How do we know when our ideas are true? - OSF

HOW DO WE KNOW WHEN OUR IDEAS ARE TRUE?

How do we know when our ideas are true? The Eureka Heuristic and The Insight Fallacy

Ruben E. Laukkonen The University of Queensland

Jonathan W. Schooler University of California, Santa Barbara

Jason M. Tangen The University of Queensland

Direct correspondence to: Ruben Laukkonen, [email protected]

Laukkonen, Schooler, & Tangen, 2018

1


Abstract Here we connect several fields including insight problem solving, metacognition, and heuristics and biases, and propose that the Aha! phenomenology is an intuitive shortcut for evaluating the ‘truthiness’ of ideas and solutions to problems. The feeling that often accompanies an insight moment has received comparatively less attention than the problem solving processes that precede an insight moment, and in some cases the Aha! moment has been explicitly described as an epiphenomenon. Our account assumes instead that the Aha! experience has an adaptive purpose, and serves as a signal regarding the veracity of an idea that pops into mind unexpectedly. Research has consistently shown that insight experiences promote greater confidence—often despite no deliberate verification or conscious reflection—and can lead to more accurate solutions to problems. The Eureka Heuristic provides an account of why Aha! moments occur in certain contexts but not others, it describes a mechanism for why the Aha! experience predicts confidence and objective performance, and also allows for predictions about situations where false insights are likely to occur. We also propose the insight fallacy to describe situations where a person incorrectly concludes that a solution or idea must be true solely based on the fact that the idea or solution was accompanied by an Aha! moment.


2


Introduction “Eureka!, I know the answer!” said Archimedes. “But how do you know?!” questioned Socrates. “Well, because I experienced a Eureka moment,” he replied.

In the fictitious exchange above, Archimedes has an idea that occurred to him suddenly, and he experienced a Eureka moment. When asked to justify his sudden confidence in a solution he had only just discovered, he recursively points back to the Eureka moment itself. While popular accounts of Eureka moments and Aha! experiences are rarely this self-aware, it is possible to detect hints of a similar deductive process in everyday decision making. In 2012, while waiting to go to a concert, mathematician Yitang Zhang discovered the solution to the twin prime problem. He said that he “…immediately knew that it would work,” and then it took several months to verify his solution (Nisbett, 2015). The 19th century polymath Henri Poincaré was immediately certain about the accuracy of a solution that came to him when stepping onto an omnibus, and mathematician Jacques Hadamard said that, “on being very abruptly awakened by an external noise, a solution long searched for appeared to me at once without the slightest instant of reflection on my part.”

The sudden certainty that accompanies Aha! moments—without any deliberate verification—is also regularly observed in the laboratory (Danek et al., 2014; Danek & Wiley, 2017; Kounios & Beeman, 2014; Bowden & Jung-Beeman, 2003; Kounios et al., 2006; Webb et al., 2016). Another classic feature of insight moments is that they arise in consciousness unexpectedly, and participants often cannot predict when an insight moment is going to occur (Metcalfe, 1986; Metcalfe & Wiebe, 1987; Laukkonen & Tangen, 2017). In a similar vein, popular anecdotes as well as a recent analysis of over 1000 respondents from the general population (Ovington et al., 2015), suggest that Eureka moments can strike unexpectedly while one is engaged in an unrelated task (Ovington et al., 2015; Laukkonen, Schooler, & Tangen, 2018

3


Snyder et al., 2004; Nisbett, 2015). Since the solution can pop into mind unexpectedly during problem solving, and the problem solver can be engaged in another task, the sudden solution was a product of processing that occurred below awareness (Bowden, 1997; Hedne et al., 2016; Smith & Kounios, 1996; Kounios & Beeman, 2014). Because the problem solving process may be hidden from awareness, when an Aha! moment occurs, the problem solver may not know ‘how’ or ‘why’ they know, and many months or even years may be required to verify the solutions (Sternberg & Davidson, 1995; Topolinski & Reber, 2010; Nisbett, 2015). These facts together raise fundamental questions about how solutions to problems are settled on, and why some solutions or ideas are suddenly accompanied by powerful phenomenology and confidence, often despite no deliberate verification by the problem solver.

The fact that people are largely unaware of the problem solving processes that underlie their final solution is also corroborated by numerous experiments (discussed in more detail later). Maier (1931) famously found that participants suddenly discovered a solution to his ‘two-string’ problem shortly after he provided a hint that the participants claimed not to notice (an effect replicated by, Landrum, 1990). Bowden (1997) found that subliminally priming the solution to an anagram lead to more reported insight experiences, without participants knowing that they were being primed. Hattori, Sloman, and Orita (2013) also found that subliminal primes improved insight problem solving across three experiments, in some cases leading to a fivefold improvement. Since problem solving processes, and therefore the reasoning behind the solution, are often hidden from awareness, what can be the source of such an immediate sense of certainty? Perhaps it is the sudden Aha! phenomenology itself that compels humans to trust that the present solution—the only solution that evoked such explosive affect—is the one they have been searching for.


4


Implicit processing that precedes an insight solution may draw on a vast and complex network of information, experiences, and beliefs that are difficult and inefficient to consciously appraise. Time pressures also discourage deliberate evaluation. When a person suddenly finds a solution, they may not have the luxury of weighing up its components before acting. Therefore it is often less important to notice ‘why you know,’ and is abundantly more efficient to use affect to signal that a viable solution has been found (Goldstein & Gigerenzer, 2002). It is plausible that the Aha! phenomenology serves as a signal that a promising solution has been discovered through unconscious processes, and that humans often use the Aha! experience as a fast and frugal substitute for an effortful and time-consuming review of the evidence.

In most problem solving contexts so far investigated, a solution accompanied by an Aha! moment is more likely to be accurate compared to solutions not accompanied by an Aha! moment (Metcalfe 1986; Metcalfe & Wiebe, 1987; Danek et al., 2014; Salvi et al., 2016; Danek & Wiley, 2017; Webb, Little, & Cropper; 2016; Hedne et al., 2016), and the more intense the Aha! moment, the more likely the solution is to be accurate (Webb et al., 2016). Therefore, the sudden increase in confidence associated with Aha! moments appears to be justified. But how does the Aha! phenomenology predict accuracy?

It is common knowledge that an expert can have intuitive expertise about their domain, for example the next best move on a chess board (Ericsson & Charness, 1994; Kahneman, 2011). Since humans do not have privileged access to problem solving processes, it seems plausible that they also have intuitions about the accuracy of their ideas. Expert intuitions are often fast and feel automatic, and so too an intuition about an idea might occur almost immediately as the idea ‘pops into mind’. It is possible to imagine that the Aha! phenomenology signals a ‘consistency’ or ‘coherence’ with what one knows and believes. Thus in the same way that the chess expert draws on their expertise to make a move—often without any conscious effort—the problem solver can Laukkonen, Schooler, & Tangen, 2018

5


evaluate a solution automatically and intuitively based on their own expertise regarding the problem domain. This mechanism may explain why there is a positive relationship between the Aha! experience and accuracy. As long as a person’s knowledge and experiences are reliable, then the greater the consistency between the solution and one’s memories and experiences, the more likely the solution will be accurate. The precise mechanisms for how the Aha! moment interacts with existing knowledge and memory is unclear, in the same way that it is not known precisely how any emotion or intuition draws on a person’s knowledge or experiences, but they certainly do so (Kahneman, 2011; Klein, 1999). A simple schematic of the Eureka Heuristic model is provided in Figure 1.

1

Problem solving begins with an accumulation of information, experiences, and beliefs, which form implicit knowledge structures.

2

Information is integrated below awareness, and new patterns and associations are formed that contribute to problem solving.

Implicit processes lead to a solution that is cognised suddenly and unexpectedly accompanied by an Aha! moment.

3 Aha! phenomenology is partly determined by the extent to which implicit knowledge structures cohere with the solution.

4

The intensity of the Aha! phenomenology is interpreted such that greater intensity results in greater judged truth of the solution.

Figure 1. The four steps in the Eureka Heuristic view of the Aha! moment.

The heuristic view also helps to explain why Aha! experiences can range from being barely noticeable to intense, and why they can occur in a such a wide variety of circumstances. Mathematicians such as Henry Poincaré and Yitang Laukkonen, Schooler, & Tangen, 2018

6


Zhang, as well as physicists Albert Einstein and Richard Feynman, report intense Eureka experiences which resulted in—as Poincaré put it—“immediate certainty.” Their considerable expertise which was developed over decades of study and practice resonated loudly with the sudden solution. Any problem solving situation where there is some uncertainty—and as long as the preceding process occurs below awareness—may be accompanied by an Aha! moment (an intuition about the solution that appears in mind). Although apparently less common, it also happens that one can have a Eureka experience where the contents of the solution are untrue (Danek & Wiley, 2017). If a person’s memory or knowledge is impoverished then false insights are bound to arise. On the other hand, if a person’s memory, conceptual structures, and assimilated knowledge are based on decades of quality experience and clear and replicable evidence (and a healthy state of mind), then the Aha! moment can signal a breakthrough discovery.

We begin by describing the definitions and phenomenology of insight in the context of problem solving and briefly consider the role of fluency in the Aha! experience. We then outline the connection to metacognition and confabulation, and then judgment and decision-making in order to specify the Eureka experience as an ecologically rational heuristic akin to affect and fluency (Goldstein & Gigerenzer, 2002). We also consider potential trade-offs involved with the Eureka heuristic, and describe the insight fallacy as any situation where one concludes that an unexpected idea or solution which come to mind must be true, solely on the grounds that an Aha! moment accompanied the solution. In the final section, we discuss how the Eureka heuristic model of the Aha! moment may contribute to a number of long-standing debates in the insight problem solving literature, and provide a framework for understanding recent empirical findings. Given the breadth of research discussed, it is inevitable that some of the richness of each area is lost. We see that part of the value of this contribution is in describing overlapping literatures that are otherwise isolated.


7


The Phenomenology of Insight One popular, historical definition of insight is a solution that appears in consciousness suddenly and unexpectedly during problem solving (Ohlsson, 1984; Sternberg & Davidson, 1995; Metcalfe & Wiebe, 1987). Öllinger and Knoblich (2009) also made a useful distinction between three dimensions of insight: a task dimension, a process dimension, and a phenomenological dimension (see also Chu, 2009). In some cases, researchers have been unclear whether they were interested in insight as a feature of some subset of problems (the task dimension), a particular cognitive process (the process dimension), or the feeling experienced when a solution was discovered (the phenomenological dimension). The different approaches meant that authors occasionally spoke past one another—and sometimes disagreed—partly owing to the fact that they were discussing different aspects of a broader construct. Regarding the task dimension of insight, there is now good reason to believe that the insight phenomenology is not specific to certain kinds of problems, at least where a participant’s own reports are valued. For instance, subjects report that the same problem can be solved with insight in some cases, and some cases not, and nearly all problems can occasionally be solved by insight (Weisberg & Alba, 1981; Webb et al., 2016; Salvi et al., 2016; Laukkonen & Tangen, 2017; Danek et al., 2017). Nevertheless, it is important to understand why some problems result in more insights than others, and how those problems are solved, which has been the focus of research on the task dimension of insight (see for example, Fleck & Weisberg, 2013; Chein et al., 2010; Weisberg & Alba, 1981).

A researcher interested in the process dimension examines the cognitive process that leads to an insight moment. The process dimension has been the predominant area of research for the past 50 years, and most researchers agree that solving classical insight problems involves a process of restructuring or representational change that occurs below awareness (Ohlsson, 1984; 1992; 2011; Knoblich et al., 1999; Knoblich et al., 2001; Kounios & Beeman, 2014; Laukkonen, Schooler, & Tangen, 2018

8


Öllinger & Knoblich, 2009; Schooler, Ohlsson, & Brooks, 1993; Schooler & Melcher, 1995). According to this view, insight problems tend to result in sudden solutions because they are initially represented incorrectly—wrong assumptions are made. To solve the problem, a new representation or perspective must be discovered, and when it is, the solution is immediately obvious, resulting in a phenomenologically sudden solution. For example, try to answer the following insight problem:

Mr. Hardy was washing windows on a high-rise office building when he slipped and fell off a sixty-foot ladder onto the concrete sidewalk below. Incredibly, he did not injure himself in any way. How is this possible?

Most people initially assume that Mr. Hardy is standing at the top of the ladder, despite the fact that the problem does not declare it. It is difficult to solve the problem because the assumption occurs automatically and often without conscious awareness. When it occurs to the problem solver that Mr. Hardy may be on the bottom rung of the ladder, the problem is solved suddenly and an Aha! moment may occur. Somewhat less is known about how other types of problems can elicit the same Aha! experience. Naturalistic studies of insight suggest that many different pathways to insight exist, and that restructuring is likely just one of many mechanisms responsible for subjectively sudden solutions (Klein & Jarosz, 2011).

Recent definitions of insight also emphasize the phenomenological dimension that accompanies some sudden solutions such as pleasure, relief, certainty, drive, and surprise (Danek & Wiley, 2017; Webb et al., 2016). These dimensions are underlying qualities of the Aha! moment—the umbrella term for the phenomenology sometimes experienced when a problem is solved. As we have already argued, although the Aha! experience is more common under certain circumstances—particularly in ill-defined creative problems—the experience is not problem-specific. Thus, over time, it became common practice to measure Laukkonen, Schooler, & Tangen, 2018

9


when insight occurs on a case-by-case basis, and to use a single set of problems to investigate both insight and non-insight solutions classified according to selfreport—a particularly popular technique when investigating neural correlates of insight (Bowden & Jung-Beeman, 2007). With regard to theory, the phenomenological dimension has been largely ignored for the past century, which has become problematic recently because Aha! moments are increasingly used to define insight, and self-reported Aha! moments are used as dependent measures (Bowden & Jung-Beeman, 2007; Kounios & Beeman, 2014). Topolinksi and Reber (2010) aptly describe this state of affairs as follows:

“Although a body of excellent research has examined the cognitive and brain processes that may lead up to an insight (see Kounios & Beeman, 2009; Gick & Lockhart, 1995; Metcalfe & Wiebe, 1987), there is no coherent explanation for the phenomenology (i.e., experience) of insight. This is astonishing, since for many researchers the phenomenology is sufficient to define insight (e.g., Kounios & Beeman, 2009; Gick & Lockhart, 1995; Metcalfe & Wiebe, 1987).”

Before we turn our attention to one account of Aha! as a sudden increase in processing fluency, let us first consider more carefully what the Aha! moment feels like. When experiments aim to detect Aha! moments, the tendency has been to describe to participants what it is, so that they can identify when it occurs. However, in doing so, there is a risk of circularity. If participants are told what an Aha! moment is, and then report that an Aha! moment occurred according to the prescribed definition, then it’s possible to miss important dimensions or features of the Aha! moment that occur naturally. Or, we risk overemphasizing certain dimensions and underemphasizing others. Addressing this concern, Danek and Wiley (2017) compared participants’ global Aha! ratings to six different dimensions that tend to accompany the insight experience. The authors found that all dimensions were significant predictors of Aha!. Pleasure (r=.66) was the best predictor of the global Aha! rating, followed by certainty (r=.58), Laukkonen, Schooler, & Tangen, 2018

10


suddenness (r=.49), relief (r=.49), drive (r=.28), and then surprise (r=.07). When describing an Aha! moment to participants, some research has employed as few as two dimensions, one of which was surprise, the weakest predictor of Aha! moments (Cushen & Wiley, 2012). This classification is clearly problematic given Danek and Wiley’s (2017) findings. However, there are also some weaknesses to purely reducing the Aha! moment to its component parts. For instance, even pleasure—the strongest predictor of Aha! moments—is a regularity, but not a rule. Negative insights, described as Uh-Oh! moments (Hill & Kemp, 2016), the experience of chagrin (Gick & Lockhart, 1995), and even D’oh! moments (Hill & Kemp, 2016), are all more or less displeasurable versions of the Aha! experience, but otherwise share most of the same features. It is perhaps better to think of Aha! as an experience like hope, or shame—sharing qualities of different emotions, but being an experience of its own and different from the sum of its parts. Different dimensions of the Aha! experience may also provide other nuanced signals about the nature of the idea that popped into mind and its consequences. ‘Truthiness’ is likely just one of the intuitions humans have about their ideas, but we argue it is a key role of the quintessential Eureka moment.

To summarize, traditional definitions of insight may refer to a type of creative problem, to the cognitive process of restructuring, or the phenomenology of an Aha! moment. Increasingly, researchers are using the phenomenology—the Aha! moment—as the defining characteristic of insight, rather than the problem, or the process (Kounios et al., 2006; Bowden & Jung-Beeman, 2007; Kounios & Beeman, 2014; Zedelius & Schooler, 2015; Salvi et al., 2016; Webb et al., 2016; Danek & Wiley, 2017; Laukkonen & Tangen, 2017). Yet, there is a distinct lack of theory surrounding the Aha! experience. It is particularly surprising that outside of the insight problem solving literature there is little mention of the Aha! moment at all, despite our impression that it is acutely relevant for both learning and decision-making (for example, Aha! moments are remembered better than non-Aha! moments, Danek et al., 2013). We suspect that there may be a widespread assumption that Aha! moments are merely an epiphenomenon Laukkonen, Schooler, & Tangen, 2018

11


not serving any specific purpose (see for example Klein & Jarosz, 2011), or perhaps simply a reaction to having found a solution, and that this may be partly responsible for the gap in the literature. One speculation is that the possibility of having intuitions about our own ideas—as opposed to ideas out in the world— has not been recognised. In the next section, we consider how processing fluency may play a role in facilitating the Aha! experience (Topolinski & Reber, 2010).

Fluency To the best of our knowledge, the fluency account of the Aha! experience is currently the only published explanation of why the Aha! phenomenology occurs, and appears to be gaining popularity (Topolinski & Reber, 2010). While the heuristic view is not a competing model of the Aha! experience, it is worthwhile to discuss the fluency account to illustrate how the two perspectives compliment each other, and where the heuristic view extends on fluency.

According to Topolinski and Reber (2010), when an unexpected solution appears in mind in an Aha! moment, a problem that was once difficult, confusing, or had disconnected elements, is suddenly resolved and processed fluently, leading to positive affect and judged truth. A number of studies find that manipulating fluency creates a sense of cognitive ease, pleasure, and confidence (Winkielman & Cacioppo, 2001; Topolinski & Strack, 2009; Reber & Schwarz, 1999). For example, statements that are presented with high figureground contrast (e.g., black letters with a white background) are more likely to be judged as true compared to low contrast statements (Reber & Schwarz, 1999). Repeated exposure makes a stimulus more pleasurable (Fang, Singh, & Ahluwalia, 2007; Reber, Winkielman, & Schwarz, 1998), and solutions presented more ‘suddenly’ following an anagram (50ms versus 150ms) are more likely to be judged as correct (Topolinski & Reber, 2010b). The authors argue that since suddenness, pleasure, and judged truth are dimensions of the Aha! experience—


12


and manipulating fluency affects each of these dimensions—then fluency is likely to be the driving force behind the experience of Aha!, or in their words, ‘the glue between its experiential features’ (Topolinski & Reber, 2010).

There are, however, some difficulties in falsifying the fluency account because self-reports of fluency are not easily obtained (as noted by Danek & Wiley, 2017). It is also the case that although the Aha! phenomenology has similar features to the effects of fluency manipulations, this does not mean it is therefore the effect of fluency. Another issue is that the intensity of Aha! moments can vary dramatically. The fluency account cannot easily explain why an Aha! moment is barely noticeable in some cases, and phenomenally large in others. Andrew Wiles describes his discovery of the solution to Fermat’s last theorem in 1994 as follows:

“At the beginning of September I was sitting here at this desk when suddenly, totally unexpectedly, I had this incredible revelation. It was the most important moment of my working life. Nothing I ever do again will… I’m sorry.”

Andrew Wiles fights back tears throughout the video, and in the end turns away from the camera because recounting the experience evokes a powerful emotional response. There is an apparent incongruity between the sheer emotional weight of some Aha! moments and effects we observe (or would expect) from changes in fluency. One possible reason is that, while fluency may be a productive account of some of the dimensions of the Aha! experience, it fails to account for all of them. According to participants’ own reports, relief is a key feature of the Aha! phenomenology that has not previously been connected to fluency (Danek & Wiley, 2017). Another dimension is drive, or inspiration, which accompanies some insight experiences (Danek & Wiley, 2017). When strong Eureka experiences occur—something difficult to evoke in the laboratory—then inspiration is likely to play a larger role than in laboratory puzzles (Danek et al., Laukkonen, Schooler, & Tangen, 2018

13


2014; Ohlsson, 1984). Archimedes was said to be so stimulated and inspired that he ran naked through the streets shouting “Eureka!”, pointing to the archetypal role of inspiration and the ‘rush of insight’ that accompanies the Eureka experience (Gick & Lockhart, 1995). Our overall impression is that fluency may play a precursory role in some of the phenomenology observed during insight moments, but it does not fully account for the dimensions or the intensity of the Aha! experience. It is worth pointing out that our own account of the Aha! experience as a decision-making heuristic is generally inclusive regarding the role of fluency. Fluency is one way to trigger some dimensions of the Aha! experience; but our Eureka heuristic is an account of how the Aha! experience is interpreted and its adaptive role in decision-making more generally.

One promising avenue of research here is in investigating whether artificial manipulations of fluency can influence the Aha! experience. Topolinski and Reber (2010) predict that it ought to be possible to generate illusions of insight by manipulating fluency in a problem solving context. We are also optimistic about this possibility, and there is already some evidence in this direction. For instance, positive affect tends to facilitate insight and creative problem solving (Isen, Daubman, & Nowicki, 1987; Rowe et al., 2007). Participants higher in positive affect also report experiencing more insights (Subramaniam, Kounios, Parrish, & Jung-Beeman. 2009). Findings of this sort may indicate that there is an element of self interpretation in the Aha! experience—positive affect confounds the pleasure dimension of the Aha! experience, so more Aha! moments are reported. We elaborate on this possibility in the next section.


14


Self Interpretation Humans often cannot reliably introspect regarding their mental processes or the true causes that underlie their behavior and attitudes, and regularly confabulate instead (see also Johansson et al., 2004; Johansson et al., 2006; Wegner & Wheatley, 1999; Brasil-Neto et al., 1992; Wegner, 2002; Schooler, 2002). This is not to say that human introspection is always incorrect, but that so-called introspections even when they are accurate, are in fact post-hoc interpretations based on implicit causal theories and an evaluation of contextual and sensory information. There are now dozens of examples where experiments find a mismatch between self-reported causes of behavior (and underlying cognitive processes) and the actual causes triggered by experimental manipulations (see Nisbett & Schachter, 1966; Latane & Darley, 1970 for famous examples, Nisbett & Wilson, 1997 and Caruthers, 2009 for reviews, and for more recent work see Dougal & Schooler, 2007, Whittlesea & Williams, 1998, Johansson et al., 2004; Johansson et al., 2005; Johansson et al., 2006). One account regarding the evolution of metacognition is that first-order cognitive processes—while usually efficacious—benefit from monitoring and troubleshooting (Shallice, 1988). This monitoring and troubleshooting of one’s own cognitions may have evolved out of the capacity to monitor and interpret the behaviors, feelings, and intentions of others (Caruthers, 2009). The evolutionary pressure to behave socially—to mind-read and to empathize with others—may have eventually allowed humans to reverse the lens of introspection and ‘mindread’ themselves. One of our central claims is that humans self interpret their phenomenology in deciding whether an idea or solution is true, and therefore the fact that self interpretation is a basic feature of human cognition provides the framework for this observation. This literature also reinforces the fact that mental processing and problem solving processes are often hidden from awareness. If humans had introspective access to their own problem solving


15


processes then there would be no need for an affective cue because the reasoning that underlies a solution would be known directly accessible.

The classic research conducted by Maier (1931) provides an illustrative example. Maier (1931) set up a famous insight problem in a laboratory with two ropes hanging from the ceiling. The ropes each had different objects attached at the bottom, such as pliers, or clamps. The task was to attach the two ropes, but it was physically impossible to reach one rope while holding the other. When participants were close to giving up, Maier would haphazardly send one of the ropes in motion. Shortly thereafter participants tended to ‘suddenly’ discover the solution: they tied an object to the rope, set it in motion, quickly grabbed the other rope and then caught the swinging object and tied the ropes together. When probed about the source of the solution, individuals explained that it simply ‘dawned on them,’ or that it was ‘the next obvious thing to try.’ Others have since replicated Maier’s basic findings (Landrum, 1990). In another famous demonstration, Nisbett and Schachter (1966) demonstrated that physiological responses to an electric shock can be misattributed by participants. Subjects were given a placebo pill that ostensibly would lead to the same physiological effects as typically reported during electric shocks: heart palpitations, breathing irregularities, hand tremors, and butterflies in the stomach. The participants who were given the placebo pill opted for fourfold more powerful electric shocks on average than a control group. When these participants were asked why they took more shocks than the average, they failed to consider that the placebo pill had any influence. One responded with, “Gee, I don’t really know…. Well, I used to build radios and stuff when I was 13 or 14, and maybe I got used to electric shock.” The subjects also reported that they did not think about the pill during the shock, and did not consider that the pill was having any physical effects. Yet, it is most likely the case that participants were attributing their symptoms to the pill instead of the shock, and were therefore willing to expose themselves to a stronger amperage. In other words, the participants were misattributing the source of their phenomenology because they could not access the true source. Laukkonen, Schooler, & Tangen, 2018

16


Misattribution of phenomenology can also occur in the context of memory judgments (Schooler, Dougal, & Johnson 1998; Schooler 2001, Dougal & Schooler, 2007). In one study, Dougal and Schooler (2007) presented participants 60 words to memorize, and then provided a set of anagrams to solve followed by a recognition judgment regarding the solution of the anagram. They found that the anagrams that were solved were more likely to be recognised compared to the anagrams that were not, suggesting that something about solving the anagram was leading to an ‘illusion of prior experience.’ In five more experiments, Dougal and Schooler (2007) replicate their basic finding and also demonstrate that the feeling of solving a problem can confound the feeling of prior recall, or “remembering remembering.” The authors consider a series of possible explanations including item selection (Watkins & Gibson, 1988), a fluency account (Lindsay & Kelley, 1996), item exposure (Watkins & Peynircioglu, 1990), and an identification heuristic account (Higham & Vokey, 2000), but argue that the view receiving the most consistent support is that participants are misattributing the phenomenological state induced by discovering a solution to an anagram with the feeling of remembering the word. Specifically, Dougal and Schooler (2007) suggest that the Aha! experience leads to a false inference of remembering, where participants incorrectly interpret their Aha! phenomenology as a signal that a word is familiar. In a series of similar experiments conducted by Whittlesea and colleagues (1990; 1998; 2000), they suggest that the feeling of surprise can also confound memory judgments, where pseudohomophones of real words (e.g., frog spelled phrawg) are more likely to be reported as old (recognised as previously seen) than words spelled correctly and non-words. The surprise experienced as a consequence of an unfamiliar letter-string that sounded like a real word may have lead to a misattribution of phenomenology so that participants felt that the word was previously seen (Whittlesea & Williams, 2000).


17


In yet another experiment, Brasil-Neto et al., (1992) used focal magnetic stimulation to move a subject’s finger, but the subjects reported making the choice to move the finger. There are many examples, with some of the strongest evidence coming from split-brain patients (see for example Gazzaniga, 1995; 2000). In each case, people demonstrably fail to introspect on the reasons for their behavior, and instead make interpretations given whatever evidence they have available (Caruthers, 2009). Recent examples of change and choice blindness are also fairly compelling. Across a number of studies, Johansson and colleagues demonstrate that when individuals make a choice, but are persuaded that they made a different choice without their awareness, they will eagerly confabulate reasons for making a choice that they did not in fact make (Johansson et al., 2004; Johansson et al., 2005; Johansson et al., 2006).

From this point of view, the Aha! experience is ripe for self interpretation (Dougal & Schooler, 2007; Whittlesea & Williams, 2000; Nisbett & Wilson, 1977; Caruthers, 2009). When faced with a problem, one begins by making conscious efforts to solve it. As information is accumulated it is also processed and integrated below awareness. These implicit processes may eventually reach a threshold where a candidate solution is cognised, sometimes while the person is working on the problem, and sometimes while they are engaged in another task. The solution is then cognised with accompanying Eureka phenomenology—but the ‘why’ or ‘how’ of the solution may remain hidden. The individual, sometimes without any other information to go on, relies on the one cue that is available— their phenomenology. Note that although there is overwhelming evidence that processes preceding an insight are hidden from awareness (the ‘how’), there is likely to be circumstances where a problem solver can accurately report why the solution that suddenly popped into mind is a correct one, particularly if the reasoning underlying the solution is simple. However, recall that in many cases where self interpretation and confabulation occurs, people believe that they are genuinely introspecting (Caruthers, 2009). Hence, when a sudden insight occurs, even if it is possible to provide a narrative about the reasons for a discovery or Laukkonen, Schooler, & Tangen, 2018

18


why it is correct, it too may be a matter of post-hoc theorising and self interpretation. It is much easier to explain why a solution is correct once the solution is known. Further research is necessary in order to test participants’ ability to verbalise accurately ‘why’ an insight solution is correct. One prediction is that it ought to be possible to have an insight experience and provide an accurate solution, but also provide an inaccurate description of why that solution is correct. This would demonstrate a dissociation such that the true reasoning behind the solution does not need to be cognised for a correct solution to be known, and the Aha! phenomenology may in some cases be more ‘rational’ than one’s own deliberate rationalising. It may also be that in some domains it is easy to see why a solution is correct as soon as it is cognised, but in other more complex domains—like those reflected in famous anecdotes—the Aha! experience can trigger immediate certainty, “without the slightest instant of reflection” as reported by Jacques Hadamard.

Another possibility is that Aha! phenomenology may be employed to induce truth misattribution, in a similar fashion to discovery misattribution (Dougal and Schooler, 2007). For instance, if an Aha! experience can be made to occur at the same time as an unrelated proposition (e.g., “lithium is the lightest of all metals”, Reber & Unkelbach, 2010) , then the Aha! experience may be falsely attributed to—or confused with—the ‘truthiness’ of the proposition. The discussed tendency to self interpret in humans lays a foundation for the heuristic view of the Aha! experience, which is described in more detail below.

The Eureka Heuristic The heuristics and biases approach has had an enormous impact on cognitive science and decision-making research over the past 40 years (Simon, 1956; 1982; Tversky & Kahneman, 1973; 1975; Gigerenzer & Goldstein, 1996; Kahneman, 2003; Gigerenzer & Gaissmaier, 2011; Slovic et al., 2007; Schooler &


19


Hertwig, 2005). The majority of this progress in understanding how humans make decisions comes from a deceptively simple idea: there are shortcuts to navigating a complex world. These mental shortcuts are known as heuristics, and the study of heuristics is the identification of these mental shortcuts and their implications. Stereotyping is a familiar example. In order to make sense of the world, humans categorise people and objects according to experiences with them. Due to the inherent fact that humans have limited exposure to the members of any category—and limited cognitive capacities—they are forced to generalize from the small subset that they have been exposed to. This process of generalizing impressions from known category members to all other members— despite a lack of experience with the new instances—is stereotyping. Stereotyping is an adaptive mental shortcut. For example, police officers, store clerks, and doctors dress a certain way. Therefore, it’s usually a safe bet that the stranger in the familiar blue uniform is in fact a police officer—but not always. One of the advantages in understanding a heuristic is in mapping the situations where they fall short. Stereotyping, like all other heuristics, have certain trade offs (Tversky & Kahneman, 1975). The tendency to generalize what are believed to be the traits or behaviors of other races, genders, or minorities can be problematic, particularly when those categories are highly amorphous and have little predictive power at the level of individuals. Humans run the risk of generalizing too far. In some cases, where the category is based on a biased sample to begin with—such as an overrepresentation in the media—then the stereotype may be entirely false, and likewise every generalisation that follows from it. Despite the fact that heuristics can and do fail in some circumstances, it is important not to underestimate just how adaptive they are. Gigerenzer and Gaissmaier (2011) argue that many heuristics are ecologically rational; they are highly adapted to the structure of the environment, and in some cases perform more accurately than sophisticated rational models.


20


For the purposes of viewing the Aha! experience within this framework, the affect heuristic affords a particularly useful analogy. Slovic et al., (2007) pointed out that, “Although analysis is certainly important in some decisionmaking circumstances, reliance on affect and emotion is a quicker, easier, and more efficient way to navigate in a complex, uncertain, and sometimes dangerous world.” The authors argue that the role of affect in decision-making and cognition has not received the attention it deserves, and the weight of its influence has been overlooked. They cite an excerpt from Damasio (1994) who studied patients that suffered brain damage in ventromedial frontal cortices that resulted in a specific impairment in the ability to “feel.” Counter intuitively, the patients experienced a marked failure to make rational decisions and to reason appropriately, despite appearing to have all other instruments traditionally considered important for reason intact. Schwarz’s (2010) theory of feelings-asinformation outlines how, for example bodily sensations and emotions, carry information that individuals regularly use to efficiently make judgments. An obvious example is the feeling of hunger, which signals a state about the body’s nutritional requirements. Schwarz (2010) notes that humans are much more sensitive to the phenomenology itself rather than the causes of the phenomenology, which is one reason why context and incidental information or situational factors can lead decisions astray. For example, mere exposure to a stimulus can make it feel more familiar, and even increase its perceived veracity (Zajonc, 1968; Weaver, Garcia, Schwarz, & Miller, 2007; Schwarz & Newman, 2017). Slovic et al., (2007) suggest that affective cues are based on impressions developed through experience, where some object or event has been associated with positive or negative affect in the past. In a new but uncertain situation, a person can draw on their existing impressions or experiences of similar situations by consulting their affective responses, assisting them to make fast and efficient judgments informed by their experiences. Studies show that affect has widespread influence on judgments and decisions even in abstract domains that seem on the surface purely analytical, including risk judgments (Fischhoff et al., 1978), gambling and probability judgments (Lowenstein et al., 2001), and a Laukkonen, Schooler, & Tangen, 2018

21


range of preference evaluations (Mellers et al., 1992; Anderson, 1981; Winkielman et al., 1997).

The analogy between the affect heuristic and the Eureka heuristic has many levels. One obvious similarity is that they both involve an interpretation of affect to guide decisions. A crucial difference, however, lies in the underlying mechanisms. Slovic et al., (2007) proposed that the affective response to a stimulus or situation draws on an “affect pool” that contains previous, related experiences and representations ‘tagged’ as either emotionally positive or negative. When in a familiar situation, a person can draw on the affect pool—by reacting to the affect they currently experience—in order to help guide decisionmaking. The Eureka heuristic may function in a similar albeit different way— where the affect heuristic draws on an “affect pool,” the Eureka heuristic draws on a “knowledge pool.” When a solution is found, and existing knowledge about the problem and relevant associations resonate with the candidate solution, then an Aha! moment occurs to signal that one’s complex, implicit knowledge structures are consistent with the current solution. Another way to conceptualize this difference is that intuitions and affective cues occur to inform us about events out in the world, whereas the insight experience occurs to inform us about events in our minds—our own ideas.

If the Aha! moment acts as an ecologically rational heuristic akin to fluency, recognition, or affect (Gigerenzer & Gaissmaier, 2011, Slovic et al., 2007; Schooler & Hertwig, 2005), then it ought to be predictive of problem solving accuracy and subjective confidence in the solution. The data are clear in both cases. Across a range of problems, solutions that are accompanied by the Aha! experience are more likely to be accurate than solutions that are not, and Aha! moments are highly predictive of confidence—despite no deliberate verification or introspection by the problem solver (Metcalfe 1986; Metcalfe & Wiebe, 1987; Danek et al., 2014; Salvi et al., 2016; Danek & Wiley, 2017; Webb, Little, & Cropper; 2016; Hedne et al., 2016). The relationship between Aha! and accuracy Laukkonen, Schooler, & Tangen, 2018

22


varies depending on the problem type. A multiplication problem like 46 x 83 is rarely solved unconsciously using associative processing, and so there is a low probability of experiencing an Aha! moment. This is likely why, for classical analytic problems, there are far fewer Aha! moments, and the correlation between Aha! and accuracy is negligible, whereas remote associates tests and insight problems—which are thought to involve associative or implicit restructuring processes—have a strong positive insight accuracy relationship (Webb et al., 2016; Ohlsson, 1984; 1992; 2011; Knöblich et al., 1999; Danek et al., 2017). The more the problem is amenable to a linear, conscious strategy, the more easily it is verified on the basis of the reasoning that lead to the solution, and there is little use for informative phenomenology. Whereas, if the problem is solved through unconscious processing, and the ‘how’ or ‘why’ of the solution may not be accessible, then the phenomenology of the Aha! moment is helpful for informing the individual of its veracity. Consistent with this view, in a comparison of the think aloud protocols leading up to successful solutions of problems associated with insight versus non-insight solutions, Schooler & Melcher (1995) found that numerous elements (e.g. arguments, re-reads) were predictive of analytic problem solutions, but very little was predictive of insight solutions.

The above conception can explain why Aha! moments are problem-general rather than problem-specific, and why certain kinds of problems elicit more Aha! moments than others (Salvi et al., 2016; Webb et al., 2016; Zedelius & Schooler, 2015). The more likely the problem is to draw on implicit knowledge structures and processing below awareness, the more likely it is to be solved with an Aha! moment. As long as this principle is met, then there are potentially myriad cognitions that can precede an Aha! moment, which is why an idea for a new painting, a line of poetry, a way to resolve a dispute, or a solution to an engineering problem, can all appear in mind in a sudden moment of insight. We stress this point because, if true, it is a crucial step in our understanding of insight in general: There is no single problem solving process that leads to Laukkonen, Schooler, & Tangen, 2018

23


insight—the experience may not be a consequence of a specific set of cognitions that take place in solving the problem. Instead, the experience of insight may be a signal that the work done behind the scenes has reached a conclusion that is likely to work given what is known. It is a kind of inference about the validity of the idea rather than a consequence of arriving at a solution. We now consider a trade off where the Eureka heuristic can lead to a false and persuasive sense that a true solution has been found.

The Insight Fallacy The mathematician and Nobel laureate John Nash was asked why he believed that he was being recruited by aliens to save the world. His response powerfully illustrates the recursive danger of the Eureka heuristic. He said that: “…the ideas I had about supernatural beings came to me the same way that my mathematical ideas did. So I took them seriously” (Nasar, 1998). Here, John Nash may have committed what we term the insight fallacy. He has concluded that an idea is true solely because it occurred to him with certain phenomenology, in this case the same phenomenology as his previous mathematical discoveries. As mentioned earlier, one of the benefits in defining heuristics and understanding the shortcuts we use is that we may also uncover the circumstances where they lead to errors. For example, the representativeness heuristic is associated with errors of base-rate neglect (Tversky & Kahneman, 1974). The availability heuristic can be lead astray by sampling biases, and anchoring and adjustment heuristics can be affected by incidental and irrelevant information (Tversky & Kahneman, 1973; 1974). What, if any, are the trade offs that occur as a consequence of using Aha! phenomenology to track the ‘truthiness’ of our ideas?


24


Since Aha! moments are in essence an affective response, another such response may provide a useful analogy. Fear is an adaptive signal of a dangerous or challenging situation, but is also sometimes unwarranted or irrational, and in severe cases, debilitating. We can see that often fear ought to be overcome, for example, so that we can fly in a plane, swim in the ocean, or speak in public. Phobias can cause overwhelming and unjustified feelings of fear. The same is not so obvious for feelings that accompany our ideas and beliefs when they arise— our Aha! moments. The Aha! moment, like fear, may be a helpful signal that perhaps we’ve discovered something true and important (Salvi et al., 2016; Webb et al., 2016; Danek & Wiley, 2017). However, if there is overwhelming contradictory evidence, or one is suffering from mental illness (John Nash was diagnosed with schizophrenia in 1959), then it is likely that no matter how intense our Aha! moment, the contents of our idea will remain untrue. Just as a person might experience a profound fear of elevators, the intensity of the fear does not make the elevator dangerous. Likewise, if one is suffering from delusions, or they have been mislead with false information, then no matter how explosive the Aha! moment, the idea is no more likely to be correct. This is a particular concern given the powerful effect that the Aha! moment has on judged truth and confidence—despite the fact that we may not know why we are suddenly so confident (Hedne et al., 2016; Danek & Wiley, 2017). It is hard to predict how widespread the impact is. But consider how many contradictory ideas there are in the world, and how many of the ideas that we hold most dearly were—at least subjectively—our own insights. One conclusion from this line of thinking is that we ought to be aware that the phenomenology accompanying our ideas is predictive, but also highly fallible (Danek & Wiley, 2017). Danek and Wiley (2017) found that 37% of incorrect solutions to magic tricks were reported as Aha! moments. The proportion of false insights in everyday life may be different—since magic tricks are a domain of negligible experience for most people—but even a fraction of 37% is alarmingly prevalent.


25


Since false insights and Aha! moments do occur, then to commit the insight fallacy is to believe that an idea is true—rather than probably true— simply because the solution was accompanied by an Aha! moment. Hedne et al. (2016) showed that when an Aha! moment occurs, subjects are less likely to accept an alternative solution to the problem, and are more likely to stick with solutions that are similar to their insight. It may be that Aha! solutions are particularly hard to revise since the underlying process is opaque (Bowden, 1997; Hedne et al., 2016; Smith & Kounios, 1996; Kounios & Beeman, 2014). After all, it is impossible to argue against in reasoning that are themselves unknown to the problem solver. The Eureka heuristic is certainly functional, but when an error does occur, the consequences can be dire. Not only are Aha! moments potentially incorrigible, they also promote inspiration, and provide a drive towards action (Danek & Wiley, 2017). Relative to an incorrect-but-analytic solution to a problem, when a false insight occurs, it may be more difficult to change the person’s mind and to prevent them from behaving as if the solution were true.

One important implication of the insight fallacy is that we should learn to be aware that the feeling of Aha! is fallible and that in certain situations, and perhaps in certain states of mind, it is advantageous to actively doubt our light bulb moments and search for support beyond our phenomenology. There are many promising avenues of research here, which we discuss further in the final section below. We propose that an important question for future research is not, ‘what are the variables that lead to more Aha! moments?’ Instead we ought to be asking, ‘what are the variables that lead to accurate Aha! moments?’. For example, it is likely that some psychological disorders are particularly prone to false insights. A person suffering from schizophrenia or any illness where delusions are present, may—like John Nash—be interpreting their phenomenology to conclude that an idea is true. If the functioning of this particular cognitive process is compromised, then the phenomenology experienced is no longer informative, and instead may lead to a strong conviction Laukkonen, Schooler, & Tangen, 2018

26


to ideas that are otherwise unjustified. Research regarding false insights are in their infancy, so identifying the situations where the Aha! moment is particularly error prone is an exciting new path for empirical work.

When can Aha! moments be trusted? In this short section, we take some liberty to speculate about how the Aha! moment may affect decision-making more broadly in society. Since the reasoning and information that underlies an Aha! solution may be hidden, then for any human problem solver, it is difficult to know when a sudden insight ought to be trusted, and when it ought to be doubted in the face of contradictory evidence. A solution discovered with an Aha! experience—provided it is in an area of considerable expertise and experience—may be drawing on a vast and complex web of associations, experiences, and memories. What if someone points out a flaw in the new solution? How should an expert weigh this information in the face of her intuitions? The expert may not be able to bring to mind exactly ‘why they know what they know’, so trying to articulate a rational argument may result in futile confabulation rather than do justice to the content of the insight (as discussed earlier). Perhaps in such cases there is reason not to hastily dismiss the Aha! solution. Many great scientific discoveries faced harsh criticism before they were accepted by the mainstream, particularly if they affected paradigmatic views. Of course, once the contents of the insight and the underlying reasoning is better understood, then the weight of the evidence ought to determine whether the solution is accepted or revised.

There may also be some domains where Aha! moments are more common for the average person. For example, any human can experience an explosive Aha! moment regarding one’s self, relationships, or worldviews. Some public domains may also be predisposed to Aha! moments: politics, culture, philosophy, morality, and religion can draw on networks of knowledge that are even


27


tenuously relevant. With regard to these more diffuse domains, any person may have strong insight experiences because each of us has some relevant knowledge structures. However, it is in precisely these areas that the Aha! experience may be most error prone. The validity of the Aha! experience—if it is cuing a consistency with existing knowledge structures—is in direct relationship with the quality of those knowledge structures and thus the information that underlies them. Depending on unique life experiences and exposure to different cultural myths and the sheer abundance of (often contradictory) information available, it is not surprising that there are revelations of almost every imaginable sort in these areas. Since insights are the product of an individual’s knowledge and experiences—and most individuals will necessarily be exposed to only a fraction of the relevant information in these domains—then accurate insights may in fact be rare. And where knowledge structures are either biased or untrue, then the Aha! moment may serve to further reinforce and motivate false beliefs. In the case of complex belief systems of cults, certain conspiracy theories, or superstitious beliefs, one can be an ‘expert’ in a domain where the knowledge structures that fuel intense Eureka moments are fictitious to begin with. Here, the phenomenology that accompanies new ideas may be altogether unreliable and act only to recursively increase the persistence of these worldviews (see Figure 2).

Fictitious knowledge structures are acquired and reinforced

Aha! experience leads to renewed inspiration and confidence

New idea is discovered accompanied by Aha! phenomenology

Figure 2. The recursive nature of the Aha! phenomenology in learning. Laukkonen, Schooler, & Tangen, 2018

28


Summary and Future Directions We have argued that the Aha! phenomenology plays an adaptive role in decision-making and problem solving, and have proposed what that role may be. The heuristic view requires a restructuring of the way we think about Aha! moments, from a reaction to an idea or solution, to an intuition about the idea or solution. Human experience is filled with rich phenomenology, bodily sensations, and emotions, that guide our decisions and help us to intuitively navigate complexity and uncertainty. It would be at odds with the greater body of psychological research if the ability to feel was important in nearly all other domains of judgment and decision making, but not with regard to our own ideas and solutions to problems (Slovic et al., 2007; Schwarz, 2010; Kahneman & Beatty, 1973; Kahneman, 2011).

Since much of complex, associative, problem solving occurs below awareness, it is perhaps unnecessary—and certainly inefficient—to review all of the reasoning and information involved with each candidate solution (Ohlsson, 1992; 2011; Schooler, Ohlsson, & Brooks, 1993; Schooler & Melcher, 1995; Bowden, 1997; Hedne et al., 2016; Smith & Kounios, 1996; Kounios & Beeman, 2014). Therefore, when it comes to solving problems, and our own ideas, knowing ‘why we know’ is not as important as simply ‘knowing that we know’, especially when time is of the essence. The Aha! phenomenology may serve as a fast and frugal means of signalling implicit support for a solution, pleasure, confidence, and the drive to act on the solution (Danek & Wiley, 2017).

Much research on insight problem solving has centered around debates about the process underlying insight and non-insight problem solving, and most authors implicate distinct cognitive processes (Ohlsson, 1984; 1992; 2011 Metcalfe & Wiebe, 1987; Jones, 2003; Gilhooly & Murphy, 2005; Weisberg & Alba, 1981; Chein et al., 2010; Fleck & Weisberg, 2013; Ollinger et al., 2013;


29


Ollinger & Knoblich, 2009, to name a few). However, as it became increasingly common to measure when insights occur on a case-by-case basis, it also became clear that understanding how insight problems are solved is not necessarily going to address the question of how or why insight experiences occur (Bowden & Jung-Beeman, 2007; Kounios & Beeman, 2014; Webb et al., 2016). The problem solving data, as well as naturalistic studies, both indicate that there may be many ways to solve a single problem, and that many different processes can lead to insights (Salvi et al., 2016; Webb et al., 2016; Danek & Wiley, 2017; Klein & Jarosz, 2011). Therefore, with regard to the insight experience—as opposed to a particular problem or problem solving process—there has been relatively less progress over the last century (Topolinksi & Reber, 2010).

The Eureka heuristic helps us understand recent data, and speaks to a number of long-standing debates in the literature. First, the Eureka heuristic model can explain why Aha! moments occur in such a wide array of problems. Any problem, be it creative or analytic, can lead to an Aha! moment, provided that some crucial steps in the solution process occurred below awareness. Traditional insight problems, remote associate problems, matchstick arithmetic, anagrams, magic tricks, and likely many others, reliably lead to Aha! moments in laboratories because they are easily amenable to unconscious processing (Ohlsson, 1984; Kounios & Beeman, 2014; Salvi et al., 2016; Danek & Wiley, 2017; Hedne et al., 2016; Knoblich et al., 1999; Dougal & Schooler, 2007). Provided that some of the operations involved with finding the solution are hidden from awareness—even in the case of traditional analytic problems—then an Aha! experience may occur to signal that successful processing (in the context of ones own knowledge) has occurred.

There has been a recent effort to better understand how we come to decide whether a claim is true or not, with some compelling evidence that truth may not be quite as important as ‘truthiness’, that is, “Truth that comes from the gut, not the book” (Colbert, 2005). Certain intuitive evaluations such as familiarity, Laukkonen, Schooler, & Tangen, 2018

30


fluency, and availability can act as shortcuts where more effortful and timeconsuming evaluations of truth are undesirable or unavailable (see Schwarz & Newman, 2017 for a recent review). Likewise, in solving everyday problems—the source of a new rattle in the engine of our car, the reason that our friend is upset with us, or the particular aroma that we encounter in a lobby of a hotel—is also going to benefit from our intuitions. The only crucial difference is that the event occurs in our own minds, rather than out in the world. The Aha! experience predicts immediate confidence, pleasure, and drive towards action, and in situations of uncertainty appears also to predict accurate solutions to problems (Metcalfe 1986; Metcalfe & Wiebe, 1987; Danek et al., 2014; Salvi et al., 2016; Danek & Wiley, 2017; Webb, Little, & Cropper; 2016; Hedne et al., 2016). There is likely a variety of ways that the nuance of the Aha! phenomenology might be helpful whenever an idea pops into mind—and variations in the phenomenology may serve other specific purposes—but here we have argued that one of the benefits of the Aha! intuition is that it signals ‘truthiness’, or the likelihood that our own idea is a correct, given what we know.

There is, however, a dark side to the Eureka heuristic. There is evidence that Aha! moments are difficult to revise Hedne et al., (2016), and Aha! solutions are more easily recalled (Danek et al., 2013). The Aha! moment may be highly functional most of the time, but when it signals a false solution, then the implications are profound. The individual may be left with a powerful sense of certainty, and also the drive and inspiration to act according to an incorrect solution. In ill-defined domains, where one has very little experience, or domains that are fictional by nature—Aha! moments may simply provide a kind of recursive illusion of progress. In such circumstances, the Aha! moment may reinforce solutions and ideas that are false, and in some cases inspire further work in the wrong problem space.


31


Throughout this paper, we have made a number of predictions and highlighted several areas for future research. This may be seen as both a strength and a weakness of the theory; a number of possible streams of research are identified, but a number of hypotheses remain without conclusive evidence aside from analogies to related research. The promise of novel experiments, and the breadth of the theories and empirical work on which our predictions are founded, we believe sufficiently outweighs the risks of falsification (the possibility of which is itself a strength). Below we reiterate some key predictions already made and provide a few more novel hypotheses.

One particularly promising direction for empirical work will be in identifying the key variables that predict the accuracy or validity of Aha! moments. Many studies have aimed to measure and manipulate the ability to solve insight problems, or increase the incidence of insight experiences (e.g., Weller et al., 2011; Thomas & Lieras, 2009; Ostafin & Kassman, 2012; Steidle & Werth, 2013; Jarosz et al., 2012). However, increasing the incidence of insights overall may also lead to more false insights. In the context of everyday life, false insights may be at least as unhelpful as accurate insights are helpful (Hedne et al., 2016; Danek et al., 2016; Danek et al., 2013). If humans are indeed using the Aha! experience as a way to inform their judgments of truth and confidence regarding their own ideas, then manipulations of Aha! phenomenology ought to also lead to changes in confidence and truth judgments regarding those ideas. One such candidate is positive affect, as noted by Topolinski and Reber (2010). Since positive affect is a key dimension of most Aha! experiences (Danek & Wiley, 2017), then participants experiencing positive affect may report more frequent and perhaps more intense reports of Aha! during problem solving. If the positive affect is not relevant to the informational content of the Aha! moment— for example if it is artificially induced alongside problem solving—then we would expect more false insights. There is evidence that positive affect does indeed lead to more reports of insight, and also improved creative problem solving, but it is not clear whether more false insights also occur (Subramaniam et al., 2009). Laukkonen, Schooler, & Tangen, 2018

32


Other practical candidates for confounding the Aha! experience include arousal, fluency, suddenness, and surprise. Another interesting question is whether it is possible to shift criterions about what counts as a true insight, and whether such individual differences exist. Those who score high on the ‘need for cognition’, for instance, may adopt a more conservative criterion and therefore experience fewer false insights, whereas those who score low may adopt a more liberal criterion (Cacioppo, John, Petty, & Richard, 1982). Another possibility is that some intoxicants or psychoactive substances may jeopardise how informative Aha! experiences are, and increase (or indeed decrease) the rate of false insights. In any case, it is possible that problem solving performance overall may be similar among different groups, but the incidence of true versus false Aha! moments may differ significantly. More broadly there is also much opportunity for extending our understanding of other phenomenology that may accompany ideas, and also how nuanced dimensions of the Aha! experience may signal different or more specific intuitions, and therefore predict different behaviors and decisions. Another key prediction is that, since humans may be relying on their phenomenology rather than deliberate reasoning to evaluate the ‘truthiness’ of their ideas, then verbalising the reasoning underlying an insight solution may lead to confabulation. For example, it is conceivable that a problem may be correctly solved in an Aha! experience, but because the true reasoning underlying the insight is implicit, then the explicit reasoning may turn out to be false. Just as humans may have mistaken interpretations about the reasons for their behaviour, they may have mistaken interpretations about the reasons for their own ideas and solutions to problems.

Concluding remarks. We are optimistic that the Eureka heuristic account of insight fits the existing data, it explains the adaptive role of the phenomenological experience that accompany insights, and why that phenomenology has predictive power. This view of the insight experience may also help us understand why Aha! moments occur across a range of different problems, and why longstanding debates about the cognitive process underlying Laukkonen, Schooler, & Tangen, 2018

33


reports of insight are perhaps misdirected if the goal is to understand the insight experience. The insight fallacy illuminates the pitfalls of relying on affect to guide problem solving in some circumstances, and helps us to understand how accurate and inaccurate ideas may arise from interpreting one’s own phenomenology and how they may be recursively reinforced by the Aha! experience. In the same way that incidental information and situational factors may bias our intuitions about the world, they may also bias our intuitions regarding solutions to problems and our own ideas. There is much promise for future research to identify the situations, states of mind, and disorders that make the ‘phenomenology of truth’—the Eureka experience—either functional or dysfunctional.


34


References Anderson, N.H., 1981. Foundations of information integration theory. Academic, New York. Bowden, E. M. (1997). The effect of reportable and unreportable hints on anagram solution and the aha! experience. Consciousness and cognition, 6(4), 545-573 . Bowden, E. M., & Jung-Beeman, M. (2003). Aha! Insight experience correlates with solution activation in the right hemisphere. Psychonomic Bulletin & Review, 10(3), 730-737. Bowden, E., & Jung-Beeman, M. (2007). Methods for investigating the neural components of insight. Methods, 42(1), 87-99. http://dx.doi.org/10.1016/ j.ymeth.2006.11.007 Brasil-Neto, J., Pascual-Leone, A., Valls-Solé, J., Cohen, L. & Hallett, M. (1992) Focal transcranial magnetic stimulation and response bias in a forced choice task. Journal of Neurology, Neurosurgery, and Psychiatry 55:964– 66. Brinõl, P. & Petty, R. (2003) Overt head movements and persuasion: A selfvalidation analysis. Journal of Personality and Social Psychology 84:1123 – 39. Cacioppo, J. T., & Petty, R. E. (1982). The need for cognition. Journal of personality and social psychology, 42(1), 116. Carruthers, P. (2009). How we know our own minds: the relationship between mindreading and metacognition. The Behavioral and Brain Sciences, 32. http://dx.doi: 10.1017/S0140525X09000545 Colbert, S. (2005). Truthiness. The Colbert Report, 17 Oct 2005. http:// www.cc.com/ video- clips/63ite2/the-colbert-report-the-word---truthiness Chein, J., Weisberg, R., Streeter, N., & Kwok, S. (2010). Working memory and insight in the nine-dot problem. Memory & Cognition, 38(7), 883-892. http://dx.doi.org/10.3758/mc.38.7.883


35


Chu, Y. (2007). Human insight problem solving: Performance, processing, and phenomenology. ProQuest. Cushen, P. & Wiley, J. (2012). Cues to solution, restructuring patterns, and reports of insight in creative problem solving. Consciousness and Cognition, 21(3), 1166-1175. http://dx.doi.org/10.1016/j.concog.2012.03.013 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. Danek, A., Fraps, T., von Müller, A., Grothe, B., & Öllinger, M. (2014). Working Wonders? Investigating insight with magic tricks. Cognition, 130(2), 174-185. http://dx.doi.org/10.1016/j.cognition.2013.11.003 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. Dougal, S., & Schooler, J. W. (2007). Discovery misattribution: when solving is confused with remembering. Journal of Experimental Psychology: General, 136(4), 577. Chicago Ericsson, K. A., & Charness, N. (1994). Expert performance: Its structure and acquisition. American psychologist, 49(8), 725. Fischhoff, B., Slovic, P., Lichtenstein, S., Reid, S., Coombs, B., 1978. How safe is safe enough? A psychometric study of attitudes towards technological risks and benefits. Policy Sciences 9, 127–152. Fleck, J. & Weisberg, R. (2013). Insight versus analysis: Evidence for diverse methods in problem solving. Journal Of Cognitive Psychology, 25(4), 436-463. http://dx.doi.org/10.1080/20445911.2013.779248 Gazzaniga, M. S. (1995). Consciousness and the cerebral hemispheres. In M. S. Gazzaniga (Ed.), The cognitive neurosciences (pp. 1391-1400). Cambridge, MA: The MIT Press. Gazzaniga, M. (2000) Cerebral specialization and inter-hemispheric communication: Does the corpus callosum enable the human condition? Brain 123:1293–326. Laukkonen, Schooler, & Tangen, 2018

36


Gick, M.L., & Lockhart, R.S. (1995). Cognitive and affective components of insight. In R.J. Sternberg & J.E. Davidson (Eds.), The Nature of insight. (pp. 197–228). Cambridge, MA: MIT Press. Gigerenzer, G., & Goldstein, D. G. (1996). Reasoning the fast and frugal way: models of bounded rationality. Psychological review, 103(4), 650. Gigerenzer, G., & Gaissmaier, W. (2011). Heuristic decision making. Annual review of psychology, 62, 451-482. Gilhooly, K. & Murphy, P. (2005). Differentiating insight from non-insight problems. Thinking & Reasoning, 11(3), 279-302.10.1080/13546780442000187 Goldstein, D. G., & Gigerenzer, G. (2002). Models of ecological rationality: the recognition heuristic. Psychological review, 109(1), 75. Hattori, M., Sloman, S. A., & Orita, R. (2013). Effects of subliminal hints on insight problem solving. Psychonomic bulletin & review, 20(4), 790-797. Hedne, M. R., Norman, E., & Metcalfe, J. (2016). Intuitive feelings of warmth and confidence in insight and noninsight problem solving of magic tricks. Frontiers in psychology, 7. Hertwig, R., Herzog, S. M., Schooler, L. J., & Reimer, T. (2008). Fluency heuristic: a model of how the mind exploits a by-product of information retrieval. Journal of Experimental Psychology: Learning, memory, and cognition, 34(5), 1191. Higham, P. A., & Vokey, J. R. (2000). Judgment heuristics and recognition memory: Prime identification and target-processing fluency. Memory & Cognition, 28, 574–584. Isen, A. M., Daubman, K. A., & Nowicki, G. P. (1987). Positive affect facilitates creative problem solving. Journal of personality and social psychology, 52(6), 1122. Johansson, P., Hall, L., Sikstrom, S., & Olsson, A. (2004). Facing changes: Choice blindness and facial attractiveness. International Journal of Psychology, 39(5-6), 287-287.


37


Johansson, P., Hall, L., Sikstrom, S., & Olsson, A. (2005). Failure to detect mismatches between intention and outcome in a simple decision task. Science, 310(5745), 116-119. doi: 10.1126/science.1111709 Johansson, P., Hall, L., Sikstrom, S., Tarning, B., & Lind, A. (2006). How something can be said about telling more than we can know: On choice blindness and introspection. Consciousness and Cognition, 15(4), 673-692. doi: 10.1016/ j.concog.2006.09.004 Jones, G. (2003). Testing two cognitive theories of insight. Journal Of Experimental Psychology: Learning, Memory, And Cognition, 29(5), 1017-1027. http://dx.doi.org/10.1037/0278-7393.29.5.1017 Jung-Beeman, M., Bowden, E., Haberman, J., Frymiare, J., Arambel-Liu, S., & Greenblatt, R. et al. (2004). Neural Activity When People Solve Verbal Problems with Insight. Plos Biology, 2(4), e97. http://dx.doi.org/10.1371/ journal.pbio. 0020097 Kahneman, D. (2003). A perspective on judgment and choice: mapping bounded rationality. American psychologist, 58(9), 697. Kahneman, D. (2015). Thinking, fast and slow. New York: Farrar, Straus and Giroux. Klein, G. A. (1999). Sources of power: How people make decisions. MIT press. Klein, G. & Jarosz, A. (2011). A Naturalistic Study of Insight. Journal Of Cognitive Engineering And Decision Making, 5(4), 335-351. http:// dx.doi.org/10.1177/1555343411427013 Knoblich, G., Ohlsson, S., Haider, H., & Rhenius, D. (1999). Constraint relaxation and chunk decomposition in insight problem solving. Journal Of Experimental Psychology: Learning, Memory, And Cognition, 25(6), 1534-1555. http://dx.doi.org10.1037//0278-7393.25.6.1534 Knoblich, G., Ohlsson, S., & Raney, G. E. (2001). An eye movement study of insight problem solving. Memory & cognition, 29(7), 1000-1009. Kounios, J. & Beeman, M. (2009). The Aha! Moment: The Cognitive Neuroscience of Insight. Current Directions In Psychological Science, 18(4), 210-216. http://dx.doi.org/10.1111/j. 1467-8721.2009.01638.x Laukkonen, Schooler, & Tangen, 2018

38


Kounios, J. & Beeman, M. (2014). The Cognitive Neuroscience of Insight. Annual Review Of Psychology, 65(1), 71-93. http://dx.doi.org/10.1146/annurevpsych-010213-115154 Kounios, J., Fleck, J., Green, D., Payne, L., Stevenson, J., Bowden, E., & JungBeeman, M. (2008). The origins of insight in resting-state brain activity. Neuropsychologia, 46(1), 281-291. http://dx.doi.org/10.1016/ j.neuropsychologia.2007.07.013 Kounios, J., Frymiare, J., Bowden, E., Fleck, J., Subramaniam, K., Parrish, T., & Jung-Beeman, M. (2006). The Prepared Mind: Neural Activity Prior to Problem Presentation Predicts Subsequent Solution by Sudden Insight. Psychological Science, 17(10), 882-890. http://dx.doi.org/10.1111/j. 1467-9280.2006.01798.x Landrum, R. E. (1990). Maier's (1931) two-string problem revisited: evidence for spontaneous transfer?. Psychological Reports, 67(3), 1079-1088. Latané, B., & Darley, J. M. (1970). The unresponsive bystander: Why doesn't he help?. Appleton-Century-Crofts. Laukkonen, R. & Tangen, J. (2017). Can observing a Necker cube make you more insightful?. Consciousness And Cognition, 48, 198-211. http://dx.doi.org/ 10.1016/j.concog.2016.11.011 Lindsay, S. D., & Kelley, C. M. (1996). Creating illusions of familiarity in a cued recall remember/know paradigm. Journal of Memory and Language, 35, 197–211. Loewenstein, G.F., Weber, E.U., Hsee, C.K., Welch, E.S., 2001. Risk as feelings. Psychological Bulletin 127, 267–28 Maier, N. R. (1931). Reasoning and learning. Psychological review, 38(4), 332. Mellers, B.A., Richards, V., Birnbaum, J.H., 1992. Distributional theories of impression formation. Organizational Behavior and Human Decision Processes 51, 313–343. Metcalfe, J., & Wiebe, D. (1987). Intuition in insight and noninsight problem solving. Memory & Cognition, 15(3), 238-246. http://dx.doi.org/10.3758/ bf03197722 Laukkonen, Schooler, & Tangen, 2018

39


Nasar, S. (1998). A Beautiful Mind: A Biography of John Forbes Nash, Winner of the Nobel Prize in Economics, 1994. Simon and Schuster. Nisbett, R. E. (2015). Mindware: Tools for smart thinking. Farrar, Straus and Giroux. Nisbett, R. E., & Schachter, S. (1966). Cognitive manipulation of pain. Journal of Experimental Social Psychology, 2(3), 227-236. Nisbett, R. & Wilson, T. (1977) Telling more than we can know: Verbal reports on mental processes. Psychological Review 84:231–95. Ohlsson, S. (1984). Restructuring revisited. Scandinavian Journal Of Psychology, 25(2), 117-129. http://dx.doi.org/10.1111/j. 1467-9450.1984.tb01005.x Ohlsson, S. (1992). Information-processing explanations of insight and related phenomena. Advances in the psychology of thinking, 1, 1-44. Ohlsson, S. (2011). Deep learning: How the mind overrides experience. Cambridge University Press. Öllinger, M., & Knoblich, G. (2009). Psychological research on insight problem solving (pp. 275-300). Berlin-Heidelberg, Germany: Springer. Öllinger, M., Jones, G., Faber, A. H., & Knoblich, G. (2013). Cognitive mechanisms of insight: the role of heuristics and representational change in solving the eight-coin problem. Journal of Experimental Psychology: Learning, Memory, and Cognition, 39(3), 931. Ostafin, B. D., & Kassman, K. T. (2012). Stepping out of history: Mindfulness improves insight problem solving. Consciousness and cognition, 21(2), 1031-1036. Ovington, L. A., Saliba, A. J., Moran, C. C., Goldring, J., & MacDonald, J. B. (2015). Do people really have insights in the shower? The when, where and who of the Aha! Moment. The Journal of Creative Behavior, 0, 1-18 Reber, R., & Schwarz, N. (1999). Effects of perceptual fluency on judgments of truth. Consciousness and Cognition, 8, 338–342.


40


Reber, R., & Unkelbach, C. (2010). The epistemic status of processing fluency as source for judgments of truth. Review of philosophy and psychology, 1(4), 563-581. Rowe, G., Hirsh, J. B., & Anderson, A. K. (2007). Positive affect increases the breadth of attentional selection. Proceedings of the National Academy of Sciences, 104(1), 383-388. 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. Salvi, C., Bricolo, E., Kounios, J., Bowden, E., & Beeman, M. (2016). Insight solutions are correct more often than analytic solutions. Thinking & Reasoning, 1-18. Schwarz, N., & Newman, E. J. (2017). Psychological Science Agenda | August 2017. Psychological Science. Schooler, J. W. (2001). Discovering memories of abuse in the light of metaawareness. Journal of Aggression, Maltreatment & Trauma, 4(2), 105-136. Schooler, J. W. (2002). Re-representing consciousness: Dissociations between experience and meta-consciousness. Trends in cognitive sciences, 6(8), 339-344. Schooler, J. W., Ohlsson, S., & Brooks, K. (1993). Thoughts beyond words: When language overshadows insight. Journal of experimental psychology: General, 122(2), 166. Schooler, L. J., & Hertwig, R. (2005). How forgetting aids heuristic inference. Psychological Review, 112(3), 610. Schooler, J. W., & Melcher, J. (1995). The ineffability of insight. In S. Smith, T. Ward, & R. Finke (Eds.), The creative cognition approach (pp. 97–133). Cambridge, Mass.: MIT Press. Shallice, T. (1988). From neuropsychology to mental structure. Cambridge University Press. Slovic, P., Finucane, M. L., Peters, E., & MacGregor, D. G. (2007). The affect heuristic. European journal of operational research, 177(3), 1333-1352. Laukkonen, Schooler, & Tangen, 2018

41


Smith, R. W., & Kounios, J. (1996). Sudden insight: All-or-none processing revealed by speed–accuracy decomposition. Journal of Experimental Psychology: Learning, Memory, and Cognition, 22(6), 1443. Snyder, A., Mitchell, J., Ellwood, S., Yates, A., & Pallier, G. (2004). Nonconscious idea generation. Psychological Reports, 94, 1325–1330. Steidle, A., & Werth, L. (2013). Freedom from constraints: Darkness and dim illumination promote creativity. Journal of Environmental Psychology, 35, 67-80. Sternberg, R., & Davidson, J. (1995). The nature of insight. Cambridge, Mass.: MIT Press. Subramaniam, K., Kounios, J., Parrish, T., & Jung-Beeman, M. (2009). A Brain Mechanism for Facilitation of Insight by Positive Affect. Journal Of Cognitive Neuroscience, 21(3), 415-432. http://dx.doi.org/10.1162/jocn. 2009.21057 Thomas, L. E., & Lleras, A. (2009). Swinging into thought: Directed movement guides insight in problem solving. Psychonomic bulletin & review, 16(4), 719-723. Topolinski, S. & Reber, R. (2010). Gaining Insight Into the "Aha" Experience. Current Directions In Psychological Science, 19(6), 402-405. http:// dx.doi.org/10.1177/0963721410388803 Topolinski, S., & Strack, F. (2009b). The architecture of intuition: Fluency and affect determine intuitive judgments of semantic and visual coherence, and of grammaticality in artificial grammar learning. Journal of Experimental Psychology: General, 138, 39– 63. Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for judging frequency and probability. Cognitive psychology, 5(2), 207-232. Tversky, A., & Kahneman, D. (1975). Judgment under uncertainty: Heuristics and biases. In Utility, probability, and human decision making (pp. 141-162). Springer Netherlands.


42


Watkins, M. J., & Gibson, J. M. (1988). On the relation between perceptual priming and recognition memory. Journal of Experimental Psychology: Learning, Memory, and Cognition,14, 477–483. Watkins, M. J., & Peynircioglu, Z. F. (1990). The revelation effect: When disguising test items induces recognition. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 1012–1020. Weaver, K., Garcia, S. M., Schwarz, N., & Miller, D. T. (2007). Inferring the popularity of an opinion from its familiarity: a repetitive voice can sound like a chorus. Journal of personality and social psychology, 92(5), 821. Webb, M. E., Little, D. R., & Cropper, S. J. (2016). Insight is not in the problem: Investigating insight in problem solving across task types. Frontiers in Psychology, 7. Wegner, D. & Wheatley, T. (1999) Apparent mental causation: Sources of the experience of the will. American Psychologist 54:480–91. Wegner, D. (2002) The illusion of conscious will. MIT Press. Weisberg, R. (1992). Metacognition and insight during problem solving: Comment on Metcalfe. Journal Of Experimental Psychology: Learning, Memory, And Cognition, 18(2), 426-431. http://dx.doi.org/ 10.1037//0278-7393.18.2.426 Weisberg, R. W., & Alba, J. W. (1981). An examination of the alleged role of" fixation" in the solution of several" insight" problems. Journal of Experimental Psychology: General, 110(2), 169. Weller, A., Villejoubert, G., & Vallée-Tourangeau, F. (2011). Interactive insight problem solving. Thinking & Reasoning, 17(4), 424-439. Wells, G. & Petty, R. (1980) The effects of overt head movements on persuasion: Compatibility and incompatibility of responses. Basic and Applied Social Psychology 1, 219–30. Whittlesea, B. W., Jacoby, L. L., & Girard, K. (1990). Illusions of immediate memory: Evidence of an attributional basis for feelings of familiarity and perceptual quality. Journal of Memory and Language, 29(6), 716-732.


43


Whittlesea, B. W., & Leboe, J. P. (2000). The heuristic basis of remembering and classification: fluency, generation, and resemblance. Journal of Experimental Psychology: General, 129(1), 84. Whittlesea, B. W., & Williams, L. D. (1998). Why do strangers feel familiar, but friends don't? A discrepancy-attribution account of feelings of familiarity. Acta psychologica, 98(2), 141-165. Winkielman, P., & Cacioppo, J.T. (2001). Mind at ease puts a smile on the face: Psychophysiological evidence that processing facilitation leads to positive affect. Journal of Personality and Social Psychology, 81, 989–1000. Winkielman, P., Zajonc, R.B., Schwarz, N., 1997. Subliminal affective priming resists attributional interventions. Cognition and Emotion 11 (4), 433– 465. Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of personality and social psychology, 9(2p2), 1.


44