GENERAL DISCUSSION

Although some conspiracy claims lack empirical support (unverified), others are on the official public record (verified). Yet, research has focused almost exclusively on belief in the former. Across three studies, we examined what predicts the capacity to distinguish between verified and unverified conspiracies, whether discrimination is a stronger predictor of cognitive sophistication than response bias, and whether indiscriminate rejection, like indiscriminate belief, is associated with lower cognitive sophistication. We found that better discriminators consistently outperform poorer discriminators on tests of general knowledge, scientific reasoning and cognitive reflection. When combined in a regression model with other predictors, general knowledge and scientific reasoning ability also remained as significant predictors. Additionally, when flipping the analytic direction, discriminability was consistently a stronger predictor of cognitive sophistication than response bias.

In Studies 2 and 3, we also classified participants into three groups: Good Discriminators, Indiscriminate Believers and Indiscriminate Sceptics. Good Discriminators outperformed the other two groups on each cognitive sophistication measure, whereas Indiscriminate Believers and Indiscriminate Sceptics did not differ significantly on any measure in Study 2. This was also the case in Study 3 once differences in discriminability were controlled for.

Prior research consistently links belief in unverified or implausible conspiracies to poorer domain‐general reasoning (Binnendyk & Pennycook, 2022; Gjoneska, 2021; Lantian et al., 2021; Pytlik et al., 2020; Yelbuz et al., 2022), implicitly reinforcing the idea that rejecting conspiracies must signal cognitive sophistication. However, conspiracy beliefs are commonly defined and measured using only unwarranted or implausible conspiracy theories. By examining belief in both verified and unverified conspiracy claims, we show that blindly dismissing conspiracy claims also appears to be associated with lower cognitive performance. In other words, effective discrimination between verified and unverified conspiracies, rather than indiscriminate belief or rejection, is a better indicator of cognitive sophistication.

This problem may run deeper still. In Studies 2 and 3, conspiracy mentality—a widely used predictor of conspiracy belief—was associated with both a liberal response bias and reduced discrimination, but its association with response bias was significantly stronger. Research relying on conspiracy mentality scales may therefore primarily capture a tendency to indiscriminately endorse conspiracy claims rather than a deficit in discrimination, further compounding the limitations of existing research in this area (see Tagand et al., 2026). That said, response bias showed small negative associations with cognitive sophistication measures in all three studies, such that those who tend to reject conspiracy claims were somewhat more cognitively sophisticated than those who tend to endorse them. This speaks against a notion of strict symmetry between sceptics and believers (see Tagand et al., 2026). The LPAs nevertheless suggest that Indiscriminate Sceptics represent a genuine participant profile that emerges naturally from the data. Crucially, this group displayed comparable cognitive sophistication to Indiscriminate Believers in Study 2, and their advantage in Study 3 disappeared when differences in discriminability were controlled for.

Prior work has indicated that poorer reasoning predicts belief in implausible conspiracies but not plausible ones (Frenken et al., 2024; Hattersley et al., 2022). We conceptually replicate this work, demonstrating that scientific reasoning and cognitive reflection are negatively related to belief in unverified conspiracies but unrelated to belief in verified conspiracies. However, general knowledge was sometimes negatively correlated with unverified conspiracy belief and positively correlated with verified conspiracy belief, suggesting that historical knowledge might help people recognize genuine conspiracies. Both types of belief were also positively correlated with conspiracy mentality and political conservatism and negatively correlated with trust in experts. This evidence suggests that processes driving belief in implausible/unverified conspiracies may differ from processes underpinning belief in plausible/verified ones (Lewandowsky et al., 2018; Stanovich & Toplak, 2025), but also that there is overlap.

We also extend this literature by focusing primarily on conspiracy discrimination, which considers belief in verified claims relative to unverified claims. Consistent with prior work (Bensley et al., 2020; Moulding et al., 2016; Pennycook et al., 2025; but see Wagner‐Egger et al., 2023), we found that verified and unverified conspiracy belief is strongly correlated, highlighting a potential trade‐off between incorrectly believing conspiracies that are false and incorrectly dismissing conspiracies that are true. Using an SDT approach, however, we could separate general credulity (response bias) from discriminability, revealing that discrimination is more closely tied to cognitive sophistication than response bias alone. Additionally, better discrimination correlated with political liberalism, higher trust in experts and lower conspiracy mentality in all three studies. Stronger discriminators also skewed male in Studies 1 and 2, and age positively predicted discrimination in Studies 2 and 3. Further research is nevertheless needed to better understand the correlates of conspiracy discrimination (see Stanovich & Toplak, 2025).

There are two caveats to our SDT approach. First, the ground truth of conspiracy claims is not directly available, and what researchers classify as true or false can reflect subjective epistemological assumptions (Uscinski et al., 2024); some unverified claims may in fact be true and some genuine conspiracies may never come to light given their secretive nature. Claims nevertheless vary in their level of evidential support, and using verified/unverified as a proxy for true/false allows for meaningful examination of how well people differentiate these claims. Second, unlike typical discrimination tasks, participants were not explicitly asked to categorize claims as verified or unverified. Our measure of discrimination is therefore indirect, reflecting beliefs formed over a lifetime of exposure to evidence.

Finally, none of this implies that all conspiracy claims are true or that conspiracy theorists expose genuine conspiracies. Conspiracies are typically uncovered by professional investigators who collect evidence like confessions and official documents (Wagner‐Egger et al., 2019, 2023). When evaluating conspiracy claims that one cannot personally investigate, however, calibrated skepticism—differentiating claims based on evidential support rather than indiscriminate belief or indiscriminate rejection—appears most appropriate.