Andrew Zhang is a sophomore 
majoring in Molecular and Cellular 
Biology. Currently, he is a research 
assistant in Dr. Huimin Zhao’s lab 
and also part of UIUC’s American 
Chemistry Society and REACT. 

How to Improve Memory

 Our abilities to handle novel 
situations and utilize critical thinking 
depends heavily on our ever-expand-
ing memory. While activities like 
problem solving and learning require 
persistence and effort, studies suggest 
there are ways to optimize our time 
and increase our efficiency to remem-
ber new things. Since the late 1800’s, 
research has been uncovering how our 
memory works. Psychological theories 
on memory paved the road for our 
understanding of memory, and many 
classrooms conducted applied research 
to test the efficacy of different learning 
techniques. Recently, neurological 
studies on memory are also corrob-
orating the evidence seen in older 
psychological studies.
 A prominent method for 
learning is the testing effect, which in-
dicates that practicing knowledge with 
test-based questions improves learning 
significantly. While exams may serve 
as a gauge for people’s knowledge in 
the classroom, researchers have begun 
to realize their potential as an effec-
tive and robust learning method. The 
testing effect is seen through improved 
long-term memory, when the memory 
is retrieved during studying. Studies 
have shown that short answer ques-
tions enhance long-term memory the 
best, while other testing methods like 
multiple choice questions or simple 
recall were not as effective (McDaniel 
et al., 2007). Methods like repeated 
studying and rereading proved less 

valuable than just one intermittent test 
(Carpenter, 2009).
 Recent neurological studies 
show increased activity in the brain 
from the testing effect, more so than 
other studying methods. For example, 
in learning Dutch-Swahili translations 
through the testing effect, participants’ 
left inferior parietal and left middle 
temporal lobes activated in fMRI (van 
den Broek et al., 2013). The same 
activity was not seen in traditional 
studying strategies, like repeating the 
lesson (van den Broek et al., 2013). 
In another study, for learning asso-
ciations between nouns, the testing 
effect activated hippocampal regions, 
the prefrontal cortex, and the poste-
rior cingulate cortex,  which are brain 
regions involved in memory retrieval 
cues (Wing, 2013). On the other hand, 
these brain regions were much less 
active in the restudy condition, sug-
gesting that the testing effect is more 
effective at utilizing brain resources to 
encode memory (Wing, 2013).
  The testing effect proved 
robust in many different kinds of 
examinations and different subjects 
(Agarwal et al., 2008). Even tests that 
are quite different from the actual 
examination proved beneficial for 
memory (Carpenter, 2009). Evidence 
led many experts to believe the testing 
effect’s ability to improve learning and 
problem solving in addition to mem-
ory. When it comes to learning and 
memorizing new things, a simple test

or two can be very helpful. The im-
portant implication is that even a bad 
testing session is more effective than 
rereading notes or textbooks.
 While tests may substantially 
improve memory, it is not necessary 
to overload oneself with large exams. 
Researchers would most likely suggest 
the opposite, that by spacing material 
into reasonable learning sessions we 
can achieve a higher retention for the 
particular subject. This idea was first 
proposed by Hermann Ebbinghaus, 
who suggested that memory follows 
a forgetting curve, when information 
fades from memory over time. This 
loss of retention is best counteracted 
by learning and reviewing during sep-
arate occasions, rather than learning 
in only one sitting (Ebbinghaus, 1913). 
This strategy for maximum retention 
became known as the spacing effect 
- the relationship between memory 
acquisition and the spacing of time to 
review the material. When studying 
is spaced out, information tends to 
encode better in long term memory. 
In other words, memory is improved 
significantly with the help of spacing.

 Spacing has seen success in a 
variety of practical situations, especial-
ly the classroom setting. For example, 
in a study conducted on 5th graders, 
students were required to learn diffi-
cult English vocabulary in one of two 
strategies: one taught in mass study 
(everythinig at once) while the other 
re-taught after a 7-day gap (spaced 
repetition) (Sobel & Kapler, 2010). The 
students performed equally well after

Figure 1. A schematic of the forgetting curve. The curve gradually 
lengthens with each review session, representing better retention with 
each review (Chung & Heo, 2018).

1



the first session of learning, but 5 
weeks after the last learning session, 
the those with spaced repetition 
performed significantly better (Sobel 
& Kapler, 2010). Another example 
was seen in a study with children who 
were tasked to remember certain toys. 
Children who were allowed to play in 
between learning each toy were able 
to memorize the toys at a significantly 
better rate compared to children who 
learned the toys all at once (Vlach et 
al., 2008).
 Recently, neurologists have 
studied memory, like the forgetting 
curve and the spacing effect, in the 
brains of animals. The hippocampus 
appears to be crucial in retaining 
memory. In one experiment (Sisti et 
al., 2007), rats were tested on a water 
maze, where they were required to 
learn and memorize the location of a 
platform in the maze. Rats were also 
injected with 5-bromo-2-deoxyuridine 
(BrdU), which labels newly synthe-
sized cells. Compared to normal rats, 
those with their hippocampus dam-
aged through irradiation performed 
significantly worse in the water maze 
only after a few weeks, and showed de-
creases in BrdU in neurons, meaning 
less formation of new neurons (Snyder 
et al., 2005). It is hypothesized that 
new neurons in the hippocampus were 
not necessary for learning, since mice 
with a damaged hippocampus per-
formed equally well with normal rats. 
However, new neurons are necessary 
for retention of memory, as seen by a 
drastic forgetting curve without them. 
A second experiment was conduct-
ed, where two groups of rats learned 
a water maze in either a single mass 
session (all at once) or with spacing. 
The rats with spaced learning per-
formed significantly better than those 
without and were correlated with more 
BrdU labeled cells in the hippocam-
pus, suggesting neurological changes 
due to the spacing effect (Sisti et al., 
2007). Overall, these studies point to 
the impact of the spacing effect on the 
preservation of new neurons, which in 
turn helps retain more information.

and also outperforming the previous 
two groups (Kang & Pashler, 2011). 
Based on these findings, it appears the 
spacing effect was not responsible for 
improving in associations. Rather, in-
terleaving is responsible for improving 
the ability to differentiate and associat-
ing pieces of information.
 Not only does interleaving 
improve associations and differenti-
ations, it has been shown to improve 
test performance in a practical setting. 
For example, in the following study 
(Rohrer & Taylor, 2007), interleaving 
improved math scores for students 
practicing math problems. Spacing 
was not controlled for (students were 
not doing multiple math problems at 
the same time), which resembles more 
a practical classroom setting. The 
students were split into three groups. 
One group learned and practiced math 
through mixed topics (interleaving). 
Another group practiced through 
blocked review, practicing one con-
cept at a time. A third group also 
used a blocked review but included 
overlearning, meaning they complet-
ed multiple problems testing a single 
concept at a given time. Referred to 
the masser group, they did twice as 
many problems as the original block 
group. The interleaving group overall 
did the same amount of problems as 
the masser group but spread at inter-
vals the same size as the original block 
review. When tested, the masser group 
performed only slightly better than 
the original block group. However, the 
interleaving group performed signifi-
cantly better than both groups. This 
suggests that additional practice is 
only useful for learning if spaced and 
mixed.
 Studies on the neurological 
basis of interleaving are novel. In 
one study, (Lin et al., 2011) partici-
pants were required to perform serial 
(ordering) tasks, requiring some but 
minimal upper body motion. In order 
to do so, participants must learn a 
specific sequence. One group learned 
through block training, and another 
through interleaving. The participants

Figure 2. Learning correlated with BrdU-labeled cells (Sisti et al., 2007).

 In recent years, it is found that 
even the spacing effect can be further 
improved upon in strategies that make 
learning and memory consolidation 
more efficient. A similar but relatively 
new approach of learning is interleav-
ing, or mixing subjects together while 
learning. For example, one can learn 
both math and English concepts in the 
same hour, alternating between the 
two subjects every couple of minutes. 
Many interleaving techniques inevita-
bly introduce spacing effects. Concepts 
from one subject are separated in time 
in order to sandwich concepts from 
a different subject. However, even in 
controlling for spacing, studies suggest 
that interleaving promotes stronger 
associations with similar concepts and 
stronger differentiation between differ-
ent concepts (Kang & Pashler, 2011). 
Basically, interleaving helps improve 
and sharpen memory.
 In one study, subjects were 
tasked to learn and identify paintings 
by the artists. One group was shown 6 
paintings of each painter all at once. A 
second group had mixed the orders of 
paintings. Both groups were then ad-
ministered distractor tasks to perform. 
When tested for the paintings later, the 
mixed group performed significantly 
better at identifying painters (Kor-
nell & Bjork, 2008). Another study 
followed up with a similar setup. This 
time, the two groups were tested with 
no mixed order, but the spacing of 
time between each painter and paint-
ing pair learned was changed. This 
resulted in no significant difference 
in performance. In the same study, 
another setup included mixed orders, 
which were shown either simultane-
ously or spaced with time. Again, the 
two groups performed equally well 

2



were studied under fMRI blood-oxy-
gen-level-dependent signals (BOLD) 
and excitability in the primary motor 
cortex (M1) through transcranial mag-
netic stimulation. During retention 
(learning phase), BOLD in prefrontal 
and sensorimotor regions and M1 
excitability were higher in the inter-
leaving group. Initially, the interleav-
ing group performed tasks with slower 
reaction time than the block training. 
However, after 5 days, the interleav-
ing group experienced faster reaction 
times. M1 excitability was still higher, 
but BOLD in prefrontal regions were 
weaker compared to the block training 
group. These results suggest that in-
terleaving produces higher activity in 
parts of the brain for learning, as seen 
by BOLD. Over time, the brain incor-
porates the information. This makes 
retrieval more efficient, requiring 
less activity in brain regions as seen 
by decreased BOLD. M1 excitability 
shows higher activation of relevant 
brain regions in completing tasks. It is 
plausible other areas of the brain are 
also easily excitable when activated 
through interleaving.

The incorporation of ideas and infor-
mation inro long term memory is in-
credibly important. To effectively use 
one’s memory, one must also be able to 
retrieve information and use it. Much 
of that brain power relies on working 
memory, which is closely tied to short 
term memory. Additionally, any new 
pieces of information must first go 
through the short term memory befo-

re it can be stored in the long term 
memory.
 The working memory allows 
the brain to act on or even modify in-
formation. For example, the brain can 
imagine breaking a chair without one 
actually breaking the chair in real life. 
Short-term memory cannot incorpo-
rate an infinite amount of information 
at the same time, however. In a very 
famous historical paper, George Miller 
estimates the limit to be 7±2 pieces of 
information (Miller 1956). However, 
the limit is actually not definite. Some 
pieces of information themselves con-
tain information, which are known as 
chunks. The chunk does not yet have 
a rigorous definition in the scientific 
community, but it is thought to be a 
group of information that the brain 
handles as one entity. In other words, 
a single chunk will consist of many 
pieces of information while taking less 
space in working memory. Howev-
er, chunks do not completely bypass 
Miller’s estimate. Further studies have 
shown the capacity of the brain to 
handle up to 4±1 chunks (Crowan, 
2010), which is less than Miller’s orig-
inal estimate. Because chunks them-
selves contain more information, each 
chunk takes up more space in working 
memory than a single item.

In a recent study conducted, partic-
ipants were required to memorize a 
sequence of numbers. Depending on 
how many numbers were contained in 
each chunk, the maximum chunks the 
brain can handle varied. When chunks 
were only one number each, the limit 
was about 7. When chunks became 
very long, around 5 numbers each, 
the brain could only handle about 3-4 
chunks (Mathy & Feldman, 2012). 
This corroborates the idea that the 
brain has a capacity for working mem-
ory, even when chunking. Despite

this, chunking still helps carry more 
information in working memory than 
individual pieces of information alone.
 The ability to use chunking 
effectively improves memory usage 
and memory consolidation dramati-
cally. For example, studies conducted 
show that chess players rely on chunk-
ing entire movesets in a given board, 
like helping players remember where 
individual pieces are on a board, given 
only a few seconds to see the board 
(Linhares & Brum, 2007). It is also 
shown that pattern recognition in 
games like chess correlates with skill 
(Linhares & Brum, 2007).
 Some neurological insights 
into chunking have corroborated with 
previous studies on its efficacy. For ex-
ample, in one study (Bor et al., 2003), 
participants were required to memo-
rize spatial patterns. One group had 
a disruption in learning at a random 
point in time. Another had a disrup-
tion specifically in between two differ-
ent sets of information, establishing 
meaningful chunks in the participants’ 
memory. The second group performed 
better, and in fMRI brain scans, their 
prefrontal cortex was also lit up more 
(Bor et al., 2003). Chunking produc-
es higher activity in brain regions 
important for processing information, 
and chunking can improve short term 
memory. Ultimately, with chunking, 
higher activity allows for better con-
solidation of information.
 While most memory and 
learning techniques were developed 
recently, there are some ancient tech-
niques still used today, like the method 
of loci. Also known as the “memory 
palace,” people would imagine putting 
pieces of information in each “room” 
of a building they are familiar with. 
Retrieval of memory simply requires 
finding the right “room.” The tech-
nique was first used by ancient Greeks 
to memorize speeches, and now it is 
used in memory competitions, allow-
ing people to effectively memorize 
large chunks of information (Dresler 
et al., 2017). Additionally, the memory 
technique is just as effective when usi-

Figure 3. Increased blood flow was higher during practice in individ-
uals with interleaving (top image, bottom row). During the retention 
phase, interleaving showed less blood flow activity compared to the 
control (bottom image, bottom row). Presumed that interleaving is 
more efficient, requiring less effort during retention (Lin et al. 2011).

Figure 4. Basic schematic of encoding and retaining memory (Esteve 
2016).

3



ng locations in virtual reality as in 
with real locations (Legge et al., 2012).
The memory technique is uniquely 
a mental construct, but it provides tan-
gible improvements for information 
consolidation.
 Using the method of loci ef-
fectively requires practice and training 
(Legge et al., 2012). To test for the 
effectiveness of the memory palace, a 
study was conducted on older subjects 
to practice memorizing a list of words. 
The subjects were trained in the meth-
od of loci during the study. The adults 
who were asked to utilize the memory 
palace technique performed signifi-
cantly better at remembering words 
compared to the control (Gross et al., 
2014). On pieces of paper, those who 
used the method of loci remembered 
words in the correct order, and even 
left spaces in between for words they 
forgot (Gross et al., 2014).
 Neurological correlates also 
indicate the effectiveness of the meth-
od of loci. For example, in a neuro-
logical study conducted on memory 
atheletes and control participants, 
those who utilized the method of loci 
performed significantly better than 
other strategies, like active or passive 
learning, even up to at least 4 months 
later (Dresler et al., 2017). In an fMRI 
scan done on the participants, during 
memory consolidation and retrieval, 
those who trained with memory of 
loci had heightened activity between 
visual lobes, temporal lobes, and 
default mode networks (Dresler et al., 
2017). It is believed that the method of 
loci promotes increased connectivity 
between different parts of the brain, 
promoting memory consolidation.
 Evidence-based research in 
effective memory techniques is rela-
tively new. While some methods were 
well-known since ancient times, most 
have only been uncovered recently. 
Neurological studies on the effects 
of memory techniques are currently 
ongoing but already substantiate the 
techniques. Despite the significantly 
improved performances from these 
techniques, many participants in these

studies believed traditional studying 
strategies were more effective. As re-
searchers begin to understand more of 
these memory techniques, it is crucial 
that people will also learn to under-
stand the importance of these tech-
niques as well. Learning new material 
can require effort, but there are always 
strategies to make learning and mem-
orizing easier and more efficient.

References
 Agarwal, Pooja K., et al., 19 Sept. 2008, “Examining the 
Testing Effect with Open- and Close- Boot Tests.” Applied Cognitive 
Psychology, vol. 22, no. 7, pp. 861-876, John Wiley & Sons, doi: 
10.1002/acp.1391. 

 Bor, Daniel, et al., 23 Jan. 2003, “Encoding Strategies 
Dissociate Prefrontal Activity from Working Demand.” Neuron, vol. 
37, no. 2, pp. 361-357, Cell Press, doi: 10.1016/S0896-6273(02)01171-6.

 Carpenter, S. K., 2009, “Cue Strength as a Moderator of 
the Testing Effect: The Benefits of Elaborative Retrieval.” Journal of Ex-
perimental Psychology: Learning, Memory, and Cognition, vol. 35, no. 
6, pp. 1563-1569, American Psychological Association, doi: 10.1037/
a0017021. 

 Chung, Bo A., and Heo, Hae J., Jan. 2018, “The effect of 
flipped learning on academic performance as an innovative method for 
overcoming Ebbinhaus’s forgetting curve.” International Conference on 
Information and Education Technology, vol. 6, pp. 56-60, Association 
for Computing Machinery, doi: 10.1145/3178158.3178206.

 Crowan, N., 2 Mar. 2021,“The Magical Mystery Four: 
How is Working Memory Capacity Limited, and Why?” Current 
Directions in Psychological Science, vol. 19, no. 1, pp. 51-57, SAGE 
Publications,doi: 10.1177/0963721409359277.

 Dresler, Martin, et al., 8 Mar. 2017, “Mnemonic training 
reshapes brain networks to support superior memory.” Neuron, vol. 93, 
no. 5, pp. 1227-1235, Cell Press, doi: 10.1016/j.neuron.2017.02.003.
 Crowan, N., 2 Mar. 2021,“The Magical Mystery Four: 
How is Working Memory Capacity Limited, and Why?” Current 
Directions in Psychological Science, vol. 19, no. 1, pp. 51-57, SAGE 
Publications,doi: 10.1177/0963721409359277.

 Dresler, Martin, et al., 8 Mar. 2017, “Mnemonic training 
reshapes brain networks to support superior memory.” Neuron, vol. 93, 
no. 5, pp. 1227-1235, Cell Press, doi: 10.1016/j.neuron.2017.02.003.

 Ebbinghaus, Hermann., 1913. “Memory: A Contribu-
tion to Experimental Psychology.” Translated by Ruger, Henry A. and 
Bussenius, Clara E.
 `Esteve, Clàudia Y., 2016.,
 “Very young learners’ vocabulary development in English : a case 
study with 4 and 5 year-old children.” 

 Gross, Alden L., et al.,13 Mar. 2014, “Do Older Adults 
Use the Method of Loci? Results from the ACTIVE Study.” Exper-
imental Aging Research, vol. 40, no. 2, pp. 140-163, Routledge,doi: 
10.1080/0361073X.2014.882204.

 Kang, Sean H. K., et al., 02 May 2011, “Learning Paint-
ing Styles: Spacing is Advantageous when it Promotes Discriminative 
Contrast.” Applied Cognitive Psychology, vol. 26, no. 1, pp. 97-103, 
John Wiley & Sons, doi: 10.1002/acp.1801. 

 Kornell, Nate, and Bjork, Robert A.,1 Jun. 2008, “Learn-
ing Concepts and Categories Is Spacing the ‘Enemy of Induction’?” 
Psychological Science, vol 19, no. 6, pp. 585-592, SAGE Publications, 
doi: 10.1111/j.1467-9280.2008.02127.x.

 Legge, Eric L. G., et al. 2012.09.002.,“Building a 
memory palace in minutes: equivalent memory performance using 
virtual versus conventional environments with the Method of Loci.” 
Acta Psychologica, vol. 141, no. 3, pp. 380-390, Elsevier, nov. 2012, doi: 

10.1016/j.actpsy.

 Lin, Chien-Ho (Janice), et al., 1 Jun. 2011, “Brain-be-
havior correlates of optimizing learning through interleaved practice.” 
NeuroImage, vol. 56, no. 3, pp. 1758-1772, Elsevier, doi: 10.1016/j.
neuroimage.

 Linhares, Alexandre, and Brum, Paulo., 10 Jan. 2010 
“Understanding Our Understanding of Strategic Scenarios: What Role 
do Chunks Play?” Cognitive Science, vol. 31, no. 6, pp. 989-1007, John 
Wiley & Sons, doi: 10.1080/03640210701703725.

  Mathy, Fabien, Feldman, Jacob., 2011.11.003. “What’s 
magic about magic numbers? Chunking and data compression in 
short-term memory.” Cognition, vol. 122, no. 3, pp. 346-362, Elsevier , 
doi: 10.1016/j.cognition.

 McDaniel, Mark A., et al., 02 Jul. 2007, “Testing the 
testing effect in the classroom.” Journal of Cognitive Psychology, vol 19, 
no. 4-5, pp. 494-513, Routledge, doi: 10.1080/09541440701326154. 

 Miller, George., Mar. 1956. “The magical number seven, 
plus or minus two: some limits on our capacity for processing informa-
tion.” The Psychological Review, vol. 63, no. 2, American Psychological 
Association,  

 Rohrer, Doug, and Taylor, Kelli. 19 Apr. 2007, “The 
shuffling of mathematics problems improves learning.” Instructional 
Science, vol. 35, no. 6, pp. 481-498, SpringerLink, doi: 10.1007/s11251-
007-9015-8.
 
 Sisti, Helene M., et al. 19 Apr. 2007 “Neurogenesis and 
the spacing effect: Learning over time enhances memory and the sur-
vival of new neurons.” Learning & Memory, vol. 14, no. 5, pp. 368-375, 
Cold Spring Harbor Laboratory Press, doi: 10.1101/lm.488707.

 Sobel, Hailey S., et al. “Spacing Effects in Real-World 
Classroom Vocabulary Learning.” Applied Cognitive Psychology, vol. 
25, no. 5, pp. 762-767, John Wiley & Sons, 22 Sep. 2020, doi: 10.1002/
acp.1747.

 Van den Broek, Gesa S. E., et al. Sep. 2013, “Neural 
correlates of testing effects in vocabulary learning.” NeuroImage, vol. 
78, pp. 94-102, Elsevier,doi: 10.1016/j.neuroimage.2013.03.071.

  Vlach, Haley. A., et al. 2008.07.013.“The spacing effect 
in children’s memory and category induction.” Cognition, vol. 109, no. 
1, pp. 162-167, Elsevier, Oct. 2008, doi: 10.1016/j.cognition.

 Wing, Erik A., et al. Oct. 2013, “Neural correlates of 
retrieval-based memory enhancement: An fMRI study of the testing 
effect.” Neuropsychologia, vol. 51, no. 12, pp. 2360-2370, Elsevier, Oct. 
2013, doi: j.neuropsychologia.

4