INSIGHT AND MEMORY

Insight enhances learning for incidental information. New evidence

supports the insight memory advantage.

Carola Salvi1,2, Nicole Keller1, Samuel E. Cooper1, Emily Leiker3 and Joseph Dunsmoor1

1. Department of Psychiatry and Behavioral Sciences, University of Texas at Austin.

2. Department of Psychological and Social Sciences, John Cabot University, Rome, Italy.

3. Department of Psychiatry, University of Pittsburgh. Pittsburgh.

[SHORTENED TITLE UP TO 50 CHARACTERS] 2

Abstract

Research on creative problem-solving finds that solutions achieved via spontaneous insight (i.e.,

Aha! moment) are better remembered than solutions reached without this sense of epiphany,

referred to as an ‘insight memory advantage.’ We hypothesized that the insight memory

advantage can spread to incidental information encoded in the moments surrounding

spontaneous insight as well. Participants (N= 291) were presented with incidental, and unrelated

to problems, scholastic facts immediately after indicating they reached a solution to a word

problem (i.e., Rebus Puzzles) but prior to entering the answer. Participants indicated whether

they reached the solution via either insight or a step-by-step analysis. Memory results showed

better performance for incidental scholastic facts presented when problem solving was

accompanied by a spontaneous (Aha! Experience) and induced (D’oh! Experience) insight,

compared to solutions reached with analysis. This finding suggests that the memory advantage

for problems solved via insight spreads to other unrelated information encoded in close temporal

proximity and has implications for novel techniques to enhance learning in educational settings.

Keywords: memory, problem-solving, insight, Aha! Moment, D-oh! Moment, creativity,

learning
[SHORTENED TITLE UP TO 50 CHARACTERS] 3

Introduction

People tend to have better memory for information they are incentivized to remember.

These include intrinsically motivating events, such as novel or emotional experiences, as well as

extrinsically motivating events, such as remembering information for reward or the risk of

punishment for forgetting (Shohamy & Adcock, 2010; Murty & Adcock, 2014). Research on

creative problem-solving shows that idea generation accompanied by a feeling of insight

enhances memory for both the problem and the solution (Danek & Wiley 2020; Becker, Cabeza,

& Kizilirmak, 2022; Danek & Wiley, 2020; Kizilirmak, & Becker 2022; Danek, Fraps, von

Müller, Grothe, & Öllinger, 2013; Kizilirmak et al., 2015). Solving a problem is an intrinsically

pleasing and rewarding event (Danek and Wiley, 2017). Such an experience is indeed often

externalized by exclamations such as Aha! when the solver achieves a solution surprise and

excitement, or D-oh! when the solution is revealed to the solver after having a failing, indexing a

feeling of obviousness but also satisfied curiosity. The memory enhancement for solutions

reached via insight, versus solutions reached without insight, is referred to as the ‘insight

memory advantage’ (Danek & Wiley, 2020). Interestingly, affective experiences not only

promote long-term memory for the target information but can extend to neutral incidental

information that happened to be presented close in time to the salient aspects of the experience

(Murphy, Dehmelt, Yonelinas, Ranganath, & Gruber, 2021). Whether the insight memory

advantage likewise spreads to incidental information encoded close in time to the moment of

insight is unknown. Here, we sought to leverage the insight memory advantage to capture

memory for incidental information (scholastic facts) presented in temporal proximity to an

epiphany.
[SHORTENED TITLE UP TO 50 CHARACTERS] 4

The idea that insight-based solutions are associated with enhanced memory has a long

history in the field of problem-solving (see e.g., Dominowski & Dallob, 1995; Osgood, 1953;

Woodworth & Schlosberg, 1954). A renewed interest in this matter was sparked by recent

empirical evidence showing that problem solutions are better recalled, specifically, when

accompanied by an Aha! moment (Danek, Fraps, von Müller, Grothe, & Öllinger, 2013; Danek

and Wiley, 2020 but also Kizilirmak, et al., 2015; see Kizilirmak & Becker, 2022 for a review).

Danek et al. (2013) found that when spontaneous (i.e., self-generated) correct solutions to magic

tricks were accompanied by an Aha! Experience they were more likely to be remembered after a

two-week period compared to solutions without Aha! experience. Later in 2020 Danek and Wiley

used the same procedure and found that is the pleasurable affective experience associated with

Aha! Moments to be responsible for the insight memory advantage. Kizilirmak, and colleagues

(2015) also found a similar memory advantage with a delay of one week on a perceptual

problem-solving task (“Mooney images” i.e., degraded picture), and one more time memory was

associated with a higher rating of pleasure when accompanied by a subjective feeling of Aha!

than without. Interestingly, they found that also the sudden revelation of the

solution that induces a comparable feeling to the ‘‘Aha!’’ (i.e., the D’oh!

Experience) for self-solved items causes a memory advantage. They argued

that this effect caused by the presented solutions accompanied by the

“Aha!” could be due to the pleasurable experience associated with the

discovery of the solution, as well as the representational change (i.e.,

restructuring).

Problem-solving is rewarding, and people like being engaged in crossword puzzles,

riddles, trivia questions, murder mysteries, and escape rooms (Oh et al., 2020). Figuring out a
[SHORTENED TITLE UP TO 50 CHARACTERS] 5

problem solution is inherently pleasurable, sometimes regardless of succeeding in it. Whether the

problem solution is achieved spontaneously or revealed, the restructuring of the initial

representation of the problem that is now seen in a new light generates a feeling of surprise and

pleasure. In addition to the affective response associated with the Aha! scholars suggest that the

memorability of problems and their solutions might be due also to the restructuring associated

with finding a new organization for problem elements which occurs both when people generate

their problem solutions but also when the solution is revealed (Danek & Wiley 2020; Kizilirmak

et al., 2015).

While used indistinctively, spontaneous and induced insights are emotionally indexed by

different exclamations such as: Aha! and D’oh! While the term Aha! Experience has been largely

used for spontaneous insights we refer to the ‘D’oh! Experience’, as those moments where

people fail to solve a problem and the solution is revealed to them. The frustration of not having

been able to figure out the solution, which when it is revealed feels obvious, mixed with the

satisfaction of now knowing the solution is indexed by the exclamation ‘D’oh!’ While this is

different from having an Aha! experience, (i.e., when the problem solution rises suddenly as an

insight) the D’oh! experience is still associated with a feeling of pleasure and reward, as matter

of fact when the solution to a problem is not revealed to us that generates frustration. Kizilirmak

referred to what we name the ‘D’oh’ experience as an induced insight, and in 2021 showed how

this phenomenon evokes a positive feeling which serves as an intrinsic reward since they are

associated with brain regions, as the striatum, associated with reward short reward and

reinforced-based learning (Knutson et al., 2001; Wittmann et al., 2005; Harunoet al., 2004;

Kahnt et al.,2009) as well as the hippocampus, (important for explicit memory, detection, and
[SHORTENED TITLE UP TO 50 CHARACTERS] 6

encoding of novel stimuli, contexts, and associations; Ranganath and Rainer, 2003) and the

amygdala (important for emotional memory (e.g., McGaugh, 2004; Phelps and LeDoux, 2005).

Collectively, these components link the insight memory advantage to research on

emotional enhancements in memory, which consistently shows superior memory for emotional

versus neutral information (Danek et al., 2013; Kizilirmak, et al., 2015 but also McGaugh, 2015).

An intriguing byproduct of salient events on memory includes effects on neutral information that

is presented in close temporal proximity (e.g., Dunsmoor, Murty, Clewett, Phelps, Davachi,

2022). Whether the insight memory advantage enhances memory for other incidental information

encoded close in time to the problem and solution is unknown. There is, however, increasing

evidence that phasic changes in motivational states, like those generated during creative

problem-solving, boost memory for irrelevant information that is presented in temporal

proximity.

Not all the events we live in life are stored in long-term memory, most are forgotten. For

example, we probably do not remember what we ate two Mondays ago, however, we are more

likely to remember what we ate on the first date with our significant other. Some episodes are

remembered for a longer time thanks to consolidation, when this information is emotionally

charged, and thus subject to stabilization in the neocortex (Squire, 1992; Dudai et al., 2000).

Initial retention occurs when a novel, or a rewarding event, happens shortly before or after the

time of memory encoding, similar to a ‘flashbulb memory’ (McGaugh, 2004; Brown, & Kulik,

1977; (Dunsmoor, Murty, Davachi, & Phelps, 2015). For example, Gruber and colleagues find

that high states of curiosity to know the answer to a trivia question improves memory both for

the answer and for incidentally presented visual stimuli (i.e., faces) presented during an

anticipation phase prior to receiving the answer. The state of high curiosity generated by a trivia
[SHORTENED TITLE UP TO 50 CHARACTERS] 7

question and the state of insight generated by realizing the solution to a puzzle share common

elements both psychologically and neurobiologically (e.g., activation throughout the

dopaminergic midbrain and striatum) (Danek & Wiley, 2017b, 2020b; Kizilirmak et al., 2019,

2021; Oh et al., 2020b; Tik et al., 2018). The state of curiosity generated by a trivia question and

the state of insight generated by realizing the solution to a puzzle share common elements both

psychologically and neurobiologically (e.g., activation throughout the dopaminergic midbrain

and striatum) (Danek & Wiley, 2017, 2020; Kizilirmak et al., 2016, 2019, 2021; Gruber,

Gelman, Ranganath, 2014; Oh et al., 2020b; Tik et al., 2018). However, there are important

distinctions in the temporal dynamics between states of curiosity generated by trivia questions

and insight-based problem-solving that may affect memory for incidental information presented

in temporal proximity.

While Gruber and colleagues find that high states of curiosity to know the answer to a

trivia question improves memory both for the answer and for incidentally presented visual

stimuli (i.e., faces) shown during an anticipation phase prior to receiving the answer, the relation

between the curiosity of knowing the solution of a problem and memory is unknown.

In the present study, we examined whether insight-based problem solving enhances

memory for unrelated incidental information presented in temporal proximity to a spontaneous

insight. Expanding upon recent work on incidental memory during states of reward and curiosity,

the incidental memoranda were scholastic facts, rather than visual items. The goal of using

scholastic facts was to draw this line of work closer to potential applications for education. For

example, if the insight memory advantage spreads to other information, then one application

could involve embedding study material with recreational problem solving to improve learning

and retention.
[SHORTENED TITLE UP TO 50 CHARACTERS] 8

To investigate this relation, we asked participants to solve a series of word puzzles

(Rebus Puzzles) and indicated whether or not they reached an answer. Then, we presented an

unrelated scholastic fact either immediately after subjects indicated they reached a solution (prior

to providing the answer, either with or without Aha! Moment), or immediately after receiving the

solution to a problem they indicated they could not solve (D’oh! Moments). Importantly, we

distinguished between solutions reached via spontaneous insight (Aha!) and problems solved

without an accompanying sense of insight, referred to as a step-by-step solution, or unsolved

problems to which a solution was provided (D’oh! Moments). Following the problem-solving

phase, participants underwent a surprise memory test for the incidentally presented scholastic

facts in a multiple-choice test. Given the conceptual and potential mechanistic overlap between

the insight memory advantage and emotional memory enhancements, we predicted an

enhancement in memory for incidental facts presented at the moment accompanying a sense of

insight versus a problem solution reached without insight. Given research indicating the insight

memory advantage also occurs for induced insights (Kizilirmak et al., 2015; 2021), we did not

predict a difference in memory for incidental facts presented after puzzles solved via insight

(Aha! Moments) versus incidental facts delivered after the solution for puzzles subjects could not

answer (an induced insight, what we refer to as D’oh! Moments).

Methods

Participants and data cleaning

Participants were recruited online via the CloudResearch platform (Littman et al., 2017)

on Amazon Mechanical Turk (MTurk). To mitigate data quality concerns that can arise with

online participants (Kennedy et al., 2020), we used CloudResearch’s Approved Participants

[SHORTENED TITLE UP TO 50 CHARACTERS] 9

feature to recruit only vetted MTurk workers (i.e., participants who have shown prior evidence of

attention and engagement). Eligibility was restricted to individuals aged 18-50, in the United

States, with American English as a first language. Participants took ~1 hr (median completion

time: 42.3 minutes), to complete the study for a total of $4 upon completion. Study procedures

were approved by the IRB at the University of Texas at Austin and all participants provided

written informed consent.

A total of 351 participants completed the study. From that sample, we excluded

participants who did not fully complete the task or did not understand the instruction (e.g.,

instead of trying to solve the problems they were typing in the scholastic fact, never attempted to

solve the Rebus puzzles (i.e., they always pressed ‘ NO’ when they have to report if they had the

solution of the problem), did not solve any problem correctly, did not remember any of the

scholastic facts, or missed the catch trials. We also removed participants who declared they

solved all the problems via insight or via step-by-step and those who solved less than M-/+2.5

SD of the problems correctly. After the data cleaning, the sample included 291 participants (177

female, 108 male, and 6 nonbinary or genderfluid; age average 34.5 ± SD 7.7). For included

subjects, we eliminated trials for which subjects’ reaction time was shorter than 2 seconds (Salvi

et al., 2015) and longer than 30 seconds and trials for which participants gave an unspecified

solution, such as scrambled letters or sentences as ‘I forgot the solution.’

Procedure

Participants were first directed to an informational page and online consent form.

Following written consent, participants completed a short demographic survey, the experimental
[SHORTENED TITLE UP TO 50 CHARACTERS] 10

paradigm (consisting of an incidental learning phase and surprise recognition memory phase),

and a series of questionnaires. Each of these components is detailed below.

Demographic information. The demographic section included a series of self-report

questions assessing participants’ age, gender identity, marital status, level of education,

occupation, country of residence, and political ideology.

Questionnaires. As this study was conducted amidst the Covid-19 pandemic, we included

a variety of questionnaires to assess the rise in anxiety, fear of COVID-19, uncertainty,

xenophobia, as well as belief in conspiracy theories. None of the questionnaire variables were of

a priori interest and did not affect the results of the experiment.

Experimental Paradigm

Following a short practice of one trial participants were asked to attempt solving 44

randomized Rebus Puzzles taken from (Gregor, 2009; MacGregor & Cunningham, 2008).

Several studies, both online and in person, show how these types of puzzles are an established

measure of insight problem-solving (e.g., Salvi, Costantini, Bricolo, Perugini & Beeman, 2015;

Salvi, Costantini, Pace & Palmiero, 2018; Salvi, et al., 2021; Salvi, Bricolo, Bowden, Kounios,

& Beeman, 2016; Threadgold et al., 2018). To solve these puzzles, participants were instructed to

identify a common phrase from the verbal and visual clues provided on screen (e.g., ‘cycle,

cycle, cycle’ would be solved as ‘tricycle’, for another example see figure 1). After each puzzle,

participants were presented with different scholastic items that were unrelated to the problem

they were asked to solve. The 44 scholastic items were selected from three topic areas: English,

history, and science. The content of these facts did not pertain to general knowledge, but rather

information that could be learned in a high school or college setting. Participants were not
[SHORTENED TITLE UP TO 50 CHARACTERS] 11

instructed to remember the scholastic items (incidental encoding), just to read the sentences

carefully.

Participants had 15s to solve each puzzle within which they had to indicate if they had

found the solution of the Rebus. If they thought, they had a solution they were asked to

‘immediately’ press the YES button, the scholastic item appeared immediately and remained on

the screen for 3sec. Following participants had to type in the problem solution and report if that

solution was achieved via insight or via step-by-step.1

If participants did not have a solution to the Rebus they pressed the NO button, or the

trial would time out and ask them if they had a solution to the Rebus. If participants pressed the

NO button the problem solution would immediately appear on-screen for 1s, followed by a

scholastic item that lasted 3s. After all trials were completed, there was a brief 1-min washout

period where participants watched a neutral video clip of a boat moving through water.

Surprise Recognition Memory Phase. Participants completed an immediate surprise

recognition memory test on the neutral facts encountered in the incidental learning phase. The

test consisted of 44 multiple-choice trials assessing their memory for the neutral facts, plus 3

catch trials that served as attentional checks. Multiple-choice trials were formatted as either

questions or fill-in-the-blanks (e.g., ‘Elements in their standard state have ______ number of

electrons and protons.’). Participants were given the 20s per question to select the correct answer

from 4 options (e.g., ‘the same,’ ‘a larger,’ ‘a smaller,’ or ‘a much larger’; correct answer: ‘the

same’).

1
The same instructions used by Salvi et al. (2015) were given to participants to explain how to distinguish solutions via insight from those via step-by-step. The
original instructions are: ‘[. . . ] by INSIGHT means that the answer suddenly (i.e., unexpectedly) came to your mind, while you were trying to solve the problem,
even though you are unable to articulate how you achieved the solution. This kind of solution is often associated with surprise exclamations such as ‘Aha!’; STEP-
BY-STEP means that you figured out the answer after you deliberately and consciously tested out different words until you found the correct one. In this case for
instance, you are able to report the steps that you used to reach the solution.’
[SHORTENED TITLE UP TO 50 CHARACTERS] 12

Figure 1. Summary of the experimental procedure (on the left) and river plot of the results (on
the right). To the left: Participants first had to report when they were ready for each problem to
appear on the screen. They were given 15 s to solve each problem. If they found a solution,
participants had to press the YES button and then the trial with scholastic material would appear
immediately. Afterward, they had to type the solution phrase manually and report how they had
solved the problem, either via insight or via step-by-step. If participants claimed that did not
have the solution, they were asked to press the NO button, if they ran out of time. In that case,
the solution would appear on the screen followed by the scholastic fact. To the right: The river
plot of the number of responses at each step of the procedure. Node one represents if participants
claimed they had a solution; node two of the YES branch represents the accuracy of those trials
where participants claimed they had a solution; on node three the solution type (i.e., insight or
step-by-step) of each accurate or inaccurate provided solution; On node four “Hit” or “Miss”
bins of the memory task for each condition.

Analytic plan. Primary analyses consisted of binomial (i.e., logistic) generalized linear mixed

models (GLMM) fitted with the lme4 library (Bates et al., 2015) in the R environment (R Core
[SHORTENED TITLE UP TO 50 CHARACTERS] 13

Team, 2018). All models contained, at minimum, memory accuracy (hit or miss, coded 1 and 0

respectively) as the outcome variable and participant as a random intercept. To test for the

significance of terms of interest, we constructed GLMMs with and without the term and used

likelihood ratio- tests (LRTs; χ2 distribution) to assess model fit improvement. Follow-up

analyses on estimated marginal means were conducted with the emmeans library (Lenth, 2021)

in line with a priori hypotheses. To assess individual model terms and estimated marginal means

for significance, we used Wald z-tests, per standard recommendations for GLMMs without

overdispersion (Bolker et al., 2009). All fitted models did not show signs of overdispersion

(dispersion ratios >= .92, all ps = 1). When applicable, we present unstandardized beta

coefficient (b) in odds-ratio (OR) form for estimated marginal means tests to facilitate the

interpretation of relative improvements of predictors on memory performance.

Results

Overall, participants indicated they had a solution to the puzzle (regardless of accuracy)

49.38% of the times (6103 observations) whereas they said they did not have a solution 50.62%

of the times (6257 observations). Trials without an attempted solution are considered the induced

insight condition (i.e., D’oh!). See Table 1 for the breakdown of problem-solving.

Table 1.
Problem Accuracy
Correct Incorrect Total
Induced Insight (D’oh!) - - 6257
Spontaneous Insight
2932 998 3930
(Aha!)
Non-Insight 1477 696 2173
[SHORTENED TITLE UP TO 50 CHARACTERS] 14

Table 1. In the table are reported the number of observations for Induced Insights, Spontaneous

Insights, and Non-Insight for problems solved correctly and incorrectly.

Participants attempted to recall 96.5% of the scholastic trials (11929 trials, 6010 after

unsolved problems, 4289 after problems solved correctly, and 1630 after problems solved

incorrectly). Failures to recall the scholastic material collapsed into failures of memorization.

Participants recalled 59.88% of scholastic facts correctly (hits 7143 trials). Out of all the hits,

50.13% were on scholastic material that was presented after induced insights (3581 trials), and

the remaining (49.87%; 3562 trials) after people declare to have found the solution to a problem

(correct and incorrect). Out of all the problems solved correctly (4289 trials), 67.38% of the

scholastic facts were recalled after a spontaneous insight (1818 trials) and 32.61% after a non-

insight solution (880 trials) (see Figure 1 and Table 2 for details).

Table 2

Unsolved Solved correctly Solved incorrectly

Induced Insight Insight Non Insight Non Total
Memory
(D’oh!) (Aha!) Insight (Aha!) Insight
Missed 2429 1034 557 466 299 4785
Hit 3581 1818 880 489 376 7144
Total 6010 2852 1437 955 675 11929
[SHORTENED TITLE UP TO 50 CHARACTERS] 15

Table 2. Table 2 shows the number of observations for scholastic facts remembered (hit) or non-

remembered (missed) presented after failing to solve a problem i.e., Induced Insight; or after

having solved a problem correctly with an insight or without an insight.

Relation between accuracy, insight, and memorization (corroboration of Danek and

Wiley’s results)

To test the main question of whether the spontaneous insight (Aha! Experience) leads to

increased memorization of the scholastic material presented after solving a problem correctly, we

constructed a GLMM predicting memory accuracy with fixed effects for reported solving

method (spontaneous insight or step-by-step) and trial number, and a three-level random effects

structure of trial nested within method, which was nested within subject. This structure allowed

for the estimation of a separate intercept and slope for each participant (multiple trials nested

within each method for each participant) and helps account for different numbers of observations

for each type of method for each participant. For this model, we excluded memory on trials in

which participants did not indicate they solved the problem. Crucially, we also included an

interaction term of method (insight vs. step-by -step) and per-trial problem accuracy (hit vs.

missed, coded as 1 or 0), which was the term of primary interest in this analysis. To account for

between-subjects differences in overall problem accuracy, we also included each participant’s

mean problem-solving accuracy as a fixed effect. Finally, we included the number of trials in

which participants indicated they did not have a solution as a fixed effect to adjust estimates for

differing numbers of available outcome observations. Addition of the method x problem

accuracy interaction resulted in significantly improved model fit, χ2(2) = 4.97 p = .025. In
[SHORTENED TITLE UP TO 50 CHARACTERS] 16

support of increased memory accuracy when solving a problem correctly using spontaneous

insight, the differences in estimated marginal means between correct vs. incorrect insight

responses was significant, OR = 1.2, 95%CI [1.01, 1.42], zwald = 2.13, p = .033, but not correct vs.

incorrect non-insight responses, OR = 0.89, 95%CI [0.72, 1.10], zwald = -1.05, p = .290. Further,

the contrast of these two differences was significant, OR = 1.34, 95%CI [1.04, 1.73], zwald = 2.26,

p = .023, in favor of an overall increased memory in relation to insight responses on correct

problems.

Figure 2
[SHORTENED TITLE UP TO 50 CHARACTERS] 17

Figure 2. On the Y axes on a scale from 50 to 70 percent how many times the scholastic fact has

been remembered, on the X axes the distribution of when the scholastic fact has been presented:

after having solved a problem correctly or incorrectly with an insight (Aha! experience) or

without an insight (Step-by-Step). All plotted values are estimated marginal means from the

GLMM, and error bars represent 95% confidence intervals adjusted for random effects. Asterixis

indicate significance difference between tests of two estimated marginal means.

Does having an insight (induced or spontaneous) predict memorization?

To test the main question of whether the spontaneous insight (Aha! Experience), induced

insight (D’oh! Experience), or non-insight solutions would lead to increased memorization of the

scholastic material presented after receiving/achieving the problem solution, we used a modified

version of the previously described GLMM predicting memory accuracy. In this model, we

included data from all trials, including those in which participants indicated they did not solve

the problem. Fixed effects for this model included type of insight (induced, spontaneous, or non-

insight/step-by-step) and trial number, and again we used a three-level random effects structure

of trial nested within method, which was nested within subject. We did not include any of the

other terms from the previous GLMM. Addition of the method fixed-effect yielded significant

model improvement, χ2(2) = 12.626 p = .001. Estimated marginal means analysis revealed that

both induced, OR = 1.22, 95%CI [1.09, 1.37], zwald = 3.57, p < .001, and spontaneous, OR = 1.14,

95%CI [1.01, 1.29], zwald = 2.16, p = .031, insights yielded increased memory performance

relative to non-insight. There was no significant difference between induced vs spontaneous

insight; OR = 1.07, 95%CI [0.91, 1.18], zwald = 1.53, p = .124.

[SHORTENED TITLE UP TO 50 CHARACTERS] 18

Figure 3

Figure 3. On the Y axes on a scale from 50 to 70 percent of how many times the scholastic fact

has been remembered, on the X axes the distribution of when the scholastic fact has been

presented: after having solved a problem correctly Step-by-Step, after a spontaneous insight

(Aha! experience) or after an induced insight (D’oh! Experience) i.e., when participants did not

solve the problem and the solution was presented to the followed by the scholastic fact ). All

plotted values are estimated marginal means from the GLMM, and error bars represent 95%
[SHORTENED TITLE UP TO 50 CHARACTERS] 19

confidence intervals adjusted for random effects. Asterixis indicate significance difference

between tests of two estimated marginal means.

Discussion

In 1917 Köhler suggested that learning may be enhanced when the solution to a problem

is comprehended suddenly thanks to a sudden insight (Köhler, 1917). It is now well known that

emotional arousal enhances memory, and thus events happening around the time of learning

affect the strength and persistence of a memory (McGaugh, 2015). As a matter of fact, in the last

two decades, several behavioral and neurological studies evidenced that people are more likely to

remember the solution to a problem when it is achieved via insight, whether it is spontaneous or

induced (Ash et al., 2012; Auble et al., 1979; Becker, Cabeza, & Kizilirmak, 2022; Danek &

Wiley, 2020; Danek et al., 2013; Kizilirmak et al., 2015, 2019; Kizilirmak, & Becker 2022). We

hypothesize that the persistence in memory of a problem solution is related to the emotional

arousal triggered by the insight. If our hypothesis was correct, we would have seen this

advantage not only for the problem solution but also for any kind of information encountered

around the moment of excitement that follows having found the solution to a problem. To this

goal, we designed an experiment where we assessed memory for scholastic material (thus

unrelated to the problem solution) presented when people were exposed to a problem solution

whether it was given to them, or they solved it via insight.

We are more likely to remember some events for a long time since an adaptive memory

system helps us ensure that the memory for these events includes information preceding and after

the specific episode: such as the street we walk down when we met the pickpocket. This system

is meant to provide us with warning signs we might have neglected along the way. The idea is
[SHORTENED TITLE UP TO 50 CHARACTERS] 20

that because of temporal proximity with an emotionally charged event (positive or negative),

some information gains relevance as a lasting memory. The neurobiological model of this long-

term memory consolidation is known as synaptic tag-and-capture (Frey and Morris, 1997). The

model is grounded on the neuroscientific finding that weak synaptic potentiation creates the

conditions for long-term memory if it is associated with a stronger synaptic potentiation that is

emotionally charged within a critical time period. Thus, the memory of the details surrounding

the emotionally charged event, which would otherwise be forgotten, shifts into a long-term

memory (reviewed by Barco et al., 2008; Frey and Morris, 1998; Redondo and Morris, 2011).

While solving a problem, or being told a solution to a problem, is not as emotionally

charged, as being pickpocketed for example, a robust body of literature shows that insight

problem-solving is associated with pleasure, reward, excitement, and activates the correspondent

neural circuitries (Danek & Wiley, 2017, 2020; Oh et al., 2020b; Salvi, 2022; Tik et al., 2018).

Such an advantage could be explained by the synaptic tag-and-capture hypothesis. This model

focuses on the cellular component of consolidation where memory encoding and initial storage

are followed by a ‘consolidation’ process that, if activated, enables traces to become stabilized

(Frey & Morris, 1997).

Our results corroborate the hypothesis that the affective component of insight enhances

memorization. As shown participants remembered better scholastic material when it was

presented after spontaneous insight, compared to step-by-step solving and after an induced

insight, but not after an inaccurate solution i.e., when the affective component of knowing the

solutions of a problem is missing. Our results can be explained by the synaptic tag-and-capture

model since the scholastic information was presented specifically right after the affective

components of insight: the Aha! and D’oh! experiences.

[SHORTENED TITLE UP TO 50 CHARACTERS] 21

Further, our experiment represents the first comparison between spontaneous and induced

insight and the effect of their emotional components (Aha! and D’oh! experiences on

memorization). Previous studies treated these two as eliciting the same insight. While they differ

in the way the problem solution is achieved, and so is the associated emotion (Aha! vs. D’oh!)

the fact that they both elicit an emotional response as a result of restructuring, they consolidate

the associated memory in comparison to the problem solutions where this feeling is missing.

Reaching a correct solution via restructuring may make the solution more memorable

because the new organization leads to a coherent and integrated representation of the problem

and solution. Our study provides more evidence that restructuring may play a part in the

privileged status of solutions in memory.

In sum our results provide a theoretical frame of earlier behavioral and neurological

studies that evidence neural circuitries involving crucial brain areas associated with memory,

emotion, and reward (Aziz-Zadeh et al., 2009; Ludmer et al., 2011; Oh et al., 2020; Salvi,

Leiker, et al., 2021; Tik et al., 2018b). Supporting the idea that the hippocampus might be more

generally implicated in the detection of associative novelty, as defined by a novel combination of

familiar items (Düzel et al., 2003; Schott et al., 2004; Davachi, 2006). Because the insight

memory advantage was seen in both induced and spontaneous insights, and novel combinations

of familiar items occurred in both (i.e., restructuring), our data suggest that the hippocampus may

be particularly sensitive to the novel meaningful relationships between familiar items. In

addition, we advanced a potential explanation of why we observe the insight memory advantage

that is rooted in a neurobiological model of this long-term memory consolidation.

Neurological evidence and future direction

[SHORTENED TITLE UP TO 50 CHARACTERS] 22

Ludmer, Dudai, and Rubin, (2011) were among the first to shed light on the neuro-

behavioral mechanisms that explain the insight memory advantage. They recorded the neural

activity of people trying to disambiguate camouflage images (similar to those used by the Gestalt

psychologists) where the underlying objects were hard to recognize, followed by brief exposures

to the un-camouflaged image (i.e., they revealed the solution), which triggered a the D’oh’

induced insight experience. When they tested participants’ memory one week later, they found

that those remembered images were pronouncedly associated with the amygdala activation,

whose activity predicted which solutions remain in long-term memory. The authors concluded

that the role of the amygdala in the study is to promote long-term memory of ‘the sudden

reorganization of internal representations’ (Ludmer at a., 2011, page 1). In another study, Zhao

and colleagues (2013) found more activity in several regions associated with memory and

emotions, including the amygdala, hippocampus, and middle frontal gyrus while participants

solved Chinese idiom riddles. It is well known that the amygdala is responsible for processing

and encoding emotions (e.g., Hamann et al., 1999; McGaugh, 2004; Phelps and LeDoux, 2005),

and its activity modulates the strength of emotional memories (Cahill and McGaugh, 1996). The

amygdala's involvement in problem-solving thought reflects the experience of insight, providing

support for the affective experience associated with the restructuring i.e., of the ‘Aha!’ or ‘D’oh!’

experience. Further evidence of the importance of memory in insight was corroborated by the

involvement of the hippocampus found in at least two more studies (Luo & Niki, 2003; Zhao et

al., 2013). The hippocampus is well known to be important for the neural manifestation of

explicit memory, and its role in memory includes the detection and encoding of novel stimuli,

associations, and contexts (Ranganath and Rainer, 2003).

[SHORTENED TITLE UP TO 50 CHARACTERS] 23

Again, in 2019 Kizilirmak and colleagues showed how people are more likely to

remember problem solutions when they are revealed to participants, inducing a D’oh!

experience, in association with the activation of the hippocampus and the amygdala. As Ludmer

and colleagues’ (2011) study points out, the amygdala activation found in these studies probably

represents the rapidly changing value of visual stimuli that are associated with a rewarding or

aversive unconditioned stimulus (Paton et al., 2006). However, the stimuli in these studies did

not have any emotional valence, nor they were paired with external rewards. Thus, is the sudden

insightful solution, which is associated with the distinct saliency of the insight, that is rewarding.

Following studies corroborated insights are associated with the reward system and dopamine

activation, therefore, rewarding (Oh et al., 2020; Salvi et al., 2015; Tik et al., 2018). Other

researchers emphasize the feelings associated with insight, or the emotional or hedonic

component of the insightful solution process (Cosmelli & Preiss, 2014; Gick & Lockhart, 1995;

Gruber, 1995; Topolinski & Reber, 2010).

The association between insight and the dopamine/reward salience network, together

with further evidence showing the activation of brain areas implicated in learning, including

limbic structures such as the hippocampus and the amygdala (Kizilirmak, 2019; Ludmer, Dudai,

& Rubin, 2011; Kizilirmak et al., 2016; Shen et al., 2018) suggests that insight problem-solving

relates to evolutionary ancient areas of the brain which are responsible for basic functions such

as reward and emotions. The positive experience of insight may have a number of practical

consequences such as, motivating future problem-solving, increasing persistence, affecting a

person’s willingness to take a risk based on the solution (Salvi & Bowden, 2019), and making

solutions more memorable (e.g., Danek, Fraps, von Müller, Grothe, & Öllinger, 2013; Danek &

Wiley, 2020).
[SHORTENED TITLE UP TO 50 CHARACTERS] 24

A recent wave of studies on the phenomenology of insight and the associated neural

markers suggests that the emotional response associated with insight may be a signal of accuracy

and be evolutionally advantageous (Danek & Salvi, 2018; Laukkonen, Webb, Salvi, Schooler, &

Tangen, 2020; Salvi, Bricolo, Bowden, Kounios, & Beeman, 2016; Salvi, 2021). Indeed, the

emotion associated with insight could be an adaptive mechanism for the reinforcement of the

exploration of new strategies when solving problems (Oh et al., 2020). Most emotions have an

adaptive function, and feelings of pleasure that accompany an insight seem to signal the probable

utility of a solution since solutions via insight are more likely to be correct and are better

remembered than those via step-by-step (Danek & Salvi, 2018; Danek & Wiley, 2020;

Laukkonen, Webb, Salvi, Schooler, & Tangen, 2020; Salvi, Bricolo, Kounios, Bowden, &

Beeman, 2016). Thus, the emotional response associated with insight may be evolutionally

advantageous (Danek & Salvi, 2018; Salvi et al., 2016; Salvi, 2021). This hypothesis would also

explain the involvement of subcortical areas responsible for alertness, reward, and emotions, but

also learning and memory, which are evolutionary and more ancient than the cortex. So, if they

are more likely to be accurate, it makes sense that they are better remembered.

In a recent review, Laukkonen et al. (2020) argue that ‘the feeling of insight is an

adaptive signal that humans use to guide their judgments about new ideas.’ Similar to the way

that fear signals danger, Aha! Moments signal accurate solutions that pop into awareness

pervasively, attracting attention and forcing us to ignore the other myriad thoughts that crowd

our train of thoughts. They named this effect the Eureka Heuristic. According to this proposal,

the intensity of the Aha! Moment provides a useful heuristic signal about the accuracy of the

idea, based on experience and existing knowledge, and which involves an interpretation of

phenomenology to guide judgments. In this study, we provided evidence in support of this

[SHORTENED TITLE UP TO 50 CHARACTERS] 25

hypothesis by showing that the emotional response associated with insight strengthens unrelated

memory. Further, our evidence supports the idea that insight problem-solving involves

subcortical areas that are evolutionarily older and responsible for processing reward, emotions,

and memory, allowing us to speculate that the phenomenology that accompanies Aha! Moments

might have an adaptive function. The Aha! serves as an indirect indicator of the accuracy or

quality of ideas and enhances memory for the ideas.

 [SHORTENED TITLE UP TO 50 CHARACTERS] 26

References

Ash, I. K., Jee, B., & Wiley, J. (2012). Investigating insight as sudden learning. Journal of Problem

Solving, 4, 150–176.

Auble, P. M., Franks, J. J., Soraci, S. A., Soraci, S. A., & Soraci, S. A. (1979). Effort toward

comprehension: Elaboration or “aha”? Memory & Cognition, 7(6), 426–434.

[Link]

Aziz-Zadeh, L., Kaplan, J. T., & Iacoboni, M. (2009). “Aha!”: The neural correlates of verbal

insight solutions. Human Brain Mapping, 30(3), 908–916.

Barco, A., de Armentia, M. L., & Alarcon, J. M. (2008). Synapse-specific stabilization of plasticity

processes: the synaptic tagging and capture hypothesis revisited 10 years

later. Neuroscience & biobehavioral reviews, 32(4), 831-851.

Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using

lme4. Journal of Statistical Software, 67(1). [Link]

Becker, M., Cabeza, R., & Kizilirmak, J. M. (2022). A cognitive neuroscience perspective on

insight as a memory process: Searching for the solution.

Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R., Stevens, M. H. H., &

White, J.-S. S. (2009). Generalized linear mixed models: A practical guide for ecology

and evolution. Trends in Ecology & Evolution, 24(3), 127–135.

[Link]

Brown, R. & Kulik, J. Flashbulb memories. Cognition 5, 73–99 (1977).

Danek, A. H., Fraps, T., Von Müller, A., Grothe, B., & Öllinger, M. (2013). Aha! experiences leave

a mark: facilitated recall of insight solutions. Psychological research, 77(5), 659-669.

 [SHORTENED TITLE UP TO 50 CHARACTERS] 27

McGaugh, J. L., Cahill, L., & Roozendaal, B. (1996). Involvement of the amygdala in memory

storage: interaction with other brain systems. Proceedings of the National Academy of

Sciences, 93(24), 13508-13514.

Danek, A. H., & Salvi, C. (2018). Moment of Truth: Why Aha! Experiences are Correct. Journal of

Creative Behavior, 54(2), 484–486. [Link]

Danek, A. H., & Wiley, J. (2017). What about false insights? Deconstructing the Aha! experience

along its multiple dimensions for correct and incorrect solutions separately. Frontiers in

Psychology, 7:2077. [Link]

Danek, A. H., & Wiley, J. (2020). What causes the insight memory advantage? Cognition,

September 2019, 104411. [Link]

Davachi L (2006) Item, context and relational episodic encoding in humans. Curr Opin Neurobiol

16:693–700

Dominowski, R. L., & Dallob, P. (1995). Insight and Problem Solving. In R. J. Sternberg & J. E.

Davidson (Eds.), The Nature of Insight (pp. 273–278). MIT Press. Retrieved from

[Link]

Dunsmoor, J. E., Murty, V. P., Davachi, L., & Phelps, E. A. (2015). Emotional learning selectively

and retroactively strengthens memories for related events. Nature, 520(7547), 345–348.

[Link]

Dunsmoor, J. E., Murty, V. P., Clewett, D., Phelps, E. A., & Davachi, L. (2022). Tag and capture:

how salient experiences target and rescue nearby events in memory. Trends in

Cognitive Sciences.

Dudai, Y. & Morris, R. G. M. in Brain, Perception and Memory: Advances in Cognitive

Neuroscience (ed. Bolhuis, J. J.) 147–162 (Oxford Univ. Press, 2000)

 [SHORTENED TITLE UP TO 50 CHARACTERS] 28

Düzel, E., Habib, R., Rotte, M., Guderian, S., Tulving, E., & Heinze, H. J. (2003). Human

hippocampal and parahippocampal activity during visual associative recognition

memory for spatial and nonspatial stimulus configurations. Journal of

Neuroscience, 23(28), 9439-9444.

Frey, U., & Morris, R. G. M. (1997). Synaptic tagging and long-term potentiation. Nature.

[Link]

Gregor, J. N. M. A. C. (2009). Categories of insight and their correlates : An exploration of

relationships among classic-type insight problems , rebus puzzles , remote associates

and esoteric analogies. Journal Of Creative Behavior, 43(4), 1966–1967.

Gruber, M. J., Gelman, B. D., & Ranganath, C. (2014). States of Curiosity Modulate Hippocampus-

Dependent Learning via the Dopaminergic Circuit. Neuron, 84, 486-496.

Hamman SB, Ely TD, Grafton ST, Kilts CD. Amygdala activity related to enhanced memory for

pleasant and aversive stimuli. Nat Neurosci. 1999;2:289–293.

Kennedy, R., Clifford, S., Burleigh, T., Waggoner, P. D., Jewell, R., & Winter, N. J. (2020). The

shape of and solutions to the MTurk quality crisis. Political Science Research

Kizilirmak, J. M., Gallisch, N., Schott, B. H., & Folta-Schoofs, K. (2021). Insight is not always the

same: differences between true, false, and induced insights in the matchstick arithmetic

task. Journal of Cognitive Psychology, 0(0), 1–18.

[Link]

Kizilirmak, J. M., Galvao Gomes da Silva, J., Imamoglu, F., & Richardson-Klavehn, A. (2015).

Generation and the subjective feeling of “aha!” are independently related to learning

from insight. Psychological Research, 80(6), 1059–1074.

[Link]
 [SHORTENED TITLE UP TO 50 CHARACTERS] 29

Kizilirmak, J. M., Schott, B. H., Thuerich, H., Sweeney-Reed, C. M., Richter, A., Folta-Schoofs, K.,

& Richardson-Klavehn, A. (2019). Learning of novel semantic relationships via sudden

comprehension is associated with a hippocampus-independent network. Consciousness

and Cognition. [Link]

Kizilirmak, J. M., Thuerich, H., Folta-Schoofs, K., & Schott, B. H. (2016). Neural correlates of

learning from induced insight: A case for reward-based episodic encoding. Frontiers in

Psychology, 7(November), 1–16. [Link]

Kizilirmak, J. M., & Becker, M. (2022). A cognitive neuroscience perspective on

insight as a memory process: encoding the solution. In L. J. Ball & F.

Valleé-Tourangeau (Eds.), Routledge International Handbook of

Creative Cognition. Preprint on PsyArXiv.

Kohler, W. (1917). Intelligenzprüfungen an Anthropoiden (The mentality of apes) (R. P. S. of

Sciences, Ed.).

Laukkonen, R. E., Webb, M. E., Salvi, C., Schooler, J. W., & Tangen, J. M. (2020). The Eureka

heuristic: Relying on insight to appraise the quality of ideas. Preprint, February, 1–44.

[Link]

Lenth, R. V. (2021). emmeans: Estimated Marginal Means, aka Least-Squares Means.

[Link]

Litman, L., Robinson, J., & Abberbock, T. (2017). [Link]: A versatile crowdsourcing data

acquisition platform for the behavioral sciences. Behavior Research Methods, 49(2),

433–442. [Link]
 [SHORTENED TITLE UP TO 50 CHARACTERS] 30

Ludmer, R., Dudai, Y., & Rubin, N. (2011). Uncovering Camouflage: Amygdala Activation Predicts

Long-Term Memory of Induced Perceptual Insight. Neuron.

[Link]

Luo, J., & Niki, K. (2003). Function of hippocampus in “insight” of problem solving.

Hippocampus, 13(3), 316–323. doi:10.1002/hipo.10069

MacGregor, J. N., & Cunningham, J. B. (2008). Rebus puzzles as insight problems. Behavior

Research Methods, 40(1), 263–268. [Link]

McGaugh, J. L. (2004). The amygdala modulates the consolidation of memories of emotionally

arousing experiences. Annu. Rev. Neurosci., 27, 1-28.

McGaugh, J. L. (2015). Consolidating memories. Annual Review of Psychology, 66, 1–24.

[Link]

Murty, V. P., & Adcock, R. A. (2014). Enriched encoding: reward motivation organizes cortical

networks for hippocampal detection of unexpected events. Cerebral Cortex, 24(8),

2160-2168.

Murphy, C., Dehmelt, V., Yonelinas, A. P., Ranganath, C., & Gruber, M. J. (2021). Temporal

proximity to the elicitation of curiosity is key for enhancing memory for incidental

information. Learning & Memory, 28(2), 34-39.

Oh, Y., Chesebrough, C., Erickson, B., Zhang, F., & Kounios, J. (2020). An insight-related neural

reward signal. NeuroImage, 116757. [Link]

Osgood, C. E. (1953). Method and theory in experimental psychology. New York: Oxford

University Press.

Phelps, E. A., & LeDoux, J. E. (2005). Contributions of the amygdala to emotion processing: from

animal models to human behavior. Neuron, 48(2), 175-187.

 [SHORTENED TITLE UP TO 50 CHARACTERS] 31

Redondo, R.L., and Morris, R.G.M. (2011). Making memories last: the synaptic tagging and capture

hypothesis. Nat. Rev. Neurosci. 12, 17–30.

Salvi, C., Markers of insight. (2021). Routledge International Handbook of Creative Cognition.

Salvi, C., Bricolo, E., Bowden, E., Kounios, J. & Beeman, M. (2016). Insight solutions are correct

more often than those achieved by analysis. Thinking and Reasoning, 6783(February).

[Link]

Salvi, C., Bricolo, E., Franconeri, S. L., Kounios, J., & Beeman, M. (2015). Sudden insight is

associated with shutting out visual inputs. Psychonomic Bulletin & Review, 22(6),

1814–1819. [Link]

Salvi, C., Costantini, G., Bricolo, E., Perugini, M., & Beeman, M. (2015). Validation of Italian

rebus puzzles and compound remote associate problems. Behavior Research Methods.

[Link]

Salvi, C., Costantini, G., Pace, A., & Palmiero, M. (2018). Validation of the Italian Remote

Associate Test. The Journal of Creative Behavior. [Link]

Salvi, C., Leiker, E. K., Baricca, B., Molinari, M. A., Eleopra, R., Nichelli, P. F., Grafman, J., &

Dunsmoor, J. E. (2021). The Effect of Dopaminergic Replacement Therapy on Creative

Thinking and Insight Problem-Solving in Parkinson’s Disease Patients. 12(March), 1–

15. [Link]

Schott, B. H., Richardson-Klavehn, A., Henson, R. N., Becker, C., Heinze, H. J., & Düzel, E.

(2006). Neuroanatomical dissociation of encoding processes related to priming and

explicit memory. Journal of Neuroscience, 26(3), 792-800.

Shohamy, D., & Adcock, R. A. (2010). Dopamine and adaptive memory. Trends in cognitive

sciences, 14(10), 464-472.

 [SHORTENED TITLE UP TO 50 CHARACTERS] 32

Shen, W., Gu, H., Ball, L. J., Yuan, Y., Yu, C., Shi, R., & Huang, T. (2020). The impact of

advertising creativity, warning-based appeals, and green dispositions on the attentional

effectiveness of environmental advertisements. Journal of Cleaner Production, 271,

122618. [Link]

Shen, W., Yuan, Y., Liu, C., & Luo, J. (2016). In search of the “Aha!” experience: Elucidating the

emotionality of insight problem-solving. British Journal of Psychology, 107(2), 281–

298.

Shohamy, D., & Adcock, R. A. (2010). Dopamine and adaptive memory. Trends in cognitive

sciences, 14(10), 464-472.

R Core Team. (2018). R: A language and environment for statistical computing. R Foundation for

Statistical Computing. [Link]

Squire, L. R. Memory, and the hippocampus: a synthesis from findings with rats, monkeys, and

humans. Psychol. Rev. 99, 195–231 (1992).

Threadgold, E., Marsh, J. E., & Ball, L. J. (2018). Normative data for 84 UK English rebus puzzles.

Frontiers in Psychology, 99–124. [Link]

Tik, M., Sladky, R., Luft, C. D. B., Willinger, D., Hoffmann, A., Banissy, M. J., Bhattacharya, J., &

Windischberger, C. (2018a). Ultra-high-field fMRI insights on insight: Neural

correlates of the Aha!-moment. Human Brain Mapping, 39(8), 3241–3252.

[Link]

Zhao, Q., Zhou, Z., Xu, H., Chen, S., Xu, F., Fan, W., & Han, L. (2013). Dynamic Neural Network

of Insight: A Functional Magnetic Resonance Imaging Study on Solving Chinese

“Chengyu” Riddles. PLoS ONE, 8(3). [Link]

Woodworth, R. S., & Schlosberg, H. (1954). Experimental psychology. Oxford and IBH Publishing.
[SHORTENED TITLE UP TO 50 CHARACTERS] 33
[SHORTENED TITLE UP TO 50 CHARACTERS] 34

Supplementary Material