psychology article summaries
xiaowuxiang
psy_memory.pdf
Adaptive memory: The survival scenario enhances item-specific processing relative to a moving scenario
Daniel J. Burns1, Joshua Hart1, Samantha E. Griffith1, and Amy D. Burns2
1Department of Psychology, Union College, Schenectady, NY, USA 2Department of Psychology, Marist College, Poughkeepsie, NY, USA
Nairne, Thompson, and Pandeirada (2007) found that retention of words rated for their relevance to survival is superior to that of words encoded under numerous other deep processing conditions. They suggested that our memory systems might have evolved to confer an advantage for survival-relevant information. Burns, Burns, and Hwang (2011) suggested a two-process explanation of the proximate mechanisms responsible for the survival advantage. Whereas most control tasks encourage only one type of processing, the survival task encourages both item-specific and relational processing. They found that when control tasks encouraged both types of processing, the survival processing advantage was eliminated. However, none of their control conditions included non-survival scenarios (e.g., moving, vacation, etc.), so it is not clear how this two-process explanation would explain the survival advantage when scenarios are used as control conditions. The present experiments replicated the finding that the survival scenario improves recall relative to a moving scenario in both a between-lists and within-list design and also provided evidence that this difference was accompanied by an item-specific processing difference, not a difference in relational processing. The implications of these results for several existing accounts of the survival processing effect are discussed.
Keywords: Adaptive memory; Survival processing; Planning; Item-specific; Relational; Free recall; Cumulative recall.
The capabilities of the human memory system have undoubtedly been sculpted by evolution as a consequence of problems faced by our ancestors (e.g., Cosmides & Tooby, 1987; Sherry & Schacter, 1987), with species adaptedness being the ulti- mate function of these capabilities. Nairne et al. (2007) reasoned that one likely consequence of memory system evolution is that information relevant to survival would be afforded special status by our memory systems, producing a memorial advantage. In their original study parti- cipants were presented with a list of words and performed one of three orienting tasks on the items. In the survival processing task participants were instructed to imagine they were stranded in
the grasslands of a foreign land without any food or supplies, and were asked to rate words in terms of their relevance to surviving in this situation. The second task was a pleasantness rating task, which was chosen because it is known to produce particularly good retention performance (e.g., Einstein & Hunt, 1980). The third task was designed to promote a level of schematic proces- sing similar to that of the survival scenario, and required participants to rate the words for their relevance to moving to a city in a foreign land. Unexpected recall of the items following a short distractor task revealed a substantial free-recall advantage for the survival condition relative to the other conditions, consistent with Nairne et al.’s
Address correspondence to: Daniel J. Burns, Department of Psychology, Union College, Schenectady, NY 12308, USA. E-mail:
This research was supported in part by an internal faculty research grant from Union College.
Memory, 2013 Vol. 21, No. 6, 695�706, http://dx.doi.org/10.1080/09658211.2012.752506
# 2013 Taylor & Francis
conjecture that human memory has adapted to help us remember survival-relevant information.
Across numerous replications comparing the survival task to a variety of different control tasks, the survival processing retention advantage has proven reliable and robust (e.g., Kang, McDermott, & Cohen, 2008; Kostic, McFarlan, & Cleary, 2012; Nairne, Pandeirada, Gregory, & Van Arsdall, 2009; Nairne, Pandeirada & Thompson, 2008, Smeets, Otgaar, Raymaekers, Peters, & Merckelbach, 2012; but see Howe & Derbish, 2010). Most of these control tasks can be divided into two types: (1) those requiring participants to imagine them- selves in a particular scenario and to rate the items for their relevance to that scenario (e.g., moving, robbery, vacation, city survival), and (2) those not involving a scenario, but nonetheless requiring a decision be made about the items (e.g., pleasantness, self-relevance, and imagery ratings, category sorting, and item generation).
For the second type of control task, those not involving scenarios, it has been suggested that the survival task may encourage the processing of both item-specific and relational information, whereas the control tasks may have encouraged only item-specific or only relational processing (Burns et al., 2011; Nairne & Pandeirada, 2008). Item-specific processing refers to the encoding of individual characteristics of items, whereas rela- tional processing refers to encoding the relation- ships between list items. Each type of processing serves a different function during retrieval (Ein- stein & Hunt, 1980; Hunt & Einstein, 1981). Relational processing is believed to facilitate retention by providing an organised plan for efficient retrieval of the items, whereas item- specific processing presumably facilitates discri- mination of individual items on the list from other items, as well as providing specific retrieval cues for individual items (e.g., Burns, 2006; Burns & Gold, 1999; Hunt & McDaniel, 1993). It has been demonstrated repeatedly that the combination of item-specific and relational processing is particu- larly beneficial to recall (e.g., Einstein & Hunt, 1980; Hunt & Einstein, 1981).
Nairne and Pandeirada (2008) compared sur- vival processing to a pleasantness rating task known to induce item-specific processing. Addi- tionally they used a categorically related list of words, ensuring that all participants would encode relational information. This procedure of requir- ing pleasantness rating of categorically related items is the technique of choice for recruiting both item-specific and relational processing, and
results in recall performance superior to that of conditions promoting only item-specific or only relational processing (e.g., Burns & Schoff, 1998; Einstein & Hunt, 1980; Hunt & Einstein, 1981; Klein, Loftus, Kihlstrom, & Aseron, 1989). The results, however, still revealed a recall advantage for the survival processing group, suggesting that survival processing produces recall above that produced by the combined processing of item- specific and relational information.
However, Burns et al. (2011) argued that the amount of relational processing in the survival and pleasantness rating groups might not have been equivalent in Nairne and Pandeirada’s (2008) experiments. The list items were selected from categories that were highly relevant to the survival scenario (fruits, vegetables, four-legged animals, and human dwellings), which may have resulted in more relational (or more congruous) processing for the survival group than for the pleasantness group. Burns et al. compared survi- val processing both to a pleasantness-rating con- dition, presumed to encourage item-specific processing, and a category-sorting group, known to promote relational processing. In the first two experiments a list of categorically related items was used, and in the remaining two experiments a list of seemingly unrelated items from ad hoc categories was used. (Previous research has shown that categorically related list items inher- ently foster relational processing, whereas unre- lated lists foster item-specific processing.)
In the first two experiments survival processing was contrasted with a condition presumed to perform only relational processing (category sort- ing) and a condition presumably performing both types of processing (pleasantness rating). In the latter two experiments survival processing was compared to a condition performing only item- specific processing (pleasantness rating) and a condition processing both types of information (category sorting). The results showed a recall advantage for survival processing over conditions performing only item-specific or only relational processing, but not over conditions involving both types of processing. Moreover, several indices of item-specific and relational processing were con- sistent with the hypothesis that item-specific and relational processing differences were responsible for the recall differences. For example, a final recognition test revealed that the recognition scores, which index item-specific processing, tended to be higher for the survival processing condition than for conditions performing only
696 BURNS ET AL.
relational processing, but not for conditions performing only item-specific processing or both types of processing. Similarly, a minute-by-minute cumulative-recall analysis showed that compared to conditions fostering only item-specific proces- sing, the survival task produced greater recall during the first few minutes of the recall period; results indicative of a relational processing differ- ence (e.g., Burns & Schoff, 1998). Compared to conditions performing only relational processing, the survival processing recall advantage occurred in the latter portions of the recall period, indicat- ing an item-specific processing difference (see Burns & Schoff).
Burns et al.’s (2011) experiments provide evidence that the proximate cause of the reten- tion advantage associated with survival processing is enhanced processing of item-specific and rela- tional information. They argued that many of the control conditions tested in previous studies resulted in the processing of only one type of information, affording survival processing a mem- orial advantage, and proposed a two-process account of the survival processing effect. Accord- ing to this account the survival scenario recruits both item-specific and relational processing, whereas control conditions have usually recruited only one type of processing or the other. When control conditions foster both types of processing the survival advantage should be eliminated.
This two-process explanation seems particu- larly useful for explaining the results of experi- ments that used non-scenario control conditions (e.g., pleasantness rating, imagery rating, genera- tion), because many of these control conditions are known to induce primarily item-specific processing (e.g., Burns, Curti & Lavin, 1993; Einstein & Hunt, 1980; Hodge & Otani, 1996). It is not clear, however, why survival processing would produce more item-specific or more rela- tional processing than control conditions invol- ving other scenarios (e.g., moving, robbery, etc.). For example, some control scenarios were created specifically to be equivalent to the survival scenario in terms of thematic structure, so one would expect a similar degree of relational processing. There is also no obvious reason to expect the survival scenario would promote more item-specific processing than other scenarios. According to the two-process account, however, the superior recall for the survival scenario is due to either more relational processing or more item- specific processing than other scenarios.
Recent findings by Kroneisen and Erdfelder (2011) suggest that the survival scenario may promote more item-specific processing than non-survival scenarios. In the first two experi- ments the standard survival and moving scenarios were compared to a simplified survival scenario involving only one problem or potential threat to survival: the search for potable water. In Experi- ment 3 participants were given the standard survival and moving scenarios but were required to generate either one or four arguments con- cerning the relevance of each list item. The results showed that when the participants were required to solve only one problem or generate only one argument, the survival scenario produced recall that was statistically equivalent to the moving scenario. The authors suggested that the typical survival scenario may promote more elaborative processing, resulting in more distinctive encoding of the items, relative to the moving scenario. Simplifying the survival tasks reduces the amount of elaborative processing, thereby eliminating the distinctiveness advantage that typically accompa- nies survival processing.
Although Kroneisen and Erdfelder (2011) discussed their results in terms of elaboration and distinctiveness*terms that may suggest the enhanced processing of either item-specific or relational processing*their actual manipulations might have primarily affected only item-specific processing. For example, increasing the number of arguments that participants generate about each item’s survival relevance seems very similar to increasing the number of word associates gener- ated for each list item, a procedure known to enhance item-specific processing (e.g., McDaniel, Moore, & Whiteman, 1998). Whereas Kroneisen and Erdfelder’s (2011) results suggest to us that the survival scenario enhances item-specific pro- cessing relative to a moving scenario (an inter- pretation that must remain tentative because they did not use direct measures of item-specific and relational processing), there is also reason to believe that survival processing may lead to greater relational processing than other scenarios under certain conditions. Both Howe and Derbish (2010) and Otgaar and Smeets (2010) showed that survival processing produced higher recall than a moving scenario when the list items were related either categorically or semantically (i.e., lists used in the Deese, Roediger and McDermott [DRM] task). Moreover, survival processing also pro- duced greater false recall of the critical lures that were related to the list words but not
SURVIVAL PROCESSING 697
presented. It has been shown that relational processing of related words increases both true and false memories, whereas item-specific proces- sing increases recall of the list items but decreases false memories (e.g., Burns, Jenkins, & Dean, 2007). Hence there is some evidence that, at least for related lists, survival processing may induce more relational processing of the list words.
The purpose of the two experiments presented here was to examine whether survival processing enhances relational or item-specific processing of seemingly unrelated list words. We focused on unrelated words because we were interested in whether the two-process explanation could ex- plain the bulk of the published research contrast- ing survival scenarios with other scenarios, nearly all of which used unrelated or minimally related items.
Similar to Burns et al. (2011), we used cumu- lative-recall curves to assess item-specific and relational processing. We observed the number of items recalled during each minute of the 10- minute recall period, and plotted the cumulative- recall curves for each condition. It is known that the following exponential equation provides a good fit of cumulative-recall curves:
n tð Þ ¼ n 8ð Þ 1 � e�kt � �
(1)
where n(t) is the number of items recalled at time t, n(8) is asymptotic level of recall, e is the base of the natural logarithm, and l is the rate of approaching asymptote (e.g., Bousfield & Sedge- wick, 1944; Indow & Togano, 1970; Roediger, Stellon & Tulving, 1977). Whereas a strong inverse relationship between asymptotic recall level, n(8), and rate of approaching asymptote, l, usually exists (e.g., Bousfield & Sedgewick, 1944; Hermann & Chaffin, 1976; Hermann & Murray, 1979; Indow & Togano, 1970; Johnson, Johnson, & Marks, 1951; Kaplan, Carvellas, & Metlay, 1969), differential processing of item- specific and relational processing produces an exception to this inverse relationship. Relational processing produces curves with a steep slope that reach asymptotic levels very quickly, whereas item-specific processing produces more gradual cumulative recall. Examples from Burns and Schoff’s (1998) experiments of the curves pro- duced by varying the amount of item-specific and relational processing are shown in Figure 1. As can be seen, conditions performing both types of processing produce high initial recall as well as a relatively gradual approach to asymptote, whereas conditions performing only relational
processing produce initially high recall that asymptotes relatively quickly. Finally, conditions fostering only item-specific processing produce relatively slow initial recall that remains steady throughout the remainder of the recall period. Thus the shapes of the cumulative-recall curves and the estimates of l and n(8) can be, and have been, used to assess both item-specific and rela- tional processing differences (see Burns, 2006; Burns & Hebert, 2005; Burns et al., 2007; Burns, Martens, Bertoni, Sweeney, & Lividini, 2006; Congleton & Rajaram, 2012).
We also gave a final recognition test following the recall test. Good recognition performance is highly dependent on item-specific processing (see Einstein & Hunt, 1980; Hunt & Einstein, 1981).
Recall Duration (Min)
C u m
u la
tiv e R
e ca
ll S
co re
0
5
10
15
20
25
30
35
Relational and Item-Specific Relational
Recall Duration (Min)
C u m
u la
tiv e R
e ca
ll S
co re
0
5
10
15
20
25
30
35
Relational and Item-Specific Item-Specific
Recall Duration (Min)
60 2 4 8 10 12 14 16
60 2 4 8 10 12 14 16
60 2 4 8 10 12 14 16
C u m
u la
tiv e R
e ca
ll S
co re
0
5
10
15
20
25
30
Relational Item-Specific
Figure 1. Mean cumulative-recall scores for conditions
tested in Burns and Schoff’s (1998) experiments that varied
in amount of item-specific and relational processing per-
formed.
698 BURNS ET AL.
We note here that the survival scenario has already been shown to produce higher recogni- tion performance than non-survival scenarios (e.g., Nairne et al, 2007), so we expected to replicate this finding.
EXPERIMENT 1
The main objective of Experiment 1 was to determine whether, as suggested by Kroneisen and Erdfelder (2011), survival processing induces more item-specific processing than the moving scenario. We hypothesised that the survival sce- nario would produce greater recall than the moving scenario and that the recall advantage would be accompanied by a recognition advan- tage. We also predicted that the survival scenario would produce a cumulative-recall curve that diverged from the curve produced by the moving group, but only after the first few minutes of recall. This pattern of cumulative recall was expected to produce fairly similar estimates of l for the two groups, but the survival scenario was expected to produce a higher estimate of n(8). These results would suggest that the survival scenario induces more item-specific processing, but not more relational processing, than the moving scenario.
Method
Participants. A total of 73 college students who were paid either $6.00 or received credit towards an out-of-class activities requirement in their introductory psychology course took part in the experiment, with 37 participants assigned ran- domly to the survival group and 36 assigned to the moving group.
Lists and design. The list of words used was the same as the list of seemingly unrelated words used by Burns et al. (2011). It consisted of four words from each of 12 ad hoc categories (e.g., things women wear, things that have an odour, and things that are round). The ad hoc list was used to allow for a more direct comparison with Burns et al.’s results. All 24 items from a random six categories were presented first, followed by the 24 items from the remaining six cate- gories. The items from each set of six categories were presented in random order, with the exception that no two items from the same
category were presented in adjacent serial posi-
tions. We presented all items from six categories
prior to presenting any items from the remain-
ing categories to keep the procedures in both
experiments consistent (Experiment 2 used a
within-list manipulation of scenario type, neces-
sitating that the list to be divided into two sets of
24 words, so we did the same in Experiment 1).
On the five-alternative-choice recognition test,
each list item was presented with four lures
which were selected from the same ad hoc
category as the list item.
Procedure. Participants were informed that they would be seated in front of a computer
where they would be required to perform a rating
task on a list of words. With the exception that
participants were required to rate the relevance of
words on a 1�4 scale instead of a 1�5 scale, the instructions for both the survival scenario and the
moving scenario were identical to those used by
Nairne et al. (2007). The survival-rating task
required participants to rate each word according
to how relevant it would be to their survival if
they were stranded in the grasslands of a foreign
land. The moving scenario instructions required
participants to rate each word according to how
relevant it would be if they were moving to a new
home in a foreign land. Words were presented for
6 seconds each, and centred on the middle of the
computer monitor. This rating scale (1 �extre- mely irrelevant, 2 �somewhat irrelevant, 3 � somewhat relevant, and 4 �extremely relevant) remained on the lower portion of the screen
during list presentation. Following list presentation, instructions were
read describing the 2-minute digit-recall task (a
filler task used by Nairne et al., 2007, and others
to delay recall). The task consisted of four trials
in which participants see seven digits (ranging
from 0 to 9) one at a time each for 1 second,
followed by a 15-second recall period in which
participants typed the digits in the order they
were shown. Following this distractor task, parti-
cipants were given 10 minutes to write the
previously presented words on a recall sheet in
any order. Participants were asked to draw a line
under the last word recalled after each minute of
recall, which allowed for the cumulative recall
analysis. An untimed final recognition test was
administered approximately 45 seconds after the
recall test.
SURVIVAL PROCESSING 699
Results and discussion
Independent groups t-tests revealed that neither
the relevance ratings nor response times differed
across conditions, t(71) �0.24, and t(71) � � 0.12, respectively (see Table 1). These results
rule out the possibility that any of the memory
measures are the result of processing time differ-
ences or congruity effects. As can be seen in Table 1 there was a
significant recall advantage for the survival group
over the moving group, t(71) �2.09, d�0.49. The cumulative-recall scores, displayed in Figure 2,
show that the two groups recalled roughly the
same number of items during the first 2 minutes
of recall. However, the survival group tended to recall more items during each of the next several minutes of the recall period. We used Equation 1 to produce individual estimates of l and n(8) for participants in both groups. The estimates (see Table 1) revealed no between-group differences in the rate of approach to asymptote, t(71) � � 1.18. However, the survival group did produce a higher estimate of asymptotic recall than the moving group, t(71) �2.15, d�0.51. These curve estimates are highly similar to those found by Burns and Schoff (1998) in their experiment that compared two conditions, both of which per- formed relational processing, but only of which performed item-specific processing.Finally, the survival group also produced significantly higher recognition performance than the moving group, t(71) �3.92, d�0.93. The recognition results suggest that the survival task enhanced item- specific processing. However, the recognition test was given after free recall, so it is possible that the recognition scores are contaminated by prior recall. To help rule out this possibility we analysed recognition performance only for those items not successfully recalled. The mean percen- tage of non-recalled items that were correctly recognised by each group is presented in Table 1. The survival group produced significantly higher conditional recognition percentages than the moving group, t(71) �4.02, d�0.95. These results show that the recognition advantage for the survival group extended to the non-recalled items.
The results replicated the significant recall advantage for the survival processing group over the moving group (e.g., Nairne et al., 2007). The unique aspect of our experiment, however, is that we analysed cumulative-recall curves, as well as recognition performance to assess potential dif- ferences in item-specific and relational proces- sing. Both measures of item-specific processing (cumulative recall and recognition performance) showed differences, converging on the conclusion that the survival scenario produced more item- specific processing than the moving scenario. The initial portions of the cumulative-recall curves, however, provided no evidence of a relational processing difference. On the basis of this pattern of results we suggest that previously reported demonstrations of recall differences between survival processing and other scenarios were the result of a differential processing of item-specific information, a conclusion consistent with that of Kroneisen and Erdfelder (2011).
Recall Duration (Min) 0 2 4 6 8 10
M e a n C
u m
u la
tiv e R
e ca
ll
0
5
10
15
20
Survival Moving
Figure 2. Mean cumulative recall percentages for the two
conditions tested in Experiment 1. Error bars represent 95%
confidence intervals. Because the comparisons of greatest
interest are between the survival and moving groups, the error
bars are based on the error term from separate ANOVAs
comparing the two groups at each minute of recall (see Loftus
& Masson, 1994).
TABLE 1
Mean performance measures for Experiment 1 (pure-list
design) as a function of type of orienting task
Type of orienting task
Survival Moving
Measure M SD M SD
Rating 2.28 0.32 2.26 0.36
Response time (ms) 2473 458 2485 462
Recall 19.19 4.56 16.72 5.49
Recognition% 93.13 5.87 84.20 12.50
Cond. Recognition% 90.11 7.97 78.48 15.63
Approach to asymptote (l) 0.63 0.25 0.73 0.43
Asymptote (n(8)) 19.06 4.30 16.44 6.01
700 BURNS ET AL.
EXPERIMENT 2
The survival processing advantage over other scenarios has been obtained in both pure-list and mixed-list designs. Experiment 1 suggests that, with pure-list designs, the survival processing effect is the result of greater item-specific proces- sing given to the survival items. It is tempting to conclude that this item-specific processing differ- ence is also responsible for the survival processing advantage found in mixed-list designs. This con- clusion may be premature, however, because there are several list-learning effects for which item-specific and relational processing differences occur in one type of design but not the other (e.g., McDaniel, Einstein, DeLosh, May & Brady, 1995; Serra & Nairne, 1993). Therefore we thought it was important to test whether the survival pro- cessing advantage that has been found with mixed-list designs is also accompanied by an item-specific processing difference.
Method
We used a within-participants (and mixed-list) manipulation of scenario type in Experiment 2, rather than the between-participants manipula- tion used in Experiment 1. Half of the 28 participants rated the first 24 words for their relevance to the survival scenario and then rated the remaining words for their relevance to the moving scenario. The other 14 participants were given the two scenarios in the reverse order. All other aspects of the procedure were identical to those of Experiment 1.
Results and discussion
Dependent groups t-tests showed that response ratings did not differ among the two conditions, t(27) � �0.79, nor did the response times, t(27) � �0.45 (see Table 2). Replicating the main result of Experiment 1, the overall recall scores revealed that words rated for their survival relevance were recalled better than the words rated for moving relevance, t(27) �2.51, d�0.51.1
Although participants recalled the words rated for survival- and moving-relevance together, we plotted cumulative recall for each set of words separately. Burns and Hebert showed, that the effects of relational and item-specific processing on the shapes of the cumulative-recall curves produced in within-list designs are similar to those found in between-list designs.2 Not surpris- ingly, with the obvious exception that recall is much lower because recall is based on 24 words
instead of 48, the cumulative-recall curves, shown in Figure 3, produced a pattern of results very similar to that of Experiment 1, resulting in no significant difference between conditions for l, t(27) � �0.40, but a significantly higher estimate of n(8) for the survival condition, t(27) �2.75, d�53. Both the standard recognition percentages and the conditionalised scores produced a sig- nificant advantage for the survival condition: t(27) �2,28, d�0.37, and t(27) �2.65, d�0.61, respectively.
Experiment 2 closely replicated the main findings of Experiment 1. The survival processing advantage in free recall was accompanied by greater recognition performance and a cumula- tive-recall advantage in the latter portion of the
TABLE 2
Mean performance measures for Experiment 2
(mixed-list design as a function of type of orienting task
Type of orienting task
Survival Moving
Measure M SD M SD
Rating 2.26 0.33 2.33 0.37
Response Time (ms) 2340 421 2377 444
Recall 12.32 4.06 10.43 3.39
Recognition% 92.71 8.76 89.29 9.52
Cond. Recognition% 92.15 9.35 85.24 12.94
Approach to Asymptote (l) 0.63 0.35 0.66 0.28
Asymptote (n(8)) 12.33 4.23 10.28 3.47
1 The original standard deviations were used in calculating
Cohen’s d, rather than the pooled standard deviations
corrected for the amount of correlation between the two
scores (see Dunlap, Cortina, Vaslow, & Burke, 1996).
2 Burns et al. (2011) did not present cumulative-recall
curves for their mixed-list experiments because participants
tended to cluster recall by categories, recalling most of the
items from one processing condition before recalling items
from another condition. Although participants did some
clustering of items in our Experiment 2, they did not do so
extensively. On average they switched from recalling items
from one condition to the other condition 7.39 times. More-
over, the cumulative-recall scores for the two different
counterbalanced orders of condition presentation (moving
first vs survival first) produced highly similar patterns, both of
which were similar to the pattern shown in Figure 3.
SURVIVAL PROCESSING 701
recall period, not the initial few minutes. These
results are exactly what would be predicted if
survival processing induces more item-specific
processing than the moving scenario. Experiment
2 suggests that mixed-list survival processing
effects can be accounted for with the two-process
explanation proposed by Burns et al. (2011).
GENERAL DISCUSSION
The memorial advantage conferred by processing
information for survival relevance may be the
consequence of evolutionary adaptations. Burns
et al. (2011) provided evidence for the proximate
mechanisms responsible for the survival advan-
tage and proposed a two-process explanation.
One potential concern with this two-process
explanation is whether it is able to explain the
retention advantage for the survival scenario over
control tasks involving other scenarios (as op-
posed to non-scenario control tasks). It is not
obvious why the survival scenario would induce
more item-specific or more relational processing
than other scenarios, some of which are very
similar in terms of thematic structure. Although
the present results do not explain why, they
clearly suggest that the survival scenario pro-
motes more item-specific processing than the
moving scenario.
Does the survival scenario also increase relational processing?
The results also suggest that the survival scenario does not increase relational processing relative to the moving scenario, although this conclusion is more tentative because there was a slight recall advantage for the survival conditions during the first few minutes of recall. This pattern of results leaves open the possibility of a slight relational processing advantage for the survival task. How- ever, it seems unlikely that such a small relational processing difference would be responsible for the obtained recall differences.
Even if our measures of relational processing showed absolutely no difference between condi- tions, we would not be able to definitively conclude that the two conditions encoded an equivalent amount of relational information. It may be that survival processing enhanced rela- tional processing, but that the relational informa- tion was not used to guide retrieval. Perhaps participants focused primarily on the extensive item-specific cues during recall, ignoring the relational cues. This explanation is similar in many respects to the differential-retrieval-process framework proposed by McDaniel, DeLosh, and Merritt (2000) to explain the bizarreness effect. That hypothesis states that ‘‘contextual factors . . . influence the extent to which various types of information are used at retrieval’’ (p. 1045), suggesting that under some conditions partici- pants may use only one type of information to guide retrieval even when other types are avail- able. But if so our results would still suggest that the survival processing recall advantage is the result of enhanced item-specific processing, be- cause the relational processing was not used to benefit recall.
There is yet another reason to expect that survival processing may sometimes enhance the encoding of relational information. As men- tioned in the introduction, research by Howe and Derbish (2010) and Otgaar and Smeets (2010) using semantically or categorically related list words, which are known to increase proces- sing of relational information, has suggested a relational processing advantage for the survival condition relative to the moving condition. Ot- gaar and Smeets argued that survival processing enhances gist processing, which increases recall of both true and false items. Gist processing is often viewed as similar to, or a form of, relational
Recall Duration (Min)
0 2 4 6 8 10
M e a n C
u m
u la
ti v e R
e c a ll
0
2
4
6
8
10
12
14
Survival Moving
Figure 3. Mean cumulative recall scores for the two condi-
tions tested in Experiment 2. Error bars represent 95%
confidence intervals. Because the comparisons of greatest
interest are between the survival and moving conditions, the
error bars are based on the error term from separate within-
participant ANOVAs comparing the two conditions at each
minute of recall (see Loftus & Masson, 1994).
702 BURNS ET AL.
processing, especially with the procedures used in their study (cf. Burns et al., 2006). Howe and Derbish also suggested that survival processing induces more relational (or associative proces- sing) because it increases the ease with which themes that relate to (or integrate) many list items are activated. These integrative themes provide an organisational structure for the list. It seems plausible, therefore, that some encoding procedures, especially those involving related lists, may produce a relational processing differ- ence between the survival scenario and other scenarios. Again, however, most studies have used unrelated or minimally related word lists, suggesting that the proximate cause of the survival processing advantage typically reported in the literature is an increase in item-specific processing.
Possible explanations for the enhanced item-specific processing associated with survival
Our results extend the explicative power of the two-process account to the research using alter- native scenarios as control conditions. They also provide an important clue about the nature of survival processing: Apparently there is some- thing unique about the survival scenario that typically fosters item-specific processing relative to other scenarios.
This finding is somewhat surprising, at least in relation to Burns et al.’s (2011) findings. They found that, when they used the same list of ostensibly unrelated words as used in the present experiments, there was no item-specific proces- sing difference between the survival condition and non-scenario control conditions (pleasantness rating and category sorting). They suggested that the unrelated list structure fostered item-specific processing for all conditions (see also Hunt & Einstein, 1981). If unrelated words fostered item- specific processing in Burns et al.’s study, why didn’t they do so to the same extent for our moving condition? We are unable to provide a definitive explanation for this interesting discre- pancy. We speculate, however, that when scenar- ios are included in the encoding task, relational (or thematic) information may be inherently encoded (e.g., by providing themes for inte- grating the items). The encoding of this rela- tional information may reduce the amount of
item-specific processing typically induced by un- related words.
Regardless of the reason, it appears that the use of unrelated lists does not equate item- specific processing across scenarios. Why might that be? One possibility alluded to by Kroneisen and Erdfelder (2011) is that the survival task requires the solving of more problems (e.g., finding food and water, and avoiding predators) than control scenarios. Although this is possible for some scenarios, other scenarios seem very closely matched on this dimension. For example, in the burglary scenario used by Kang et al. (2008) and others, participants were required to find people to help them rob a bank as well as to gather supplies for the robbery. Similarly, Nairne and Pandeirada (2010) used nearly identically worded scenarios, changing only a couple of words pertaining to the nature of the problem, not the number of problems (e.g., finding medic- inal plants vs finding relevant antibiotics).
A second possibility that is consistent with Klein, Robertson, and Delton’s (2010) findings is that survival processing involves more planning than other scenarios, with the additional planning resulting in more item-specific processing. Klein et al. showed that the extent to which camping scenarios involved planning largely determined free recall differences, with a future camping scenario producing recall superior to that of a survival scenario. At first glance, however, it would seem that planning would be more likely to result in greater relational processing, not greater item-specific processing; anyone who has planned extensively for a future event realises that successful planning depends heavily on good organisational skills.
If planning is not the critical factor responsible for the survival advantage, then perhaps the survival task induces more self-relevant proces- sing than moving. Burns et al. (2011) speculated that survival processing might induce more self- referential processing than control conditions, and Klein (2012) has provided some evidence in favour of this explanation. The present finding that survival increases item-specific processing relative to other scenarios fits nicely with this self-referential processing explanation. The self- reference task has been shown to increase both item-specific and relational processing relative to different control tasks (e.g., Klein & Loftus, 1988), so it is possible that the survival scenario involved more self-referential processing, which produced both the recall and item-specific
SURVIVAL PROCESSING 703
processing advantage. Of course the problem with this explanation is that there was no difference in amount of relational processing between the survival and moving conditions. One solution is that, as suggested above, the use of scenarios induces relational processing for both conditions, thereby eliminating, or at least minimising, the relational processing advantage for the survival condition.
Still another possibility is that survival proces- sing instils thoughts of dying (i.e., thoughts of not surviving), thereby placing participants in a mor- tality-salient state. Hart and Burns (2012) offered this suggestion and showed that free recall of a list of words increases after participants are placed in a mortality-salient state. They suggested that mortality salience may lead individuals to process information more deeply or complexly than usual, and tentatively suggested that this deep proces- sing might be partly responsible for the survival processing advantage. Clearly more research is needed to fully understand why the survival scenario improves item-specific processing.
Future directions
One concern for future research is to explain a discrepancy between the results of Burns et al. (2011) and those of Otgaar and Smeets (2010). The former authors found that pleasantness rating of a categorically related list produced recall equivalent to that of survival processing, whereas Otgaar and Smeets used categorically related items and found that survival processing re- mained superior to pleasantness rating, thus posing a challenge to the two-process account. There were several procedural differences be- tween studies, including the fact that Burns et al. intermixed items from different categories within the list, whereas Otgaar and Smeets blocked items by category. Another difference is that the list used by Burns et al. contained 4 items from each of 12 categories, whereas the category size used by Otgaar and Smeets was much larger (10 items from each of 6 categories). Both of these procedural differences likely resulted in considerably more relational processing for Otgaar and Smeets’ participants. In fact, Englekamp, Biegelmann, and McDaniel (1998) showed that increases in category size increase relational pro- cessing but have little effect on item-specific processing, regardless of orienting task (cf. Hunt & Seta, 1984). It is possible, therefore, that
item-specific processing differences were elimi- nated in both studies. However, the survival processing effect might have persisted in Otgaar and Smeet’s experiment because, under condi- tions of abundant relational information, survival processing utilises that information more so than pleasantness rating. This conclusion is consistent with Otgaar and Smeets’ conclusion that the survival task resulted in more gist processing. It would be interesting to replicate Otgaar and Smeets’ (2010) experiment, using the various measures of item-specific and relational proces- sing employed in our experiments.
Of course, it is also possible that control tasks that foster both types of processing minimise, but do not eliminate, the survival processing recall advantage. Perhaps the absence of a survival processing effect in the Burns et al. (2011) studies represents a failure to detect the small effect. Logically, however, control tasks that foster both types of processing must reduce the survival processing effect relative to tasks resulting in only one type of processing. We know this is true because the former tasks have consistently produced higher recall than the latter tasks (e.g., Einstein & Hunt, 1980; Hunt & Einstein, 1981).
Another interesting topic for future research concerns false memories. As noted in the intro- duction, both Howe and Derbish (2010) and Otgaar and Smeets (2010) found that survival processing not only increases recall and recogni- tion of items from DRM lists, it also increase false memories. Burns et al. (2006, 2007) used several of the indices of relational and item-specific processing that we used in the present experi- ments to assess the type of processing given to both list items and critical lures in the DRM task. At least as determined by the indices used, the results suggested that list items in the DRM task tended to receive more relational processing than the critical lures but that the critical lures actually received more item-specific processing*a finding that contradicts most theoretical accounts of false memories. Moreover, manipulations that in- creased relational processing of the DRM list items resulted in even greater item-specific pro- cessing of the critical lures, whereas manipula- tions increasing item-specific processing of the list items decreased item-specific processing (and false recall) of the lures.
Perhaps, then, survival processing increases both true memories and false memories for different reasons? If, as Otgaar and Smeets (2010), and Howe and Derbish (2010) speculate,
704 BURNS ET AL.
survival processing increases gist or relational
processing of the list items under some conditions,
this additional relational processing may actually
enhance item-specific processing of the critical
lures. If this is true, then one task for researchers
theorizing about the evolutionary significance of
false memories (e.g., Howe, 2011; Howe &
Derbish, 2010) is to consider how increasing the
item-specific content of those false recollections
might be adaptive.
Manuscript received 16 July 2012
Manuscript accepted 19 November 2012
First published online 24 December 2012
REFERENCES
Bousfield, W. A., & Sedgewick, C. H. W. (1944). An analysis of restricted associative responses. Journal of General Psychology, 30, 149�165.
Burns, D. J. (2006). Assessing distinctiveness: Measures of item-specific and relational processing. In R.R. Hunt & J.B. Worthen (Eds.), Distinctiveness and memory (pp. 109�130). New York, NY: Oxford University Press.
Burns, D. J., Burns, S. A., & Hwang, A. J. (2011). Adaptive memory: Determining the proximate me- chanisms responsible for the memorial advantages of survival processing. Journal of Experimental Psychol- ogy: Learning, Memory, and Cognition, 37, 206�218.
Burns, D. J., Curti, E. T., & Lavin, J. C. (1993). The effects of generation on item and order retention in immediate and delayed recall. Memory & Cogni- tion, 21, 846�852.
Burns, D. J., & Gold, D. E. (1999). An analysis of item gains and losses in retroactive interference. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25, 978�985.
Burns, D. J., & Hebert, T. (2005). Using cumulative- recall curves to assess the extent of relational and item-specific processing. Memory, 13, 189�199.
Burns, D. J., Jenkins, C. L., & Dean, E. E. (2007). Falsely recalled items are rich in item-specific information. Memory & Cognition, 35, 1630�1640.
Burns, D. J., Martens, N. J., Bertoni, A. A., Sweeney, E. J., & Lividini, M. D. (2006). An item gains and losses analysis of false memories suggests critical items receive more item-specific processing than list items. Journal of Experimental Psychology: Learning, Memory, and Cognition, 32, 277�289.
Burns, D. J., & Schoff, K. M. (1998). Slow and steady often ties the race: The effects of item-specific and relational processing on cumulative recall. Journal of Experimental Psychology: Learning, Memory, and Cognition, 24, 1041�1051.
Congleton, A., & Rajaram, S. (2012). The origin of the interaction between learning method and delay in the testing effect: The roles of processing and conceptual retrieval organisation. Memory & Cog- nition, 40, 528�539.
Cosmides, L., & Tooby, J. (1987). From evolution to behavior: Evolutionary psychology as the missing link. In J. Dupré (Ed.), The latest on the best: Essays on evolution and optimality (pp. 276�306). Cambridge, MA: The MIT Press.
Dunlap, W. P., Cortina, J. M., Vaslow, J. B., & Burke, M. J. (1996). Meta-analysis of experiments with matched groups or repeated measures designs. Psychological Methods, 1, 170�177.
Einstein, G. O., & Hunt, R. R. (1980). Levels of processing and organisation: Additive effects of individual item and relational processing. Journal of Experimental Psychology: Human Learning and Memory, 6, 588�598.
Engelkamp, J., Biegelmann, U., & McDaniel, M. A. (1998). Relational and item-specific information: Trade-off and redundancy. Memory, 6, 307�333.
Hart, J., & Burns, D. J. (2012). Nothing concentrates the mind: Thoughts of death improve recall. Psycho- nomic Bulletin & Review, 19, 264�269.
Hermann, D. J., & Chaffin, R. J. S. (1976). Number of available associations and rate of association for categories in semantic memory. Journal of General Psychology, 95, 227�231.
Hermann, D. J., & Murray, D. J. (1979). The role of category size in continuous recall from semantic memory. Journal of General Psychology, 101, 205�218.
Hodge, M. H., & Otani, H. (1996). Beyond category sorting and pleasantness rating: Inducing relational and item-specific processing. Memory & Cognition, 24, 110�115.
Howe, M. L. (2011). The adaptive nature of memory and its illusions. Current Directions in Psychological Science, 20, 312�315.
Howe, M. L., & Derbish, M. H. (2010). On the susceptibility of adaptive memory to false memory illusions. Cognition, 115, 252�267.
Hunt, R. R., & Einstein, G. O. (1981). Relational and item-specific information in memory. Journal of Verbal Learning and Verbal Behavior, 20, 497�514.
Hunt, R. R., & McDaniel, M. A. (1993). The enigma of organisation and distinctiveness. Journal of Memory and Language, 32, 421�445.
Indow, T., & Togano, K. (1970). On retrieving sequence from long-term memory. Psychological Review, 77, 317�331.
Hunt, R. R., & Seta, C. E. (1984). Category size effects in recall: The roles of relational and individual item information. Journal of Experimental Psychology: Learning, Memory, and Cognition, 10, 454�464.
Johnson, D. M., Johnson, R. C., & Mark, A. L. (1951). A mathematical analysis of verbal recall. Journal of General Psychology, 44, 121�128.
Kang, S. H. K., McDermott, K. B., & Cohen, S. M. (2008). The mnemonic advantage of processing fitness-relevant information. Memory & Cognition, 36, 1151�1156.
Kaplan, I. T., Carvellas, T., & Metlay, W. (1969). Searching for words in letter sets of varying size. Journal of Experimental Psychology, 82, 377�380.
Klein, S. B. (2012). A role for self-referential processing in tasks requiring participants to imagine survival on the savannah. Journal of Experimental Psychology: Learning, Memory, and Cognition, 38, 1234�1242.
SURVIVAL PROCESSING 705
Klein, S. B., & Loftus, J. (1988). The nature of self- referent encoding: The contributions of elaborative and organisational processes. Journal of Personality and Social Psychology, 55, 5�11.
Klein, S. B., Loftus, J., Kihlstrom, J. F., & Aseron, R. (1989). Effects of item-specific and relational in- formation on hypermnesic recall. Journal of Experi- mental Psychology: Learning, Memory, and Cognition, 15, 1192�1197.
Klein, S. B., Robertson, T. E., & Delton, A. W. (2010). Facing the future: Memory as an evolved system for planning future acts. Memory & Cognition, 3, 13�22.
Kostic, B., McFarlan, C. C., & Cleary, A. M. (2012). Extensions of the survival advantage in memory: Examining the role of ancestral context and implied social isolation. Journal of Experimental Psychology: Learning, Memory, and Cognition, 38, 1091�1098.
Kroneisen, M., & Erdfelder, E. (2011). On the plasticity of the survival processing effect. Journal of Experi- mental Psychology: Learning, Memory, and Cogni- tion, 37, 1553�1562.
Loftus, G. R., & Masson, M. J. (1994). Using confidence intervals in within-subject designs. Psychonomic Bulletin & Review, 1, 476�490.
McDaniel, M. A., DeLosh, E. L., & Merritt, P. S. (2000). Order information and retrieval distinctive- ness: Recall of common versus bizarre material. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26, 1045�1056.
McDaniel, M. A., Einstein, G. O., DeLosh, E., May, C., & Brady, P. (1995). The bizarreness effect: It’s not surprising, it’s complex. Journal of Experimental Psychology: Learning, Memory, Cognition, 21, 422�435.
McDaniel, M. A., Moore, B. A., & Whiteman, H. L. (1998). Dynamic changes in hypermnesia across early and late tests: A relational/item-specific ac- count. Journal of Experimental Psychology: Learn- ing, Memory, and Cognition, 24, 173�185.
Nairne, J. S., & Pandeirada, J. N. S. (2011). Congruity effects in the survival processing paradigm. Journal
of Experimental Psychology: Learning, Memory, and Cognition, 37, 539�549.
Nairne, J. S., & Pandeirada, J. N. S. (2008). Adaptive memory: Is survival processing special? Journal of Memory and Language, 59, 377�385.
Nairne, J. S., & Pandeirada, J. N. S. (2010). Adaptive memory: Ancestral priorities and the mnemonic value of survival processing. Cognitive Psychology, 61, 1�22.
Nairne, J. S., Pandeirada, J. N. S., Gregory, K. J., & Van Arsdall, J. E. (2009). Adaptive memory: Fitness relevance and the hunter-gatherer mind. Psycholo- gical Science, 20, 740�746.
Nairne, J. S., Pandeirada, J. N. S., & Thompson, S. R. (2008). Adaptive memory: The comparative value of survival processing. Psychological Science, 19, 176� 180.
Nairne, J. S., Thompson, S. R., & Pandeirada, J. N. S. (2007). Adaptive memory: Survival processing en- hances retention. Journal of Experimental Psychol- ogy: Learning, Memory, and Cognition, 33, 263�273.
Otgaar, H., & Smeets, T. (2010). Adaptive memory: Survival processing increases both true and false memory in adults and children. Journal of Experi- mental Psychology: Learning, Memory, and Cogni- tion, 36(4), 1010�1016.
Roediger, H. L., Stellon, C. C., & Tulving, E. (1977). Inhibition from part-list cues and rate of recall. Journal of Experimental Psychology: Human Learn- ing and Memory, 3, 174�188.
Serra, M., & Nairne, J. S. (1993). Design controversies and the generation effect: Support for an item-order hypothesis. Memory & Cognition, 21, 34�40.
Sherry, D. F., & Schacter, D. L. (1987). The evolution of multiple memory systems. Psychological Review, 94, 439�454.
Smeets, T., Otgaar, H., Raymaekers, L., Peters, M. V., & Merckelbach, H. (2012). Survival processing in times of stress. Psychonomic Bulletin & Review, 19, 113�118.
706 BURNS ET AL.
Copyright of Memory is the property of Psychology Press (UK) and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
