Judgment under radical uncertainty: Epistemic rational heuristics

• Published date: 01 December 2023
• Author: Anna Grandori
• Date: 01 December 2023
• DOI: http://doi.org/10.1111/emre.12624

EDITORIAL

Judgment under radical uncertainty: Epistemic rational heuristics

INTRODUCTION

The path leading to this Special Issue (SI) is a long one.

For me personally, it is a life long research program:

the first international article I ever published

(Grandori, 1984) made two points: that decision models

can be more fruitfully seen as contingent decision strate-

gies (depending on the state of knowledge and the configu-

ration of interests) rather than “rival paradigm”; and that

there was a conspicuous “hole”in our decision-making

equipment: models thatcombine the discovery and genera-

tive capacity of heuristic reasoning with logically defend-

able procedures. The guest editors of the present SI were

involved already in various roles in a prior SI in memory

of Herbert Simon (Foss, 2001;Grandori,2001;

Hatchuel, 2001), intended to develop Simon in directions

that, albeit present in his thought, have been neglected.

One of those was his methodological and epistemological

legacy (Simon, 1977). In that part of his writings, it is clear

that the “bounds”to rationality are not just limits in com-

putational capacity but include the limits to objective per-

fect knowledge that characterizes any knowing activity,

including scien tific discovery. Therefore, an eff ective

response to uncertainty is not always to resort to mental

“shortcuts”(as heuristics are intended in cognitive psychol-

ogy) (Kahneman et al., 1982) but can be the use of sound

“methods for discovery”(as heuristics are intended in

logics and philosophy of science) (e.g., Kiss, 2006;

Lakatos, 1970). For example, taking issue with a diffused

“mystical view towards discovery,”Simon retrieved from

Pierce the heuristic of “retroduction”(most commonly

referred to as “abduction”) as one of the logically sound

procedures “for the systematic processes leading to dis-

cover”(Simon, 1973).

There can be little doubt that when the decisions to be

taken are not only uncertain and unique, but also impor-

tant, following logically sound methods is desirable—as

we have been reminded by “black swans”and other poorly

imaginable, high risk events, such as Covid and the war in

Ukraine. Probably, we would not like fast and frugal heu-

ristics to guide sanitary vaccination policies. We would

like strong rather than weak thought, powerful and low

error rather than effort saving heuristics to guide responsi-

ble decision makers and allof us.

However, a common tenet in our field is that the

stronger uncertainty becomes, the weaker the form of

rationality that can be used. Behavior is expected to shift

from more “complex”towards “simpler”ways of

reasoning that have been demonstrated to work well in a

specific domain or type of problem, such as estimating

distances or entering an international market

(Bingham & Eisenhardt, 2011; Gigerenzer et al., 2022).

This approach has come to be defined as “ecological

rationality”: the natural selection of behaviors based on

experience, based on adaptation to specific environments,

in which they demonstrably performs as nicely or even

better than other strategies.

The problem is that important uncertain decisions—

from entrepreneurial ventures to grand public

challenges—are typically of a kind that does not permit

the ecological selection of good rules andthe accumulation

of experience (Felin & Zenger, 2009). In addition, experi-

ential rationality shares with deductive optimizing ratio-

nality the feature of not being generative of new solutions.

Hence, an alternative form of rationality is needed that in

the course of being heuristic and generative is based on

sound logic and valid knowledge (Grandori, 2010).

Smith (2003), in his Nobel Lecture, characterized

such a broad family of models, alternative to those based

on ecological rationality, as forms of “constructivist

rationality,”including all models based on the “conscious

and deliberate use of reason”in an intendedly logically

and epistemically sound way. Therefore, constructivist

rationality should not include only models based on

deduction or maximization, with which rationality is

often confused. Actually, as Savage (1954) made crystal

clear that only a limited sub-sets of “simple”problems

may be rationally addressed by utility maximizing proce-

dures: where the “small world”represented in a model is

a satisfactory representation of the real “grand world.”

Analogously, in spite of the current popularity of a

Bayesian approach nurtured by subjective probabilities,

foundational works in logic have argued that where there

is no frequency, no experience, no statistics, and no logi-

cal basis for assigning probability is more rational not to

assign numbers at random rather than to settle every-

thing by “subjective”probabilities: this is arguably the

meaning of “Knightian uncertainty”as a situation where

probabilities “cannot”be assigned. If interpreted in that

way, Knight’s insight is consistent with Popper’s decisive

criticism of probabilism as applied to the logic of discov-

ery in empirically based science, that is, as applied to

unbounded problems: in those settings, the probability of

a hypothesis being true is always zero, as the number pos-

sible falsifiers, the denominator, is always infinite. And

this is why “Knightian uncertainty”can be considered a

DOI: 10.1111/emre.12624

European Management Review. 2023;20:619–625. wileyonlinelibrary.com/journal/emre © 2023 European Academy of Management. 619