Central to Johnson et al.'s Conviction Narrative Theory (CNT) is the assumption that reasoners adopt “a narrative that feels ‘right’ to explain the available data” and use “that narrative to imagine plausible futures” (target article, Abstract). Their discussion of what makes a narrative “feel right” focuses on structural aspects of the narrative and neglects the role of metacognitive experiences in the construction and evaluation of narratives. This commentary addresses this gap, drawing on feelings-as-information theory (Schwarz, Reference Schwarz, Van Lange, Kruglanski and Higgins2012; Schwarz & Clore, Reference Schwarz, Clore, Kruglanski and Higgins2007).

Narratives are more compelling when they are easy to construct. People monitor the dynamics of their own thought processes and have more confidence in their thoughts when processing is easy (Koriat, Reference Koriat, Zelazo, Moscovitch and Thompson2007; Schwarz, Reference Schwarz, Mikulincer, Shaver, Borgida and Bargh2015). Numerous content, context, and person variables can influence the subjective experience of fluent processing (Reber, Schwarz, & Winkielman, Reference Reber, Schwarz and Winkielman2004; Schwarz, Jalbert, Noah, & Zhang, Reference Schwarz, Jalbert, Noah and Zhang2021). Some influences are integral to what the person thinks about (e.g., the coherence or complexity of a narrative), whereas others are incidental (e.g., momentary distractions or the readability of a print font). Because people are more sensitive to their feelings than to the source of their feelings, they frequently misread incidental feelings as bearing on what they are thinking about. The inferences they draw from processing fluency are guided by lay theories of mental operations, which are usually correct. People assume, for example, that a narrative is easier to process when it is familiar, coherent, and compatible with other things they know than when it is not. Hence, they infer higher familiarity (Song & Schwarz, Reference Song and Schwarz2009) and coherence (Topolinski & Strack, Reference Topolinski and Strack2009) when processing is easy, and are less likely to notice a claim's incompatibility with their own knowledge (Song & Schwarz, Reference Song and Schwarz2008), even when the experience is solely due to variations in color contrast, print font, or ease of pronunciation. Because familiarity is an important input into judgments of risk and trust, people are also more likely to invest in stocks when the ticker symbol is easy to pronounce (Alter & Oppenheimer, Reference Alter and Oppenheimer2006; Green & Jame, Reference Green and Jame2013), indicating reduced risk perception, and to trust online sellers when their usernames are easy to process (Silva, Chrobot, Newman, Schwarz, & Topolinski, Reference Silva, Chrobot, Newman, Schwarz and Topolinski2017). Moreover, fluent processing due to repetition (Hasher, Goldstein, & Toppino, Reference Hasher, Goldstein and Toppino1977), cultural familiarity (Oyserman, Reference Oyserman, Baumeister and Forgas2019), audio quality (Bild et al., Reference Bild, Redman, Newman, Muir, Tait and Schwarz2021), color contrast (Reber & Schwarz, Reference Reber and Schwarz1999), rhyme (McGlone & Tofighbakhsh, Reference McGlone and Tofighbakhsh2000), related nonprobative photos (Newman & Zhang, Reference Newman, Zhang, Greifeneder, Jaffé, Newman and Schwarz2021), and many other incidental variables increases the acceptance of claims as true (Brashier & Marsh, Reference Brashier and Marsh2020; Schwarz & Jalbert, Reference Schwarz, Jalbert, Greifeneder, Jaffé, Newman and Schwarz2021).

This suggests that only fluent narratives are compelling enough to elicit strong “convictions,” whereas disfluent narratives are likely to seem implausible (Schwarz, Sanna, Skurnik, & Yoon, Reference Schwarz, Sanna, Skurnik and Yoon2007). Debiasing research supports this conclusion. After an outcome is known, people believe that they could have predicted it (Fischhoff, Reference Fischhoff1975). To attenuate this hindsight bias, it is often recommended to consider how the event might have turned out otherwise (Larrick, Reference Larrick, Koehler and Harvey2004), thus prompting the development of an alternative narrative. Because thinking of just a few reasons for alternative outcomes is more difficult than thinking of many, this debiasing strategy is successful when people are asked for a few reasons, but backfires when they are asked for many (Sanna, Schwarz, & Stocker, Reference Sanna, Schwarz and Stocker2002). Even when people successfully generate many reasons for alternative outcomes, the associated difficulty convinces them that the obtained outcome was almost unavoidable. All three results – the emergence of a hindsight bias, its attenuation through generating a few alternatives, and its exaggeration by generating many alternatives – reflect judges' reliance on their own metacognitive experiences and can be obtained with different incidental and integral manipulations of processing fluency (Schwarz et al., Reference Schwarz, Sanna, Skurnik and Yoon2007).

In addition to providing information about one's processing dynamics, fluent processing is experienced as pleasant and elicits a spontaneous positive affective response that can be captured with psychophysiological measures (Winkielman & Cacioppo, Reference Winkielman and Cacioppo2001). Hence, a given stimulus is liked more when it is easy to process, for example, due to repetition (Zajonc, Reference Zajonc1968), visual (Reber, Winkielman, & Schwarz, Reference Reber, Winkielman and Schwarz1998) or conceptual primes (Belke, Leder, Strobach, & Carbon, Reference Belke, Leder, Strobach and Carbon2010), or individual differences in color vision (Álvaro, Moreira, Lillo, & Franklin, Reference Álvaro, Moreira, Lillo and Franklin2015). This affective pathway from fluent processing to liking parallels the influence of happy and sad moods (Schwarz, Reference Schwarz, Higgins and Sorrentino1990) and figures prominently in aesthetic preference and judgments of beauty (Reber et al., Reference Reber, Schwarz and Winkielman2004). The common intuition that beauty and truth go hand in hand reflects that processing fluency serves as an input into both judgments (Schwarz, Reference Schwarz, Proust and Fortier2018).

CNT holds that “explanatory fit is experienced affectively” (sect. 6.2). Processing fluency provides affective as well as non-affective information that informs assessments of what one is thinking about – when thoughts flow smoothly, people nod along (Schwarz & Jalbert, Reference Schwarz, Jalbert, Greifeneder, Jaffé, Newman and Schwarz2021). The observed parallel influence of integral and incidental variables implies that assessments of explanatory fit are highly error prone – because people are more sensitive to their feelings than to the source of their feelings, narratives can feel “right” (or “wrong”) for reasons that are substantively unrelated to the content of the narrative.

For researchers, this source insensitivity provides two useful methodological tools. In Johnson et al.'s examples, narrative content and associated feelings cannot be separated – simulating a narrative that leads to a positive (negative) future elicits positive (negative) feelings and either the content of the narrative or the feeling it elicits may drive subsequent choices. To separate the contributions of content and feelings, one can (1) elicit an incidental feeling, as in the above examples, or (2) undermine the informational value of integral feelings by leading people to attribute their feelings to an incidental source, which attenuates or eliminates the impact of feelings (e.g., Novemsky, Dhar, Schwarz, & Simonson, Reference Novemsky, Dhar, Schwarz and Simonson2007; Schwarz & Clore, Reference Schwarz and Clore1983). Without such manipulations, CNT's causal claims about the roles of narrative content versus the accompanying (affective or non-affective) feelings, and their relative contributions at different stages, remain ambiguous. The further development of CNT will benefit from tackling this task.