Content uploaded by Jude Weinstein-Jones
Author content
All content in this area was uploaded by Jude Weinstein-Jones on Feb 04, 2014
Content may be subject to copyright.
Evolutionary psychologists (e.g., Tooby & Cosmides,
1992, 2005) propose that our cognitive processes have
been shaped by adaptation. One premise is that given the
ultimate goal of staying alive and reproducing, humans
may have evolved to heighten attention in situations where
their survival is threatened or simply made salient. Fol-
lowing this reasoning, one might further predict that hu-
mans would exhibit enhanced retention for information
that is processed while they are thinking about survival, or
when they are in a survival-mode. In order to test this pre-
diction, Nairne, Thompson, and Pandeirada (2007) asked
participants to imagine a scenario in which they would
have to secure food and shelter and protect themselves
from predators, and then rate words for their relevance
to the scenario. This encoding task led to higher levels
of recall than did control conditions in which the same
words were rated for relevance to moving to a foreign
land, pleasantness, or self-relevance, and the latter two
tasks have been shown in prior research to be among the
best encoding tasks for enhancing later recall and recogni-
tion, as discussed below.
Nairne, Thompson, and Pandeirada’s (2007) findings
are remarkable for several reasons. First, the findings stem
from one of the few paradigms that have been established
to empirically test an evolutionary-based explanation of
memory performance. Second, the findings provide, at
least at first glance, some rather convincing evidence in
favor of the idea that memory for information processed
while one is in a survival mode is superior to memory for
information that is deeply processed by virtue of other en-
coding manipulations. Since Craik and Lockhart’s (1972)
influential article, differences in free recall rates follow-
ing various encoding manipulations have typically been
attributed to depth of processing during encoding. That is,
semantically processed information is generally recalled
at a higher rate than is information processed for shallow
attributes, such as phonemic, visual, or auditory proper-
ties (but see Roediger & Gallo, 2002, for some difficul-
ties with this account). Though there are various means by
which one can deeply encode information, Packman and
Battig (1978) showed that rating words for pleasantness
resulted in significantly higher recall than that obtained
in six other tasks, including rating words for meaning-
fulness. Subsequently, Challis, Velichkovsky, and Craik
(1996) showed that rating words for relevance to oneself
produced higher rates of free recall than did a variety of
other semantic rating tasks. As such, pleasantness and
self-reference rating tasks have generally been perceived
as the most conducive encoding tasks studied in this tra-
dition; it is therefore noteworthy that Nairne, Thompson,
and Pandeirada have revealed the existence of an encoding
task that results in higher levels of free recall than those
found with other tried and true methods.
Because research in this area is just beginning, it is not
entirely clear whether the survival advantage is reliable or
extends to other materials. At the outset of our project, no
reports of replication in independent laboratories or with a
different set of stimuli had appeared. To preface, in Exper-
iment 1, we first established that the survival advantage is
indeed replicable and generalizes to a new set of words.
913 Copyright 2008 Psychonomic Society, Inc.
Can the survival recall advantage be explained
by basic memory processes?
Ya n a W e i n s t e i n
University College London, London, England
a n d
Ju l i e M. Bu g g a n d He n r Y l. ro e d i g e r iii
Washington University, St. Louis, Missouri
Nairne, Thompson, and Pandeirada (2007) demonstrated a striking phenomenon: Words rated for relevance
to a grasslands survival scenario were remembered better than identical words encoded under other deep pro-
cessing conditions. Having replicated this effect using a novel set of words (Experiment 1), we contrasted the
schematic processing and evolutionary accounts of the recall advantage (Experiment 2). Inconsistent with the
schematic processing account, the grasslands survival scenario produced better recall than did a city survival
scenario requiring comparable schematic processing. Recall in the grasslands scenario was unaffected by a
self-reference manipulation. The findings are consistent with an evolutionary account that attributes the recall
advantage to adaptive memory biases.
Memory & Cognition
2008, 36 (5), 913-919
doi: 10.3758/MC.36.5.913
Y. Weinstein, y.weinstein@ucl.ac.uk
914 We i n s t e i n , Bu g g , a n d Ro e d i g e R
situation. Some of the words may be relevant and others may not—
it’s up to you to decide.”
Moving
. “In this task we would like you to imagine that you are
planning to move to a new home in a foreign land. Over the next few
months, you’ll need to locate and purchase a new home and transport
your belongings. We are going to show you a list of words, and we
would like you to rate how relevant each of these words would be for
you in accomplishing this task. Some of the words may be relevant
and others may not—it’s up to you to decide.”
Pleasantness
. “In this task, we are going to show you a list of
words, and we would like you to rate the pleasantness of each word.
Some of the words may be pleasant and others may not—it’s up to
you to decide.”
On each trial of the rating task, a word appeared in the center of
the screen, and participants were asked to rate it using a 5-point
scale where 1 5 totally irrelevant or totally unpleasant and 5 5
extremely relevant or extremely pleasant. The rating scale appeared
on the screen below each presented item, and participants indicated
their response by selecting the appropriate key on the number pad.
Critically, participants were unaware that memory for the items that
they rated would subsequently be tested. They were asked to respond
within 5 sec and reaction times (RTs) were recorded. If a response
was not made within 5 sec, the next trial started automatically and
data from the trial did not contribute to RT and ratings means pre-
sented below. Such trials were extremely rare and resulted in only 10
missing values across both experiments.
After participants rated 12 words in one scenario, they received
instructions for the next rating task, until all three scenarios had
been completed. Following completion, participants performed a
2-min filler task in which they were asked to recall as many U.S.
states as possible (Washington University participants) or calculate
math sums (University College London participants). After the filler
task, participants were asked to write down as many words as they
could remember from the rating tasks, in any order. Ten minutes
were given for this recall task.
Results and Discussion
The primary purpose of Experiment 1 was to examine
the replicability of the survival advantage in free recall
obtained by Nairne, Thompson, and Pandeirada (2007).
Levels of free recall achieved in each of the three encod-
ing conditions are shown in Figure 1. Consistent with
their findings, the survival advantage was obtained using
a within-subjects design. That is, a signif icant main effect
of rating scenario was found for free recall [F(2,142) 5
3.62, MSe 5 223, p , .05].1 Recall was 5.2% greater
for words rated in terms of their relevance to survival as
compared with their relevance to moving [t(71) 5 2.27,
SEM 5 2.30, p , .05] and 6.3% higher than for words
rated for pleasantness [t(71) 5 2.37, SEM 5 2.63, p ,
.05], as indicated by planned paired-samples t tests. This
pattern of findings represents a conceptual replication of
the work of Nairne, Thompson, and Pandeirada and ex-
tends the survival advantage to a novel set of word stimuli,
obviating concern that the effect might be bound to a spe-
cific set of stimuli.
To confirm that the recall pattern that we obtained was
not driven by differences between the conditions in either
RTs to produce ratings or mean ratings, we conducted
within-subjects ANOVAs on the mean RT and ratings data,
which are shown in Table 1.2 Rating scenario had a signifi-
cant effect on RTs [F(2,136) 5 6.58, MSe 5 80,725, p ,
.01]; post hoc comparisons indicated that the times taken
to rate items for relevance to survival and moving did not
Experiment 2 was designed to test a more theoretically
interesting question: By what process is memory perfor-
mance facilitated when participants evaluate the relevance
of words to a survival scenario? Nairne, Thompson, and
Pandeirada’s (2007) view is that the advantage reflects a
deeply rooted adaptive bias that is activated when partici-
pants are asked to read the survival scenario and evaluate
words on its basis. Another view is that the survival pro-
cessing advantage reflects the operation of basic memory
processes such as enhanced schematic or self-referential
processing, not abstract, evolutionary biases. In Experi-
ment 2, we contrasted these competing accounts.
EXPERIMENT 1
Participants rated lists of 12 words in three different
encoding conditions and later performed a surprise free
recall test. Two of the encoding conditions involved rating
words for their relevance to a particular scenario (survival
or moving), whereas the other condition involved rating
words for pleasantness. The encoding and recall tasks were
similar in nature to those used in Nairne, Thompson, and
Pandeirada (2007); no changes were made to the word-
ing of the orienting tasks, although we used a completely
within-subjects design incorporating all three rating sce-
narios and a new set of word stimuli. Nairne, Thompson,
and Pandeirada had not included all three orienting tasks
in a within-subjects design, so this experiment represents
a modest extension as well as replication of prior work.
Method
Participants
Seventy-two paid participants took part in this experiment, of
whom 36 were undergraduate students from Washington University
in St. Louis and 36 were volunteers recruited at University College
London.
Materials and Design
Stimuli were 36 concrete nouns randomly selected from a list of
words that were generated by submitting a range of 400–700 for the
frequency, concreteness, imageability, and meaningfulness criteria
in the MRC Psycholinguistic Database (see the Appendix). This list
was then randomly split into three lists of 12 words. As can be seen,
we made no special effort to select words for survival relevance.
A within-subjects design was used, with rating scenario (survival/
moving/pleasantness) as the independent variable. All participants
studied the same three lists of 12 words, with the order of lists, order
of rating scenario, and assignment of word lists to scenario coun-
terbalanced, though the order of words within each list was held
constant. Altogether, this produced 36 counterbalancing conditions,
and 2 participants were assigned to each of these conditions.
Procedure
Participants were informed that they would be rating words on
the basis of various characteristics. They were then given one of the
following three rating instructions, which were identical in word-
ing to those used in Nairne, Thompson, and Pandeirada’s (2007)
experiments.
Survival
. “In this task we would like you to imagine that you are
stranded in the grasslands of a foreign land, without any basic sur-
vival materials. Over the next few months, you’ll need to find steady
supplies of food and water and protect yourself from predators. We
are going to show you a list of words, and we would like you to rate
how relevant each of these words would be for you in this survival
su R v i va l Re c a l l ad va n t a g e 915
words rated for pleasantness as opposed to survival, alter-
native explanations of the survival advantage based simply
on scenario-word relevance may be discounted.
To summarize, Experiment 1 provided a conceptual
replication of the results of Nairne, Thompson, and Pan-
deirada (2007), demonstrating a significant memory
advantage for survival over moving and pleasantness en-
coding using a novel set of words and a completely within-
subjects design. Experiment 2 was conducted to evaluate
the processes that underlie this memory advantage.
EXPERIMENT 2
One specific account of the recall advantage is that
evaluating words for their relevance to the survival sce-
nario encourages greater schematic processing than do
conditions such as rating pleasantness, and this in turn
facilitates memory performance (Bransford & Johnson,
1972). A straightforward evaluation of the schematic pro-
cessing account would pit the survival scenario against
a second scenario that is presumed to encourage equally
rich schematic processing. Nairne, Thompson, and Pan-
deirada (2007) attempted to do this by using a control task
in which words were rated for their relevance to moving
to a foreign land. The survival scenario produced supe-
rior retention when compared with the moving scenario,
suggesting that the advantage is not simply the result of
enhanced schematic processing. This evidence might be
considered tentative, however, because it could be argued
that the two scenarios were not fully equated for schematic
processing. Whereas the survival scenario engages par-
ticipants in a rather unusual context, likely forcing them
to perform online schematic processing to assign the rel-
evance of each item, the moving scenario is not as novel.
Rather than engaging in schematic processing online, par-
ticipants could rely on existing schemas and stored experi-
ences involving moving to rate words for relevance to the
moving scenario.
An alternative approach for evaluating a schematic pro-
cessing account is to use two scenarios that involve almost
identical wording, with one referencing a context that evo-
lutionary theorists would anticipate as inducing the sur-
vival mode (surviving predators in the grasslands) and the
other involving a context that would be less likely to induce
this ancestral survival mode (surviving attackers in a city).
This approach results in two similar scenarios that are more
equally matched in the level of online schematic process-
ing required. If the previously observed recall advantage
simply reflects disproportionately greater engagement in
schematic processing while studying the survival scenario
relative to prior comparison conditions (e.g., moving), then
the advantage should be negated when one compares the
two present scenarios, which are closely matched in the
level of schematic processing required. Indeed, because
college students are more accustomed to living in cities
than in grasslands, one might predict that they could use
schemas to elaborate words and concepts more readily in
this scenario than in the grasslands survival scenario. If
so, one might even expect better recall in the city survival
scenario than in the grasslands survival scenario.
differ ( p 5 .44), whereas pleasantness ratings were made
significantly faster than survival ratings [t(68) 5 2.85,
SEM 5 45, p , .01]. Although one might argue that the
additional time spent producing the ratings in the survival
condition relative to the pleasantness condition might ac-
count for the survival recall advantage over pleasantness,
we believe that a simple processing time explanation of
this sort is unlikely. Such an explanation would also predict
equivalent recall levels for the survival and moving condi-
tions, because the time taken to produce ratings in the sur-
vival and moving conditions was equivalent. As confirmed
earlier, this was not the case. In addition, prior research
in the levels-of-processing tradition has shown that type
of task is critically important in producing differences in
recall, and time spent on the encoding task has negligible
effects (Craik & Tulving, 1975; Hyde & Jenkins, 1973).
Rating scenario also had a significant effect on mean rat-
ings [F(2,136) 5 33.07, MSe 5 0.17, p , .001]. Although
survival and moving ratings did not differ (p 5 .49), words
were rated as significantly more pleasant than relevant to
survival [t(68) 5 7.34, SEM 5 0.06, p , .001]. Clearly,
the latter two rating tasks cannot be treated as equivalent:
In the case of the survival task, ratings are made in relation
to a particular scenario, whereas pleasantness ratings are
made by assigning valences. In both cases, however, we
might expect that higher ratings would lead to higher levels
of recall because of a congruity effect (Craik & Tulving,
1975). Because the pattern of ratings that we found here
would be expected to enhance, not attenuate, memory for
Figure 1. Mean proportion of words recalled by rating scenario
in Experiment 1. The error bars represent standard errors of the
means.
.10
.20
.30
.40
.50
.60
Survival Moving Pleasantness
Rating Scenario
Mean Proportion Recalled
Table 1
Mean Ratings and Reaction Times (RTs, in Milliseconds)
by Rating Scenario in Experiment 1
Rating Scenario RT SE Rating SE
Survival 1,618 49 2.62 0.06
Moving 1,656 57 2.50 0.07
Pleasantness 1,489 40 3.04 0.05
916 We i n s t e i n , Bu g g , a n d Ro e d i g e R
each of the two rating tasks (survival, moving) twice, and counter-
balancing was achieved by rotating the order of scenarios across
two conditions (moving, survival, moving, survival and survival,
moving, survival, moving).
Procedure
The overall procedure and instructions were identical to those of
Experiment 1, except that participants received new rating instruc-
tions after every nine words. The 1st person moving and 1st person
grasslands survival scenarios were identical to those used by Nairne,
Thompson, and Pandeirada (2007) and by us in Experiment 1. The
rest of the instructions were created by changing as few words as
possible in order to retain the grammatical and syntactical structure
of the text. Instructions for all the conditions are presented below.
Note that italics are used here purely to draw attention to the differ-
ences between rating scenarios and were not presented in the instruc-
tions that the participants received.
Grasslands survival (1st person)
. “In this task we would like
you to imagine that you are stranded in the grasslands of a for-
eign land, without any basic survival materials. Over the next few
months, you’ll need to f ind steady supplies of food and water and
protect yourself from predators. We are going to show you a list of
words, and we would like you to rate how relevant each of these
words would be for you in this survival situation. Some of the words
may be relevant and others may not—it’s up to you to decide.”
Grasslands survival (3rd person)
. “In this task we would like
you to imagine that a friend is stranded in the grasslands of a for-
eign land, without any basic survival materials. Over the next few
months, they’ll need to find steady supplies of food and water and
protect themselves from predators. We are going to show you a list
of words, and we would like you to rate how relevant each of these
words would be for them in this survival situation. Some of the words
may be relevant and others may not—it’s up to you to decide.”
City survival (1st person)
. “In this task we would like you to
imagine that you are stranded in the city of a foreign land, without
any basic survival materials. Over the next few months, you’ll need
to find steady supplies of food and water and protect yourself from
attackers. We are going to show you a list of words, and we would
like you to rate how relevant each of these words would be for you in
this survival situation. Some of the words may be relevant and others
may not—it’s up to you to decide.”
City survival (3rd person)
. “In this task we would like you to
imagine that a friend is stranded in the city of a foreign land, without
any basic survival materials. Over the next few months, they’ll need
to find steady supplies of food and water and protect themselves
from attackers. We are going to show you a list of words, and we
would like you to rate how relevant each of these words would be for
them in this survival situation. Some of the words may be relevant
and others may not—it’s up to you to decide.”
Moving (1st person)
. “In this task we would like you to imag-
ine that you are planning to move to a new home in a foreign land.
Over the next few months, you’ll need to locate and purchase a new
home and transport your belongings. We are going to show you a
list of words, and we would like you to rate how relevant each of
these words would be for you in accomplishing this task. Some of
the words may be relevant and others may not—it’s up to you to
decide.”
Moving (3rd person)
. “In this task we would like you to imagine
that a friend is planning to move to a new home in a foreign land. Over
the next few months, they’ll need to locate and purchase a new home
and transport their belongings. We are going to show you a list of
words, and we would like you to rate how relevant each of these words
would be for them in accomplishing this task. Some of the words may
be relevant and others may not—it’s up to you to decide.”
After the recall task, participants completed a Shipley vocabu-
lary test (not analyzed here) and a demographic questionnaire. They
were also asked to indicate the frequency with which they watched
survival-related television programs (e.g., Survivor, Lost) using the
following scale: never, sometimes, frequently, always.
On the other hand, the evolutionary account generates
an opposing prediction. This line of thinking proposes that
our attitude toward survival has been shaped by evolution,
and thus threats to survival that were endured by our ances-
tors are more salient to us than threats that have appeared
relatively recently in our evolutionary history (Öhman &
Mineka, 2001). Early research into phobias suggested that
humans are more likely to be phobic regarding predators
and open spaces than guns and dangerous vehicles even
though the latter are more realistic threats in the modern
environment (Marks, 1969; Seligman, 1971; but see Fox,
Griggs, & Mouchlianitis, 2007, for recent data on a com-
peting account). The prediction that follows from this rea-
soning is that certain survival scenarios should activate
the adaptive bias more strongly than others. Accordingly,
in the present experiment, the evolutionary account an-
ticipates recall to be greater when participants are encour-
aged to think about survival in the grasslands rather than
survival in a more modern context such as a city.
A secondary goal of Experiment 2 was to evaluate an-
other explanation of the recall advantage that, like the
schematic processing account, is rooted in existing mem-
ory principles. This account attributes the recall advan-
tage to greater engagement in self-referential processing
in the survival scenario than in control conditions. One
way in which self-referential processing is thought to af-
fect memory is that it facilitates high levels of elaboration
(Rogers, Kuiper, & Kirker, 1977). The idea is that when
relating a concept to ourselves, we are able to engage in
more item-specific processing because we are so profi-
cient at thinking about the self (Kihlstrom, 1993). If the
survival advantage is partially dependent upon engage-
ment in self-referential processing, one might hypothesize
that it should weaken if participants are asked to consider
the relevance of words to someone else’s survival instead
of their own. To evaluate this prediction, we also manipu-
lated perspective in Experiment 2 in such a way that in-
structions were either phrased in the 1st person (imagine
yourself in this scenario) or in the 3rd person (imagine a
friend in this scenario).
Method
Participants
Eighty-eight participants were drawn from the University College
London participants’ pool and either were paid the equivalent of $4
for their time or volunteered their time without monetary reward.
Paid/unpaid participants were split evenly among all conditions.
Materials and Design
We manipulated both survival context (city, grasslands) and per-
spective (1st person, 3rd person) between subjects: Participants
rated words for relevance either to survival in the grasslands (n 5
40) or to survival in the city (n 5 48); and half of each of these
groups imagined themselves in this survival scenario, whereas the
other half imagined a friend in the scenario. All participants also
rated words for relevance to a moving scenario as an additional con-
trol to establish the basic survival advantage. Moving instructions
remained the same regardless of survival scenario, but varied as to
whether participants imagined themselves or a friend to match the
survival scenario. We used the same 36 words from Experiment 1
and distributed them into four blocks of 9 words. The words were
rated in the same order by all participants. Participants performed
su R v i va l Re c a l l ad va n t a g e 917
nario in the 3rd person actually performed 5.3% better
than those who read the same scenario in the 1st person.
Neither of these comparisons approached significance
(ps . .25). Obviously, the pattern of data showing, if any-
thing, greater recall in the 3rd person than in the 1st per-
son grasslands survival scenario cannot be accommodated
by the notion that the basic survival advantage is created
solely by self-referential processing.
In order to verify that the obtained pattern of recall was
not dependent on ratings or RTs,4 we performed ANOVAs
on the mean RT and ratings data with survival context (city,
grasslands) and perspective (1st, 3rd person) as between-
subjects variables. The data are shown in Table 2. This
analysis yielded no significant comparisons (all ps . .14):
Scenario instructions did not affect mean ratings or time
spent making these ratings, suggesting that the different
conditions were indeed comparable.
In summary, Experiment 2 demonstrated that the recall
advantage originally obtained by Nairne, Thompson, and
Pandeirada (2007) persists even when the control condition
is closely matched for schematic processing. This finding
is particularly striking because the difference between en-
coding instructions was limited to two words, with city re-
placing grasslands and attackers replacing predators. This
recall pattern is anticipated by the evolutionary account of
the survival advantage and not the schematic processing ac-
count. As previously explicated, the evolutionary account
Results and Discussion
We first compared mean recall for the three instructional
conditions. The grasslands survival condition produced
highest recall (.38), the city survival condition produced
next best recall (.31), and the moving condition produced
the least recall (.28). We confirmed that an overall sur-
vival advantage (mean of grasslands and city survival)
over moving was again obtained: Across all 88 partici-
pants, recall was on average 6.7% higher for words that
were rated for relevance to any survival scenario than for
words that were rated for relevance to the moving scenario
[F(1,87) 5 15.38, MSe 5 128, p , .001]. Of the 88 par-
ticipants, 53 showed this advantage, while 21 showed the
opposite pattern; an exact binomial sign test confirmed
that these proportions were significantly different ( p ,
.001). Furthermore, the survival advantage remained sig-
nificant in the 42 participants who reported never watch-
ing survival-related television programs [F(1,41) 5 7.30,
MSe 5 158, p , .05]. The overall finding of an advantage
of survival over moving conceptually replicates that of
Experiment 1; rates of recall in the moving condition were
excluded from all further analyses. As discussed below,
the grasslands survival scenario produced better recall
than did the city survival scenario.
Of primary interest in Experiment 2 were two theoreti-
cally motivated questions. The first considered the degree
to which disproportionate schematic processing accounts
for the survival recall advantage; the second considered
the role of self-referential processing in the effect. To
preface, the survival recall advantage persisted when the
control condition was matched for the degree of schematic
processing required. Furthermore, this advantage was gen-
erally similar for both the 1st and 3rd person perspectives,
disfavoring a pure self-referential processing account.
The recall data for the relevant conditions are shown
in Figure 2. We carried out a 2 2 2 ANOVA with
survival context (city, grasslands), perspective (1st, 3rd
person), and incentive (paid, unpaid) as between-subjects
variables.3 This analysis revealed a main effect of survival
context [F(1,80) 5 9.72, MSe 5 162, p , .005]: Recall
was 7.8% higher for words rated for relevance to grass-
lands survival than for words rated for relevance to city
survival. Instructions that differed from the grasslands
survival scenario only in terms of evolutionary relevance
were significantly less conducive toward subsequent re-
call, disfavoring a schematic processing account.
As for the self-reference manipulation, no main effect
of perspective on recall was found (F , 1), although there
was a marginal interaction between survival context and
perspective [F(1,80) 5 3.12, MSe 5 162, p 5 .08]. Our
hypothesis regarding this manipulation was that if self-
referential processing is responsible for the survival ad-
vantage, there should be a reduction in recall when the
instructions ask participants to imagine someone other
than themselves in the survival scenario. Although this
was indeed the case in the city survival condition, with
participants who imagined themselves in this scenario re-
calling 3.0% more words than did those who imagined a
friend, participants who read the grasslands survival sce-
Figure 2. Mean proportion of words recalled by rating scenario
in Experiment 2. The error bars represent standard errors of the
means.
.10
.20
.30
.40
.50
.60
Grasslands Survival City Survival
Encoding Condition
Mean Proportion Recalled
1st person
3rd person
Table 2
Mean Ratings and Reaction Times (RTs, in Milliseconds)
by Rating Scenario in Experiment 2
Rating Scenario RT SE Rating SE
Survival: Grasslands (n 5 40)
1st person 1,761 111 2.51 0.14
3rd person 1,818 92 2.54 0.10
Survival: City (n 5 48)
1st person 1,640 108 2.49 0.14
3rd person 1,641 90 2.28 0.10
918 We i n s t e i n , Bu g g , a n d Ro e d i g e R
readily accessible schemas (e.g., Bransford & Johnson,
1972). This outcome clearly did not occur.
Although we have disconfirmed the schematic process-
ing account, another account based on existing memory
principles remains to be tested. The difference in recall
between the grasslands and city survival conditions might
be explained in terms of a proxy mechanism involving
emotional arousal. It is well established that emotional
arousal can lead to enhanced retention. The grasslands
survival scenario may have evoked greater levels of emo-
tional arousal, boosting recall relative to the potentially
less arousing city survival scenario. Such an explanation
is not, however, incompatible with an evolutionary ac-
count. It has been proposed, for example, that our fear
module is responsive to evolutionarily primed stimuli
(Öhman & Mineka, 2001). Similarly, an evolutionary per-
spective posits that it would be adaptive to retain informa-
tion about emotionally arousing stimuli in order to react
to them more appropriately in the future (for a review, see
Hamann, 2001). Further research is needed in order to
tease apart these explanations.
The third finding that we established in the present study
is that a self-reference manipulation (asking participants
to imagine a friend instead of themselves in the survival
scenario) had little effect on the levels of recall achieved.
Similarly, Nairne, Thompson, and Pandeirada (2007, Ex-
periment 4) found a significant memory advantage for
words rated for relevance to survival in comparison with
words rated for relevance to oneself. Taken together, the
implication is that the survival advantage cannot be ex-
plained simply in terms of the degree to which the encod-
ing task induces self-referential processing. Interestingly,
while preparing this manuscript, we became aware of two
additional pieces of research that converge with our own.
Kang, McDermott, and Cohen (in press) have found that
survival processing enhances retention even when words
are rated for relevance to a character’s survival in a video
clip. Furthermore, Nairne, Pandeirada, and Thompson
(2008) have shown that survival processing is better than
a host of “deep” processing conditions, including self-
reference. These data further strengthen the conclusion
that a self-reference explanation of the survival advantage
does not suffice.
One caveat of the finding above is that some research
indicates that the effect of a self-reference manipulation
may be related to the degree of intimacy of the subject
in the 3rd person instructions (e.g., Bower & Gilligan,
1979), although a meta-analysis found an overall signifi-
cant effect of self-reference across 17 studies that used a
highly intimate 3rd person (e.g., one’s mother; Symons &
Johnson, 1997). In order to counter this concern, we ran
an additional 18 participants in the 3rd person grasslands
survival condition, with identical instructions except that
a friend was replaced with a stranger. There was no dif-
ference in recall of words rated in the survival scenario
between this condition and the analogous 1st person con-
dition ( p 5 .76). Thus, the recall advantage does not ap-
pear to be limited to thinking about oneself in a survival
scenario, but rather persists when words are processed for
relevance to a friend or even a stranger.
predicts that survival contexts that would have been more
relevant to our ancestors (e.g., grasslands) should activate
the adaptive memory bias more strongly than those that
would have been less relevant (e.g., modern city). Experi-
ment 2 also demonstrated that the recall advantage was not
limited to thinking about one’s own survival. We further
consider the implications of these findings for both the
evolutionary and schematic processing accounts below.
GENERAL DISCUSSION
In two experiments, we have established three primary
points. First, using a novel set of word stimuli, we have
shown that processing words for relevance to survival results
in better recall than either pleasantness ratings or process-
ing words for relevance to a moving scenario. This finding
confirms the survival recall advantage originally demon-
strated by Nairne, Thompson, and Pandeirada (2007).
Second, we have shown that the recall advantage per-
sists when the control condition is closely matched for the
level of schematic processing required but differs from
the original scenario in terms of evolutionary relevance.
Specifically, rating words for relevance to survival in a
modern, urban environment appears to be less conducive
to recall than does rating words for relevance to survival in
the grasslands. It is thus unlikely that the survival advan-
tage can be explained in terms of the schematic process-
ing account, which posits that the scenario instructions
create a schema that is conducive toward elaboration and
therefore leads to better encoding and retrieval. Instead,
our finding is in line with an evolutionary account, ac-
cording to which our memory has adapted very well to
self-preservation in the type of setting in which we might
have found ourselves until very recently (i.e., the grass-
lands), but has not yet evolved to function optimally in
more modern contexts (i.e., the city). This is a reasonable
prediction in light of the recent appearance of city-like
urban settlements just over 5,000 years ago (“City,” 2007),
an insignificant event in terms of the larger evolutionary
time scale of our development.
By pitting the grasslands survival scenario against an
analogous scenario set in the city, we eliminated some of
the apparent differences between the survival and mov-
ing scenarios. We more closely matched the two scenar-
ios on at least four factors: the subject matter to which
the scenarios referred; the grammatical structure and
vocabulary of the scenario descriptions; the emotional
arousal associated with the scenarios; and participants’
first-hand experience with each scenario. In this way, the
city scenario is an improvement over the original moving
scenario. Nonetheless, we acknowledge that disagree-
ment may remain regarding the equivalence of the city
and grasslands survival scenarios. However, students
in St. Louis and London are presumably more familiar
with cities than with grasslands. If students applied and
elaborated relevant schemas more readily in the former
than in the latter scenario, one might have expected a city
survival advantage to emerge rather than a grasslands
survival advantage, as in other studies that have shown
better retention for material when it can be encoded by
su R v i va l Re c a l l ad va n t a g e 919
Öhman, A., & Mineka, S. (2001). Fears, phobias, and preparedness:
Toward an evolved module of fear and fear learning. Psychological
Review, 108, 483-522.
Packman, J. L., & Battig, W. F. (1978). Effects of different kinds of
semantic processing on memory for words. Memory & Cognition, 6,
502-508.
Roediger, H. L., III, & Gallo, D. A. (2002). Levels of processing:
Some unanswered questions. In M. Naveh-Benjamin, M. Moscovitch,
& H. L. Roediger III (Eds.), Perspectives on human memory and cog-
nitive aging: Essays in honour of Fergus Craik (pp. 28-47). New York:
Psychology Press.
Rogers, T. B., Kuiper, N. A., & Kirker, W. S. (1977). Self-reference
and the encoding of personal information. Journal of Personality &
Social Psychology, 35, 677-688.
Seligman, M. E. P. (1971). Phobias and preparedness. Behavior Ther-
apy, 2, 307-320.
Symons, C. S., & Johnson, B. T. (1997). The self-reference effect in
memory: A meta-analysis. Psychological Bulletin, 121, 371-394.
Tooby, J., & Cosmides, L. (1992). The psychological foundations
of culture. In J. H. Barkow, L. Cosmides, & J. Tooby (Eds.), The
adapted mind: Evolutionary psychology and the generation of culture
(pp. 19-136). New York: Oxford University Press.
Tooby, J., & Cosmides, L. (2005). Conceptual foundations of evolution-
ary psychology. In D. M. Buss (Ed.), The handbook of evolutionary
psychology (pp. 5-67). Hoboken, NJ: Wiley.
NOTES
1. Note that for all analyses described below, a 3 2 mixed subjects
ANOVA with rating scenario and sample (United Kingdom vs. United
States) as factors was initially conducted for each of the three dependent
measures. The main effect of sample was not significant for any measures,
nor was the interaction of sample and rating scenario (all ps . .1). There-
fore, we collapsed across the two samples for all reported analyses.
2. Because of a programming error, RT and ratings data are missing
for 3 participants.
3. There was a significant main effect of incentive, showing that paid
participants recalled 6.3% more words from the survival scenario than
did unpaid volunteers [F(1,80) 5 6.92, MSe 5 5.46, p , .05]. This factor
did not interact with any other factors, and we therefore collapsed across
this variable in subsequent analyses.
4. Ratings data were lost for 1 participant because of a programming
error.
In summary, then, the adaptive survival advantage in free
recall appears to be reliable and robust to changes in stimuli,
rating tasks, and perspective. Critically, the recall advantage
cannot be explained on account of at least two basic mem-
ory principles, schematic processing and self-reference. The
primary finding, that recall is significantly higher when the
instructions orient participants to an evolutionarily relevant
(grasslands) as opposed to a modern (city) context accords
nicely with an adaptive memory explanation of the sort pos-
ited by Nairne, Thompson, and Pandeirada (2007; see also
Nairne & Pandeirada, in press, for further discussion of this
approach to the study of memory).
AUTHOR NOTE
This research was partially supported by BBSRC Studentship
07322/-101970 and a generous grant from the Bogue Fellowship, Uni-
versity College London. J.M.B. was supported by National Institute
on Aging Grant 5T32AG00030 during the writing of this article. The
authors thank James Nairne for comments on an earlier version of the
manuscript. Correspondence concerning this article should be addressed
to Y. Weinstein, Department of Psychology, University College London,
Gower Street, London WC1E 6BT, England (e-mail: y.weinstein@ucl
.ac.uk).
REFERENCES
Bower, G. H., & Gilligan, S. G. (1979). Remembering information
related to one’s self. Journal of Research in Personality, 13, 420-432.
Bransford, J. D., & Johnson, M. K. (1972). Contextual prerequisites
for understanding: Some investigations of comprehension and recall.
Journal of Verbal Learning & Verbal Behavior, 11, 717-726.
Challis, B. H., Velichkovsky, B. M., & Craik, F. I. M. (1996). Levels-
of-processing effects on a variety of memory tasks: New findings and
theoretical implications. Consciousness & Cognition, 5, 142-164.
“City” (2007). In Encyclopædia Britannica. Retrieved May 29, 2007,
from Encyclopædia Britannica Online: www.britannica.com/eb/
article-9106247.
Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A
framework for memory research. Journal of Verbal Learning &Verbal
Behavior, 11, 671-684.
Craik, F. I. M., & Tulving, E. (1975). Depth of processing and the
retention of words in episodic memory. Journal of Experimental Psy-
chology: General, 104, 268-294.
Fox, E., Griggs, L., & Mouchlianitis, E. (2007). The detection of
fear-relevant stimuli: Are guns noticed as quickly as snakes? Emo-
tion, 7, 691-696.
Hamann, S. (2001). Cognitive and neural mechanisms of emotional
memory. Trends in Cognitive Sciences, 5, 394-400.
Hyde, T. S., & Jenkins, J. J. (1973). Recall for words as a function of
semantic, graphic, and syntactic orienting tasks. Journal of Verbal
Learning & Verbal Behavior, 12, 471-480.
Kang, S. H. K., McDermott, K. B., & Cohen, S. M. (in press). The
mnemonic advantage of processing fitness-relevant information.
Memory & Cognition.
Kihlstrom, J. F. (1993). What does the self look like? In T. K. Srull & R. S.
Wyer, Jr. (Eds.), The mental representation of trait and autobiographi-
cal knowledge about the self (pp. 79-90). Hillsdale, NJ: Erlbaum.
Marks, I. M. (1969). Fears and phobias. London: Heineman.
Nairne, J. S., & Pandeirada, J. N. S. (in press). Adaptive memory:
Remembering with a stone-age brain. Current Directions in Psycho-
logical Science.
Nairne, J. S., Pandeirada, J. N. S., & Thompson, S. R. (2008). Adap-
tive memory: The comparative value of survival processing. Psycho-
logical Science, 19, 176-180.
Nairne, J. S., Thompson, S. R., & Pandeirada, J. N. S. (2007).
Adaptive memory: Survival processing enhances retention. Journal
of Experimental Psychology: Learning, Memory, & Cognition, 33,
263-273.
APPENDIX
a i R m a c a R o n i
a u t h o R m a g a z i n e
B a c t e R i a o p i u m
B o d y p e a c e m a k e R
B o u l d e R p i a n o
B u n g a l o W p i p e
c a n d y p R i e s t
c e l l a R p R o p e Rt y
c o n t R a c t s h R i e k
c R a d l e s i c k n e s s
d o l l s k i n
e n g i n e s l i p p e R
g e m s q u a R e
g l u t t o n s t R a W B e R Ry
g o l d s u n s e t
h o s t ag e t i m e p i e c e
i n s t i t u t e t o m B
j o u R n a l W o R l d
(Manuscript received November 26, 2007;
revision accepted for publication January 22, 2008.)