In a world that increasingly depends on scientific knowledge and faces detrimental threats such as the climate crisis, a scientifically literate citizenry is of great importance (Dunwoody, 2009). An increasing number of people use social media to inform themselves about scientific topics (ARD/ZDF-Forschungskommission, 2023; Pew Research Center, 2018). However, studies show that the reception of information on these platforms does not lead to actual knowledge acquisition (Amsalem & Zoizner, 2023). Instead, social media usage seems to selectively increase users’ self-assessment of knowledge (so-called meta-knowledge) which can lead to an overestimation of one’s knowledge (Lee et al., 2021; Schäfer, 2020).

The term meta-knowledge is derived from the broader term metacognition (Flavell, 1979), which describes an individual’s self-perception of his or her own cognitive processes. This selfperception arises through so-called metacognitive monitoring processes (Nelson & Narens, 1990). Since humans do not have complete and direct access to their object-level cognitions, monitoring processes rely on a multitude of heuristics such as subjective experiences arising during the processing of information (experience-based monitoring) and naive theories about which situations lead to knowledge acquisition (information-based monitoring; Koriat, 2015). Unfortunately, these heuristics are not always correct and might favor miscalibrated metacognitions. Erroneous metacognitive monitoring processes might also underlie findings that argue that social media use increases meta-knowledge independent of objective knowledge gains (Lee et al., 2021; Schäfer, 2020). However, the existing literature lacks explicit theories about which monitoring processes play a role in social media usage. This dissertation therefore theoretically models different instances of metacognitive monitoring on social media and presents empirical investigations thereof in Studies 1 and 2.

Study 1 presents a naturalistic field experiment that contrasted (meta-)knowledge acquisition between a social media and a more traditional media context. We found that while participants on social media learned less than participants receiving the same information in an email newsletter, participants’ meta-knowledge mirrored these differences. This contradicts notions that social media generally favors an "illusion of knowledge" (Schäfer, 2020). Furthermore, learning success on social media could be increased to the level of the newsletter condition through fostering elaborate information processing, but this increase in objective knowledge was not adequately mirrored in meta-knowledge levels. Study 2 investigated how one specific characteristic of social media messages, that is, attitudinal congruence between the message and the recipient’s prior opinion, influences meta-knowledge ratings. Participants in this study felt more knowledgeable after viewing the congruent, compared to the incongruent, stimuli. This was explained by differences in processing fluency (Oppenheimer, 2008) that influenced meta-knowledge through experience-based monitoring. Taken together, these studies suggest that individuals reflected learning differences between social media and traditional media in their meta-knowledge. However, this perception can be distorted by characteristics of the message, such as attitude congruence and was not responsive to usage styles which did affect objective knowledge acquisition. In addition to the object-level influencing the meta-level through metacognitive monitoring, Nelson and Narens’ (1990) metamemory framework also assumes the reverse flow of information, that is, metacognitions influencing actual behavior or objective cognitions. These processes are called metacognitive control. I examine these control processes at two different stages of informal learning processes on social media. First, I see them as antecedents of learning in situations where meta-knowledge influences future information processing (Griffin et al., 1999). Second, I see them as outcome of learning when meta-knowledge influences non-information-related behaviors such as political participation (Lee et al., 2021). In Study 3 of this dissertation, I tested whether higher climate change meta-knowledge increases or decreases selective exposure to and selective elaboration of related information. I also examined effects on non-information-related behaviors such as climate-related political participation. Although I found no group differences in any of the dependent variables, correlational and mediation analyses tentatively support a positive relationship between meta-knowledge and engagement with further information. With respect to noninformation-related behaviors, however, our results are less conclusive and call for further research.

By combining an extensive literature review, novel theoretical considerations, and empirical findings, this dissertation provides a comprehensive theoretical framework and empirical investigation of metacognitive processes in informal science learning on social media. In the General Discussion, I argue that an overestimation of one’s knowledge through social media use is most likely due to inappropriate experience-based monitoring processes. Furthermore, effective learning seems possible when social media users receive the appropriate information and engage in cognitive elaborate processing. As a practical consequence, users should be educated on how to better monitor their own knowledge acquisition on social media. In addition, they should be informed about how to make learning on social media more effective. In this way, they can use this medium in a more conscious way, either to increase learning success or to reflect a lack of learning in their meta-knowledge judgments.