Insight in the Conspiracist’s Mind

Introduction

Believers of conspiracy theories are regularly depicted either as passive, gullible victims of their (social media) echo chambers or as “wishful thinkers” who just believe what makes them feel good or what allows them to belong to a social group (Cassam, 2018; Cinelli et al., 2022; Douglas et al., 2017; Sunstein & Vermeule, 2009). The rest of us, it is often assumed, are largely led by “pure” epistemic motivations (Pronin, 2007): We stick to evidence when forming our beliefs and actively seek additional information whenever we are not sure about them. Scientists would be forgiven for going along with this portrayal, since they, as epistemic authorities, are often the epitome of what the conspiracist distrusts and rejects. Add to that the utter weirdness of some conspiracy beliefs (Williams, 2021), and it is perhaps understandable that we, as scientists, often focus on the errors in conspiracy theories, be it the factual, the moral, or the cognitive ones (such as irrational biases; Adam-Troian, & Caroti, 2019; Brotherton & French, 2015; Cassam, 2018; Douglas et al., 2016; Georgiou et al., 2019, 2021b; Rizeq et al., 2021; Stasielowicz, 2022; van Prooijen & van Vugt, 2018).

Stumped by the attraction that conspiracy theories can have on people, it is easy for us to slip into a discourse about the “mentally ill” or “cognitively deficient” conspiracists, still common in the media and even in parts of the scientific literature (Bratich, 2008; Groh, 1987; Robins et al., 1997; Stasielowicz, 2022). Jumping to such conclusions can easily lead us into the so-called Enlightenment trap (Gray et al., 2022), where we paint ourselves as the ones with intellectual modesty, while they have the “foreclosed mind” (but where the “humble ones” may turn out to be the more prejudiced ones; see Colombo et al., 2021). This othering of people who believe in conspiracy theories can be especially harmful because research shows that conspiracy beliefs are more prevalent, as we will see, in already disadvantaged or epistemically excluded communities.

Such othering is also scientifically problematic because to do good research as social scientists, we need to be able to rely on a minimum of trust in our research subject (that goes both ways). Trust is required even if we are not interested in changing conspiracists’ minds but merely want to study their minds, whether by inviting them to the lab for experimental studies or by engaging with them in ethnographic fieldwork, surveys, or in-depth interviews.

It is why we welcome the growing calls to shift away from a sole focus on the errors of conspiracists in recent years (Alper, 2022; Alper & Imhoff, 2022; Douglas & Sutton, 2022; Harambam & Aupers, 2015; Hornsey et al., 2022). There is an increasing realization that this focus is both a nonstarter when trying to reconnect conspiracists to society and a simplification of the conspiracist’s actual psychology. It leads to a one-sided emphasis on how to debunk isolated, pernicious beliefs (Cook & Lewandowsky, 2012; Ecker et al., 2022) rather than on how those beliefs are actively acquired and shaped through the believers’ epistemic, information-seeking practices. Indeed, it is clear that while conspiracy believers do tend to believe in more than one conspiracy theory (Lewandowsky et al., 2013), they will not simply believe anything (Douglas et al., 2019; Klein et al., 2018). In fact, in informal self-reports, conspiracy believers usually show a high need to be cognitively stimulated and are extremely curious about exploring their domains of interest (McDonald-Gibson, 2022). In modern times, this need is expressed through active participation in online discussion forums, where conspiracists collect and weigh the evidence for their theories (de Wildt & Aupers, 2023; Klein et al., 2018, 2019; Levy, 2022). Correlational, questionnaire-based evidence, too, suggests they pride themselves on their independent, analytical, and discerning thinking (Georgiou et al., 2021c; Tomas et al., 2022) and on their openness to experience (Swami et al.,2010, 2013).

Instead of readily dismissing these reports as empty posing or as rhetorical tricks to give themselves the semblance of scientists, we start in this perspective paper from the position that there is more to the conspiracist’s frequent call to just “Do your own research” (Carrion, 2018; Levy, 2022). As we will see, looking at their epistemic motivations (curiosity), epistemic practices, and epistemic experiences (such as discovery or aha experiences), reveals more sophisticated epistemic processes in conspiracists than traditionally assumed. Many previous works have reserved a role for epistemic needs in the adoption of conspiracy beliefs, but the focus has almost exclusively been on establishing that conspiracists have a closed-minded, “self-sealing” thinking style characterized by an avoidance of uncertainty and a craving for immediate closure or order (Biddlestone et al., 2022; Goertzel, 1994; Hornsey et al., 2022). Little attention has been given to applying to conspiracists what we know about actual epistemic, uncertainty-driven processes from basic experimental psychology. Our hope is that doing so will add experimental, mechanistic ideas to a predominantly correlational literature.

In what follows, we aim to lay the theoretical groundwork for such a new perspective on conspiracy theories. In the next part, we survey the conspiracists’ epistemic (re)searches to show that a theoretical account focused on epistemic practices and experiences is called for. Indeed, we end this part by discussing why those practices and experiences are hard to fit with popular ideas—from the public sphere and scientific literature—on conspiracy belief formation. The section on “The nature of discovery” contains the core of our proposition and applies findings from the experimental psychology of epistemic experiences (curiosity and aha experience) to conspiracy theories. Next, we give an overview of the new, empirically tractable hypotheses that our analysis offers. In the subsequent section, we consider what implications our proposal has concerning the deeper, societal roots of conspiracy thinking. Here, we hypothesize that the need for autonomous epistemic agency and discovery and so the attraction of conspiracy theorizing builds up in people experiencing rising uncertainties and/or perceived epistemic exclusion. With the latter, we mean the sense that one’s own knowledge-gathering and insights have no place in society. Before concluding, we rebut possible objections one could muster against our proposition. Along the way, we illustrate how epistemic needs are connected to existential needs (need to have some control over the world) and social needs (need to belong to and be recognized by a social group) as they contribute to conspiracy beliefs.

With this, we hope to show that epistemic processes of curiosity and aha are promising but overlooked elements in explaining conspiracy thinking. It is an effort to bring out the inner logic of the formation of conspiracy belief systems. One that starts from a commonality in human experience—the need for autonomous epistemic action and discovery—but that also clarifies the potential of those very processes to lead to a narrowed world and an ossified mind. It redirects the focus from detecting (and targeting) cognitive or personality profiles that cause people to “fall for” conspiracy theories to identify the conditions under which conspiracy-like beliefs—characterized by what we will call short epistemic arcs—tend to be proactively adopted.

Charting the Conspiracist’s (Re)Searches

Conspiracy theories are commonly defined as beliefs “that certain events or situations are secretly manipulated behind the scenes by powerful forces with negative intent” (European Commission, 2020). Based on large-scale surveys, it is estimated that about one in four US adults endorses at least some conspiracy ideas (Freeman & Bentall, 2017; Mitchell et al., 2020). One conspiracy theory that has recently gained a lot of disciples is QAnon, sometimes called an “Ur-conspiracy theory” because it integrates so many older, more scattered conspiracy theories. However, informal observers (Hon, 2020) have remarked that QAnon is not just a conspiracy theory but also a true knowledge-generating and problem-solving community. Cryptic messages and “mysteries” (known as “Q drops”) launched by Q and others in the QAnon conspiracy network spur individuals to “do their research,” to go hunting for clues, connections, and explanations. Conspiracy theories become game-like in this sense (Hacker, 2021; Hon, 2020), applying a clever, non-patronizing way of recruiting and engaging members.

The starting cue is a challenge, a question, an expectation violation, or something that does not fit one’s model of the world (yet). It incites curiosity, which can be described as a perceived gap or uncertainty (Loewenstein, 1994), but a gap for which one also has the feeling or expectation that it is resolvable (Van de Cruys, Damiano, et al., 2021). Curiosity is thought to be driven by expected learning (or information) gains (Gottlieb & Oudeyer, 2018). It is the feeling that it is possible, with some effort, to unearth the regular pattern underlying this challenge. Indeed, empirical work shows that curiosity is intrinsically associated with agency, that is, the need to actively search for the answer oneself (Metcalfe et al., 2021), as opposed to just wanting to receive the solution.

But what are those actions? There is a lot of speculation, but not much is known yet about how conspiracists seek information and research their problems. Although Google, Youtube, and the like certainly can lead people down dangerous rabbit holes (Hosseinmardi et al., 2021; Roose, 2019), the internet has greatly democratized our capacity to do our own research. Crucially, we hypothesize that online search is experienced by the internaut as a largely autonomous epistemic practice that can lead to real personal discoveries, especially if it takes some effort and trial-and-error. Support for this comes from a study with non-conspiracist participants finding that the very act of making internet searches leads to an illusion of knowledge (overconfidence), compared to just passively being presented with the same knowledge (Fisher et al., 2015 experiments 4a-c). Search results seem to be experienced as pieces of “knowledge” that you have uncovered through your knowledge-seeking activities. The fact that online search is actually, to a large extent, governed by algorithms and other sources of bias (Johnson, 2021; Narayanan & De Cremer, 2022) does not seem to matter much (or is too opaque) in one’s experience of this research activity. Indeed, Fisher et al. (2015) showed that even if you explicitly and specifically direct people what to search for, the overconfidence effect of the act of searching holds.

Still, presumably, the conspiracist has a salient intuition that ease or “fluency” with which one found something is not a good guide here: Information gathered without some work (“your own research”) done, seems untrustworthy, not unlike the scientist’s attitude. This may explain the effort they go through in finding the “right” channels and alternative search engines and tactics (Golebiewski & Boyd, 2019; Urman et al., 2022), dismissing mainstream ones as biased and doctored for fluency. Official stories of events are suspect for them, presumably precisely because they are too easy to come by. Instead, perceived truth is attached to the results of a proactive information-seeking effort (see subsequently). Indeed, some real effort and creativity are needed to find and fit the same conspiracy-based explanatory “patterns” and “rules” with recurring central people, places, and events to new data or challenges. However baseless or bizarre their claims may be, conspiracists often have well-developed and sophisticated arguments for their theories (e.g., 9/11-Inside-Job believers; Clarke, 2002; Dentith, 2018; Meuer et al., 2022). Conspiracists build their own quasi-scientific methods and communication channels, complete with conferences (Fenster, 1999), publications (Garry et al., 2021), and tutorials on the proper epistemic practices, that is, how to properly collect and analyze data (Lee et al., 2021; Levy, 2022). Indeed, many conspiracists go to much greater lengths to collect and evaluate evidence to support their alternative explanation (and undermine the official story) than adherents of the official story do (Brashier, 2022; Harris, 2018).

As far as most of the information-seeking and inference activity of the conspiracist is expressed in the secluded activity of browsing and annotating the web, it is of course hard to track in situ. While the product of this activity is often posted on public, minable forums (Klein et al., 2019; Perry & DeDeo, 2021), personal web search patterns are not open but amassed and owned by Big Tech companies. Nonetheless, epistemic strategies on the internet are beginning to be examined empirically, at least in typical, non-conspiracist participants (Lydon-Staley et al., 2021; Robertson et al., 2021) with custom tracking software. Conspiracists may not be eager to have their web activity monitored by distrusted “establishment” scientists, but this initial resistance might be surmountable. Anthropologists and sociologists doing fieldwork in conspiracy communities have noticed that conspiracists are often pleasantly surprised that scientists show an interest in their thinking (Harambam & Aupers, 2015). There may be great scientific value in the study of their information-seeking experiences and practices through qualitative, ethnographic methods (Franks et al., 2017; Lee et al., 2021) as well as through machine learning techniques to mine online search data for patterns of information-seeking. We could use the latter data to give some much-needed empirical substance to frequent speculations that conspiracists would show stronger perseverative behavior (i.e., dwelling or circling back on the same topics) and increased path-dependency (i.e., early information strongly biases later searches) compared to the average internaut (Molnar & Loewenstein, 2021). Although social media posts (e.g., on Reddit) have been mined already (Klein et al., 2019; Perry & DeDeo, 2021; Zannettou et al., 2017), actual internet search data, be it for “free-ranging” or task-dependent searches, has not been looked at yet.

Because our perspective-taking often fails when it comes to conspiracists, we must emphasize basic qualitative research as well, poignantly denoted as perspective-getting (see Eyal et al., 2018): Actually asking conspiracists about their experiences and practices when forming their beliefs. That is not because we assume they necessarily have good insight into the processes that produced their beliefs (people rarely do; Nisbett & Wilson, 1977) but because those experiences serve as our explanandum, constraining our theories. This obviously does not mean being blind to the potential harm of people’s ideas, but it does require approaching them in a charitable, curious way, temporarily suspending judgments of truth and morality.

The Nature of Discovery

A great visual and visceral illustration of our sense of insight or “aha” can be found in so-called Mooney (or “two-tone”) images. These are distorted images that are created by grayscaling, blurring, and thresholding photographs such that only irregular, disconnected patches of black and white seem to remain (see Figure 1). The content is completely obscured or “camouflaged,” until, often with some help, the actual object can be discovered and organized or segmented from the background. At that point, people often have a positive insight or “aha experience” (a “click” of understanding). Usually, they cannot unsee the object; phenomenally, they cannot return to the original disorganized percept of the same input (and will remember the solution when later shown the image).

OPEN IN VIEWER

Figure 1.

Two-tone or so-called mooney images are created by blurring and thresholding grayscale photographs (see the source photograph on the next page). They are examples of one-shot learning: Once you find or are confronted with the solution you cannot unsee it. “Discovery” of the familiar structure in the image, usually gives a positive feeling of insight or Aha-Erlebnis. If unsuccessful, take a look at the solution/source in Figure 2 and return to this one to (hopefully) experience the Aha.

OPEN IN VIEWER

Figure 2.

The source (Solution) image for Figure 1.

From the literature on the topic (Danek et al., 2013; Kounios & Beeman, 2009; Topolinski & Reber, 2010; Van de Cruys, Damiano, et al., 2021; Van de Cruys et al., 2018), we know that an Aha-Erlebnis is an acute positive feeling associated with an epistemic experience (“pieces of a puzzle clicking together”). Hence, aha experiences have also been invoked as a key factor in aesthetic appreciation of stimuli (Muth & Carbon, 2013). In psycho-aesthetics (Graf & Landwehr, 2015; Van de Cruys, Bervoets, & Moors, 2021; Wänke & Hansen, 2015), it has become apparent that disfluencies are important contributors to subjective appreciation: What counts is not so much the momentary or overall fluency of a work of art (i.e., its symmetry, familiarity, or predictability), but rather the relative processing fluency. So an initial increase in uncertainty, experienced as disfluency, non-comprehension, or effort, is a necessary element in the generation of an aha or insight experience (Auble et al., 1979) that explains the inverted U-shaped curve relating complexity/unpredictability and liking (Walker, 1981).

The aha experience is empirically shown not only to increase liking but also the perceived truth of a stimulus (Laukkonen et al., 2020). In aha experiences or discoveries, new beliefs are formed—new regularities discovered in the world—but not just by making ideas or materials as fluent as possible to process. Discovery implies an initial obstacle to understanding, that is, disfluencies instead of mere fluency, that is subsequently overcome. It is an expectation violation, surprise, or conflict that is subsequently subsumed in one’s mental model again (possibly by integration of a new model). This renders the conflict or disparate inputs predictable again, by application of the right “frame,” usually in the form of a sparse explanation of the data (Gaver & Mandler, 1987). Since the early work on ahas by the Gestalt psychologists, this mental shift has been called restructuring (Duncker, 1945; Wertheimer, 1943). The stimuli or problems used to induce ahas in the lab are diverse (Sprugnoli et al., 2017), going from classical object-based puzzles (Duncker’s candle problem, matchstick problems, etc.; e.g., Duncker, 1945) to image-based puzzles (e.g., Mooney images illustrated earlier) to word-based problems, such as the remote associate's task (searching for a fourth word that links three given words; e.g., Stuyck et al., 2021), anagrams, or sentence-based problems (e.g., “The breakfast was excellent because the thread was sticky.” The solution is spider web here; Auble et al., 1979).

Evidently, a subjective sense of discovery or insight, for instance during the conspiracist’s online research, does not necessarily equal truth. For example, in research with Mooney images (Van de Cruys, Damiano, et al., 2021; Van de Cruys et al., 2018), one finds that people at times are very convinced of some illusory content or pattern they discovered in an image. Of course, we have the ground truth for a Mooney image (its source photograph), but after distortion, the image might have just as well come from another object. Indeed, participants usually also have specific ideas about how their perceived object is constructed (coincidentally, of course) from cues in the Mooney image. Similarly, in ahas induced by (purportedly) solving a magic trick, Danek and Wiley (2016) found that false insights regularly happen. Experimentally, it is also relatively easy to induce false ahas, for example, when semantic priming leads to ahas for inaccurate solutions of anagrams (Grimmer et al., 2022). Indeed, anyone will be able to relate, somewhat begrudgingly, to the visceral rushes of insights that later turn out to be inaccurate (Keil, 2006). An aha can be had, but rigorous follow-up research may not be able to validate it.

Ahas as Tools in Conspiracy Thinking

The potential for false insights that have all the characteristics of real ones—increased confidence or perceived truth and a sense of beauty or positive affect—already indicates how useful aha experiences can be as tools for changing minds, for example, in the context of conspiracy theories. But there are three additional features of aha that make it extremely suited for belief formation.

The first is the carryover effect. Consider a typical study of aha’s effects on appreciation or perceived truth. In such an experiment, people are presented with statements that contain a jumbled word (aha-inducer; Laukkonen et al., 2020), for example, “ithlium is the lightest of all metals.” People will rate the accuracy of those statements as higher when they contain such an aha compared to control conditions with a non-jumbled word without an aha (or with only a delay in the appearance of the word). Note that induction of aha is done orthogonal to the object that is evaluated. In the lithium example, for example, solving an anagram of a single word in a statement influences the perceived truth value of a statement unrelated to this puzzle-solving. The aha generalizes beyond the concrete materials used to induce it to materials contiguous with them. While the spatiotemporal extent of this carryover effect has not been examined yet, the potential for suspicious beliefs to piggyback on innocuous ahas at least seems there.

The second feature of aha that is relevant to conspiracy thinking is connected to the so-called generation effect in classical memory studies (Slamecka & Graf, 1978). The idea here, confirmed by empirical studies, is that people remember things better if they have been able to infer or (re)construct them themselves. The principle is used from art to advertising, when a stimulus is actually rendered in a distorted or incomplete way—made less instead of more fluent—in order for people to contribute something while trying to capture the regularities embedded in it. There is a telling convergence here, in that an artwork, an idea, or a brand is made to be experienced as more pleasing, more truthful, and more memorable. The principle can again be illustrated with a perceptual example (Figure 3), although it applies to cognitive processing in general.

OPEN IN VIEWER

Figure 3.

Ambiguous figure from hebb (1949). Why do figures like this appeal? Absences become meaningful (as obstacles or prediction errors) as they allow you to contribute something and complete the figure. In a matter of milliseconds, your curiosity is piqued by the squiggly line, which deviates (prediction error) from what you expect an average/randomly drawn line looks like. This invites you to make additional saccades (a mental resolving effort), which is followed by the aha upon discovery of the two faces

This of course relates to the conspiracy theory’s potential to easily apply to new data, that is, new challenges or disfluencies. The generative potential of conspiracy ideas, also called their cognitive fertility (Thi Nguyen, 2021), is a central subjective marker that we have understood something. Again, it implies that understanding is not about passively absorbing information but about active and creative (re)construction of information (Elgin, 2002; Thi Nguyen, 2021). Being able to construct your own “pieces of knowledge” using a self-devised variation of the conspiracy theory seems to be a key mechanism in the adoption of conspiracy theories (Brashier, 2022; Garry et al., 2021). It gives you a sense of cognitive labor fulfilled and so of trustable knowledge, even if the active involvement or effort is a “mere” mental action with minimal overt action (e.g., eye movement to resolve an image or get a “clue”). It is the resulting sense of ownership of the beliefs formed that is the secret weapon of conspiracy theories and it suggests conspiracy theories are not engineered for clarity or fluency, but for discovery.

This reasoning also aligns with the empirical finding that people “often discount or dismiss the opinions of others too much, and give credence to a belief when it is attributed to themselves” (Dunning, 2019; Morin et al., 2021; Trouche et al., 2018). Make people discover it themselves, make them actively fill in the blanks, and they will reliably incorporate it into their core belief structures, as “part of themselves,” as it were. The boosting of value and ownership by one’s own effort is also reminiscent of the so-called Ikea-effect: We seem to attach more value to something we constructed ourselves (Norton et al., 2012; Stafford, 2021; Tiehen, 2022).

In attaching a new sense of truth to things, discovery experiences seem to inevitably come with a strong mind projection fallacy (James, 1910; Jaynes, 1990), in the sense that what we discover seems to be immediately projected unto the world itself. It is seen as a “fact about reality,” rather than merely to our subjective judgments or experience. Mooney images illustrate this as well because once people have discovered the solution—illusory or not—they seem to situate the content in the image itself, rather than just in their experience of it. Even though just seconds ago, the object was not there for them, ahas seem to firmly anchor ideas or viewpoints “out there” in reality. It is this strong combination of belief projection and belief ownership that characterizes real belief adoption ushered in by aha experiences.

The third and related feature of ahas that may be exploited in conspiracy theories is that they have what Thi Nguyen (2021) has called thought-terminating properties. The discovery experience subjectively marks that some cognitive work has been carried out to completion or closure, and so it receives a mark of understanding or (new) clarity, which tends to stop the need for any further thinking or cognitive elaboration. For example, experiments with Mooney images suggest that this closure after aha is associated with a narrowing or reduced dispersion of the pattern of eye movements (Król & Król, 2018), indicative of reduced information seeking. This raises the (as yet untested) hypothesis that the aha causes a stronger reliance on one’s top-down expectations and schemata, as opposed to new sensory information. The logic here is that the discovery just made validates one’s mental models because substantial information gain was made. This validation increases confidence in one’s models. A confidence that then generalizes or carries over to parts of one’s models that were not specifically at stake in the particular aha.

In sum, these “darker sides” of ahas may be (implicitly) instrumentalized in conspiracy thinking. This list suggests that epistemic experiences using curiosity-based prompts or challenges do more than just pull new people into conspiracy thinking. They are also what makes people stay for the longer term—using aha experiences to consolidate conspiracy beliefs and fulfill people’s sense-making needs. This idea is consistent with Perry and DeDeo’s (2021), who used machine learning to analyze the content of conspiracy-related posts on social media and found that long-term engagement is predicted by posts that provide “synthetic and systematic explanations for the way things are. . .[and] show how a number of distinct pieces of evidence connect together.” Relatedly, in a content analysis of a series of real-world online conspiracy articles, in this case coded by humans instead of machines, Meuer et al. (2022) found, contrary to their expectations, that those articles did not provide less argumentation or less detail on underlying explanatory processes for events or standpoints, suggesting that the epistemic or sense-making experiences are indeed crucial.

Toward Testable Hypotheses

We are now equipped to more precisely formulate the new hypotheses afforded by our account and how they can be subjected to empirical tests. At the core of our account is the role of epistemic arcs in explaining the pull of conspiracy thinking, where an epistemic arc can be described as a sequence of an experience of curiosity followed by an epistemic act (covert internal search or overt information-seeking), and subjective insight. The epistemic act could be minimal, as in the example of the web searches discussed earlier (Fisher et al., 2015). More concretely, the mechanism relies on the creation and resolution of uncertainty. In typical participants, Ruan et al. (2018) found that people liked stimuli like trivia texts more when uncertainty was first created by teasing people with missing information or questions (curiosity). It is well-established that curiosity raised by teasing people also induces better learning (Fandakova & Gruber, 2021; Wade & Kidd, 2019). Hence, an obvious question is whether these processes play out differently in conspiracy thinkers: Do they tend to get curious or get positive feelings associated with uncertainty resolution more easily than typical participants (for non-conspiracy-related materials)?

More crucially, one could look at whether epistemic acts, like internet searches, as Fisher et al. (2015) discussed earlier, increase belief confidence more in individuals with stronger conspiracy beliefs than in typical subjects. Of course, we could also introduce a complete epistemic arc. This would entail explicitly adapting conspiracy (vs. non-conspiracy) texts so as to contain solvable disfluencies (e.g., a few solvable jumbled words) orthogonal to the content of the stories (see also Laukkonen et al., 2020). According to our account, the epistemic act and experience elicited by those challenges should lead to increased plausibility or belief of the story, particularly in conspiracy-prone individuals. As the aha-eliciting process would, in principle, be orthogonal to the plausibility of the content, this procedure would also test the carryover effect discussed previously.

In-depth content analysis of existing online conspiracy texts also has the potential to provide more targeted empirical tests of our hypothesis that conspiracy theories are implicitly engineered for discovery (epistemic emotions) or resolvable disfluencies. For example, one could focus a content analysis on forms of disfluencies such as expectation violations in these narratives compared to non-conspiracy news articles. A recent study by van Prooijen et al. (2022) reports that people find conspiracy narratives more entertaining than non-conspiracy control stories, as rated with an “entertainment” scale composed mostly from dimensions of curiosity (i.e., items about epistemic emotions such as How interesting was this article? How captivating was it? How boring? How engaging? How attention-grabbing?). Given that such epistemic emotions are traditionally connected to uncertainty creation and subsequent resolution—like in the structure of a whodunit story—it would be worth analyzing conspiracy narratives in this way.

We may be able to scale said analysis of the temporal dynamics in uncertainty by automatizing it via natural language processing, given that manual annotation of the unpredictabilities in text segments would be time-intensive (given a large enough corpus of written conspiracy vs. non-conspiracy narratives) and subjective (requiring multiple raters). Recently developed Large Language Models can compute the predicted probability distribution of the next word, at each point in the narrative, and hence can be used to derive a time series of prediction errors for a text (Kumar et al., 2022). Our hypothesis would be that conspiracy narratives would show more variability in predictability, leading to more epistemic emotions across texts, than non-conspiracy control texts.

A second possible way of exploiting online conspiracy-related data to test this account would be looking at the use of internet memes in conspiracy circles. It is well-known that memes are popular, witty, or ironic communication devices of online culture, also prevalent in conspiracy networks (Hernandez Aguilar, 2023). Interestingly, they exemplify aha experiences because, like many forms of humor, they often include an expectation violation (so-called pattern with variation) that is easily overcome by minimal but non-negligible cognitive effort, often resulting in an aha experience and humor. Hence, memes share the generation effect—eliciting your “own” contribution—with other aha experiences. Memes, like other insights, do not spell out the specific ways by which people may reach the “insight,” but the discovery is prepackaged in the sense that with the right starting conditions and minimal cues, the meme creator controls the outcomes (Rao, 2022). The active contribution (“doing your research”) is real, but the autonomy (“own” research) is illusory. A meme can incite a brief insight, but it is not a real explanation. It piques and satisfies curiosity and brings momentary closure, potentially helping to conceal gaps in conspiracy explanations (cf. illusion of explanatory depth).

As vehicles of minimal discovery, memes seem to have naturally evolved as tools of conspiracy theories, to successfully communicate ideas and change minds on social media. In the way they compress ideas and make them easily consumable (Fiadotava et al., 2023; Hernandez Aguilar, 2023), they can of course be exploited for both emancipatory and more nefarious purposes. But as tools for the adoption or radicalization of particular beliefs, they may provide a tractable way to test our ideas in online media, currently the naturalistic context for conspiracy theories. We predict that exposure to conspiracy-related memes can be identified as entry or radicalization points in the process of embracing a conspiracy world.

Beyond memes, one could examine whether conspiracists are more captivated by counterintuitive ideas. Research on the spread of misinformation on social media shows that novel and counterintuitive messages are shared more frequently (Berger & Milkman, 2012; Thorson, 2008). We would hypothesize that this is because people have the feeling they made a discovery and that conspiracists might be especially sensitive to this.

In the previous one, we have consistently referred to the internet as the naturalistic setting for conspiracy theories. But of course conspiracy theories predate the internet. As the epistemic processes we describe generalize beyond the internet, an interesting avenue of research might be to perform historical research, into media use (newspapers, radio, libraries, archives, etc.) by pre-Internet conspiracists, to get a sense of their epistemic needs and practices. That said, the internet undeniably amplified both the cues and the means of doing research (e.g., 9/11 truthers). It is already widely considered to drive conspiracy theories by (a) making exposure to misinformation simpler and (b) facilitating the gathering of like-minded people with unusual views, but we are introducing a new angle: the internet facilitates epistemic arcs. This seems especially obvious in the case of QAnon, which instrumentalized the internet in unprecedented ways to activate people and escalate its effects on society.

Epistemic Progress and Epistemic Exclusion

Where does this discovery-based account of conspiracy thinking lead us when we want to track the deeper origins of conspiracy thinking? Given the hypothesized centrality of having one’s own epistemic experiences in the development of conspiracy thinking, it seems very plausible that this type of thinking is rooted in a deep sense of epistemic exclusion. This urges us to look at both sides of the epistemic divide in society. As many scholars have observed, conspiracy theories are only the symptoms of wider problems in society (Hacker, 2021). It turns the problem of conspiracy theorists or “pirate scientists,” into a challenge for science and other authoritative epistemic voices in society. People feel epistemically excluded when they feel their explanations, stories, or experiences are marginalized, rejected, or even inexpressible in the language of the dominant discourse in a society (Barkun, 2015; Fricker, 2017; Harambam, 2021). They feel that their individual and social knowledge-creating capacities have no place in society. One can predict that communities or people that are relatively epistemically excluded would tend to search for their own forms of knowledge creation and discovery, even if it is in “epistemic aberrations” such as conspiracy theories. Surveys have shown that the adoption of conspiracy belief is indeed more prevalent in epistemically disadvantaged communities. This can be due to being economically or ethnically marginalized (Crocker et al., 1999; Goertzel, 1994; Imhoff, 2015; Imhoff et al., 2022; Parsons et al., 1999; van Prooijen, Staman, & Krouwel, 2018), but can have other causes as well. The important element here is that one feels excluded, not that one is actually excluded, although of course the two often coincide. Here, we sidestep the sociological discussions on whether or not such feelings are merited or not.

For example, with technological advancement in our information society, more and more areas of our lives are dependent on technical expertise—be it scientific, political, juridical, or financial—and on technological interventions, automation, and, increasingly, artificial intelligence (Fischhoff, 2013). While the benefits of these evolutions are well-touted and clear beyond the example, they also threaten to create a democratic deficit. They introduce a lack of insight into what governs our lives and the society we live in, with, paradoxically, a concomitant decrease in perceived agency or control. These evolutions render more and more parts of life partly or completely opaque for a growing number of people.

One notable example of this concerns anti-vax sentiments in mothers of newborns. A recent study shows that mothers often feel that their maternal instinct and knowledge creation (“maternal epistemology”) are rejected in the technical and medical sphere and dominant in society (Carrion, 2018). Many of these women seem to have had one or more bad experiences in the medical world, with complaints or symptoms that were not taken seriously or even altogether denied to exist (Carrion, 2018; Castel, 2022). This history of rejection and ensuing distrust can be the seed for radicalization in conspiracy thinking and the rejection of conventional medicine (e.g., in antivax sentiments) in favor of spirituality ideas (or “conspirituality”; Ward & Voas, 2011). Similar links between negative medical experiences, distrust in medical services, and conspiracy beliefs have been reported in other socially disadvantaged groups (Benkert et al., 2019; Bogart et al., 2021; Jaiswal & Halkitis, 2019). In interviews, these people often are careful not to claim that they have all the evidence to consider their deviating beliefs as the truth; they only claim the right to engage in their own truth-seeking activities (Carrion, 2018; Harambam & Aupers, 2015; Stewart, 1999).

Times of increased change and uncertainties in life ¹ —such as during a pandemic, a first pregnancy, in illness, or old age—may raise the need for autonomous discovery and insight experiences because those experiences provide renewed clarity on how to think and act. So conspiracy theories do not just “fill a gap” in their will-less victims. Times with unexpected increases in uncertainty in how to go about one’s life and the anxiety that comes with it create a need to redress this by one’s own epistemic practices: Information-seeking actions that create knowledge and cause a reduction of uncertainty. Indeed, personal discoveries point to new regularities that allow us to regain a modicum of predictability and control in our life. It is this feeling of epistemic progress that is sufficient, irrespective of whether this progress holds true in the final reckoning; see the earlier discussion on ahas and false feelings of truth.

Frantic checking of the news (also epistemic actions) during pandemics or terrorist attacks, is unlikely to provide the strong insights that people long for in these times (Head et al., 2020). Conspiracy theories, however, might do so, by setting up quests—epistemic arcs as we called them—that do run to their aha-completion. While from the outside, conspiracy theories may seem far-fetched and complex, from the inside, they seem to evoke a sense of manageable knowledge, a domain that can be understood using one’s own epistemic actions. In this way, it may protect you from depression by reinforcing your epistemic agency. Since, in these epistemic quests, your actions are clearly shown to have reliable, satisfying effects, your perceived lack of control or helplessness does not generalize, as it often does in depression (cf. learned helplessness; Lieder et al., 2013). There is correlational data as well as anecdotal self-reports consistent with this protective effect of conspiracy theories for distress or depression (Fountoulakis et al., 2021; Garry et al., 2021), but those findings did not look at causality, nor at the crucial epistemic agency that we propose is behind the effect.

However, note that there is longitudinal evidence that conspiracy theories sometimes decrease well-being in the longer term (Liekefett et al., 2021). This does not need to be at odds with an immediate distress-reducing effect of conspiracy theories (and so possibly epistemic agency), as has been noted before (Douglas et al., 2017). Conspiracy theories provide a sense of control by discovering yourself explanations for things you cannot control. There may be comfort in being able to work out the predictable principles of evil or experienced opposition in the world, even if that evil is unavoidable so does not allow for control. When evil is quasi-unavoidable, we seem to prefer an unjust, structured world with clearly (predictably) localized evil—one that we can get insight into—over a random world (Janoff-Bulman & Yopyk, 2004; Skinner, 2000; Stroeken, 2004; Sutton & Douglas, 2014). Indeed, long-standing experiments show that animals in general prefer and are less distressed by predictable aversive stimuli like electroshocks over unpredictable ones, even if those stimuli are inescapable (Badia et al., 1979; Weiss, 1970). Our preference for predictable evil probably stems from both our capacity to (somatically) prepare for and so soften the blow of negative events and our experience that in the vast majority of everyday contexts, insight into the underlying structure does provide a handle for control as well, even if only in future encounters. This is why epistemic needs or experiences and existential ones cannot be neatly separated from everyday experience.

By enabling epistemic agency in people, conspiracy theories might function as a kind of proximal promise (proxy) for broader agency with respect to adversity. This testable hypothesis would align with the recent finding that solving problems with an aha creates a greater tolerance for uncertainty immediately after the experience, as measured by increased risky choice-making (Y. Yu et al., 2022). We would predict that, while people with conspiracy beliefs may often have a higher intolerance of uncertainty by default (Larsen et al., 2021; Marchlewska et al., 2018, but see, Molding et al., 2016), engagement with conspiracy materials, especially if they appeal to their epistemic agency as described here, would indeed lead to a short-lived lowering of their intolerance of uncertainty as well as a more optimistic outlook. However, in the long-term, conspiracy adoption may, at least in some cases (Liekefett et al., 2021), raise or sustain distress for a number of reasons, such as the full realization of the bleak contents of the discovered worldview, harmful decisions made on the basis of this view (Douglas, 2021), or scorn from family or friends because of these ideas. In addition, getting habituated to shorter epistemic arcs may be the information equivalent of temporal discounting in reward learning (i.e., rewards are perceived to be less valuable the more distant in the future). Expecting the act (effort) of information-seeking to pay off and resolve quickly (at a certain rate) is detrimental to any epistemic quests that span longer arcs, which most undertakings in life require. This may explain why conspiracists seem to escape life to find solace in their conspiracy world and its (online) world-discovering practices (Stewart, 1999).

Objections

Before discussing some implications and future directions, we briefly highlight four possible objections to the view we present.

The Role of Diversity

We assume that the processes described here do not differ across cultures or ethnicities, but our paper is of course based on research in (experimental) psychology, conducted primarily with Western individuals in affluent, Western societies, where conspiracy theories are most apparent and investigated. While the need for autonomous epistemic processes is presumably universal across cultures, this need may not necessarily be expressed in the form of conspiracy theories. As we describe in the fifth section, there are reasons to think that particularly modern, highly specialized, and technological Western societies create the sense of epistemic exclusion that we deem important for the emergence of conspiracy thinking. Hence, we would caution against hasty generalization to non-Western societies.

We ourselves are a neurodiverse and multidisciplinary team, but we all come from and are working in a Western cultural context. We relied on crucial insights from female scholars and from studies done with minority participants when developing our view, extending our scope beyond the usual studies considered in the field, and incorporating elements of standpoint epistemology (Carrion, 2018; Fricker, 2017; S. Harding, 1991). Nonetheless, our paper cannot escape the well-documented underrepresentation of women and minority scholars in citations (Dworkin et al., 2020). For the topic at hand, it is all the more important that the field, in both its researchers and the populations it studies, comes to better reflect the diverse stories of epistemically disadvantaged communities.

Conclusion

We proposed that at the core of the conspiracist’s urge to do their own research is a universal human need for autonomous discovery, for making epistemic progress, using one’s own epistemic actions spurred by the visceral cues that we see in curiosity and discovery (or aha) experiences. If the aha experience indeed comprises a confluence of positive value, perceived truth, and agency—a suddenly increased grip on one’s world—we may want to describe the epistemic arc as a building block of sense-making. Susceptible to deception, yes, but nonetheless essential. It is the experience of overcoming disfluencies through one’s active epistemic foraging, which in turn validates the resulting pieces of knowledge. The fruits of one’s search (the final beliefs) are not even the main point here; rather, the attraction is in having the agency to be able to engage in these quests.

If making subjective epistemic progress is indeed a vital need and not just a researcher’s odd inclination, it may even compensate for actual hardship or lack of control in life. Using conspiracy theories as a response to increased uncertainty or lack of control may sound absurd, given that these worldviews give little actual added grip on your environment. But that is precisely why a focus on epistemic agency and aha experiences makes so much sense. We call for more research into the mechanisms of the aha experience and how they play out in the conspiracist’s information-seeking and belief formation.

If our perspective is borne out empirically, scientific knowledge may need to be sufficiently validated by knowledge creation (discovery) at the individual and the social level to make it “lived” and actionable. This may seem a misplaced statement to a scientist who will insist that the validity of knowledge does not depend on any individual or social psychological processes. But it may be what true adoption of beliefs entails—to feel it as true instead of “just” to know it to be true. Further research will be necessary to see whether, by providing people with accessible science-based aha experiences (e.g., of the kind carefully constructed in the best popular science YouTube channels), we can increase and generalize trust in science and skepticism toward conspiracy ideas.

Keeping would-be conspiracists on board in our shared knowledge-producing systems is sure to remain an enormous challenge for science and society. If our account is on track, the key is then to reduce the perceived epistemic exclusion and to give people a sense of autonomous discovery as well as community-based problem-solving, so they do not need to chase “insights” through conspiracy theories, esotericism, and the like (Parmigiani, 2021). If we manage this, there is great potential for decentralized, community-based problem-solving, a kind of epistemic crowd-sourcing that can be redirected toward issues of great societal relevance. For example, citizen science projects are addressing environmental health issues (English et al., 2018), and amateur researchers in the “open source intelligence” movement (known as OSINT) have documented Russian troop movements during the 2022 Russian invasion of Ukraine (Schwartz, 2022).

The centrality of epistemic progress tells us that such forms of decentralized knowledge creation will always be present in a society, something that is well-understood by, for instance, anthropologists working in traditional communities (Stroeken, 2004) but underestimated in modern society. This may be linked to what some have called the meaning crisis in Western civilization (Vervaeke et al., 2017). Decentralized knowledge creation is also the lifeblood of science of course, so let us not smother it in society, even if it gets a little weird. Let us study how people do their research, capitalize on this hunger, and try to lift their practices by sharing much more than just the products of our own scientific research (Ballantyne et al., 2022).

From afar or close by, most of us know people that have been “sucked into” spiraling systems of questionable beliefs. Despite the ring of this metaphor, the fundamental cognitive capacities of these people did not suddenly break down as they became conspiracy believers. Our curious, questioning nature is also what can pull us out of a conspiracy tunnel, as examples show (Castel, 2022; Garry et al., 2021; McDonald-Gibson, 2022). Ultimately, the view on offer is an optimistic one: The “tunneling drive” is the exception if we only give people the mental space to enact their explorative, sense-making drives.