Memory & Cognition (2022) 50:1683–1693

https://doi.org/10.3758/s13421-022-01293-3

Negative recency effects in delayed recognition: Spacing,

consolidation, and retrieval strategy processes
Rona Sheaffer1 · Daniel A. Levy1

Accepted: 14 February 2022 / Published online: 11 March 2022

© The Psychonomic Society, Inc. 2022

Abstract
While items learned immediately before testing are generally remembered better than prior items in a study list, in
delayed testing this relationship is reversed, yielding a negative recency effect. To adjudicate between the strategic
rehearsal and spacing accounts of this phenomenon, we examined performance of 169 participants on a delayed rec-
ognition test following multiple sessions requiring the study and immediate free recall testing of 16 lists of 16 words.
This revealed a strong effect of the amount of spacing between initial study position and initial free recall position on
the degree of negative recency, supporting the spacing account. Furthermore, these spacing effects were nonmonotonic,
suggesting that they are mediated by consolidation processes. Additional analyses indicate that strategies and rehearsal
opportunities may also contribute to the effects of within-list encoding position on subsequent long-term memory, but
for recall more than for recognition.

Keywords Recognition · Spacing · Recall · Consolidation · Rehearsal

Imagine that you are at a departmental function, meeting Recently, Kahana and colleagues (Kuhn et al., 2018)
unfamiliar researchers from other labs and trying to remem- have appealed to the spacing effect to explain an interest-
ber their names. You even use the trick you once read about, ing aspect of free recall. In immediate free recall of word
of repeating the person’s name in conversation with them: lists, the most recently studied items are generally bet-
“Hi George . . . nice to make your acquaintance, George . ter recalled than mid-list items, an effect often attributed
. . I really love your work, George . . .”. Now imagine your to the retrieval of final items from short-term memory
chagrin the next day, when you bump into George in the (Davelaar et al., 2005; Talmi et al., 2005). When cogni-
corridor, but can’t for the life of you remember his name. tively demanding activity intervenes between study and
Why didn’t the name repetition trick help consolidate your test, this effect dissipates. Furthermore, when after ini-
memory of that name? tial immediate free recall and subsequent cognitive activ-
There’s a good explanation for your forgetfulness—you ity, a final free recall test is administered, the prominent
ignored the spacing effect. Repeated exposure to information recency effect observed in immediate free recall is actu-
strengthens memory for it, but spaced repetitions are most ally reversed, with list-end items being remembered more
effective (Madigan, 1969; Melton, 1970; and more recently poorly than earlier items (Craik, 1970). In other words,
Cepeda et al., 2006; Delaney et al., 2010). Had you repeated there is a positive within-list recency effect for imme-
George’s name throughout the event and not in a massed diate retrieval, and a negative within-list recency effect
fashion, the chances of retaining that information for future for delayed re-retrieval. Initially, this negative recency
recall would increase significantly. effect was adduced to provide evidence in favor of dual-
store models of word list memory. Craik argued that
items from the end of a word list were held in short-term
memory buffer for the least amount of time. Therefore,
* Daniel A. Levy
daniel.levy@idc.ac.il although they had a higher probability of being imme-
diately retrieved, they have the least strength in longer-
1
Baruch Ivcher School of Psychology, Interdisciplinary term memory. Single store models sought alternative
Center Herzliya, 8 HaUniversita St, 46150 Herzliya, Israel

13
Vol.:(0123456789)
1684 Memory & Cognition (2022) 50:1683–1693

explanations of the negative recency effect. Contrary to recall (Craik, 1970). In contrast, Cohen (1970) reported pos-
accounts that attribute it to strategic rehearsal processes itive recency effects in delayed recognition following imme-
(e.g., Tan & Ward, 2000), Kuhn et al. (2018) provide evi- diate free recall tests. Engle (1974) noted that in the study of
dence1 that negative recency is a function of spacing (i.e., Craik et al. (1970), inclusion in recognition success analysis
the amount of time or the number of cognitive steps inter- was conditional on being successfully recalled in the imme-
vening between initial study and immediate free recall). diate test, while in Cohen (1970) inclusion in the recognition
This challenges the strategic rehearsal processes account, analysis was not conditional on prior production in imme-
which ascribes importance only to an item's list position diate free recall. Engle (1974) therefore directly compared
in the presentation phase and not to its output position in conditional and unconditional analyses, as well as manip-
the immediate recall phase. Kuhn and colleagues propose ulating presentation rate. He reported that words recalled
that since items presented late in an encoding list are likely in initial free recall (IFR) exhibited negative recency at all
to be immediately recalled early on, those items have the presentation rates, but that was not the case for non-recalled
least amount of spacing between activations, and therefore words. Engle (1974) also reported that subsequent recogni-
the subsequent memory for them is weaker than for items tion confidence ratings for words that had been recalled in
earlier in the encoding sequence. However, the crucial fac- IFR increased linearly with IFR output position, reflecting
tor is not study list position alone, but the degree of spac- greater subsequent memory strength. Similarly, McCabe and
ing resulting from the study list position and the output Madigan (1971) reported that the final (recency) item in a
position in the immediate free recall test. This account can five-element sequence of word pairs was identified with the
accommodate the negative recency phenomenon within a lowest confidence level in delayed recognition after immedi-
single-store model of list memory retrieval patterns. ate free recall. In contrast, Engle and Durban (1977), exam-
If the spacing of prior exposures is responsible for the long- ining both auditory and visual presentations, found a small
term negative-recency effect, we should expect to see that positive recency effect, and Darley and Murdock (1971),
effect not only on subsequent (final) free recall, but on other who employed a three-alternative forced choice recognition
expressions of memory traces, such as recognition tests. This test, found no recency effects.
is because the second exposure via recall is said to serve as a Notably, delayed recognition accuracy (or confidence) as
further encoding event (Craik, 1970; Kuhn et al., 2018). That a function of specific study-output lag for each IFR item was
becomes apparent upon considering that retrieval in initial not reported in any of the abovementioned studies. Lacking
free recall is a paradigmatic case of retrieval practice (recently that data, even in those studies that do report a negative
reviewed by McDermott, 2021). Although retrieval practice recency effect, it is difficult to adjudicate between Kuhn
affects recollection more strongly than familiarity, in a multi- et al.’s (2018) claim that spacing is responsible for long-
list paradigm like the current one, it was found to improve term negative recency in delayed recognition and an alter-
overall recognition (Chan & McDermott, 2007, Exp. 3). Simi- native explanation—that list-end items strategically receive
larly, spacing effects are reported for recognition memory, more immediate rote rehearsal (McCabe & Madigan, 1971;
especially for its recollective aspect (Glenberg, 1976; Hintz- Reitman, 1970). In that alternative account, those list-end
man, 1969; Parkin & Russo, 1993; Zhao et al., 2015). There- items are also said to be produced in initial output positions
fore, to the extent that the mechanisms responsible for the based on registration in short-term memory, while primacy
spacing effect yield negative recency for recall, they should items, when immediately remembered, are retrieved from
yield negative recency for recognition. If the effect of study- long-term memory formed during initial encoding, and so
recall spacing duration is found for subsequent (final) recog- are also more successfully identified in delayed recognition
nition testing as well, it would provide additional support for tests.
the explanation put forward by Kuhn and colleagues; absent Furthermore, there is reason to suspect that in a final rec-
such a finding, the evidence in favor of the spacing account ognition test a different pattern of effects might be observed
and against the strategic retrieval account would be equivocal. than those reported by Kuhn et al. (2018) for final free recall.
List serial position effects on delayed (subsequent/final) Numerous studies by Zacks and colleagues (e.g., Swallow
recognition tests have been explored in several classic stud- et al., 2009; reviewed in Radvansky & Zacks, 2017) have
ies. Craik et al. (1970) reported a negative recency effect in documented the strong impact of proximity to event bound-
delayed recognition following immediate free recall tests, aries on subsequent memory. In recognition memory, the
parallel to the negative recency effect found for final free most potent retrieval cue is the memorandum itself, which
is used as the mnemonic probe—a copy cue (Tulving, 1983).
1
In a test of recognition, the strategic retrieval processes
Based on the analysis of a large database of free recall perfor-
employed in free recall may be overshadowed by copy cue
mance (the Penn Electrophysiology of Encoding and Retrieval Study
[PEERS]); see references to numerous studies of these data provided strength and proximity to boundaries— i.e., study list end
by Kuhn et al. (2018). and/or to an early output position in the initial free recall

13
Memory & Cognition (2022) 50:1683–1693 1685

stage—which lead to strong encoding might yield the best Earlier PEERS publications report details of the method
final recognition performance. (for a more complete description, see Healey et al., 2014;
Therefore, examination of negative recency in delayed Lohnas & Kahana, 2014). In brief, each item was on the
recognition, and identification of its possible mechanism screen for 3,000 ms, followed by jittered 800–1,200-ms
if occurring, require detailed information about the lags interstimulus interval.4 After the last item in the list, there
between study presentation and production in IFR. Fortu- was a 1,200–1,400-ms jittered delay, after which a tone
nately, the PEERS data set includes a number of experimen- sounded, a row of asterisks appeared, and the participant
tal sessions in which delayed recognition tests were adminis- was given 75 seconds to vocally recall the just-presented
tered at the conclusion of learning and immediate free recall items in any order. If a session was randomly selected for
testing of all lists. Furthermore, unlike earlier reports, this final free recall (FFR) test, participants performed the FFR
data enable examination of the exact serial order of recall in test following the 16th immediate free recall (IFR) test. An
immediate free recall, enabling more precise quantification instruction screen informed participants that they had 5 min-
of the study-to-free-recall lag. We were therefore able to utes to recall all the items from the preceding lists in any
assess whether the negative recency effect in delayed recog- order. An old/new recognition test was administered after
nition is indeed a function of spacing between initial encod- either FFR or after the last list’s IFR test. In total, 320 words
ing and immediate free recall in the lists presented during were presented serially on the computer screen, with target/
the earlier phases of the experiment. lure ratio varying with session, and targets comprising 80%,
In addition to investigating whether final recognition per- 75%, 62.5%, or 50% of the total trial items. Participants were
formance would support the spacing hypothesis of Kuhn instructed to indicate for each word whether the test word
et al. (2018) by performing the same analyses they executed, had been presented previously. Recognition trials were self-
but on the recognition data, we took advantage of the scope paced. Feedback on accuracy and reaction time was provided
of the PEERS database to conduct additional analyses track- after each trial.
ing recognition probabilities as a function of list recency
and within-list recency. The most important of these analy-
ses examines the degree of monotonicity of the effects of Results
spacing. As detailed below, this revealed some unexpected
findings. Before turning to the analyses relevant to the question
under consideration, it is instructive to get a sense of the
overall pattens of performance in delayed recognition. One
Method interesting pattern is the probability of final recognition as
a function of the item’s serial position during encoding,
The current investigation is based recognition data col- considering both the list that each item came from and
lected as part of Experiment 1 of the Penn Electrophysiol- the item’s serial position within that list, for a total of 256
ogy of Encoding and Retrieval Study (PEERS).2 One hun- (16 × 16) possible positions. We therefore conducted a
dred seventy-one participants participated in Experiment 1 repeated-measures analysis of variance (ANOVA), with
(consisting of seven experimental sessions).3 Two of these list number (Lists 1–16) and within-list serial position
participants were excluded from the present analyses due (Slots 1–16) as within-subject factors. Mauchly’s test
to data corruption, leaving N = 169. Each of the seven ses- of sphericity indicated that degrees of freedom in both
sions consisted of 16 lists of 16 words presented one at a main effects required correction; Greenhouse–Geiser
time on a computer screen. Each study list was followed by epsilon was applied accordingly. We found a main effect
an immediate free recall test, and each session ended with of between-list position, F(11.34, 1905.58) = 47.92, p <
a comprehensive recognition test. Half of the sessions were .001, and a main effect of within-list position, F(12.72,
randomly chosen to include a final free recall test, which 2136.08) = 10.02, p < .001, but no interaction between
took place before the recognition test. those factors, F(83.19, 13976.51) = 1.09, p = .270. As
can be seen in Fig. 1a (top), there is a positive long-term

2
Raw data from the PEERS studies was obtained from Michael 4
Words were either presented concurrently with a task cue indicat-
Kahana’s web page (http://​memory.​psych.​upenn.​edu). ing one of two judgments (size or animacy) to be made for that word,
3
Kuhn et al. (2018) include the data of PEERS Experiments 2 and 3 or with no encoding task. Kuhn et al. (2018) report that the encod-
in their analyses. Since the former involved continual-distractor free ing task had minimal effects on recall performance, and, importantly,
recall, and the latter included some participants performing external- nearly identical negative-recency effects across encoding conditions.
ized free recall (reporting intrusions), we did not include those data in We have therefore followed their method and aggregated data across
our analyses. these conditions.

13
1686 Memory & Cognition (2022) 50:1683–1693

Fig. 1  Final recognition as a function of serial position during encoding. All 16 lists of Fig. 1a (up) are averaged together in Fig. 1b (down), to
create a within-list serial position curve. Error bars indicate SEM

recency effect across lists, such that, as might be expected, the across-list position is not relevant, nor did it interact
participants recognized more items from recent than from with the within-list effects, we will not further comment
remote lists; this is reflected in the strong positive-slope on that finding.
linear trend in the across list effect, F = 347.1, p < .001, To provide a clearer picture of the shape of the within-
and the absence of a significant quadratic trend, F = 3.61, list serial position curve, we aggregated the data across all
p > .05. This detrimental effect of study–test delay, with 16 lists; in this analysis, we included all items, whether
the concomitant number of intervening lists (possibly or not they were successfully recalled in the initial free
leading to retroactive interference on subsequent recogni- recall (IFR) stage. The shape of the average serial posi-
tion), is not surprising. Moreover, since for the investiga- tion curve presented in Fig. 1b (bottom) illustrates that
tion of the spacing effect proposal the key factor is the there is also a primacy effect within each list. In fact,
within-list position at study (and at initial free recall), and recognition probability appears to decrease with each

13
Memory & Cognition (2022) 50:1683–1693 1687

Fig. 2  Final recognition as a function of recency of encoding, for three classes of items: those recalled in early output positions in IFR (first half
of recalls), those recalled in late output positions in IFR (last half of recalls), and items that were not recalled in IFR

additional item on the list, until reaching its lowest value presented here (including Fig. 1, above) consider final rec-
at the middle of the list (Item 9), followed by the asymp- ognition data only for PEERS Experiment 1 sessions that did
totic mid-list plateau observed in classic serial position not include the FFR test.5 Moreover, the analyses consider
curves for immediate free recall. At the penultimate recognition data separately for items that participants did or
point, recognition probability rises again (Item 15). This did not initially recall during their IFR trials, except for the
is reflected in the presence of both linear (F = 49.32, p detailed spacing analysis, which considers the number of
< .001) and quadratic (F = 26.46, p < .001) components items between encoding and subsequent IFR, and is there-
in the within-list effect. However, the comparison with fore based solely on the former items.
Fig. 3 and the free recall data provided by Lohnas and To test the hypothesis that spacing between encoding and
Kahana (2014); Fig. 3a) indicates that this spike is not an initial retrieval influences final recognition, we classified the
indication of advantage in principle, but rather a function initially recalled items into two categories according to their
of the greater relative proportion for the words in this output positions, defining early output positions as the first half
position achieving successful initial free recall (~81% vs. of outputs, and late output positions as the second half (as in
an average of 59% for items in Positions 9–14), for which Kuhn et al., 2018). This partitioning resulted in three classes
subsequent recognition is better. Notably, and once again of items: not initially recalled, recalled early, and recalled late.
unlike classic serial position curves for immediate free Figure 2 shows the probability of final recognition for these
recall, there is a sharp decline in final recognition suc- three item types as a function of recency of encoding, consid-
cess for the final 16th item, despite items in that position ering both the item’s list number and the item’s serial position
having been initially recalled at a rate of ~94% (Lohnas & within that list. It illustrates three major effects in final recog-
Kahana, 2014; Fig. 3a). The implications of these perfor- nition. First, participants correctly recognized almost all the
mance trends will be explored in the Discussion. items that they recalled during IFR, as was reported for FFR by
To examine the proposal of Kuhn et al. (2018) regarding Kuhn et al. (2018). Second, participants recognized more items
spacing effects on negative recency, we proceeded to ana-
lyze recency effects, examining memory for later list items
compared to earlier items. We assume that the mere retrieval
of an item produces learning (e.g., Karpicke & Roediger, 5
In practice, we found very similar negative-recency effects for
2008). Therefore, to avoid confounds of the effects of final words in both FFR and no-FFR sessions, but still believe it preferable
free recall before long-term recognition, all the analyses to avoid that potential confound.

13
1688 Memory & Cognition (2022) 50:1683–1693

Table 1  Results of a mixed model analysis comparing recognition

probability between the first and last half of retrievals, separately for
each of the serial positions (retrieval position half was used as fixed
factor and participant as a random factor)

Serial Posi- N (first half) N (second df F p value

tion half)

1 166 (3) 166 (3) (1,330) .142 .706

2 165 (4) 165 (4) (1,327) 1.033 .310
3 166 (3) 166 (3) (1,330) .227 .634
4 163 (5) 167 (2) (1,328) .077 .782
5 164 (4) 165 (4) (1,326) 3.162 .076
6 161 (6) 165 (4) (1,323) 1.125 .290
7 163 (4) 165 (4) (1,326) .583 .446
8 164 (3) 166 (3) (1,327) .127 .722
9 164 (4) 167 (2) (1,329) .566 .452
10 160 (9) 166 (3) (1,323) 2.965 .086
11 166 (3) 166 (3) (1,329) 2.666 .103
12 167 (2) 165 (4) (1,329) 14.718 .000
13 164 (5) 167 (2) (1,329) 5.158 .024
14 169 (0) 167 (0) (1,333) 7.145 .008
15 166 (3) 144 (5) (1,307) 54.677 .000
16 168 (1) 101 (6) (1,266) 34.224 .000

Note. Columns 2 and 3 present the number of participants included in

the analysis for each position (participants who recalled one or more
Fig. 3  Upper part of panel: Final recognition probabilities as a func- items), not including excluded outliers. The numbers in parentheses
tion of within-list position, averaged across all 16 lists of Fig. 2. represent the number of participants that were excluded
Error bars depict 1 SEM (calculated using the methods described in
Cousineau, 2005, and Morey, 2008). The large SEM seen at Posi-
tions 15 and 16 for items recalled late and unrecalled items results attenuated when the analysis is restricted to either not-recalled
from a large drop in the number of items recalled at these posi-
tions (see Appendix, Supplementary Table 1, for the exact number
items or items recalled late in the recall period. Thus, this find-
of items recognized in each position). Lower part of panel: Final ing provides additional support for the spacing-based account
free recall probabilities, as reported by Kuhn et al., 2018. For both of negative recency (Craik, 1970; Kuhn et al., 2018).
recall and recognition, the negative recency effect is strongest (i.e., This inference is further supported by an analysis focusing
retrieval probability is relatively weaker) for items recalled in early
output positions but attenuated (i.e., retrieval probability is relatively
on the initially recalled items (the two top curves in Fig. 3),
stronger) for items recalled in later output positions which allowed us to specifically compare items recalled in
the first half of recalls to those recalled in the last half, as a
function of the encoding serial position (ranging from 1-16).
from recent than from remote lists. However, this effect was The number of items in each of these serial positions varied
seen primarily for items that participants failed to recall in IFR. greatly. Only 107 participants (out of 169) recalled one or more
Third, we found a pronounced within-list negative-recency items from the 1­ 6th encoding position in late output positions,
effect, which was most dominant for items that were recalled whereas all 169 did so when considering early output posi-
in early output positions during IFR. To provide a clearer tions. This was simply because items from encoding posi-
picture of the relationship between this negative recency and tion 16 were much more likely to be produced early in the
item type, in Fig. 3 we aggregated the data across all 16 lists.6 recall sequence. Therefore, we subjected these data to mixed
Like the results of Kuhn et al. (2018); Fig. 2a) regarding free model analyses, which is most appropriate for unbalanced data
recall, Fig. 3 demonstrates that the negative recency effect is (Baayen et al., 2008; Tibon & Levy, 2015). For each of the
serial positions, outlier participants whose recognition per-
formance was 3 SD below group mean were excluded7. Item

6
Figure 3 portrays mean performance for each serial position across
all participants rather than means of individual participants’ perfor-
7
mance, because of the different numbers of participants contributing On average, three participants were excluded from each serial posi-
data to the later positions, which required the mixed-models analysis tion. See Columns 2 and 3 on Table 1 for the exact number of partici-
conducted on this data, as detailed. pants excluded from each position.

13
Memory & Cognition (2022) 50:1683–1693 1689

Fig. 4  Probability of final recognition as a function of the spacing culated using the methods described in Cousineau, 2005, and Morey,
between initial presentation and initial recall. A positive correlation 2008). Note that the large SEM seen at the right end of the graph
was found, demonstrating that the probability of recalling an item derives from the scarcity of items recalled after such a large spacing
during final recognition increases as the number of items between the (see Appendix, Supplementary Table 2, for the exact number of items
initial presentation and IFR increases. Error bars depict 1 SEM (cal- recognized for each spacing)

serial position during initial encoding, and IFR output position number of intervening items; possible values range from 0
(first vs. second half) were fixed factors, with participant as to 30). As explained in the Introduction, the spacing account
a random factor (West, 2009). This analysis yielded a main of negative recency posits that memory for items should
effect of item serial position, F(15, 5171) = 16.38, p < .001, improve consistently as the spacing between the two learn-
and a main effect of IFR output position, F(1, 5171) = 49.96, ing episodes (initial encoding and IFR) increases. There-
p < .001. Importantly, and consistent with the spacing-based fore, end-of-list items, which have shorter spacings than
account of negative recency, there was a significant interaction early list items on average, should be remembered less well,
between serial position and IFR output position, F(15, 5171) = and this effect should be strongest for end-of-list items that
9.66, p < .001. As shown in Fig. 3, the proportion of items rec- are recalled early (shortest spacings of all). Consistent with
ognized in Encoding Positions 1–11 was not affected by IFR the spacing account, Fig. 4 portrays the positive correlation
output position, whereas the proportion of recognized items between spacing and recognition probability, such that final
from Encoding Positions 12–16 was significantly lower if pro- recognition performance rises with the spacing between an
duced in the first half of IFR output positions than if produced item’s position during study and its position in IFR. Follow-
in the last half. This is confirmed by the mixed model analyses ing the methods used by Kuhn et al. (2018), we computed
that compared recognition probability between the first and last several variations of the correlation between spacing amount
half, separately for each of the serial positions (see Table 1 for and recognition probability separately for each participant.8
the complete statistical report, and Supplementary Table 1 for Across all possible spacings (0–30), the distribution of cor-
the detailed descriptive statistics). This indicates the specific- relation coefficients was significantly positive (mean correla-
ity of the negative recency effect on final recognition to items tion = 0.27), t(167) = 15.41, p < .001. However, Fig. 4 indi-
previously produced in the first half of IFR output positions, as cates that such correlations might not be equivalent across
might be expected given the shorter spacing between the initial
encoding and retrieval of these items.
8
To gauge the spacing account more directly, we exam- Several participants were excluded from the correlational analyses
ined the probability of correct final recognition as a func- due to perfect performance (constant recognition probability), which
does not/did not allow correlation calculation. The exact number of
tion of the exact spacing between the initial presentation of participants excluded from each analysis can be deduced from the
the items and their recall during the IFR (measured as the degrees of freedom reported.

13
1690 Memory & Cognition (2022) 50:1683–1693

all spacing positions. We therefore dismantled the general 2002) of the activation of the word representations; associa-
correlation into three separate examinations of spacing-rec- tion with multiple contexts might increase the likelihood
ognition correlations for spacing amounts 0–10, 11–20, and of the subsequent recall and recognition of a studied probe
21–30. Indeed, these subset correlations demonstrate that (Smith, 1982; Smith & Handy, 2016). A third possibility,
the overall positive correlation derives completely from the suggested by many earlier studies of within-session spacing
leftmost part of the graph (spacings 0–10), mean r = .34, effects (reviewed by Delaney et al., 2010), is that participants
t(166) = 15.54, p < .001, as clearly seen in Fig. 4, whereas use different encoding strategies for each encounter with a
the correlation in the other two subcategories is either null stimulus in massed versus spaced presentations. This claim
(mean r = −.02), t(160) = 0.67, p = .502, or negative (mean is supported by participant reports (Delaney et al., 2010).
r = −.73), t(168) = 29.42, p < .001.9 This seems to indicate Resolution of this question—whether consolidation fac-
that spacing between prior exposures indeed improves sub- tors, temporal context changes, or encoding processing dif-
sequent recognition performance, but that once sufficient ferences are responsible for negative recency in delayed rec-
spacing is provided, further spacing does not provide addi- ognition—might be provided by the non-monotonic effects
tional memory strength. in the current data (Fig. 4). This finding accords with an
early report of non-monotonic spacing effects on retrieval.
This may be found in Glenberg (1976, Exp. 3), in which
Discussion memory for trigrams was tested at lags of 8, 32 or 64 items
after a second presentation, which lagged 0, 1, 8, 20, or
We assessed the spacing account of the delayed recency 40 items after the initial presentation. Glenberg reported a
effect in retrieval of studied word lists (Kuhn et al., 2018), nonmonotonic spacing advantage at the two longer reten-
by examining whether its predictions would hold not only tion intervals, which approached asymptote following a lag
for final free recall but also for final recognition testing. of 8 items between study presentation – comparable to our
We found that this was indeed the case: in comprehensive finding of an asymptotic trend beginning at a lag of ~10
delayed recognition testing conducted on a large corpus of items. Beyond that report, however, the variety of paradigms
studied words, the probability of a word being correctly rec- employed in earlier research makes it difficult to draw direct
ognized was significantly influenced by the spacing between comparisons. As noted by Benjamin and Tullis (2010), few
its initial presentation and its initial immediate free recall. studies of spacing effects have made use of recognition test-
As in the analyses reported by Kuhn et al. (2018), this was ing. Additionally, nonmonotonicity in the effects of spacing
found both in comparison of halves of recall sessions and for on the order of days and even months between presentations
individual spacings. However, a more nuanced examination (as summarized, e.g., in the meta-analysis of Cepeda et al.,
of this relationship demonstrated that unlike the case in final 2006) are likely to involve very different processes than
free recall (Kuhn et al., 2018, Fig. 2b), this relationship was lags of a maximum of 75 seconds as in the present study. A
not monotonic. Increases from immediate repetition up to more fundamental difference between the present paradigm,
~10 intervening items led to greater likelihood of correct and earlier research is that in a preponderance of studies,
subsequent endorsement, but additional spacing did not fur- the second (and sometimes third, etc.) encounter with to-
ther improve performance. be-remembered information is in an additional study trial
Considering both the consonance and the dissonance identical to the initial study trial. Under such circumstances,
between the recognition and recall spacing effects may be differences in strategic encoding processes between spaced
instructive regarding the possible mechanistic bases of the and massed presentations (Delaney et al., 2010) might very
spacing effect in general (Greene, 1989). Although this well play a role. In contrast, in the current paradigm, the
issue is not discussed by Kuhn et al. (2018), the spacing second exposure to the information is when it is produced
account accords well with the notion that immediate post- in the IFR. In this case, the participants are not engaging
encoding consolidation processes, including changes on the in intentional encoding, but rather in intentional retrieval.
neural level, are required for the stability of newly formed It therefore seems that the two most robust candidates for
memories (Ben-Yakov & Dudai, 2011; Dudai et al., 2015). explaining the current results are consolidation factors and
Alternatively, the advantage of spaced repetitions might be temporal context changes.
in the changing temporal context (Howard & Kahana, 1999, We propose that for recognition memory, which enable
retrieval judgments based on familiarity, the benefits to
repeated encoding might significantly depend on meta-
bolic re-potentiation of the neuronal networks represent-
9
The negative correlation in the third subcategory (spacing 21–30) ing the words to be remembered (Feng et al., 2019; Xue
should be viewed with caution due to the scarcity of items recalled in
this subcategory and the large standard deviation; see Supplementary et al., 2011). Once those neurons have had enough time to
Table 2. recoup their ability to conduct the processes required for

13
Memory & Cognition (2022) 50:1683–1693 1691

Hebbian plasticity, little further benefit might be derived encoding that underlies the testing effect (Karpicke & Roe-
from additional delay (Smolen et al., 2016). While tem- diger, 2008), leading to stronger subsequent recognition in a
poral context factors may also impact on recognition, in delayed test. This distinction cannot account for the asymp-
the final recognition test paradigm, those influences might totic profile of spacing effects observed for all the other input
be attenuated. In this paradigm, the multiple temporally position items in the recognition test but may be relevant to
structured study–test stages of the preceding part of the understanding poor delayed recognition of the final input
experiment are disrupted by the presentation of a com- position items. We noted above that the negative recency
prehensive set of recognition probes in random order, so effect was initially adduced to provide evidence for dual-
those factors might be overshadowed by the strength of store models of word list memory (Craik, 1970); the finding
the recognition test copy cue, as noted in the Introduction. of Kuhn et al. (2018) that the crucial factor is not study list
Therefore, in the current data as well in the comparable position but rather study-to-initial recall spacing provides
retention intervals in Glenberg (1976), spacing benefits an alternative explanation that accommodates single-store
become asymptotic after ~10 items. In contrast, for recall, models. Seemingly, this anomalously poor delayed recogni-
the greater the difference between the temporal context tion of the final list items might be seen as providing support
of the initial (study) and subsequent (initial free recall) for a limited version of dual-store models, applying only to
exposures to a given target word, the greater the variety of the final list item. It seems to us, though, that resolution of
associative cues available, with the greater chance that it the larger issue is better served by paradigms that combine
will be recollected in the final free recall test. As opposed anatomical and physiological assays and interventions with
to the recognition test in which the probes are presented behavioral patterns (e.g., Innocenti et al., 2013; Kloth et al.,
randomly, disrupting the possibility of using temporal con- 2020; Talmi et al., 2005) in providing evidence for or against
text retrieval of one target to cue additional targets, in free process dissociation.
recall participants may use temporal context to perform In conclusion, performance on delayed comprehensive
further retrievals of items that were temporally proximal recognition tasks generally supports the spacing account of
to just-retrieved words. Thus, while both temporal context the negative-recency effect observed in retrieval of word lists
and consolidation process factors affect both free recall (Kuhn et al., 2018). Moreover, the pattern of nonmonoto-
and recognition, the two tasks may be sensitive to these nicity in negative recency effects hints to a link between
influences, to different degrees. Consolidation factors these behavioral findings and cellular-molecular processes
might be more relevant for final familiarity-based recog- implicated in the early consolidation of new learning. The
nition (yielding an asymptotic limit to spacing benefit), rich pattern of results revealed by the large dataset including
while temporal context factors might play a greater role in multiple tests of a large number of participants indicates that
recollection-based final free recall (in which Kuhn et al., rehearsal and retrieval strategy factors may also play a role
2018, observed no asymptote in spacing benefits). in affecting what we will subsequently remember and what
One aspect of the current data may require explanation we will subsequently forget.
through an additional process. Examination of the overall
recognition probabilities for each serial position (Fig. 1b; Supplementary Information The online version contains supplemen-
tary material available at https://d​ oi.o​ rg/1​ 0.3​ 758/s​ 13421-0​ 22-0​ 1293-3.
that graph also includes the words not retrieved in initial
free recall) reveals that there is a sharp drop in recognition Acknowledgements The authors thank Michael Kahana and Lynn
probability for the last list item; indeed, it is the item least Lohnas for helpful guidance, and Noam Isachar for assistance with
likely to be later recognized, despite the very strong recency programming.
effect reported for initial free recall of items in that posi-
tion (~94%; Lohnas & Kahana, 2014; Fig. 3a). Although the Funding This work was supported by the Israel Science Foundation
grant 2497/18 to D.A.L.
differences are numerically small, given the power inher-
ent in the dataset, this anomaly should not be ignored. We
suggest that it results from the fact that initial free recall
began immediately after list presentation. The last list item,
enjoying the standard recency advantage in immediate free
References
recall, may be reported from short-term memory, and there- Baayen, R. H., Davidson, D. J., and Bates, D. M. (2008). Mixed-effects
fore subject to minimal retrieval processing and remembered modeling with crossed random effects for subjects and items.
less well in the long run (Crowder, 1976; McCabe & Madi- Journal of Memory and Language, 59, 390–412. https://​doi.​org/​
gan, 1971; Reitman, 1970; Tan & Ward, 2000). In contrast, 10.​1016/j.​jml.​2007.​12.​005
Benjamin, A. S., & Tullis, J. (2010). What makes distributed practice
primacy and mid-list words that are not held in a short-term effective?. Cognitive Psychology, 61(3), 228–247. https://​doi.​org/​
store require IFR retrieval by strategic recall processes. 10.​1016/j.​cogps​ych.​2010.​05.​004
Those retrieval processes might serve as the type of deep

13
1692 Memory & Cognition (2022) 50:1683–1693

Ben-Yakov, A., & Dudai, Y. (2011). Constructing realistic engrams: Hintzman, D. L. (1969). Recognition time: Effects of recency, fre-
Poststimulus activity of hippocampus and dorsal striatum pre- quency, and the spacing of repetitions. Journal of Experimental
dicts subsequent episodic memory. Journal of Neuroscience, 31, Psychology, 79, 192–194. https://​doi.​org/​10.​1037/​h0026​943
9032–9042. https://​doi.​org/​10.​1523/​jneur​osci.​0702-​11.​2011 Howard, M. W., & Kahana, M. J. (1999). Contextual variability and
Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). serial position effects in free recall. Journal of Experimental
Distributed practice in verbal recall tasks: A review and quantita- Psychology: Learning, Memory, and Cognition, 25, 923–941.
tive synthesis. Psychological Bulletin, 132(3), 354. https://d​ oi.o​ rg/​ https://​doi.​org/​10.​1037/​0278-​7393.​25.4.​923
10.​1037/​0033-​2909.​132.3.​354 Howard, M. W., & Kahana, M. J. (2002). A distributed representation
Chan, J. C., & McDermott, K. B. (2007). The testing effect in recogni- of temporal context. Journal of Mathematical Psychology, 46,
tion memory: A dual process account. Journal of Experimental 269–299. https://​doi.​org/​10.​1006/​jmps.​2001.​1388
Psychology: Learning, Memory, and Cognition, 33(2), 431–437. Innocenti, I., Cappa, S. F., Feurra, M., Giovannelli, F., Santarnecchi,
https://​doi.​org/​10.​1037/​0278-​7393.​33.2.​431 E., Bianco, G., Cincotta, M., & Rossi, S. (2013). TMS interfer-
Cohen, R. L. (1970). Recency effects in long-term recall and recogni- ence with primacy and recency mechanisms reveals bimodal
tion. Journal of Verbal Learning and Verbal Behavior, 9, 672– episodic encoding in the human brain. Journal of Cognitive
678. https://​doi.​org/​10.​1016/​s0022-​5371(70)​80031-7 Neuroscience, 25, 109–116. https://​d oi.​o rg/​1 0.​1 162/​j ocn_a_​
Cousineau, D. (2005). Confidence intervals in within-subject designs: 00304
A simpler solution to Loftus and Masson’s method. Tutorials in Karpicke, J. D., & Roediger, H. L. (2008). The critical importance
Quantitative Methods for Psychology, 1, 42–45. https://​doi.​org/​ of retrieval for learning. Science, 319 (5865), 966–968. https://​
10.​20982/​tqmp.​08.3.​p182 doi.​org/​10.​1126/​scien​ce.​11524​08
Craik, F. I. M. (1970). The fate of primary memory items in free recall. Kloth, N., Lemke, J., Wiendl, H., Meuth, S. G., Duning, T., & Johnen,
Journal of Verbal Learning and Verbal Behavior, 9, 658–664. A. (2020). Serial position effects rapidly distinguish Alzheimer’s
https://​doi.​org/​10.​1016/​s0022-​5371(70)​80042-1 from frontotemporal dementia. Journal of Neurology, 267, 975–
Craik, F. I. M., Gardiner, J. M., & Watkins, M. J. (1970). Further evi- 983. https://​doi.​org/​10.​1007/​s00415-​019-​09662-w
dence for a negative recency effect in free recall. Journal of Ver- Kuhn, J. R., Lohnas, L. J., & Kahana, M. J. (2018). A spacing account
bal Learning and Verbal Behavior, 9, 554–560. https://d​ oi.o​ rg/1​ 0.​ of negative recency in final free recall. Journal of Experimental
1016/​s0022-​5371(70)​80101-3 Psychology: Learning, Memory, and Cognition, 44, 1180–1185.
Crowder, R. G. (1976). Principles of learning and memory. Erlbaum. https://​doi.​org/​10.​1037/​e5024​12013-​799
Darley, C. F., & Murdock, B. B. (1971). Effects of prior free recall Lohnas, L. J., & Kahana, M. J. (2014). Compound cuing in free recall.
testing on final recall and recognition. Journal of Experimental Journal of Experimental Psychology: Learning, Memory, and
Psychology, 91(1), 66–73. https://​doi.​org/​10.​1037/​h0031​836 Cognition, 40, 12–24. https://​doi.​org/​10.​1037/​a0033​698
Davelaar, E. J., Goshen-Gottstein, Y., Ashkenazi, A., Haarmann, Madigan, S. A. (1969). Intraserial repetition and coding processes in
H. J., & Usher, M. (2005). The demise of short-term memory free recall. Journal of Verbal Learning and Verbal Behavior, 8,
revisited: Empirical and computational investigations of recency 828–835. https://​doi.​org/​10.​1016/​s0022-​5371(69)​80050-2
effects. Psychological Review, 112, 3–42. https://​doi.​org/​10.​ McCabe, L., & Madigan, S. (1971). Negative effects of recency in
1037/​0033-​295x.​112.1.3 recall and recognition. Journal of Verbal Learning and Ver-
Delaney, P. F., Verkoeijen, P. P., & Spirgel, A. (2010). Spacing and bal Behavior, 10(3), 307–310. https://​doi.​org/​10.​1016/​S0022-​
testing effects: A deeply critical, lengthy, and at times discursive 5371(71)​80059-2
review of the literature. Psychology of Learning and Motivation, McDermott, K. B. (2021). Practicing retrieval facilitates learning.
53, 63–147. https://​doi.​org/​10.​1016/​S0079-​7421(10)​53003-2 Annual Review of Psychology, 72, 609–633. https://​doi.​org/​10.​
Dudai, Y., Karni, A., & Born, J. (2015). The consolidation and trans- 1146/​annur​ev-​psych-​010419-​051019
formation of memory. Neuron, 88, 20–32. https://​doi.​org/​10.​ Melton, A. W. (1970). The situation with respect to the spacing of rep-
1016/j.​neuron.​2015.​09.​004 etitions and memory. Journal of Verbal Learning and Memory, 9,
Engle, R. W. (1974). Negative recency in delayed recognition. Jour- 596–606. https://​doi.​org/​10.​1016/​s0022-​5371(70)​80107-4
nal of Memory and Language, 13(2), 209–216. https://​doi.​org/​ Morey, R. D. (2008). Confidence intervals from normalized data: A
10.​1016/​S0022-​5371(74)​80045-9 correction to Cousineau (2005). Tutorials in Quantitative Meth-
Engle, R. W., & Durban, E. D. (1977). Effects of modality of pres- ods for Psychology, 4, 61–64. https://d​ oi.o​ rg/1​ 0.2​ 0982/t​ qmp.0​ 4.2.​
entation on delayed recognition. Perceptual and Motor Skills, p061
45, 1203–1210. https://​doi.​org/​10.​2466/​pms.​1977.​45.​3f.​1203 Parkin, A. J., & Russo, R. (1993). On the origin of functional differ-
Feng, K., Zhao, X., Liu, J., Cai, Y., Ye, Z., Chen, C., & Xue, G. ences in recollective experience. Memory, 1, 231–237. https://d​ oi.​
(2019). Spaced learning enhances episodic memory by increas- org/​10.​1080/​09658​21930​82582​35
ing neural pattern similarity across repetitions. The Journal of Radvansky, G. A., & Zacks, J. M. (2017). Event boundaries in memory
Neuroscience, 39, 5351–5360. https://​doi.​org/​10.​1523/​JNEUR​ and cognition. Current Opinion in Behavioral Sciences, 17, 133–
OSCI.​2741-​18.​2019 140. https://​doi.​org/​10.​1016/j.​cobeha.​2017.​08.​006
Glenberg, A. M. (1976). Monotonic and nonmonotonic lag effects in Reitman, W. (1970). What does it take to remember? In D. Norman
paired-associate and recognition memory paradigms. Journal (Ed.), Models of human memory (pp. 469–509). Academic Press.
of Verbal Learning and Verbal Behavior, 15(1), 1–16. https://​ Smith, S. M. (1982). Enhancement of recall using multiple environ-
doi.​org/​10.​1016/​S0022-​5371(76)​90002-5 mental contexts during learning. Memory & Cognition, 10, 405–
Greene, R. L. (1989). Spacing effects in memory: Evidence for a 412. https://​doi.​org/​10.​3758/​bf031​97642
two-process account. Journal of Experimental Psychology: Smith, S. M., & Handy, J. D. (2016). The crutch of context-depend-
Learning, Memory, and Cognition, 15, 371–377. https://​doi.​ ency: Effects of contextual support and constancy on acquisition
org/​10.​1037/​0278-​7393.​15.3.​371 and retention. Memory, 24, 1134–1141. https://​doi.​org/​10.​1080/​
Healey, M. K., Crutchley, P., & Kahana, M. J. (2014). Individual 09658​211.​2015.​10718​52
differences in memory search and their relation to intelligence. Smolen, P., Zhang, Y., & Byrne, J. (2016). The right time to learn:
Journal of Experimental Psychology: General, 143, 1553–1569. Mechanisms and optimization of spaced learning. Nature Reviews
https://​doi.​org/​10.​1037/​a0036​306 Neuroscience, 17, 77–88. https://​doi.​org/​10.​1038/​nrn.​2015.​18

13
Memory & Cognition (2022) 50:1683–1693 1693

Swallow, K. M., Zacks, J. M., & Abrams, R. A. (2009). Event bounda- Xue, G., Mei, L., Chen, C., Lu, Z. L., Poldrack, R., & Dong, Q. (2011).
ries in perception affect memory encoding and updating. Journal Spaced learning enhances subsequent recognition memory by
of Experimental Psychology: General, 138, 236–257. https://​doi.​ reducing neural repetition suppression. Journal of Cognitive
org/​10.​1037/​a0022​160 Neuroscience, 23, 1624–1633. https://​doi.​org/​10.​1162/​jocn.​2010.​
Talmi, D., Grady, C. L., Goshen-Gottstein, Y., & Moscovitch, M. 21532
(2005). Neuroimaging the serial position curve: A test of single- Zhao, X., Wang, C., Liu, Q., Xiao, X., Jiang, T., Chen, C., & Xue,
store versus dual-store models. Psychological Science, 16, 716– G. (2015). Neural mechanisms of the spacing effect in episodic
723. https://​doi.​org/​10.​1111/j.​1467-​9280.​2005.​01601.x memory: A parallel EEG and fMRI study. Cortex, 69, 76–92.
Tan, L., & Ward, G. (2000). A recency-based account of the primacy https://​doi.​org/​10.​1016/j.​cortex.​2015.​04.​002
effect in free recall. Journal of Experimental Psychology: Learn-
ing, Memory, and Cognition, 26, 1589–1626. https://​doi.​org/​10.​ Open practices statement The data and materials for all experiments
1037/​0278-​7393.​26.6.​1589 are available (http://​memory.​psych.​upenn.​edu), and this analysis was
Tibon, R., & Levy, D. A. (2015). Striking a balance: analyzing unbal- not preregistered.
anced event-related potential data. Frontiers in Psychology, 6,
555. https://​doi.​org/​10.​3389/​fpsyg.​2015.​00555
Publisher’s note Springer Nature remains neutral with regard to
Tulving, E. (1983). Elements of episodic memory. Clarendon Press.
jurisdictional claims in published maps and institutional affiliations.
West, B. T. (2009). Analyzing longitudinal data with the linear mixed
model’s procedure in SPSS. Evaluation & the Health Professions,
32, 207–228. https://​doi.​org/​10.​1177/​01632​78709​338554

13