ArticlePDF Available

Abstract and Figures

Kornell and Bjork (Psychological Science 19:585-592, 2008) found that interleaving exemplars of different categories enhanced inductive learning of the concepts based on those exemplars. They hypothesized that the benefit of mixing exemplars from different categories is that doing so highlights differences between the categories. Kang and Pashler (Applied Cognitive Psychology 26:97-103, 2012) obtained results consistent with this discriminative-contrast hypothesis: Interleaving enhanced inductive learning, but temporal spacing, which does not highlight category differences, did not. We further tested the discriminative-contrast hypothesis by examining the effects of interleaving and spacing, as well as their combined effects. In three experiments, using photographs of butterflies and birds as the stimuli, temporal spacing was harmful when it interrupted the juxtaposition of interleaved categories, even when total spacing was held constant, supporting the discriminative-contrast hypothesis. Temporal spacing also had value, however, when it did not interrupt discrimination processing.
Content may be subject to copyright.
Why interleaving enhances inductive learning:
The roles of discrimination and retrieval
Monica S. Birnbaum &Nate Kornell &
Elizabeth Ligon Bjork &Robert A. Bjork
#Psychonomic Society, Inc. 2012
Abstract Kornell and Bjork (Psychological Science 19:585
592, 2008) found that interleaving exemplars of different
categories enhanced inductive learning of the concepts based
on those exemplars. They hypothesized that the benefit of
mixing exemplars from different categories is that doing so
highlights differences between the categories. Kang and Pash-
ler (Applied Cognitive Psychology 26:97103, 2012) obtained
results consistent with this discriminative-contrast hypothesis:
Interleaving enhanced inductive learning, but temporal spac-
ing, which does not highlight category differences, did not. We
further tested the discriminative-contrast hypothesis by exam-
ining the effects of interleaving and spacing, as well as their
combined effects. In three experiments, using photographs of
butterflies and birds as the stimuli, temporal spacing was
harmful when it interrupted the juxtaposition of interleaved
categories, even when total spacing was held constant,
supporting the discriminative-contrast hypothesis. Temporal
spacing also had value, however, when it did not interrupt
discrimination processing.
Keywords Categorization .Induction .Interleaving .
Spacing .Metacognition
People accumulate a great deal of knowledge via inductive
learning. Children, for example, learn concepts such as boat
or fruit by being exposed to exemplars of those categories
and inducing the commonalities that define the concepts.
Later in life, we might learn to distinguish between dif-
ferent species of butterflies or birds, as in the present
research. Such inductive learning is critical in making
sense of events, objects, and actionsand, more generally,
in structuring and understanding our world. In the present
research, we examined how exemplars of to-be-learned cate-
gories should be sequenced and spaced in order to optimize
inductive learning.
Kornell and Bjork (2008) investigated the effect of study
schedules on inductive learningspecifically, learning
artistspainting styles from exemplars of their paintings.
Images of six paintings by each of 12 artists were presented
for study, with the artists name displayed below each paint-
ing. The paintings by half of the artists were blocked (i.e., all
six paintings by a given artist were shown consecutively),
whereas the paintings by the other six artists were inter-
leaved (i.e., mixed together). After the learning phase, par-
ticipants were shown new paintings by each of the 12 artists
and were asked to identify which artist had painted each
new painting. Kornell and Bjork found that interleaving
artistspaintings led to better performance on this inductive
task than did blockingeven though participants consis-
tently believed that blocking, rather than interleaving, had
been more helpful for learning the artistsstyles.
Kornell and Bjorks(2008) findings were replicated by
Kornell, Castel, Eich, and Bjork (2010) with older adults as
participants. Furthermore, Zulkiply, McLean, Burt, and Bath
(2012) found similar results when their participants read
case studies exemplifying different psychological disorders
in an inductive-learning experiment, and Vlach, Sandhofer,
and Kornell (2008) found similar results when three-year-
old children learned the names of novel objects on the basis
of induction. Additionally, Kornell and Bjorks results have
been replicated by Kang and Pashler (2012), Zulkiply and
Burt (in press), and Wahlheim, Dunlosky, and Jacoby
(2011). All of these findings seem to fit within an extensive
literature on the spacing effectthat is, the finding that
M. S. Birnbaum (*):E. L. Bjork :R. A. Bjork
Department of Psychology, University of California, Los Angeles,
1285 Franz Hall, Box 951563, Los Angeles, CA 90095-1563,
N. Kornell
Department of Psychology, Williams College,
Williamstown, MA 01267, USA
Mem Cogn
DOI 10.3758/s13421-012-0272-7
items studied once and restudied after a delay are recalled
better in the long term than are items studied and restudied
in quick succession (for reviews, see Cepeda, Pashler, Vul,
Wixted, & Rohrer, 2006; Dempster, 1988; Glenberg, 1979).
In all of the recent studies demonstrating the benefits of
spacing for inductive learning (with the exception of Vlach
et al., 2008, who did not have an interleaved condition in
addition to their spaced condition), however, interleaving
mixing exemplars from different categories togetherwas
what enhanced learning, rather than temporal spacing
per sea point to which we will return shortly.
A major goal of the present research was to discover
why interleaving appears to enhance inductive learning.
Intuitively, it might seem that studying a single category
in a block would be beneficial, as learners would notice
similarities within the category. Consistent with this idea,
early studies showed that mixing exemplars from different
categories resulted in poorer learning than did grouping
exemplars of the same category (Kurtz & Hovland, 1956;
Whitman & Garner, 1963). More recently, Goldstone
(1996) found better performance when categories alternat-
ed on 25 % of the trials than when they alternated on
75 % of the trialsthat is, less-frequent alternation appeared
to produce more learning. (We will discuss these findings and
why they may differ from the more recent results in the
General Discussion.)
Discriminative-contrast hypothesis
If blocked studying facilitates noticing similarities among
intracategory exemplars, interleaved studying might facili-
tate noticing the differences that separate one category from
another. In other words, perhaps interleaving is beneficial
because it juxtaposes different categories, which then high-
lights differences across the categories and supports dis-
crimination learningan idea that we refer to as the
discriminative-contrast hypothesis. As Goldstone (1996)
pointed out, frequent alternation of categories has the ad-
vantage of highlighting features that serve to distinguish
categories. Conversely, infrequent alternation of categories
has the advantage of highlighting information that remains
constant across the members within a category(p. 615). If
the ultimate goal of category learning is to be able to classify
new examples into the appropriate categories, then knowing
what distinguishes categories is crucial.
In a recent study, Kang and Pashler (2012) investigated
the degree to which noticing differences is the driver
behind the benefit of interleaving. Using three of the 12
artists employed by Kornell and Bjork (2008), they repli-
cated the blocked and interleaved conditions. In their first
experiment, they also included a condition in which the
blocked items were temporally spaced (using irrelevant
cartoon drawings).
They found a benefit of interleaving
over blocking (replicating Kornell & Bjork, 2008), but no
benefit of temporal spacing over blocking. On the basis of
this finding, they concluded that the value of interleaving
lies in juxtaposing different categories, a process that they
referred to as discriminative contrast. Consistent with this
conclusion, they also found that simultaneous presenta-
tions of multiple paintings by the same artist did not
benefit learning (Exp. 1), whereas simultaneous presenta-
tions of multiple paintings by different artists were bene-
ficial (Exp. 2)again, presumably, because they promoted
discriminative contrast. Wahlheim et al. (2011) also found
benefits of simultaneous presentation of different catego-
ries for inductive learning.
Kang and Pashlers(2012) findings are consistent with
the view that the value of interleaving lies in promoting
discrimination between categories. We believe, however,
that their findings do not seal the case. Their argument rests
on two findings. The first has to do with the effects of
simultaneous presentation. These results clearly supported
the value of discriminative contrast, but presenting one item
at a time is different from presenting all items at once
doing the latter places greater demands on working memory,
and the discrimination processes that appear to be at work
during a simultaneous presentation may not happen, or may
not happen as effectively, during single presentations. Their
second finding was a lack of benefit of temporal spacing,
but that finding also did not pin down the cause of the
interleaving benefit. That is, although a spacing effect did
not occur, had it occurred, such a spacing effect would not
have ruled out a role for discriminative contrast when exam-
ples are interleaved.
The present experiments
The crucial question, in our view, has to do with comparing
different versions of the interleaved condition, because the
interleaved condition is where the discriminative-contrast hy-
pothesis makes strong predictions. Specifically, we tested the
hypothesis thatin an interleaved conditionpreventing
items from being juxtaposed with one another would hurt
inductive learning. We tested this prediction, in Experiment
A note on terminology: We refer to mixing the presentation of similar
types of trials together as interleaving and its opposite as blocking. We
refer to spacing study trials apart using unrelated activities as spacing
and its opposite as contiguous presentation. For example, presenting
examples of one type of algebra problem mixed with other types of
algebra problems would be interleaving; presenting different examples
of one type of algebra problem mixed with historical biographies
would be spacing, because the activity separating the algebra problems
is the study of historical biographiesnot the study of examples of
another type of algebra problem.
Mem Cogn
1, by presenting interleaved exemplars with, or without, un-
related trivia questions inserted between successive exem-
plars. The basic logic was that if inserting trivia questions
did not disrupt inductive learning, it would then be difficult to
conclude that discrimination processes are responsible for the
benefit of interleaving.
Our study was also motivated by practical questions.
Creating a condition that was both spaced and interleaved
allowed us to investigate an optimal combination of both
manipulations. Zulkiply and Burt (in press) tested an inter-
leaved schedule in which each exemplar was temporally
spaced out with 30 s of unrelated filler task. They found
that although interleaving led to a benefit in inductive learn-
ing, as compared to blocking, temporal spacing did not
affect the benefit of interleaving. In our Experiment 2,we
tested a similar condition, but used species of butterflies as
the to-be-learned categories and employed 10-s intervening
filler tasks. In Experiments 2and 3, we examined the
possible interaction of spacing and interleaving.
Experiment 1
In Experiment 1, we presented four photos each of eight
species of birds in an interleaved order during the learning
phase, using the following three study conditions. In the
contiguous condition, no trivia questions were interpolated
between the interleaved presentations of exemplars. In the
alternating-trivia condition, the presentation order of the
bird exemplars was the same as in the contiguous condition,
but the presentations of successive exemplars were separat-
ed by the insertion of a trivia question. An unrelated task
was inserted rather than having no filler task during the
time delay so as to prevent rehearsal of the previously
viewed image. In the grouped-trivia condition, single
exemplars from each of the eight species of birds were
presented in randomly ordered groups, with no trivia
questions presented inside of a group, but successive
groups of exemplars were separated by a series of eight
trivia questions.
Importantly, between successive exemplars of a given
species, exactly the same events transpired in the grouped-
trivia and alternating-trivia conditions; that is, an average of
eight trivia questions and seven photos intervened between
successive exemplars of a given species in each condition.
Thus, the total spacing was held constant in these two
conditions, but the alternating-trivia condition interrupted
comparison processes (by placing trivia questions between
the photos), whereas the grouped-trivia condition did not.
The discriminative-contrast hypothesis predicts that the
alternating-trivia condition, which disrupted discrimination
processes, should impair performance as compared to the
grouped-trivia condition.
Participants A group of 102 participants (52 female, 50
male) were recruited to participate via Amazons Mechan-
ical Turk, a website that allows people to sign up to
complete small tasks for pay. Participants were paid 80
cents for participating, which took an average of about
12 min, and the participants averaged 31.2 years of age
(range: 1860). Of the participants, 59 were from the
United States, 23 were from India, and the remaining 20
came from 17 other countries.
The numbers of participants in the contiguous, grouped-
trivia, and alternating-trivia conditions were, respectively,
35, 25, and 42, with the differing numbers of participants
arising through random assignment.
Materials The materials consisted of five different photo-
graphs of each of eight different species of birds (see Wahlheim
et al., 2011). As is illustrated in Fig. 1, each bird was shown
against a brown background, with its species name shown
below the photo during learning. The names of the species
were accurate but simplified; for example, Cave Swallow was
changed to Swallow (see Fig. 1). The materials also included
trivia questions (e.g., What is the name of the spear-like object
that is thrown during a track meet?), which came from Nelson
and Narens (1980).
Procedure The study was conducted online. After reading
the instructions, participants were shown 32 photographs of
birds on their computer screens, one at a time, for 4 s each.
The eight species were interleaved, such that each group of
eight trials included one photo from each species, arranged
in a random order.
In the contiguous condition, the photos were presented
contiguously. In the other two conditions, 32 trivia questions
were also presented for 8 s each, accompanied by the fol-
lowing instruction: Try to think of the answer. We will ask
you to recall it later.
In the alternating-trivia condition, a trivia question was
presented before every photo. This task was intended to
keep participants occupied with a task not related to bird
photos. In the grouped-trivia condition, a series of eight
trivia questions was presented followed by a series of eight
photos, followed by a different set of eight trivia questions,
and so forth. With the trials ordered in this fashion, all three
conditions ended when the last photo was presented.
After the last photo was presented, participants were
asked to play the computer game Tetris for 3 min as a
distractor task and were then given an inductive-learning
test. During the test, a previously unpresented exemplar of
each species was presented, along with the names of all
eight species, which were shown below the photos. Partic-
ipants were asked to select the name of the species
Mem Cogn
represented by the presented exemplar and to type it on the
computer keyboard. Participants had unlimited time to re-
spond, and no feedback was provided.
Correct performance on the final test, which is illustrated in
Fig. 2, was significantly affected by participantslearning
conditions, F(2, 99) 08.53, p<.001, η2
p¼:15. Further-
more, a TukeyKramer post-hoc test showed that accuracy
was significantly lower in the alternating-trivia condition
than in the other two conditions, for which performance
did not differ significantly.
Experiment 1went beyond previous findings by holding
spacing constant between the grouped-trivia and alternating-
trivia conditions and directly manipulating participantsabil-
ity to engage in discrimination processes in an interleaved
presentation condition. The results were consistent with prior
studies (Kang & Pashler, 2012; Wahlheim et al., 2011)in
supporting the discriminative-contrast hypothesis: Interrupt-
ing discrimination processing, which was accomplished in the
present experiment by inserting trivia questions between in-
terleaved exemplars, impaired inductive learning, whereas
Contiguous Grouped-
Proportion Correct
Fig. 2 Proportions correct on the final test in Experiment 1. Error bars
represent one standard error of the mean
Fig. 1 Photographs used to depict exemplars of the eight bird species
used in Experiment 1. The top row (left to right) depicts Fieldfare,
Finch, Grosbeak, and Oriole. The middle row depicts Swallow, Tit-
mouse, Veery, and Warbler. Although all species were presented in an
interleaved order, the bottom row depicts four Fieldfare exemplars.
During the learning phase, the name of the exemplars species was
shown below each exemplar
Mem Cogn
inserting groups of triviaquestions between successive groups
of exemplarsand, thus, separating exemplars from within a
given species temporallydid not impair inductive learning
significantly. One could argue that introducing another task
increased interference, and thus a true comparison between
the contiguous and grouped-trivia conditions could not be
made. The grouped-trivia and alternating-trivia conditions,
however, had comparable levels of interference, and a com-
parison between these two conditions reveals the importance
of discriminative contrast in inductive learning. Additionally,
interference and task were held constant across all three con-
ditions in Experiment 3.
Experiment 2
In Experiment 2, we examined how interleaving and spac-
ing might interact in producing their effects on inductive
learning. The results of Experiment 1suggested that spac-
ing, as implemented in the grouped-trivia condition, might
have impaired performance slightly relative to the contigu-
ous condition, but Experiment 1did not include a noninter-
leaved condition of the type that would permit an assessment
of the value of interleaving, relative to blocking, as a function
of temporal spacing. Experiment 2was designed to permit
such an assessment.
The participants in Experiment 2studied four exemplars
of each of 16 species of butterflies in the context of a 2 × 2
mixed design. The species were randomly assigned to either
the interleaved or blocked condition. In addition, for half of
the participants, the butterfly exemplars were presented
contiguously, replicating previous research (e.g., Kornell &
Bjork, 2008); for the other half of the participants, trivia
questions were inserted during the 10-s intervals between
the presentations of successive butterfly exemplars, which
created a spaced condition.
We predicted an interaction between spacing and inter-
leaving. When exemplars are presented contiguously, the
discriminative-contrast hypothesis, as well as previous re-
search, predicts that interleaving should be beneficial as
compared to blocking. When, however, exemplars are
spaced apart with trivia questions, the expected impairment
of discrimination processing as a consequence of blocking
might not be observed, because such processing has already
been neutralized by the presence of the interpolated trivia
questions. To the extent that trivia questions prevent dis-
crimination processing across the board, the discriminative-
contrast hypothesis does not predict a benefit of interleaving
over blocking. (Such a finding would be consistent with
Kang & Pashlers, 2012, finding that spacing did not en-
hance blocked learning, but Kang & Pashler did not com-
pare interleaved vs. blocked study when both conditions
were spaced.)
We were also interested in participantsmetacognitive
judgments regarding which study schedule, interleaved or
blocked, was more helpful to their learning. Kornell and
Bjorks(2008) participants rated blocking as being more
effective than interleaving, despite having just completed a
test on which they had performed better for categories
learned under interleaved rather than under blocked condi-
tions. We expected that the present participants would make
the same judgment error for both the contiguous and spaced
conditions, because blocking appears to create a greater
sense of fluency of induction than does interleaving.
Participants A group of 114 undergraduates from the Uni-
versity of California, Los Angeles, served as participants for
partial course credit.
Materials The materials were 80 photographs depicting 16
species of butterflies, examples of which are illustrated in
Fig. 3. The participants studied four exemplars of each
species and were tested on a fifth exemplar. The labels
presented to participants for each species were based on
the actual name of each species, but some names were
shortened, made into one word, or changed entirely if the
real name described physical characteristics of the butterfly.
The names employed were Admiral, American, Baltimore,
Cooper, Eastern Tiger, Hairstreak, Harvester, Mark, Painted
Lady, Pine Elfin, Pipevine, Sprite, Tipper, Tree Satyr, Viceroy,
and Wood Nymph.
In the spaced condition, a different individual trivia ques-
tion, drawn from the same pool that had been used in
Experiment 1, was presented between the successive pre-
sentations of butterfly photographs.
Design Half of the 16 species were presented in a blocked
fashion (i.e., the four exemplars from each species were
presented with no exemplars from other species in between),
and the other half were presented in an interleaved fashion
(i.e., the presentations of the four exemplars from each
species were mixed with presentations of exemplars from
other species). Additionally, equal numbers of participants
were randomly assigned to either a temporally spaced (n0
57) or a contiguous (n057) presentation mode, in which no
intervening filler tasks were presented between the exem-
plars. Thus, the experiment employed a 2 × 2 mixed facto-
rial design, with presentation order (blocked vs. interleaved)
manipulated within participants and presentation mode
(spaced vs. contiguous) manipulated between participants.
Procedure During the learning phase, 64 different photo-
graphs of butterflies appeared, according to the following
organizational scheme: I I B B I I B B I I B B I I B B, where
Mem Cogn
I and B represent sequences of four photos that were pre-
sented in either an interleaved (I) or a blocked (B) fashion.
Within each I I sequence, one exemplar from each of eight
species was presented, meaning that each exemplar of a
given species was interleaved among exemplars of the other
species. The eight photos were ordered randomly within
each I I grouping. In each B B sequence, all four exemplars
of a single species were presented consecutively, followed
by all four exemplars of another species. Thus, in the inter-
leaved condition, the four exemplars belonging to a given
species were distributed across all four I I sequences, where-
as in the blocked condition, all four exemplars of a given
species were presented within one B sequence, without
exemplars of other species intervening.
Each photo was presented on a computer screen for 4 s
with the species name shown below the photo. For partic-
ipants assigned to the spaced presentation mode, a 10-s
interval occurred between successive presentations of but-
terfly photos. During this interval, a trivia question appeared
on the screen, along with the instruction to try to retrieve the
answer and then remember it for a later memory test. For
participants assigned to the contiguous-presentation mode,
all photos immediately succeeded one another, with no
intervening trivia questions appearing on the screen. Thus,
whereas all participants studied half of the butterfly species
in an interleaved fashion and half in a blocked fashion, for
half of the participants a trivia question appeared between
successive photos of butterflies, and for the other half of the
participants, no trivia questions were presented.
When the learning phase ended, participants were imme-
diately tested, as follows. A new exemplar of one of the 16
species was presented with the names of all 16 species
shown below the photo. Participants were asked to select
the species represented by the presented exemplar and enter
it on the computer keyboard. All 16 species were tested in
this manner in a random order. A unique random order was
used for each participant. The participants had unlimited
time to respond, and no feedback was provided during the
testing phase.
After the testing phase, participants read a brief descrip-
tion of the experiment that they had just completed, which
defined the terms blocked and interleaved. They were then
asked to answer the following question, which appeared on
the screen: Which do you think helped you learn more,
blocking or interleaving?
Results and discussion
Induction performance Performance on the final induction
test, which is illustrated in Fig. 4, was analyzed using a 2
(blocked vs. interleaved) × 2 (contiguous vs. spaced) mixed
Fig. 3 Photographs used to depict exemplars of butterfly species. The
top row depicts four exemplars of the Viceroy species. The bottom row
depicts exemplars from four different species (Sprite, Cooper, Balti-
more, and Tipper). The first row corresponds to a typical sequence
during blocked study; the bottom row corresponds to a typical se-
quence during interleaved study. During the learning phase, the name
of the exemplars species was shown below each exemplar
Contiguous Spaced
Proportion Correct
Fig. 4 Proportions correct on the final test in Experiment 2. Error bars
represent one standard error of the mean
Mem Cogn
analysis of variance (ANOVA). As is indicated in Fig. 4,a
significant overall advantage of interleaving over blocking
was obtained, F(1, 112) 012.34, p<.001, η2
Importantly, as is further indicated in Fig. 4, a significant
interaction between presentation order (interleaved vs.
blocked) and presentation mode (contiguous vs. spaced) was
also obtained, F(1, 112) 010.909, p<.01, η2
p¼:089. As
predicted, performance in the interleaving condition was su-
perior to that in the blocking condition when exemplars were
presented contiguously (the means were .36 and .21, respec-
tively). When trivia questions were used to space out the
presentation of successive exemplars however, performance
in the interleaving condition was not significantly different
from that in the blocking condition (the means were .29 and
.28, respectively). This pattern of results is consistent with the
discriminative-contrast hypothesis.
A planned comparison of performance in the spaced versus
contiguous conditions, limited to only the blocked condition,
showed that spacing was beneficial for this type of presenta-
tion, t(112) 02.018, p<.05, d0.379, whereas Kang and
Pashler (2012) had found no significant effect of spacing
during blocked learning. Why the findings from these two
studies diverged in this way is unclear, but the difference
emphasizes the importance of Experiment 1in providing
direct support for the discriminative-contrast hypothesis.
Another planned comparison showed that in the absence of
spacing, the induction learning produced by interleaving was
superior to that produced by blocking, t(56) 04.794, p<.0001,
d00.678, replicating prior findings (e.g., Kang & Pashler,
2012;Kornell&Bjork,2008). Combined, these two planned
comparisons demonstrated that, as compared to a baseline
condition in which presentations were not interleaved or
spaced, adding spacing or interleaving enhanced learning.
Adding both interleaving and spacing, however, did not lead
to the best performanceinterleaving alone didso clearly,
the benefits of spacing and interleaving were subadditive.
Zulkiply and Burt (in press) also found a benefit of a contigu-
ous interleaved study schedule as compared to a spaced
blocked schedule. This provides further support for the
accounts that discrimination processes are more critical to
inductive learning than is pure temporal spacing.
One potential criticism of Experiment 2is that spacing
was confounded with retention interval: For blocked cate-
gories, in particular, the interval between the last study trial
and the final induction test was longer in the spaced than in
the contiguous condition (this was not the case in the inter-
leaved condition, in which all species were studied at both
the beginning and end of the session). To examine the
potential effects of this confounding, we analyzed test accu-
racy as a function of study order by computing accuracy
scores for the first two blocked species (which always
occurred consecutively) and the last two blocked species
(which also always occurred consecutively). Study block
did not have a significant effect on performance, nor did it
interact with spacing. Thus, order effects did not seem to
have an impact on our results. (Also, Exp. 1had been
designed to avoid such a confounding problem.)
Temporal spacing did not enhance learning for inter-
leaved species. At the same time, spacing did significantly
enhance learning in the blocked condition. We interpreted
this pattern as indicating that two separable benefits of
distributed learning occurred. In the interleaving condition,
spacing did not provide a benefit in addition to interleaving
because spacing prevented the beneficial discriminative
contrast processing that otherwise would have occurred. In
contrast, in the blocked condition, spacing was valuable for
the reasons that have been used to explain spacing effects in
noninductive learningin particular, the value of study-
phase retrieval (e.g., Thios & DAgostino, 1976). That is,
delay allows time for forgetting, making retrieval of previ-
ous exemplars from memory more difficult, but thereby
enhancing learning when such retrievals are successful.
Our interpretation of these opposing findings might seem
inconsistent, because forgetting is viewed as bad in the
interleaved condition, but good in the blocked condition.
There are two reasons to believe that these claims are
actually consistent with each other. First, forgetting was
not catastrophic in the blocked condition (where we believe
that it was beneficial) because blocked items came from the
same category, and thus, similar items could serve as
reminders of each other. Seeing a Cooper, for example,
could serve as a reminder of what the previous Cooper had
looked like. Seeing a Tipper, however, would not bring to
mind what the previous Cooper looked like. Therefore, it
would be easier to recall the previous butterfly photograph
(after a trivia question has been presented) in the blocked
condition than in the interleaved condition. Owing to such
reminders, retrieval between trials should be better in the
blocked than in the interleaved condition, so some forgetting
would likely be valuable in the former condition.
A second reason why the two claims are consistent is that
when the same information is presented twice (as with
repetition learning), spacing enhances learning, but no spac-
ing effect occurs when two separate pieces of information
are presented. Two items from the same category (e.g., two
Tippers) contain similar information and share the same
category label, and thus, spacing would be expected to
enhance learning in the blocked condition. In contrast, two
items from different categories contain far less similar in-
formation (and have different labels); thus, there would be
little reason to expect a benefit of temporal spacing in the
interleaved condition.
Related to such considerations, the subadditivity of spac-
ing and interleaving that we observed in the present exper-
iment is analogous, perhaps, to the subadditivity of spacing
Mem Cogn
and variation found by Appleton-Knapp, Bjork, and Wickens
(2005) in participantsmemory for advertisements of hypo-
thetical products. In that research, recall was facilitated when
the formats of advertisements for the same product were
varied, but only when the advertisements were shown close
together. Spacing also facilitated recall, but only when the
exact same advertisement was repeated. When variation and
spacing were combined, therefore, the effects on later recall
were subadditive. Appleton-Knapp et al. reasoned that study-
phase retrieval processes are crucial in supporting long-term
retention: When spacing and variation were combined, the
participants were no longer able to retrieve the earlier version
of the advertisement. The present induction task differed in
several ways from Appleton-Knapp et al.srecalltask,buta
similar dynamic may have been at work in the present exper-
iment: When spacing and interleaving were combined, the
participants became unable, when shown an exemplar of a
given species, to remember the prior exemplars of that species.
As a result, they found it difficult to recognize and remember
commonalities within a given species, with their performance
suffering accordingly.
Metacognitive judgments After completing the final test,
participants were asked whether they believed that blocking
or interleaving had been more effective for their learning of
the different species. In the contiguous condition, 65 % (37/
57) of the participants favored blocking. A binomial test using
an approximation of the normal distribution showed this ratio
to be significantly different from chance (i.e., from 50 %), z0
2.12, p<.05. The ratings were almost identical in the spaced
condition, in which 67 % (38/57) of the participants rated
blocking as more effective, z02.38, p<.05. Thus, although
blocking was counterproductive in the contiguous condition
and had almost no overall effect in the spaced condition,
participants in both groups believed that blocking the exem-
plar presentations had helped them learn more effectively than
did interleaving the exemplar presentations. This finding rep-
licates previous research demonstrating that the conditions
that produce the most fluent processing during learning tend
to result in the highest judgments of learning, even when the
participantsactual performance shows quite the opposite
pattern (e.g., Kornell & Bjork, 2008; Kornell et al., 2010;
Simon & Bjork, 2001; Wahlheim et al., 2011; Zechmeister
& Shaughnessy, 1980).
Experiment 3
In Experiment 2, the optimal way to study was not to
combine spacing and interleaving, because spacing seemed
to interrupt discrimination processing. This finding has a
practical message: It is crucial, in inductive learning, to
juxtapose examples of different categories. Assuming,
therefore, that interleaving is the superior condition for
inductive learning, a question remains unanswered: When
all items are contiguously interleaved, is more spacing better
than less? In Experiment 1, performance was not signifi-
cantly different between the grouped-trivia condition and
the contiguous (baseline) condition, suggesting that the
amount of temporal spacing may not matter, provided that
items are contiguously interleaved. In Experiment 1, how-
ever, the temporal spacing between exemplars of the same
category differed between the contiguous and grouped-trivia
conditions. Thus, in Experiment 3, we investigated the
effect of spacing on interleaved inductive learning while
holding constant total time spent studying.
The participants in Experiment 3studied pictures of butter-
flies, and the exemplars of all species were presented in an
interleaved manner. Instead of using trivia questions to manip-
ulate spacing, however, we varied the lag between exemplars
of a given category. In the small- and large-spacing conditions,
respectively, the lag was filled with three or 15 exemplars of
butterflies from other species. This manipulation allowed us to
examine the effects of varying temporal spacing while holding
juxtaposition, time on task, and time until test constant across
all conditions. We did not expect to find differences in perfor-
mance, on the basis of the discriminative-contrast hypothesis;
our question was whether temporal spacing might enhance
learning even when all of the to-be-learned categories were
presented in an interleaved manner.
Participants A group of 53 undergraduates from the Uni-
versity of California, Los Angeles, served as participants for
partial course credit.
Materials The materials used in this experiment were iden-
tical to those used in Experiment 2.
Procedure The participants studied four exemplars from each
of 16 species of butterflies for 4 s each, with the label below
each photo, and they were then immediately tested with new
exemplars, using the procedure from Experiment 2.Inthe
small-spacing condition (n026), participants studied photos
of a given species separated by an average of three other trials,
whereas in the large-spacing condition (n027), successive
photos of a given species were separated by an average of 15
other trials. Within each block of four exemplars, the order of
species was randomized, with the constraint that no two
exemplars from the same species were presented consecutive-
ly between blocks. The numbers of between-category juxta-
positions were the same for both spacing conditions. In the
large-spacing condition, exemplars of a given species were
juxtaposed next to exemplars of only three other species, in
Mem Cogn
order to match the number of between-category juxtapositions
that occurred in the small-spacing condition.
As is shown in Fig. 5, the participants in the large-spacing
condition performed significantly better (M0.42, SD 0
.198) than those in the small-spacing condition (M0.29,
SD 0.125), t(51) 02.74, p0.008, d00.752.
Thus, the results of Experiment 3indicate that larger spac-
ing can lead to better inductive learning than does smaller
spacing. Furthermore, it is important to note that this benefit of
spacing cannot be attributed to greater discrimination process-
ing, because the degrees of juxtaposition did not vary across
conditions. Instead, this pattern of results is consistent with the
study-phase retrieval account of the spacing effect.
On the basis of the Experiment 1and 2results, we attributed
the value of interleaving to enhanced discrimination process-
ing, but the value may actually come from multiple sources. It
is important to remember that, as compared to blocking, inter-
leaving increases the amount of spacing between exemplars,
and, as is demonstrated by the results of Experiment 3,
increases in spacing may play some role in the value of
interleaving for inductive learning. We do acknowledge that
there may be an upper limit to how much spacing can occur,
even with contiguous interleaved presentation, before perfor-
mance suffers. This effect has not yet been tested, and it should
be. Our present findings do, however, indicate that temporal
spacing can provide a unique benefit to inductive learning,
separate from the benefits of temporal juxtaposition.
General discussion
When people learn natural categories via induction, it is
important to learn both the similarities among exemplars
within a category and the differences between categories.
The results obtained in the present research suggest that, to a
great extent, the reason why interleaving enhances inductive
learning is that the juxtaposing of exemplars of different
categories that interleaving affords promotes the learning of
differences. Temporal spacing is a natural consequence of
interleaving, but the present results suggest that spacing can
have both benefits and costs in inductive learning.
In Experiment 1, presenting trivia questions between
photographs of birds impaired learning, but only when doing
so prevented the juxtaposition of exemplars from different
categories. This finding thus supported the discriminative-
contrast hypothesis (Kang & Pashler, 2012; Kornell & Bjork,
2008). Spacing provided no benefit when it interrupted jux-
taposition between categories in Experiment 2, although
spacing was beneficial in the blocked condition of Experi-
ment 2when it did not interrupt juxtapositioncontrary to
the prior findings by Kang and Pashler. Thus, we further
investigated the value of spacing in Experiment 3, obtaining
results that showed that, when the presentation conditions for
inductive learning were already interleaved, more spacing
was superior to less spacing.
The present findings have practical implications, but first we
turn to a consideration of their theoretical implications. First,
the present findings lend strong support to the discriminative-
contrast hypothesis, suggesting that interleaving is valuable
in inductive learning because it allows people to grasp the
differences between categories. The downside of temporal
spacing appears to be that it can prevent between-category
differentiation, as it did when exemplars were interleaved in
Experiment 1.
Spacing also enhanced learning, however, when exem-
plars were blocked in Experiment 2and when they were
interleaved in Experiment 3, suggesting that multiple mech-
anisms may contribute to the value of distributed practice
(spacing and interleaving) in inductive learning. It seems
that there is value in making it difficult to retrieve a prior
instance of the same category, as long as retrieval does not
become impossible.
In the conceptual framework that we have adopted, block-
ing facilitates processing of similarities within a category,
whereas interleaving facilitates processing of differences be-
tween categories. Categories can be defined by both similar-
ities and differences, and these two dimensions are not
completely independent (see Goldstone, 1996,foradiscus-
sion of the degree to which these concepts are isolated vs.
interrelated). Zulkiply and Burt (in press) found that blocking
was more effective for learning highly discriminable catego-
ries, whereas interleaving was critical for inducing learning of
low-discriminability categories. This set of findings can be
seen as supporting the hypothesis that it is important to pro-
mote the processing of similarities within a category as well as
of differences between categories.
Small Spacin
e Spacin
Proportion Correct
Fig. 5 Proportions correct on the final test in Experiment 3. Error bars
represent one standard error of the mean
Mem Cogn
In this framework, the way that learning is assessed
may be important. In the present experiments, the final
test required participants to indicate the appropriate spe-
cies for each of a set of new exemplars. Interleaving may
have been effective, in part, because the test required
discrimination, and interleaving promotes discrimination.
Another test, one that emphasized discrimination less and
similarity moresuch as an inference rather than a clas-
sification test (see Chin-Parker & Ross, 2002)might
have produced an advantage of blocking, or at least a
smaller advantage of interleaving, even if the training
conditions had been the same.
Although the present pattern of results seems inconsistent
with prior studies of inductive learning that have demon-
strated a benefit of blocked practice (Goldstone, 1996; Kurtz
& Hovland, 1956; Whitman & Garner, 1963), the concep-
tual framework just described may offer an explanation of
these apparently divergent results. In the present research,
similarities within butterfly species (as illustrated in Fig. 3)
may have been relatively easy to process, even during
interleaved practice. If so, any advantage of blocking for
noting similarities may have been unneededonly making
an already easy process easierwhereas interleaving helped
with the more difficult process of discriminating among
species. In contrast, an informal analysis of the previous
experiments that showed a benefit for blocking indicates
that the materials and procedures employed made identify-
ing similarities within categories more difficult, which
according to the present frameworkshould result in block-
ing being more advantageous. This speculation is tentative,
however, because the present research differed from that
prior work in other ways. Among other differences, the
previous studies had used artificial categories, more practice
trials, and fewer categories, and they did not involve a final
test (i.e., performance was measured during learning).
Practical implications
The foremost practical implication of the present findings is
that successful inductive learning entails the juxtaposing of
exemplars from different categories so as to promote dis-
criminative contrast. This implication may extend well be-
yond visual categories of the type employed in the present
studies. Taylor and Rohrer (2010), for example, asked stu-
dents to learn four kinds of mathematics problems using a
task that could be categorized as requiring inductive learn-
ing, because participants never saw the same problem twice
during the learning phase. Taylor and Rohrer found that
interleaving enhanced learningas measured by the ability
to solve new examples of each type of problem on a delayed
testeven when the amount of spacing had been the same
in the interleaved and blocked conditions. It could be argued
that the main challenge for students in the Taylor and Rohrer
experiment was to learn to discriminate among the cate-
gories of problems so that later, on a test, they could
recognize a category of problem and retrieve the correct
solution procedure. Blocked practice may have, as the
authors suggested, led to the studentsfailure to discrim-
inate between different kinds of problems(p. 846). On
the basis of Taylor and Rohrers findings, it seems likely
that a variety of learning tasks in the classroom might
profit from interleavedrather than blockedschedules
of practice or study.
The present results also suggest that temporal spacing
can present a danger to inductive learning. Spacing and
interleaving have both been described as desirable diffi-
culties”—that is, manipulations that create challenges and
difficulties for the learner, but enhance long-term retention
and transfer (Bjork, 1994). As in the case of the results
obtained by Appleton-Knapp et al. (2005), described ear-
lier, the present results illustrate that two desirable diffi-
culties are not always more desirable than one. When
spacing interfered with discrimination processes, it impaired
inductive learning.
Spacing does not have to interfere with discriminative
contrast, however. Spacing was beneficial both when exem-
plars were blocked in Experiment 2and when they were
interleaved in Experiment 3. In part, the value of spacing
may depend on how spacing is achieved: Increasing the lag
between interleaved items may be wiser than inserting gaps
between inductive-learning trials. The bottom line is that
interleaving is valuable for inductive learning, and that
spacing can be valuable, too, if it does not interfere with
discriminative contrast.
Concluding comment
In educational settings, formal and informal, it is often impor-
tant to learn a new concept or category by observing, or
studying, exemplars of that category or concept. There may
be situations in which it is advantageous to block such learn-
ing trials, as has been demonstrated using artificial laboratory
tasks, particularly when it is difficult to discern similarities
among the exemplars of a category. In the case of naturalistic
categories, however, the data thus far are fairly unanimous in
demonstrating the benefits of interleaving over blocking
(Kang & Pashler, 2012;Kornell&Bjork,2008; Kornell et
al., 2010; Taylor & Rohrer, 2010; Wahlheim et al., 2011;
Zulkiply & Burt, in press;Zulkiplyetal.,2012).
Importantly, however, even when interleaving is beneficial,
it may seem counterproductive: The great majority of the
participants in the present study, as well as those in prior
studies (Kornell & Bjork, 2008;Kornelletal.,2010;Wahlheim
et al., 2011; Zulkiply et al., 2012), judged that they had learned
more effectively with blocked than with interleaved study.
Thus, a bit of practical advice to learners and educators seems
Mem Cogn
warranted: If your intuition tells you to block, you should
probably interleave.
Author note Grant No. 29192G from the McDonnell Foundation
supported this research. Thanks to Michael Garcia, Sean Kang, and
John Nestojko for their contributions to, and conversations about, this
article. Andrei Baiu, Colin Curzi, Brian Kim, Alexander Manter,
and Conor Ryan contributed to the design and data collection in
Experiment 2. Chris Wahlheim provided the stimuli for Experiment 1.
Appleton-Knapp, S., Bjork, R. A., & Wickens, T. D. (2005). Examin-
ing the spacing effect in advertising: Encoding variability, retriev-
al processes and their interaction. Journal of Consumer Research,
32, 266276. doi:10.1086/432236
Bjork, R. A. (1994). Memory and metamemory considerations in the
training of human beings. In J. Metcalf & A. P. Shimamura (Eds.),
Metacognition: Knowing about knowing (pp. 185205). Cam-
bridge, MA: MIT Press.
Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006).
Distributed practice in verbal recall tasks: A review and quantita-
tive synthesis. Psychological Bulletin, 132, 354380. doi:10.1037/
Chin-Parker, S., & Ross, B. H. (2002). The effect of category learning
on sensitivity to within-category correlations. Memory & Cogni-
tion, 30, 353362. doi:10.3758/BF03194936
Dempster, F. N. (1988). The spacing effect: A case study in the failure
to apply the results of psychological research. American Psychol-
ogist, 43, 627634. doi:10.1037/0003-066X.43.8.627
Glenberg, A. M. (1979). Component-levels theory of the effects of
spacing of repetitions on recall and recognition. Memory & Cog-
nition, 7, 95112. doi:10.3758/BF03197590
Goldstone, R. L. (1996). Isolated and interrelated concepts. Memory &
Cognition, 24, 608628. doi:10.3758/BF03201087
Kang, S. H. K., & Pashler, H.(2012). Learning painting styles: Spacing is
advantageous when it promotes discriminative contrast. Applied
Cognitive Psychology, 26, 97103. doi:10.1002/acp.1801
Kornell, N., & Bjork, R. A. (2008). Learning concepts and categories:
Is spacing the enemy of induction?Psychological Science, 19,
585592. doi:10.1111/j.1467-9280.2008.02127.x
Kornell, N., Castel, A. D., Eich, T. S., & Bjork, R. A. (2010). Spacing
as the friend of both memory and induction in young and older
adults. Psychology and Aging, 25, 498503. doi:10.1037/
Kurtz, H. K., & Hovland, C. I. (1956). Concept learning with differing
sequences of instances. Journal of Experimental Psychology, 51,
239243. doi:10.1037/h0040295
Nelson, T. O., & Narens, L. (1980). Norms of 300 general infor-
mation questions: Accuracy of recall, latency of recall, and
feeling-of-knowing ratings. Journal of Verbal Learning and
Verbal Behavior, 19, 338368. doi:10.1016/S0022-5371
Simon, D. A., & Bjork, R. A. (2001). Metacognition in motor
learning. Journal of Experimental Psychology: Learning,
Memory, & Cognition, 27, 907912. doi:10.1037/0278-7393.
Taylor, K., & Rohrer, D. (2010). The effects of interleaved practice.
Applied Cognitive Psychology, 24, 837848. doi:10.1002/
Thios, S. J., & DAgostino, P. R. (1976). Effects of repetition as a
function of study-phase retrieval. Journal of Verbal Learning
and Verbal Behavior, 15, 529536. doi:10.1016/0022-5371
Vlach, H. A., Sandhofer, C. M., & Kornell, N. (2008). The spacing
effect in childrens memory and category induction. Cognition,
109, 163167. doi:10.1016/j.cognition.2008.07.013
Wahlheim, C. N., Dunlosky, J., & Jacoby, L. L. (2011). Spacing
enhances the learning of natural concepts: An investigation of
mechanisms, metacognition, and aging. Memory & Cognition,39,
750763. doi10.3758/s13421-010-0063-y
Whitman, J. R., & Garner, W. R. (1963). Concept learning as a
function of form of internal structure. Journal of Verbal Learning
and Verbal Behavior, 2, 195202. doi:10.1016/S0022-5371
Zechmeister, E. B., & Shaughnessy, J. J. (1980). When you know that
you know and when you think that you know but you dont.
Bulletin of the Psychonomic Society, 15, 4144.
Zulkiply, N., & Burt, J. S. (in press). The exemplar interleaving effect
in inductive learning: Moderation by the difficulty of category
discriminations. Memory & Cognition. doi:10.3758/s13421-012-
Zulkiply, N., McLean, J., Burt, J., & Bath, D. (2012). Spacing and
induction: Application to exemplars presented as auditory and
visual text. Source: Learning and Instruction, 22, 215221.
Mem Cogn
... The interleaving benefit has been demonstrated across many different types of concepts and age groups. For example, interleaved study or practice has been demonstrated to benefit the learning of perceptual categories, such as artists' painting styles (Kang & Pashler, 2012;Kornell & Bjork, 2008), butterfly species (Birnbaum et al., 2013), and chest radiographic patterns (Rozenshtein et al., 2016). Interleaved study has also been shown to benefit the learning of cognitive concepts, such as mathematics formulae (Rohrer, 2012;Taylor & Rohrer, 2010), the application of non-parametric statistics to different research designs (Sana et al., 2017), the diagnosing of clinical disorders in case patients (Zulkiply et al., 2012), and the classification of organic chemistry compounds (Eglington & Kang, 2017). ...
... One of the key theories as to why interleaving may benefit category learning is that an interleaved schedule juxtaposes examples from different categories and draws learners' attention to the features that discriminate between categories while blocking by categories draws learners' attention to the features that are shared within a category (sequential-attention theory, Carvalho & Goldstone, 2017; discriminative-contrast theory, Kang & Pashler, 2012). Following this line of logic, multiple studies have shown that inserting other tasks between examples to disrupt learners' ability to easily identify discriminating features between categories can attenuate or even eliminate the interleaving benefit (Birnbaum et al., 2013;Sana et al., 2017). ...
... This dual-task manipulation lowered overall performance but did not interact with schedule. This specific manipulation, however, also served to disrupt the juxtaposition between consecutive paintings, possibly changing the nature of the task (see Birnbaum et al., 2013;Kang & Pashler, 2012;Sana et al., 2017). In the present studies, we address these issues. ...
... A convincing body of evidence demonstrates that the sequence of exemplar presentations (i.e., interleaved or blocked practice) has a significant impact on category induction performance (Kornell and Bjork, 2008;Vlach et al., 2008;Kornell et al., 2010;Wahlheim et al., 2011;Kang and Pashler, 2012;Zulkiply et al., 2012;Birnbaum et al., 2013;Zulkiply and Burt, 2013;Carvalho and Goldstone, 2014;Rohrer et al., 2015;Mathy and Feldman, 2016;Eglington and Kang, 2017;Foster et al., 2019). When stimuli are presented in blocked practice, exemplars of the same category are presented sequentially. ...
... The results indicated that participants' classifications of the new paintings were more accurate for artists for whom the paintings had been presented in interleaved practice than in blocked practice. The interleaving effect has been replicated many times (Kornell et al., 2010;Wahlheim et al., 2011;Kang and Pashler, 2012;Birnbaum et al., 2013;Verkoeijen and Bouwmeester, 2014;Metcalfe and Xu, 2016). For example, Birnbaum et al. (2013) used pictures of butterflies from different species to evaluate the acquisition of categories representing natural species using interleaved and blocked practice, and also confirmed the presence of the interleaving effect. ...
... The interleaving effect has been replicated many times (Kornell et al., 2010;Wahlheim et al., 2011;Kang and Pashler, 2012;Birnbaum et al., 2013;Verkoeijen and Bouwmeester, 2014;Metcalfe and Xu, 2016). For example, Birnbaum et al. (2013) used pictures of butterflies from different species to evaluate the acquisition of categories representing natural species using interleaved and blocked practice, and also confirmed the presence of the interleaving effect. Kornell et al. (2010) reported comparable results for older adults, and Wahlheim et al. (2011) demonstrated similar findings using pictures of diverse types of birds, rather than landscapes. ...
Full-text available
Interleaved practice (i.e., exemplars from different categories are intermixed within blocks) has been shown to enhance induction performance compared to blocked practice (i.e., exemplars from the same category are presented sequentially). The main aim of the present study was to examine explanations of why interleaved practice produces this benefit in category induction (known as the interleaving effect). We also evaluated two hypotheses, the attention attenuation hypothesis and the discriminative-contrast hypothesis, by collecting data on participants’ fixation on exemplars, provided by eye-tracking data, and manipulating the degree of discriminative-contrast. In Experiments 1 and 2, participants were instructed to learn the style of 12 new artists in blocked and interleaved practice in fixed-paced and self-paced learning conditions, respectively. We examined fixation durations for six positions (temporal sequence of exemplars presented in each block) using eye-tracking. The results of the two experiments, based on eye-tracking data, suggested that attention attenuation may not be the primary mechanism underlying the interleaving effect in category induction. In Experiment 3, we manipulated the degree of discriminative-contrast to examine the impact on the interleaving effect in category induction. The results showed that the main effect of the degree of discriminative-contrast was significant, and performance in the high-contrast condition was significantly better than those in the medium-contrast and low-contrast conditions. Thus, the current results support the discriminative-contrast hypothesis rather than the attention attenuation hypothesis.
... The effectiveness of SIM-based generative retrieval may be partly driven by inductive and deductive cognitive processes of modeling and simulation, but other researchers have also reported that retrieval practices support inductive and deductive learning even without the use of modeling and simulation tools. While we did not directly measure the impact of SIM-based generative retrieval on basic reasoning skills (i.e., inductive and deductive thinking), recent reviews by Birnbaum et al. (2013); Brunmair and Richter (2019), and Firth et al. (2021) cite that interleaving retrieval practices improve inductive learning-a mental process of acquiring conceptual knowledge from the synthesis of exemplars (Prince and Felder, 2007) that is often known as abstraction in both CT and ST literatures (Wing, 2006;Dunbar and Klahr, 2012;Thagard, 2012;Yaşar et al., 2017;Denning and Tedre, 2019). Wissman et al. (2018) found spaced-out retrieval to improve deductive learning-a mental process of testing factual knowledge, a formula, concept, or theory to various scenarios (Prince and Felder, 2007) as described earlier in terms of distributive processing of information. ...
... This is consistent with the literature on retrieval practices (Agarwal et al., 2020). More importantly, our study complements findings of other recent studies such as Prince and Felder (2007), Birnbaum et al. (2013), Eglington and Kang (2018); Brunmair and Richter (2019);Firth et al. (2021), and Samani and Pan (2021) to point out that retrieval practices enhance core elements of computational and scientific thinking. While other researchers had reported a favorable impact of retrieval practices on inductive and deductive thinking, a more general realization that-taken together with our findings-retrieval practices enhance core CT and ST skills is the most important conclusion of our study. ...
Full-text available
The notion of teaching experts’ habits of mind (e.g., computational thinking and scientific thinking) to novices seems to have inspired many educators and researchers worldwide. In particular, a great deal of efforts has been invested in computational thinking (CT) and its manifestations in different fields. However, there remain some troubling spots in CT education as far as how to teach it at different levels of education. The same argument applies to teaching scientific thinking (ST) skills. A remedy has been suggested to narrow CT and ST skillsets down to core cognitive competencies so they can be introduced in early and middle grades and continue to be nurtured during secondary and post-secondary years. Neuroscientists suggest that the act of (computational) thinking is strongly linked to the acts of information storage/retrieval by our brain. Plus, years of research have shown that retrieval practices promote not only knowledge retention but also inductive reasoning and deductive reasoning. Not surprisingly, these reasoning skills are core elements of both CT and ST skillsets. This article will mesh the findings of a teacher professional development with the existing literature to lay a claim that retrieval practices enhance CT and ST skills. The study offered training to secondary school teachers (n = 275) who conducted classroom action research to measure the impact of retrieval practices on teaching and learning of STEM and CT concepts. We used a quasi-experimental research design with purposeful sampling and a sequential mixed-methods approach focusing on the impact of professional development on teacher outcomes and, in turn, on student outcomes. A survey of teacher participants showed that the majority (96%) of survey respondents (n = 232) reported a good understanding of retrieval strategies, and how relevant ideas can be implemented and tested in the classroom. A large number of action research (target-control) studies by teachers (n = 122) showed that students who learned STEM and CS concepts through retrieval practices consistently scored 5–30% higher than those using the usual blocked practice. In most cases, the difference was statistically significant (p < 0.05). While the study contributes to retrieval practices literature, those looking for best practices to teach core CT and ST skills should benefit from it the most. The study concludes with some recommendations for future research based on the limitations of its current findings.
... Participants were instructed to study 12 artists' painting styles by viewing 72 paintings, with 6 paintings from each artist. For 6 artists, their paintings were presented in a massed format (i.e., participants first studied the 6 paintings from one artist, and then studied the 6 paintings from another artist, and so on), whereas for the other 6 artists, their paintings were presented in an interleaved format (i.e., participants first studied studies (e.g., Birnbaum et al., 2013;Kornell et al., 2010;Tauber et al., 2013;Yan et al., 2016Yan et al., , 2017Zulkiply et al., 2012). ...
... The stimuli in the first learning task were identical to those in Experiment 1. The stimuli in the second learning task were 80 photographs of 16 species of butterflies, with 5 photographs for each species, taken from Birnbaum et al. (2013). All photographs were resized to 900 × 750 pixels. ...
Full-text available
Observing category exemplars in an interleaved manner is more beneficial for inductive learning than blocked (massed) presentation, a phenomenon termed the interleaving effect on inductive learning. However, people tend to erroneously believe that massed is more beneficial than interleaved learning, and learners prefer the former during self-regulated learning. We report four experiments designed to investigate whether explicit instructions, which include individual performance feedback and the interleaving effect results from previous research, can (1) correct metacognitive illusions regarding the interleaving effect, (2) promote self-employment of interleaving, and (3) facilitate category learning. In addition, the current study explored (4) whether the intervention effect is long-lasting and (5) transferable to learning of categories in other domains. Experiments 1–4 established the effectiveness of the instruction intervention to enhance metacognitive appreciation of the interleaving effect, to promote self-employment of interleaving, and to facilitate learning of new categories. The intervention effect was long-lasting (at least 24 h; Experiment 2), and transferable to learning of categories in different domains (Experiments 3 and 4). These findings support the practical use of the instruction intervention.
... (Fabien Mathy) classified as a joyful melody, while the scale "si la sol fa mi re do" is typically classified as a somber one. A large number of studies have confirmed the in-5 tuition that the order in which information is presented can alter perception (Birnbaum et al., 2012;Carvalho & Goldstone, 2014a;Clapper, 2014;Rohrer, 2009Rohrer, , 2012, representation (Corcoran et al., 2011;Elio & Anderson, 1981, 1984Mathy & Feldman, 2016;Sana et al., 2016;Zeithamova & Maddox, 2009), and learning (Bloom & Shuell, 1981;Carpenter & Mueller, 2013;Carvalho & Gold-10 stone, 2015a; Helsdingen et al., 2011;Kornell & Bjork, 2008;Kornell et al., 2010;Lipsitt, 1961;Mathy & Feldman, 2009;Samuels, 1969;Sandhofer & Doumas, 2008;. More specific experimental studies have investigated how the temporal context influences the way categories are learned and encoded (Brady, 2008;Carvalho & Goldstone, 2020, 2014bJones & Sieck, 2003;15 Kang & Pashler, 2012; Li et al., 2012;Mack & Palmeri, 2015;Mcdaniel et al., 2013;Qian & Aslin, 2014;Yan et al., 2017;Zotov et al., 2011;; however, only few of them have attempted to model and account for order effects through the conception and use of computational models. ...
... Finally, our study only examined within-category orders, in particular rulebased vs. similarity-based orders. Other types of orders, such as interleaved vs. blocked study or dissimilarity-based vs. similarity-based orders, has been largely proven to impact the way we learn and represent categories as well (Birnbaum et al., 2012;Carvalho & B. Albuquerque, 2012;Carvalho & Gold-585 stone, 2014b, 2015a ;Carpenter & Mueller, 2013;de Zilva & Mitchell, 2012;Kang & Pashler, 2012;Kornell & Bjork, 2008;Kornell et al., 2010;Kost et al., 2015;Kurtz & Hovland, 1956;Mathy & Feldman, 2009;Rawson et al., 2014;Rohrer, 2009Rohrer, , 2012Sana et al., 2016;Taylor & Rohrer, 2010;Yan et al., 2017;Wahlheim et al., 2011Wahlheim et al., , 2012. A 590 fourth perspective includes the study of the ability of the existing categorization models to predict whether different study sequences will be more or less beneficial to learners. ...
Full-text available
Most categorization models are insensitive to the order in which stimuli are presented. However, a vast array of studies have shown that the sequence received during learning can influence how categories are formed. In this paper, the objective was to better account for effects of serial order. We developed a model called Ordinal General Context Model (OGCM) based on the Generalized Context Model (GCM), which we modified to incorporate ordinal information. OGCM incorporates serial order as a feature along ordinary physical features, allowing it to account for the effect of sequential order as a form of distortion of the feature space. The comparison between the models showed that integrating serial order during learning in the OGCM provided the best account of classification of the stimuli in our data sets.
... Several studies showed that over 60% of the students judged blocked practice to be a more effective strategy than interleaved practice, even though interleaved practice enhanced their category learning more than blocked practice (Birnbaum et al., 2013;Kornell et al., 2010;Zulkiply et al., 2012). This suggests that many students are unable to infer the efficacy of interleaved practice from their task experiences (Kornell & Bjork, 2008). ...
Full-text available
In higher education, many students make poor learning strategy decisions. This, in part, results from the counterintuitive nature of effective learning strategies: they enhance long-term learning but also cost high initial effort and appear to not improve learning (immediately). This mixed-method study investigated how students make learning strategy decisions in category learning, and whether students can be supported to make effective strategy decisions through a metacognitive prompt, designed to support accurate monitoring of effort and learning. Participants (N = 150) studied painting styles through blocked and interleaved practice, rated their perceived effort and perceived learning across time, and chose between either blocked or interleaved practice. Half of the participants (N = 74) were provided with a metacognitive prompt that showed them how their subjective experiences per strategy changed across time and required them to relate these experiences to the efficacy of learning strategies. Results indicated that subjective experiences with interleaved practice improved across time: students’ perceived learning increased as their perceived effort decreased. Mediation analysis revealed that the increased feeling of learning increased the likelihood to select interleaved practice. The percentage of students who chose interleaved practice increased from 13 to 40%. Students’ learning strategy decisions, however, did not benefit from the metacognitive prompt. Qualitative results revealed that students initially had inaccurate beliefs about the efficacy of learning strategies, but on-task experiences overrode the influence of prior beliefs in learning strategy decisions. This study suggests that repeated monitoring of effort and learning have the potential to improve the use of interleaved practice.
... Critically, benefits that result from spaced presentation of examples from the same category and benefits that arise from discriminative contrast of different categories are not mutually exclusive, as CPS argue they are. Both can occur (Birnbaum et al., 2013) and perhaps it is time to theoretically bridge spacing and interleaving theories (Yan et al., 2020). Unfortunately, CPS's two proposed mechanisms are mutually exclusive-learners can either be engaging in discriminative contrast or resting, but not both-and hence does not account for the broader evidence base. ...
Full-text available
In their recent paper, Chen et al. (Educational Psychology Review, 2021) propose that rest periods between deliberate learning characterize the spacing effect and the alternation between skills without rest characterizes the interleaving effect. In this commentary, we show that this theory is inadequate in two aspects. First, the operationalization of their constructs are problematic—their mechanism of rest-from-deliberate-learning mismatches their operationalization (e.g., they code deliberate learning activities that should not allow for working memory recovery as rest-from-deliberate-learning), and their definition of whether stimuli require discriminative contrast appears to depend on the study outcome. Second, their systematic review neglects a large body of literature that is incompatible with their theory. For example, they neglect classic spacing studies on vocabulary learning, and their theory of spacing effects as being a result of working memory recovery cannot account for lag effects or interactions found in the literature. We conclude that there are almost certainly mechanistic differences between spacing and interleaving effects, but the mechanisms are likely not mutually exclusive, as defined by Chen and colleagues.
Concurrent with the publication of the Eleventh Edition of the American Society of Clinical Psychopharmacology (ASCP) Model Curriculum for Training Directors and Teachers of Psychopharmacology in Psychiatric Residency Programs in December 2020, as well as continued advances in translational neuroscience and psychopharmacologic practice, it is time to focus on changes in how psychopharmacology is taught (Glick. Model psychopharmacology curriculum for training directors and teachers of psychopharmacology in psychiatric residency programs. American Society of Clinical Psychopharmacology, 2020). Other than individual papers on teaching psychopharmacology, the last major journal publication with a focus on psychopharmacology education was the 2005 Special Issue of Academic Psychiatry (Zisook et al. Acad Psychiatry. 29(2):141–54, 2005; Glick and Zisook. Acad Psychiatry. 29(2):141–54, 2005). This chapter covers educational objectives for a psychopharmacology curriculum. In addition, this chapter discusses what to teach, how to teach, when to teach, in what settings, and how to evaluate for clinical competency. Our goal is to provide ideas for both new and established residency programs to help with developing, innovating, or updating their psychopharmacology didactic and clinical education programs.
Full-text available
Although current exemplar models of category learning are flexible and can capture how different features are emphasized for different categories, they still lack the flexibility to adapt to local changes in category learning, such as the effect of different sequences of study. In this paper, we introduce a new model of category learning, the Sequential Attention Theory Model (SAT‐M), in which the encoding of each presented item is influenced not only by its category assignment (global context) as in other exemplar models, but also by how its properties relate to the properties of temporally neighboring items (local context). By fitting SAT‐M to data from experiments comparing category learning with different sequences of trials (interleaved vs. blocked), we demonstrate that SAT‐M captures the effect of local context and predicts when interleaved or blocked training will result in better testing performance across three different studies. Comparatively, ALCOVE, SUSTAIN, and a version of SAT‐M without locally adaptive encoding provided poor fits to the results. Moreover, we evaluated the direct prediction of the model that different sequences of training change what learners encode and determined that the best‐fit encoding parameter values match learners’ looking times during training.
Full-text available
examine 2 . . . contributors to nonoptimal training: (1) the learner's own misreading of his or her progress and current state of knowledge during training, and (2) nonoptimal relationships between the conditions of training and the conditions that can be expected to prevail in the posttraining real-world environment / [explore memory and metamemory considerations in training] (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Full-text available
Recent research demonstrates a spacing effect in inductive learning. Spacing different individual exemplars apart in time, rather than massing them together, aids in the learning of categories. Experiment 1 examined whether it is interleaving or temporal spacing that is critical to the spacing effect in the situation wherethe memory load is high, and the results favored interleaving. Experiment 2 examined the effect of the difficulty of the category discrimination on presentation style (massed vs. spaced) in inductive learning, and the results demonstrated that spacing (i.e., interleaving of exemplars from different categories) is advantageous for low-discriminabilty categories, whereas massing is more effective for high-discriminability categories. In contrast to these performance measures, massing was judged by participants to be more effective than spacing in both discriminability conditions, even when performance for low-discriminability categories showed the opposite.
Research on judgments of verbal learning has demonstrated that participants' judgments are unreliable and often overconfident The authors studied judgments of perceptual-motor learning. Participants learned 3 keystroke patterns on the number pad of a computer, each requiring that a different sequence of keys be struck in a different total movement time. Practice trials on each pattern were either blocked or randomly interleaved with trials on the other patterns, and each participant was asked, periodically, to predict his or her performance on a 24-hr test. Consistent with earlier findings, blocked practice enhanced acquisition but harmed retention. Participants, though, predicted better performance given blocked practice. These results augment research on judgments of verbal learning and suggest that humans, at their peril, interpret current ease of access to a perceptual-motor skill as a valid index of learning.
The spacing effect would appear to have considerable potential for improving classroom learning, yet there is no evidence of its widespread application. I consider nine possible impediments to the implementation of research findings in the classroom in an effort to determine which, if any, apply to the spacing effect. I conclude that the apparent absence of systematic application may be due, in part, to the ahistorical character of research on the spacing effect and certain gaps in our understanding of both the spacing effect and classroom practice. However, because none of these concerns seems especially discouraging, and in view of what we do know about the spacing effect, classroom application is recommended.
It is an established finding that spacing repetitions generally facilitates memory for the repeated events. However, the effect of spacing of exemplars on inductive learning is not really known. Two experiments using textual material were conducted to investigate the effect of spacing on induction. Experiment 1 and 2 extended the generality of recently observed spacing benefits to texts, regardless of whether they were visually or aurally presented. Interestingly, participants in both experiments judged massing to be more effective than spacing though their performance showed the opposite. Possible explanations for the superiority of the spaced condition over the massed condition in inductive learning, practical implications of the present study and suggestions for future research are discussed.
F. N. Dempster's (see record 1989-03118-001) recommendation that distributed practice be implemented in the classroom fails to consider such impediments as costs and effects on educators' quality of work life. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
The spacing effect would appear to have considerable potential for improving classroom learning, yet there is no evidence of its widespread application. I consider nine possible impediments to the implementation of research findings in the classroom in an effort to determine which, if any, apply to the spacing effect. I conclude that the apparent absence of systematic application may be due, in part, to the ahistorical character of research on the spacing effect and certain gaps in our understanding of both the spacing effect and classroom practice. However, because none of these concerns seems especially discouraging, and in view of what we do know about the spacing effect, classroom application is recommended. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
College students rated the likelihood of recall of individual words presented for free recall learning. Predictions were made using a 7-point scale immediately following an item's presentation in the list. To-be-rated items included those presented 1 time as well as items presented twice in either a massed (MP) or distributed (DP) manner. Twice-presented items were rated as more likely to be recalled than items presented once, and they were recalled as such. However, although MP items were judged more likely to be recalled than DP items, they were not. The finding that Ss misjudged when they knew MP items suggests why processing may be less for massed than for distributed presentations. Results support the attenuation of attention hypothesis regarding the spacing effect in free recall. (12 ref) (PsycINFO Database Record (c) 2012 APA, all rights reserved)
A concept-attainment experiment was run, using 16 stimuli formed from 4 dichotomous variables, and in which Ss were required to learn a concept (subset of stimuli) of 8 of the 16. Two kinds of subsets of stimuli were used, one with a favorable form of internal structure involving a simple contingency between variables, and one with an interaction between variables. Three methods of presenting stimuli were used: the positive instances alone, both positive and negative instances with each kind grouped together, and both positive and negative instances intermixed.The results showed that the good form of internal structure strongly facilitated concept attainment except when the stimuli were intermixed. This result is interpreted to mean that the intermixing prevents perception of the characteristics of the subsets of stimuli. We have further argued that a primary function of presentation of negative instances in concept learning is to define the larger set of stimuli from which the particular subset is selected; and that if the subset itself defines the larger set, then negative instances can do no good and may make the problem more difficult.
Previous research shows that interleaving rather than blocking practice of different skills (e.g. abcbcacab instead of aaabbbccc) usually improves subsequent test performance. Yet interleaving, but not blocking, ensures that practice of any particular skill is distributed, or spaced, because any two opportunities to practice the same task are not consecutive. Hence, because spaced practice typically improves test performance, the previously observed test benefits of interleaving may be due to spacing rather than interleaving per se. In the experiment reported herein, children practiced four kinds of mathematics problems in an order that was interleaved or blocked, and the degree of spacing was fixed. The interleaving of practice impaired practice session performance yet doubled scores on a test given one day later. An analysis of the errors suggested that interleaving boosted test scores by improving participants' ability to pair each problem with the appropriate procedure. Copyright © 2009 John Wiley & Sons, Ltd.