Content uploaded by Meike Kroneisen
Author content
All content in this area was uploaded by Meike Kroneisen on Sep 12, 2018
Content may be subject to copyright.
RESEARCH REPORT
On the Plasticity of the Survival Processing Effect
Meike Kroneisen and Edgar Erdfelder
University of Mannheim
Nairne, Thompson, and Pandeirada (2007) discovered a strong and rather general memory advantage for word
material processed in a survival-related context. One possible explanation of this effect conceives survival
processing as a special form of encoding: Nature specifically “tuned” our memory systems to process and
remember fitness-relevant information. We tested this explanation by studying whether the survival process-
ing effect is robust against encoding manipulations that do not affect the fitness relevance of information.
Three experiments replicated a strong survival processing effect under standard conditions but showed that the
mnemonic benefit of survival processing diminishes or even vanishes when participants focus on a single
problem (Experiments 1 and 2) or technique (Experiment 3) of survival. We argue that it is not survival
processing per se that facilitates recall but the richness and distinctiveness with which information is encoded.
Keywords: memory, evolution, survival processing effect, recall
The evolution of Homo sapiens shaped human memory to solve
important adaptive problems in everyday life of our ancestors. This
view has been endorsed by evolutionary psychologists (e.g., Klein
et al., 2009; Klein, Cosmides, Tooby, & Chance, 2002; Klein,
Robertson, & Delton, 2010; Tooby & Cosmides, 1992, 2005) and
is shared by many cognitive psychologists nowadays. According to
Anderson and Schooler (1991, 2000), for example, our memory
systems are optimally designed to reflect the statistical structure of
the environment. Even the apparently maladaptive mechanism of
forgetting may prove adaptive in certain contexts (Schooler &
Hertwig, 2005). Following these ideas, Nairne, Thompson, and
Pandeirada (2007) suggested that nature “tuned” our memory
systems to process and remember fitness-relevant information.
Because of these properties of human memory, our ancestors had
survival advantages and were more likely to reproduce.
To test this prediction, Nairne et al. (2007) asked their partici-
pants to imagine a survival scenario (being stranded on the grass-
lands of a foreign land) and then rate a number of objects with
respect to their relevance in this situation. A subsequent surprise
retention test showed that survival-based processing resulted in
significantly better retention than other highly effective encoding
procedures such as pleasantness ratings of words (Packman &
Battig, 1978), self-relevance ratings (Symons & Johnson, 1997) or
non-survival-based processing of otherwise comparable scenarios
(i.e., moving to a foreign land). The mnemonic benefit of survival
processing could also be established relative to other, more novel
and exciting control scenarios, for example, planning and execut-
ing a bank heist (Kang, McDermott, & Cohen, 2008) or vacation-
ing to a fancy resort (Nairne, Pandeirada, & Thompson, 2008).
Moreover, and again in line with the predictions, direct compari-
son of a city survival scenario (not relevant for our hunter–
gatherer ancestors) with a grasslands survival scenario (highly
relevant for our hunter–gatherer ancestors) yielded superior mem-
ory performance in the latter (Nairne & Pandeirada, 2010; Wein-
stein, Bugg, & Roediger, 2008).
These results show that the survival processing effect is real,
replicable, and stronger than effects of several other powerful
encoding procedures (see Nairne, 2010, for a recent review).
However, as argued by Nairne and Pandeirada (2008a, p. 378),
“these experiments have revealed very little about the proximate
mechanisms that actually produce the survival benefit. Is the
survival processing special, arising from the action of some kind of
special mnemonic adaptation, or can we explain the advantage
using traditional explanatory tools? For example, one might claim
that survival processing is simply another form of ‘deep process-
ing,’ albeit a particularly good one, leading to enhanced elabora-
tion or distinctive encodings.”
Surprisingly, to the best of our knowledge, this latter idea has
not been subjected to empirical tests so far. In the present article,
we thus took up Nairne and Pandeirada’s (2008a) suggestion and
analyzed the survival processing effect in the depth of processing
framework originally put forward by Craik and Lockhart (1972).
More precisely, we refer to the “richness of encoding” and “cue
distinctiveness” perspectives that have evolved from the depth of
processing framework (Craik & Tulving, 1975; Hunt, 2003; Hunt
& Smith, 1996; Hunt & Worthen, 2006; Moscovitch & Craik,
1976; Watkins, 1978; Watkins & Watkins, 1975). Craik and Tulv-
ing (1975) argued that the crucial factor in depth of processing is
not merely semantic encoding of information—as previously sug-
gested by Craik and Lockhart (1972) — but rather the richness and
distinctiveness with which information is encoded. Consistent with
this claim, they demonstrated that using semantically more com-
This article was published Online First July 25, 2011.
Meike Kroneisen and Edgar Erdfelder, Department of Psychology, Uni-
versity of Mannheim.
We would like to thank the Social Psychology and Methodology De-
partment of the University of Freiburg for support in data collection. We
are also grateful to Benjamin E. Hilbig for helpful comments and sugges-
tions on a draft of this article.
Correspondence concerning this article should be addressed to Meike
Kroneisen, Psychology III, University of Mannheim, Mannheim D-68131,
Germany. E-mail: kroneisen@psychologie.uni-mannheim.de
Journal of Experimental Psychology: © 2011 American Psychological Association
Learning, Memory, and Cognition
2011, Vol. 37, No. 6, 1553–1562 0278-7393/11/$12.00 DOI: 10.1037/a0024493
1553
plex sentence frames for sentence compatibility judgments of a list
of words facilitates subsequent recall of these words. Compared
with simple sentences, complex sentences provide more opportu-
nities to generate a multitude of unique idiosyncratic associations
for each to-be-remembered word during encoding, which, in turn,
may later serve as powerful retrieval cues. In a similar vein,
Moscovitch and Craik (1976) showed that semantic encoding
questions that are unique for each to-be-remembered word en-
hance later recall more than encoding questions shared by several
words of the list. A similar phenomenon, called cue overload effect
(Watkins & Watkins, 1975, 1976), has been established with
respect to retrieval cues. The larger the number of items associated
with the same retrieval cue, the worse the recall of these items
(Watkins, 1978). Following these ideas, Hunt and Smith (1996)
explicitly manipulated the number of cues shared by a target.
Unique cue–target relationships produced highest levels of recall,
particularly when the cues were self-generated during encoding.
The terms distinctive, elaborative processing and richness of
encoding are often used to indicate the number of unique cues
encoded jointly with each to-be-remembered item. As clearly
shown in the depth-of-processing research tradition, unique cues,
if activated in the retrieval context, boost recall. A survival sce-
nario as studied by Nairne and co-workers can be seen as fostering
distinctive processing and elaborative encoding. When evaluating
the relevance of words in a complex grasslands survival context,
participants may easily think of a number of different perceptual
and functional object features or other possible associations to a
word. Perhaps most important, many possible uses of the objects
indicated by the words come to mind, enhancing the pragmatic
distinctiveness of the words. Consequently, each single word can
be connected to many self-generated cues that subsequently pro-
mote successful recall in the retrieval context. If this reasoning is
correct, then it is not the evolutionary significance of survival per
se that explains the survival processing effect. Rather, the degree
to which survival processing invites elaborative, distinctive forms
of encoding would predict the mnemonic benefit of survival pro-
cessing. We call this the richness of encoding (RE) hypothesis of
the survival processing effect.
Note that the RE hypothesis is consistent with an evolutionary
account of survival processing. In typical everyday contexts—
including those of our hunter-gatherer ancestors—richness of en-
coding is likely to be strongly correlated with importance (i.e.,
fitness relevance) of information. In general, the higher the adap-
tive value of information in certain ecological contexts, the more
elaborate the encoding of this input. Hence, it seems reasonable
that evolution shaped a memory mechanism consistent with the RE
hypothesis. In other words, tests of the RE hypothesis should not
be considered tests against evolutionary accounts of the survival
processing effect. Rather, they address the proximate memory
mechanism that presumably moderates the strength of survival
processing effects on memory performance.
To this end, we evaluated the RE hypothesis in three experiments
similar in design to the studies of Nairne and co-workers. In Exper-
iments 1 and 2, the relevance of words was rated with respect to a
hypothetical context, and memory for these words was tested in a
subsequent surprise recall test. In Experiment 1, we manipulated the
original survival scenario by reducing the number of survival prob-
lems and compared each of these scenarios within-subjects to a
moving scenario lacking survival-relevant information. In Experiment
2, we replicated the effects of Experiment 1 using a randomized
between-subjects design. In Experiment 3, in contrast, we left the
scenarios unchanged and manipulated the task of the participants.
Specifically, we combined the original survival and moving scenarios
with a relevance-argument-generation task rather than a relevance-
rating task. For each item, we asked half of the participants to generate
four arguments highlighting the relevance of the item with respect to
the scenario, whereas the other participants were asked to generate a
single argument per item only. In line with the RE hypothesis, we
predicted that the survival processing effect diminishes when the
number of survival problems addressed in the scenario (Experiments
1 and 2) or the number of to-be-generated arguments for each item
(Experiment 3) is reduced.
Experiment 1
If the survival processing advantage hinges on the richness and
distinctiveness of encoding, manipulating the number of survival
problems mentioned in the scenario should impact its effective-
ness. In general, the smaller the number of survival problems
relevant in the encoding context, the fewer unique associations will
be generated for each word of the list. As is well-supported by the
extensive literature sketched above, this should adversely affect
subsequent recall.
Hence, we modified the survival scenario used by Nairne and
co-workers, thereby reducing the number of fitness-related aspects
relevant during encoding. In our modified survival scenario, only
a single—albeit very serious and urgent—survival problem, lack
of potable water, was relevant for the participants.
Method
Participants. Eighty-one students (65 women, 16 men) from
the University of Mannheim participated. They received a mone-
tary compensation. Their age ranged from 19 to 37 years (M⫽
23.31, SD ⫽3.92).
Apparatus and materials. Stimulus materials (i.e., words to
be rated for their relevance) were taken from Experiment 1 of
Nairne et al. (2007) and consisted of 30 typical words chosen from
30 unique categories. They were translated into German and ran-
domly split in two subsets of 15 words each, separately for each
participant. Stimulus presentation was controlled by personal com-
puters. The survival and moving descriptions were identical to
those used by Nairne et al. (2007) except for our new modified
survival scenario. For the latter, only a single aspect of survival,
lack of water, was relevant. The other survival problems referred
to in the original scenario (i.e., predators and lack of food) were
not mentioned (see the Appendix for details). All materials were
presented in German.
Design. A2⫻2 mixed design was used. Participants were
randomly assigned to either the original (n
o
⫽42) or the short (n
s
⫽
39) survival scenario group (between-subjects factor: original vs.
short). In addition to the survival scenario, participants of both groups
rated words with respect to the original moving scenario of Nairne et
al. (2007; within-subject factor: survival vs. moving).
The experiment consisted of 30 trials. Participants rated the first
15 words using the original survival scenario or the short version
of the survival scenario (depending on the group factor original vs.
short) and the following 15 words using the original moving
1554 KRONEISEN AND ERDFELDER
scenario. Free recall performance (proportion correct) and rating
latencies served as dependent variables. In this design, the survival
processing advantage corresponds to the difference in recall per-
formance between the survival and the moving scenario, separately
for the two versions of the survival scenario (see Nairne et al.,
2007).
Procedure. Participants were asked to rate words according
to their relevance for the scenarios. Instructions for the original
survival and moving scenarios were identical to those used by
Nairne et al. (2007). The instructions for the short survival sce-
nario were as follows (for English translations of all scenarios, see
the Appendix):
In this task, we would like you to imagine that you are stranded in the
grasslands of a foreign land. After searching along the surrounding
area and the debris flushed to the shore along with you, you are bound
to realize that you have a major problem of survival: you have no
potable water. 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.
Note that only the italicized part of this text deviates from the
original survival scenario (see Appendix). Stimuli were presented
one at a time for 5 s each, and participants were asked to rate the
words on a 5-point scale, with 1 indicating that the word was
totally irrelevant and 5 indicating that the word was extremely
relevant to the current scenario. They had to respond within 5 s. If
they did not respond within this time limit, a warning message
occurred. The first target item was preceded by a short practice
phase in which five words were rated with respect to the survival
scenario. After providing relevance ratings for 15 target items with
respect to the survival scenario and another 15 target items with
respect to the moving scenario, a 10-min intelligence test was
administered that served as distractor task. Next, free recall in-
structions appeared unexpectedly for the participants. Participants
were instructed to write down the words they had rated before, in
any order, on a blank sheet of paper. The final recall phase lasted
for 10 min. The total experiment took approximately 25 min. At
the end of the experiment, participants were debriefed and thanked
for their participation.
Results
The significance level was set to ␣⫽.05 for all statistical tests.
All pvalues reported refer to two-tailed tests, even in case of
directed predictions. Relevance ratings were provided for 99% of
the presented words. The number of words for which participants
failed to provide a rating within 5 s did not differ between the two
groups (i.e., original and short).
The mean proportions of correct free recall for both groups and
scenarios are shown in Figure 1.
1
Replicating Nairne et al. (2007),
we found a significant main effect of scenario, F(1, 79) ⫽11.85,
2
⫽0.13, p⬍.05. In addition, the interaction between scenario
and group was significant, F(1, 79) ⫽4.30,
2
⫽0.05, p⬍.05.
Planned comparisons revealed better retention for the original
survival scenario condition compared with the moving scenario
condition (M
Survival
⫽8.36, SE ⫽0.28; M
Moving
⫽6.57, SE ⫽
0.34), t(79) ⫽3.97,
2
⫽0.17, p⬍.05.
2
In contrast, no significant
survival processing advantage was found in the group using the
short survival scenario (M
Survival
⫽6.87, SE ⫽0.40; M
Moving
⫽
6.46, SE ⫽0.35), t(79) ⫽0.95,
2
⫽0.01, p⫽.35. Moreover,
consistent with the RE hypothesis, free recall for the original
survival scenario was significantly better than for the short version
of the survival scenario, t(79) ⫽2.55,
2
⫽0.08, p⬍.05.
Table 1 presents the median response times for the relevance
ratings, separately for each scenario. Response times for the rat-
ings in the survival scenarios were significantly longer than those
in the moving scenario, for both the original survival scenario
group, t(79) ⫽3.95,
2
⫽0.16, p⬍.05, and the short survival
scenario group, t(79) ⫽3.37,
2
⫽0.13, p⬍.05.
Figure 2 displays recall performance as a function of the relevance
ratings provided in the encoding phase. Obviously, higher relevance
ratings are associated with higher levels of recall. With overall recall
performance of the participants controlled, the partial correlation
between ratings and recall was significant in all four conditions of the
design—that is, for the original survival scenario (r⫽.17; p⬍.05),
the moving scenario in the original version group (r⫽.32; p⬍.05),
the short survival scenario (r⫽.58; p⬍.05), and the moving scenario
in the short version group (r⫽.59; p⬍.05).
Discussion
In sum, Experiment 1 nicely replicated the survival recall ad-
vantage found in several previous experiments (see Butler, Kang,
& Roediger, 2009; Kang et al., 2008; Nairne et al., 2007, 2008;
Nairne & Pandeirada, 2008a, 2008b, 2010; Nairne, Pandeirada,
Gregory, & Van Arsdall, 2009; Otgaar & Smeets, 2010; Otgaar,
Smeets, & van Bergen, 2010; Weinstein et al., 2008). However, a
modification of Nairne et al.’s (2007) survival scenario focusing
on a single major survival problem—lack of water—produced
significantly poorer recall performance compared with the original
survival scenario. In fact, a significant survival processing advan-
tage relative to the moving scenario control condition could not be
found for the short version of the scenario. As is well known,
however, insignificance does not prove nonexistence of an effect.
For a two-tailed test, ␣⫽.05, df ⫽79, and small (f
2
⫽.02 or
2
⫽
0.019) versus medium effect sizes (f
2
⫽.15 or
2
⫽0.13; see
Cohen, 1988), the power of our planned contrast ttest is approx-
imately .241 and .931, respectively (Faul, Erdfelder, Lang, &
Buchner, 2009). In other words, our test detects medium and large
1
The two words whiskey and juice were excluded from all further
analyses. Given that in the short version of the survival scenario partici-
pants were told that finding water—or anything drinkable—is the only
critical task for survival, it seems trivial that this group remembers the
words whiskey and juice best. Note that the basic pattern of results is not
affected by excluding these words. Specifically, the critical interaction
effect is also significant when both words are included in the analyses.
2
To increase statistical power, we did not use within-groups matched-
pairs ttests based on separate error variance estimates for each of the
groups. Our planned comparisons are based on the overall within-groups
error term for mean differences and thus have df ⫽79. Note that, in our
application, the conclusions are not affected by this methodological deci-
sion. Separate matched-pairs ttests yield essentially the same results:
survival scenario advantage in the group using the original scenario:
t(41) ⫽4.59
2
⫽0.34, p⬍.05; survival scenario advantage in the group
using the short scenario: t(38) ⫽0.84,
2
⫽0.02, p⫽.41. A high level of
statistical power is of course important, especially for the latter (insignif-
icant) comparison. Therefore, we believe that using planned comparisons
in combination with the overall error term is most appropriate.
1555
SURVIVAL PROCESSING EFFECT
survival processing effects reliably but not necessarily small ef-
fects. Hence, we may conclude that the survival processing advan-
tage either vanishes for the short scenario or diminishes to a small
effect size, consistent with the RE hypothesis.
Replicating Kang et al. (2008), we also found that participants
took more time to rate words in both survival scenarios than in the
moving scenario. Although not strictly predicted, longer rating
latencies in the survival scenario are in line with the RE hypoth-
esis. In general, distinctive processing takes time. Elaborative
encoding often, but not always, requires more time than more
shallow forms of encoding (for examples and exceptions, see Craik
& Tulving, 1975). Hence, it is not surprising to find longer rating
latencies for those scenarios that are hypothesized to invite more
elaborate forms of encoding.
A similar prediction can be made with respect to the relation
between ratings and recall performance. The higher the rating, the
more arguments have probably been found in support of the
relevance claim. Because self-generated arguments may later serve
as retrieval cues, positive correlations between relevance ratings
and recall performance can be expected (Hunt & Smith, 1996).
This is in line with the findings: Higher relevance ratings were
associated with higher recall rates. Again, our results replicate
similar data patterns previously reported by Nairne et al. (2007)
and Butler et al. (2009). Obviously, items that are congruent with
or relevant for the processing task are remembered better. Analo-
gous results have been reported in the depth-of-processing litera-
ture (Craik & Tulving, 1975; Moscovitch & Craik, 1976; Schul-
man, 1974).
Overall, our results mirror the predictions of the richness of
encoding hypothesis. However, a possible objection against Ex-
periment 1 refers to the mixed-list design used in the encoding
phase. Although often used in research on the survival processing
effect because of its efficiency (Nairne et al., 2007, 2008), this
design has the disadvantage of producing a confound between the
type of scenario (survival vs. moving) and the list position of the
items (early vs. late). As is well known, position effects may affect
performance in free recall tests in various ways. Hence, it remains
an open question whether the results of Experiment 1 replicate
under conditions that eliminate this confound. We addressed this
issue in Experiment 2.
Experiment 2
Experiment 2 was designed to replicate the first experiment with
a completely randomized between-subjects design.
Method
Participants. Eighty-one students (37 women, 44 men) from
the University of Mannheim participated. They received a small
monetary compensation. Their age ranged from 19 to 39 years
(M⫽24.98, SD ⫽4.5).
Apparatus and materials. Scenarios and stimulus materials
were identical to Experiment 1 except for 12 buffer words that
were added to the item list, six at the beginning and six at the end
of the list. Buffer words were included to absorb primacy and
recency effects typically found in free recall of word lists. They
were drawn from the German version of the Battig and Montague
category norms (Mannhaupt, 1983). All words, except the buffer
words, were presented in random order.
Design. Participants were randomly assigned to one of three
different groups. The original survival scenario was used in Group 1
(n
1
⫽25), the short version of the survival scenario in Group 2 (n
2
⫽
28), and the moving scenario in Group 3 (n
3
⫽28). Again, free recall
performance and rating latencies served as dependent variables.
Figure 1. Mean proportion of correct recall for each scenario, separately for each group of Experiment 1. The
error bars represent standard errors of the means.
Table 1
Means and Standard Deviations of Participants’ Median Rating
Latencies in Experiment 1
Group/condition
Rating latency (ms)
MSD
Original
Survival 2,242.21 476.96
Moving 1,910.76 426.51
Short
Survival 2,320.95 541.36
Moving 2,038.74 549.22
1556 KRONEISEN AND ERDFELDER
Procedure. The procedure was identical to Experiment 1
except that all words were rated with respect to the same scenario.
The experiment lasted approximately 30 min.
Results
Again, we made use of ␣⫽.05 and two-tailed comparisons in
all statistical tests reported. Ratings were provided for 99% of the
presented words. The number of words for which participants
failed to provide a rating within 5s did not differ between the
groups. The 12 buffer words were not included in the analyses.
Figure 3 shows the proportion of correct free recall for the three
groups: original survival (Group 1: M⫽13.88; SE ⫽0.75), short
survival (Group 2: M⫽12.32; SE ⫽0.75), and moving (Group 3:
M⫽11.18; SE ⫽0.56). The type of scenario produced a clear
difference in recall rates, F(2, 78) ⫽3.76,
2
⫽.05, p⬍.05. As
indicated by planned comparisons, the survival-processing advantage
occurred between the original survival scenario and the moving sce-
nario, t(78) ⫽2.74,
2
⫽0.09, p⬍.05, but not between the short
survival condition and the moving condition, t(78) ⫽1.19,
2
⫽0.02,
p⫽.24.
3
Again, there was also an advantage of the original survival
scenario compared with the short scenario; this time, however, it fell
short of significance, t(78) ⫽1.58,
2
⫽.03, p⫽.12.
Median response times for the ratings are presented in Table 2. An
analysis of variance (ANOVA) showed a significant effect of sce-
nario, F(2, 78) ⫽7.13,
2
⫽.08, p⬍.05. Response times for the
ratings in the original survival scenario were slowest and differed
significantly from the response times in the moving condition, t(78) ⫽
3.58,
2
⫽0.14, p⬍.05. Participants in the short survival condition
also needed more time to rate the items than participants in the
moving condition, t(78) ⫽2.77,
2
⫽0.09, p⬍.05.
Figure 4 illustrates recall performance as a function of partici-
pants’ ratings in the encoding phase. Again, higher ratings are
associated with higher levels of recall. With the overall recall
performance of the participants controlled, the partial correlation
between rating and recall rates was significant for the original
survival scenario group (r⫽.09; p⬍.05), the short survival
scenario group (r⫽.15; p⬍.05), and the moving scenario group
(r⫽.25; p⬍.05).
Discussion
Experiment 2 replicated Experiment 1 nicely. A modification of
the survival scenario focusing on a single aspect of survival
resulted in poorer memory performance than the original survival
scenario. Compared with the moving control scenario, the memory
advantage of survival processing was significant for the original
scenario but failed to be significant for the short version of the
survival scenario. Moreover, we replicated the correlation between
relevance ratings and recall performance and the rating latency
differences between scenarios found in Experiment 1.
Experiment 2 rules out possible objections against Experiment
1, namely, that results were affected by list position effects. In
Experiment 2, we controlled primacy and recency effects using
primacy and recency buffers. Also eliminated was the confound
between list position and scenario inherent in Experiment 1 and
other experiments on the survival processing effect. Nevertheless,
we were able to replicate the results of Experiment 1.
In sum, Experiments 1 and 2 showed that the survival process-
ing effect diminishes when richness of encoding (RE) is cut down
by simplifying the survival scenario. Obviously, this is only one
way to manipulate RE. If the RE hypothesis of the survival
processing effect is correct, other manipulations of the relevant
encoding processes should work as well. For example, rather than
comparing different scenarios, one could keep the scenario un-
changed and manipulate the task participants perform while pro-
cessing the relevance of items with respect to the scenario. Does a
simplification of this task also result in a reduction of the survival
processing effect, even if the original survival scenario is left
unchanged? Experiment 3 was designed to address this issue.
3
We use planned comparisons based on the overall error term for the
same reasons already outlined in Footnote 2. Two-group ttests based on
separate error variance estimates for each pair of groups would have lower
statistical power. Note that our procedure does not require assumptions in
addition to those we usually make in mean comparisons. Using the overall
error term presupposes homogeneity of error variance across all three
groups, an assumption we also make in ANOVAs. It is important to note
that there was no reason to reject this assumption for our data.
Figure 2. Mean proportions of correct recall for each scenario in Experiment 1, separately for each rating
category (5-point scale, with 1 ⫽totally irrelevant and 5 ⫽extremely relevant). The error bars represent standard
errors of the means.
1557
SURVIVAL PROCESSING EFFECT
Experiment 3
In Experiment 3, we replaced the relevance-rating task em-
ployed in the encoding phases of Experiments 1 and 2 by a
relevance-argument-generation task. Specifically, participants
were asked to write down either four arguments or one argument
highlighting the relevance of an item for either the original sur-
vival scenario or the original moving scenario. For obvious rea-
sons, the four-argument condition provides more opportunities for
distinctive and elaborate forms of encoding than the one-argument
condition. According to the RE hypothesis, therefore, the memory
advantage of survival processing should be much more pro-
nounced for the former condition than for the latter.
The one-argument condition of Experiment 3 resembles the
short survival scenario rating condition of Experiments 1 and 2 in
that participants are asked to focus on a single aspect of survival in
both cases. Whereas participants think about a single survival
problem in the short survival rating condition (i.e., one or more
uses of an item to solve a single survival problem), they typically
think about a single technique of survival in the one-argument
condition (i.e., a single use of an item to solve one or more survival
problems). For example, when processing the target item screw-
driver, one of our participants produced the argument “hunt,”
indicating that she could use a screwdriver for hunting animals.
Given this similarity between the short survival rating condition
and the one-argument condition, we expected that the survival
processing effect would diminish or even vanish in the one-
argument condition of Experiment 3 as it did in the short scenario
rating conditions of Experiments 1 and 2.
Method
Participants. Fifty-three students from the University of
Mannheim and another 53 students from the University of
Freiburg participated in the experiment. Three participants were
excluded due to a failure to comply with task instructions. The
analyses reported are based on the remaining 103 students (78
women, 25 men) who received participation credit for taking part
in the experiment. Their age ranged from 18 to 37 years (M⫽
22.26, SD ⫽3.20).
Apparatus and materials. Stimulus materials were identical
to those in Experiment 2. All words, except the buffer words, were
presented in random order. The survival and moving scenario
descriptions were identical to those used by Nairne et al. (2007).
All materials were presented in German.
Design. A 2 (type of scenario: moving vs. survival) ⫻2
(number of to-be-generated arguments: one vs. four) completely
randomized between-subjects design with four groups in total was
used (Group 1: survival scenario, one argument, n
1
⫽26; Group 2:
survival scenario, four arguments, n
2
⫽23; Group 3: moving
scenario, one argument, n
3
⫽26; Group 4: moving scenario, four
arguments, n
4
⫽28). The experiment consisted of 42 trials, in-
cluding the 12 buffer words. A total of 30 trials per participant
were included in the analyses. Free recall performance and re-
sponse times in the relevance-argument-generation task served as
dependent variables.
Procedure. Depending upon experimental condition, partic-
ipants first read the original survival or the original moving sce-
nario. Next, also depending upon condition, participants were
asked to think of either one or four arguments highlighting the
relevance of each item with respect to the scenario and to generate
two key words per argument. There was no time limit for this task.
Relevant key words were to be typed into the computer in any
order. After the participants provided the key words for one item
and pressed the “Enter” key, the next target item appeared at the
center of the screen, followed by the next argument-generation
task. A short practice phase for one word preceded the main task.
Following the main task including 42 items—a primacy buffer of
six items, 30 target items, and a recency buffer of six items—a
10-min intelligence test was administered that served as a distrac-
Figure 3. Mean proportion of correct recall for each scenario, separately for each group of Experiment 2. The
error bars represent standard errors of the means.
Table 2
Means and Standard Deviations of Participants’ Median Rating
Latencies in Experiment 2
Group
Rating latency (ms)
MSD
Original survival 2,044.66 435.65
Short survival 1,992.29 465.03
Moving scenario 1,695.46 122.21
1558 KRONEISEN AND ERDFELDER
tor task. Next, free recall instructions appeared unexpectedly for
the participants. The final recall phase lasted 7 min, a time frame
that did not impose any time pressure on the participants (see also
Nairne and Pandeirada, 2008a, p. 379). The total experiment took
approximately 60 min for the four-arguments groups and about 40
min for the one-argument groups. At the end of the experiment,
participants were debriefed and thanked for their participation.
Results
As in case of Experiments 1 and 2, our statistical tests are
two-tailed and make use of ␣⫽.05 without exception. A prelim-
inary analysis of free recall rates showed, as expected, that sample
background (University of Freiburg vs. University of Mannheim)
did not produce a main effect or any interaction effect with one or
both factors of the design, all F(1, 95) ⬍2.12, all p⬎.15. Hence,
all subsequent analyses were performed for the total sample only.
The mean proportions of correct free recall for the four exper-
imental groups are shown in Figure 5. Significant main effects
emerged for the scenario factor, F(1, 99) ⫽4.35,
2
⫽0.04, p⬍
.05, and the number-of-arguments factor, F(1, 99) ⫽9.61,
2
⫽
0.09, p⬍.05. The interaction between scenario and number of
arguments was also significant, F(1, 99) ⫽3.96,
2
⫽0.04, p⬍
.05. Planned comparisons based on the overall error term revealed
better retention for the survival scenario group compared with the
moving scenario group (M
Survival
⫽21.52, SE ⫽0.81; M
Moving
⫽
18.21, SE ⫽0.91), t(99) ⫽2.86,
2
⫽0.07, p⬍.05, in the
four-argument condition. The same effect did not occur when
we compared both scenarios for the one-argument condition
(M
Survival
⫽17.38, SE ⫽0.85; M
Moving
⫽17.31, SE ⫽0.80),
t(99) ⫽0.07,
2
⬍0.001, p⫽.95.
To test for differences in response times in the relevance-
argument-generation task between conditions, we computed me-
dian response times for each participant (Table 3). As expected,
participants required more time to find four arguments rather than
one argument, F(1, 99) ⫽145.62,
2
⫽0.59, p⬍.05. In contrast,
neither the main effect of scenario, F(1, 99) ⫽0.60,
2
⫽0.01,
p⫽.44, nor the interaction of type of scenario and number of
arguments was significant, F(1, 99) ⫽0.07,
2
⫽0.001, p⫽.80.
Discussion
In Experiment 3, we aimed at testing the RE hypothesis using a
relevance-argument-generation task rather than a relevance-rating
Figure 4. Mean proportions of correct recall for each scenario, separately for each rating category (5-point
scale, with 1 ⫽totally irrelevant and 5 ⫽extremely relevant) in Experiment 2. The error bars represent standard
errors of the means.
Figure 5. Mean proportion of correct recall for each scenario, separately
for each group (four-argument group vs. one-argument group) of Experi-
ment 3. The error bars represent standard errors of the means.
Table 3
Means and Standard Deviations of Participants’ Median Rating
Latencies in Experiment 3
Group
Response time (ms)
MSD
Four-argument
Survival 41,503.22 14,192.59
Moving 44,237.57 19,075.13
One-argument
Survival 10,152.88 3,557.29.
Moving 11,515.50 11,499.12
1559
SURVIVAL PROCESSING EFFECT
task. We reasoned that, compared with the moving scenario, the
original survival scenario of Nairne and coworkers fosters distinc-
tive processing and richness of encoding when participants are
required to find four arguments highlighting the relevance of an
item in the respective context, much like the rating task employed
in Experiments 1 and 2 does. In contrast, when participants are
asked to generate a single argument only, differences between both
scenarios in terms of richness of encoding diminish, simply be-
cause the variety of encoding options provided by the survival
scenario is reduced to a single survival aspect for each item.
Hence, if the RE hypothesis is correct, the survival processing
advantage should diminish or perhaps even vanish in the one-
argument condition compared with the four-argument condition. In
line with these predictions, a clear survival processing advantage
was detected in the four-argument condition but not in the one-
argument condition. In fact, the survival processing advantage was
negligible in the one-argument condition (
2
⬍0.001) and far
from being significant. Again, however, one should keep in mind
that insignificance does not prove nonexistence of an effect. For a
two-tailed test, ␣⫽.05, df ⫽99, and a small effect size (f
2
⫽.02
or
2
⫽0.019; see Cohen, 1988), the power of our planned
contrast ttest is .295 in Experiment 3 (Faul et al., 2009). The test
power increases to .973 for a medium effect size (f
2
⫽.15 or
2
⫽
0.13; see Cohen, 1988) under the same conditions. This means that
we were able to detect medium and larger effect sizes reliably. Just
like in Experiments 1 and 2, however, we cannot rule out a small
survival processing advantage in the one-argument condition, even
though the effect was far from being significant. This result cor-
roborates the RE hypothesis. The RE hypothesis predicts a sub-
stantial reduction, not necessarily an elimination, of the survival
processing effect when encoding processes focus on a single
survival aspect.
General Discussion
Based on the richness-of-encoding explanation of the survival
processing effect, our experiments explored the plasticity of the
survival processing effect induced by manipulations of richness of
encoding. In Experiment 1 and 2, we modified the original survival
scenario such that only a single archaic survival problem, lack of
potable water, became relevant. The idea behind this modification
was to keep the survival relevance of the encoding context un-
changed while at the same time diminishing the degree to which
the scenario invited distinctive, elaborate forms of encoding. We
showed that this modification resulted in substantially reduced
recall rates compared with the original survival scenario. Addi-
tionally, we were unable to find a significant recall difference
between the short survival scenario and the moving scenario in two
experiments, despite sufficient statistical power for effects of at
least medium size. The moving scenario is often used as a control
scenario to establish survival processing advantages (Nairne et al.,
2007; Otgaar & Smeets, 2010; Otgaar, Smeets, & van Bergen,
2010; Weinstein et al. 2008). We may thus conclude that the
survival processing effect either vanishes for the short version of
the survival scenario or diminishes to a small size.
In Experiment 3, we instructed our participants to generate
either one argument or four arguments highlighting the relevance
of an item with respect to the survival or the moving scenario. Both
scenarios matched those used by Nairne and coworkers in their
experiments. In line with the results of the first two experiments,
this manipulation results in a clear survival processing advantage
in the four–argument condition, whereas the same effect disap-
pears in the one-argument condition.
Our results shed light on the proximate memory processes
underlying the survival processing effect. They cast doubt on the
view that it is indeed survival relevance that matters for the
survival processing advantage. As shown in the present research,
this effect is not found for encoding contexts that focus on a single
(though very essential and urgent) survival problem (short scenar-
ios in Experiments 1 and 2) or on a single technique of survival
(one-argument condition of Experiment 3). Hence, processing of
survival relevance is not sufficient for the emergence of the sur-
vival processing effect, even if scenarios, materials, tasks, and
memory tests match those used by Nairne and co-workers. More-
over, the survival processing effect is correlated with the relevance
ratings provided in the encoding phase and, although to a weaker
degree, with the time required to provide the ratings. Apparently,
elaborate forms of encoding are involved in the strong survival
processing effect typically reported in the literature. Increasing the
number of survival problems addressed in the survival scenario
(Experiments 1 and 2) or the number of survival techniques
processed per item (Experiment 3) invites richer, more distinctive
forms of encoding and thereby boosts the survival processing
advantage. We conclude that it is the richness and distinctiveness
of encoding that matters for the emergence of the survival pro-
cessing effect, consistent with the RE hypothesis originally sug-
gested by Nairne and Pandeirada (2008a).
Clearly, further research is needed to test whether the survival-
recall advantage can be manipulated in other ways that also affect
richness of encoding and, in addition, whether richness of encod-
ing is the only proximate mechanism involved in the survival
processing effect or just one mechanism among others. Hence, our
results should be considered first steps toward understanding the
nature of the survival processing effect.
References
Anderson, J. R., & Schooler, L. J. (1991). Reflections of the environment
in memory. Psychological Science, 2, 396–408. doi:10.1111/j.1467-
9280.1991.tb00174.x UID 1992-15346-001
Anderson, J. R., & Schooler, L. J. (2000). The adaptive nature of memory.
In E. Tulving & F. I. M. Craik (Eds.), Handbook of memory (pp.
557–570). New York, NY: Oxford University Press. UID 2000-00111-
034
Butler, A. C., Kang, S. H. K., & Roediger, H. L. (2009). Congruity effects
between materials and processing tasks in the survival processing par-
adigm. Journal of Experimental Psychology: Learning Memory and
Cognition, 35, 1477–1486. doi:10.1037/a0017024 UID 2009-19590-008
Cohen, J. (1988). Statistical power analysis for the behavioral sciences.
Hillsdale, NJ: Erlbaum.
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. doi:10.1016/S0022-5371(72)80001-X UID
1973-20189-001
Craik, F. I. M., & Tulving, E. (1975). Depth of processing and the retention
of words in episodic memory. Journal of Experimental Psychology:
General, 104, 268–294. doi:10.1037/0096-3445.104.3.268 UID 1976-
00185-001
Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2009). Statistical
power analysis using G
ⴱ
Power 3.1: Tests for correlation and regression
1560 KRONEISEN AND ERDFELDER
analyses. Behavior Research Methods, 41, 1149–1160. doi:10.3758/
BRM.41.4.1149
Hunt, R. R. (2003). Two contributions of distinctive processing to accurate
memory. Journal of Memory and Language, 48, 811–825. doi:10.1016/
S0749-596X(03)00018-4 UID 2003-00771-010
Hunt, R. R., & Smith, R. E. (1996). Accessing the particular from the
general: The power of distinctiveness in the context of organization.
Memory & Cognition, 24, 217–225. doi:10.3758/BF03200882 UID
1996-04341-008
Hunt, R. R., & Worthen, J. B. (2006). Distinctiveness and memory. Oxford,
NY: Oxford University Press. UID 2007-06322-000
Kang, S. H. K., McDermott, K. B., & Cohen, S. M. (2008). The mnemonic
advantage of processing fitness-relevant information. Memory & Cog-
nition, 36, 1151–1156. doi:10.3758/MC.36.6.1151 UID 2008-14529-
011
Klein, S. B., Cosmides, L., Gangi, C., Jackson, B., Tooby, J., & Costabile,
K. (2009). Evolution and episodic memory: An analysis and demonstra-
tion of a social function of episodic recollections. Social Cognition, 27,
283–319. doi:10.1521/soco.2009.27.2.283 UID 2009-05885-007
Klein, S. B., Cosmides, L., Tooby, J., & Chance, S. (2002). Decisions and
the evolution of memory: Multiple systems, multiple functions. Psycho-
logical Review, 109, 306–329. doi:10.1037/0033-295X.109.2.306 UID
2002-12576-004
Klein, S. B., Robertson, T. E., & Delton, A. W. (2010). Facing the future:
Memory as an evolved system for planning future acts. Memory &
Cognition, 38, 13–22. doi:10.3758/MC.38.1.13 UID 2009-23676-003
Mannhaupt, H. R. (1983). Produktionsnormen fu¨r verbale Reaktionen zu
40 gela¨ufigen Kategorien [German category norms for verbal ı´tems in 40
categories]. Sprache und Kognition, 4, 264–278.
Moscovitch, M., & Craik, F. I. M. (1976). Depth of processing, retrieval
cues, and uniqueness of encoding as factors in recall. Journal of Verbal
Learning & Verbal Behavior, 15, 447–458. doi:10.1016/S0022-
5371(76)90040-2 UID 1977-04980-001
Nairne, J. S. (2010). Adaptive memory: Evolutionary constraints on re-
membering. In B. H. Ross (Ed.), The psychology of learning and
motivation: Advances in research and theory, (Vol. 53, pp. 1–32). San
Diego, CA: Elsevier. doi:10.1016/S0079-7421(10)53001-9 UID 2010-
14184-001
Nairne, J. S., & Pandeirada, J. N. S. (2008a). Adaptive memory: Is survival
processing special? Journal of Memory and Language, 59, 377–385.
doi:10.1016/j.jml.2008.06.001 UID 2008-12852-011
Nairne, J. S., & Pandeirada, J. N. S. (2008b). Adaptive memory: Remem-
bering with a stone-age brain. Current Directions in Psychological
Sciences, 17, 239–243. doi:10.1111/j.1467-8721.2008.00582.x UID
2008-10757-001
Nairne, J. S., & Pandeirada, J. N. S. (2010). Ancestral priorities and the
mnemonic value of survival processing. Cognitive Psychology, 61, 1–22.
doi:10.1016/j.cogpsych.2010.01.005 UID 2010-10966-002
Nairne, J. S., Pandeirada, J. N. S., Gregory, K. J., & Van Arsdall, J. E.
(2009). Adaptive memory and the hunter-gatherer mind. Psychological
Science, 20, 740–746. doi:10.1111/j.1467-9280.2009.02356.x UID
2009-08575-013
Nairne, J. S., Pandeirada, J. N. S., & Thompson, S. (2008). Adaptive
memory: The comparative value of survival processing. Psychological
Science, 19, 176–180. doi:10.1111/j.1467-9280.2008.02064.x UID
2008-01753-014
Nairne, J. S., Thompson, S. R., & Pandeirada, J. N. S. (2007). Adaptive
memory: Survival processing enhances retention. Journal of Experimen-
tal Psychology: Learning, Memory, and Cognition, 33, 263–273. doi:
10.1037/0278-7393.33.2.263 UID 2007-02740-001
Otgaar, H., & Smeets, T. (2010). Adaptive memory: Survival processing
increases both true and false memory in adults and children. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 36, 1010–
1016. doi:10.1037/a0019402 UID 2010-12650-012
Otgaar, H., Smeets, T., & van Bergen, S. (2010). Picturing survival
memories: Enhanced memory after fitness-relevant processing occurs
for verbal and visual stimuli. Memory & Cognition, 38, 23–28. doi:
10.3758/MC.38.1.23 UID 2009-23676-004
Packman, J. L., & Battig, W. F. (1978). Effects of different kinds of
semantic processing on memory for words. Memory & Cognition, 6,
502–508. doi:10.3758/BF03198238 UID 1980-04662-001
Schooler, L. J., & Hertwig, R. (2005). How forgetting aids heuristic
inference. Psychological Review, 112, 610–628. doi:10.1037/0033-
295X.112.3.610 UID 2005-08166-004
Schulman, A. I. (1974). Memory for words recently classified. Memory &
Cognition, 2, 47–52.
Symons, C. S., & Johnson, B. T. (1997). The self-reference effect in
memory: A meta-analysis. Psychological Bulletin, 121, 371–394. doi:
10.1037/0033-2909.121.3.371 UID 1997-03609-003
Tooby, J., & Cosmides, L. (1992). The psychological foundations of
culture. In J. H. Barkow, L. Cosmides, & J. Tooby (Eds.), The adapted
mind: Evolutionary theory and the generation of culture (pp. 19–136).
New York, NY: Oxford University Press. UID 1992-98504-001
Tooby, J., & Cosmides, L. (2005). Conceptual foundations of evolutionary
psychology. In D. Buss (Ed.), The handbook of evolutionary psychology
(pp. 5–67). Hoboken, NJ: Wiley. UID 2005-08200-001
Watkins, M. J. (1978). Engrams as cuegrams and forgetting as cue-
overload: A cueing approach to the structure of memory. In C. R. Puff
(Ed.), The structure of memory. (pp. 347–372). New York, NY: Aca-
demic Press.
Watkins, M. J., & Watkins, O. C. (1976). Cue-overload theory and the
method of interpolated attributes. Bulletin of the Psychonomic Society, 7,
289–291. UID 1976-25332-001
Watkins, O. C., & Watkins, M. J. (1975). Buildup of proactive inhibition
as a cue-overload effect. Journal of Experimental Psychology: Human
Learning and Memory, 1, 442–452. doi:10.1037/0278-7393.1.4.442
UID 1975-28897-001
Weinstein, Y., Bugg, J. M., & Roediger, H. L. (2008). Can the survival
recall advantage be explained by the basic memory processes? Memory
& Cognition, 36, 913–919. doi:10.3758/MC.36.5.913 UID 2008-09383-
002
(Appendix follows)
1561
SURVIVAL PROCESSING EFFECT
Appendix
Scenarios for Experiment 1, 2, and 3
Survival
In this task, we would like you to imagine that you are stranded
in the grasslands 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 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.
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.
Short Survival
In this task, we would like you to imagine that you are stranded
in the grasslands of a foreign land. After searching along the
surrounding area and the debris flushed to the shore along with
you, you are bound to realize that you have a major problem of
survival: you have no potable water. 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.
Received June 29, 2010
Revision received February 21, 2011
Accepted March 31, 2011 䡲
1562 KRONEISEN AND ERDFELDER
A preview of this full-text is provided by American Psychological Association.
Content available from Journal of Experimental Psychology Learning Memory and Cognition
This content is subject to copyright. Terms and conditions apply.
