How to describe environments?

Simon (Reference Simon1990) stressed that “Human rational behavior …is shaped by a scissors whose two blades are the structure of task environments and the computational capabilities of the actor” (p. 7). Together with Newell, he developed a research program that characterizes those blades.

Narratives could serve to capture and transmit actionable insights for tackling ill-structured problems (Simon & Newell, Reference Simon and Newell1958). In such problems, “the objective function, the goal, is vague and nonquantitative” (p. 5), key variables are non-numerical, and “computational algorithms … are not available” (p. 5). Simon and Newell (Reference Simon and Newell1958) contrasted such problems to well-structured ones, which “can be formulated explicitly and quantitatively, and …be solved by known and feasible computational techniques” (p. 5). Their typology of well-structured versus ill-structured problems is orthogonal to the Knightian (Knight, Reference Knight1921/1971) contrast between uncertainty and risk Johnson et al. focus on: Both ill-structured and well-structured problems can come with uncertainty. What matters for decision-making is not just uncertainty – and be it Donald Rumsfeld's surprising “unknown unknowns” (Mousavi & Gigerenzer, Reference Mousavi and Gigerenzer2017, p. 363) – but also the degree and type of structure a problem exhibits. Most challenging real-life problems do not come with clearly delineated boundaries; rather, they are undefined, they can shapeshift and evolve dynamically over time, or be interchained with, and leading into, new problems. Particularly social task environments – which Johnson et al. discuss – exhibit such structural fuzziness.

How do people select among different strategies for action?

One may speculate that narratives allow recognizing key elements of ill-structured problems. Recognizing one's task environment and patterns therein may help knowing what strategies one should rely on. Indeed, Simon (e.g., Simon, Reference Simon1990; Simon & Chase, Reference Simon and Chase1973) stressed how important recognition processes are for human decision-making and rationality. My hypothesis is that narratives serve what is known as strategy selection in the decision sciences (e.g., Marewski & Schooler, Reference Marewski and Schooler2011).

Moreover, narratives may inform us not only about what we should do, but also about what we can do in the first place. The latter function (‘the can') is a prerequisite for selection (‘the should'): Narratives likely transmit actionable insights; they are vehicles for passing on heuristics and other decision-making strategies across generations; they aid filling our toolbox of strategies. Think of holy books, fairy tales, or folk wisdom: “Do not put all your eggs in the same basket” (Hafenbrädl, Waeger, Marewski, & Gigerenzer, Reference Hafenbrädl, Waeger, Marewski and Gigerenzer2016, p. 218), “Do to others as you would have them do to you” (The Bible, Luke, 6:31), and other well-known simple guiding principles are heuristic decision-strategies (e.g., Marewski & Hoffrage, Reference Marewski, Hoffrage and Viale2020). Polya (Reference Polya1945/2014) – a founding father of the study of heuristics whose insights shaped Newell's and Simon's work on heuristics (Dick, Reference Dick2015) – saw how ancient proverbs and heuristics of more modern days correspond.

Many such decision tools have fuzzy scopes and their task environments fuzzy boundaries. Such lack of definiteness could be functional: One may speculate that fuzziness allows for those narrated descriptions of tools and environments to attach to different kinds of ill-structured real-world situations a decision-maker may experience.

Indeed, communication, selection, and pattern detection all form part of Conviction Narrative Theory; there may be room for an integrative research program.

Developing Conviction Narrative Theory's narrative

Such a program warrants precise definitions of concepts concerning the environmental and mental blades, and those theoretical elements must fit to each other, analogously to how two blades must fit to cut. Currently, Johnson et al. invoke uncertainty to characterize environments only broadly. Future developments of Conviction Narrative Theory could specify different types of environments further, with structure being one dimension. Likewise, one could zoom into specific models of heuristics and organize the toolbox of different types of heuristics as a function of the description of environments available. For instance, most fast-and-frugal heuristics (Gigerenzer, Todd, & the ABC Research Group, Reference Gigerenzer and Todd1999) are algorithmic models that can be applied to well-defined problems that feature uncertainty. Classification and probabilistic inference tasks are examples. Heuristics that come as qualitative guiding principles – as in proverbs – may help to manage diverse ill-defined tasks, with more fuzzy boundaries. For example, qualitative guiding principles such as the one described in Luke (6:31) may serve leadership and cooperation alike, and be it in business or in warfare (see also Marewski & Katsikopoulos, Reference Marewski and Katsikopoulos2022).

Currently, Conviction Narrative Theory invokes heuristics too nonspecifically, brushing across Tversky and Kahneman's (Reference Tversky and Kahneman1974) error-prone shortcuts and heuristics as ecologically grounded models of bounded rationality (e.g., Gigerenzer & Goldstein, Reference Gigerenzer and Goldstein1996). Only the latter reflect Simon's (Reference Simon1990) scissors (see Petracca, Reference Petracca2021, for a discussion). Also accounts of more complex tools need detail. Bayesianism is an example: Johnson et al. do not distinguish between Lindley's (Reference Lindley1983) “Universal Bayes” (Gigerenzer & Marewski, Reference Gigerenzer and Marewski2015, p. 431), allegedly applying to all environments, and Savage's (Reference Savage1954/1972) domain-specific Bayesian decision theory. The latter sets apart “small” and “grand” worlds (e.g., p. 84), complementing Knight's contrast between risk and uncertainty (see Binmore, Reference Binmore2017), and taking into account “human possibility” (p. 16), not too inconsistent with Simon's notion of bounded rationality. Future work could deep-dive into such earlier theories of decision-making, examining if and how they would fit Conviction Narrative Theory.

All of the above might translate into two tasks: (1) Theory integration, and (2) setting apart, in detail, Conviction Narrative Theory from alternative lines of thought. Tackling both tasks may aid to better transmit the narrative Conviction Narrative Theory itself represents, allowing researchers to better understand and work with that narrative.