Professional Physical Scientists Display Tenacious Teleological
Tendencies: Purpose-Based Reasoning as a Cognitive Default
Deborah Kelemen, Joshua Rottman, and Rebecca Seston
Boston University
Teleological explanations account for objects and events by reference to a functional consequence or
purpose. Although they are popular in religion, they are unpopular in science: Physical scientists in
particular explicitly reject them when explaining natural phenomena. However, prior research provides
reasons to suspect that this explanatory form may represent a default explanatory preference. As a strong
test of this hypothesis, we explored whether physical scientists endorse teleological explanations of
natural phenomena when their information-processing resources are limited. In Study 1, physical scientists
from top-ranked American universities judged explanations as true or false, either at speed or without time
restriction. Like undergraduates and age-matched community participants, scientists demonstrated increased
acceptance of unwarranted teleological explanations under speed despite maintaining high accuracy on control
items. Scientists’ overall endorsement of inaccurate teleological explanation was lower than comparison
groups, however. In Study 2, we explored this further and found that the teleological tendencies of professional
scientists did not differ from those of humanities scholars. Thus, although extended education appears to
produce an overall reduction in inaccurate teleological explanation, specialization as a scientist does not, in
itself, additionally ameliorate scientifically inaccurate purpose-based theories about the natural world. A
religion-consistent default cognitive bias toward teleological explanation tenaciously persists and may have
subtle but profound consequences for scientific progress.
Keywords: teleology, design, explanation, dual processing, agency
Supplemental materials: http://dx.doi.org/10.1037/a0030399.supp
“Inquiry into final causes is sterile and like a virgin consecrated to God,
produces nothing.”
—Francis Bacon, De Augmentis Scientarum, Book III
Aristotle classically argued that an adequate scientific response
to any “why” question about nature requires reference to four types
of causes. Significant among these was the “efficient cause,” the
antecedent source of an object or event. But although efficient
causes are familiar as the guiding focus of scientific discovery and
explanation in the contemporary physical sciences, Aristotle him-
self viewed them as secondary when it came to adequately
accounting for the existence and properties of natural objects.
For him, the pinnacle of explanation for all living and nonliving
natural phenomena lay with identification of the “final cause”
or “the end that for the sake of which a thing is done (telos),”
in other words, the object or event’s goal or function. Thus, for
Aristotle, leaves on plants exist in order to provide shade,
flames flicker because their natural end point lies heavenwards,
and water exists in order to sustain life on Earth (see Aristole,
c. 350 BC/1930).
Aristotle’s appeal to goals and functions as a basis of explana-
tion, what is nowadays termed teleological explanation, was pro-
miscuous. He viewed teleology as a fundamental and general
principle of explanation broadly applicable to living and nonliving
natural phenomena of all kinds. This liberality made sense in
context of his underlying theoretical assumptions about the general
makeup of the cosmos: He viewed it as akin to a living organism.
Just as an organism’s component biological processes and organs
seem intrinsically charged with the purpose to maintain the vital
organism as a whole, so natural entities act and function with the
end of preserving the integrity of the universe. Physical scientists
nowadays reject these animistic causal assumptions as well as the
other empirically unverifiable metaphysical beliefs that have his-
torically licensed teleological explanation in science (e.g., the
theory that nature is an artifact of divine design). This rejection
began in the Renaissance, and with it, as Bacon’s epigram sug-
gests, teleological explanation increasingly fell into disrepute. The
critique was that it is logically flawed and nonexplanatory because,
stripped of any animistic or intentional theoretical underpinnings,
statements like “atoms react in order to maintain stability” violate
temporal constraints by treating an entity’s consequence as if it
could be its own cause in backward causal fashion.
This article was published Online First October 15, 2012.
Deborah Kelemen, Joshua Rottman, and Rebecca Seston, Department of
Psychology, Boston University.
This research is an output of the Cognition, Religion and Theology
Project at the University of Oxford funded by the John Templeton Foun-
dation. It was also funded by National Science Foundation (NSF) DRL-
1007984 (awarded to Deborah Kelemen) and NSF Graduate Research
Fellowship DGE-0741448 (awarded to Joshua Rottman). The views ex-
pressed are those of the authors. Many thanks to Holly Jacobs for data
collection assistance, Matt Rogers of Colorado State University for advice,
and the BU Child Cognition Lab research assistants for all of their help.
Correspondence concerning this article should be addressed to Deborah
Kelemen, Boston University, Department of Psychology, 64 Cummington
Mall, Boston, MA 02215. E-mail: dkelemen@bu.edu
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Journal of Experimental Psychology: General © 2012 American Psychological Association
2013, Vol. 142, No. 4, 1074 –1083 0096-3445/13/$12.00 DOI: 10.1037/a0030399
1074
More recently, attempts have been made to reestablish the
scientific legitimacy of teleological explanation, at least for evo-
lutionary biology (e.g., Allen, Bekoff, & Lauder, 1998; Hempel &
Oppenheim, 1948; Mayr, 1985; Nagel, 1961; Neander, 1991;
Perlman, 2004; Sober, 1984; Wright, 1976).
1
Nevertheless, outside
of those domains in which intentional causality can be implicitly
assumed (e.g., intentionally designed artifacts and goal-directed
behavior), the status of teleological explanation in science remains
highly controversial, most markedly in the physical sciences: In
such disciplines, explanatory references to the goals and purposes
of inanimate natural phenomena are not only tainted by quasi-
religious overtones of design and animism but also deemed super-
fluous because all phenomena can be more straightforwardly
mechanistically explained by reference to antecedent physical-
causal conditions (e.g., Barrow & Tipler, 1986; Burtt, 1932; Co-
rey, 1993; Perlman, 2004; Talanquer, 2007; White, 1992; Wicken,
1981). But although contemporary physical scientists might ex-
plicitly reject teleological explanation, it is unclear whether its
influence on thought and scientific inquiry is really so easy to
escape. The question at the heart of the present article is, can it
ever be escaped?
Contemporary chemistry and physics may have been purged of
teleological explanation, but that has not undermined its appeal
among populations that are relatively untutored in science. In fact,
a broad tendency to see purpose in nature may run quite deep in the
human psyche. A body of research has demonstrated that a bias
toward teleological explanation is established from quite early in
development. Young children are “promiscuously teleological,”
displaying strong, generalized preferences for teleological rather
than physical-causal explanations of living and nonliving natural
objects from preschool (DiYanni & Kelemen, 2005; Kelemen,
1999a, 1999b, 2003; Kelemen & DiYanni, 2005; but see Greif,
Kemler Nelson, Keil, & Gutierrez, 2006; Keil, 1992). Such find-
ings are also echoed in various adult populations, suggesting
children’s bias is not simply a symptom of immaturity that is
automatically extinguished by development. In particular, Roma-
nian Roma adults with minimal schooling (Casler & Kelemen,
2008) and Alzheimer’s patients with degraded semantic memories
(Lombrozo, Kelemen, & Zaitchik, 2007) also display broad tele-
ological intuitions about natural entities and their properties. Un-
like undergraduates, when they are asked to choose between ex-
planations for the existence of natural objects or their properties,
they tend to endorse teleological ideas (e.g., prehistoric rocks were
pointy so that animals would not sit on them and smash them)
rather than physical-causal alternatives (e.g., they were pointy
because material built up over time).
Taken together, these results are consistent with a proposal
that teleological explanation represents something of a develop-
mentally persistent cognitive default. This proposal breaks from
traditional accounts of conceptual development, which tend to
characterize conceptual change as a process of revision and re-
placement in which earlier intuitive theories are effectively re-
structured by the acquisition of more veridical scientific accounts
such that they subsequently become unavailable for explanation
and inference (e.g., Piaget, 1983; see also Carey, 1985, 1991; Chi,
1992; Gopnik & Wellman, 1994). In contrast, it seems possible
that a coexistence account of theory change might be more accu-
rate: Rather than being supplanted by the elaboration of scientific
theoretical ideas, it is feasible that an early arising intuitive tele-
ological construal of nature might remain as a lifelong bias, which
may be inhibited and concealed by later constructed beliefs, but
which is never fully displaced (e.g., Dunbar, Fugelsang, & Stein,
2007; E. M. Evans, Legare, & Rosengren, 2011; Kelemen &
Rosset, 2009; Zaitchik & Solomon, 2008).
2
This is akin to dual-
processing models that characterize early developing intuitions as
heuristics that can be increasingly overridden in later development
by effortful processing, but which can nevertheless persistently
reemerge in cases when intuitions are favored or forced (e.g., J. St.
B. T. Evans, 2008, 2011; Kahneman, 2011; Stanovich, West, &
Toplak, 2011).
This coexistence proposal makes a clear prediction: Even highly
educated individuals with substantial countervailing scientific con-
tent knowledge and intellectual bias should occasionally betray
signs that they preferentially default to scientifically unwarranted
teleological interpretations of natural phenomena. However, to
date, a strong test of this prediction has never been conducted
despite suggestive findings from college-educated populations. In
earlier work, Kelemen and Rosset (2009) found that although
university undergraduates favor physical-causal over teleological
alternatives on the kinds of reflective reasoning tasks used to test
children, they show striking tendencies to accept inaccurate tele-
ological statements about living and nonliving natural phenomena
when their cognitive resources are taxed. That is, when asked to
judge explanations at speed—a manipulation that precludes inhi-
bition of initial nonreflective explanatory reactions—undergradu-
ates showed heightened tendencies to accept inaccurate teleolog-
ical explanations (61%) relative to their unspeeded counterparts
(52%). By contrast, their speeded responses to inaccurate control
explanations revealed no equivalent levels of inaccuracy or per-
formance decrement.
As noted, these results are suggestive evidence regarding the
status of teleological explanation as a default. Nevertheless, they
remain inconclusive for several reasons. First, even with exposure
to multiple college-level science classes, the scientific knowledge
of undergraduates is relatively weak. Indeed, independent assess-
ments of biological and geoscience content knowledge in the
undergraduates studied by Kelemen and Rosset (2009) confirmed
the fragmentary nature of their understanding. Second, there is, in
general, no strong reason to assume that teleological explanations
should be absent from undergraduate populations. Such students
1
The dominant defense of teleological explanation in evolutionary bi-
ology is the etiological argument, which proposes that it is legitimate to
make statements that explain a biological trait in terms of its functional
effect (e.g., hearts are for pumping blood) if that functional capacity caused
an organism with a trait of that structural type to get selected during the
evolutionary process (Millikan, 1989; Neander, 1991; Wright, 1976). It is
unlikely that most adults engage in such mechanistically based historical
assumptions when making teleological statements about biological traits
and organs, however, given that natural selection is generally misunder-
stood as a foresightful rather than a blind process. Although biologists use
teleological language routinely, it remains controversial. The distaste has
prompted various attempts to strip the term teleology from the life sciences
and use teleonomy instead (e.g., Mayr, 1985).
2
Interpretively, then, the reason why Alzheimer’s patients demonstrate
promiscuous teleological beliefs is not because the degradation of their
knowledge base prompts them to construct new teleological ideas resem-
bling those once constructed in childhood but because the purpose-based
explanatory preferences and beliefs that were developed in childhood never
disappeared.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1075
TELEOLOGICAL EXPLANATION IN SCIENTISTS
have not necessarily adopted the reflective, normative stance that
teleology is scientifically inappropriate. Unlike professional scien-
tists, it is probable that they have not even given careful thought to
the issue. Patterns in Kelemen and Rosset’s findings again sug-
gested this. Although students found teleological explanations
even more appealing under cognitive load, their endorsement of
teleological explanation was already over 50% when they were
given time to reflect on their answers in an unspeeded condition.
In the present research, we therefore present the clearest explo-
ration of the coexistence proposal—and the tenacious entrench-
ment of a teleological bias—by examining an expert population
whose knowledge of physical mechanisms is unquestionable and
for whom scientific physical-causal theorizing is not only privi-
leged but normative and routine
3
: professional physical scientists
at high-ranking American research universities. Do physical sci-
entists default to scientifically unwarranted teleological explana-
tions when their abilities to censor their automatic explanatory
reactions are compromised by being required to respond at speed?
To explore this, we asked academically active physical scientists
to perform a speeded explanation judgment task in which they
judged the correctness of warranted and unwarranted explanations
of various phenomena under speeded conditions or without time
limits. In addition, we also explored the potential theoretical com-
mitments that might underpin this bias given the alternative pos-
sibility that the tendency is atheoretical and reflects a conceptually
nondecomposable innate interpretive stance (e.g., Keil, 1992,
1995), modular heuristic (Atran, 1995), or basic inferential ten-
dency triggered by structure-function fit (e.g., Lombrozo et al.,
2007). With regard to the idea that the teleological bias has folk
theoretical underpinnings and occurs because of underlying causal
commitments promoting beliefs that natural phenomena exist for
purposes, two related alternatives seem likely. One is the quasi-
spiritual agentive theory that the Earth is a goal-directed living
organism—in contemporary parlance, a cosmological belief in
Mother Earth or “Gaia.” The other is the notion that nature is an
artifact of supernatural design—an idea that is, of course, also well
represented in the major religions. Both of these kinds of theories
have explicitly licensed scientific assumptions about final causa-
tion throughout history. Relevantly, recent studies have yielded
evidence that both may play a role in promoting teleological
beliefs: In young children, promiscuous teleology has been found
to relate to beliefs about intentional design in nature (Kelemen &
DiYanni, 2005; see also Diesendruck & Haber, 2009), and in
undergraduates, evidence has suggested connections to intuitive
Gaia beliefs (Kelemen & Rosset, 2009; but see Lombrozo et al.,
2007). Although it was anticipated that contemporary physical
scientists would not overtly endorse the existence of Gaia or God
with any particular enthusiasm, any evidence that the teleological
bias is a product of agency-based causal explanatory beliefs rather
than a more conceptually primitive heuristic has important impli-
cations for not only theory but also educational practice (Kelemen,
2012).
Study 1
Method
Participants. The final sample included 80 physical scientists
(39 women, mean age ⫽ 36 years, SD ⫽ 4; mean years since
Ph.D. ⫽ 8 years, SD ⫽ 4) who were actively publishing scholars
in chemistry, geoscience, and physics departments at high-ranking
American colleges and universities (e.g., Boston University,
Brown, Columbia, Harvard, MIT, Yale). Two control populations
included 179 Boston-area college undergraduates (107 women,
mean age ⫽ 19 years, SD ⫽ 1) and 49 members of the Boston
community who were age-matched to the scientists but held only
bachelor’s degrees (28 women, mean age ⫽ 38 years, SD ⫽ 5).
Participants in all groups were native English speakers, under 45
years, and without self-reported color-blindness or dyslexia. An
additional 71 participants were excluded for failing to (a) respond
to at least 75% of the test items and/or (b) accurately respond to at
least 80% of the control items (nine science; 51 college; 11
community). The whole study took approximately 1 hr to com-
plete. College undergraduates, who had the highest rate of exclu-
sion, received experimental course credit for their participation; all
other participants received a cash payment.
Procedure. Stimuli were 100 one-sentence explanations for
“why things happen,” presented consecutively on a laptop using
PsyScope software (Cohen, MacWhinney, Flatt, & Provost, 1993).
Participants judged them as “true” or “false” using two response
keys. There were 30 test sentences and 70 control sentences.
Test sentences described scientifically unwarranted teleological
explanations for various natural phenomena (e.g., “The sun radi-
ates heat because warmth nurtures life”; see Table 1 for more
examples and supplementary online material for a full list). Con-
trol sentences were of four types that were included to track
participants’ response biases, their overall abilities to read at
speed, and their accuracy at judging the truth or falsity of explan-
atory statements in general. They comprised 20 true causal expla-
nations (e.g., “Conception occurs because sperm and eggs fuse
together”), 10 true teleological explanations (e.g., “Children wear
mittens in the winter in order to keep their hands warm”), 30 false
causal explanations (e.g., “Snowflakes are white because they are
symmetrical”), and 10 false teleological explanations (e.g., “Win-
dow blinds have slats so that they can capture dust”). In contrast to
teleological test sentences, which involved inaccurate explanations
of natural phenomena, false teleological control sentences con-
cerned the social-conventional and artifact domains in which te-
leological explanation is appropriate. They were false by virtue of
incongruity. To catch response strategies based on skimming sen-
tences for content words rather than reading them fully, control
sentences, like test sentences, invoked closely associated concepts
throughout.
Participants in each group were randomly assigned to speeded
and unspeeded conditions. In the speeded condition, participants
had a maximum of 3,200 ms to respond—a speed determined, via
piloting, to be two standard deviations above the average reading
time for all sentences. In the unspeeded condition, participants
were asked to make a judgment after careful consideration and
received no time limit. In both conditions, the stimulus progressed
immediately after the participant’s response (or, in the speeded
condition, after 3,200 ms had passed if the participant had not yet
3
Note that these characteristics do not extend to professional biologists
who habitually (and controversially) use teleological language and are
therefore an ambiguous population for study (see Allen et al., 1998; Mayr,
1985; Sober, 1984).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1076
KELEMEN, ROTTMAN, AND SESTON
responded). Sentences were presented consecutively in 10 ten-
sentence blocks with a 3-s pause between blocks. Each block
contained seven control items and three test items in random order.
Two blocks of practice sentences were not included in analyses.
In order to explore individual differences in general suscepti-
bility to the teleological bias, participants also completed several
additional measures. People with poorer inhibitory control may
generally have greater difficulties suppressing automatic intuitive
reactions in favor of less intuitive, tutored responses, and so we
examined whether poor inhibitory control would increase tenden-
cies to endorse teleological ideas. Participants therefore completed
a computer-based 48-item Stroop color task (Stroop, 1935)to
measure inhibitory control. Their task was to quickly identify the
print color of a written color word when the meaning of the written
word and its print color were different (incongruent) or the same
(congruent). Scores were calculated by subtracting the average
reaction time to congruent items from the average reaction time to
incongruent ones to create a difference score. Higher Stroop dif-
ference scores indicated lower inhibitory control and were used to
examine whether Stroop scores would positively predict teleolog-
ical sentence endorsement. Stroop data were not included in anal-
yses if participants were less than 80% accurate on incongruent
Stroop trials (one science, 11 college, one community).
Gaps in scientific knowledge base could also increase tenden-
cies to endorse intuitive teleological ideas because alternative
physical-causal explanations would be unavailable in individuals’
semantic memories. To measure scientific knowledge, participants
completed the 20-item multiple-choice Conceptual Inventory of
Natural Selection (Anderson, Fisher, & Norman, 2002) and 18
multiple-choice items of the Geoscience Concept Inventory (Li-
barkin & Anderson, 2006).
Finally, in order to explore whether the tendency to explain
nature in teleological terms is theoretical insofar as it derives from
more basic causal intuitions about agency and design in nature
(e.g., Kelemen, 2004; Kelemen & Rosset, 2009) rather than being
a primitive cognitive stance or heuristic (Atran, 1995; Keil, 1992;
Lombrozo & Carey, 2006; Lombrozo et al., 2007), we explored
whether participants with stronger explicit religious beliefs in God
and spiritual “Gaia” beliefs in Mother Nature showed greater
susceptibility to teleological errors. Participants rated their beliefs
in God and “Nature is a powerful being” ona1(strongly disagree)
to5(strongly agree) Likert scale.
Results
Explanation judgment task. We first explored the effects of
restricted processing on endorsements of unwarranted teleological
explanations. A 2 (condition: speeded vs. unspeeded) ⫻ 3 (group:
scientists vs. community vs. college) analysis of variance
(ANOVA) on inaccurate test sentence endorsements
4
revealed
main effects of group, F(2, 302) ⫽ 58.59, p ⬍ .001,
2
⫽ .28, and
condition, F(1, 302) ⫽ 22.57, p ⬍ .001,
2
⫽ .07, with no
interaction. Planned post hoc analyses exploring the group effect
revealed that, replicating prior findings (Kelemen & Rosset, 2009),
college participants displayed substantial acceptance of unwar-
ranted teleological explanations of natural phenomena (M ⫽ 51%,
SD ⫽ 21%), with community participants endorsing them to an
equivalent degree (M ⫽ 47%, SD ⫽ 22%; p ⫽ .23). The physical
scientists’ general acceptance for teleological explanations (M ⫽
22%, SD ⫽ 19%) was lower than that of both control groups (p ⬍
.001 for both). The pattern therefore suggests that maturation does
not decrease the general appeal of teleological explanation, but
becoming a scientist does.
Despite differences between group means, post hoc analyses of
the condition effect revealed that all participant groups, including
the physical scientists, showed higher acceptance of inaccurate
teleological explanations under speed. Notably, speeded scientists’
teleological endorsements (M ⫽ 29%, SD ⫽ 22%) were approxi-
mately twice those of their unspeeded colleagues (M ⫽ 15%,
SD ⫽ 14%). Indeed, as demonstrated by the lack of statistical inter-
action between group and condition, the effect of restricted pro-
cessing remained constant across group differences in education
and age (see Figure 1). In short, even if years of higher education
generally reduce scientifically unwarranted teleological explana-
tions of natural phenomena, an intuitive bias in favor of teleolog-
ical explanations resolutely perseveres. Physical scientists reveal
this tendency when their cognitive resources are taxed.
We examined control sentence performance to see whether the
effect of speeded processing on test sentence endorsement oc-
curred because participants experienced generalized difficulties in
reading and evaluating any kind of explanation at speed. A 2
(condition: speeded vs. unspeeded) ⫻ 2 (sentence type: test vs.
control) ⫻ 3 (group: scientists vs. community vs. college)
4
Test sentences were reviewed by two independent scientific experts
prior to testing and deemed inaccurate. However, during testing, three test
sentences received unexpectedly high levels of endorsement by unspeeded
scientists. Six independent scientific experts were then interviewed about
all test items but produced no consensus view on these three sentences
(while also displaying difficulty justifying judgments that the statements
were accurate, whereas other similarly worded statements were not). To
avoid potentially inflating any report of teleological bias, we have excluded
these sentences from all analyses. It should be noted that including them
does not alter the pattern of results reported here. All of the removed
sentences could be regarded as implying that nature self-regulates in a
Gaia-like fashion (see Study 2, Kelemen & Rosset, 2009).
Table 1
Examples of Test and Control Sentences
Sentence
type Subtype Item
Test Trees produce oxygen so that animals can
breathe.
Germs mutate in order to become drug
resistant.
Moss forms around rocks in order to stop soil
erosion.
The Earth has an ozone layer in order to
protect it from UV light.
The sun makes light so that plants can
photosynthesize.
Control TT Women put on perfume in order to smell
pleasant.
FT Lamps shine brightly so that they can produce
heat.
TC Soda fizzes because carbon dioxide gas is
released.
FC Oceans have waves because they contain a lot
of seawater.
Note. TT ⫽ true teleological; FT ⫽ false teleological; TC ⫽ true causal;
FC ⫽ false causal.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1077
TELEOLOGICAL EXPLANATION IN SCIENTISTS
ANOVA on incorrect responses revealed a main effect of sentence
type, F(1, 302) ⫽ 720.50, p ⬍ .001,
2
⫽ .71. This indicated that
participants erred more often on test sentences (M ⫽ 42%, SD ⫽
24%) than on control sentences (M ⫽ 7%, SD ⫽ 5%), where error
rates were very low. Analysis of a significant Sentence Type ⫻
Condition interaction, F(1, 291) ⫽ 18.61, p ⬍ .001,
2
⫽ .06, was
also consistent with the pattern above: Whereas speeded respond-
ing created substantial disparities in accuracy on test sentences
(14% difference), the disparity on control sentences was not com-
parable (3% difference). The overall pattern of these findings
confirms that participants’ heightened level of error on speeded
teleological test sentences was not a result of general difficulties in
reading and judging all kinds of explanation at speed.
Individual differences. Within each participant group, indi-
viduals randomly assigned to speeded and unspeeded conditions
did not differ on scientific knowledge, inhibitory control, or per-
sonal belief scores. Conditions were therefore collapsed for all
subsequent analyses.
One-way ANOVAs revealed that, as expected, a group differ-
ence emerged with respect to biological content knowledge, F(2,
305) ⫽ 57.75, p ⬍ .001,
2
⫽ .28, and geoscience content
knowledge, F(2, 305) ⫽ 55.38, p ⬍ .001,
2
⫽ .27. Post hoc
analyses confirmed that physical scientists had more biological
and geoscience content knowledge than either college or commu-
nity participants (ps ⬍ .001) who did not differ from each other
(ps ⬎ .48). Importantly, however, despite prior research suggest-
ing relationships between inhibitory control and education (Gan-
guli et al., 2010; Stern et al., 1994), the groups did not differ in
their inhibitory control capacities, F(1, 292) ⫽ 0.29, p ⫽ .75.
Physical scientists’ lower overall level of teleological endorsement
was therefore not a result of having enhanced abilities to inhibit
natural, automatic responses.
Next, we examined whether the groups differed in their
beliefs in God and Mother Nature and found that they did: God,
F(2, 305) ⫽ 14.78, p ⬍ .001,
2
⫽ .09; Mother Nature, F(2,
305) ⫽ 43.20, p ⬍ .001,
2
⫽ .22. Physical scientists were
disinclined to believe in both, giving lower ratings of explicit
belief than both college and community participant groups (ps ⬍
.003), who did not differ from each other (p ⬎ .26; see Table 2).
In sum, physical scientists who, as a group, were less prone to
endorse inaccurate teleological explanations also had a greater
scientific knowledge base and lower explicit beliefs in God and
Mother Nature.
Because the college and community samples did not differ on
any measure, their data were collapsed for individual-differences
analyses within each group (henceforth, the combined group is
referred to as CC). As Table 3 shows, linear regressions of scien-
tific content knowledge and inhibitory control on unwarranted
teleological sentence endorsement (controlling for the effects of
condition) revealed that, in both the CC and scientist groups,
poorer scientific knowledge predicted individual susceptibility to
teleological error (CC: R
2
⫽ .17, p ⬍ .001; scientists: R
2
⫽ .51,
p ⬍ .001). Among CC participants, weaknesses in both biological
knowledge and geoscience knowledge increased test sentence en-
Figure 1. Mean percentage of unwarranted teleological test sentences
accepted (top panel) and incorrect control sentence responses (bottom
panel) by college, community, and scientist participants in Study 1 and
humanities participants in Study 2. Error bars represent standard errors of
the means.
Table 2
Group Means (and Standard Deviations) for Predictor Variables
Group
Biological
knowledge
Geoscience
knowledge
Inhibitory
control
Belief
in God
Belief in
Mother Nature
College 59 (19) 50 (16) 168 (116) 3.3 (1.4) 3.7 (1.2)
Community 61 (20) 48 (17) 179 (126) 3.0 (1.5) 3.9 (1.4)
Scientists 84 (16) 73 (19) 178 (115) 2.2 (1.5) 2.2 (1.5)
Humanities 76 (17) 49 (14) 193 (127) 2.2 (1.4) 2.5 (1.6)
Note. Predictor variables are as follows: percentage correct on the Conceptual Inventory of Natural Selection
(Anderson et al., 2002) and Geoscience Concept Inventory (Libarkin & Anderson, 2006); Stroop inhibitory
control score (milliseconds); Belief in God and Mother Nature (1 ⫽ Strongly disagree;5⫽ Strongly agree).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1078
KELEMEN, ROTTMAN, AND SESTON
