HEURISTIC METHODS FOR UPDATING SMALL WORLD REPRESENTATIONS IN STRATEGIC SITUATIONS OF KNIGHTIAN UNCERTAINTY

骑士不确定性战略情境下小世界表示更新的启发式方法

ALBERTO FEDUZI
阿尔贝托·费杜齐

PHILIP FAULKNER
菲利普·福克纳

JOCHEN RUNDE
约亨·伦德

University of Cambridge 剑桥大学

LAURE CABANTOUS City, University of London 伦敦大学劳尔·卡邦图（LAURE CABANTOUS）城市（注：此处“City”可能为专有名词或排版问题，按原文保留）

CHRISTOPH H. LOCH University of Cambridge 剑桥大学 克里斯托夫·H·洛克

Recent studies on the construction and use of “small world representations” in strategic decision-making under Knightian uncertainty say little about how such representations should be updated over the implementation phase. This paper draws on the psychology of reasoning to take a step toward answering this question. We begin by theorizing small world representations and how the scenario spaces they contain are constructed and may be updated over time. We then introduce two well-known heuristic methods of inquiry, disconfirmation and counterfactual reasoning; translate them into practical procedures for updating scenario spaces; and compare the relative performance of these procedures in strategic situations of Knightian uncertainty. Our principal findings are that the procedure based on counterfactual reasoning is superior to that based on disconfirmation with respect to (a) counteracting the confirmation bias, (b) promoting the exploration of the set of imaginable scenarios, and (c) facilitating action to mitigate or exploit the consequences of what would otherwise have been Black Swans. We close with some broader implications for the study of strategic decision-making under Knightian uncertainty. 关于在奈特不确定性下的战略决策中“小世界表征”的构建与运用的最新研究，鲜少涉及这类表征在实施阶段应如何更新。本文借鉴推理心理学，向回答这一问题迈进了一步。我们首先对小世界表征进行理论化分析，探讨其包含的情景空间是如何构建以及随时间推移可能如何更新；接着引入两种广为人知的启发式探究方法——证伪（disconfirmation）和反事实推理，并将它们转化为更新情景空间的实用程序；最后比较这些程序在奈特不确定性战略情境中的相对表现。我们的主要发现是，基于反事实推理的程序在以下方面优于基于证伪的程序：（a）抵消确认偏差，（b）促进对可想象情景集合的探索，（c）便于采取行动以减轻或利用原本可能成为“黑天鹅”事件的后果。我们就此结束，并对奈特不确定性下战略决策研究的更广泛启示进行了总结。

It is widely recognized that decision-makers operating in situations of Knightian uncertainty (Alvarez & Barney, 2005, 2007; Knight, 1921; McMullen, & Shepherd, 2006; Packard & Clark, 2020; Reymen, Andries, Berends, Mauer, Stephan, & Van Burg, 2015) employ simplified cognitive representations, sometimes called “small world representations” (SWRs), to anticipate the longer-term consequences of their strategic decisions (Gavetti, Greve, Levinthal, & Ocasio, 2012; Grégoire, Barr, & Shepherd, 2010; 人们普遍认为，在奈特不确定性（Knightian uncertainty）情境下开展决策的决策者（Alvarez & Barney, 2005, 2007; Knight, 1921; McMullen, & Shepherd, 2006; Packard & Clark, 2020; Reymen, Andries, Berends, Mauer, Stephan, & Van Burg, 2015）会采用简化的认知表征方式——有时被称为“小世界表征”（SWRs）——来预判其战略决策的长期后果（Gavetti, Greve, Levinthal, & Ocasio, 2012; Grégoire, Barr, & Shepherd, 2010;

Levinthal, 2011, 2018; Maitland & Sammartino, 2015a, 2015b; Packard, Clark, & Klein, 2017; Porac, Thomas, & Baden-Fuller, 1989). There has been a good deal of research on the initial construction of such representations. Much of this work has investigated factors that affect decision-makers’ ability to perform this task effectively, such as their experience, expertise, and cognitive processes (Gary, Wood, & Pillinger, 2012; Gavetti & Levinthal, 2000; Grégoire et al., 2010; Lovallo, Clarke, & Camerer, 2012; Maitland & Sammartino, 2015a; Simon, 1997) and the quality and quantity of the information available to them (Gavetti & Rivkin, 2007; Maitland & Sammartino, 2015b). 莱文瑟尔，2011年，2018年；梅特兰和萨马蒂诺，2015a，2015b；帕卡德、克拉克和克莱因，2017年；波拉克、托马斯和巴登-富勒，1989年）。关于此类表征的初始构建已有大量研究。其中大部分研究调查了影响决策者有效执行这一任务的因素，例如他们的经验、专业知识和认知过程（加里、伍德和皮林格，2012年；加维蒂和莱文瑟尔，2000年；格雷瓜尔等人，2010年；洛瓦洛、克拉克和卡梅勒，2012年；梅特兰和萨马蒂诺，2015a；西蒙，1997年）以及他们可获得的信息的质量和数量（加维蒂和里夫金，2007年；梅特兰和萨马蒂诺，2015b）。

The question of how SWRs should be updated during the implementation phase of strategic decisions has, however, received considerably less attention, perhaps because adaptive responses to emerging information are often seen as extempore and ad hoc rather than the result of systematic analysis (Klingebiel & De Meyer, 2013; Langley, Mintzberg, Pitcher, Posada, & Saintmacary, 1995). The question is nonetheless a pressing one in situations of Knightian uncertainty, where the scope for constructing adequate SWRs ab initio is often highly limited, and where decisionmakers relying on heuristics to guide their actions (Bingham & Eisenhardt, 2011; Busenitz & Barney, 1997; Maitland & Sammartino, 2015a; Shepherd, Haynie, & McMullen, 2012)1 often find it difficult to adapt their cognitive representations in a timely way (Tripsas & Gavetti, 2000). 然而，在战略决策的实施阶段，社会稳定风险评估（SWRs）应如何更新的问题却受到了相当少的关注，这或许是因为对新出现信息的适应性回应常被看作是即兴和临时的，而非系统性分析的结果（Klingebiel & De Meyer, 2013；Langley, Mintzberg, Pitcher, Posada, & Saintmacary, 1995）。但在奈特不确定性情境中，这一问题依然非常紧迫：此时从头构建充分社会稳定风险评估的空间往往极为有限，而依赖启发式方法指导行动的决策者（Bingham & Eisenhardt, 2011；Busenitz & Barney, 1997；Maitland & Sammartino, 2015a；Shepherd, Haynie, & McMullen, 2012）1 常常难以及时调整自身的认知表征（Tripsas & Gavetti, 2000）。

The purpose of this paper is to begin answering this question. To this end, we draw on the psychology of reasoning (e.g., Evans, 2016; Manktelow, 2012; Poletiek, 2001) to theorize how two well-known heuristic methods of inquiry—disconfirmation and counterfactual reasoning—might inform procedures for updating SWRs. We show that, in strategic situations of Knightian uncertainty, the procedure based on counterfactual reasoning is likely to be more effective in supporting the process of updating SWRs than the one based on disconfirmation with respect to (a) counteracting the confirmation bias, (b) promoting exploration of the set of imaginable scenarios, and (c) facilitating action to mitigate or exploit the consequences of what would otherwise have been Black Swans (Taleb, 2007). 本文旨在开始回答这一问题。为此，我们借鉴推理心理学（例如，Evans, 2016；Manktelow, 2012；Poletiek, 2001）的理论，探讨两种广为人知的启发式探究方法——证伪和反事实推理——如何为情景依赖型风险评估（SWRs）的更新程序提供指导。我们发现，在奈特不确定性的战略情境中，基于反事实推理的程序在以下三个方面可能比基于证伪的程序更有效地支持SWRs的更新过程：（a）抵消证实偏差，（b）促进对可想象情景集合的探索，以及（c）助力采取行动以减轻或利用本可能成为黑天鹅事件（Taleb, 2007）的后果。

Our argument begins with a detailed exposition of SWRs, focusing particularly on how the associated scenario spaces are constructed and may change as they are updated over time. We then provide a brief account of disconfirmation and counterfactual reasoning as they appear in the psychology literature, translate them into practical procedures for updating SWRs, and compare the relative performance of these procedures in strategic situations of Knightian uncertainty. We conclude with a discussion of the wider implications of our work. 我们首先详细阐述情景-愿望-资源（SWRs），特别关注相关情景空间的构建方式以及在随时间更新过程中可能发生的变化。接着，我们简要介绍心理学文献中出现的证伪和反事实推理，并将其转化为更新SWR的实用程序，然后在奈特不确定性的战略情境中比较这些程序的相对表现。最后，我们讨论了本研究的更广泛意义。

SMALL WORLD REPRESENTATIONS AND KNIGHTIAN UNCERTAINTY

小世界表征与奈特不确定性

The concept of SWRs can be traced to the “small world” idea introduced by Leonard Savage (1954) in his landmark Foundations of Statistics. While the underlying technicalities are considerable and continue to be debated to this day (Binmore, 2008; Bradley, 2017; Shafer, 1986), a small world is most easily understood as a decision situation that has been distilled to a choice of one from a set of courses of action or “acts,” each of which lead to one of a set of “consequences” depending on which of a set of contingent and exogenously determined “states of the word” is obtained. Consequences are regarded as “anything at all about which the person could possibly be concerned” (Savage, 1954: 14) and states as descriptions of the world that leave “no relevant aspect undescribed” (Savage, 1954: 9). The decision-maker is assumed to know the sets of acts, states, and consequences, and the only uncertainty concerns which of the set of states will turn out to be the true one, what is sometimes called state uncertainty (Bradley & Drechsler, 2014). SWRs（小世界关系）的概念可以追溯到伦纳德·萨维奇（Leonard Savage）1954年在其里程碑式著作《统计学基础》中提出的“小世界”思想。尽管其背后的技术细节相当复杂且至今仍存在争议（Binmore, 2008；Bradley, 2017；Shafer, 1986），但“小世界”最容易被理解为一种决策情境，它被简化为从一系列行动或“行为”中选择其一，而每个行为会根据一系列偶然且外生决定的“世界状态”（此处原文“word”应为笔误，应为“world”）中的哪一个成为现实，进而导向一系列“结果”中的某一个。结果被视为“人可能关心的任何事物”（萨维奇，1954：14），而状态则是对世界的描述，且“不遗漏任何相关方面”（萨维奇，1954：9）。决策者被假定知晓行为、状态和结果的集合，唯一的不确定性在于哪一个状态会成为真实状态，这有时被称为状态不确定性（Bradley & Drechsler, 2014）。

It has often been pointed out, even by Savage (1954: 16) himself, that many, if not most, practical decision situations are too complicated to be reduced to small worlds by the decision-maker (Binmore, 2008, 2009; Bradley & Drechsler, 2014; Gigerenzer & Gaissmaier, 2011). That is, in many real-world situations it is simply beyond human ability to arrive at exhaustive lists of the relevant acts, states, and consequences, never mind describe the members of these lists with any high degree of completeness or precision (Gilboa & Schmeidler, 1995, 2001; Levinthal, 2011). Some authors have accordingly written about the problem of decision-making in “large” (Binmore, 2008; Drechsler, 2012) or “grand” (Bradley, 2017) worlds, by which they mean situations in which small world formulations are not feasible. Since strategic decisions typically involve Knightian uncertainty, they will invariably be of this kind (Levinthal, 2011). 人们常指出，即使是萨维奇本人（1954：16）也认为，许多（即便不是大多数）实际决策场景过于复杂，决策者无法将其简化为“小世界”（Binmore, 2008, 2009；Bradley & Drechsler, 2014；Gigerenzer & Gaissmaier, 2011）。也就是说，在许多现实场景中，人类根本无法穷尽列出相关的行动、状态和结果，更不用说以高度的完整性或精确性描述这些列表中的成员（Gilboa & Schmeidler, 1995, 2001；Levinthal, 2011）。因此，一些学者将这种决策问题描述为“大”世界（Binmore, 2008；Drechsler, 2012）或“宏大”世界（Bradley, 2017）——他们所指的是无法采用“小世界”表述的场景。由于战略决策通常涉及奈特不确定性，因此必然属于此类（Levinthal, 2011）。

However, none of this should be allowed to obscure that the broad acts-states-consequences framework provides a natural and quite common way to structure decision-making problems, and in particular that strategic decision-makers often do seem to employ highly simplified cognitive representations that approximate small worlds to guide their actions (Csaszar & Levinthal, 2016; Gavetti et al., 2012). We will follow recent contributions to the strategy and entrepreneurship literature in calling such approximations SWRs (Levinthal, 2011, 2018; Maitland & Sammartino, 2015a; Packard et al., 2017), which we interpret as rough analogs of Savage small worlds insofar as they share the same broad acts—states-consequences structure. 然而，这些都不应掩盖这样一个事实：广泛的行为-状态-结果框架为构建决策问题提供了一种自然且相当常见的方式，特别是战略决策者似乎经常采用高度简化的认知表征，这些表征近似于“小世界”来指导他们的行动（Csaszar & Levinthal, 2016; Gavetti et al., 2012）。我们将遵循战略与创业文献中的最新研究，将这种近似称为SWRs（Levinthal, 2011, 2018; Maitland & Sammartino, 2015a; Packard et al., 2017），我们将其解释为萨维奇“小世界”的粗略类似物，因为它们共享相同的广泛行为-状态-结果结构。

For our purposes, there are two key differences between small worlds and SWRs. The first is that the descriptions of SWR analogs of acts, states, and consequences tend to be partial and vague by comparison to what is assumed in small worlds, and where, importantly, the membership of these three categories is typically incomplete (Alvarez & Barney, 2005; Gavetti & Menon, 2016; Levinthal, 2011; Maitland & Sammartino, 2015a; Packard et al., 2017; Reymen et al., 2015). That is, whereas the uncertainty facing the decision-maker in a small world is limited to state uncertainty, SWRs typically reflect the myriad additional uncertainties characteristic of many real-world decisions, including uncertainties about the nature and number of possible acts, states, and consequences, as well as uncertainties about the probabilities of states. We henceforth use the term Knightian uncertainty to refer to situations in which decision-makers face these additional types of uncertainty to significant degrees.2 就我们的研究目的而言，小世界（small worlds）与情境化世界表征（SWRs）之间存在两个关键差异。首先，与小世界中所假设的情况相比，SWRs对行为、状态和结果的类似物描述往往是部分且模糊的，并且重要的是，这三类（行为、状态、结果）的成员通常是不完整的（Alvarez & Barney, 2005; Gavetti & Menon, 2016; Levinthal, 2011; Maitland & Sammartino, 2015a; Packard et al., 2017; Reymen et al., 2015）。也就是说，小世界中决策者面临的不确定性仅限于状态不确定性，而SWRs通常反映了许多现实世界决策中特有的无数额外不确定性，包括对可能行为、状态和结果的性质及数量的不确定性，以及对状态概率的不确定性。我们此后将“奈特不确定性”（Knightian uncertainty）这一术语用于指代决策者在显著程度上面临这些额外类型不确定性的情境。2

The second key difference between small worlds and SWRs concerns the treatment of time, and thus the scope for learning within each formulation. In a small world, the specification of acts, states, and consequences entails that everything that might happen at all subsequent stages of a decision problem is laid out from the outset. The result is a highly attenuated rendering of time in which “genuine” learning—such as uncovering new, formerly unimagined, possibilities—is precluded (Binmore, 2008; Feduzi & Runde, 2014). SWRs impose no such requirement on their analog acts, states, and consequences, and are therefore consistent with the ongoing, contingent, and adaptive nature of strategic decisions in which it is rarely possible to know everything relevant about what the future might hold from the outset (Alvarez et al., 2013; Packard et al., 2017). SWRs, therefore, offer room for the acquisition of new information over time, and thus the possibility of updating the SWR over the course of the implementation of a strategic decision. 小世界与SWRs之间的第二个关键区别在于对时间的处理，以及每种表述中学习的可能性。在小世界模型中，行为、状态和结果的规范意味着决策问题后续所有可能发生的事情从一开始就被确定下来。这导致时间被高度弱化，其中“真正的”学习——如发现新的、以前未想象过的可能性——被排除（Binmore, 2008；Feduzi & Runde, 2014）。SWRs对其类似的行为、状态和结果没有这样的要求，因此与战略决策的持续性、偶然性和适应性本质一致，在这种决策中，从一开始就可能很少有可能了解未来可能发生的一切相关信息（Alvarez et al., 2013；Packard et al., 2017）。因此，SWRs为随时间获取新信息提供了空间，从而有可能在战略决策的实施过程中更新SWR。

Decision-makers’ ability to construct SWRs in a way that provides a useful guide to action (Csaszar & Levinthal, 2016; Levinthal, 2011) depends on both the cognitive attributes of the decision-maker and the nature of the informational environment (Gary et al., 2012; Gavetti & Rivkin, 2007; Maitland & Sammartino, 2015a, 2015b). Foremost among the former are capacities and dispositions such as imagination, creativity, and curiosity, which are crucial to envisioning possible futures, as well as an actor’s prior knowledge, experience, and expertise, and the reasoning and problem-solving methods used (associative thinking, analogical reasoning, fast-and-frugal thinking, etc.) (Alvarez & Porac, 2020; Arikan, Arikan, & Koparan, 2020; Brandenburger, 2017; Felin & Zenger, 2017; Gavetti & Levinthal, 2000; Grégoire et al., 2010; Lovallo et al., 2012; Maitland & Sammartino, 2015a; Pontikes & Rindova, 2020; Porac & Tschang 2013; Rindova & Courtney, 2020; Simon, 1997). 决策者构建战略意外响应（SWRs）以有效指导行动的能力（Csaszar & Levinthal, 2016; Levinthal, 2011）取决于决策者的认知属性和信息环境的性质（Gary et al., 2012; Gavetti & Rivkin, 2007; Maitland & Sammartino, 2015a, 2015b）。前者最为关键的是想象力、创造力和好奇心等能力与特质，这些对于展望可能的未来至关重要；此外，行动者的先前知识、经验和专业技能，以及所使用的推理和问题解决方法（如联想思维、类比推理、快省思维等）（Alvarez & Porac, 2020; Arikan, Arikan, & Koparan, 2020; Brandenburger, 2017; Felin & Zenger, 2017; Gavetti & Levinthal, 2000; Grégoire et al., 2010; Lovallo et al., 2012; Maitland & Sammartino, 2015a; Pontikes & Rindova, 2020; Porac & Tschang 2013; Rindova & Courtney, 2020; Simon, 1997）。

The quality and quantity of the information available to the decision-maker is also of paramount importance to the construction of effective SWRs (Gavetti & Rivkin, 2007; Maitland & Sammartino, 2015b). At the very early stages of an entrepreneurial endeavor, after a major shock or, more generally, in rapidly changing environments, the information accessible to strategic decision-makers will often be highly incomplete, nondefinitive, and ambiguous (Alvarez & Barney, 2007; Forbes, 2007; Gavetti & Rivkin, 2007; Maitland & Sammartino, 2015b; Miller, 2008; Shepherd, Williams, & Patzelt, 2015). Moreover, the limited information that is available is liable to change constantly, not least because actors’ responses to changes in their environment may lead to it changing still further (Alvarez & Porac, 2020; Berglund, Bousfiha, & Mansoori, 2020; Gavetti, Helfat, & Marengo, 2017; Packard & Clark, 2020; Patvardhan & Ramachandran, 2020). 决策者可获取的信息的质量和数量对于有效战略工作回顾（SWRs）的构建也至关重要（Gavetti & Rivkin, 2007; Maitland & Sammartino, 2015b）。在创业活动的早期阶段、重大冲击之后，或者更普遍地说，在快速变化的环境中，战略决策者可获取的信息往往高度不完整、非确定性且模糊（Alvarez & Barney, 2007; Forbes, 2007; Gavetti & Rivkin, 2007; Maitland & Sammartino, 2015b; Miller, 2008; Shepherd, Williams, & Patzelt, 2015）。此外，现有的有限信息容易不断变化，这在很大程度上是因为行动者对环境变化的反应可能会进一步导致信息发生变化（Alvarez & Porac, 2020; Berglund, Bousfiha, & Mansoori, 2020; Gavetti, Helfat, & Marengo, 2017; Packard & Clark, 2020; Patvardhan & Ramachandran, 2020）。

Thus far, however, the literature has had little to say about issues concerning the updating of SWRs, in particular the prescriptive question of how decisionmakers should update their SWRs over the course of the implementation of strategic decisions. On the one hand, contributors taking a forward-looking, predictive approach to strategic decision-making have focused primarily on the initial construction of SWRs and their impact on subsequent behavior (Gavetti & Levinthal, 2000; Gavetti & Rivkin, 2007; Lovallo et al., 2012; Maitland & Sammartino, 2015a, 2015b). On the other, contributors emphasizing the adaptive nature of strategic decision-making have downplayed the role of cognitive representations by adopting a nonpredictive, “backward-looking” approach (Gavetti et al., 2012: 26; Patvardhan & Ramachandran, 2020), perhaps because adaptations are seen as “comparatively ad hoc, serendipitous, and/or messy” rather than the product of careful thought (Klingebiel & De Meyer, 2013: 133; see also Cyert & March, 1963/ 1992; Langley et al., 1995). 然而，到目前为止，文献中对战略工作回顾（SWRs）更新相关问题的论述较少，特别是决策者在战略决策实施过程中应如何更新其战略工作回顾的规范性问题。一方面，采用前瞻性、预测性战略决策方法的学者主要关注战略工作回顾的初始构建及其对后续行为的影响（Gavetti & Levinthal, 2000；Gavetti & Rivkin, 2007；Lovallo et al., 2012；Maitland & Sammartino, 2015a, 2015b）。另一方面，强调战略决策适应性本质的学者则通过采用非预测性的“回顾性”方法，淡化了认知表征的作用（Gavetti et al., 2012: 26；Patvardhan & Ramachandran, 2020），这或许是因为适应性被视为“相对临时、偶然和/或混乱的”，而非深思熟虑的产物（Klingebiel & De Meyer, 2013: 133；另见Cyert & March, 1963/1992；Langley et al., 1995）。

There is a significant gap in the literature, therefore, especially in situations of Knightian uncertainty, where the ability to anticipate the future and construct adequate SWRs at the outset is highly limited, and decision-makers often have difficulty adapting their cognitive representations effectively over time (Tripsas & Gavetti, 200o). This issue is a pressing one even when decision-makers create, at least in part, the future through their own actions (Alvarez & Barney, 2007; Gavetti & Porac, 2018) by adopting a “shaping-oriented forward-looking” approach (Patvardhan & Ramachandran, 2020: 671). Our aim in what follows is to take some steps toward filling this gap, by recognizing the merits of engaging in forward-looking behavior while adopting the view that strategic decision-making under Knightian uncertainty is nevertheless a dynamic, adaptive process in which decision-makers are called on to systematically update their cognitive representations as additional information emerges. 在文献中存在一个显著的空白，因此，特别是在奈特不确定性（Knightian uncertainty）的情况下，人们在一开始就能够预测未来并构建适当的战略工作回顾（SWRs）的能力极为有限，决策者往往难以随着时间的推移有效地调整其认知表征（Tripsas & Gavetti, 2000）。即使决策者至少部分通过自身行动创造未来（Alvarez & Barney, 2007；Gavetti & Porac, 2018），采用“塑造导向的前瞻性”方法（Patvardhan & Ramachandran, 2020: 671），这一问题仍然十分紧迫。我们接下来的目标是通过认识到进行前瞻性行为的价值，同时认为在奈特不确定性下的战略决策仍然是一个动态的、适应性的过程——在这个过程中，随着新信息的出现，决策者需要系统地更新其认知表征——来朝着填补这一空白迈出一些步伐。

THEORIZING THE SCENARIO SPACE

场景空间的理论化

In addressing how decision-makers should update their SWRs we will focus specifically on the task of updating the SWR analogs of small world states and the state space, which, to avoid confusion, we henceforth refer to as scenarios and the scenario space, respectively. We do so because, as we will show, the way in which decision-makers update the scenario space is pivotal to the process of updating SWRs. By scenarios we mean relatively coarse and partial descriptions of future states of affairs that may be imagined by a decision-maker in the course of constructing and updating an SWR. Scenarios are therefore hypotheses about what may occur in the future, that specify how currently unknown and contingent parts of reality might unfold (Jungerman & Thuring, 1988; Porter, 1985; Schoemaker, 1993). The scenario space at a given point in time is then the set of scenarios the decision-maker currently regards as possible in relation to a particular decision. 在探讨决策者应如何更新其情景权重比（SWRs）时，我们将特别关注更新小世界状态及其状态空间的情景权重比模拟的任务。为避免混淆，我们此后分别将其称为“情景”和“情景空间”。我们这样做是因为，正如我们将展示的，决策者更新情景空间的方式对于更新情景权重比的过程至关重要。我们所说的“情景”是指决策者在构建和更新情景权重比时可能设想的未来事态的相对粗略和部分的描述。因此，情景是关于未来可能发生的事情的假设，它具体说明当前未知和偶然的现实部分可能如何发展（Jungerman & Thuring, 1988; Porter, 1985; Schoemaker, 1993）。在特定时间点的情景空间，就是决策者当前针对某一特定决策可能认为可行的所有情景的集合。

There are three general points to highlight about scenarios and the scenario space. The first is that scenarios have to be constructed and are thus the product of managerial imagination and other cognitive attributes. We will assume that this is done on the basis of projections from some set of imagined influences, an influence being any category of event or condition that could contribute causally to the realization of a state of affairs. For example, when deciding whether to introduce a new product in a particular market, relevant influences might include future events such as competitor moves and regulatory changes, and background conditions such as prevailing technologies and existing market demographics. The second point, one we have already emphasized, is that scenario spaces are almost always incomplete (Gilboa, Minardi, & Samuelson, 2017; Shackle, 1979), typically running to only a relatively small number of scenarios (Maitland & Sammartino, 2015a; see also Kahneman & Lovallo, 1993; Lovallo & Kahneman, 2003). 关于情景及情景空间，有三个要点需要强调。首先，情景必须被构建，因此是管理想象力和其他认知属性的产物。我们假设这是基于对某些想象中的影响因素的预测来完成的，影响因素是任何可能对事态实现产生因果贡献的事件或条件类别。例如，在决定是否在特定市场推出新产品时，相关影响因素可能包括未来事件（如竞争对手的行动和监管变化）以及背景条件（如现有技术和市场人口结构）。第二个要点，也是我们已经强调过的，是情景空间几乎总是不完整的（Gilboa, Minardi, & Samuelson, 2017；Shackle, 1979），通常只包含相对较少的情景（Maitland & Sammartino, 2015a；另见Kahneman & Lovallo, 1993；Lovallo & Kahneman, 2003）。

The third point is that while an incomplete scenario space rules out the use of classical probabilities, since these require a complete list of mutually exclusive eventualities (Bradley & Drechsler, 2014; Feduzi & Runde, 2014; Ghirardato, 2001), it does not preclude scenarios being the subject of qualitative judgments of probability. That is, strategic decision-makers may yet be able to arrive at judgments to the effect that a scenario is possible (has a nonzero probability) or is impossible (has a zero probability), or to make judgments of the form “Scenari $A$ is highly probable,” “Scenario $A$ is more probable than Scenario $B$ ,” or “Scenario $A$ is as probable as Scenario $B ^ { \prime \prime }$ on the basis of the available evidence (for a theory of qualitative probability on these lines, see Keynes, 1921/1973). We will proceed on the basis that strategic decision-makers are able to make judgments of this kind when assessing their scenarios, and, to avoid the risk of these being confused with probability judgments as standardly interpreted, we henceforth refer to them as judgments of credibility. 第三点是，尽管不完整的情景空间排除了经典概率的使用（因为经典概率需要完整的互斥事件列表（Bradley & Drechsler, 2014; Feduzi & Runde, 2014; Ghirardato, 2001），但这并不排除情景成为概率定性判断的对象。也就是说，战略性决策者仍然可能能够得出这样的判断：某个情景是可能的（具有非零概率）或不可能的（具有零概率），或者做出“情景A很可能发生”、“情景A比情景B更可能发生”或“情景A与情景B可能性相同”（基于现有证据）的判断（关于此类定性概率理论，参见Keynes, 1921/1973）。我们将基于以下假设继续：战略性决策者在评估其情景时能够做出这类判断，并且为避免这些判断被误认为是标准解释下的概率判断，我们此后将其称为可信度判断。

We are now in a position to theorize how the scenario space may be constructed and updated over time. To facilitate the analysis, we draw on some foundational concepts introduced by Faulkner, Feduzi, and Runde (2017). Faulkner et al. (2017: 1282) defined a known as “any feature of the world that an individual has knowledge of,” and an unknown or gap in knowledge as “any feature of the world that an individual lacks knowledge of.” A feature of the world is any fact “about past, present and future reality” (Faulkner et al., 2017: 1281), such as the current state of the shale gas market in the United 我们现在能够从理论上探讨情景空间如何随时间构建和更新。为便于分析，我们借鉴了福克纳（Faulkner）、费杜齐（Feduzi）和朗德（Runde）（2017）提出的一些基础概念。福克纳等人（2017：1282）将“个体已知的世界特征”定义为“个体已知的任何世界特征”，而“知识空白”则定义为“个体缺乏知识的任何世界特征”。世界特征是关于“过去、现在和未来现实”的任何事实（福克纳等人，2017：1281），例如美国页岩气市场的当前状态。

States, Euro CO emission standards applying to diesel vehicles in present-day London, or the closing level of the FTSE on July 1, 2025. They further distinguish between a known unknown—a gap in knowledge that an individual knows about and is aware of at the relevant time”—and an unknown unknown—“a gap in knowledge that an individual is not aware of at that time, either because they do not know about that gap in knowledge or because, despite knowing of it, they are unaware of it” (Faulkner et al., 2017: 1283). 国家、当今伦敦适用的柴油车欧盟二氧化碳排放标准，或2025年7月1日富时100指数的收盘水平。他们进一步区分了“已知的未知”——即个人在相关时间已知并意识到的知识缺口——和“未知的未知”——即个人在当时并未意识到的知识缺口，原因可能是他们不知道存在这个缺口，或者尽管知道这个缺口的存在，但却未意识到它（Faulkner等人，2017：1283）。

Building on these concepts, Faulkner et al. (2017: 1283) introduced the notion of hypothetical values associated with a gap in knowledge, where a hypothetical value is “any value—outcome, state of affairs, result, quantity and so on—that could conceivably be thought to be a candidate for the actual or true value of the unknown under consideration.” Thus, if the unknown concerns the Euro CO emission standards applying to diesel vehicles in London, for example, the associated hypothetical values would likely include numerous different quantities such as $1 . 0 0 \mathrm { g } /$ km, $2 . 7 2 \ : \mathrm { g / k m }$ , and so on. Finally, the set of hypothetical values associated with an unknown is “the entire collection of values that could conceivably be regarded as the true value of that unknown by any person within the group or community concerned” (Faulkner et al., 2017: 1283). This set therefore represents the universe of conceivable hypothetical values at a given point in time, comprising all values that could potentially be imagined by a decision-maker from that community, given the prevailing background conditions (state of knowledge, cognitive abilities, etc.). 基于这些概念，福克纳等人（2017：1283）引入了与知识缺口相关的假设值概念，其中假设值是“任何可能被认为是所考虑的未知事物的实际或真实值的候选值——结果、事态、结果、数量等等”。因此，例如，如果未知问题涉及适用于伦敦柴油车辆的欧盟二氧化碳排放标准，相关的假设值可能包括许多不同的数量，如1.00克/公里、2.72克/公里等等。最后，与未知事物相关的假设值集合是“在相关群体或社区中的任何人都可能认为是该未知事物真实值的所有值的集合”（福克纳等人，2017：1283）。因此，这个集合代表了在特定时间点上可想象的假设值的范围，包括在当时的背景条件（知识状态、认知能力等）下，该社区的决策者可能想象出的所有值。

To relate these concepts to scenarios and scenario spaces, consider a decision-maker facing a strategic decision problem. The future course of events as they pertain to that decision is a known unknown to the decision-maker, with hypothetical values associated with this unknown consisting of alternate futures that might be envisioned. The crucial link to the SWR framework is that these hypothetical values are potential scenarios—that is, imagined future states of affairs that might be included in the decision-maker’s scenario space. The set of hypothetical values with respect to the unknown future is the set of all conceivable scenarios, comprising each and every scenario that any person in the broader community, put in the position of decision-maker, might conceivably imagine at the present time. We will call this the set of imaginable scenarios. 为了将这些概念与情景及情景空间联系起来，我们可以考虑一位面临战略决策问题的决策者。与该决策相关的未来事件走向对决策者而言是“已知的未知”，而与这一未知相关的假设性取值包括可能被设想的替代未来。与情景加权回归（SWR）框架的关键联系在于，这些假设性取值是潜在的情景——即可能被纳入决策者情景空间的想象中的未来事态。关于未知未来的假设性取值集合，是所有可想象情景的集合，包括当前任何处于决策位置的更广泛社区中的人可能设想的每一个情景。我们将这一集合称为“可想象情景集”。

To further characterize the scenario space, we make three assumptions about the scenarios a specific decision-maker facing a strategic decision problem comes up with: 为了进一步描述场景空间，我们对特定决策者在面临战略决策问题时提出的场景做出三个假设：

1. The scenarios consciously imagined by the decision-maker are typically fewer in number than conceivably could have been imagined. In particular, the number of items in the set of imaginable scenarios will generally far exceed the number of items in the decision-maker’s scenario space.
   决策者有意识地设想的场景，其数量通常比可能想象到的要少。特别是，可想象场景集合中的项目数量通常会远远超过决策者场景空间中的项目数量。

2. The scenarios consciously imagined by the decision-maker could include scenarios they regard as impossible.

3. 决策者有意识地设想的情景可能包括他们认为不可能的情景。

4. Each of the scenarios consciously imagined by the decision-maker is either genuinely possible or impossible at any point in time, and where this possibility or impossibility is something that depends on the way the world is rather than on what the decision-maker believes about the world.3

5. 决策者有意识地设想的每一种情景在任何时候要么是真正可能的，要么是不可能的，而这种可能性或不可能性取决于世界的实际情况，而不是决策者对世界的看法。3

Given these assumptions, at any given point in time during a decision problem each element of the set of imaginable scenarios can be classified according to three attributes: (a) whether the decisionmaker has consciously considered that scenario; (b) if they have, whether they regard it as possible; (c) and whether that scenario is genuinely possible. The resulting classification of scenarios is depicted in Table 1, which assigns every conceivable scenario to one of six cells. 基于这些假设，在决策问题的任何特定时间点，可想象场景集中的每个元素都可以根据三个属性进行分类：(a) 决策者是否有意识地考虑过该场景；(b) 如果他们考虑过，是否认为该场景是可能的；(c) 该场景是否真正可能。场景的最终分类如图表1所示，该图表将每个可想象的场景分配到六个单元格中的一个。

Cells 1 and 2 of Table 1 contain scenarios imagined and regarded as possible by the decision-maker, split between those that are correctly (cell 1), and those that are mistakenly (cell 2), regarded as such. Taken together, the scenarios in these two cells therefore constitute the decision-maker’s scenario space, indicated by the shaded area. The scenarios in cells 3 and 4 are those consciously imagined by the decisionmaker but excluded from the scenario space because they are regarded as impossible, correctly so in cell 4 but mistakenly so in cell 3. The scenarios in cells 5 and 6 also lie outside the scenario space, having not even entered the imagination of the decision-maker. 表1中的单元格1和2包含决策者设想并认为可能的场景，分为被正确（单元格1）和错误（单元格2）视为可能的场景。综合来看，这两个单元格中的场景构成了决策者的场景空间，以阴影区域表示。单元格3和4中的场景是决策者有意识地设想的，但因被认为不可能而被排除在场景空间之外，其中单元格4是正确的，而单元格3是错误的。单元格5和6中的场景也不在场景空间内，甚至未进入决策者的想象。

As a benchmark against which to assess the accuracy of a decision-maker’s scenario space, we define a scenario space as correct when it contains all, and only, those conceivable scenarios that, at the present time, are genuinely possible future states of affairs. In Table 1, a correct scenario space therefore requires that cells 3 and 5 in the left-hand column, and cell 2 in the right-hand column, are empty. This standard is unlikely ever to be met in practice, the scenario space falling short of being correct whenever a decision-maker commits an “error of omission” in failing to include a genuinely possible scenario, or an “error of commission” by including a genuinely impossible scenario. 作为评估决策者情景空间准确性的基准，我们将情景空间定义为正确的，当且仅当它包含且仅包含当前真正可能的未来事态的所有可设想情景。因此，在表1中，正确的情景空间要求左侧列中的单元格3和5，以及右侧列中的单元格2为空。在实践中，这一标准几乎不可能被满足，每当决策者因未能纳入真正可能的情景而犯下“遗漏错误”，或因纳入真正不可能的情景而犯下“包含错误”时，情景空间就无法达到正确的标准。

TABLE 1 The Set of Imaginable Scenarios and the Scenario Space
表1 可想象场景集合与场景空间

Source: Adapted from Faulkner et al., 2017. 改编自福克纳等人，2017年。

In light of the earlier discussion of the factors influencing decision-makers’ ability to construct and update SWRs, it is the awareness of relevant influences that is likely to be crucial to their capacity to populate the scenario space with genuinely possible scenarios. Recall that a scenario is a projected future state of affairs that flows from a set of imagined influences. It then follows that a decision-maker’s gaps in knowledge about particular influences implies corresponding shortcomings in the scenario space, most notably the omission, either through being unimagined or regarded as impossible, of scenarios that might genuinely arise from these unrecognized influences. The scale of the impact on the scenario space depends in part on whether such influences are known unknowns, or unknown unknowns, for the decision-maker. In the former case, the decisionmaker will at least be in a position to contemplate hypothetical values associated with the relevant influence, even if they are not able to identify all, or even a representative spread, of them and construct the associated scenarios. The latter case is likely to be more damaging, since a decision-maker who is unaware of having a gap in knowledge about an influence will not be in a position even to begin contemplating its range of associated hypothetical values and the scenarios these may lead to. 考虑到之前关于影响决策者构建和更新情景加权响应（SWRs）能力的因素的讨论，关键在于决策者对相关影响因素的认知，这可能对其在情景空间中填充真正可能出现的情景的能力至关重要。需要回顾的是，情景是基于一组想象中的影响因素而推导出来的未来事态发展。由此可推断，决策者对特定影响因素的知识缺口意味着情景空间中存在相应的不足，最显著的是，由于这些未被识别的影响因素可能产生的情景要么因未被想象到，要么被视为不可能而被遗漏。

情景空间受到的影响程度部分取决于决策者是否将这些影响因素视为“已知的未知因素”或“未知的未知因素”。在前一种情况下，决策者至少能够考虑与相关影响因素相关的假设性取值，即使他们无法识别所有这些取值，甚至无法识别其代表性的分布范围，也能构建相关情景。而后一种情况可能更具破坏性，因为如果决策者没有意识到自己对某一影响因素存在知识缺口，他们甚至无法开始考虑该影响因素相关的假设性取值范围，以及这些取值可能导致的情景。

While Table 1 provides a useful way of representing the scenario space, and of evaluating the scope and accuracy of its contents at a given point in time, the static nature of the analysis limits what can be said about the updating of scenario spaces during the implementation of a strategic decision and how the scenario space might change as a result. By updating the scenario space, we mean the process by which the decision-maker revises their beliefs about the future in response to new or newly considered information and thereby may be led to include new or exclude existing scenarios in or from the scenario space.4 From the perspective of Table 1, a decision-maker updating their beliefs implies the vertical movement of scenarios between cells; that is, from a cell in one row of the table to a cell in another row within the same column. The possibilities are as follows: 虽然表1提供了一种表示情景空间、并在特定时间点评估其内容范围和准确性的有用方式，但分析的静态性质限制了我们对战略决策实施过程中情景空间的更新以及情景空间可能发生的变化所能进行的描述。我们所说的更新情景空间，是指决策者根据新的或新考虑的信息修订其对未来的信念，从而可能被引导将新情景纳入情景空间或从情景空间中排除现有情景。4 从表1的角度来看，更新信念的决策者意味着情景在单元格之间的垂直移动；也就是说，从表中某一行的单元格移动到同一列中另一行的单元格。可能性如下：

1. Scenarios previously unimagined are now considered and regarded as possible (scenarios in cells 5 and 6 move to cells 1 and 2);
2. 以前难以想象的场景现在被认为是可能的（单元格5和6中的场景移至单元格1和2）；

TABLE 2 Updating Beliefs: Movements Within the Table
表2 更新信念：表内变动

2. Scenarios previously considered possible are now forgotten, rendering them unimagined (scenarios in cells 1 and 2 move to cells 5 and 6);

3. 之前被认为可能的场景现在被遗忘，使得它们变得难以想象（单元格1和2中的场景移至单元格5和6）；

4. Scenarios previously considered impossible are now considered possible (scenarios in cells 3 and 4 move to cells 1 and 2);

5. 之前被认为不可能的场景现在被认为是可能的（单元格3和4中的场景移至单元格1和2）；

6. Scenarios previously considered possible are now considered impossible (scenarios in cells 1 and 2 move to cells 3 and 4);

7. 此前被认为可能的情景现在被认为是不可能的（单元格1和2中的情景移至单元格3和4）；

8. Scenarios previously unimagined are now considered and regarded as impossible (scenarios in cells 5 and 6 move to cells 3 and 4);

9. 以前无法想象的场景现在被认为是不可能的（单元格5和6中的场景移至单元格3和4）；

10. Scenarios previously considered impossible are now forgotten, rendering them unimagined (scenarios in cells 3 and 4 move to cells 5 and 6).

11. 之前被认为不可能的场景如今已被遗忘，变得难以想象（单元格3和4中的场景移至单元格5和6）。

Of these six possibilities, only the first four are associated with changes to the decision-maker’s scenario space. The first and third involve additions, augmenting the scenario space with scenarios now considered possible that had previously not been imagined by the decision-maker, or that had been imagined but disregarded as impossible. The second and fourth involve eliminations from the scenario space, removing scenarios previously considered possible but now either regarded as impossible or disregarded entirely. 在这六种可能性中，只有前四种与决策者的情景空间变化相关。第一种和第三种涉及新增内容，通过纳入决策者此前未设想过或虽有设想但被视为不可能的情景，来扩充情景空间。第二种和第四种涉及情景空间的剔除，移除了此前被认为可能但现在要么被视为不可能，要么被完全忽略的情景。

These findings lead to Table 2, a dynamic version of Table 1 that depicts ways the scenario space may change during the updating process. The doubleheaded arrows represent the different movements of scenarios that may occur as a result of the decisionmaker revising their beliefs during the process of updating, with the double-headed nature of each arrow reflecting that, for each, movement may be in either direction. The shaded arrows denote movements implying changes to the membership of the scenario space, while the unshaded arrows indicate movements that leave the contents of the scenario space unaltered. 这些发现引出了表2，它是表1的动态版本，描述了场景空间在更新过程中可能发生变化的方式。双头箭头代表决策者在更新过程中修正其信念时可能发生的不同场景移动，每个箭头的双头特性反映了对于每个移动，可能是双向的。阴影箭头表示意味着场景空间成员发生变化的移动，而未阴影的箭头则表示使场景空间内容保持不变的移动。

It is worth highlighting two further points concerning changes to the scenario space and the updating process, both arising from the implementation of strategic decisions taking place over (possibly long periods of) time. The first is that developments in the world may result in conceivable scenarios that were formerly genuinely impossible becoming possible, or vice versa.5 In Tables 1 and 2 this would imply horizontal movement of scenarios between cells; that is, from a cell in one column to a cell in the other column within the same row. Crucially, however, such developments imply no change to the membership of the decision-maker’s scenario space. At most, and perhaps unbeknown to the decisionmaker, the effect on the scenario space is to shift scenarios between cell 1 and cell 2, either because a scenario the decision-maker (until now wrongly) regards as possible is now rendered genuinely possible (moving from cell 2 to cell 1), or else a scenario the decision-maker (until now rightly) regards as possible is now rendered genuinely impossible (moving from cell 1 to cell 2). Other cases, where scenarios move either between cells 3 and 4, or between cells 5 and 6, have no impact on the scenario space. 值得强调的是，关于情景空间的变化和更新过程，还有两点需要说明，这两点都源于战略决策的长期（可能是长期）实施。首先，世界的发展可能会使以前真正不可能的可想象情景变为可能，反之亦然。5 在表1和表2中，这意味着情景在单元格之间的横向移动；也就是说，从同一行中某一列的单元格移动到另一列的单元格。然而，至关重要的是，此类发展并不意味着决策者的情景空间成员发生变化。最多，也许决策者并不知情，情景空间的变化仅在于情景在单元格1和单元格2之间移动，要么是因为决策者（到目前为止错误地）认为可能的情景现在真正变得可能（从单元格2移动到单元格1），要么是因为决策者（到目前为止正确地）认为可能的情景现在真正变得不可能（从单元格1移动到单元格2）。在其他情况下，情景在单元格3和4之间或单元格5和6之间移动，对情景空间没有影响。

The second point is that the world may develop in ways leading either to the emergence of novel scenarios, meaning possible future states of affairs that were previously unimaginable given the prevailing background conditions and so previously outside the set of imaginable scenarios; or to the outright elimination of scenarios, meaning possible future states of affairs that were previously imaginable that then become entirely unimaginable given the prevailing background conditions.6 In effect, our earlier analysis assumes that the set of imaginable scenarios is fixed and unchanging over time, with the members of this set distributed between the six cells in Tables 1 and 2. Once it is acknowledged that this set may change over time, however, entirely new scenarios may enter, or existing scenarios exit, the table. 第二点是，世界的发展可能有两种走向：一种是出现全新的情景，即考虑到当前背景条件，这些可能的未来事态之前是难以想象的，因此不在可想象情景的范围内；另一种是现有情景的彻底消失，即原本可想象的可能未来事态，在当前背景条件下变得完全不可想象。6 实际上，我们之前的分析假设可想象情景的集合是固定且随时间不变的，该集合中的成员分布在表1和表2的六个单元格中。然而，一旦承认这个集合可能随时间变化，全新的情景就可能出现，或者现有情景会从表格中消失。

TABLE 3 Updating the Scenario Space
表3 更新场景空间

The implication for the scenario space during updating is that scenarios that were previously wholly unimaginable may be added to the scenario space, or scenarios that were formerly in the scenario space may become unimaginable. To allow for these possibilities we now amend Table 2, adding doubleheaded, shaded arrows from cells 1 and 2 that comprise the scenario space, to an area outside the table denoting previously unimaginable scenarios. We have also removed the two unshaded arrows that denoted changes that would leave the scenario space unaltered. The resulting Table 3 therefore fully characterizes the ways the scenario space might change during updating. 更新过程中场景空间的含义是，之前完全无法想象的场景可能会被添加到场景空间中，或者以前在场景空间中的场景可能会变得无法想象。为了考虑这些可能性，我们现在修改表2，添加从构成场景空间的单元格1和2出发的双头、阴影箭头，指向表示之前无法想象的场景的表格外部区域。我们还移除了两个未阴影的箭头，这些箭头表示会使场景空间保持不变的变化。因此，最终的表3充分描述了更新过程中场景空间可能发生变化的方式。

We conclude this section with some brief comments on the decision-maker’s objective in updating the scenario space. In simple terms, updating aims to improve the accuracy of the scenario space, drawing on new and existing information to (a) reevaluate earlier assessments that are likely to have been generated under significant information, time, and other constraints; and (b) take account of developments in the world that have occurred since any previous updating. The key issue here, however, is what improving the accuracy of the scenario space entails. 我们以对决策者更新情景空间目标的简要评论结束本节。简单来说，更新的目的是提高情景空间的准确性，利用新的和现有信息来（a）重新评估那些可能在大量信息、时间和其他限制下生成的早期评估；以及（b）考虑自上次更新以来世界上发生的发展。然而，这里的关键问题是，提高情景空间的准确性意味着什么。

Since a perfectly accurate scenario space is one that is correct in the sense defined earlier, improving the accuracy of the scenario space, in principle at least, involves moving closer toward the correct scenario space. This means seeking to include a greater number of genuinely possible scenarios in the scenario space (i.e., reducing errors of omission) by moving scenarios in the left-hand column of Table 3 upward via the arrows into cell 1, while also trying to reduce the number of genuinely impossible scenarios in the scenario space (i.e., reducing errors of commission) by moving them downward in the right-hand column of the table from cell 2 to cell 4 (in the first instance). By definition, this approach to updating leads in the limit to a correct scenario space. 由于完全准确的情景空间是指在前面定义的意义上正确的情景空间，因此提高情景空间的准确性（至少在原则上）涉及向正确的情景空间逐步靠近。这意味着通过将表3左侧列中的情景通过箭头向上移动到单元格1，以纳入更多真正可能的情景（即减少遗漏误差），同时还试图减少情景空间中真正不可能的情景的数量（即减少错误包含误差），首先将表右侧列中的情景从单元格2向下移动到单元格4。根据定义，这种更新方法在极限情况下会导致正确的情景空间。

Of course, given strategic decision-makers’ cognitive make up, the nature of the informational environment in which they are operating, and other factors that may be in play, it is unlikely that updating could achieve anything close to a correct scenario space in practice. However, that is not to say that strategic decision-makers could not be more effective in updating their scenario spaces in many cases, or that the procedures they might use to do so are equally effective. To pursue these matters, we now turn to the psychology literature on human reasoning, in particular to its findings in relation to hypothesis-testing behavior. 当然，考虑到战略决策者的认知构成、他们所处信息环境的性质以及其他可能起作用的因素，在实践中，更新不太可能接近一个正确的情景空间。然而，这并不意味着战略决策者在许多情况下不能更有效地更新他们的情景空间，也不意味着他们可能使用的更新程序同样有效。为了探讨这些问题，我们现在转向关于人类推理的心理学文献，特别是其关于假设检验行为的研究结果。

DISCONFIRMATION AND COUNTERFACTUAL REASONING IN HYPOTHESIS-TESTING

假设检验中的否定与反事实推理

Hypothesis testing—a key function of the mind aimed at establishing whether, based on the available evidence, a hypothesis is supported and should therefore be retained, or is refuted and should therefore be revised or abandoned (Evans, 2007)—is an important area of research for psychologists studying human reasoning (Poletiek, 2001). Since scenarios are themselves hypotheses about alternative futures, updating the scenario space is de facto an exercise in hypothesis testing. It is therefore natural to ask what might be learned from the psychology of reasoning about how decision-makers may revise their scenario spaces during the course of implementing strategic decisions. 假设检验——一种旨在根据现有证据确定假设是否得到支持（因此应保留）或被反驳（因此应修订或放弃）的关键心理功能（Evans, 2007）——是研究人类推理的心理学家的重要研究领域（Poletiek, 2001）。由于情景本身是关于替代未来的假设，更新情景空间实际上就是一种假设检验练习。因此，很自然会问：从推理心理学中可以了解到决策者在实施战略决策过程中可能如何修订其情景空间。

In what follows, we focus on two simple hypothesis-testing heuristics—disconfirmation and counterfactual reasoning—that people sometimes actually employ when solving decision tasks and that have been extensively studied in the psychology of reasoning (Cherubini, Castelvecchio, & Cherubini, 2005; Farris & Revlin, 1989a, 1989b; Gale & Ball, 2009; Gorman & Gorman, 1984; Gorman, Stafford, & Gorman, 1987; Klayman & Ha, 1987; Oaksford & Chater, 1994; Tukey, 1986). In the present section we characterize the two heuristics, using Wason’s (1960) famous 2-4-6 experiment (see also Evans, 2016) to illustrate the steps involved in their implementation.’ In the following section we then transpose the two heuristics to the strategic decision-making context, developing procedures for updating the scenario space based on each heuristic. 在下文，我们将聚焦于两种简单的假设检验启发式方法——证伪和反事实推理——这两种方法有时人们在解决决策任务时确实会使用，并且在推理心理学领域已得到广泛研究（Cherubini, Castelvecchio, & Cherubini, 2005；Farris & Revlin, 1989a, 1989b；Gale & Ball, 2009；Gorman & Gorman, 1984；Gorman, Stafford, & Gorman, 1987；Klayman & Ha, 1987；Oaksford & Chater, 1994；Tukey, 1986）。在本节中，我们将描述这两种启发式方法，并以沃森（1960）著名的2-4-6实验（另见Evans, 2016）为例，说明它们实施过程中的步骤。在下一节中，我们将把这两种启发式方法应用到战略决策的背景中，开发基于每种启发式方法更新情景空间的程序。

Disconfirmation and Counterfactual Reasoning

否定与反事实推理

Wason’s (1960) 2-4-6 task is a rule-discovery problem that has been widely used in experimental studies of hypothesis-testing behavior. In this task the experimenter privately chooses a rule—“three numbers in increasing order of magnitude” in Wason’s (1960) original study—that generates particular triples of numbers. Subjects are then presented with a triple consistent with that rule—2-4-6 in Wason’s (1960) study—and invited to uncover the rule. To do so, they are required to generate hypotheses about what the rule might be and to test those hypotheses by providing triples of their own to the experimenter, who then tells them whether the triples conform to the chosen rule. When subjects believe they have identified the rule, they are instructed to declare it to the experimenter, who tells them whether they are correct. If the rule the subject declares is incorrect, the task continues. The task ends when a subject correctly identifies the rule, the allotted time runs out, or the subject gives up. Wason（1960）的2-4-6任务是一个规则发现问题，在假设检验行为的实验研究中被广泛应用。在这个任务中，实验者私下选择一个规则——在Wason（1960）的原始研究中是“三个按大小递增的数字”——该规则生成特定的数字三元组。然后，受试者会被呈现一个与该规则一致的三元组（在Wason（1960）的研究中是2-4-6），并被邀请去发现规则。为此，他们需要对规则可能是什么提出假设，并通过向实验者提供自己生成的三元组来检验这些假设，实验者随后会告知这些三元组是否符合所选规则。当受试者认为自己已经识别出规则时，他们会被要求向实验者声明该规则，实验者会告知他们是否正确。如果受试者声明的规则不正确，任务将继续。当受试者正确识别出规则、规定时间结束或受试者放弃时，任务结束。

Subjects employ a variety of hypothesis-testing strategies in attempting to solve Wason’s (1960) task and its variants (Tukey, 1986), including disconfirmation and counterfactual reasoning (Farris & Revlin, 1989a, 1989b; Gorman & Gorman, 1984). These last two strategies represent an eliminative approach in being aimed at rejecting, rather than confirming, a hypothesis under consideration. 被试在尝试解决沃森（1960）任务及其变体（图基，1986）时采用了多种假设检验策略，包括证伪和反事实推理（法里斯 & 雷夫林，1989a，1989b；戈尔曼 & 戈尔曼，1984）。后两种策略代表了一种排除法，其目的是拒绝而非证实正在考虑的假设。

Disconfirmation is a basic form of Popperian-style falsification (Popper, 1959) that recommends that the subject start with a working hypothesis explaining some observed phenomenon, assume this hypothesis is correct, and then perform a series of tests that involve deliberately looking for new evidence that is inconsistent with the working hypothesis. If such evidence is found, the working hypothesis is eliminated, a new hypothesis is generated, and the process begins again. The higher the number of tests passed by any working hypothesis, in the sense of not being eliminated on the basis of new evidence, the greater the subject’s confidence in it. 证伪是波普尔式证伪（波普尔，1959）的基本形式，它建议主体从一个解释某种观察到的现象的工作假设开始，假定该假设正确，然后进行一系列测试，这些测试包括刻意寻找与工作假设不一致的新证据。如果发现了这样的证据，工作假设就会被排除，新的假设会被生成，然后这个过程重新开始。在不被新证据排除的意义上，任何工作假设通过的测试数量越多，主体对它的信心就越大。

Table 4 specifies the steps involved in disconfirmation, and classifies each in terms of the three broad activities involved in hypothesis testing—namely hypothesis formation, evidence collection, and feedback evaluation. The final column illustrates the use of the heuristic in Wason’s (1960) 2-4-6 task, enumerating the implementation of each step from the beginning of the experiment. 表4详细说明了证伪过程中的步骤，并根据假设检验涉及的三大类活动（即假设形成、证据收集和反馈评估）对每个步骤进行了分类。最后一列展示了该启发式方法在沃森（1960）的2-4-6任务中的应用，列举了实验开始时每个步骤的具体实施方式。

Counterfactual reasoning also directs the subject to attempt to eliminate the current working hypothesis. However, unlike disconfirmation it does so by looking for evidence in favor of a specific alternative hypothesis (Farris & Revlin, 1989a, 1989b; Gorman et al., 1987; Oaksford & Chater, 1994; Tukey, 1986). The subject again starts with a working hypothesis explaining some observed phenomenon and performs a series of tests. However, in this case the subject assumes that the working hypothesis is false. In particular, they generate an alternative to the working hypothesis that is consistent with the existing evidence, and then deliberately look for new evidence that would be consistent with the alternative hypothesis and inconsistent with the original working hypothesis. If such evidence is found, the original working hypothesis is eliminated and the alternative becomes the new working hypothesis, and the process begins again. The higher the number of tests passed by any working hypothesis, the greater the subject’s confidence in it. The steps and activities involved in counterfactual reasoning are given in Table 5, together with their implementation from the beginning of Wason’s (1960) 2-4-6 task. 反事实推理还引导被试尝试排除当前的工作假设。然而，与证伪不同的是，它通过寻找支持特定替代假设的证据来实现这一点（Farris & Revlin, 1989a, 1989b；Gorman et al., 1987；Oaksford & Chater, 1994；Tukey, 1986）。被试再次从解释某个观察现象的工作假设开始，并进行一系列测试。但在这种情况下，被试假设工作假设是错误的。具体来说，他们生成一个与现有证据一致的工作假设的替代方案，然后刻意寻找与该替代假设一致但与原始工作假设不一致的新证据。如果发现了这样的证据，原始工作假设就会被排除，替代假设成为新的工作假设，然后该过程重新开始。任何工作假设通过的测试数量越多，被试对其的信心就越大。反事实推理涉及的步骤和活动列于表5中，同时也包括从Wason（1960）的2-4-6任务开始的实施过程。

TABLE 4 Disconfirmation
表4 不确认

USING DISCONFIRMATION AND COUNTERFACTUAL REASONING IN UPDATING SCENARIO SPACES

使用不确认和反事实推理更新场景空间

We now transpose disconfirmation and counterfactual reasoning as represented in the psychology literature into a strategic decision-making context, addressing in particular the question of how a decision-maker might update their scenario space over the course of the implementation of a strategic decision. To this end, Table 6 describes procedures that operationalize each heuristic method for use in this context. 我们现在将心理学文献中所描述的否定确认（disconfirmation）和反事实推理（counterfactual reasoning）转化为战略决策情境，特别探讨决策者在战略决策实施过程中如何更新其情景空间。为此，表6描述了在该情境中对每种启发式方法进行操作化的程序。

TABLE 5 Counterfactual Reasoning
表5 反事实推理

TABLE 6 Disconfirmation-Based and Counterfactual Reasoning-Based Procedures for Updating Scenario Spaces
表6 基于否定确认和反事实推理的场景空间更新流程

The disconfirmation-based (henceforth D-based) procedure begins with the decision-maker assuming for the purpose of the exercise that their existing scenario space is correct in the sense of containing all, and only, those scenarios that are genuinely possible future states of affairs. The decision-maker then selects one of these scenarios and looks for new evidence against it, meaning information that, if uncovered, would reduce the credibility associated with that scenario. Once this search is completed the decision-maker reevaluates the credibility of all scenarios in the scenario space in light of the new evidence. When the evidence is so strong that a scenario can no longer be considered possible and must be removed from the scenario space, the decision-maker is obliged to generate an alternative scenario that is not currently in the scenario space but is consistent with the now expanded evidence, by imagining a new influence they had not considered before. The procedure is then repeated as resources allow or until the updated scenario space is deemed adequate. 基于否定的（以下简称D型）程序始于决策者在进行该过程时假设其现有情景空间是正确的，即该情景空间包含且仅包含那些真正可能的未来事态。决策者随后从这些情景中选择一个，并寻找反对该情景的新证据，即如果发现这些信息，会降低该情景可信度的信息。一旦完成搜索，决策者会根据新证据重新评估情景空间中所有情景的可信度。当证据如此确凿，以至于某个情景不再被认为是可能的，必须从情景空间中移除时，决策者必须通过设想一个之前未考虑过的新影响，生成一个当前不在情景空间中但与现已扩展的证据一致的替代情景。然后，该过程在资源允许的情况下重复进行，或直到更新后的情景空间被认为足够为止。

In the case of the counterfactual reasoning—based (henceforth CR-based) procedure, the decisionmaker begins by assuming the existing scenario space is incorrect in the sense of containing one or more scenarios that are in fact impossible, or omitting one or more scenarios that are genuinely possible. The decision-maker must then generate an alternative scenario not currently in the scenario space, consistent with the existing evidence, by imagining a new influence they had not considered before. Once this is done, the decision-maker looks for new evidence in favor of this newly created scenario—that is, information that, if uncovered, would increase the credibility associated with that scenario. Having completed this search, the decision-maker then evaluates the impact of the new evidence on both the new and existing scenarios, retaining the new scenario, and eliminating existing scenarios, as justified by the evidence. The procedure is then repeated as resources allow or until the scenario space is deemed adequate. 在基于反事实推理（以下简称CR）的流程中，决策者首先假设现有场景空间存在错误，即包含一个或多个实际上不可能的场景，或遗漏了一个或多个真正可能的场景。然后，决策者必须通过设想一个之前未考虑过的新影响因素，生成一个当前不在场景空间内但与现有证据一致的替代场景。完成这一步后，决策者会寻找支持这一新创建场景的新证据——即如果被发现，将提高该场景可信度的信息。完成这一搜索后，决策者评估新证据对新场景和现有场景的影响，根据证据保留新场景并排除现有场景。在资源允许的情况下，或直到场景空间被认为足够充分，该流程会重复进行。

EVALUATING THE COMPARATIVE PERFORMANCE OF D- AND CR-BASED PROCE-DURES IN UPDATING SCENARIO SPACES UNDER KNIGHTIAN UNCERTAINTY

在奈特不确定性下评估基于D和CR的程序在更新场景空间中的比较性能

We now examine the relative performance of the D- and CR-based procedures in updating scenario spaces under conditions of Knightian uncertainty. To prepare the ground, we begin with findings from the psychology literature on the impact of ecological considerations on the effectiveness of disconfirmation and counterfactual reasoning in Wason’s (1960) 2-4-6 task and some of its variants. After characterizing the environment in which strategic decisions take place and the ways in which this differs from the experimental settings of Wason-style tasks, we evaluate the two procedures relative to three sets of criteria germane to improving the accuracy of the scenario space and so promoting more effective decision-making in strategic settings. 我们现在考察在奈特不确定性条件下，基于D和CR的程序在更新情景空间时的相对表现。为了铺垫，我们首先从心理学文献中关于生态因素对沃森（1960）2-4-6任务及其部分变体中证伪和反事实推理有效性的影响的研究结果入手。在描述战略决策发生的环境，以及这种环境与沃森式任务实验设置的差异之后，我们根据三组与提高情景空间准确性相关的标准，评估这两种程序，从而促进战略环境中的更有效决策。

On the Performance of Disconfirmation and Counterfactual Reasoning

关于否定确认与反事实推理的表现研究

In his original study, Wason (1960) found that subjects performed poorly in the 2-4-6 task, something he attributed to their tendency to rely on what we refer to as a confirmation heuristic of testing hypothesized rules by generating triples consistent with those rules.8 For example, subjects presented with the triple 2-4-6 often came up with the hypothesis that the experimenter’s rule was “three even numbers,” and proceeded to test this hypothesis by generating further triples consistent with this rule, such as 12-14- 16, 18-20-22, and 24-26-28. Given that such triples are also consistent with the experimenter’s rule of “three numbers in increasing order of magnitude,” they elicited a series of positive responses from the experimenter that led subjects to the erroneous conclusion that they had correctly identified the experimenter’s rule. 在他最初的研究中，瓦森（Wason，1960）发现被试在2-4-6任务中表现不佳，他将此归因于被试倾向于依赖所谓的证实性启发式，即通过生成与假设规则一致的三元组来检验假设的规则8。例如，面对三元组2-4-6的被试常常会假设实验者的规则是“三个偶数”，并通过生成更多符合该规则的三元组（如12-14-16、18-20-22和24-26-28）来检验这一假设。由于这些三元组也符合实验者“三个数字按大小递增”的规则，它们从实验者那里得到了一系列肯定的反馈，导致被试错误地认为自己正确识别了实验者的规则。

Wason (1960: 139) proposed that the key to success in the 2-4-6 task is to attempt to eliminate, rather than confirm, the hypothesis under consideration—a suggestion that generated a stream of studies exploring whether use of some form of disconfirmation or counterfactual reasoning might improve subjects’ hypothesis-testing performance (Farris & Revlin, 1989a, 1989b; Gorman et al., 1987; Oaksford & Chater, 1994; Tukey, 1986). The main finding of this work concerns the critical importance of ecological considerations: that rather than one heuristic dominating in all circumstances, the performance of different heuristics in Wason-style tasks varies with the structure of the task environment (McKenzie, 2004). 沃森（1960：139）提出，在2-4-6任务中取得成功的关键在于尝试证伪而非证实所考虑的假设——这一观点引发了大量研究，探究某种形式的证伪或反事实推理是否能提高被试的假设检验表现（Farris & Revlin, 1989a, 1989b；Gorman et al., 1987；Oaksford & Chater, 1994；Tukey, 1986）。这项研究的主要发现涉及生态考量的关键重要性：在所有情境中并非单一启发式占主导地位，而是不同启发式在沃森式任务中的表现会随任务环境的结构而变化（McKenzie, 2004）。

This finding emerges from the results of two quite different types of study. In the first, particular attributes of the 2-4-6 task, most notably the nature of the relationship between the experimenter’s true rule and the subject’s working hypothesis, are shown to impact the efficacy of different heuristics (Klayman & Ha, 1987; Oaksford & Chater, 1994). Thus, in the case described above in which the subject’s hypothesized rule is a restricted version of the experimenter’s true rule, it follows as a matter of logic that both disconfirmation and counterfactual reasoning can be effective in facilitating the discovery of the experimenter’s true rule by promoting elimination of an incorrect working hypothesis. In contrast, in the opposite case in which the experimenter’s rule is a restricted version of the subject’s hypothesized rule, disconfirmation and counterfactual reasoning are invariably ineffective. Between these two extremes lies a continuum of cases in which the efficacy of the heuristics varies according to the form of the relationship between the true and hypothesized rules. 这一发现来自两种截然不同类型研究的结果。在第一项研究中，2-4-6任务的特定属性（最显著的是实验者真实规则与被试工作假设之间关系的性质）被证明会影响不同启发式方法的有效性（Klayman & Ha, 1987；Oaksford & Chater, 1994）。因此，在上述案例中，当被试的假设规则是实验者真实规则的一个受限版本时，根据逻辑可以推断，证伪和反事实推理都能通过促进排除错误的工作假设，有效地帮助发现实验者的真实规则。相比之下，在实验者的规则是被试假设规则的一个受限版本的相反情况下，证伪和反事实推理始终无效。在这两种极端情况之间存在一个连续体，其中启发式方法的有效性根据真实规则与假设规则之间关系的形式而变化。

The importance of the task environment is also evident in the results of studies of the second kind, looking at the performance of the two heuristics in an experimental setting. Here, minor variations in the procedures adopted have been found to have considerable impact on their effectiveness. For example, while Tweney et al. (1980) observed no difference between the performance of subjects using disconfirmation and those using confirmation with Wason’s (1960) original task protocol, Gorman and Gorman (1984) found that subjects instructed to use disconfirmation performed better than those instructed to use confirmation, or simply choose their own strategy, with the protocol altered so that the experimenter provides feedback on rule guesses only at the end of the experiment. Similar variations arose in experiments in which subjects were induced to employ counterfactual reasoning through the use of a facilitatory manipulation of the 2-4-6 task called “dual goal instructions,” in which subjects had to discover two rules. Thus, while Vallée-Tourangeau, Austin, and Rankin (1995) found that counterfactual reasoning typically improves performance over subjects choosing their own strategy in this case, Wharton, Cheng, and Wickens (1993) found that this was not necessarily true, the variation in results being the product of minor differences between the two studies in the information given to subjects regarding the relationship between the two rules (on this point, see Gale & Ball, 2009). 任务环境的重要性在第二类研究的结果中也很明显，这类研究考察了两种启发式方法在实验环境中的表现。在这里，研究发现，所采用程序的微小变化会对它们的有效性产生相当大的影响。例如，虽然特维尼等人（1980）观察到，使用证伪法和使用证实法的被试在沃森（1960）的原始任务方案中表现没有差异，但戈尔曼和戈尔曼（1984）发现，在实验者仅在实验结束时才对规则猜测提供反馈的方案调整下，被指示使用证伪法的被试表现优于被指示使用证实法或仅选择自己策略的被试。在一些实验中也出现了类似的变化，在这些实验中，通过对2-4-6任务进行促进性操作（称为“双重目标指令”），诱导被试进行反事实推理，要求被试发现两条规则。因此，虽然瓦莱-图朗热奥、奥斯汀和兰金（1995）发现，在这种情况下，反事实推理通常比被试选择自己策略的表现更好，但沃顿、程和威肯斯（1993）发现这并不一定成立，结果的差异是两项研究在向被试提供的关于两条规则之间关系的信息方面存在微小差异的结果（关于这一点，见盖尔和鲍尔，2009）。

Strategic Situations of Knightian Uncertainty

奈特不确定性的战略情境

The overriding message we take from the literature just reviewed is the need to be sensitive to the environment in which strategic decision-making takes place when assessing the merits of the D- and CRbased procedures for updating scenario spaces. The present section is accordingly devoted to characterizing environments of this kind, which we henceforth refer to as strategic situations of Knightian uncertainty. We focus on three characteristics we regard as having the greatest bearing on our subsequent evaluation of the two procedures, explaining each with reference to differences between strategic situations of Knightian uncertainty and the environments of Wason-style experiments. 我们从刚刚回顾的文献中得出的核心结论是，在评估基于 D 和 CR 的情景空间更新程序的优劣时，必须对战略决策所处的环境保持敏感。因此，本节致力于描述这类环境，我们此后称之为奈特不确定性下的战略情境。我们将着重阐述三个被认为对后续评估这两种程序影响最大的特征，并通过对比奈特不确定性下的战略情境与沃森式实验环境的差异来解释每个特征。

The first characteristic concerns the nature of the task facing the strategic decision-maker. Unlike Wason’s (1960) subjects, who were called on to discover what they knew to be a single true rule and to render this rule with complete accuracy, the aim for strategic decision-makers is to come up with a more accurate scenario space. Doing so means simultaneously entertaining multiple scenarios in order to increase the number of genuinely possible, while decreasing the number of genuinely impossible, scenarios in the scenario space, and where scenarios are necessarily coarse and partial, rather than wholly accurate, descriptions of possible futures. Of course, to commit to the existence of a number of genuinely possible scenarios at any one point in time is to commit to a world in which the course of history is not fully determined, and on this point we share the intuitions of Knight (1921) himself (see also Lawson, 1988). 第一个特点涉及战略决策者所面临任务的性质。与沃森（1960）的实验对象不同——他们被要求发现自己已知的单一真实规则并以完全准确的方式呈现该规则——战略决策者的目标是构建一个更准确的情景空间。要做到这一点，意味着要同时考虑多个情景，以增加情景空间中真正可能发生的情景数量，同时减少真正不可能发生的情景数量；而且，这些情景必然是对可能未来的粗略且部分的描述，而非完全准确的描述。当然，在任何特定时间点，承认存在若干真正可能的情景，就意味着承认历史进程并非完全确定，在这一点上，我们与奈特（1921）本人的直觉一致（另见劳森，1988）。

The second characteristic is that, unlike Wason’s (1960) 2-4-6 task in which the true rule remains constant throughout the course of the experiment, in strategic situations of Knightian uncertainty the set of genuinely possible scenarios a decision-maker is aiming to incorporate in their scenario space may change over time. This has two important implications for decision-making. First, it may happen that none of the scenarios included in the scenario space at the outset of a decision problem (say, time $t )$ remain genuinely possible and thus a relevant guide to action at a later point of the implementation phase (say, time $t + \ 1$ ), even if the scenario space at time $t$ was correct. Second, as the body of evidence upon which the scenario space was built at time $t$ might have become completely irrelevant by time $t ~ + ~ 1$ decision-makers updating their scenario spaces might have to generate wholly new scenarios that are disjointed from all of the initial scenarios. 第二个特征是，与沃森（1960）的2-4-6任务不同，在该任务中真实规则在整个实验过程中保持不变，而在奈特不确定性的战略情境中，决策者试图纳入其情境空间的真正可能的情景集合可能会随时间变化。这对决策有两个重要影响。首先，在决策问题开始时（例如时间t），情境空间中包含的情景中可能没有一个仍然是真正可能的，因此在实施阶段的稍后时间点（例如时间t+1），即使在时间t的情境空间是正确的，该情景也可能不再是相关的行动指南。其次，由于在时间t构建情境空间所依据的证据体到时间t+1时可能已完全不相关，更新其情境空间的决策者可能需要生成与所有初始情景不相交的全新情景。

The final characteristic we highlight is that, in contrast to Wason’s (1960) subjects who received perfectly reliable, “deterministically accurate,” feedback from the experimenter during the task (Klayman & Ha, 1987), the new information available to decisionmakers in strategic situations of Knightian uncertainty will usually be partial, subject to doubt, open to multiple and possibly conflicting interpretations, and at odds with at least some of the existing information. That is, the evidence decision-makers have access to when updating their scenario spaces will often be profoundly ambiguous and nondefinitive, two qualities we regard as major sources of Knightian uncertainty. 我们强调的最后一个特征是，与沃森（1960）的被试不同——他们在实验过程中从实验者那里获得了完全可靠、“确定性准确”的反馈（克莱曼和哈，1987）——在奈特不确定性的战略情境中，决策者可获得的新信息通常是不完整的、存疑的、存在多种可能相互冲突的解释，并且与至少部分现有信息相悖。也就是说，决策者在更新其情景空间时可获取的证据往往具有深刻的模糊性和非确定性，这两种特质我们认为是奈特不确定性的主要来源。

By evidence being ambiguous, we mean that it leaves a decision-maker uncertain about whether it increases or decreases the credibility of a scenario. The greater the degree of ambiguity, the more uncertain a decision-maker is in this regard. The significance of ambiguity for the procedures described in Table 6 lies in the fact that it leaves decision-makers susceptible to errors when analyzing how new evidence bears on the credibility of scenarios. These errors may take the form of false positives, where new evidence is wrongly judged to increase the credibility of a scenario, or false negatives, where new evidence is wrongly judged to decrease the credibility of a scenario. 我们所说的证据模糊性，是指它会让决策者不确定该证据是增加还是降低某个场景的可信度。模糊性程度越高，决策者在这方面的不确定性就越大。表6中描述的程序中，模糊性的重要性在于，它会使决策者在分析新证据如何影响场景可信度时容易出错。这些错误可能表现为假阳性（即新证据被错误地判断为增加了某个场景的可信度）或假阴性（即新证据被错误地判断为降低了某个场景的可信度）。

By evidence being nondefinitive, we mean that while it may lead to a change in the credibility of a scenario, it falls short of being able to justify the exclusion of a scenario in the “all-or-none manner” (Evans, 2016: 2082) characteristic of the evidence obtained in Wason-style tasks. The significance of nondefinitive evidence thus lies in the fact that it precludes the analog of what Bacon (1620/1905) called crucial experiments that would allow scenarios to be quickly and definitively eliminated from the scenario space. Note that the two qualities identified here are not independent: ambiguity (of any degree) is sufficient, but not necessary, for evidence to be nondefinitive, while nondefinitiveness is neither necessary nor sufficient for ambiguity (of any degree). 我们所说的证据非决定性，是指虽然它可能会导致某个场景的可信度发生变化，但不足以以“全有或全无”的方式排除某个场景（Evans, 2016: 2082），而这种方式正是华生式任务（Wason-style tasks）中获得的证据的特征。因此，非决定性证据的意义在于，它排除了培根（1620/1905）所说的“关键实验”的类比可能性——关键实验本可以让场景从场景空间中被快速且明确地排除。需要注意的是，这里确定的两种性质并非独立：任何程度的歧义都足以使证据具有非决定性，但并非必要条件；而证据的非决定性既不是歧义（任何程度）的必要条件，也不是充分条件。

Evaluating the D- and CR-Based Procedures

基于D和CR的程序评估

We now evaluate our D-and CR-based procedures for updating scenario spaces in strategic situations of Knightian uncertainty. To do so, we need to specify appropriate criteria against which to judge their performance. This is a complex issue, since the effectiveness of any method for improving the accuracy of scenario spaces might meaningfully be assessed against a number of different criteria. In what follows, we concentrate on three criteria that seem to us particularly relevant to effective decision-making in strategic settings: (a) counteracting the confirmation bias, (b) promoting exploration of the set of imaginable scenarios, and (c) facilitating action to mitigate or exploit the consequences of what would otherwise have been Black Swans. 我们现在评估基于D和CR的程序在奈特不确定性战略情境中更新情景空间的效果。为此，我们需要明确适当的标准来评判其性能。这是一个复杂的问题，因为任何提高情景空间准确性的方法的有效性，可能需要从多个不同的标准来有意义地评估。接下来，我们重点关注三个我们认为在战略环境中对有效决策特别相关的标准：(a) 抵消确认偏差，(b) 促进对可想象情景集合的探索，以及(c) 促进采取行动以减轻或利用原本可能是黑天鹅事件的后果。

Counteracting the confirmation bias: The ambiguity effect. Although the term has often been used in more specific ways (Nickerson, 1998), we adopt a broad definition of the confirmation bias as a general “inclination to retain, or a disinclination to abandon, a currently favored hypothesis” (Klayman, 1995: 386). We have selected counteracting the confirmation bias as our first criterion in view of management research findings that strategic decision-makers, similarly to Wason’s (1960) subjects, are prone to this bias in their reasoning (Bazerman & Moore, 2013; Gavetti & Rivkin, 2005; Miller, 2008; Ng, Westgren, & Sonka, 2009; Russo & Schoemaker, 1992; Weick & Sutcliffe, 2007). The paucity of research on how SWRs should be updated notwithstanding, this literature has provided strong prima facie evidence that strategic decision-makers will display the same tendency when updating their scenario spaces—namely, to be inclined toward retaining the current scenario space by seeking and interpreting information in ways favorable to existing scenarios.

抵消确认偏差：模糊效应

尽管“确认偏差”一词常被用于更具体的情境中（Nickerson, 1998），但我们采用其广义定义，即“倾向于保留而非放弃当前偏好假设的一种普遍倾向”（Klayman, 1995: 386）。鉴于管理研究的发现——战略决策者与Wason（1960）的实验对象类似，在推理过程中易受该偏差影响（Bazerman & Moore, 2013；Gavetti & Rivkin, 2005；Miller, 2008；Ng, Westgren, & Sonka, 2009；Russo & Schoemaker, 1992；Weick & Sutcliffe, 2007），我们将“抵消确认偏差”选为首要标准。尽管关于情景更新（SWRs）应如何更新的研究尚不充分，但现有文献已提供了强有力的初步证据：战略决策者在更新情景空间时，同样会表现出保留当前情景空间的倾向——即通过以有利于现有情景的方式寻求和解读信息，来维持对当前情景空间的偏好。

While both disconfirmation and counterfactual reasoning may be effective in counteracting the confirmation bias in situations of low ambiguity, such as in Wason’s (1960) original experiment, the effectiveness of both of our procedures in this regard is likely to be reduced when ambiguity is high. The reason for this lies in the reluctance of decision-makers to attach significance to evidence they believe to be ambiguous, for fear of it leading to errors in updating the credibility of scenarios. In other words, the possibility of error may impede decision-makers from incorporating ambiguous evidence into their judgments to the extent they would have were that evidence known to be unambiguous (Gorman, 1986, 1989; Klayman, 1995; Koehler, 1993; McKenzie, 2004). 虽然在低模糊性情境（如沃森1960年的原始实验）中，证伪和反事实推理都可能有效抵消确认偏差，但当模糊性较高时，我们这两种方法在这方面的有效性可能会降低。原因在于决策者不愿将他们认为模糊的证据赋予重要性，因为担心这会导致对情景可信度更新的错误。换句话说，错误的可能性可能会阻碍决策者将模糊证据纳入其判断，其程度会比已知证据明确时更低（Gorman, 1986, 1989; Klayman, 1995; Koehler, 1993; McKenzie, 2004）。

Crucially, however, the two procedures differ in the extent to which their performance in counteracting the confirmation bias is affected by ambiguity. The D-based procedure starts by assuming that the existing scenario space is correct and then attempting to eliminate scenarios from the scenario space on the basis of negative evidence. Ambiguity implies the possibility of false negatives, which has been shown to lead to “hypothesis perseveration” in experiments where decision-makers are aware of the possibility of error in the feedback they receive (Gorman, 1989: 389). Here, decision-makers tend either to immunize their hypotheses against disconfirmation by treating disconfirming data as errors, or to correctly recognize that there was no error but then devote so much time to replicating experiments to be sure of this that they failed to test their hypotheses adequately (Gorman, 1986, 1989). In short, the threat of false negatives provides grounds for decisionmakers to resist giving up on their existing scenarios immediately in the face of negative evidence. The more ambiguous the evidence, the less inclined decision-makers relying on the D-based procedure will be to give up on scenarios in their existing scenario space, and thus the less effective the procedure will be in offsetting the confirmation bias. 然而，关键的是，这两种程序在抵消确认偏差的过程中，其受模糊性影响的程度有所不同。基于D的程序首先假设现有场景空间是正确的，然后根据负面证据试图从场景空间中排除场景。模糊性意味着存在假阴性的可能性，这在实验中已被证明会导致“假设坚持”，即决策者意识到他们收到的反馈中可能存在错误（Gorman，1989：389）。在这里，决策者要么将反证数据视为错误，从而使自己的假设免受反驳，要么正确认识到没有错误，但随后花费大量时间重复实验以确认这一点，结果未能充分测试其假设（Gorman，1986，1989）。简而言之，假阴性的威胁为决策者在面对负面证据时立即放弃其现有场景提供了理由。证据越模糊，依赖基于D的程序的决策者就越不愿意放弃其现有场景空间中的场景，因此该程序在抵消确认偏差方面的效果就越差。

Ambiguity also reduces the extent to which the CR-based procedure offsets the confirmation bias, but the reduction is likely to be less significant in this case. The CR-based procedure begins by assuming that the existing scenario space is incorrect and then looking for positive evidence in favor of alternative scenarios. Consequently, ambiguous evidence once more renders decision-makers susceptible to errors when analyzing the bearing of evidence on the credibility of a scenario, in this case to false positives rather than false negatives. The reluctance of decisionmakers to accept such evidence again suggests that the more ambiguous the evidence, the less inclined decision-makers relying on the CR-based procedure will be to reduce the credibility of, and possibly to give up on, scenarios in their existing scenario space, and the less effective the procedure will be in offsetting the confirmation bias. 歧义还会降低基于CR的程序抵消确认偏差的程度，但在这种情况下，这种降低可能不太显著。基于CR的程序首先假设现有的场景空间是不正确的，然后寻找支持替代场景的积极证据。因此，当分析证据对场景可信度的影响时（在这种情况下是假阳性而非假阴性），模棱两可的证据再次使决策者容易出错。决策者不愿接受此类证据再次表明，证据越模糊，依赖基于CR的程序的决策者就越不愿意降低现有场景空间中场景的可信度，甚至可能放弃这些场景，而该程序抵消确认偏差的效果也会越差。

However, unlike the D-based procedure, there are two closely related tendencies that counteract the ambiguity of evidence reducing the power of the CRbased procedure to offset the confirmation bias: first, individuals being more inclined to interpret ambiguous evidence in a way that confirms rather than disconfirms their beliefs (McKenzie, 2004), and second, individuals being more willing to incorporate ambiguous evidence when it confirms rather than disconfirms their beliefs (Klayman, 1995). Recall that, by design, the CR-based procedure instructs decisionmakers to assume their existing scenario space is incorrect at the point scenarios are tested, and to then seek to confirm the newly generated alternative scenario. So long as decision-makers are faithful to the instruction to find evidence that confirms the alternative scenario, they will be more likely to interpret and then accept ambiguous evidence as being in favor of it. Any inclination to dismiss alternative scenarios on grounds of the potential falsity of evidence in their favor will thus be tempered. 然而，与基于D的程序不同，存在两种密切相关的趋势，它们会抵消证据模糊性的影响，从而削弱基于CR的程序抵消确认偏差的效力：首先，个体更倾向于以确认而非否定自身信念的方式解读模糊证据（McKenzie, 2004）；其次，个体在证据确认而非否定自身信念时，更愿意纳入模糊证据（Klayman, 1995）。需要回顾的是，基于CR的程序在设计上要求决策者在测试场景时，假定其现有场景空间是错误的，然后寻求确认新生成的替代场景。只要决策者忠实于“寻找支持替代场景的证据”这一指令，他们就更有可能将模糊证据解读为有利于该替代场景，并进而接受它。任何基于“支持替代场景的证据可能不真实”而否定这些替代场景的倾向，都将因此受到抑制。

Of course, false positives may lend alternative scenarios more credibility than they deserve, and thus lead to an excessive reduction of the credibility of the existing scenarios. However, given that our starting point was that strategic decision-makers suffer from confirmation bias, this effect is actually welcome in the present context for encouraging them to question their existing scenarios. We thus arrive at the following proposition. 当然，假阳性可能会让替代场景获得超出其应得的可信度，从而导致对现有场景可信度的过度降低。然而，鉴于我们的出发点是战略决策者存在确认偏差，这种效应在当前情境下实际上是受欢迎的，因为它能促使他们质疑现有场景。因此，我们得出以下命题。

Proposition 1. In strategic situations of Knightian uncertainty in which evidence is highly ambiguous, the CR-based procedure helps strategic decisionmakers offset the confirmation bias and thereby question their existing scenario space more acutely than does the D-based procedure. 命题1. 在存在高度模糊证据的奈特不确定性战略情境中，基于CR的程序有助于战略决策者抵消确认偏差，从而比基于D的程序更敏锐地质疑其现有的情景空间。

Exploring the set of imaginable scenarios. Prompted by the fact that, in strategic situations of Knightian uncertainty, the set of genuinely possible scenarios is liable to change significantly during the implementation of a decision, as well as evidence that decision-makers tend to be disinclined to generate alternative hypotheses, and when they do do so, to generate an unduly low number of them (Dougherty, Gettys, & Thomas, 1997; Fischhoff, Slovic, & Lichtenstein, 1978; Gettys, Pliske, Manning, & Casey, 1987; Heath, Larrick, & Klayman, 1998), the second criterion against which we assess the performance of our two procedures is their ability to promote exploration of the set of imaginable scenarios. In particular, we will consider how the rate at which new scenarios are generated and included in the scenario space when using each procedure is likely to be affected by the nondefinitiveness and the high degree of ambiguity of evidence. 探索可想象场景的集合。鉴于在奈特不确定性的战略情境中，真正可能的场景集合在决策实施过程中可能会发生显著变化，以及决策者往往不愿生成替代假设，即便生成，数量也过低（Dougherty, Gettys, & Thomas, 1997；Fischhoff, Slovic, & Lichtenstein, 1978；Gettys, Pliske, Manning, & Casey, 1987；Heath, Larrick, & Klayman, 1998），我们评估两种程序性能的第二个标准是它们促进对可想象场景集合进行探索的能力。具体而言，我们将考虑在使用每种程序时，证据的非确定性和高度模糊性可能如何影响新场景生成并纳入场景空间的速率。

Before we proceed, it is worth noting that the two procedures are specified to counteract a closely related tendency exhibited by decision-makers when generating hypotheses, that of coming up with alternatives that are overly similar to one another and that occupy the same general frame (Heath et al., 1998). Thus, both procedures stipulate that when the decision-maker is required to generate a new scenario, this is done by imagining an influence they had not considered before rather than simply by contemplating an additional scenario associated with influences already considered. By forcing the decision-maker to consider a new influence, the two procedures aim to promote the generation of a more diverse and heterogenous body of scenarios in the scenario space. 在我们开始之前，值得注意的是，这两种程序是为了应对决策者在生成假设时表现出的一种密切相关的倾向而制定的，即提出彼此过于相似且占据相同总体框架的替代方案（Heath等人，1998）。因此，这两种程序都规定，当要求决策者生成一个新场景时，应该通过设想一个他们之前未曾考虑过的影响因素来实现，而不是仅仅通过思考一个与已考虑过的影响因素相关的额外场景。通过迫使决策者考虑一个新的影响因素，这两种程序旨在促进在场景空间中生成更多样化和异质的场景集合。

Turning to the rate at which new scenarios are generated and included in the scenario space, the D-based procedure is fundamentally reactive in that only when sufficient negative information is found does it move from the elimination of existing scenarios from the scenario space to the formulation of new scenarios. Rather than being a necessary element of the process of scenario testing, the generation of new scenarios thus emerges only as a byproduct of the process of eliminating existing scenarios. The CR-based procedure is instead proactive as it reverses the order of these activities, moving from the generation of new scenarios to the elimination of existing ones when justified by the evidence. The generation of new scenarios is thus integral to the process of scenario testing. 转向新场景生成并纳入场景空间的速率，基于D的流程本质上是被动的，因为只有当发现足够的负面信息时，它才会从从场景空间中消除现有场景转向制定新场景。因此，新场景的生成并非场景测试过程的必要要素，而只是消除现有场景过程的副产品。相反，基于CR的流程是主动的，因为它颠倒了这些活动的顺序，在有证据支持时，从生成新场景转向消除现有场景。因此，新场景的生成是场景测试过程不可或缺的一部分。

This difference in the order of activities fundamentally affects the relative performance of the two procedures in promoting exploration of the set of imaginable scenarios when information is nondefinitive. Using the D-based procedure, decisionmakers generate a new scenario only when the search for evidence against a scenario in the current scenario space yields evidence sufficient to eliminate an existing scenario. However, while this eventuality may sometimes arise, the nondefinitive nature of information means that new (negative) evidence will often only throw doubt on a scenario, and that it will take time to acquire the disconfirming evidence required to reduce the credibility of a scenario sufficiently for it to be eliminated. This is all the more so, the higher the degree of credibility decision-makers initially attach to an existing scenario. The speed of this process of generation, and accordingly the overall number of new scenarios contemplated and eventually included in the scenario space over time, is thus significantly—and negatively—affected by the nondefinitive nature of the available evidence. 这种活动顺序的差异从根本上影响了两种程序在信息非确定性时促进对可想象场景集合进行探索的相对表现。使用基于D的程序时，决策者仅在当前场景空间中搜索针对某一场景的反证时获得了足以排除现有场景的证据后，才会生成新场景。然而，尽管这种情况有时会发生，但信息的非确定性意味着新的（负面）证据往往只会对某个场景产生怀疑，而且需要时间才能获得足以充分降低场景可信度以将其排除所需的反证。决策者最初对现有场景赋予的可信度越高，这种情况就越严重。因此，生成这一过程的速度，以及随着时间推移最终纳入场景空间的新场景的总体数量，会受到可用证据非确定性的显著（且负面）影响。

In contrast, decision-makers using the CR-based procedure begin with the assumption that their existing scenarios are deficient and are required to generate a new scenario that currently lies outside their scenario space, and then look for evidence in support of it. As before, the acquisition of new (positive) evidence will often lead to only gradual increases in the credibility of a new scenario and corresponding decreases in the credibility of existing ones. However, since this procedure does not require an existing scenario to be eliminated for a new one to be generated, the nondefinitive nature of information does not impact the rate at which decision-makers contemplate alternative scenarios. 相比之下，使用基于CR的流程的决策者从这样的假设开始：他们现有的情景是有缺陷的，并且需要生成一个当前不在其情景空间内的新情景，然后寻找支持它的证据。与之前一样，获取新的（正面的）证据通常只会导致新情景的可信度逐渐增加，而现有情景的可信度相应降低。然而，由于该流程不需要为了生成新情景而消除现有情景，因此信息的非确定性不会影响决策者思考替代情景的速度。

Moreover, the nondefinitive nature of information should not dramatically affect the rate at which newly imagined scenarios are actually included in the scenario space. For, unlike the D-based procedure, where the nondefinitive nature of information is likely to significantly prolong the time it takes to find sufficient disconfirming evidence to eliminate a scenario, in the case of the CR-based procedure this adverse effect is largely ameliorated by the relatively small amount of positive evidence needed to include a new scenario in the scenario space. In addition, with decision-makers explicitly directed to look only for positive evidence in favor of new scenarios, the CR-based procedure pushes decision-makers to take them seriously and to find support for their inclusion in the scenario space. We thus arrive at the following proposition. 此外，信息的非确定性不应显著影响新设想情景被实际纳入情景空间的速率。因为，与基于D的程序不同——在该程序中，信息的非确定性可能会显著延长寻找足够证伪证据以排除某个情景所需的时间——在基于CR的程序中，这种不利影响在很大程度上得到了缓解，因为纳入新情景到情景空间所需的正面证据相对较少。此外，由于决策者被明确引导只寻找支持新情景的正面证据，基于CR的程序促使决策者认真对待这些情景，并为将其纳入情景空间寻找支持。因此，我们得出以下命题。

Proposition 2a. In strategic situations of Knightian uncertainty in which evidence is nondefinitive, the CR-based procedure will lead to a more rapid accretion of new scenarios in the scenario space compared to the D-based procedure. 命题2a。在存在奈特不确定性的战略情境中，当证据不明确时，基于CR的程序将比基于D的程序更快地在情景空间中积累新的情景。

Turning to the highly ambiguous nature of the information typically available to strategic decisionmakers, this reduces the effectiveness of both procedures in promoting the exploration of the set of imaginable scenarios. However, the reduction due to ambiguity is likely to be less significant for the CRbased procedure than it is for the D-based procedure. 转向战略决策者通常可获得的信息的高度模糊性，这降低了两种程序在促进对可想象情景集进行探索方面的有效性。然而，与基于D的程序相比，基于CR的程序因模糊性导致的降低可能不那么显著。

Decision-makers using the D-based procedure generate a new scenario only when the newly acquired negative evidence reduces the credibility of a scenario already included in the scenario space sufficiently for it to be eliminated. The existence of ambiguity reduces the speed at which the degree of credibility is decreased since, given decisionmakers’ reluctance to attach significance to evidence they believe to be ambiguous, disconfirming data tend to be discounted when there is the possibility of false negatives. The fear of committing an error of omission will thus lead decision-makers to stick with their existing scenarios until data become more reliable. The consequence is that ambiguity reduces the speed at which new scenarios are generated, thereby having a further depressing effect on the rate at which new scenarios are contemplated and eventually included in the scenario space. 使用基于D的流程的决策者仅在新获得的负面证据降低了场景空间中已包含的某个场景的可信度，使其足以被排除时，才会生成新的场景。歧义的存在会降低可信度降低的速度，因为考虑到决策者不愿重视他们认为模糊的证据，当存在假阴性的可能性时，否定性数据往往会被低估。因此，担心犯遗漏错误会导致决策者坚持现有的场景，直到数据变得更可靠。其结果是，歧义降低了生成新场景的速度，从而进一步抑制了新场景被考虑并最终纳入场景空间的速率。

In contrast, the CR-based procedure encourages decision-makers to generate new scenarios that currently lie outside their scenario space, and to look only for evidence that might increase their credibility. As the proactive generation of new scenarios is integral to the process of testing existing scenarios, ambiguity does not affect the rate at which new scenarios are generated. However, the possibility of false positives arising from evidence being ambiguous might reduce the rate at which the degree of credibility of an alternative scenario is increased, thereby preventing its inclusion in the scenario space. The fear of committing an error of commission may thus lead decision-makers to dismiss the alternative scenarios unless and until data become more reliable. Ambiguity might therefore reduce the rate at which new scenarios are included in the scenario space. 相比之下，基于CR的流程鼓励决策者生成当前不在其情景空间内的新情景，并仅寻找可能提高其可信度的证据。由于主动生成新情景是测试现有情景的过程中不可或缺的部分，模糊性不会影响生成新情景的速度。然而，由于证据模糊而产生误报的可能性可能会降低替代情景可信度提升的速度，从而阻止其被纳入情景空间。因此，担心犯下错误行为的决策者可能会驳回替代情景，直至数据更加可靠。因此，模糊性可能会降低新情景被纳入情景空间的速度。

Nevertheless, there are two reasons the depressing effect of ambiguity on the rate at which new scenarios are included in the scenario space is likely to be less significant for the CR- than for the D-based procedure. First, as discussed above, decision-makers are more likely to take into account confirming data than disconfirming data, when data might be subject to error, and to interpret ambiguous evidence in a way that confirms rather than disconfirms their beliefs. Again, this tendency implies that the inclination to dismiss new scenarios on grounds of the potential falsity of evidence in their favor will be tempered. Of course, ambiguity might lead decision-makers to give new scenarios more credibility than they deserve. However, this effect is actually welcome in the present context as it positively affects the speed at which decision-makers using the CR-based procedure include new scenarios in the scenario space. 然而，有两个原因使得模糊性对新情景纳入情景空间速率的抑制作用，在CR-（确认-基于）程序中可能不如基于D（证伪-基于）的程序那么显著。首先，如前所述，当数据可能存在误差时，决策者更倾向于考虑支持性数据而非否定性数据，并以确认而非否定自身信念的方式解读模糊证据。同样，这种倾向意味着，因潜在证据对新情景不利而否定它们的倾向会有所缓和。当然，模糊性也可能使决策者赋予新情景超出其应有程度的可信度。不过，在当前语境下，这种效果实际上是有益的，因为它能积极提升使用CR-程序的决策者将新情景纳入情景空间的速度。

Second, decision-makers who suffer from confirmation bias are more likely to tolerate the possibility of errors of commission—including new scenarios on the basis of false positives—than the possibility of errors of omission—excluding existing scenarios on the basis of false negatives. This is because a new scenario can be added to the scenario space as long as a minimum amount of positive evidence is found and, unless the credibility of an existing scenario is reduced enough for it to be eliminated, without excluding any existing scenario. This leads to the following proposition. 其次，存在确认偏差的决策者更倾向于容忍“误报”（包括基于误报的新场景）的可能性，而非“漏报”（基于漏报排除现有场景）的可能性。这是因为只要找到最低限度的正面证据，就可以将新场景添加到场景空间中，并且除非现有场景的可信度被降低到足以将其排除的程度，否则不会排除任何现有场景。这引出了以下命题。

Proposition 2b. In strategic situations of Knightian uncertainty in which evidence is highly ambiguous, the CR-based procedure will lead to a more rapid accretion of new scenarios compared to the D-based procedure. 命题2b。在存在奈特不确定性的战略情境中，当证据高度模糊时，基于CR的程序将比基于D的程序更快地积累新场景。

Black Swans and taking action. In strategic situations of Knightian uncertainty, decision-makers’ scenario spaces are prone to rapid obsolescence, and the likelihood of Black Swans (Taleb, 2007) and their associated risks and opportunities will be high. The question then arises as to whether, under such circumstances, the D- or CR-based procedure will be more effective in promoting the uncovering of what would otherwise have been Black Swans and thereby prompting action aimed at mitigating or exploiting their consequences. 黑天鹅与采取行动。在奈特不确定性的战略情境中，决策者的场景空间容易迅速过时，黑天鹅（塔勒布，2007）及其相关风险和机遇的发生可能性将很高。那么问题就来了：在这种情况下，基于D或CR的程序是否更能有效地揭示原本可能是黑天鹅的事件，从而促使采取行动以减轻或利用其后果。

According to Taleb’s (2007: xviixviii) well-known formulation, there are three conditions for an event to qualify as a Black Swan: that it is “an outlier, as it lies outside of the realm of regular expectations, because nothing in the past can convincingly point to its possibility”; that it “carries an extreme impact”; and that “in spite of its outlier status, human nature makes us concoct explanations for its occurrence after the fact, making it explainable and predictable.” We will ignore the third condition, which seems neither an essential property of unanticipated events (Faulkner et al., 2017; Runde, 2009) nor particularly relevant to our concern about mitigating or exploiting their consequences. A natural interpretation of Black Swans from the point of view of this paper, then, is that they correspond to events or states of affairs that occur and have an extreme impact on a strategic decision at a specific point of the implementation phase (say, time $t + \mathit { 1 } )$ , but were not represented in the scenario space when it was constructed or last updated (say, time t). 根据塔勒布（2007：xvii-xviii）的著名表述，一个事件要符合黑天鹅的定义，需满足三个条件：一是“它是一个异常值，因为它超出了常规预期的范围，过去没有任何事物能令人信服地指向它的可能性”；二是“它具有极端影响”；三是“尽管具有异常值的地位，但人性使我们在事后为其发生编造解释，使其变得可解释和可预测”。我们将忽略第三个条件，因为它似乎既不是意外事件的本质属性（福克纳等人，2017；伦德，2009），也与我们关注的减轻或利用其后果的问题无关。因此，从本文的角度对黑天鹅的自然解读是：它们对应于在特定实施阶段（例如时间\( t + 1 \)）发生并对战略决策产生极端影响，但在情景空间构建或最后一次更新时（例如时间\( t \)）未被纳入考虑的事件或事态。

There are three points to note about this interpretation. First, a Black Swan as experienced by one decision-maker may not come as even a mild surprise to another. For example, the eponymous black swan (the bird) that was experienced by the Dutch explorer Willem de Vlamingh in his first encounter with the species in Western Australia in 1697 (Taleb, 2007) would have been just another black swan to the indigenous population accustomed to their presence. Second, and contrary to many people’s intuition, events experienced as Black Swans do not necessarily have to be highly improbable. Context, and especially changes in context, often matter here, as can be seen from de Vlamingh’s first encounter with a black swan actually being highly probable once he had entered its natural habitat. Likewise, in fast-changing environments, events or situations that come as significant surprises on first exposure may in fact have been highly probable given changed conditions (and may rapidly become the “new normal”). Finally, on our understanding of Black Swans, they may be positive as well as negative in their impact. 关于这种解读，有三点需要注意。首先，一个决策者经历的“黑天鹅”事件，对另一个决策者而言可能甚至不会带来轻微的意外。例如，1697年荷兰探险家威廉·德弗拉明（Willem de Vlamingh）在西澳大利亚首次遇到黑天鹅（这种鸟类）时（塔勒布，2007），对于早已习惯黑天鹅存在的原住民来说，这不过是又一只黑天鹅而已。其次，与许多人的直觉相反，被视为“黑天鹅”的事件并不一定必须是极不可能发生的。这里，背景（尤其是背景的变化）往往起着重要作用，这一点从德弗拉明首次遇到黑天鹅的经历中可见一斑——一旦他进入黑天鹅的自然栖息地，这种经历其实是极有可能发生的。同样，在快速变化的环境中，首次接触时看似重大意外的事件或情况，在条件改变后实际上可能是极有可能发生的（并且可能迅速成为“新的常态”）。最后，就我们对“黑天鹅”的理解而言，它们的影响可能是正面的，也可能是负面的。

We are now in a position to compare the relative performance of the D- and CR-based procedures in relation to uncovering what would otherwise have been Black Swans. To facilitate this comparison, we will adapt the two procedures by requiring decisionmakers, whenever prompted to generate a new scenario, to do so by imagining a crucial influence they had not considered before. By a crucial influence we mean one that would raise a new scenario with extreme consequences for the decision under consideration, and thereby prompt the generation of new possible courses of action. 我们现在可以比较基于D和CR的程序在揭示原本可能被忽视的黑天鹅事件方面的相对表现。为了便于这种比较，我们将对这两种程序进行调整，要求决策者在被提示生成新情景时，通过设想一个他们之前未曾考虑过的关键影响因素来完成。这里的关键影响因素指的是会为所考虑的决策带来极端后果的新情景，并因此促使产生新的可能行动方案。

Take first the D-based procedure, which requires decision-makers to generate new scenarios only when the newly acquired negative evidence eliminates a scenario from the existing scenario space. As summarized in propositions 2a and 2b, when information is ambiguous and nondefinitive, this feature limits the extent to which the procedure promotes the inclusion of new scenarios in the scenario space and consequently reduces the likelihood of uncovering possible high-impact future events or states of affairs before they occur. Further, by requiring decision-makers to look only for evidence against existing scenarios, the D-based procedure anchors managerial attention to prevailing expectations rather than encouraging search for information, including as-yet unimagined crucial influences that might indicate a future very different from the one expected. In this way, the procedure again works against receiving early clues about potential Black Swans. On both counts, the D-based procedure would hinder decision-makers in their efforts to anticipate in a timely way environmental changes that might have an extreme impact, and who will consequently often be left having to make ad hoc adaptive responses as events unfold. In other words, their ability to update SWRs in a way that allows them to systematically devise, and then implement, previously unconsidered courses of action appropriate to radically new conditions is limited. 首先考虑基于D的程序，该程序要求决策者仅在新获得的负面证据从现有情景空间中排除某个情景时，才生成新情景。如命题2a和2b所述，当信息模糊且非决定性时，这一特征会限制该程序促进新情景纳入情景空间的程度，从而降低在潜在高影响未来事件或事态发生前发现它们的可能性。此外，由于要求决策者仅寻找与现有情景相反的证据，基于D的程序将管理注意力锚定在普遍预期上，而非鼓励对信息的搜索，包括那些尚未被想象到的、可能表明未来与预期截然不同的关键影响因素。通过这种方式，该程序再次阻碍了对潜在黑天鹅事件早期线索的接收。从这两个方面来看，基于D的程序会阻碍决策者及时预测可能产生极端影响的环境变化，因此他们往往不得不随着事件的展开采取临时适应性反应。换句话说，他们更新情景权重比（SWRs）的能力受到限制，这种能力本应使他们能够系统地设计并实施适用于全新条件的、此前未考虑过的行动方案。

In contrast, the CR-based procedure offers clear benefits for decision-makers exposed to Black Swans. As shown in propositions 2a and 2b, by encouraging the constant and proactive generation of alternative scenarios, the procedure leads to a larger number of new scenarios being generated and eventually included in the scenario space over time. As decision-makers’ attention is now directed to imagining crucial influences and associated high-impact scenarios, the procedure also promotes the early uncovering of a larger number of potential Black Swans. The instruction to look only for evidence in favor of these new scenarios encourages decision-makers to take them seriously and to be skeptical of the existing scenarios as a guide to action. Whenever a new high-impact scenario is included in the scenario space, and whether or not one or more existing scenarios are eliminated, decision-makers are called on to update their SWRs and identify and, eventually, implement alternative courses of action to exploit or mitigate what would otherwise have been Black Swans. 相比之下，基于CR的流程为面临黑天鹅事件的决策者提供了明显优势。如命题2a和2b所示，通过鼓励持续主动地生成替代情景，该流程会在较长时间内产生并最终纳入更多新情景。随着决策者的注意力转向想象关键影响因素及相关高影响情景，该流程还能促进更早发现更多潜在的黑天鹅事件。要求仅寻找支持这些新情景的证据的指令，鼓励决策者认真对待这些新情景，并对现有情景作为行动指南持怀疑态度。每当有新的高影响情景被纳入情景空间，无论是否淘汰一个或多个现有情景，决策者都须更新其情景权重比（SWR），识别并最终实施替代行动方案，以利用或减缓原本可能成为黑天鹅事件的情况。

We conclude that in strategic situations of Knightian uncertainty in which decision-makers’ cognitive representations quickly become obsolete and are seldom adapted in a timely way, the CR-based procedure is superior to the D-based procedure in helping update SWRs in a way that favors the early uncovering of what would otherwise have been Black Swans, and the systematic identification and implementation of previously unconsidered courses of action. We thus arrive at our final proposition. 我们得出结论：在奈特不确定性的战略情境中，决策者的认知表征迅速过时且很少得到及时调整，基于CR的程序在帮助更新SWRs方面优于基于D的程序，这种更新方式有利于早期发现原本可能被视为黑天鹅的事件，以及系统地识别和实施之前未考虑过的行动方案。因此，我们得出最终命题。

Proposition 3. In strategic situations of Knightian uncertainty in which evidence is nondefinitive and highly ambiguous, the CR-based procedure is more effective than the D-based procedure in uncovering, and so facilitating action to mitigate or exploit the extreme consequences of, what would otherwise have been Black Swans. 命题3. 在存在奈特不确定性的战略情境中，当证据非确定性且高度模糊时，基于CR的程序比基于D的程序更有效地揭示那些原本可能是黑天鹅事件的极端后果，从而促进采取行动减轻或利用这些极端后果。

DISCUSSION

讨论

Our analysis contributes to the literature on how decision-makers operate in situations of Knightian uncertainty in four ways. First, we have set out a detailed conception of SWRs and of key differences between these and Savage’s (1954) small worlds, an exercise that made it possible to account for and systematize various types of uncertainty that contribute to the Knightian uncertainty often faced by strategic decision-makers, and to highlight various implications for learning in strategic contexts. As the notion of SWRs has been gaining some traction in the behavioral strategy and entrepreneurship literature (e.g., Csaszar & Levinthal, 2016; Levinthal, 2011, 2018; Maitland & Sammartino, 2015a; Packard et al., 2017), and it is often used interchangeably with terms such as mental models, simplified cognitive representations, cognitive frames, and so on (Gavetti, Levinthal, & Rivkin, 2005; Grégoire et al., 2010; Maitland & Sammartino, 2015b), we hope that this exercise will make a useful contribution to the foundations of further work on cognitive representations in management and organizational contexts. At the same time, we hope that our conceptualization of Knightian uncertainty might be of interest to those working toward a more nuanced treatment of uncertainty in entrepreneurial studies (Packard et al., 2017; Ramoglou & Tsang, 2016; Townsend et al., 2018). 我们的分析从四个方面为关于决策者如何在奈特不确定性情境下运作的文献研究做出了贡献。首先，我们详细阐述了SWRs（简化世界表示）的概念，以及它们与萨维奇（1954）提出的“小世界”之间的关键差异。这一研究使我们能够解释并系统化各种类型的不确定性——这些不确定性是战略决策者常面临的奈特不确定性的来源——并强调了战略情境中学习的多重影响。由于SWRs的概念在行为战略和创业文献中逐渐受到关注（例如，Csaszar & Levinthal, 2016；Levinthal, 2011, 2018；Maitland & Sammartino, 2015a；Packard et al., 2017），并且常与心理模型、简化认知表征、认知框架等术语互换使用（Gavetti, Levinthal, & Rivkin, 2005；Grégoire et al., 2010；Maitland & Sammartino, 2015b），我们希望这一研究能为管理和组织情境下关于认知表征的进一步研究奠定基础。同时，我们也希望我们对奈特不确定性的概念化能为创业研究中更细致处理不确定性的工作提供参考（Packard et al., 2017；Ramoglou & Tsang, 2016；Townsend et al., 2018）。

Second, we have developed a theoretical framework designed to lay bare the relationship between SWRs and the scenarios they contain, and thereby to distinguish between the set of imaginable scenarios and the scenario space, and to elaborate how scenario spaces may be updated over time. We offer this framework as a potential resource for further research into strategic decision-making under Knightian uncertainty that touches on the question of how SWRs should be updated. For example, our emphasis on the mechanisms through which the scenario space changes over time, and on establishing a clear connection between the exploration of the set of imaginable scenarios and the problem of Black Swans, might complement recent research on how to operate under Knightian uncertainty (Alvarez & Barney, 2005; Townsend et al., 2018), and then especially studies inspired by Shackle (1961, 1972, 1979; e.g., Packard et al., 2017; Porac & Tschang, 2013). Further, our framework might shed light on how assumptions about the kind of uncertainty faced by strategic decision-makers play into debates on the nature of entrepreneurial opportunities (Alvarez & Barney, 2007; Ramoglou & Tsang, 2016) and the merits of using predictive or nonpredictive approaches to strategic decision-making (Packard & Clark, 2020; Patvardhan & Ramachandran, 2020). There are also connections to be made with the work of behavioral strategists who have embraced Gilboa and Schmeidler’s (1995, 2001) case-based decision theory (e.g., Gavetti & Rivkin, 2007; Lovallo et al., 2012), given that even some of the original authors of this theory have recently argued that, in the presence of Black Swans, it may be necessary to complement case-based reasoning with scenario thinking (Gilboa et al., 2017). 其次，我们构建了一个理论框架，旨在揭示情景加权风险（SWRs）与其所包含情景之间的关系，从而区分可想象情景集合与情景空间，并阐述情景空间如何随时间更新。我们提供这一框架作为进一步研究奈特不确定性下战略决策的潜在资源，该研究涉及如何更新情景加权风险的问题。例如，我们强调情景空间随时间变化的机制，并明确建立可想象情景集合的探索与黑天鹅问题之间的联系，这可能与近期关于如何在奈特不确定性下运作的研究（Alvarez & Barney, 2005；Townsend et al., 2018）形成互补，尤其是受Shackle启发的研究（1961, 1972, 1979；例如，Packard et al., 2017；Porac & Tschang, 2013）。此外，我们的框架可能有助于阐明战略决策者面临的不确定性类型假设如何影响关于创业机会本质的辩论（Alvarez & Barney, 2007；Ramoglou & Tsang, 2016），以及在战略决策中使用预测性或非预测性方法的优劣（Packard & Clark, 2020；Patvardhan & Ramachandran, 2020）。此外，我们的框架还与行为战略学家的研究相关，他们采纳了Gilboa和Schmeidler（1995, 2001）的基于案例的决策理论（例如，Gavetti & Rivkin, 2007；Lovallo et al., 2012），因为该理论的一些原始作者最近认为，在存在黑天鹅的情况下，可能有必要将基于案例的推理与情景思维相结合（Gilboa et al., 2017）。

Third, we have extended recent work on the role of heuristics in strategic decision-making under Knightian uncertainty (Bingham & Eisenhardt, 2011; Maitland & Sammartino, 2015a; Shepherd et al., 2012) by showing how heuristic methods of inquiry from the psychology of reasoning might inform the updating of SWRs. Apart from contributing directly to research on how strategic decision-makers construct and use representations to operate in situations of Knightian uncertainty (Csaszar & Levinthal, 2016; Gavetti & Rivkin, 2007; Maitland & Sammartino, 2015a, 2015b; Porac, et al., 1989), we believe that our work might also be of interest to researchers working on cognate topics. These topics include the benefits and costs of comprehensive decisionmaking processes in turbulent environments, the meaning of rationality in situations of Knightian uncertainty, and the microfoundations of forwardlooking behavior in strategy and organizational theory (Alvarez et al., 2013; Felin, Kauffman, Koppl, & Longo, 2014; Felin & Zenger, 2017; Forbes, 2007; 第三，我们通过展示推理心理学中的启发式探究方法如何为战略决策风险（SWRs）的更新提供信息，扩展了关于奈特不确定性下战略决策中启发式作用的近期研究（Bingham & Eisenhardt, 2011; Maitland & Sammartino, 2015a; Shepherd et al., 2012）。除了直接为研究战略决策者如何在奈特不确定性情境中构建和使用表征来运作（Csaszar & Levinthal, 2016; Gavetti & Rivkin, 2007; Maitland & Sammartino, 2015a, 2015b; Porac等人, 1989）做出贡献外，我们认为我们的研究还可能引起从事相关主题研究的学者的兴趣。这些主题包括动荡环境中综合决策过程的收益与成本、奈特不确定性情境下理性的含义，以及战略和组织理论中前瞻性行为的微观基础（Alvarez等人, 2013; Felin, Kauffman, Koppl, & Longo, 2014; Felin & Zenger, 2017; Forbes, 2007;

Gavetti, 2012; Gavetti & Porac, 2018; Miller, 2008; Patvardhan & Ramachandran, 2020; Porac & Tschang, 2013; Shepherd & Rudd, 2014). Gavetti, 2012; Gavetti & Porac, 2018; Miller, 2008; Patvardhan & Ramachandran, 2020; Porac & Tschang, 2013; Shepherd & Rudd, 2014).

Fourth, we have demonstrated the importance of ecological considerations, in addition to the decisionmaker’s cognitive makeup, when attempting to assess the heuristics decision-makers use—or should use— to operate in situations of Knightian uncertainty (Gigerenzer, Todd, & The ABC Research Group, 1999; Simon, 1990). In particular, we have argued that the CR-based procedure is more effective than the D-based procedure when updating SWRs in strategic situations of Knightian uncertainty in which the information is ambiguous and nondefinitive. We hope that the approach we have used here might provide a template for further research in management aimed at taking into account the informational environment when assessing the merits of using heuristics in situations of uncertainty (Artinger, Petersen, Gigerenzer, & Weibler, 2015; Sarasvathy, Ramesh, & Forster, 2014; Shepherd et al., 2012; Zhang & Cueto, 2017). 第四，我们已经证明，在试图评估决策者在奈特不确定性（Knightian uncertainty）情境中使用（或应该使用）的启发式方法时，除了决策者的认知构成外，生态考量的重要性也不容忽视（Gigerenzer, Todd, & The ABC Research Group, 1999; Simon, 1990）。具体而言，我们认为，在信息模糊且不明确的奈特不确定性战略情境中更新社会福利比（SWRs）时，基于CR的程序比基于D的程序更有效。我们希望我们在此使用的方法能为管理领域的进一步研究提供一个模板，该研究旨在在评估不确定性情境中使用启发式方法的价值时考虑信息环境（Artinger, Petersen, Gigerenzer, & Weibler, 2015; Sarasvathy, Ramesh, & Forster, 2014; Shepherd et al., 2012; Zhang & Cueto, 2017）。

While we believe that our discussion of CR- and D-based procedures for updating scenario spaces provides some useful results for the literature on how decision-makers operate in strategic situations of Knightian uncertainty, we have only scratched the surface of the issues involved. We close with three broad avenues in which our account might be further developed. First, in this paper we have restricted ourselves to evaluating the relative effectiveness of the CR- and D-based procedures in strategic situations of Knightian uncertainty in which the information is ambiguous and nondefinitive. Our conclusion that the former is more effective than the latter is specific to this informational environment, however, and it is quite possible that the D-based procedure may be superior to the CR-based procedure in other contexts. There is considerable scope for further research here investigating how these procedures perform across a variety of informational environments (for different characterizations of the informational environment, see Todd, Gigerenzer, & The ABC Research Group, 2012), or indeed when additional situational constraints are added, such as information-acquisition costs. 虽然我们认为，针对更新情景空间的基于CR和D的程序的讨论为有关决策者在奈特不确定性战略情境中如何运作的文献提供了一些有用的结果，但我们仅触及了所涉问题的表面。我们以三个广泛的发展方向作结，我们的论述可在此基础上进一步完善。首先，在本文中，我们仅限于评估在信息模糊且非确定性的奈特不确定性战略情境中，基于CR和D的程序的相对有效性。然而，我们得出的前者比后者更有效的结论仅针对这一信息环境，而且在其他情境中，基于D的程序很可能优于基于CR的程序。这里有相当大的进一步研究空间，以调查这些程序在各种信息环境中的表现（有关信息环境的不同特征，见Todd、Gigerenzer和ABC研究小组，2012年），或者在添加额外情境约束（如信息获取成本）时的表现。

Second, our analysis has ignored various ways in which differences in the distribution of judgments of credibility between scenarios in the scenario space may be important. There is scope, for example, to explore how the specification of the D- and CR-based procedures in Table 6 might be improved by taking such differences into account, such as modifying the D-based procedure in a way that requires the decision-maker to look for evidence against a highcredibility scenario in the scenario space rather than simply an arbitrary scenario. Similarly, in evaluating the relative performance of the two procedures against a given criterion, our analysis could be extended to examine how the results depend on the distribution of credibility among the scenarios in the existing scenario space. Finally, the set of criteria we use to assess the relative performance of the procedures could be expanded to include consideration of the degree of credibility of the new scenarios introduced into the scenario space or, more generally, of those in the resulting scenario space as a whole. 其次，我们的分析忽略了场景空间中不同可信度判断分布的各种方式可能具有重要性的情况。例如，存在探索空间，以考虑这些差异来改进表6中基于D和CR的程序的规范，例如以要求决策者在场景空间中寻找反对高可信度场景的证据而不是简单地寻找任意场景的方式修改基于D的程序。同样，在评估两种程序相对于给定标准的相对性能时，我们的分析可以扩展到研究结果如何依赖于现有场景空间中场景之间的可信度分布。最后，我们用于评估程序相对性能的标准集可以扩大，以包括对引入场景空间的新场景的可信度程度的考虑，或者更一般地说，对整个结果场景空间中场景的可信度程度的考虑。

Finally, while we have focused on two specific heuristic methods of inquiry that might inform procedures for updating SWRs in the present paper, comparative exercises of the kind we have conducted could also be performed in respect of other methods that have been proposed in the philosophy of science, such as a positive-test strategy (Klayman & Ha, 1987), Bayesian epistemology (Earman, 1992), and Bacon’s (1620/1905) and Mill’s (1875/1967) methods of eliminative induction (for a review of how various philosophies of science may inform individuals’ hypothesis-testing behavior, see Tukey, 1986). An exercise on these lines would be particularly useful to explore the cases in which, rather than relying on only one method, it may be necessary to adopt a mixed-methods approach to solve a particular task. For example, some studies (Tweney et al., 1980) have suggested that disconfirmation may be useful only at later stages of a task, once at least some hypotheses have been identified and confirmed. If so, it would then be necessary to study which heuristic methods may be more effective in constructing and revising SWRs at different stages of the strategic decision-making process. 最后，虽然本文重点探讨了两种可能为更新科学工作回顾（SWRs）程序提供思路的特定启发式探究方法，但我们开展的这类比较研究也可针对科学哲学中提出的其他方法进行，例如正检验策略（Klayman & Ha, 1987）、贝叶斯认识论（Earman, 1992），以及培根（1620/1905）和密尔（1875/1967）的消除归纳法（关于各种科学哲学如何影响个体假设检验行为的综述，可参见Tukey, 1986）。沿着这一思路进行的研究对探索某些案例尤为有用——在这些案例中，仅依赖单一方法可能不足，而采用混合方法来解决特定任务或许是必要的。例如，一些研究（Tweney et al., 1980）表明，证伪可能仅在任务的后期阶段有用，此时至少已识别并确认了一些假设。若如此，就有必要研究在战略决策过程的不同阶段，哪些启发式方法在构建和修订SWRs时更为有效。

REFERENCES

参考文献

Alvarez, S. A., & Barney, J. B. 2005. How do entrepreneurs organize firms under conditions of uncertainty? Journal of Management, 31: 776793.
Alvarez, S. A., & Barney, J. B. 2005. How do entrepreneurs organize firms under conditions of uncertainty? Journal of Management, 31: 776793.

Alvarez, S. A., & Barney, J. B. 2007. Discovery and creation: Alternative theories of entrepreneurial action. Strategic Entrepreneurship Journal, 1: 1126.
阿尔瓦雷斯，S. A.，& 巴尼，J. B. 2007. 发现与创造：创业行动的替代理论。《战略创业期刊》，1: 1126.

Alvarez, S. A., Barney, J. B., & Anderson, P. 2013. Forming and exploiting opportunities: The implications of discovery and creation processes for entrepreneurial and organizational research. Organization Science, 24: 301317.
阿尔瓦雷斯，S. A.，巴尼，J. B.，& 安德森，P. 2013. 形成与利用机会：发现和创造过程对创业与组织研究的影响。《组织科学》，24：301-317。

Alvarez, S. A., & Porac, J. F. 2020. Imagination, indeterminacy, and managerial choice at the limit of knowledge. Academy of Management Review, 45: 735744.
Alvarez, S. A., & Porac, J. F. 2020. 想象力、不确定性与知识边界处的管理选择。《管理学会评论》，45: 735-744。

Arikan, A. M., Arikan, I., & Koparan, I. 2020. Creation opportunities: Entrepreneurial curiosity, generative cognition, and Knightian uncertainty. Academy of Management Review, 45. doi: 10.5465/amr.2018.0252
Arikan, A. M., Arikan, I., & Koparan, I. 2020. 创造机会：创业好奇心、生成性认知与奈特不确定性。《管理学会评论》，45。doi: 10.5465/amr.2018.0252

Artinger, F., Petersen, M., Gigerenzer, G., & Weibler, J. 2015. Heuristics as adaptive strategies in management. Journal of Organizational Behavior, 36: S33S52.
Artinger, F., Petersen, M., Gigerenzer, G., & Weibler, J. 2015. 启发式作为管理中的适应性策略。《组织行为学杂志》，36：S33-S52。

n, F. 90. No . In R. L. El J. S (Eds., Trans.), The philosophical works of Francis Bacon: 212387. London, U.K.: Routledge. (Original work published 1620)
n，F. 90。无。在 R. L. El J. S（编，译），《弗朗西斯·培根哲学著作集》：212387。英国伦敦：劳特利奇出版社。（原著出版于1620年）

Bazerman, M. H., & Moore, D. A. 2013. Judgment in managerial decision making. Hoboken, NJ: John Wiley & Sons.
Bazerman, M. H., & Moore, D. A. 2013. 管理决策中的判断。霍博肯，新泽西州：约翰·威利父子出版公司。

Berglund, H., Bousfiha, M., & Mansoori, Y. 2020. Opportunities as artifacts and entrepreneurship as design. Academy of Management Review, 45: 825846.
Berglund, H., Bousfiha, M., & Mansoori, Y. 2020. Opportunities as artifacts and entrepreneurship as design. Academy of Management Review, 45: 825846.

Bewley, T. F. 2002. Knightian decision theory. Part I. Decisions in Economics and Finance, 25: 79110. (Original work published 1986)
Bewley, T. F. 2002. 奈特决策理论。第一部分。《经济学与金融学决策》，25：79-110。（原作发表于1986年）

Bingham, C. B., & Eisenhardt, K. M. 2011. Rational heuristics: The “simple rules” that strategists learn from process experience. Strategic Management Journal, 32: 14371464.
宾厄姆，C. B.，& 艾森哈特，K. M. 2011. 理性启发式：战略家从过程经验中习得的“简单规则”。《战略管理杂志》，32：1437-1464。

Binmore, K. 2008. Rational decisions in large worlds. Annales d’Economie et de Statistique, 86: 2541.
宾莫尔，K. 2008. 大世界中的理性决策。《经济与统计年刊》，86：2541.

Binmore, K. 2009. Rational decisions. Princeton, NJ: Princeton University Press.
宾莫尔，K. 2009. 理性决策。新泽西州普林斯顿：普林斯顿大学出版社。

Bradley, R. 2017. Decision theory with a human face. Cambridge, U.K.: Cambridge University Press.
布拉德利，R. 2017. 有人情味的决策理论。英国剑桥：剑桥大学出版社。

Bradley, R., & Drechsler, M. 2014. Types of uncertainty. Erkenntnis, 79: 12251248.
布拉德利（Bradley, R.）和德雷克斯勒（Drechsler, M.），2014年。不确定性的类型。《认识》（Erkenntnis），79卷：12251248。

Brandenburger, A. 2017. Where do great strategies really come from? Strategy Science, 2: 220225.
Brandenburger, A. 2017. Where do great strategies really come from? Strategy Science, 2: 220225.

Busenitz, L., & Barney, J. 1997. Differences between entrepreneurs and managers in large organizations: Biases and heuristics in strategic decision-making. Journal of Business Venturing, 12: 930.
Busenitz, L.和Barney, J. 1997. 大型组织中企业家与管理者的差异：战略决策中的偏见与启发式方法。《商业风险杂志》，12: 930。

Cherubini, P., Castelvecchio, E., & Cherubini, A. M. 2005. Generation of hypotheses in Wason’s 246 task: An information theory approach. Quarterly Journal of Experimental Psychology, 58A: 309332.
凯鲁比尼，P.，卡斯泰尔韦基奥，E.，& 凯鲁比尼，A. M. 2005. 沃森246任务中的假设生成：一种信息论方法。《实验心理学季刊》，58A：309-332.

Csaszar, F., & Levinthal, D. A. 2016. Mental representation and the discovery of new strategies. Strategic Management Journal, 37: 20132049.
Csaszar, F., & Levinthal, D. A. 2016. 心理表征与新策略的发现。《战略管理杂志》，37: 20132049。

Cyert, R., & March, J. G. 1992. A behavioral theory of the firm. Malden, MA: Blackwell. (Original work published 1963)
Cyert, R., & March, J. G. 1992. A behavioral theory of the firm. Malden, MA: Blackwell. (Original work published 1963)

Dekel, E., Lipman, B. L., & Rustichini, A. 1998. Recent developments in modelling unforeseen contingencies. European Economic Review, 42: 523542.
Dekel, E., Lipman, B. L., & Rustichini, A. 1998. 不可预见突发事件建模的最新进展。《欧洲经济评论》，42: 523-542。

Dougherty, M. R. P., Gettys, C. F., & Thomas, R. P. 1997. The role of mental simulation in judgments of likelihood. Organizational Behavior and Human Decision Processes, 70: 135148.
道格蒂, M. R. P., 盖蒂斯, C. F., & 托马斯, R. P. 1997. 心理模拟在可能性判断中的作用. 组织行为与人类决策过程, 70: 135148.

Drechsler, M. 2012. Rationality, decisions and large worlds [Unpublished doctoral dissertation]. The London School of Economics and Political Science.
德雷舍尔，M. 2012. 理性、决策与大世界 [未发表博士论文]. 伦敦政治经济学院.

Earman, J. 1992. Bayes or bust? A critical examination of Bayesian confirmation theory. Cambridge, MA: MIT Press.
厄尔曼，J. 1992. 《贝叶斯还是破产？对贝叶斯确证理论的批判性考察》。马萨诸塞州剑桥：麻省理工学院出版社。

Ellsberg, D. 1961. Risk, ambiguity, and the Savage axioms. Quarterly Journal of Economics, 75: 643669.
埃尔斯伯格，D. 1961. 风险、模糊性与萨维奇公理。《经济学季刊》，75: 643-669.

Evans, J. St. B. T. 2007. Hypothetical thinking: Dual processes in reasoning and judgment. New York, NY: Psychology Press.
埃文斯，J. St. B. T. 2007. 假设性思维：推理与判断中的双重过程。纽约，纽约州：心理学出版社。

Evans, J. St. B. T. 2016. Reasoning, biases and dual processes: The lasting impact of Wason (1960). Quarterly Journal of Experimental Psychology, 69: 20762092.
埃文斯，J. St. B. T. 2016. 推理、偏差与双重加工过程：沃森（1960）的持久影响。《实验心理学季刊》，69：2076-2092。

Farris, H. H., & Revlin, R. 1989a. Sensible reasoning in two tasks: Rule discovery and hypothesis evaluation. Memory & Cognition, 17: 221232.
Farris, H. H., & Revlin, R. 1989a. Sensible reasoning in two tasks: Rule discovery and hypothesis evaluation. Memory & Cognition, 17: 221232.

Farris, H. H., & Revlin, R. 1989b. The discovery process: A counterfactual strategy. Social Studies of Science, 19: 497513.
Farris, H. H., & Revlin, R. 1989b. The discovery process: A counterfactual strategy. Social Studies of Science, 19: 497513.

Faulkner, P., Feduzi, A., & Runde, J. 2017. Unknowns, black swans and the risk/uncertainty distinction. Cambridge Journal of Economics, 41: 12791302.
福克纳（Faulkner, P.）、费杜齐（Feduzi, A.）和朗德（Runde, J.）。2017年。未知因素、黑天鹅与风险/不确定性的区别。《剑桥经济学杂志》，41卷：12791302。

Feduzi, A., & Runde, J. 2014. Uncovering unknown unknowns: Towards a Baconian approach to management decision making. Organizational Behavior and Human Decision Processes, 124: 268283.
Feduzi, A.，& Runde, J. 2014. 揭示未知的未知：迈向培根式的管理决策方法。《组织行为与人类决策过程》，124: 268-283.

Felin, T., Kauffman, S. A., Koppl, R., & Longo, G. 2014. Economic opportunity and evolution: Beyond landscapes and bounded rationality. Strategic Entrepreneurship Journal, 8: 269282.
Felin, T., Kauffman, S. A., Koppl, R., & Longo, G. 2014. 经济机会与进化：超越景观与有限理性。《战略创业杂志》，8：269-282。

Felin, T., & Zenger, T. R. 2017. The theory-based view: Economic actors as theorists. Strategy Science, 2: 258271.
Felin, T. 和 Zenger, T. R. 2017. 基于理论的视角：作为理论家的经济行为体。《战略科学》，2：258271。

Fischhoff, B., Slovic, P., & Lichtenstein, S. 1978. Fault trees: Sensitivity of estimated failure probabilities to problem representation. Journal of Experimental Psychology. Human Perception and Performance, 4: 330344.
菲什霍夫，B.，斯洛维奇，P.，& 利希滕斯坦，S. 1978. 故障树：估计失效概率对问题表述的敏感性。《实验心理学杂志：人类感知与表现》，4：330-344。

Forbes, D. P. 2007. Reconsidering the strategic implications of decision comprehensiveness. Academy of Management Review, 32: 361376.
福布斯，D. P. 2007. 重新考虑决策全面性的战略意义。《管理学会评论》，32：361-376。

Gale, M., & Ball, L. J. 2009. Exploring the determinants of dual goal facilitation in a rule discovery task. Thinking & Reasoning, 15: 294315.
盖尔（Gale, M.）和鲍尔（Ball, L. J.）. 2009. 规则发现任务中双目标促进作用的决定因素探索. 《思维与推理》（Thinking & Reasoning）, 15: 294-315.

Gary, M. S., Wood, R. E., & Pillnger, T. 2012. Enhancing mental models, analogical transfer, and performance in strategic decision making. Strategic Management Journal, 33: 12291246.
Gary, M. S., Wood, R. E., & Pillinger, T. 2012. Enhancing mental models, analogical transfer, and performance in strategic decision making. Strategic Management Journal, 33: 12291246.

Gavetti, G. 2012. Toward a behavioral theory of strategy. Organization Science, 23: 267285.
Gavetti, G. 2012. 战略行为理论的发展。《组织科学》，23: 267285.

Gavetti, G., Greve, H. R., Levinthal, D., & Ocasio, W. 2012. The behavioral theory of the firm: assessment and prospects. Academy of Management Annals, 6: 140.
Gavetti, G., Greve, H. R., Levinthal, D., & Ocasio, W. 2012. The behavioral theory of the firm: assessment and prospects. Academy of Management Annals, 6: 140.

Gveti, G. Helfat, C. E., & Marego, L. 017. Sear, shaping, and the quest for superior performance. Strategy Science, 2: 194209.
Gveti, G. Helfat, C. E., & Marego, L. 017. Sear, shaping, and the quest for superior performance. Strategy Science, 2: 194209.

Gavetti, G., & Levinthal, D. 2000. Looking forward and looking backward: Cognitive and experiential search. Administrative Science Quarterly, 5: 113137.
Gavetti, G.，& Levinthal, D. 2000. 向前看与向后看：认知与经验搜索。《行政科学季刊》，5：113-137。

Gavetti, G., Levinthal, D., & Rivkin, J. W. 2005. Strategy making in novel and complex worlds: the power of analogy. Strategic Management Journal, 26: 691712.
Gavetti, G., Levinthal, D., & Rivkin, J. W. 2005. Strategy making in novel and complex worlds: the power of analogy. Strategic Management Journal, 26: 691712.

Gavetti, G., & Menon, A. 2016. Evolution cum agency: Toward a model of strategic foresight. Strategy Science, 1: 207233.
Gavetti, G., & Menon, A. 2016. 演化与能动性：迈向战略预见模型。《战略科学》，1：207233。

Gavetti, G., & Porac, J. F. 2018. On the origin of great strategies. Strategy Science, 3: 352365.
Gavetti, G., & Porac, J. F. 2018. 论伟大战略的起源。《战略科学》，3：352365。

Gavetti, G., & Rivkin, J. W. 2005. How strategists really think. Harvard Business Review, 83: 5463.
加维蒂（Gavetti, G.）和里夫金（Rivkin, J. W.），2005年。战略家的真实思维方式。《哈佛商业评论》，83: 5463。

Gavetti, G., & Rivkin, J. W. 2007. On the origin of strategy: Action and cognition over time. Organization Science, 18: 420439.
Gavetti, G.，& Rivkin, J. W. 2007. 论战略的起源：随时间推移的行动与认知。《组织科学》，18：420-439。

Gettys, C. F., Pliske, R. M., Manning, C., & Casey, J. T. 1987. An evaluation of human act generation performance. Organizational Behavior and Human Decision Processes, 39: 2351.
Gettys, C. F., Pliske, R. M., Manning, C., & Casey, J. T. 1987. An evaluation of human act generation performance. Organizational Behavior and Human Decision Processes, 39: 2351.

Ghirardato, P. 2001. Coping with ignorance: unforeseen contingencies and non-additive uncertainty. Economic Theory, 17: 247276.
吉拉达托，P. 2001. 应对无知：意外突发事件与非可加性不确定性。《经济理论》，17: 247-276。

Gigerenzer, G., & Gaissmaier, W. 2011. Heuristic decision making. Annual Review of Psychology, 62: 451482.
吉仁泽（Gigerenzer, G.）和盖斯马耶（Gaissmaier, W.），2011年。启发式决策。《心理学年度评论》，62卷：451-482。

Gigerenzer, G., & Todd, P. M., & The ABC Research Group. 1999. Simple heuristics that make us smart. New York, NY: Oxford University Press.
吉仁泽（Gigerenzer, G.）、托德（Todd, P. M.）及ABC研究小组（The ABC Research Group）. 1999. 让我们变聪明的简单启发式方法. 纽约州纽约市：牛津大学出版社.

Gilboa, I., Minardi, S., & Samuelson, L. 2017. Cases and scenarios in decisions under uncertainty. Working paper No. ECO/SCD-2017-1200, HEC Paris.
吉尔博阿（I. Gilboa）、米纳尔迪（S. Minardi）和萨缪尔森（L. Samuelson）。2017年。《不确定性决策中的案例与情景》。工作论文第ECO/SCD-2017-1200号，巴黎高等商业学院（HEC Paris）。

Gilboa, I., & Schmeidler, D. 1995. Case-based decision theory. Quarterly Journal of Economics, 110: 605639.
吉尔博阿（I. Gilboa）和施梅德勒（D. Schmeidler），1995. 基于案例的决策理论。《经济学季刊》，110：605-639。

Gilboa, I., & Schmeidler, D. 2001. A theory of case-based decisions. Cambridge, U.K.: Cambridge University Press.
吉尔博阿（I. Gilboa）和施梅德勒（D. Schmeidler）. 2001. 基于案例的决策理论. 英国剑桥: 剑桥大学出版社.

Gorman, M. E. 1986. How the possibility of error affects falsification on a task that models scientific problemsolving. British Journal of Psychology, 77: 8596.
戈尔曼，M. E. 1986. 误差可能性如何影响模拟科学问题解决任务中的证伪。《英国心理学杂志》，77：8596。

Gorman, M. E. 1989. Error, falsification and scientific inference: An experimental investigation. Quarterly Journal of Experimental Psychology, 41: 385412.
戈尔曼，M. E. 1989. 误差、伪造与科学推理：一项实验研究。《实验心理学季刊》，41: 385-412.

Gorman, M. E., & Gorman, M. E. 1984. A comparison of disconfirmatory, confirmatory anda control straty on Wason’s 246 task. Quarterly Journal of Experimental Psychology, 36: 629648.
戈尔曼，M. E.，& 戈尔曼，M. E. 1984. 对沃森246任务中证伪、证实和控制策略的比较。《实验心理学季刊》，36: 629-648。

Gorman, M. E., Stafford, A., & Gorman, M. E. 1987. Disconfirmation and dual hypotheses on a more difficult version of Wason’s 246 task. Quarterly Journal of Experimental Psychology, 39: 128.
戈尔曼（Gorman, M. E.）、斯塔福德（Stafford, A.）和戈尔曼（Gorman, M. E.）. 1987. 对沃森246任务更难版本的不确认和双重假设. 《实验心理学季刊》, 39: 128.

Grégoire, D. A., Barr, P. S., & Shepherd, D. A. 2010. Cognitive processes of opportunity recognition: The role of structural alignment. Organization Science, 21: 413 431.
格雷瓜尔，D. A.，巴尔，P. S.，&谢泼德，D. A. 2010. 机会识别的认知过程：结构对齐的作用。《组织科学》，21：413-431。

Heath, C., Larrick, R. O., & Klayman, J. 1998. Cognitive repairs: How organizational practices can compensate for individual shortcomings. Research in Organizational Behavior, 20: 137.
希思（Heath, C.）、拉里克（Larrick, R. O.）和克莱曼（Klayman, J.）。1998年。认知修复：组织实践如何弥补个体缺陷。《组织行为研究》，20：137。

Hertwig, R., & Pachur, T. 2015. Heuristics, history of. In J. D. Wright (Ed.), International encyclopedia of the social & behavioral sciences: 829835. Oxford, U.K.: Elsevier.
Hertwig, R., & Pachur, T. 2015. 启发式方法，历史沿革。载于 J. D. Wright（编），《国际社会与行为科学百科全书》：829835。英国牛津：爱思唯尔出版社。

Jungerman, H., & Thuring, M. 1988. The labyrinth of the experts’ minds: Some reasoning strategies and their pitfalls. Annals of Operations Research, 16: 117130.
容格曼，H.，& 图林，M. 1988. 专家思维的迷宫：一些推理策略及其陷阱。《运筹学纪事》，16：117-130。

Kahneman, D., & Lovallo, D. 1993. Timid Choices and boFreasts:iv perpectivenski Management Science, 39: 1731.
Kahneman, D., & Lovallo, D. 1993. 胆怯的选择与过度自信：来自管理科学的视角，39: 1731.

Keynes, J. M. 1973. A treatise on probability. Reprinted in: The collected writings of John Maynard Keynes, vol. VIII, London, U.K.: Macmillan. (Original work published in 1921.)
凯恩斯，J. M. 1973. 概率论论文集。重印于：《约翰·梅纳德·凯恩斯全集》第VIII卷，英国伦敦：麦克米伦出版公司。（原著发表于1921年。）

Klayman, J. 1995. Varieties of confirmation bias. Psychology of Learning and Motivation, 32: 385418.
克莱曼，J. 1995. 确认偏差的种类。《学习与动机心理学》，32：385-418。

Klayman, J., & Ha, Y. W. 1987. Confirmation, disconfirmation and information in hypothesis testing. Psychological Review, 94: 211228.
克莱曼，J.，& 哈，Y. W. 1987. 假设检验中的确认、不确认与信息。《心理评论》，94：211228。

Klingebiel, R., & De Meyer, A. 2013. Becoming aware of the unknown: Decision making during the implementation of a strategic initiative. Organization Science, 24: 133153.
Klingebiel, R., & De Meyer, A. 2013. 意识到未知：战略举措实施过程中的决策。《组织科学》，24：133153。

Knight, F. H. 1921. Risk, uncertainty and profit. Chicago, IL: University of Chicago Press.
奈特，F. H. 1921. 风险、不确定性与利润。伊利诺伊州芝加哥：芝加哥大学出版社。

Koehler, J. J. 1993. The influence of prior beliefs on scientific judgments of evidence quality. Organizational Behavior and Human Decision Processes, 56: 2855.
科勒，J. J. 1993. 先前信念对科学证据质量判断的影响。《组织行为与人类决策过程》，56：2855。

Langley, A., Mintzberg, H., Pitcher, P., Posada, E., & Saintmacary, J. 1995. Opening up decision making: The view from the black stool. Organization Science, 6: 260279.
兰利（Langley, A.）、明茨伯格（Mintzberg, H.）、皮彻（Pitcher, P.）、波萨达（Posada, E.）和圣马卡里（Saintmacary, J.）. 1995. 开放决策制定：从黑凳子视角看. 《组织科学》（Organization Science）, 6: 260-279.

Levinthal, D. A. 2011. A behavioral approach to strategy— What’s the alternative? Strategic Management Journal, 32: 15171523.
莱文瑟尔，D. A. 2011. 一种行为战略方法——还有其他选择吗？《战略管理杂志》，32: 1517-1523.

Levinthal, D. A. 2018. From strategy to strategic organization. In M. Augier, C. Fang, & V. P. Rindova, (Eds.), Behavioral strategy in perspective (Advances in strategic management): 7177. Bingley, U.K.: Emerald Publishing.
莱文瑟尔，D. A. 2018. 从战略到战略组织。载于M. 奥吉埃、C. 方及V. P. 林多瓦（编），《行为战略的视角》（《战略管理进展》）：7177。英国宾利：爱墨瑞得出版公司。

Lawson, T. 1988. Probability and uncertainty in economics analysis. Journal of Post Keynesian Economics, 11: 3865.
劳森，T. 1988. 经济学分析中的概率与不确定性。《后凯恩斯经济学杂志》，11：3865。

Lovallo, D., Clarke, C., & Camerer, C. 2012. Robust analogizing and the outside view: Two empirical tests of case-based decision making. Strategic Management Journal, 33: 496512.
洛瓦洛（Lovallo, D.）、克拉克（Clarke, C.）和卡梅勒（Camerer, C.）。2012年。稳健类比与外部视角：基于案例决策的两项实证检验。《战略管理杂志》，33: 496-512。

Lovallo, D., & Kahneman, D. 2003. Delusions of success: How optimism undermines executives’ decisions. Harvard Business Review, 81: 5663.
洛瓦洛（Lovallo, D.）和卡尼曼（Kahneman, D.），2003年。成功的错觉：乐观如何破坏高管的决策。《哈佛商业评论》，81：5663。

Maitland, E., & Sammartino, A. 2015a. Decision making and uncertainty: The role of heuristics and experience in assessing a politically hazardous environment. Strategic Management Journal, 36: 15541578.
梅特兰，E.，& 萨马蒂诺，A. 2015a. 决策与不确定性：启发式和经验在评估政治危险环境中的作用。《战略管理杂志》，36: 1554-1578.

Maitland, E., & Sammartino, A. 2015b. Managerial cognition and internationalization. Journal of International Business Studies, 46: 733760.
梅特兰，E.，& 萨马蒂诺，A. 2015b. 管理认知与国际化。《国际商务研究杂志》，46: 733760.

Manktelow, K. 2012. Thinking and reasoning. An introduction to the psychology of reason, judgment and decision making. Hove, U.K.: Psychology Press.
曼克特洛，K. 2012. 思维与推理：理性、判断与决策心理学导论。英国霍夫：心理学出版社。

McKenzie, C. R. M. 2004. Hypothesis testing and evaluation. In D. J. Koehler & N. Harvey (Eds.), Blackwell handbook of judgment and decision making: 200 Malden,U.. Blackwell ublihig
麦肯齐，C. R. M. 2004. 假设检验与评估。载于 D. J. 科勒 & N. Harvey（编），《布莱克韦尔判断与决策手册》：200 马尔登，美国。布莱克韦尔出版公司

McMullen, J. S., & Shepherd, D. A. 2006. Entrepreneurial action and the role of uncertainty in the theory of the entrepreneur. Academy of Management Review, 31: 131152.
麦克马伦，J. S.，& 谢泼德，D. A. 2006. 创业行动与不确定性在创业理论中的作用。《管理学会评论》，31: 131152。

Mill, J. S. 1967. A system of logic (8th ed.). London, U.K.: Longmans. (Original work published 1875)
密尔，J. S. 1967. 《逻辑体系》（第8版）。英国伦敦：朗曼斯。（原著出版于1875年）

Miller, C. C. 2008. Decisional comprehensiveness and firm performance: Towards a more complete understanding. Journal of Behavioral Decision Making, 21: 598620.
Miller, C. C. 2008. 决策全面性与企业绩效：迈向更完整的理解。《行为决策杂志》，21：598620。

Mousavi, S., & Gigerenzer, G. 2017. Heuristics are tools for uncertainty. Homo Oeconomicus, 34: 361379.
穆萨维（Mousavi, S.）和吉仁泽（Gigerenzer, G.），2017年。启发式是应对不确定性的工具。《经济人》（Homo Oeconomicus），34卷：361-379页。

Mukerji, S. 1997. Understanding the non-additive probability decision model. Economic Theory, 9: 2346.
Mukerji, S. 1997. 理解非可加概率决策模型。《经济理论》，9：2346。

Ng, D., Westgren, R., & Sonka, S. 2009. Competitive blind spots in an institutional field. Strategic Management Journal, 30: 349369.
Ng, D., Westgren, R., & Sonka, S. 2009. 制度领域中的竞争性盲点。《战略管理杂志》，30：349-369。

Nickerson, R. S. 1998. Confirmation bias: A ubiquitous phenomenon in many guises. Review of General Psychology, 2: 175220.
Nickerson, R. S. 1998. Confirmation bias: A ubiquitous phenomenon in many guises. Review of General Psychology, 2: 175220.

Oaksford, M., & Chater, N. 1994. Another look at eliminative and enumerative behaviour in a conceptual task. European Journal of Cognitive Psychology, 6:149169.
Oaksford, M., & Chater, N. 1994. 概念任务中排除式和枚举式行为的再考察。《欧洲认知心理学杂志》，6：149-169。

Packard, M. D., & Clark, B. B. 2020. On the mitigability of uncertainty and the choice between predictive and non-predictive strategy. Academy of Management Review. doi: 10.5465/amr.2018.0198
帕卡德（M. D.）和克拉克（B. B.），2020年。论不确定性的可缓解性及预测性与非预测性策略的选择。《管理学会评论》。doi: 10.5465/amr.2018.0198

Packard, M. D., Clark, B. B., & Klein, P. G. 2017. Uncertainty types and transitions in the entrepreneurial process. Organization Science, 28: 840856.
帕卡德（M. D.）、克拉克（B. B.）和克莱因（P. G.）. 2017. 创业过程中的不确定性类型与转变. 《组织科学》, 28: 840-856.

Patvardhan, S., & Ramachandran, J. 2020. Shaping the future: Strategy making as artificial evolution. Organization Science, 31: 671697.
Patvardhan, S.， & Ramachandran, J. 2020. 塑造未来：作为人工进化的战略制定。《组织科学》，31：671697。

Poletiek, F. H. 2001. Hypothesis-testing behaviour. Hove, U.K.: Psychology Press.
Poletiek, F. H. 2001. 假设检验行为。英国霍夫：心理学出版社。

Popper, K. R. 1959. The logic of scientific discovery. London, U.K.: Hutchinson.
波普尔，K. R. 1959. 科学发现的逻辑。英国伦敦： Hutchinson。

Pontikes, E. G., & Rindova, V. P. 2020. Shaping markets through temporal, constructive, and interactive agency. Strategy Science, 5: 149159.
Pontikes, E. G., & Rindova, V. P. 2020. 通过时间性、建设性和互动性能动性塑造市场。《战略科学》，5：149159。

Porac, J. F., Thomas, H., & Baden-Fuller, C. 1989. Competitive groups as cognitive communities: The case of Scottish knitwear manufacturers. Journal of Management Studies, 26: 397414.
波拉克，J. F.，托马斯，H.，& 巴登-富勒，C. 1989. 竞争性群体作为认知社区：以苏格兰针织品制造商为例。《管理研究杂志》，26: 397-414.

Porac, J. F., & Tschang, F. T. 2013. Unbounding the managerial mind: It’s time to abandon the image of managers as “small brains”. Journal of Management Inquiry, 22: 250254.
波拉克，J. F.，& 张（音译），F. T. 2013. 解放管理思维：是时候摒弃管理者“头脑狭隘”的形象了。《管理探究期刊》，22：250-254。

Porter, M. E. 1985. Competitive advantage: Creating and sustaining superior performance. New York, NY: Free Press.
波特，M. E. 1985. 竞争优势：创造并保持卓越绩效。纽约，纽约州：自由出版社。

Ramoglou, S., & Tsang, E. W. K. 2016. A realist perspective of entrepreneurship: Opportunities as propensities. Academy of Management Review, 41: 410434.
拉莫格卢（Ramoglou, S.）和曾（Tsang, E. W. K.），2016年。创业的现实主义视角：机会作为倾向。《管理学会评论》，41: 410434。

Reymen, I. M., Andries, P., Berends, H., Mauer, R., Stephan, U., & Van Burg, E. 2015. Understanding dynamics of strategic decision making in venture creation: A process study of effectuation and causation. Strategic Entrepreneurship Journal, 9: 351379.
Reymen, I. M., Andries, P., Berends, H., Mauer, R., Stephan, U., & Van Burg, E. 2015. Understanding dynamics of strategic decision making in venture creation: A process study of effectuation and causation. Strategic Entrepreneurship Journal, 9: 351379.

Rindova, V. P., & Courtney, H. 2020. To shape or adapt: Knowledge problems, epistemologies and strategic postures under Knightian uncertainty. Academy of Management Review. doi: 10.5465/amr.2018.0291.
Rindova, V. P., & Courtney, H. 2020. To shape or adapt: Knowledge problems, epistemologies and strategic postures under Knightian uncertainty. Academy of Management Review. doi: 10.5465/amr.2018.0291.

Runde, J. 1998. Clarifying Frank Knight’s discussion of the meaning of risk and uncertainty. Cambridge Journal of Economics, 22: 539546.
Runde, J. 1998. 澄清弗兰克·奈特对风险和不确定性含义的讨论。《剑桥经济学杂志》，22: 539-546。

Runde, J. 2009. Dissecting the black swan. Critical Review, 21: 491505.
Runde, J. 2009. 剖析黑天鹅。《批判性评论》，21：491505。

Russo, J., & Schoemaker, P. 1992. Managing overconfidence. Sloan Management Review, 33: 717.
Russo, J., & Schoemaker, P. 1992. 管理过度自信。《斯隆管理评论》，33：717。

Sarasvathy, S. D., Ramesh, A., & Forster, W. R. 2014. The ordinary entrepreneur. In F. Welter & T. Baker (Eds.), The Routledge companion to entrepreneurship: 227244. Abingdon, VA: Routledge Press.
Sarasvathy, S. D., Ramesh, A., & Forster, W. R. 2014. The ordinary entrepreneur. In F. Welter & T. Baker (Eds.), The Routledge companion to entrepreneurship: 227244. Abingdon, VA: Routledge Press.

Savage, L. J. 1954. The foundations of statistics. London, U.K.: John Wiley & Sons.
萨维奇，L. J. 1954. 统计学基础. 英国伦敦: 约翰·威利父子出版公司.

Schoemaker, P. J. H. 1993. Multiple scenario development: Its conceptual and behavioral foundation. Strategic Management Journal, 14: 192213.
Schoemaker, P. J. H. 1993. 多情景开发：其概念与行为基础。《战略管理杂志》，14：192-213。

Shackle, G. L. S. 1961. Decision, order, and time in human affairs. Cambridge, U.K.: Cambridge University Press.
沙克尔，G. L. S. 1961. 人类事务中的决策、秩序与时间。英国剑桥：剑桥大学出版社。

Shackle, G. L. S. 1972. Epistemics and economics: A critique of economic doctrines. Cambridge, U.K.: Cambridge University Press.
沙克尔，G. L. S. 1972. 认识论与经济学：对经济理论的批判。英国剑桥：剑桥大学出版社。

Shackle, G. L. S. 1979. Imagination and the nature of choice. Edinburgh, U.K.: Edinburgh University Press.
Shackle, G. L. S. 1979. 想象力与选择的本质. 英国爱丁堡: 爱丁堡大学出版社.

Shafer, G. 1986. Savage revisited. Statistical Science, 1: 463485.
谢弗，G. 1986. 重审萨维奇。《统计科学》，1: 463-485。

Shepherd, D. A., Haynie, J. M., & McMullen, J. S. 2012. Confirmatory search as a useful heuristic? Testing the veracity of entrepreneurial conjectures. Journal of Business Venturing, 27: 637651.
Shepherd, D. A., Haynie, J. M., & McMullen, J. S. 2012. Confirmatory search as a useful heuristic? Testing the veracity of entrepreneurial conjectures. Journal of Business Venturing, 27: 637651.

Shepherd, D. A., Williams, T. A., & Patzelt, H. 2015. Thinking about entrepreneurial decision making: Review and research agenda. Journal of Management, 41: 1146.
Shepherd, D. A., Williams, T. A., & Patzelt, H. 2015. Thinking about entrepreneurial decision making: Review and research agenda. Journal of Management, 41: 1146.

Shepherd, N. G., & Rudd, J. M. 2014. The influence of context on the strategic decision-making process: A review of the literature. International Journal of Management Reviews, 16: 340364.
Shepherd, N. G., & Rudd, J. M. 2014. The influence of context on the strategic decision-making process: A review of the literature. International Journal of Management Reviews, 16: 340364.

Simon, H. A. 1990. Invariants of human behavior. Annual Review of Psychology, 41: 119.
西蒙，H. A. 1990. 人类行为的不变量。《心理学年度评论》，41: 119.

Simon, H. A. 1997. Administrative behavior (4th ed.). New York, NY: Free Press.
西蒙，H. A. 1997. 行政行为（第4版）。纽约，纽约州：自由出版社。

Taleb, N. N. 2007. The black swan: The impact of the highly improbable. London, U.K.: Penguin books.
塔勒布，N. N. 2007. 黑天鹅：如何应对不可预知的未来。英国伦敦：企鹅出版社。

Todd, P. M., & Gigerenzer, G., & The ABC Research Group. 2012. Ecological rationality: Intelligence in the world. New York, NY: Oxford University Press.
托德（Todd, P. M.）、吉格伦泽（Gigerenzer, G.）及ABC研究小组。2012年。《生态理性：世界中的智慧》。纽约，纽约州：牛津大学出版社。

Townsend, D. M., Hunt, R. A., McMullen, J. S., & Sarasvathy, S. D. 2018. Uncertainty, knowledge problems, and entrepreneurial action. Academy of Management Annals, 12: 659687.
汤森德，D. M.，亨特，R. A.，麦克马伦，J. S.，& 萨拉维蒂，S. D. 2018. 不确定性、知识问题与创业行动。《管理学会年鉴》，12：659-687。

Tripsas, M., & Gavetti, G. 2000. Capabilities, cognition and inertia: Evidence from digital imaging. Strategic Management Journal, 21: 11471161.
Tripsas, M., & Gavetti, G. 2000. 能力、认知与惯性：来自数字成像领域的证据。《战略管理杂志》，21：1147-1161。

Tukey, D. D. 1986. A philosophical and empirical analysis of subjects’ modes of inquiry in Wason’s 2-4-6 task. Quarterly Journal of Experimental Psychology, 38: 533.
Tukey, D. D. 1986. 对Wason的2-4-6任务中被试探究模式的哲学与实证分析。《实验心理学季刊》，38: 533.

Tweney, R. D., Doherty, M. E., Worner, W. J., Pliske, D. B., Mynatt, C. R., Gross, K. A., & Arkkelin, D. L. 1980. Strategies of rule discovery in an inference task. Quarterly Journal of Experimental Psychology, 32: 109 123.
特韦尼（Tweney, R. D.）、多尔蒂（Doherty, M. E.）、沃纳（Worner, W. J.）、普利斯克（Pliske, D. B.）、迈纳特（Mynatt, C. R.）、格罗斯（Gross, K. A.）和阿尔克林（Arkkelin, D. L.）. 1980. 推理任务中的规则发现策略. 《实验心理学季刊》, 32: 109-123.

Vallée-Tourangeau, F., Austin, N. G., & Rankin, S. 1995. Inducing a rule in Wason’s 2-4-6 task: A test of the information quantity and goal-complementarity hypotheses. Quarterly Journal of Experimental Psychology, 48A: 895914.
Vallée-Tourangeau, F., Austin, N. G., & Rankin, S. 1995. 诱导Wason 2-4-6任务中的规则：信息数量和目标互补性假设的检验。《实验心理学季刊》，48A：895-914。

Wason, P. C. 1960. On the failure to eliminate hypotheses in a conceptual task. Quarterly Journal of Experimental Psychology, 12: 129140.
沃森，P. C. 1960. 论概念任务中消除假设的失败。《实验心理学季刊》，12: 129-140.

Weick, K. E., & Sutcliffe, K. M. 2007. Managing the unexpected. Resilient performance in an age of uncertainty (2nd ed.). San Francisco, CA: John Wiley & Sons.
Weick, K. E., & Sutcliffe, K. M. 2007. Managing the unexpected. Resilient performance in an age of uncertainty (2nd ed.). San Francisco, CA: John Wiley & Sons.

Wharton, C. M., Cheng, P. W., & Wickens, T. D. 1993. Hypothesis-testing strategies: Why two goals are better than one. Quarterly Journal of Experimental Psychology, 46: 743758.
沃顿（Wharton, C. M.）、程（Cheng, P. W.）与威肯斯（Wickens, T. D.）. 1993. 假设检验策略：为何两个目标优于一个目标. 《实验心理学季刊》, 46: 743-758.

Zhang, S., & Cueto, J. 2017. The study of bias in entrepreneurship. Entrepreneurship Theory and Practice, 41: 419454. Zhang, S., & Cueto, J. 2017. The study of bias in entrepreneurship. Entrepreneurship Theory and Practice, 41: 419454.

X

X

Alberto Feduzi (a.feduzi@jbs.cam.ac.uk) is management practice associate professor at Cambridge Judge Business School. He holds a PhD from the University of Cambridge and his research focuses on judgment and decisionmaking under uncertainty. 阿尔贝托·费杜齐（a.feduzi@jbs.cam.ac.uk）是剑桥大学贾奇商学院的管理实践副教授。他拥有剑桥大学博士学位，研究方向为不确定性条件下的判断与决策。

Philip Faulkner (pbf1000@cam.ac.uk) is an emeritus fellow of Clare College, Cambridge; and a fellow of Cambridge Judge Business School. A coeditor of the Cambridge Journal of Economics, his research focuses on the nature of technology and decision-making under extreme uncertainty. 菲利普·福克纳（pbf1000@cam.ac.uk）是剑桥克莱尔学院的名誉研究员，同时也是剑桥Judge商学院的研究员。他是《剑桥经济学杂志》的联合编辑，研究重点是技术的本质以及极端不确定性下的决策制定。

Jochen Runde (j.runde@jbs.cam.ac.uk) is professor of economics and organization at Cambridge Judge Business School, and professorial fellow at Girton College, Cambridge. He is coeditor of the Cambridge Journal of Economics and his main areas of research are decision-making under uncertainty, social ontology and the ontology of technology, and explanation in the social sciences. 约亨·伦德（j.runde@jbs.cam.ac.uk）是剑桥大学贾奇商学院的经济学与组织学教授，同时也是剑桥大学格顿学院的教授研究员。他是《剑桥经济学杂志》的联合编辑，主要研究领域包括不确定性下的决策制定、社会本体论、技术本体论以及社会科学中的解释问题。

Laure Cabantous (laure.cabantous.1@city.ac.uk) is professor of strategy and organization at Bayes Business School, City, University of London. She studies decision-making and leadership practices in organizations using a performative perspective. She also has an interest in embodied ways of knowing and modes of reasoning in organizational settings. Laure Cabantous (laure.cabantous.1@city.ac.uk) 是伦敦城市大学贝叶斯商学院的战略与组织学教授。她从表演性视角研究组织中的决策和领导实践，还对组织环境中具身认知方式和推理模式感兴趣。

Christoph H. Loch (c.loch $@$ jbs.cam.ac.uk) is professor of operations and technology management and former dean of Cambridge Judge Business School. His research examines innovation processes, project management, strategy execution, and the motivation of professional employees. 克里斯托夫·H·洛克（c.loch @ jbs.cam.ac.uk）是运营与技术管理教授，曾任剑桥Judge商学院院长。他的研究涉及创新流程、项目管理、战略执行以及专业员工的激励机制。

Copyright of Academy of Management Review is the property of Academy of Management and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder’s express written permission. However, users may print, download, or email articles for individual use. 《管理学会评论》的版权归管理学会所有，未经版权所有者明确许可，其内容不得复制、发往多个网站或发布到邮件列表。不过，用户可打印、下载或通过电子邮件发送文章供个人使用。