
r
t
i
c
a
Economics
e
c 0.015
Sociology
n
i
e
c
os 0.010
ur
ne
e 0.005
t
i
v
ogni
c 0.000
n
i
1986
1991
1996
2001
2006
%
Year of publication
Evolutionary Psychology – ISSN 1474-7049 – Volume 5(3). 2007. - 528-

---

Evolutionary Theory in Cognitive Neuroscience


How does evolution’s penetration into neuroscience and cognitive neuroscience
compare to its penetration into other disciplines of psychological science such as social
psychology? One way is to compare the change-over-time slopes and effect sizes for
evolution among neuroscience (slope = 0.0039, r = .65), cognitive neuroscience (slope =
0.0023, r = .49), and social psychology journals (slope = 0.0019; r = .47; i.e., Journal of
Personality and Social Psychology, 1985-2004; cf. Webster, 2007b). This comparison
suggests that evolution’s penetration into evolutionary cognitive neuroscience has
increased at a rate that is roughly equivalent to its penetration into personality and social
psychology over the last two decades; however, its penetration into neuroscience in general
appears to have happened at an even faster rate.

Despite these interesting trends, the present research is not without its limitations.
Chief among these are the fact that keyword searches are an imperfect measure of the
extent to which a given topic such as evolution truly influences a given journal article. Not
only are such keyword searches indiscriminant with regard to where the keyword occurs
(e.g., title, abstract, keyword listing), but also with regard to the frequency with which the
keyword occurs throughout the article. Thus, an article that mentioned evolution at least
once was given the same weight as an article that may have mentioned evolution several
times. The potential measurement limitation of the keyword search method, however, was
partially mitigated by the fact that it allowed for an uncommonly large sample of articles to
be surveyed efficiently (i.e., 18,051 in Study 1 and 4,859 in Study 2). Although the present
findings suggest that evolution is having an increasing influence on cognitive neuroscience,
assessing the exact depth of that influence will require more detailed analyses in the future.
The present findings add to a growing literature that suggests that evolutionary
psychology is gaining in stature and influence. In addition to making inroads into social
psychology and cognitive neuroscience, evolutionary psychology is also becoming a more
empirical science, as evidenced by a recent survey of publication trends over the first 25
years of its flagship journal, Evolution and Human Behavior (Webster, 2007c). With its
increased empirical focus, evolutionary psychology stands to gain further acceptance, both
as a viable sub-discipline in psychology and neuroscience, and as an interdisciplinary field
that helps foster and promote evolutionary perspectives in several sciences. It is hoped that
the present study will not only inform readers on the current state of evolution’s influence
in neuroscience and cognitive neuroscience, but also inspire readers to pursue research in
the emerging field of evolutionary cognitive neuroscience.

Acknowledgements: This material is based upon work supported by the National Institute
of Mental Health under Training Grant Award PHS 2 T32 MH014257 entitled
“Quantitative Methods for Behavioral Research” (to M. Regenwetter, principal
investigator). This research was performed while the author was a postdoctoral trainee in
the Quantitative Methods Program of the Department of Psychology, University of Illinois
at Urbana-Champaign. Any opinions, findings, and conclusions or recommendations
expressed in this article are those of the author and do not necessarily reflect the views of
the National Institute of Mental Health.

Received 15 May 2007; Revision submitted 26 July 2007; Accepted 3 August 2007
Evolutionary Psychology – ISSN 1474-7049 – Volume 5(3). 2007. - 529-

---