Skip to navigation – Site map

HomeIssues6:2When does policy learning lead to...

When does policy learning lead to policy change? Exploring the causal chain from learning to change

Sandra Plümer
p. 200-219

Abstract

Policy learning is a crucial mechanism for policy change. Yet, there is still uncertainty about the conditions under which learning actually leads to change. This article clarifies the causal chain from policy learning to policy change in two steps. First, it develops a so-called “Learning Product Framework” which distinguishes three central features of learning products: policy belief change, policy preference change, and policy output change. Second, it presents a “Typology of Causal Pathways between Learning and Change”, leading to four different learning-induced policy changes. In the first pathway, policy beliefs have changed, but preferences and outputs remain unchanged, resulting in policy stability rather than policy change. In the second pathway, policy beliefs and preferences have changed, but the output has not been altered, also leading to policy stability. In the third pathway, beliefs, preferences, and outputs have changed, but they are not aligned, resulting in “Non-Congruent Policy Change”. Only in the fourth pathway are all three features aligned and fulfilled, leading to “Congruent Policy Change”. This conceptual clarification confirms previous findings that policy learning alone is not sufficient for policy change. It demonstrates the combination of cognitive, behavioral, and social mechanisms needed for learning-induced policy change.

Top of page

Full text

I would like to thank the anonymous referees for their insightful and helpful comments, and the editors of the journal for their excellent support throughout the submission and publication process. Special thanks go to the guest editors of this issue for their valuable feedback on multiple occasions.

1. Introduction

1Learning is fundamental to policymaking. Policymakers, both individually and collectively, learn about the severity of problems, possible solutions, policy instruments, and the overall goals of policy. From an information processing perspective, policy learning is defined as “the circulation and consumption of policy issue-related information and knowledge” (Zaki, Wayenberg, & George, 2022, p. 22), and refers to cognitive and behavioral processes. Policy learning is closely related to policy change. Indeed, “learning is a critical mechanism for policy change and improving how information, ideas, and knowledge are integrated into governance processes” (Gerlak et al., 2020; Heikkila et al., 2023, p. 2).

2Policy change resulting from policy learning is known as “learning-induced policy change” (Moyson, 2017). It is distinct from other pathways to policy change, such as those resulting from exogenous pressures or systemic shocks (Jenkins-Smith et al., 2018; Kamkhaji & Radaelli, 2017). Yet, there is still uncertainty about the conditions under which learning actually leads to change. Taking learning as the independent variable that affects policy change (Goyal & Howlett, 2024), this article seeks to clarify the causal chain from policy learning to policy change. It addresses the research question: When does policy learning lead to policy change?

3Scholars have explored the learning-change relationship for more than 30 years. Despite this longstanding research tradition, conceptual ambiguities persist regarding the distinction between learning and change, blurring the causal chain (Goyal & Howlett, 2019, 2024). Nonetheless, significant progress has been made. Contributions on the social dimension of the learning-change relationship emphasize the need for learning and change to materialize across individuals, collectives, and subsystems (e.g., Dunlop & Radaelli, 2017; Moyson et al., 2017). Additionally, the behavioral dimension is explored when examining the role of policy actors in transforming learning into change (e.g., Goyal & Howlett, 2024; Goyal, 2022), while the cognitive dimension focuses on the mental processes that translate learning into change (e.g., Moyson, 2017; Nowlin, 2021). Although these three dimensions and their findings are prevalent throughout the literature, they have not been systematically combined. This article aims to address this gap by leveraging the explanatory power of combining the three dimensions to draw a more holistic picture of the learning-change relationship.

4For this purpose, the article adopts the perspective of learning products (Heikkila & Gerlak, 2013). Learning products address the outcomes or results of learning, answering the questions ‘what is learned?’ and ‘to what effect?’ (Bennett & Howlett, 1992). Research on learning products has developed a variety of frameworks, including categorizations of learning-induced policy changes (P. A. Hall, 1993), different types of learning products (Heikkila & Gerlak, 2013), learning types and heuristics that highlight potential policy change (Dunlop & Radaelli, 2013; Trein & Vagionaki, 2021), and modes to evaluate the quality of learning products and the resulting changes (Dunlop, 2017a). While these studies enhance our understanding of the multiple facets of learning products, these insights have not yet been extensively applied to elaborate on the causal chain between learning and change. Therefore, this article adopts the learning product perspective to further explore this causal chain.

5The theoretical contribution of this article is two-fold: First, it develops a “Learning Product Framework” which distinguishes three central features of learning products: policy belief change, policy preference change, and policy output change. Second, based on this framework, it presents a “Typology of Causal Pathways between Learning and Change”, leading to four different learning-induced policy changes: In the first pathway, policy beliefs have changed but preferences and outputs remain unchanged, resulting in policy stability rather than policy change. In the second pathway, policy beliefs and preferences have changed, but the output has not been altered, also leading to policy stability. In the third pathway, beliefs, preferences, and outputs have changed, but they are not aligned, resulting in what I term “Non-Congruent Policy Change”. Only in the fourth pathway are all three features fulfilled, leading to “Congruent Policy Change”.

6The conceptual clarification developed in this article confirms the previous finding that policy learning alone is not sufficient to lead to policy change. It identifies the specific combination of cognitive, behavioral, and social mechanisms needed for learning-induced policy change. The framework and typology developed in this article serve as useful heuristics in two ways. First, they facilitate the systematic analysis of different learning outcomes – policy beliefs, preferences, and outputs – that had not previously been distinguished in such detail. Second, they identify the levels at which learning products might be flawed and thus fail to ‘travel’. Here, the causal chain from learning to change is interrupted, severely limiting learning-induced policy change (Dunlop, 2017a, 2017b; Dunlop & Radaelli, 2018b). Both the framework and the typology can be used to deepen the theoretical understanding of the learning-change relationship and to systematically analyze learning-induced policy changes in empirical research. The application of these tools in future studies can provide a robust ‘proof of concept’.

7The article proceeds as follows. After reviewing the relevant literature (section 2), the causal chain from learning to change is clarified in two steps. In a first step, the article develops the “Learning Product Framework” that distinguishes three features of learning products (section 3) and in a second step, it presents the “Typology of Causal Pathways from Learning to Change” with four different learning-induced policy changes (section 4). The article ends with a conclusion (section 5).

2. Literature review

8In this section, I will review two strands of literature: First, I will outline articles that investigate the causal chain from policy learning to policy change and point to different dimensions of the relationship. Second, I will review studies that analyze the causal chain from the perspective of learning products. In addition to summarizing the relevant existing literature, this section also presents important concepts that will be used in the theoretical part of the article (sections 3 and 4) and clarifies the analytical perspective of this contribution.

2.1 Exploring the causal chain from policy learning to policy change

  • 1 This is also visible in some articles’ titles, framing learning as change, e.g. “Do Disasters Lead (...)

9In 1992, Colin J. Bennett and Michael Howlett published one of the first contributions to examine the relationship between policy learning and policy change. The authors emphasized analytical difficulties in both identifying learning and tracing the causal chain from learning to change: “We may only know that learning is taking place because policy change is taking place” (Bennett & Howlett, 1992, p. 290). More than 30 years later, Zaki (2024) notes that most of the literature still does not distinguish precisely between learning and change, leaving the causal chain blurred.1

10This conclusion, however, highlights a certain degree of analytical imprecision rather than a lack of scientific progress in the field. Indeed, numerous contributions address the relationship between learning and change. These include ‘early’ attempts to theorize learning and change (e.g., P. A. Hall, 1993; Rose, 1991) as well as more recent research on the topic (e.g., Busetti & Righettini, 2023, Dunlop & Radaelli, 2020; Moyson et al., 2017). Furthermore, an entire special issue is dedicated to theorizing learning and change (Moyson et al., 2017).

11The most comprehensive approach on this topic to date is presented by Dunlop and Radaelli (2017), who elaborate on the micro and macro dimensions of the learning-change relationship. Using Coleman’s (1986) ‘bath-tub’ model, they map the causal chain through three mechanisms: macro-to-micro mechanisms concerning the foundations of individual behavior, micro-to-micro mechanisms highlighting interactions between individuals in collective settings, and micro-to-macro mechanisms covering the scaling up of individual and group behavior to the societal level. The meta-framework demonstrates that the relationship between learning and change must materialize multidimensionally across individuals, collectives, and the governance system (Moyson et al., 2017; Zaki, Wayenberg, & George, 2022). The same observation applies to policy learning alone: individual learning needs to aggregate and translate into collective learning to bring about change (Gerlak & Heikkila, 2011; Heikkila & Gerlak, 2013). This research underscores the social dimension of the learning-change relationship.

12Empirical studies investigating the relationship between policy learning and policy change confirm its multidimensionality and highlight the importance of agency (Zaki, 2024). For instance, Busetti and Righettini (2023) identify four possible outcomes in their municipal case studies on the Italian food stamp program implemented during COVID-19: learning and change, learning without change, change without learning, and no change/no learning. They emphasize that entrepreneurship, political strategy, and proximity between learners and decision-makers are fundamental for learning to translate into change.

13The importance of policy entrepreneurship is also confirmed by Goyal and Howlett (2024), who investigate how learning enables change as policy innovation. Goyal (2022) further argues for the necessity of a “transnational policy entrepreneur” in learning-change processes, conceptualizing learning as policy diffusion. Ritter et al. (2018) show that the absence of a policy entrepreneur explicitly hinders learning-induced policy change. Similarly, Hatch and Mead (2021) introduce the concept of “entrepreneur catalyzed learning”, highlighting the critical role of advocates for learning and change. These studies emphasize the behavioral dimension of the learning-change relationship and examine interaction processes at the micro, meso, and – to some extent –, the macro levels that link learning to change. Despite these insights, scholars emphasize that the mechanisms of the causal chain remain unclear (Dunlop & Radaelli, 2020; Zaki, Wayenberg, & George, 2022), particularly regarding “whether, how, and when” learning leads to change (Goyal & Howlett, 2024, p. 2).

14Further investigating the ‘when’, a limited number of studies focuses on the cognitive processes of learners that explain the conditions under which learning leads to change. While research has provided comprehensive overviews of cognitive biases during learning (see, e.g., Heikkila et al., 2023), these findings are rarely applied to the causal chain from learning to change. An exception is Moyson (2017), who explores the cognitive consistency of policy learning and its impact on the likelihood of policy change. According to this research, the causal chain from learning to change involves changed beliefs that translate into changed policy preferences. This study significantly enhances our understanding of the cognitive dimension of the learning-change relationship by systemically distinguishing between policy beliefs and policy preferences (see also 2.2).

15In sum, current literature highlights that policy learning alone is not sufficient for policy change. The social dimension of the learning-change relationship explains that individual learning is not enough; it must spread to a group of policymakers and manifest in the fabric of the system. The behavioral dimension of the learning-change relationship shows that learning requires agency to develop into policy change, with actions, interactions, and key policy entrepreneurs playing pivotal roles. The cognitive dimension of the learning-change relationship demonstrates that a change in beliefs does not necessarily lead to policy change. Policy preferences must align with the changed beliefs. Although these three dimensions and their findings are prevalent throughout the literature, they have not been systematically combined. This article aims to address this gap by exploiting the explanatory power of combining the three dimensions to draw a more holistic picture of the learning-change relationship.

2.2 Analyzing learning products and their contribution to policy change

16To clarify the chain between learning and change, researchers can either focus on learning products or on learning processes (Heikkila & Gerlak, 2013). Learning products, on the one hand, represent the outcomes or results of learning (Ormrod, 2008; Zaki, Wayenberg, & George, 2022). This analytical perspective answers two main questions in the learning literature (Bennett & Howlett, 1992): ‘what is learned?’ and ‘to what effect?’. Examples of learning outcomes include new or revised legislations and reforms. Studying learning processes, on the other hand, involves examining a potential change in the “mechanisms of behavior” (Domjan, 2010; G. Hall, 2003). In other words, this analytical perspective seeks to uncover the mechanisms that drive behavior and behavioral change. Despite these different focuses, research has shown that the products and processes of learning are interconnected, as the learning process can shape the nature of the learning product (Crow & Albright, 2019; Gerlak & Heikkila, 2011; Zaki, Wayenberg, & George, 2022).

  • 2 On the other hand, studying learning as a process only implicitly addresses policy change. These st (...)

17While recognizing the strong connections between learning products and learning processes, this article exclusively focuses on learning products, as they explicitly highlight potential policy change resulting from learning. By examining the outcomes of learning, it is possible to compare learning products before and after the learning process, clearly identifying both the amount and direction of changes. This approach allows for a thorough assessment of the potential policy change induced by learning (e.g., Weible et al., 2023).2

18However, the literature dealing specifically with learning products is limited. Instead, numerous articles analyze learning products alongside the processes of learning, often using the collective learning framework (Heikkila & Gerlak, 2013). This framework acknowledges both dimensions of learning but primarily emphasizes the procedural dynamics. For example, in an application of the framework, Osei‐Kojo et al. (2022) describe the three phases of learning in a case study on COVID-19 mitigation in Ghana, with only a brief mention of the behavioral learning products (see also Koebele, 2019).

19Contributions that exclusively concentrate on learning products address various facets, such as magnitudes or types of learning-induced changes (see below). Nevertheless, these insights have not been extensively applied to elaborate on the causal chain between learning and change. The present article therefore uses these observations to clarify the causal mechanisms between learning and change.

20Previous articles have focused on four main areas. First, scholars have developed categorizations of learning-induced policy changes. P. A. Hall (1993), for example, proposed three different “orders of change“ resulting from policy learning: first-order changes indicating small adjustments in a policy, second-order changes altering policy instruments and, third-order changes amending the basic goals of a policy. Additionally, Sabatier and Weible (2007) refer to major and minor policy change in order to highlight the scope of change. Major policy change involves changes in beliefs about governance modes within a particular policy field, while minor policy change pertains to alterations of beliefs about policy instruments. Policy learning typically leads to minor policy change (Pierce et al., 2020; Sabatier & Jenkins-Smith, 1999), with some exceptions (e.g., Bandelow et al., 2019).

21Second, moving from these general classifications to a more specific understanding, Heikkila and Gerlak (2013) investigated different types of learning products, which hint at various forms of policy changes. They distinguish between cognitive and behavioral learning products. Cognitive products are new or reinforced policy beliefs, further subdivided by Moyson (2017) into policy beliefs and policy preferences (Fishbein & Ajzen, 2010; VandenBos, 2007; see sections 3 and 4). These cognitive learning products precede new programs and policies as behavioral learning products. The typology of cognitive and behavioral learning products offers a tool to differentiate between cognitive products that are inward-oriented, and behavioral products that are outward-oriented. These outward-oriented learning products manifest in new legislation that may induce policy change (Zaki, Wayenberg, & George, 2022). As section 3 will show, these distinctions are crucial for the conceptual clarification of the causal chain from learning to change.

22Third, scholars have identified various learning types and heuristics. While these often focus on learning processes and the ‘how’ of learning (e.g., learning in reflexivity, Dunlop & Radaelli, 2013), they can also help to describe what is learnt and to what effect. Building on the three learning types developed by Bennett and Howlett (1992), these types include political learning (May, 1992), power-oriented learning (Trein & Vagionaki, 2021), social learning (P. A. Hall, 1993), policy-oriented learning (Sabatier, 1988), and organizational learning (Stark & Head, 2019). A recent and comprehensive review of policy learning scholarship identifies over 60 different policy learning types (Zaki, Wayenberg, & George, 2022), while another review finds that over 50 percent of learning articles introduce or rely on learning types (Squevin et al., 2021). But what do these types tell us about learning products? Distinguishing between policy- and power-oriented learning, for example, indicates whether policymakers have learned about political strategies to increase their own influence (strategy change) or if they have updated their policy ideas to improve programs (idea change; Vagionaki & Trein, 2019). Conversely, organizational learning involves acquiring knowledge about organizational proceedings (organizational change; Argyris & Schön, 1996). These types and heuristics assist in defining what I will subsequently call “policy output change” (see section 3).

23Fourth, scholars have evaluated the quality of learning products, enabling forecasts of the extent and quality of learning-induced policy changes. Adopting a more normative tone than the analytical approaches mentioned above and viewing policy learning (in the form of belief change) as a desirable feature of policymaking, they argue that policy learning does not necessarily lead to improved policy outcomes (Dunlop, 2017b; Radaelli, 2009). Policymakers might draw “erroneous inferences” (Levitt & March, 1988) or react destructively to new information (Örtenblad, 2002), leading to “pathologies of learning” or “dysfunctional learning” (Dunlop, 2017a; Dunlop & Radaelli, 2018a).

24Learning pathologies can also be found within different learning types. In epistemic learning, for example, policymakers may learn the ‘wrong’ lessons (Dunlop, 2017a), or counter-mobilize epistemic expertise (Rietig, 2018), both of which lead to limited learning and change. In hierarchical contexts, learning may be blocked where “administrative, institutional and sociopolitical aspects in the policy system (…) block the transfer of the newly acquired knowledge into policy outputs” (Vagionaki, 2018, pp. 194-195). More recently, Möck and Feindt (2023) investigated learning mode misfits within different learning types.

3. Developing the “Learning Product Framework”

25Based on the literature review in the previous section, I will now present the first step of the conceptual clarification of the learning-change relationship: the development of a “Learning Product Framework”. This framework distinguishes three central features of learning products. These features evaluate (1) whether the observed learning product reflects a change in policy beliefs (3.1), (2) whether it signifies a change in policy preferences (3.2), and (3) whether it results in a change in policy outputs (3.3). The first and the second feature address cognitive learning products (policy beliefs and policy preferences), while the third feature encompasses the transition to behavioral learning products (updating of existing policies or new policies) as an indication of policy change (Heikkila & Gerlak, 2013; Heikkila et al., 2023).

26Overall, the framework combines previous findings on the different dimensions of the learning-change relationship (see 2.1): The first and the second feature address the cognitive dimension, while the third feature refers to the behavioral and social dimensions of learning. Moreover, the framework builds on the perspective of learning products established in the existing literature (see 2.2).

3.1 First feature: Policy belief change

27The first feature covers the cognitive dimension of the learning-change relationship and assesses the extent to which a learning product is a manifestation of policy belief change. Before addressing policy belief change, it is essential to define what a policy belief is. Beliefs capture fundamental values as well as policy-related ideas (Sabatier & Weible, 2007) and function as sources through which actors filter and interpret the world (Nohrstedt et al., 2023; Sabatier, 1993). As part of the ideational turn in public policy (Béland, 2019; Béland & Cox, 2010) and the ‘ideas school’ (P. A. Hall, 1993; Kamkhaji & Radaelli, 2022), beliefs have become a central analytical category in the study of policy learning and policy change (Cairney, 2020). According to the Advocacy Coalition Framework (ACF), a belief-oriented framework used to explain learning and change, individuals hold a three-tiered belief system consisting of deep core beliefs (fundamental normative values and ontological ideas, e.g., beliefs about human nature), policy core beliefs (normative and empirical policy-related beliefs, e.g., beliefs about priorities in a subsystem or the seriousness of a specific policy problem), and secondary beliefs (beliefs about policy tools or instruments to solve policy problems, e.g., on the use of bans in a specific policy area; Nohrstedt et al., 2023).

28A change in any of these beliefs indicates that they have been altered into the direction of the new information received (Nowlin, 2021). From this information processing perspective on learning, “learning is a function of the strength of prior beliefs and the weight given to new information” (Nowlin, 2021, p. 1020). When policymakers process new information, they evaluate it against the backdrop of their current policy beliefs (Bullock, 2009). In case of learning as belief change, the weight given to the new information surpasses the current beliefs.

29Based on this systematic approach to integrate insights from information processing into the learning literature, Nowlin (2024) further develops his approach and forwards a refined conceptualization of both beliefs and learning. First, the author introduces the concepts of central tendency (the content of the belief itself) and variance (the certainty associated with the belief, meaning how certain someone is that the belief is ‘true’). The author posits that beliefs are a function of the beliefs as such (central tendency) and the (un-)certainty (variance) associated with them. Second, he illustrates different belief distributions where (a) beliefs are strongly held and the likelihood of learning is very limited, (b) beliefs are strong but uncertainty exists, allowing for some potential learning, and (c) beliefs with complete uncertainty, which are most likely to change (see also Lablih et al., 2024). Third, these findings are connected to the three-tiered belief system of the ACF, indicating the likelihood of change for the different belief layers. Policy core beliefs, on the one hand, are strong but low to medium levels of uncertainty might still exist, leading to a medium likelihood of change. Secondary beliefs, on the other hand, vary between normal to high levels of uncertainty with a high likelihood of change (Nohrstedt et al., 2023; Sabatier & Jenkins-Smith, 1999).

30While learning has often been theorized as belief change, research has also shown that it leads to belief reinforcement. Belief reinforcement is possible because individuals, characterized by bounded rationality (Simon, 1991), are subject to various biases when processing information (Baumgartner & Jones, 2015). A well-known bias is motivated reasoning (Kahan, 2013; Kunda, 1990), which describes a cognitive response in which individuals favor information that supports existing beliefs and reject information that contradicts them. Consequently, individuals either ignore information that does not align with their beliefs (selective exposure, Hart et al., 2009) or interpret pieces of information in a way that supports current beliefs (biased assimilation, Munro & Ditto, 1997). Other cognitive biases closely related to these mechanisms include confirmation biases (Nickerson, 1998), availability biases (Sunstein, 2006), and negativity biases (Baumeister et al., 2001; for an overview of other cognitive biases, see Heikkila et al., 2023). These phenomena represent cognitive barriers to potential policy belief changes in the context of policy learning.

31Cognitive biases are not only researched on a theoretical level. Empirical research has also demonstrated instances of belief reinforcement. For example, Weible et al. (2023) analyze oil and gas development in Colorado, USA, and mostly find policy belief reinforcement despite rare instances of belief change. Other studies, both adopting an ACF perspective on learning and using the literature that develops a learning theory, confirm the possibility of learning as belief reinforcement (Moyson, 2017; Nowlin, 2024; Pattison, 2018).

32While acknowledging the potential for learning as belief reinforcement, empirical research shows that policymakers can indeed change their beliefs through learning. In such cases, individuals overcome cognitive biases and engage in policy learning over time, as Montpetit and Lachapelle (2017) demonstrate in their micro-level study on shale gas development in two Canadian provinces. Furthermore, Beach et al. (2021) highlight the importance of analogical reasoning as a cognitive tool for transferring lessons from the past to the present in epistemic learning. Their findings suggest that analogical reasoning plays a crucial role in epistemic learning processes, even amid other cognitive constraints, which are particularly strong in complex and high-stake policy situations. Additionally, anxiety about possible devastating policy-related consequences can positively and significantly influence policy learning (Lablih et al., 2024). This micro-level research suggests that while cognitive biases may restrict learning, their impact can be mitigated through cognitive tools and emotional states, which help counteract these biases.

33Other studies adopting a meso-level perspective also find belief changes in the context of learning. With a recent and strong research focus on the COVID-19 pandemic, numerous articles investigate learning during crises and confirm belief changes among policymakers (e.g., Busetti & Righettini, 2023; Lee et al., 2020; Zaki, Pattyn, & Wayenberg, 2022; Zaki & Wayenberg, 2020; for an overview see Zaki & Wayenberg, 2023). For example, Crow et al. (2023) investigate COVID-19 policy responses across six US states and find ‘preemptive policy learning’ where policymakers have learned in anticipation of an emerging hazard. In addition to learning during and after the pandemic, belief changes are observed in the context of other crises, such as the European financial crises (Kamkhaji & Radaelli, 2017), and across various policy fields, such as pension politics (Leifeld, 2013) and water policy (Leach et al., 2014).

34To conclude, learning as belief reinforcement prevents learning-induced policy changes, whereas learning as belief change creates the possibility for such changes. Therefore, the first feature assesses whether learning manifests as a change in beliefs, thereby enabling potential policy change.

3.2 Second feature: Policy preference change

35Similar to the first feature, the second feature addresses the cognitive dimension of the learning-change relationship. This feature examines whether a change in policy beliefs (first feature) leads to a change in policy preferences (Moyson, 2017). By following the information processing perspective on learning, it measures the extent to which a learning product reflects a change in policy preferences.

36First, it is essential to define the concept of policy preferences. Policy preferences refer to the goals of actors within a specific policymaking context (Lindberg et al., 2019). They constitute significant elements in the policymaking process because existing policies are refined or new policies drafted based on these (changed) policy preferences (Goldstein & Keohane, 1993; Jenkins-Smith et al., 2018). In this way, policy preferences play an important role in policy formulation and influence how policy actors frame problems, advocate for or against certain policy instruments, and engage with others (Cairney & Weible, 2017; Sabatier & Weible, 2007).

37Distinguishing between policy beliefs (first feature) and policy preferences (second feature) requires understanding them as two distinct cognitive learning products. In fact, the majority of the existing literature does not clearly distinguish between the updating of policy beliefs and policy preferences. For example, Jenkins-Smith et al. (2018) define policy core beliefs but also refer to them as policy preferences within the ACF (see also Nohrstedt et al., 2023). This conflation occurs because the ACF, following the approach of Festinger (1975), posits an “internally consistent system” of policy beliefs and policy preferences, where a change in beliefs necessarily leads to a change in preferences.

38On the other hand, and consistent with the argumentation presented in this article, several scholars acknowledge the differences between beliefs and preferences. Referring to the above-mentioned three-tiered belief system in the ACF, Bolognesi et al. (2024) write: “Such approach (…) involves quite strong assumptions about how actors manage to link the different beliefs levels when forming their preferences” (own emphasis). Despite recognizing that the formation of preferences is different from the formation of beliefs, the authors do not further analyze the relationship between the two. One possible explanation is that some researchers, without clearly articulating it, may understand policy preferences as behavioral learning products, while policy beliefs are considered cognitive learning products, rendering further differentiation unnecessary (Heikkila & Gerlak, 2013; Heikkila et al., 2023). However, both beliefs and preferences can be considered cognitive learning products as they involve the internal processing of new information and knowledge (Zaki, Wayenberg, & George, 2022). This perspective aligns with VandenBos (2007) and Fishbein and Ajzen (2010), who emphasize that policy learning and learning-induced change require actors to first modify their existing beliefs, then transform their preferences (conceptualized as ‘attitudes’), and finally alter their behavioral intentions and concrete behavior (see section 3.3).

39A pioneering contribution that clearly differentiates between policy beliefs and policy preferences in a policy learning context is the study by Moyson (2017). According to this contribution, the statement ‘“this policy has had more negative effects than expected’” reflects a (changed) policy belief, while a sentence highlighting a (changed) policy preference is ‘“I am less favorable toward this policy than before’” (Moyson, 2017, p. 321). Therefore, the policy belief about the seriousness of a problem or the effects of a policy differs from the policy preference, which articulates a concrete attitude and opinion on a particular policy, typically formulated as ‘I am in favor …’ or ‘I am against …’.

  • 3 The article refrains from a repetition of the biases.

40In order to ensure internal consistency between beliefs and preferences, policymakers who have changed their policy beliefs are assumed to also change their policy preferences accordingly (see above, Festinger, 1975; Gawronski & Strack, 2012). In such cases, this is referred to as “consistent policy learning” (Moyson, 2017): “Policy learning is consistent when actors align their policy preferences with their updated beliefs” (Moyson, 2017, p. 321). Nonetheless, policy actors can also maintain their policy preferences or modify them in the opposite direction, leading to “inconsistent policy learning”. Therefore, cognitive biases are observed not only in regard to the first feature (belief change) but also to the second feature (preference change).3

41In line with Moyson (2017, p. 338), who concludes that “learning consistency is an important condition of learning-induced policy changes”, this article argues that researchers need to investigate whether changed policy beliefs (first feature) transition to changed policy preferences (second feature) when exploring learning-induced policy changes.

3.3 Third feature: Policy output change

42The third feature addresses the behavioral and social dimensions of the learning-change relationship. It examines whether a change in policy preferences (second feature) leads to a change in policy outputs. In other words, it explores whether cognitive lessons learned transition into tangible policy outputs. By ‘lessons of learning’, I refer to the products of learning that have policy implications (May, 1992), whether in terms of the social construction of problems and policies (social learning, P. A. Hall, 1993) or the use of policy instruments (instrumental learning, May 1992; for other learning types and the ‘what’ of learning see the overviews by Zaki, Wayenberg, & George, 2022 as well as Riche et al., 2021, and section 2).

  • 4 Despite this widely shared assumption, the policy learning literature also describes instances wher (...)

43Analyzing the chain from beliefs and preferences to policy outputs, the feature investigates the transition from cognitive learning products to behavioral learning products (Heikkila & Gerlak, 2013). Cognitive learning products hence precede the emergence of behavioral learning products. This assumption aligns not only with the majority of the policy learning literature (Cook & Yanow, 1993; Heikkila & Gerlak, 2013, p. 492; Zollo & Winter, 2002), but also with cognitive psychology (Fishbein & Ajzen, 2010; VandenBos, 2007).4 Thus, this feature addresses the behavioral dimension of learning (see 2.1).

44Furthermore, the third feature shifts from an information processing perspective on learning (Moyson, 2017; Nowlin, 2021) to a perspective that focuses on the implementation of learning products and the lessons learned (Brown & Stark, 2022; Schofield, 2004; Stark & van der Arend, 2023). This approach allows for an analysis of how lessons learned are integrated into subsequent policymaking. It builds on the social dimension of learning (see section 2.1), emphasizing the need for the learning-change relationship to materialize multidimensionally across individuals, collectives, and the governance system (Moyson et al., 2017; Zaki, Wayenberg, & George, 2022).

45This social dimension is closely linked to recent scholarly emphasis on the importance of the policy process in studying learning. This includes both the political context of learning and collective decision-making processes (Dunlop & Radaelli, 2013; Kamkhaji & Radaelli, 2022; Trein & Vagionaki, 2021). In this vein, Zaki, Wayenberg, and George (2022, p. 20) propose the concept of ‘policy-embedded learning’, “where learning is submerged with the fabric and context of the policy process”. While this notion is often used to investigate learning processes, this article emphasizes the significance of the policy process for the study of learning products.

46Exploring the implementation of lessons learned is essential when investigating learning-induced policy changes, as the policy process may lead to instances where lessons learned do not result in refined or new policies. In these cases, “policy actors learn but opt not to disseminate the lessons to decision-makers or fail to incorporate that knowledge into governance practices” (Heikkila et al., 2023, p. 2; Vagionaki, 2018). This means that cognitive learning products covering feature one (belief change) and feature two (preference change) might emerge but will not be converted into behavioral learning products (new or refined policies). This implementation process is referred to as “translation” (Clarke et al., 2015; Freeman, 2008, 2009). From this perspective, implementation capacity is a crucial component of “policy learning effectiveness”, which involves the implementation of policy learning lessons (Stark & van der Arend, 2023). Schofield (2004) and Biegelbauer (2016) term this process as ‘managerial learning’.

47The implementation of lessons learned is connected to the decision-making phase of the policy cycle (Lasswell, 1956), and focuses on policymaking actors who possess the authority to implement these lessons. P. A. Hall (1993) identifies governments as the key policymakers responsible for the translation of lessons learned into visible policy change, as they have the power to initiate new policy programs. Similarly, the ACF views governmental authority to make policy decisions as a critical resource in the policymaking process (Jenkins-Smith et al., 2018; Nohrstedt, 2011). When governmental policymakers do not engage in this activity, it results in the non-implementation of lessons learned due to policy inaction (Brown & Stark, 2022; McConnell & ’t Hart, 2019). Consequently, the capacity of lessons to ‘travel’ is significantly limited (Stark & van der Arend, 2023, p. 1).

48To explore policy (in-)action within governments, scholars have employed the concept of core executives (Andeweg et al., 2020; Dunleavy & Rhodes, 1990; Rhodes, 1995). Core executives are defined as “all those organizations and structures which primarily serve to pull together and integrate central government policies, or act as final arbiters within the executive of conflicts between different elements of the government machine” (Andeweg et al., 2020; Dunleavy & Rhodes, 1990, p. 4; Elgie, 2011). These “different elements” refer to executive, legislative, and bureaucratic actors who interact within the core executive to engage in policy decision-making (Bach & Wegrich, 2020). Despite potential rivalries among these actors (Pierre & Peters, 2017; 't Hart & Wille, 2006), core executives must maintain a minimum level of interaction in order to engage in policy action and make policy decisions (Page, 2012). This interaction shapes everyday policymaking and contributes to the transition of lessons learned into new or refined policy programs, ultimately leading to learning induced-policy change.

49To conclude, the third feature shifts the analytical perspective not only toward implementing lessons learned into subsequent policymaking but also toward central governmental actors who have the authority to decide on new or refined policies. Recognizing the importance of these actors does not mean neglecting the role of other actors in the policymaking process, such as interest groups, who seek to influence central decision-makers and engage with those actors in policy communities (Herweg et al., 2023) or advocacy coalitions (Nohrstedt et al., 2023).

4. Developing the “Typology on Causal Pathways between Learning and Change”

50Building on the “Learning Product Framework”, I will now present the second step of the conceptual clarification – the “Typology of Causal Pathways between Learning and Change”, leading to different learning-induced policy changes. The typology specifies the causal workings of distinct pathways between learning and change, and accounts for the different forms of learning-induced policy changes as outcomes of these causal pathways. The pathways and their outcomes vary with regard to which of the learning features (see section 3) are fulfilled.

51Compared to other recent typologies on this topic, such as the one developed by Busetti and Righettini (2023), the typology herein departs from the observation that policy learning, defined as a change in beliefs, has occurred. In this way, the typology adopts a ‘positive’ view on learning, specifically covering instances where learning as belief change has initially taken place. The emphasis on learning as belief change aligns with the aim of this article: explaining the causal chain from policy learning to policy change. If learning as belief change cannot be observed, then learning-induced policy change cannot be identified either. Consequently, several other learning and change-related phenomena remain outside the scope of this typology: learning as belief reinforcement (Pattison, 2018; Weible et al., 2023), which validates the status quo and therefore does not lead to policy change; non-learning (Radaelli, 2009), which leads to either no policy change at all or to policy change resulting from partisan effects, coercion or other factors; and learning that covers the alteration of policy preferences but not a change in policy beliefs (Jenkins-Smith et al., 2018; see also section 3).

52Table 1 displays four causal pathways from policy learning to policy change, including different learning-induced policy changes as outcomes of these pathways. The pathways distinguish distinct configurations of the learning product features outlined in section three (feature 1: policy belief change; feature 2: policy preference change; feature 3: policy output change).

53In the first pathway, feature one is fulfilled, while features two and three are not fulfilled, resulting in policy stability rather than policy change. In the second pathway, features one and two are met, while feature three is not fulfilled, also leading to policy stability. In the third pathway features one, two, and three are met, but they are not aligned, leading to “Non-Congruent Policy Change”. Only in the fourth pathway are all three features fulfilled and properly aligned, leading to “Congruent Policy Change”.

Table 1: Causal pathways between learning and change, including different learning-induced policy changes

Table 1: Causal pathways between learning and change, including different learning-induced policy changes

Source: Own compilation. A hook () indicates the feature is fulfilled, a cross () indicates the feature is not met.

54In the following section, I will detail the different causal pathways and the resulting learning-induced policy changes. This includes empirical examples from the existing literature used for illustrative purposes. While some examples of each pathway can be found in the existing literature, the underlying causal chains have not been thoroughly elaborated. Consequently, this section not only empirically demonstrates the theoretical pathways developed in this article but also categorizes the examples according to their pathway logic. It should be noted that these examples primarily focus on individual learners, even though these individuals are part of collective settings.

55In the first pathway, policy beliefs have changed (feature one), but preferences and outputs (features two and three) remain unchanged. This configuration of features leads to policy stability rather than policy change.

56Moyson (2017) provides an empirical example of the first pathway, analyzing policy learning and change related to the European liberalization process of the rail and electricity sectors. Using Belgium as a case study, the author conducted a web survey and collected data from over 50 public and private organizations involved in the liberalization process. He finds that “many policy actors adapt their beliefs about policy outcomes without aligning their policy preferences” (Moyson 2017, p. 335). According to the author, the statement “‘I believe that new evidence demonstrates the environmental inefficiency of this policy’” indicates changed beliefs, while the following expression shows that policy actors have not aligned their policy preferences with the changed beliefs: “‘I maintain my support for this policy because the economic efficiency of policies is more important than their environmental efficiency’”. As an ACF-inspired study, the second sentence reveals that policymakers did not change their preferences with respect to the policy core, which involves the overall goals in the subsystem (economic versus environmental efficiency). Because policy preferences remain unchanged in this example, individuals did not advocate for any policy changes. In a collective learning and decision-making setting where the majority of policymakers do not change their policy preferences in accordance with their changed beliefs, this configuration of features leads to an absence of learning-induced policy changes.

57In the second pathway, policy beliefs and preferences (features one and two) have changed, but the output (feature three) remains unchanged. Similar to the first pathway, this scenario leads to policy stability, albeit with a different configuration of features.

58This pathway is highlighted by Busetti & Righettini (2023), who investigate learning and change regarding the implementation of the emergency food stamp program during the COVID-19 lockdown (Busetti & Righettini, 2023). The authors study the program in three Italian municipalities (Padua, Vicenza, Verona) with a variety of data sources, including legislative and administrative documents, newspapers, and interviews. They develop a learning matrix that portrays different relationships between learning and change, using the three case studies to illustrate various outcomes. One such relationship is learning without change, which was particularly evident in Padua. Although the social service managers responsible for implementing the program reported learning, administrative inertia prevented policy change.

59The absence of transitions from either changed policy beliefs to policy preferences and policy outputs, as seen in the first pathway, or from changed policy beliefs and policy preferences to policy outputs, as highlighted by the second pathway, can be attributed to several factors that are collectively referred to as “policy inaction” (e.g., Brown & Stark, 2022; McConnell & ’t Hart, 2019). These factors include a lack of support from other powerful actors (imposed inaction) or a lack of appropriate tools and resources for action (reluctant inaction). Additionally, policymakers may consciously decide not to act (calculated inaction) (e.g., as a result of a cost-benefit calculation; McConnell & ’t Hart, 2019, p. 650).

60Transferring these types of policy inaction to core executives (Andeweg et al., 2020; Dunleavy & Rhodes, 1990; Rhodes, 1995) who hold the de facto decision-making power and authority to implement lessons learned highlights specific reasons for their inaction. Imposed inaction occurs when powerful courts do not approve policy decisions made by core executives. Reluctant inaction arises from a lack of personnel resources in bureaucracies to draft legislation. Calculated inaction results from a strategic agreement among the members of the core executive not to act, often due to high implementation costs or competition between policy programs. These various forms of policy inaction result in non-decisions (Bachrach & Baratz, 1963), preventing learning-induced policy change.

61In the third pathway, beliefs, preferences, and outputs have changed, fulfilling all three features. However, the change in output is not congruent with the prior belief and preference changes. In other words, beliefs, preferences and outputs are not properly aligned, resulting in what this article denotes as “Non-Congruent Policy Change”. In technical terms, non-congruent policy change occurs when the behavioral learning product (policy) does not align with the cognitive learning products (beliefs and preferences). This discrepancy underscores the need for an effective transition from beliefs and preferences into policies, often also referred to as “translation” (Clarke et al., 2015; Freeman, 2008, 2009). This translation has two key dimensions: first, whether the adopted policy reflects the changed beliefs and preferences at all, and second, the extent to which the policy mirrors these new beliefs and preferences. This perspective emphasizes the “dynamic capacity” of policy learning, requiring lessons learnt to ‘travel’ across time and place (Stark & van der Arend, 2023). In this context, ‘time’ and ‘place’ refer to both the decision-making phase of the policy cycle (Lasswell, 1956) and the governmental actors authorized to implement the lessons learned (P. A. Hall, 1993). In cases of non-congruent policy change, the core executive does not remain inactive, as in the first and second pathways. Instead, it takes action to draft and adopt legislation on the learned policy issue. However, this action does not ensure that the changed beliefs and preferences are properly reflected in the final policy, resulting in non-congruent policy change.

62This pathway particularly highlights the political dimension of policy learning as discussed in various learning concepts, such as power-oriented learning (Trein & Vagionaki, 2021) and learning through bargaining (Dunlop & Radaelli, 2013). While learning through bargaining can provide valuable insights into the preferences and costs of cooperation, thereby facilitating learning-induced change, it can also hinder the effective application of lessons learned (Dunlop & Radaelli, 2018b). In competitive bargaining situations, the resource allocation of interest-driven actors plays an important role in the outcomes of these negotiations (Dunlop & Radaelli, 2018b, 2022). Unequal distribution of resources and power can prevent the transition from changed beliefs and preferences into an aligned policy output (Dunlop & Radaelli, 2018b). Sabatier and Weible (2007) also emphasize the importance of resources in negotiating policy change, identifying legal authority, financial resources, and persuasive ability as key factors.

63Finding studies that provide empirical examples for the third pathway can be challenging, as existing literature often does not differentiate between the various levels of learning products in sufficient detail or provides limited information to classify cases as leading to non-congruent change. However, some case studies highlight the closely related political dimension of learning, particularly in the context of bargaining. For example, Eberlein and Radaelli (2010) discuss various aggregation and transformation techniques in the context of bargaining within EU regulatory policy. Dunlop and Radaelli (2016) examine the dynamics of learning through bargaining as the dominant learning mode during the implementation of the European Semester fiscal surveillance system.

64The fourth pathway resembles the third as policy beliefs, preferences, and outputs have changed. The key difference is that the three features are aligned, resulting in what I term “Congruent Policy Change”. Here, the behavioral learning product (policy) is congruent with the cognitive learning products (beliefs and preferences).

65Plümer (2024) provides an example of the fourth pathway by analyzing policy learning and change in German school and education policy. The case study examines the debate on the amount of secondary schooling necessary to receive the German School Leaving Certificate. Drawing on a variety of data sources, including interviews and policy documents, the study outlines the causal chain from learning to change: First, policymakers changed their beliefs regarding the amount of schooling needed for the Certificate, shifting from a belief in a school system emphasizing fast educational processes to one giving more time for students to learn and develop their personalities. Second, policymakers aligned their policy preferences with these new beliefs, with the majority of policymakers favoring the extension of secondary schooling from eight to nine years. Third, members of the core executive engaged in policy action by drafting and adopting the thirteenth school amendment law, which extended the study period from eight to nine years, exemplifying congruent policy change.

Conclusion

66When does policy learning lead to policy change? Taking policy learning as an independent variable that affects policy change, this article clarified the causal chain from policy learning to policy change in two steps. First, it presented a “Learning Product Framework” that distinguishes three central features of learning products: policy belief change, policy preference change, and policy output change. Second, based on this framework, the article presented a “Typology of Causal Pathways between Learning and Change”, leading to four different learning-induced policy changes. Each pathway entails a different configuration of the learning product features.

67The conceptual clarification developed in this article confirms the previous finding that policy learning alone is not sufficient to lead to policy change. It demonstrates the specific combination of cognitive, behavioral, and social mechanisms necessary for learning-induced policy change: Policy actors first need to learn a lesson at the level of beliefs that aligns with received information (feature one; cognitive mechanism). This lesson must then be translated into policy preferences (feature two; cognitive mechanism). Finally, governmental policymakers must make policy decisions that reflect these preferences (feature three; behavioral and social mechanism). In this way, cognitive learning products lead to behavioral learning products, enabling learning-induced policy change and congruent policy change. Normatively, congruent policy change may be seen as ‘desirable’ policymaking because the outcomes of learning become visible in public policymaking and impact society as a whole (Berglund et al., 2022).

68The framework and typology developed in this article serve as useful heuristics in two ways. They facilitate the systematic analysis of different learning outcomes (policy beliefs, preferences, and outputs) that have not previously been distinguished in such detail. Additionally, they identify the levels at which learning products might be flawed and thus fail to ‘travel’. Here, the causal chain from learning to change is interrupted, severely limiting learning-induced policy change (Dunlop, 2017a, 2017b; Dunlop & Radaelli, 2018b).

69The theoretical approach of this article is both its strength and its weakness. On the one hand, it allows for a focus on clarifying the causal chain, using various dimensions of the learning-change relationship and the learning product perspective from existing literature. Further research can extend the theoretical argument, particularly by refining the transitions from policy belief change to policy preference change, and by drawing more extensively on the theory of reasoned action to investigate human decision-making (Fishbein & Ajzen, 2010; LaCaille, 2020). Furthermore, the crucial third feature policy output change, which enables congruent policy change could be analyzed in greater depth, with an emphasis on the contextual conditions that facilitate this translation. On the other hand, this article is not an empirical analysis. Future research can provide a ‘proof of concept’ by empirically testing both the framework and the typology. This is particularly challenging as policymakers themselves often struggle to distinguish between policy beliefs and policy preferences. Nonetheless, examining learning products and the interplay between cognitive, behavioral, and social mechanisms in diverse empirical settings could further enhance our understanding of the causal chain from policy learning to policy change.

Top of page

Bibliography

Andeweg, R. B., Elgie, R., Helms, L., Kaarbo, J., & Müller-Rommel, F. (2020). The Political Executive Returns: Re-Empowerment and Rediscovery. In R. B. Andeweg, R. Elgie, L. Helms, J. Kaarbo & F. Müller-Rommel (Eds.), The Oxford Handbook of Political Executives (pp. 1-25). Oxford University Press.

Argyris, C., & Schön, D. A. (1996). Organizational learning II: Theory, method and practice. Organization development series. Addison-Wesley.

Bach, T., & Wegrich, K. (2020). Politicians and Bureaucrats in Executive Government. In R. B. Andeweg, R. Elgie, L. Helms, J. Kaarbo & F. Müller-Rommel (Eds.), The Oxford Handbook of Political Executives (pp. 523-546). Oxford University Press.

Bachrach, P., & Baratz, M. S. (1963). Decisions and Nondecisions: An Analytical Framework. American Political Science Review, 57(3), 632-642.

Bandelow, N. C., Vogeler, C. S., Hornung, J., Kuhlmann, J., & Heidrich, S. (2019). Learning as a Necessary but Not Sufficient Condition for Major Health Policy Change: A Qualitative Comparative Analysis Combining ACF and MSF. Journal of Comparative Policy Analysis: Research and Practice, 21(2), 167-182.

Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (2001). Bad is stronger than good. Review of General Psychology, 5(4), 323-370.

Baumgartner, F. R., & Jones, B. D. (2015). The politics of information: Problem definition and the course of public policy in America. University of Chicago Press.

Beach, D., Schäfer, D., & Smeets, S. (2021). The Past in the Present—The Role of Analogical Reasoning in Epistemic Learning About How to Tackle Complex Policy Problems. Policy Studies Journal, 49(2), 457-483.

Béland, D. (2019). How Ideas and Institutions Shape the Politics of Public Policy. Cambridge University Press.

Béland, D., & Cox, R. H. (2010). Ideas and Politics in Social Science Research. Oxford University Press, USA.

Bennett, C. J., & Howlett, M. (1992). The lessons of learning: Reconciling theories of policy learning and policy change. Policy Sciences, 25(3), 275-294.

Berglund, O., Dunlop, C. A., Koebele, E. A., & Weible, C. M. (2022). Transformational change through Public Policy. Policy & Politics, 50(3), 302-322.

Biegelbauer, P. (2016). How different forms of policy learning influence each other: case studies from Austrian innovation policy-making. Policy Studies, 37(2), 129-146.

Bolognesi, T., Lieberherr, E., & Fischer, M. (2024). Identifying and explaining policy preferences in Swiss water management. Policy & Politics, 1-28.

Brown, P. R., & Stark, A. (2022). Policy inaction meets policy learning: four moments of non-implementation. Policy Sciences, 55(1), 47-63.

Bullock, J. G. (2009). Partisan Bias and the Bayesian Ideal in the Study of Public Opinion. The Journal of Politics, 71(3), 1109-1124.

Busetti, S., & Righettini, M. S. (2023). Policy learning from crises: lessons learned from the Italian food stamp programme. Policy & Politics, 51(1), 91-112.

Cairney, P. (2020). Understanding public policy: Theories and issues (2nd edition). Textbooks in policy studies. Red Globe Press.

Cairney, P., & Weible, C. M. (2017). The new policy sciences: combining the cognitive science of choice, multiple theories of context, and basic and applied analysis. Policy Sciences, 50(4), 619-627.

Clarke, J., Bainton, D., Lendvai, N., & Stubbs, P. (2015). Making policy move: Towards a politics of translation and assemblage. Policy Press.

Coleman, J. S. (1986). Social Theory, Social Research, and a Theory of Action. American Journal of Sociology, 91(6), 1309-1335.

Cook, S. D. N., & Yanow, D. (1993). Culture and Organizational Learning. Journal of Management Inquiry, 2(4), 373-390.

Crow, D. A., & Albright, E. A. (2019). Intergovernmental relationships after disaster: state and local government learning during flood recovery in Colorado. Journal of Environmental Policy & Planning, 21(3), 257-274.

Crow, D. A., Albright, E. A., Ely, T., Koebele, E. A., & Lawhon, L. (2018). Do Disasters Lead to Learning? Financial Policy Change in Local Government. Review of Policy Research, 35(4), 564-589.

Crow, D. A., DeLeo, R. A., Albright, E. A., Taylor, K., Birkland, T., Zhang, M., Koebele, E. A., Jeschke, N., Shanahan, E. A., & Cage, C. (2023). Policy learning and change during crisis: Covid-19 policy responses across six states. Review of Policy Research, 40(1), 10-35.

Domjan, M. (2010). The principles of learning and behavior (6th ed.). Wadsworth/Cenage.

Dunleavy, P., & Rhodes, R. A. W. (1990). CORE EXECUTIVE STUDIES IN BRITAIN. Public Administration, 68(1), 3-28.

Dunlop, C. A. (2017a). Pathologies of policy learning: what are they and how do they contribute to policy failure? Policy & Politics, 45(1), 19-37.

Dunlop, C. A. (2017b). Policy learning and policy failure: definitions, dimensions and intersections. Policy & Politics, 45(1), 3-18.

Dunlop, C. A., & Radaelli, C. M. (2013). Systematising Policy Learning: From Monolith to Dimensions. Political Studies, 61(3), 599-619.

Dunlop, C. A., & Radaelli, C. M. (2016). Policy learning in the Eurozone crisis: Modes, power and functionality. Policy Sciences, 49, 107-124.

Dunlop, C. A., & Radaelli, C. M. (2017). Learning in the bath-tub: The micro and macro dimensions of the causal relationship between learning and policy change. Policy and Society, 36(2), 304-319.

Dunlop, C. A., & Radaelli, C. M. (2018a). Does Policy Learning Meet the Standards of an Analytical Framework of the Policy Process? Policy Studies Journal, 46(S1), 48-68.

Dunlop, C. A., & Radaelli, C. M. (2018b). The lessons of policy learning: Types, triggers, hindrances and pathologies. Policy & Politics, 46(2), 255-272.

Dunlop, C. A., & Radaelli, C. M. (2020). Policy Learning in Comparative Policy Analysis. Journal of Comparative Policy Analysis: Research and Practice.

Dunlop, C. A., & Radaelli, C. M. (2022). Policy Learning in Comparative Policy Analysis. Journal of Comparative Policy Analysis: Research and Practice, 24(1), 51-72.

Eberlein, B., & Radaelli, C. M. (2010). MECHANISMS OF CONFLICT MANAGEMENT IN EU REGULATORY POLICY. Public Administration, 88(3), 782-799.

Elgie, R. (2011). CORE EXECUTIVE STUDIES TWO DECADES ON. Public Administration, 89(1), 64-77.

Festinger, L. (1975). A theory of cognitive dissonance. Stanford University Press.

Fishbein, M., & Ajzen, I. (2010). Predicting and changing behavior: The reasoned action approach. Psychology press.

Freeman, R. (2008). Learning in Public Policy. In R. E. Goodin, M. Moran & M. Rein (Eds.), The Oxford handbook of public policy. Oxford University Press.

Freeman, R. (2009). What is ‘translation’? Evidence & Policy, 5(4), 429-447.

Gawronski, B., & Strack, F. (Eds.). (2012). Cognitive consistency: A fundamental principle in social cognition. The Guilford Press.

Gerlak, A. K., & Heikkila, T. (2011). Building a Theory of Learning in Collaboratives: Evidence from the Everglades Restoration Program. Journal of Public Administration Research and Theory, 21(4), 619-644.

Gerlak, A. K., Heikkila, T., & Newig, J. (2020). Learning in environmental governance: Opportunities for translating theory to practice. Journal of Environmental Policy & Planning, 22(5), 653-666.

Goldstein, J., & Keohane, R. O. (Eds.). (1993). Ideas and Foreign Policy: Beliefs, Institutions, and Political Change. Cornell University Press.

Goyal, N. (2022). Policy Diffusion Through Multiple Streams: The (Non‐)Adoption of Energy Conservation Building Code in India. Policy Studies Journal, 50(3), 641-669.

Goyal, N., & Howlett, M. (2019). Framework or metaphor? Analysing the status of policy learning in the policy sciences. Journal of Asian Public Policy, 12(3), 257-273.

Goyal, N., & Howlett, M. (2024). Types of learning and varieties of innovation: how does policy learning enable policy innovation? Policy & Politics, 1-22.

Hall, G. (2003). The psychology of learning. In L. Nadel (Ed.), Encyclopedia of cognitive science (2nd ed., pp. 837-845). Nature Publishing Group.

Hall, P. A. (1993). Policy Paradigms, Social Learning, and the State: The Case of Economic Policymaking in Britain. Comparative Politics, 25(3), 275.

Hart, W., Albarracín, D., Eagly, A. H., Brechan, I., Lindberg, M. J., & Merrill, L. (2009). Feeling validated versus being correct: A meta-analysis of selective exposure to information. Psychological Bulletin, 135(4), 555-588.

Hatch, M. E., & Mead, J. W. (2021). Learning from laboratory mistakes: How policy entrepreneurs catalyze city ordinance repeals in the United States. Public Policy and Administration, 36(3), 361-378.

Heikkila, T., & Gerlak, A. K. (2013). Building a Conceptual Approach to Collective Learning: Lessons for Public Policy Scholars. Policy Studies Journal, 41(3), 484-512.

Heikkila, T., Gerlak, A. K., & Smith, B. (2023). Diagnosing individual barriers to collective learning: how governance contexts shape cognitive biases. Journal of European Public Policy, 1-24.

Herweg, N., Zahariadis, N., & Zohlnhöfer, R. (2023). The Multiple Streams Framework: Foundations, Refinements, and Empirical Applications. In C. M. Weible (Ed.), Theories of the policy process (5th ed.). Routledge.

Jenkins-Smith, H. C., Nohrstedt, D., Weible, C. M., & Ingold, K. (2018). Advocacy Coalition Framework: An Overview of the Research Program. In C. M. Weible & P. A. Sabatier (Eds.), Theories of the policy process (4th ed., pp. 135-171). Westview Press.

Kahan, D. M. (2013). Ideology, motivated reasoning, and cognitive reflection. Judgment and Decision Making, 8(4), 407-424.

Kamkhaji, J. C. (2022). Policy Learning and the Euro. Springer.

Kamkhaji, J. C., & Radaelli, C. M. (2017). Crisis, learning and policy change in the European Union. Journal of European Public Policy, 24(5), 714-734.

Kamkhaji, J. C., & Radaelli, C. M. (2022). Don’t think it’s a good idea! Four building sites of the ‘ideas school’. West European Politics, 45(4), 841-862.

Koebele, E. A. (2019). Policy learning in collaborative environmental governance processes. Journal of Environmental Policy & Planning, 21(3), 242-256.

Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108(3), 480-498.

Lablih, M., Bundi, P., & Portmann, L. (2024). Does anxiety increase policy learning? Policy Studies Journal, Article psj.12529. Advance online publication.

LaCaille, L. (2020). Theory of reasoned action. In M. D. Gellman (Ed.), Encyclopedia of Behavioral Medicine (2nd ed. 2020, pp. 2231-2234). Springer International Publishing; Imprint: Springer.

Lasswell, H. (1956). The Decision Process: Seven Categories of Functional Analysis. University of Maryland Press.

Leach, W. D., Weible, C. M., Vince, S. R., Siddiki, S. N., & Calanni, J. C. (2014). Fostering Learning through Collaboration: Knowledge Acquisition and Belief Change in Marine Aquaculture Partnerships. Journal of Public Administration Research and Theory, 24(3), 591-622.

Lee, S., Hwang, C., & Moon, M. J. (2020). Policy learning and crisis policy-making: Quadruple-loop learning and COVID-19 responses in South Korea. Policy and Society, 39(3), 363-381.

Leifeld, P. (2013). Reconceptualizing Major Policy Change in the Advocacy Coalition Framework: A Discourse Network Analysis of German Pension Politics. Policy Studies Journal, 41(1), 169-198.

Levitt, B., & March, J. G. (1988). Organizational learning. Annual Review of Sociology, 14, 319-340.

Lindberg, M. B., Markard, J., & Andersen, A. D. (2019). Policies, actors and sustainability transition pathways: A study of the EU’s energy policy mix. Research Policy, 48(10), 103668.

May, P. J. (1992). Policy Learning and Failure. Journal of Public Policy, 12(4), 331-354.

McConnell, A., & ’t Hart, P. (2019). Inaction and public policy: understanding why policymakers ‘do nothing’. Policy Sciences, 52(4), 645-661.

Möck, M., & Feindt, P. H. (2023). Learning mode misfits in policy learning: typology, case study and lessons learnt. Journal of European Public Policy, 1-26.

Montpetit, É., & Lachapelle, E. (2017). Policy learning, motivated scepticism, and the politics of shale gas development in British Columbia and Quebec. Policy and Society, 36(2), 195-214.

Moyson, S. (2017). Cognition and policy change: The consistency of policy learning in the advocacy coalition framework. Policy and Society, 36(2), 320-344.

Moyson, S., Scholten, P., & Weible, C. M. (2017). Policy learning and policy change: Theorizing their relations from different perspectives. Policy and Society, 36(2), 161-177.

Munro, G. D., & Ditto, P. H. (1997). Biased Assimilation, Attitude Polarization, and Affect in Reactions to Stereotype-Relevant Scientific Information. Personality and Social Psychology Bulletin, 23(6), 636-653.

Nickerson, R. S. (1998). Confirmation Bias: A Ubiquitous Phenomenon in Many Guises. Review of General Psychology, 2(2), 175-220.

Nohrstedt, D. (2011). Shifting Resources and Venues Producing Policy Change in Contested Subsystems: A Case Study of Swedish Signals Intelligence Policy. Policy Studies Journal, 39(3), 461-484.

Nohrstedt, D., Ingold, K., Weible, C. M., Koebele, E. A., Olofsson, K. L., Satoh, K., & Jenkins-Smith, H. C. (2023). The Advocacy Coalition Framework: Progress and Emerging Areas. In C. M. Weible (Ed.), Theories of the policy process (5th edition). Routledge.

Nowlin, M. C. (2021). Policy Learning and Information Processing. Policy Studies Journal, 49(4), 1019-1039.

Nowlin, M. C. (2024). Policy beliefs, belief uncertainty, and policy learning through the lens of the Advocacy Coalition Framework. Policy & Politics, 1-21.

Ormrod, J. E. (2008). Human learning (5th ed.). Pearson.

Örtenblad, A. (2002). Organizational Learning: A Radical Perspective. International Journal of Management Reviews, 4(1), 71-85.

Osei‐Kojo, A., Kenney, P. L., Damoah, C. M., & Ahenkan, A. (2022). Collective learning and COVID‐19 mitigation in Ghana. Review of Policy Research, 39(3), 255-281.

Page, E. (2012). Policy without politicians: Bureaucratic influence in comparative perspective. Oxford University Press.

Pattison, A. (2018). Factors Shaping Policy Learning: A Study of Policy Actors in Subnational Climate and Energy Issues. Review of Policy Research, 35(4), 535-563.

Pierce, J. J., Peterson, H. L., & Hicks, K. C. (2020). Policy Change: An Advocacy Coalition Framework Perspective. Policy Studies Journal, 48(1), 64-86.

Pierre, J., & Peters, B. G. (2017). The shirking bureaucrat: a theory in search of evidence? Policy & Politics, 45(2), 157-172.

Plümer, S. (2024). Kongruentes Policy-Lernen als lernbedingter Policy-Wandel. Springer Fachmedien Wiesbaden.

Radaelli, C. M. (2009). Measuring policy learning: regulatory impact assessment in Europe. Journal of European Public Policy, 16(8), 1145-1164.

Rhodes, R. A. W. (1995). Introducing the Core Executive. In R. A. W. Rhodes & P. Dunleavy (Eds.), Prime minister, cabinet and core executive (pp. 1-8). St. Martin's Press.

Riche, C., Aubin, D., & Moyson, S. (2021). Too much of a good thing? A systematic review about the conditions of learning in governance networks. European Policy Analysis, 7(1), 147-164.

Rietig, K. (2018). Learning in the European Commission’s Renewable Energy Policy-Making and Climate Governance. In C. A. Dunlop, C. M. Radaelli & P. Trein (Eds.), Learning in public policy: Analysis, modes and outcomes (pp. 51-74). Palgrave Macmillan.

Ritter, A., Hughes, C. E., Lancaster, K., & Hoppe, R. (2018). Using the Advocacy Coalition Framework and Multiple Streams policy theories to examine the role of evidence, research and other types of knowledge in drug policy. Addiction (Abingdon, England), 113(8), 1539-1547.

Rose, R. (1991). What is Lesson-Drawing? Journal of Public Policy, 11(1), 3-30.

Sabatier, P. A. (1988). An advocacy coalition framework of policy change and the role of policy-oriented learning therein. Policy Sciences, 21(2-3), 129-168.

Sabatier, P. A. (1993). Policy Change over a Decade or More. In P. A. Sabatier & H. C. Jenkins-Smith (Eds.), Policy Change and Learning: An Advocacy Coalition Approach (pp. 13-39). Westview Press.

Sabatier, P. A., & Jenkins-Smith, H. C. (1999). The Advocacy Coalition Framework: An Assessment. In P. A. Sabatier (Ed.), Theories of the Policy Process (1. Auflage, pp. 117-166). Westview Press.

Sabatier, P. A., & Weible, C. M. (2007). The Advoacy Coalition Framework: Innovations and Clarifications. In P. A. Sabatier & C. M. Weible (Eds.), Theories of the Policy Process (2. Auflage, pp. 189-222). Westview Press.

Schofield, J. (2004). A Model of Learned Implementation. Public Administration, 82(2), 283-308.

Simon, H. A. (1991). Bounded Rationality and Organizational Learning. Organization Science, 2(1), 125-134.

Squevin, P., Aubin, D., Montpetit, É., & Moyson, S. (2021). Closer than they look at first glance: A systematic review and a research agenda regarding measurement practices for policy learning. International Review of Public Policy, 3(2), 146-171.

Stark, A., & Head, B. (2019). Institutional amnesia and public policy. Journal of European Public Policy, 26(10), 1521-1539.

Stark, A., & van der Arend, J. (2023). The traveller's guide to policy learning. Journal of European Public Policy, 1-22.

Sunstein, C. R. (2006). The Availability Heuristic, Intuitive Cost-Benefit Analysis, and Climate Change. Climatic Change, 77(1-2), 195-210.

't Hart, P., & Wille, A. (2006). Ministers and Top Officials in the Dutch Core Executive: Living Together, Growing Apart? Public Administration, 84(1), 121-146.

Thunus, S., & Schoenaers, F. (2017). How does policy learning occur? The case of Belgian mental health care reforms. Policy and Society, 36(2), 270-287.

Trein, P., & Vagionaki, T. (2021). Learning heuristics, issue salience and polarization in the policy process. West European Politics, 1-21.

Vagionaki, T. (2018). Blocked Learning in Greece: The Case of Soft-Governance. In C. A. Dunlop, C. M. Radaelli & P. Trein (Eds.), Learning in public policy: Analysis, modes and outcomes (pp. 191-214). Palgrave Macmillan.

Vagionaki, T., & Trein, P. (2019). Learning in Political Analysis. Political Studies Review, 18(2), 304-319.

VandenBos, G. R. (2007). APA Dictionary of Psychology. American Psychological Association.

Weible, C. M., Olofsson, K. L., & Heikkila, T. (2023). Advocacy coalitions, beliefs, and learning: An analysis of stability, change, and reinforcement. Policy Studies Journal, 51(1), 209-229.

Zaki, B. L. (2024). Policy learning governance: a new perspective on agency across policy learning theories. Policy & Politics, 1-18.

Zaki, B. L., Pattyn, V., & Wayenberg, E. (2022). Policy learning type shifts during creeping crises: A storyboard of COVID‐19 driven learning in Belgium. European Policy Analysis, Article epa2.1165. Advance online publication.

Zaki, B. L., & Wayenberg, E. (2020). Shopping in the scientific marketplace: COVID-19 through a policy learning lens. Policy Design and Practice, 1-18.

Zaki, B. L., & Wayenberg, E. (2023). Policy learning and the COVID‐19 crisis: A systematic review of scholarship and key lessons for research and practice. Australian Journal of Public Administration, Article 1467-8500.12598. Advance online publication.

Zaki, B. L., Wayenberg, E., & George, B. (2022). A Systematic Review of Policy Learning: Tiptoeing through a Conceptual Minefield. Policy Studies Yearbook, 12(1).

Zollo, M., & Winter, S. G. (2002). Deliberate Learning and the Evolution of Dynamic Capabilities. Organization Science, 13(3), 339-351.

Top of page

Notes

1 This is also visible in some articles’ titles, framing learning as change, e.g. “Do Disasters Lead to Learning? Financial Policy Change in Local Government” (own emphasis), by Crow et al. (2018).

2 On the other hand, studying learning as a process only implicitly addresses policy change. These studies focus on the internal workings of learning processes rather than explicitly investigating their effect on legislation (e.g., Thunus & Schoenaers, 2017).

3 The article refrains from a repetition of the biases.

4 Despite this widely shared assumption, the policy learning literature also describes instances where change precedes and triggers policy learning, reversing the commonly accepted causal mechanism, see Kamkhaji (2022), Kamkhaji & Radaelli (2017).

Top of page

List of illustrations

Title Table 1: Causal pathways between learning and change, including different learning-induced policy changes
Credits Source: Own compilation. A hook (✓) indicates the feature is fulfilled, a cross (⨯) indicates the feature is not met.
URL http://journals.openedition.org/irpp/docannexe/image/4798/img-1.png
File image/png, 46k
Top of page

References

Bibliographical reference

Sandra Plümer, When does policy learning lead to policy change? Exploring the causal chain from learning to changeInternational Review of Public Policy, 6:2 | 2024, 200-219.

Electronic reference

Sandra Plümer, When does policy learning lead to policy change? Exploring the causal chain from learning to changeInternational Review of Public Policy [Online], 6:2 | 2024, Online since 11 September 2024, connection on 13 January 2026. URL: http://journals.openedition.org/irpp/4798; DOI: https://doi.org/10.4000/138rk

Top of page

About the author

Sandra Plümer

University of Bielefeld
sandra.pluemer@uni-bielefeld.de

Top of page

Copyright

CC-BY-4.0

The text only may be used under licence CC BY 4.0. All other elements (illustrations, imported files) may be subject to specific use terms.

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search