Preprint submitted to Neuropsychologia December 8, 2016
1 of 26
New Insights into Insight: Neurophysiological Correlates of the Difference
Between the Intrinsic “Aha” and the Extrinsic “Oh Yes” Moment
Katrin Rothmalera,*, Roland Nigburb, Galina Ivanovac,d,e a Department of Computer Sciences, Humboldt-Universität zu Berlin, Unter den Linden 6, 10099 Berlin, Germany.
b Department of Psychology, Otto-von-Guericke-University Magdeburg, Postfach 4120, 39106 Magdeburg, Germany. c Department of Psychology, Humboldt-Universität zu Berlin, Unter den Linden 6, 10099 Berlin, Germany.
d Institute for Applied Informatics at Leipzig University, Hainstraße 11, 04109 Leipzig, Germany. e Information Systems Institute, Leipzig University, Grimmaische Straße 12, 04109 Leipzig, Germany.
___________________________________________________________________________
Abstract
Insight refers to a situation in which a problem solver immediately changes his understanding
of a problem situation. This representational change can either be triggered by external
stimuli, like a hint or the solution itself, or by internal solution attempts. In the present paper,
the differences and similarities between these two phenomena, namely “extrinsic” and
“intrinsic” insight, are examined. To this end, electroencephalogram (EEG) is recorded while
subjects either recognize or generate solutions to German verbal compound remote associate
problems (CRA). Based on previous studies, we compare the alpha power prior to insightful
solution recognition with the alpha power prior to insightful solution generation. Results
show that intrinsic insights are preceded by an increase in alpha power at right parietal
electrodes, while extrinsic insights are preceded by a respective decrease. These results can
be interpreted in two ways. In consistency with other studies, the increase in alpha power
before intrinsic insights can be interpreted as an increased internal focus of attention.
Accordingly, the decrease in alpha power before extrinsic insights may be associated with a
more externally oriented focus of attention. Alternatively, the increase in alpha power prior to
intrinsic insights can be interpreted as an active inhibition of solution-related information,
while the alpha power decrease prior to extrinsic insights may reflect its activation.
Regardless of the interpretation, the results provide strong evidence that extrinsic and
intrinsic insight differ on the behavioral as well as the neurophysiological level.
Keywords: problem solving; solution recognition; EEG; alpha power; right hemisphere; insight
___________________________________________________________________________
*Corresponding author, email: [email protected]
Rothmaler et al.: New Insights Into Insight
2 of 26
1 Introduction
Throughout history, some of the most important scientific achievements were accomplished
by a sudden flash of inspiration that instantly changed the discoverer’s understanding of the
problem situation. The discovery of Archimedes’ principle is a popular example of such an
insight experience. According to the anecdote, Archimedes of Syracuse was asked to develop
a non-destructive testing method for the royal crown. After a few unsuccessful attempts, he
finally found the solution while taking a bath. Various storytellers claim that Archimedes got
in the tub and saw the water level rise, which immediately triggered the answer to his
problem. However, it is also possible that taking the bath was simply relaxing him, enabling
his mind to draw connections where he never expected them. Unfortunately, we cannot ask
Archimedes whether the rising water level or the total relaxation elicited his “aha” moment.
Nevertheless, we can raise the question if it would have made any difference. We can even
go one step further and ask: would it have been the same if someone had told Archimedes the
solution?
The central question of the present study is whether there is any difference between an active
insightful solution generation, referred to as intrinsic insight, and a more passive solution
recognition, referred to as extrinsic insight. According to Wertheimer (1945), extrinsic
insight is unlikely to produce the restructuring necessary for real insightful understanding.
Thereby, restructuring designates “the process of arriving at a new understanding of the
problem situation” (Dominowski and Dallob, 1995, p. 50). Behavioral studies support
Wertheimer’s hypothesis by showing that participants who find solutions to insight problems
on their own exhibit a far better recall rate than subjects confronted with the correct solution
after failing to solve a problem (cf. Dominowski and Buyer, 2000). In addition, Metuki et al.
(2012) demonstrate that transcranial direct current stimulation (tDCS) over the left
dorsolateral prefrontal cortex significantly improves solution recognition of difficult verbal
insight problems but not their solution generation. Electrophysiological evidence reviewed by
Dietrich and Kanso (2010) suggests that even two opposing ERP results can be traced back to
a confusion of intrinsic and extrinsic insight: While intrinsic insight is associated with a
positive event-related potential after stimulus onset (P200-600) over the superior temporal
gyrus (Qiu et al., 2008), extrinsic insight is related to a negative one (N320) (Qiu et al.,
2006). These results already indicate that extrinsic and intrinsic insight differ on the
behavioral as well as the neurophysiological level. Nevertheless, various neuroscientific
Rothmaler et al.: New Insights Into Insight
3 of 26
insight studies assume that the presentation of a solution or a solution hint results in the same
“aha” moment as an internally generated solution attempt (cf. Luo et al., 2004; Luo and Niki,
2003; Mai et al., 2004; Qiu et al., 2006; Shen et al., 2013). Although several authors already
questioned this assumption (for reviews, Bowden et al., 2005; Kounios and Beeman, 2014;
Luo and Knoblich, 2007), so far no neuroscientific study has investigated this essential issue.
This paper will close this gap by examining the differences between the intrinsic “aha” and
the extrinsic “oh yes” moment on a neurophysiological level.
1.1 Operationalization of Insight
One definition of insight is given by Dominowski and Dallob (1995, p. 33) who describe it as
a “form of understanding (of a problem and its solution) that can result from restructuring, a
change in a person’s perception of a problem situation”. This definition covers both: the
active insightful solution generation and the more passive insightful solution recognition. In
the following, the former will be called intrinsic insight, while the latter will be referred to as
extrinsic insight. Another way to define insight is to distinguish it from alternative problem
solving strategies. Typically, one contrasts insight and analysis. Analytic problem solving is
characterized by a methodological and strategic processing towards the solution (cf. Wegbreit
et al., 2012). Analytic problem solvers gradually approach the solution and are well aware of
their solution path (cf. Metcalfe and Wiebe, 1987). In contrast, insightful solutions arise
suddenly and unpredictably (cf. Metcalfe and Wiebe, 1987), they often involve a mental
impasse (cf. Duncker, 1945) and the inability to report the processing that led to the solution
(cf. Maier, 1931; Schooler et al., 1993).
For several years, it was common practice to use two distinct classes of tasks to study
insightful and analytic problem solving: insight and non-insight (or analytic) problems (cf.
Dominowski and Dallob, 1995). A popular example of an insight problem is Duncker’s
candle-box-task. The task is to attach three candles side by side at eye level on a door. To do
so, the problem solver has three small pasteboard boxes containing small candles, tacks and
matches, respectively. The solution consists of emptying the boxes and tacking them to the
door so that they can be used as a platform for the candles (Duncker, 1945). On the other
hand, the Wason selection task is a famous example of an analytic problem. This task
consists of four cards, each of which has a letter on one side and a number on the other. The
problem solver sees just one side of the cards. These visible faces show „A“, „B“, „4“ and
„7“. He has to decide which cards he needs to turn to check the following rule: If a card has
Rothmaler et al.: New Insights Into Insight
4 of 26
an „A“ on one side, then it must have a „4“ on the other side. The solution is to turn the card
showing “A” and the card showing “4” (Wason, 1966).
With the use of two mutually exclusive classes of problems, it was implicitly assumed that
problems of a particular class can only be solved with the associated problem solving
strategy. However, this assumption does not hold true for all insight problems. Consider, for
instance, the following brain-teaser: “If you have black socks and brown socks in a drawer,
mixed in a ratio of 4 to 5, how many socks will you have to take out to make sure that you
have a pair of the same color?“ (Bowden et al., 2005, p. 323). Bowden et al. (2005) argue that
this insight problem can also be solved with a “What if”-strategy: “What if I take out a black
sock then a brown sock? I would only need one more sock of either color to have a pair of the
same color.“ (Bowden et al., 2005, p. 323). This observation led to a new experimental
paradigm: instead of seeking exclusive insight problems, Bowden and Jung-Beeman
deliberately utilized tasks that could be solved by either insight or analysis (1998). To
separate solutions with and without insight, they used the subjective experience of their
participants. Thus, subjects were asked to rate on a trial-by-trial basis whether they achieved
their answer via insight or analysis (cf. Bowden, 1997). This insight judgment procedure has
been successfully applied in various following studies (for review, Kounios and Beeman,
2014) and it is especially valuable for neuroimaging studies that require an adequate
reference state for analysis (for reviews, Bowden et al., 2005; Kounios and Beeman, 2009).
Contrary to the scientific consensus regarding solution generation, there is still an ongoing
debate whether solution recognition can either be achieved via insight or analysis. While
some researchers doubt the existence of an insightful solution understanding (cf. Wertheimer,
1945), others deny the existence of its analytic comprehension (cf. Luo and Niki, 2003; Mai
et al., 2004; Qiu et al., 2006; Shen et al., 2013). However, Bowden and Jung-Beeman (2003a)
successfully demonstrated that subjects can recognize solutions with and without “aha”
moment. Therefore, we will use the subjective ratings of the participants in the present study
to distinguish these different kinds of understanding.
1.2 State of the Art and Hypotheses
As mentioned previously, intrinsic insight involves restructuring, a change in the initial
problem representation. This definition implies that the initial understanding of the problem
situation is invalid, probably misdirected by past experiences, context or familiarity (cf.
Rothmaler et al.: New Insights Into Insight
5 of 26
Dominowski and Dallob, 1995). To solve a problem via insight, the problem solver needs to
think outside the box and establish new and innovative associative or semantic relations
(Jung-Beeman et al., 2004). In their neurological model of intrinsic insight, Bowden and
colleagues (2005) suggest that the initial processing of a problem already activates these
remote associations. Yet, this activation is weak and remains unconscious. The behavioral
experiment of Bowden (1997) provides evidence for such an unconscious pre-activation of
solution-related information in connection with intrinsic insight. In his study, participants
were asked to solve a series of anagrams. Prior to some anagrams, either the solution, a
semantically related or an unrelated word was presented. This prior hint was presented either
too brief to be detected, too brief to be identified (i.e. unreportable) or long enough to be
reported. After each solution, the subjects rated their subjective experience of insight on a 10-
point scale. Since these insight ratings are not interpretable for reportable hints, they were
excluded from the analysis. For undetectable hints, all hints lead to reliably higher ratings of
insight than without hint presentation. Thereby, no differences between the hint types can be
observed. The authors conclude that the mere presentation of a stimulus (even if it is
completely uninformative) attracts attention and elicits a preparedness response that
manifests itself in the subjects’ insight ratings. In contrast, a reliable effect of hint type on
insight rating is found for unreportable hints. As expected, the presentation of a hint results in
significantly higher insight ratings than without hint presentation. Furthermore, solution hints
lead to marginally higher insight ratings than semantically related hints, which, in turn, lead
to significantly higher ratings than unrelated hints. These results indicate that the
unreportable processing of solution-related concepts is crucial for insight experiences.
Moreover, the unconscious processing of information seems to be associated with insight
experiences even if the information is completely unrelated to the solution.
Bowden and colleagues (2005) further localize the weak activation of solution-related
information within the right hemisphere. This hypothesis is supported by the work of Fiore
and Schooler (1998) who demonstrate that hints to classical insight problems are more
valuable when presented to the left visual hemifield than to the right hemifield. In addition,
Bowden and Jung-Beeman (1998) reveal a connection between solution priming, solution
recognition and the right hemisphere. They use semantic association problems that demand a
single word as their solution and give the participants 15s to solve them. Afterwards, a
lateralized target word is presented that is either the correct solution or a completely
unrelated word. Subjects are asked to name the target word or to decide if it represents the
Rothmaler et al.: New Insights Into Insight
6 of 26
correct solution. Participants name solution words faster than unrelated words. This solution
priming is greater for solutions that are presented to the left visual field than for solutions that
appear in the right hemifield. If subjects fail to solve the problem, the solution priming is
even limited to the left visual field. In addition, participants recognize solutions to unsolved
problems faster if they are presented at the left visual hemifield. These results are consistent
with previous findings that associate the right hemisphere with a coarse semantic coding (cf.
Beeman, 1993; Chiarello et al., 1990). This coarse semantic coding produces large and weak
semantic fields comprising a variety of information including concepts that are only distantly
related to the input word and context (cf. Beeman, 1993), which makes it especially
important for insight solutions (cf. Jung-Beeman et al., 2004).
According to Bowden and colleagues (2005) the weak activation of non-obvious but
solution-relevant information in the right hemisphere is overshadowed by the strong
activation of obvious, problem-related concepts that do not lead to a solution. To solve the
problem, the solver needs to “switch the focus of processing to the unconscious activation”
(Bowden et al., 2005, p. 324). Hence, insightful problem solving is associated with an
increased internal focus of attention (cf. Jung-Beeman et al., 2004; Salvi et al., 2015). With
their eye movement study, Salvi et al. (2015) provide direct evidence for such a connection
between internally oriented attention and intrinsic insight. In their experiment, subjects
solved compound remote associate problems (a special kind of semantic association task)
and, on a trial-by-trial basis, indicated whether they achieved the solution via insight or
analysis. Immediately prior to insight solutions, subjects blink longer and look away from the
problem more frequently than prior to analytic solutions. Thus, they shift their attention away
from the visual stimuli towards internal processing. Salvi et al. (2015) used an experimental
paradigm similar to Jung-Beeman et al. (2004) who recorded high density EEGs while
subjects solved compound remote associate problems. Relative to analysis, the authors find
more alpha band activity around 9.8Hz at right parietal regions for insight from -1.4 to -0.4s
before the solution response. An increase in alpha power is traditionally believed to reflect
cortical idling or cortical inhibition (for reviews, Pfurtscheller, 1999; Pfurtscheller et al.,
1996). Consequently, enhanced alpha power over the parietal-occipital cortex indicates idling
or inhibition of the visual cortex, wherefore Jung-Beeman et al. interpret their alpha effect as
a selective gating of visual inputs to the right hemisphere that allows weaker processing
about more distant associations to gain strength (cf. Jung-Beeman et al., 2004). Taken
together, intrinsic insights are preceded by an increase in alpha power over right parietal
Rothmaler et al.: New Insights Into Insight
7 of 26
regions, they are associated with an increased internal focus of attention and involve a weak
and unconscious pre-activation of solution-related information within the right hemisphere.
We assume that extrinsic insights initially involve the same processing as intrinsic insights: a
weak activation of solution-related concepts combined with a strong activation of obvious but
misleading problem-related information. This hypothesis is supported by a study of Bowden
and Jung-Beeman (2003a) revealing a connection between extrinsic insight, the right
hemisphere and solution priming. The authors used a similar experimental paradigm to the
one in a previous study (1998): Their participants worked on a verbal problem, they were
confronted with a target word that was either the solution or an unrelated word, they named
the target word and decided whether it represented the correct solution. Additionally, they
rated if they recognized the solution with or without “aha” moment. As expected, following
unsolved problems, the participants exhibited more solution priming for solutions that they
later recognized with insight than for solutions that they recognized without it. This
association was even stronger for solutions that were presented to the left visual field.
Consequently, both extrinsic and intrinsic insight seem to involve an unconscious pre-
activation of solution-related information within the right hemisphere.
However, while intrinsic insight involves an attention shift away from the visual stimuli
towards internal processing, extrinsic insight may require the exact opposite. In terms of
CRAs, solution generation differs from solution recognition, because the problem solver is
confronted with the solution word itself and just needs to draw the right connections between
the problem words and the solution. Assuming that these connections are already weakly
activated for insightful solution recognition, there is no need for a shift of attention. Instead,
the problem solver shall maintain a rather externally oriented attention towards the visual
stimulus, i.e. the solution.
For analytic solution recognition, no evidence for such a specific, unconscious pre-activation
of solution-related information within the right hemisphere is found. To be more precise,
participants showed less solution priming for analytic than for insightful solution recognition
and there was no difference between the hemispheres (cf. Bowden and Jung-Beeman, 2003a).
Consequently, participants need to engage in a more active search to find out how the
solution word forms a compound with each of the three problem words. We assume that this
active search for associations results in an inhibition of visual input reflecting an increased
internal focus of attention. Accordingly, we expect more alpha power over right parietal
Rothmaler et al.: New Insights Into Insight
8 of 26
regions prior to analytic than prior to insightful solution recognition. In other words, we
presume that intrinsic and extrinsic insight have the opposed effect on alpha power.
Consequently, a cross interaction between modality, i.e. solution generation or recognition,
and problem solving strategy, i.e. insight or analysis, is anticipated.
2 Materials and Methods
To explore the differences and similarities between extrinsic and intrinsic insight
systematically, the experimental paradigm of Jung-Beeman et al. (2004) was modified. As in
their study from 2004, subjects worked on verbal puzzles, so-called compound remote
associate problems (CRA). They were asked to respond with a bimanual keystroke as soon as
they found a possible solution and to indicate whether they solved the problem with or
without “aha” moment. Unlike Jung-Beeman et al. (2004), we gave participants 20s instead
of 30s to solve a problem. A pilot study showed that approximately 50% of the problems are
solved within this time limit. This was important for our study, because subjects were not
only asked to actively solve a problem but also to passively recognize its solution if they
exceeded the time limit. Hence, if a participant failed to respond within 20s, the problem
words disappeared and the solution word was presented. To assure comparability (all CRAs
were solvable) and due to the limited number of trials, all solution words were accurate.
However, participants were not aware of their correctness. They were instructed to respond
with a bimanual keystroke only if (and at the same time as soon as) they understood the
displayed solution and to specify whether they felt an “oh yes” experience when recognizing
it. That way, time segments prior to these keystrokes could be analyzed and the insight-
specific activity of solution generation and recognition could be contrasted.
2.1 Subjects
Twenty-four right-handed, healthy native German speakers participated voluntarily in the
experiment. The eleven female and thirteen male subjects were aged between 19 and 28 and
received 8 euros per hour as expense allowance. Prior to the experiment, all participants
completed a screening questionnaire such that any neurological or psychological disorders, a
history of brain damage, substance abuse and medication could be ruled out. All experiments
were carried out in the morning and had a duration of approximately 90 minutes plus
preparation and follow-up. The study was approved by the ethics committee and all subjects
signed an informed consent form.
Rothmaler et al.: New Insights Into Insight
9 of 26
2.2 Stimulus Material
In our experiment, native German speakers were confronted with a German version of the
compound remote associate problems introduced by Bowden and Jung-Beeman (2003b) who
adapted them from Mednick’s Remote Associates Test (c.f. Mednick, 1962). They consist of
three different problem words, for instance “age”, “mile” and “sand”, to which a fourth
solution word must be found. This solution word (in the example, “stone”) needs to form a
compound with each problem word (i.e. “stone age”, “milestone” and “sandstone”). CRA
problems are especially well suited to study insight experiences with neuroscientific methods
for a number of reasons. First of all, CRA problems can be solved with and without insight
such that the non-insight condition can be used as a baseline to extract insight-specific neural
activity. Since they are simpler than classical insight problems, a better control of confound
variables can be provided. For the same reason, they can be solved quickly such that a large
number of problems can be presented. Finally, they are physically compact which enables the
presentation on a computer screen and minimizes eye movement artifacts (cf. Bowden and
Jung-Beeman, 2003b).
To obtain a set of German compound remote associate problems, the English riddles were
first translated. If possible, this translation was directly used as a German CRA. If not, the
translation was adjusted whenever it seemed reasonable or replaced with a newly invented
German compound remote associate problem. Since German word compositions often
involve so-called joint elements1, some additional restrictions were made. Firstly, the German
word compositions should always consist of two autonomous words. Secondly, the subjects
should always be able to attach the solution word prior or post the problem word without
making any adjustments. Within these restrictions, 108 German CRA problems were
generated. Thereby, all solution words were unique. That means, they solved only a single
CRA problem and they were never used as a problem word.
2.3 Procedure
Initially, all subjects completed a training session with five German CRA problems.
Afterwards, questions concerning the experimental procedure were clarified and the 103 1For instance, the English compound “safety pin” equals the German compound “Sicherheitsnadel” that consists of the nouns “Sicherheit” and “Nadel” and the joint element “s”.
Rothmaler et al.: New Insights Into Insight
10 of 26
remaining problems were presented in a random order in three consecutive blocks. Each
experimental trial started with the question “ready?” shown in the middle of a computer
screen. With an arbitrary keystroke a fixation cross appeared that was replaced after 500ms
by a German compound remote associate problem. The subjects had 20 seconds to solve each
problem and they were instructed to respond with a bimanual button press as soon as they
found a possible solution. With this keystroke, the three problem words disappeared and a
fixation cross was presented for five seconds. Afterwards, the participants had to decide
whether they solved the problem with or without “aha” moment on a three step scale, and
they were invited to pronounce their solution word. Finally, they were supposed to specify
how confident they were that their solution was correct ranging from “certain” to “fairly” and
“uncertain”.
Figure 1: Diagram showing the experiment procedure. The course of each trial depends on whether or not
a solution is generated within 20 seconds (first crossing point marked with a gray circle) and
whether or not a solution is recognized within 5 seconds (second crossing point).
If a problem could not be solved within 20 seconds, the word “timeout” appeared on the
screen for 1.5 seconds. Then, the solution word was shown and the subjects had to respond
with a bimanual button press as soon as they understood the presented solution. If they
reacted within five seconds, a fixation cross appeared for five seconds. Afterwards they had
to decide whether they recognized the solution with or without an “oh yes” experience on a
three step scale and the next trial started. If they failed to understand the presented solution
Rothmaler et al.: New Insights Into Insight
11 of 26
within the five second time limit, the words “second timeout” appeared for 1.5 seconds
before the next trial started automatically. Figure 1 illustrates the whole experimental
procedure.
In compliance with Bowden and Jung-Beeman (2003b), all words were presented at the
center of a black screen in normal horizontal orientation in yellow (RGB colors: 255,255,0)
14 point Arial font in order to minimize eye movements. Moreover, all words that appeared
on the screen were capitalized to assure comparability2. Prior to the experiment, the
participants were familiarized with the differentiation between insightful and analytic
solution generation and recognition. To avoid any influences by the investigator, the
description length or other confounding variables, the instructions were presented in a written
form on the screen. They were based on the definitions used by Jung-Beeman et al. (2004)
and by Bowden and Jung-Beeman (2003b). Afterwards, the experiment participants were
asked to briefly recapitulate their own definition of intrinsic and extrinsic insight.
2.4 Exclusion Criteria
First of all, trials that included a premature verbalization of the solution, that showed reaction
times less than 100ms or that involved any other disturbances3 were removed from analysis.
Second, subjects that were unable to correctly classify the different problem solving and
solution recognition strategies were excluded. To this end, the participants’ descriptions of
insightful and analytical solution generation and recognition were compared with theoretical
criterions of insight and analysis. These theoretical criterions were derived from the
descriptions that Jung-Beeman et al. (2004) and Bowden and Jung-Beeman (2003a) used for
their insight judgment procedure. They addressed three different aspects: the nature of the
solution path (unawareness and suddenness versus awareness and strategic thinking),
confidence (obviousness versus the need to mentally check the solution) and affect (feeling of
“aha”). Subjects whose descriptions contradicted any theoretical criterion of insight or
2 This was necessary because some problem words were adverbs or adjectives that are usually written in lower case letters, while others were nouns that start with a capital letter in German.
3 For instance, interactions with the investigator.
Rothmaler et al.: New Insights Into Insight
12 of 26
analysis, that named not at least one or that completely misunderstood the classification task4
were not included in any further analysis.
2.5 Behavioral Data Analysis
Initially, the response frequency was examined, i.e. the percentage of solutions derived via
insight or analysis, the percentage of solutions recognized with or without insight, the
proportion of correct and confident solutions, the amount of timeouts and the overall error
rate. Afterwards, differences in reaction time between insight and analysis were investigated.
Since each participant typically exhibited a different number of insightful and analytic trials,
the data was post-experimentally counter-balanced to assure comparability. That means, for
the predominant category, as many trials were chosen pseudorandomly as the less frequent
category provided. To receive sufficiently good estimates, a minimal number of five trials
was specified. Subjects who did not fulfill this criterion were excluded from the analysis of
reaction times (for an overview of the number of trials provided by each subject that was
included into analysis, see table A.1 in the appendix). For all other participants, the mean
response time of insightful and analytic solution generation and recognition was calculated.
With these individual means, two paired t-tests with significance levels of α = 0.05 were
performed: One examined differences in reaction time between insightful and analytic
solution generation, while the other one contrasted insightful and analytic solution
recognition. The assumption of normality was controlled by a Lilliefors test with a
significance level of α = 0.05.
2.6 EEG Analyses
63 electrodes were placed according to the extended 10-20 system and recorded with the
bridged mastoids as reference. Impedance was kept below 5kΩ. Moreover, the
electrooculogram was conducted consisting of two electrodes at the left and right side of the
eyes and two electrodes above and below the dominant eye. EEG analyses were performed
with BrainVision Analyzer 2 and Matlab, while some functions of EEGLAB were utilized for
the topographic mappings. The data was bandpass filtered with a passband from 0.1 Hz to
100 Hz and an initial automatic raw data inspection was performed. These preprocessing
steps ensured that neither slow drifts nor severe artifacts would affect the independent
4 For instance, some subjects confounded a lack of understanding with analytic solution recognition.
Rothmaler et al.: New Insights Into Insight
13 of 26
component analysis that was carried out to correct eye movement artifacts. This ICA was
followed by a highpass filtering to 1 Hz and a second raw data inspection with more
conservative criteria that assured that no smaller artifacts contaminated the data. After this
preprocessing, breaks between the different trials were eliminated and the alpha band power
(8-13 Hz) was calculated using a continuous wavelet transform with a complex Morlet
wavelet.
To test whether intrinsic and extrinsic insights were preceded by opposed alpha effects, time
segments were extracted that started 2s prior to the insightful or the analytic generation or
recognition of a solution and ended 200ms after it. Thereby, trials with reaction times of less
than 2s were excluded. As an effect over right parietal sides was expected, electrodes P6, P8
and PO8 were identified as region of interest (ROI). If a time segment contained any artifact
within this ROI it was excluded from analysis. By visual inspection, a relevant time window
prior to solution generation and recognition was determined. As for the behavioral analyses,
the insightful and analytic trials were counter-balanced and a minimal number of five trials
was specified. Counter-balancing was particularly important for the present analyses, because
it assured comparable signal-to-noise ratios. For every subject that provided enough trials, the
insightful and analytic solution generation and recognition segments were separately
averaged and the mean within the specified time window and ROI was calculated. Since
power values are usually not normally distributed, the data was log-transformed to assure
normality. To assess modality-specific differences between insight and analysis, two paired t-
tests with significance levels of α = 0.05 were performed. Since less alpha band power was
expected for insightful than for analytic solution recognition, a left-tailed t-test was carried
out for this modality. In contrast, a right-tailed t-test was performed for solution generation,
because more alpha band power was anticipated for insightful than for analytic solution
generation. To test the interaction between problem solving strategy and modality, a repeated
measures analysis of variances was performed that included the factor “strategy” with the
levels “insight” and “analysis” and the factor “modality” with the levels “generation” and
“recognition”. For this particular analysis, the number of trials was counterbalanced not only
for analysis and insight but also for solution generation and recognition. As a matter of
course, only those subjects were included for whom enough trials were available in all four
conditions. As before, a significance level of α = 0.05 was chosen for all tests performed.
Rothmaler et al.: New Insights Into Insight
14 of 26
3 Results
3.1 Exclusion Criteria
For intrinsic insight, all subjects were able to correctly classify insightful and analytic trials.
Only with respect to the (unnecessary) reexamination of an insight solution, some minor
deviations could be noticed: Three test persons expressed the compulsive need to internally
check their solution, while three other participants stated that they sporadically reexamined
their solution word. This possible response delay needs to be considered in the interpretation
of the results and also in the EEG signal analyses. Nevertheless, no subject had to be
excluded from analysis. For extrinsic insight, however, discrepancies between the subjects’
and the theoretical definition of analytical comprehension were detected. Five test persons
stated that they classified trials as analytical if they could not remember all three problem
words or if they did not understand the presented solution. Thus, they confounded a lack of
understanding with analytic solution recognition and had to be excluded from further
analyses.
3.2 Behavioral Data
In total, participants solved 51.2% of all the presented CRA problems. They classified 42.4%
of their solutions as insightful, 38.5% of their solutions as analytic and 19.1% as neither of
both. 19 of the 24 subjects performed the subsequent recognition task correctly. For problems
they failed to solve, these participants recognized 84.7% of the solutions. They classified
46.9% of their recognitions as insightful, 38.2% of their recognitions as analytic and 14.8%
as neither of both. The majority of the verbalized solutions were correct (87.9%), with an
even better accuracy for insightful solutions only (92.2%). Moreover, a connection between
insightful problem solving and confidence could be found: in 84.7% subjects were certain
that their insight solution was accurate, whereas for analytic solutions only 66.1% could be
achieved. This self-evaluation was appropriate since 96.1% of the confident solutions were
correct.
Rothmaler et al.: New Insights Into Insight
15 of 26
Figure 2: Boxplots of the response times averaged over trials for insightful and analytic solution generation
(left) and recognition (right).
Obviously, solution generation and recognition differ with respect to their response times.
While the solution to 84.7% of the unsolved problems was recognized within five seconds,
just 51.2% of all puzzles could be solved within a time limit of 20 seconds. This difference is
neither astonishing nor surprising. However, the examination of reaction times also revealed
a commonality. Figure 2 shows the distribution of the grand means of the response times for
insightful and analytic solution generation (left graphic) and recognition (right graphic),
while table 1 contains the corresponding sample means. As can be seen, insightful responses
are on average faster than analytic responses for both solution generation and recognition. To
be more precise, subjects needed on average 6.67s to solve a problem via insight, whereas
9.44s were required for analysis. For insightful solution recognition, on average 2.34s were
sufficient, while 3.29s were spent for analytic comprehension.
Grand Mean
Insight Analysis t n df SD p Generation 6.67s 9.44s -7.07 22 21 1.84 0.0057 Recognition 2.34s 3.29s -6.77 19 18 0.61 0.0241
Table 1: grand means of the response times for insightful and analytic solution generation and recognition
along with the p-values of the corresponding t-tests, t-statistics, degrees of freedom, standard deviations
and n, the numbers of subjects included into analysis.
This observation is supported by the corresponding paired t-tests that became significant with
t(21)=-7.07, p=0.0057 for solution generation and with t(18)=-6.77 p=0.0241 for solution
recognition, respectively. A normal distribution of the difference variables could be assumed
since the corresponding Lilliefors tests revealed high p-values of 0.5. For solution generation,
Insight AnalysisGeneration
0
5
10
15
20
Resp
onse
Tim
e in
ms
Insight AnalysisRecognition
0
1
2
3
4
5
Resp
onse
Tim
e in
ms
Rothmaler et al.: New Insights Into Insight
16 of 26
22 subjects provided the minimal number of trials, while 19 subjects fulfilled this
requirement for solution recognition. Table 1 summarizes the analyses of reaction times.
3.3 EEG Results
Like Jung-Beeman et al. (2004), we found more right parietal alpha band activity prior to
solutions that were derived via insight than prior to solutions that were solved by analysis.
This alpha effect lasted from approximately -2000ms to -1500ms before the solution. The
right graphic of figure 3 shows a topographic mapping of the alpha power difference (insight
minus analysis) within this time window, while the left graphic depicts the time course of the
mean alpha power within the specified region of interest.
Figure 3: left: topographic mapping of the alpha band power difference (insight - analytic) from -2000ms to -
1500ms prior to solution generation; right: alpha band power prior to solution generation averaged over
electrodes PO8, P8 and P6, trials and subjects for insightful (red) and analytic (blue) trials. The
associated standard errors are shaded in the corresponding color.
A right-tailed paired t-test confirmed the significance of this intrinsic insight effect with
t(21)=1.96, p=0.0318 (for more details, see table 2). To assure that this effect was no time
confound due to the different speeding of response times for insightful and analytic solutions,
we performed a second analysis that included only solutions with response times longer than
7s. For this subset of solutions, no differences in reaction time could be determined (t(16)=-
1.19, p=0.2506, α=0.05), whereas we still found significantly more alpha power for insightful
than for analytic solutions within the specified time window and ROI (t(16)=1.92, p=0.0362,
α=0.05; for more details, see table A.2 in the appendix).
Rothmaler et al.: New Insights Into Insight
17 of 26
Grand Mean
Insight Analysis t n df SD p Generation 16.22µV2 13.50µV2 1.96 22 21 0.26 0.0318 Recognition 16.04µV2 22.76µV2 -2.42 15 14 0.40 0.0149
Table 2: grand means of the alpha power for insightful and analytic solution generation and recognition along
with the p-values of the corresponding t-tests, t-statistics, degrees of freedom, standard deviations and
n, the numbers of subjects included into analysis.
As anticipated, the opposite effect was observed for extrinsic insight. That is, less alpha
power was found prior to insightful than prior to analytic solution recognition within the
specified time interval and ROI (t(14)=-2.42, p=0.0149). Analogous to figure 3, figure 4
illustrates the topography of the alpha power difference (left subplot) and the alpha power
time course (right subplot) for solution recognition, while table 2 includes the corresponding
means, p-values, t-statistics, standard deviations, degrees of freedom and sample sizes.
Figure 4: left: topographic mapping of the alpha band power difference (insight - analytic) from -2000ms to -
1500ms prior to solution recognition; right: alpha band power prior to solution recognition averaged
over electrodes PO8, P8 and P6, trials and subjects for insightful (red) and analytic (blue) trials. The
associated standard errors are shaded in the corresponding color.
As for solution generation, this effect might be biased by the differences in reaction time.
Since the exclusion of fast recognitions could not balance the speed of responses, we cannot
replicate our findings for a subset of equally speeded recognitions. However, if the
differences in response time biased our results, there would be a significant difference
between fast and slow solution recognitions. To study this possibility in detail, we contrasted
the alpha power of immediate (response within 3s) and delayed solution recognitions
(response after 3s) balancing, of course, the number of insightful and analytic trials. For this
particular analysis, no minimal number of trials was set to maintain at least 13 subjects for
Rothmaler et al.: New Insights Into Insight
18 of 26
analysis. We find no significant difference in alpha power between immediate and delayed
responses within the specified time window and ROI (t(12)=-1.31, p=0.2164 α=0.05, for
more details see table A.3 in the appendix). Due to the limited number of trials and subjects,
we further performed a repeated measures ANOVA that included strategy, i.e. insight or
analysis, as an additional factor. While neither the main effect of the factor reaction time, i.e.
immediate or delayed, nor the interaction between the factors becomes significant (F=2.24,
p=0.1600 and F=0.03, p=0.8576, respectively), the analysis reveals a significant main effect
of the factor strategy (F=18.10, p=0.0011). Hence, we are able to replicate our extrinsic
insight effect while factoring out the influence of the speed of responses (for more details, see
table A.4 and figure A.1 in the appendix). These results indicate that the difference in alpha
power prior to analytic and insightful solution recognition cannot be explained by the
differences in reaction time.
Figure 5: alpha power means of analytic (gray) and
insightful (black) solution generation (left
side) and recognition (right side) along with
their average (dotted line).
Source SS F p n Strategy 0.0002 0.0026 0.9598 15 Modality 0.1318 1.1265 0.3065 15 Interaction 0.7599 6.0622 0.0274 15
Table 3: Results of the 2-way repeated-measures
ANOVA: degrees of freedom (df) and means
of squares (MS) are not listed, because for
factors with just two levels df=1 and, with
this, SS=MS holds.
Furthermore, a two-way analysis of variance with repeated measures revealed a significant
interaction between the two factors “strategy” and “modality”, while none of the main effects
became significant. Table 3 comprises the corresponding p-values, sums of squares, F-
statistics and n, the number of subjects included into analysis, while figure 5 demonstrates the
associated alpha power means. Thereby, the mean analytic alpha power is depicted in gray,
while the mean insightful alpha power is colored in black. The average of the analytic and
insightful means is illustrated as a dotted line. Apparently, insight positively influences the
Generation RecognitionModality
16
18
20
22
24
Alp
ha P
ower
in 7
V2
Analysis Insight Average
Rothmaler et al.: New Insights Into Insight
19 of 26
alpha power for solution generation, while it affects the alpha power for solution recognition
negatively. As illustrated by the dotted line, these opposite effects cancel each other out,
wherefore no significant main effect was obtained.
In addition to these findings, figure 3 and figure 4 show a crossover of alpha power for the
intrinsic (from approximately -800ms to -900ms prior to solution generation) and for the
extrinsic alpha effect (starting about 1000ms prior to solution recognition). Yet, an analysis
of the respective time intervals yielded no significant results.
4 Discussion
As repeatedly shown, compound remote associate problems can be solved with and without
insight. The present study replicates this result and demonstrates further prove that CRAs can
also be recognized in either an analytic or an insightful way. Moreover, the behavioral data
reveals significant differences with respect to analytic and insightful response times. For both
insightful solution generation and recognition, shorter mean reaction times are obtained. This
observation is in line with three other studies that reported shorter mean (or median) reaction
times for insightful problem solving than for analysis (cf. Kounios et al., 2006; Salvi et al.,
2015; Subramaniam et al., 2009). Unlike these authors, Jung-Beeman et al. (2004) cannot
report any differences in response time. As neither of the studies provide any information
about the reaction time distribution, this discrepancy might be a result of the longer time
period subjects had to solve each problem (30s instead of the 20s time limit in the present
study and the 15s time limit in the studies cited above). In conclusion, the observed
differences in reaction times provide additional evidence that insight and analysis represent,
in fact, two distinctive problem-solving strategies. As insight has a negative effect on
response times for both solution generation and recognition, the results emphasize,
furthermore, that intrinsic and extrinsic insight are subtypes of the same cognitive
phenomenon. Meanwhile, the different scaling of the extrinsic and intrinsic reaction times
already indicates that differences between the two insight subtypes exist.
This presumption is supported by electrophysiological findings. In accordance with Jung-
Beeman and colleagues (2004), we find significantly more alpha band activity prior to
insightful than prior to analytic solution generation at right parietal electrodes. While these
authors localize this intrinsic insight effect in a time interval from -1310ms to -560ms relative
to solution, we identify a corresponding effect a bit earlier, i.e. from -2000ms to -1500ms.
Rothmaler et al.: New Insights Into Insight
20 of 26
This slight temporal shift might be the result of individual differences or response delays. As
mentioned in section 3.1., three of our subjects indicated after the experiment that they
always felt the need to mentally check their solutions, while two other participants admitted
to sporadically review their solution words. As hypothesized, we observe the opposed alpha
effect for extrinsic insight. That is, we find significantly less right parietal alpha power for
insight than for analysis from approximately -2000ms to -1500ms relative to solution
recognition. For both the intrinsic and the extrinsic insight effect, no indicators for a time
confound due to the different speeding of response times were found. Moreover, a repeated
measures ANOVA confirmed the hypothesis that intrinsic and extrinsic insight have an
opposite effect on alpha power with a significant cross interaction between problem solving
strategy, i.e. insight or analysis, and modality, i.e. solution generation or recognition.
In line with the results of Bowden and Jung-Beeman (2003a), we believe that both intrinsic
and extrinsic insight involve the same weak and unconscious pre-activation of solution-
related information within the right hemisphere. As pointed out by Jung-Beeman et al.
(2004), the insight-specific alpha power increase prior to solution generation may reflect a
selective gating of visual information that enables an increased focus towards internal
processing. This internal focus of attention strengthens the weak activation of non-obvious,
solution-related concepts that were overshadowed by the strong activation of obvious but
misleading problem-related information and allows them to emerge into consciousness.
Following this train of thought, the insight-specific decrease in alpha power prior to solution
recognition may reflect a rather externally oriented focus of attention towards the solution
word itself. Since the correct solution is presented and the information that is crucial for its
understanding is already weakly activated, no shift of attention towards internal processing is
necessary. In contrast, prior research does not suggest a comparable pre-activation of
solution-related information for analytic solution recognition. Thus, an individual needs to
engage in a more active search to find the right connection between the solution and the
problem words. This active search might be accompanied by an increased internal focus of
attention and the inhibition of visual information that is reflected in the enhanced right
parietal alpha band activity prior to analytic solution recognition. Thus, pursuing the
interpretation of Jung-Beeman et al. (2004), the opposed alpha effects for intrinsic and
extrinsic insight may suggest that the two phenomena are associated with opposed foci of
attention.
Rothmaler et al.: New Insights Into Insight
21 of 26
However, an alternative interpretation is possible. While an increase in alpha power is
traditionally believed to reflect cortical deactivation or cortical idling (for reviews,
Pfurtscheller, 1999; Pfurtscheller et al., 1996), evidence accumulates that alpha
synchronization may also reflect top-down inhibitory control processes (for review, Klimesch
et al., 2007). As Klimesch et al. (2007 p. 63) point out: “ERS is elicited in situations, where
subjects withhold or control the execution of a response and is obtained over sites that
probably are under, or exert top-down control”. Consequently, the increase in right parietal
alpha power prior to intrinsic insights may also reflect the active inhibition of solution-related
information. This interpretation is in line with the theory that an ill-defined problem space is
a crucial component of insight experiences (for review, Knoblich et al., 1999). According to
the theory, an individual automatically generates an internal representation of a problem
situation that is biased by context, familiarity and past experience. This initial representation
activates potentially useful knowledge elements that implicitly define a problem space within
which an appropriate solution is sought. For insight solutions, this initial problem space does
not contain the correct solution. As a consequence, the problem solver encounters an
impasse. This impasse can be overcome by changing the initial problem representation and,
herewith, the defined problem space. Once the problem space contains the correct solution,
the problem can be solved easily resulting in the subjective insight experience (cf. Knoblich
et al., 1999). Thus, in terms of the problem space theory, the right parietal alpha power
increase prior to intrinsic insights may be interpreted as an active inhibition of knowledge
elements that lie outside of the (still ill-defined) problem space. Accordingly, the decrease in
right parietal alpha power prior to extrinsic insight might represent the activation of
knowledge elements that were previously suppressed. This interpretation is supported by the
localization of the effects because the right hemisphere performs the coarse semantic coding
that is required to solve or understand a compound remote associate problem (cf. Beeman,
1993; Chiarello et al., 1990).
To determine if the opposed alpha effects reflect opposed inhibitory processes and/or
opposed foci of attention, further research is required. For instance, an eye movement study
could provide direct evidence for a relationship between extrinsic insight and an increased
external focus of attention (cf. Salvi et al., 2015), while the analysis of functional couplings
between parieto-occipital and prefrontal areas could provide further insights into the
involvement of top-down processes (cf. Sauseng et al., 2005).
Rothmaler et al.: New Insights Into Insight
22 of 26
5 Conclusion
This study is the first to investigate the differences and similarities between intrinsic and
extrinsic insight on a behavioral as well as a neurophysiological level. As a matter of fact,
intrinsic and extrinsic insight share some general characteristics. On the one hand, they both
lead to shorter mean reaction times than their analytic counterparts. On the other hand, a
neural correlate within the alpha frequency range can be determined for both that coincides in
terms of timing and localization. These findings indicate that the intrinsic “aha” and the
extrinsic “oh yes” moment are, in fact, two subtypes of insight experiences rather than two
completely independent cognitive phenomena. However, the disparate scaling of the extrinsic
and intrinsic reaction times and the opposite directions of the aforementioned alpha effects
also suggest that both subtypes differ substantially. While intrinsic insights are preceded by
an increase in alpha band power over right parietal regions, extrinsic insights are preceded by
a right parietal alpha power decrease. These opposed alpha effects may reflect opposed foci
of attention or, alternatively, opposed inhibitory processes. Regardless of the interpretation of
the effects, our results provide strong evidence that intrinsic and extrinsic insights differ on
the behavioral as well as the neurophysiological level. Thus, they have crucial implications
for the design of prospective insight experiments and the interpretation of recent research.
Instead of treating the intrinsic “aha” and the extrinsic “oh yes” as interchangeable, they
should be taken for what they are: two distinguishable subtypes of insight experiences that
share some characteristics but that need to be investigated separately. In conclusion,
Archimedes’ principle would have been uncovered either way: if someone had told him the
solution or if he had discovered it all on his own. Archimedes’ brain, however, would
probably have known the difference.
Rothmaler et al.: New Insights Into Insight
23 of 26
Appendix
Solution Generation Solution Recognition
Reaction Time Alpha Power Reaction Time Alpha Power Alpha Power
Subject I A t-test I A t-test I A t-test I A t-test ANOVA 1 31 28 28 29 28 28 17 16 16 6 11 6 6 2 13 18 13 6 14 6 11 20 11 5 7 5 5 3 14 23 14 14 23 14 19 19 19 12 16 12 12 5 36 14 14 34 13 13 11 19 11 8 18 8 8 6 14 16 14 14 15 14 25 16 16 12 15 12 12 8 18 23 18 18 23 18 15 13 13 14 13 13 13 9 21 34 21 21 34 21 - - - - - - 21
10 25 19 19 25 19 19 19 7 7 - - - 19 11 30 6 6 30 6 6 28 13 13 16 10 10 6 12 18 23 18 17 15 15 27 19 19 21 14 14 14 13 11 12 11 11 12 11 31 19 19 15 12 12 11 14 13 23 13 13 19 13 23 18 18 - - - 13 15 13 15 13 12 15 12 40 14 14 5 12 5 5 16 32 14 14 31 13 13 - - - - - - 13 17 24 26 24 23 26 23 19 18 18 6 17 6 6 18 33 24 24 32 24 24 - - - - - - 24 19 24 33 24 22 29 22 6 35 6 - - - 22 20 28 15 15 26 14 14 29 7 7 12 7 7 7 21 29 16 16 28 16 16 6 25 6 - - - 16 22 26 22 22 25 21 21 34 5 5 15 5 5 5 23 12 28 12 12 25 12 8 13 8 5 11 5 5 24 20 18 18 19 18 18 21 21 21 14 16 14 14
Table A.1: Overview of the number of trials provided by each subject included into analysis for the different
conditions and tests that were carried out. The abbreviation “I” stands for insight and “A” for analysis.
Grand Mean
Insight Analysis t n df SD p Reaction Time 11.21s 11.86s -1.19 17 16 2.23 0.2506 Alpha Power 14.93µV2 11.23µV2 1.92 17 16 0.43 0.0362
Table A.2: grand means of the reaction times and the alpha power for insightful and analytic solutions with
responses longer than 7s along with the p-values of the corresponding t-tests, t-statistics, degrees of
freedom, standard deviations and n, the numbers of subjects included into analysis.
Rothmaler et al.: New Insights Into Insight
24 of 26
Grand Mean
Immediate Delayed t n df SD p Alpha Power 17.60µV2 22.40µV2 -1.31 13 12 0.45 0.2164
Table A.3: grand means of the alpha power for immediate (reaction time shorter than 3s) und delayed
solution recognitions (reaction time longer than 3s) along with the p-values of the corresponding t-test,
the t-statistic, the degree of freedom, the standard deviation and n, the number of subjects.
Source SS F p n Strategy 3.7475 18.104 0.0011 13 Modality 0.4792 2.2438 0.1600 13 Interaction 0.0048 0.0336 0.8576 13
Table A.4: grand means of the alpha power for immediate (reaction time shorter than 3s) und delayed
solution recognitions (reaction time longer than 3s) along with the p-values of the corresponding t-test,
the t-statistic, the degree of freedom, the standard deviation and n, the number of subjects.
Figure A.1: alpha power means of analytic (gray) and insightful (black) immediate (left side) and delayed
(right side) solution recognition along with their average (dotted line).
Acknowledgments
The authors would like to thank the Department of Psychology at the Humboldt-Universität
zu Berlin for their great support.
Immediate DelayedReaction Time
15
20
25
30
Alp
ha P
ower
in 7
V2
Analysis Insight Average
Rothmaler et al.: New Insights Into Insight
25 of 26
References
Beeman, M., 1993. Semantic processing in the right hemisphere may contribute to drawing inferences from discourse. Brain Lang. 44, 80–120. doi:10.1006/brln.1993.1006 Bowden, E.M., 1997. The effect of reportable and unreportable hints on anagram aolution and the Aha! Experience. Conscious. Cogn. 6, 545–573. doi:10.1006/ccog.1997.0325 Bowden, E.M., Jung-Beeman, M., 2003a. Aha! Insight experience correlates with solution activation in the right hemisphere. Psychon. Bull. Rev. 10, 730–737. Bowden, E.M., Jung-Beeman, M., 2003b. Normative data for 144 compound remote associate problems. Behav. Res. Methods Instrum. Comput. 35, 634–639. Bowden, E.M., Jung-Beeman, M., 1998. Getting the right idea: semantic activation in the right hemisphere may help solve insight problems. Psychol. Sci. 9, 435–440. doi:10.1111/1467-9280.00082 Bowden, E.M., Jung-Beeman, M., Fleck, J., Kounios, J., 2005. New approaches to demystifying insight. Trends Cogn. Sci. 9, 322–328. doi:10.1016/j.tics.2005.05.012 Chiarello, C., Burgess, C., Richards, L., Pollock, A., 1990. Semantic and associative priming in the cerebral hemispheres: some words do, some words don’t... sometimes, some places. Brain Lang. 38, 75–104. Dietrich, A., Kanso, R., 2010. A review of EEG, ERP, and neuroimaging studies of creativity and insight. Psychol. Bull. 136, 822–848. doi:10.1037/a0019749 Dominowski, R.L., Buyer, L.S., 2000. Retention of problem solutions: The re-solution effect. Am. J. Psychol. 113, 249–274. Dominowski, R.L., Dallob, P., 1995. Insight and Problem Solving, in: The Nature of Insight, 1. A Bradford Book, Cambridge, London, pp. 33–62. Duncker, K., 1945. On Problem Solving, 5th ed, Psychological Monographs. The American Psychological Association. Fiore, S.M., Schooler, J.W., 1998. Right hemisphere contributions to creative problem solving: Converging evidence for divergent thinking. Right Hemisphere Lang. Comprehension Perspect. Cogn. Neurosci. 349–371. Jung-Beeman, M., Bowden, E.M., Haberman, J., Frymiare, J.L., Arambel-Liu, S., Greenblatt, R., Reber, P.J., Kounios, J., 2004. Neural Activity When People Solve Verbal Problems with Insight. PLoS Biol. 2, 500–510. doi:10.1371/journal.pbio.0020097 Klimesch, W., Sauseng, P., Hanslmayr, S., Gruber, W., Freunberger, R., 2007. Event-related phase reorganization may explain evoked neural dynamics. Neurosci. Biobehav. Rev. 31, 1003–1016. doi:10.1016/j.neubiorev.2007.03.005 Knoblich, G., Ohlsson, S., Haider, H., Rhenius, D., 1999. Constraint Relaxation and Chunk Decomposition in Insight Problem Solving. J. Exp. Psychol. Learn. Mem. Cogn. 25, 1534–1555. Kounios, J., Beeman, M., 2014. The cognitive neuroscience of insight. Annu. Rev. Psychol. 65, 71–93. doi:10.1146/annurev-psych-010213-115154 Kounios, J., Beeman, M., 2009. The Aha! Moment: The cognitive neuroscience of insight. Curr. Dir. Psychol. Sci. 18, 210–216. doi:10.1111/j.1467-8721.2009.01638.x Kounios, J., Frymiare, J.L., Bowden, E.M., Fleck, J.I., Subramaniam, K., Parrish, T.B., Jung-Beeman, M., 2006. The prepared mind: Neural activity prior to problem presentation predicts subsequent solution by sudden insight. Psychol. Sci. 17, 882–890. doi:10.1111/j.1467-9280.2006.01798.x Luo, J., Knoblich, G., 2007. Studying insight problem solving with neuroscientific methods. Methods 42, 77–86. doi:10.1016/j.ymeth.2006.12.005 Luo, J., Niki, K., 2003. Function of hippocampus in “insight” of problem solving. Hippocampus 13, 316–323. doi:10.1002/hipo.10069
Rothmaler et al.: New Insights Into Insight
26 of 26
Luo, J., Niki, K., Phillips, S., 2004. Neural correlates of the “Aha! reaction.” NeuroReport 15, 2013–2017. Maier, N.R.F., 1931. Reasoning in humans. II. The solution of a problem and its appearance in consciousness. J. Comp. Psychol. 12, 181–194. doi:10.1037/h0071361 Mai, X.-Q., Luo, J., Wu, J.-H., Luo, Y.-J., 2004. “Aha!” effects in a guessing riddle task: An event-related potential study. Hum. Brain Mapp. 22, 261–270. doi:10.1002/hbm.20030 Mednick, M.T., 1962. Research creativity in psychology graduate students. J. Consult. Psychol. 27, 265–266. Metcalfe, J., Wiebe, D., 1987. Intuition in insight and noninsight problem solving. Mem. Cognit. 15, 238–246. Metuki, N., Sela, T., Lavidor, M., 2012. Enhancing cognitive control components of insight problems solving by anodal tDCS of the left dorsolateral prefrontal cortex. Brain Stimulat., Human Brain Stimulation in Cognitive Neuroscience 5, 110–115. doi:10.1016/j.brs.2012.03.002 Pfurtscheller, G., 1999. Event-related desynchronization, 6th ed, Handbook of Electroencephalography and Clinical Neurophysiology. Elsevier, Amsterdam. Pfurtscheller, G., Stancák Jr, A., Neuper, C., 1996. Event-related synchronization (ERS) in the alpha band—an electrophysiological correlate of cortical idling: a review. Int. J. Psychophysiol. 24, 39–46. Qiu, J., Li, H., Yang, D., Luo, Y., Li, Y., Wu, Z., Zhang, Q., 2008. The neural basis of insight problem solving: An event-related potential study. Brain Cogn. 68, 100–106. doi:10.1016/j.bandc.2008.03.004 Qiu, J., Luo, J., Wu, Z., Zhang, Q., 2006. A further study of ERP effects of “insight” in a riddle guessing task. Acta Psychol. Sin. 38, 507–514. Salvi, C., Bricolo, E., Franconeri, S.L., Kounios, J., Beeman, M., 2015. Sudden insight is associated with shutting out visual inputs. Psychon. Bull. Rev. 1–6. doi:10.3758/s13423-015-0845-0 Sauseng, P., Klimesch, W., Doppelmayr, M., Pecherstorfer, T., Freunberger, R., Hanslmayr, S., 2005. EEG alpha synchronization and functional coupling during top-down processing in a working memory task. Hum. Brain Mapp. 26, 148–155. doi:10.1002/hbm.20150 Schooler, J.W., Ohlsson, S., Brooks, K., 1993. Thoughts beyond words: When language overshadows insight. J. Exp. Psychol. Gen. 122, 166. Shen, W., Liu, C., Zhang, X., Zhao, X., Zhang, J., Yuan, Y., Chen, Y., 2013. Right hemispheric dominance of creative insight: An event-related potential study. Creat. Res. J. 25, 48–58. doi:10.1080/10400419.2013.752195 Subramaniam, K., Kounios, J., Parrish, T.B., Jung-Beeman, M., 2009. A brain mechanism for facilitation of insight by positive affect. J. Cogn. Neurosci. 21, 415–432. Wason, P.C., 1966. Reasoning, in: Foss, B.M. (Ed.), New Horizons in Psychology. Penguin Books, Harmondsworth, pp. 135–151. Wegbreit, E., Suzuki, S., Grabowecky, M., Kounios, J., Beeman, M., 2012. Visual attention modulates insight versus analytic solving of verbal problems. J. Probl. Solving 4, 94–115. doi:10.7771/1932-6246.1127 Wertheimer, M., 1945. Productive thinking. Chicago University Press.