Revitalizing double‐loop learning in organizational contexts: A systematic review and research agenda

• Published date: 01 December 2023
• Author: Mercedes‐Victoria Auqui‐Caceres,Andrea Furlan
• Date: 01 December 2023
• DOI: http://doi.org/10.1111/emre.12615

REVIEW ARTICLE

Revitalizing double-loop learning in organizational contexts: A

systematic review and research agenda

Mercedes-Victoria Auqui-Caceres | Andrea Furlan

Department of Economics and Management,

University of Padova, Italy

Correspondence

Mercedes-Victoria Auqui-Caceres, Department

of Economics and Management, University of

Padova, Italy.

Email: mercedesvictoria.auquicaceres@unipd.it

[Correction added on 18 October 2023, after

first online publication: Mercedes-Victoria’s

surname has been corrected from ‘Caceres

Auqui’to ‘Auqui-Caceres’]

Abstract

Argyris & Schön’s notion of two types of learning, single-loop (SLL) and double-

loop learning (DLL), is arguably one of the most popularized categorizations of

organizational learning (OL). However, while the concept of DLL is widely cited,

it has left a superficial impact on the literature and practice. We argue that the

limited impact of DLL is due to two features of DLL: the complexity of its defini-

tion and the difficulty in its implementation. This study identifies and organizes

critical insights in the literature related to the conceptualization, measurement,

and generation of DLL. To address these topics, we review and synthesize the

findings of 128 studies on DLL published between 1974 and 2021. We aim to

reduce the confusion surrounding DLL and the proliferation of empirical studies

on DLL that ignore its original notion. We propose a framework that makes

explicit the misconceptions, wrong assumptions, and barriers in conceptualizing,

measuring, and generating DLL, and it also provides insights into how to over-

come these limitations and serves as a platform for future research on DLL.

KEYWORDS

double-loop learning, organizational learning, systematic literature review

INTRODUCTION

Scholars generally agree that one of the most important

typologies of organizational learning (OL) distinguishes

between single-loop learning (SLL) and double-loop

learning (DLL). This distinction was initially made by

Argyris & Schön (1974,1978). SLL occurs when organi-

zational members attempt to correct the mismatches

between intentions and outcomes simply by changing

their actions without questioning or altering the govern-

ing values (values, belief, and assumptions) underlying

those actions, whereas DLL occurs when “mismatches

are corrected by first examining and altering the govern-

ing variables and then the actions”(Argyris, 1999, p. 68).

Argyris & Schön’s(

1974,1978) ultimate aim was to

help organizations to develop DLL capabilities. Their

effort is aimed at transforming an organization into a

“DLL organization”. Although the concept of DLL was

introduced in 1974, it has been cited by more recent

theories on OL. For example, both Pedler et al.’s(

1989)

model on the “learning company”and Senge’s(

1990)

model on the “learning organization”refer to the concept

of DLL. Pedler et al. (1989, p. 97) recognize that the type

of learning behind a learning company (i.e., an organiza-

tion able to innovate) should be of “double-loop nature”.

Similarly, Senge (1990) argues that to become a learning

organization (i.e., an organization able to survive in a hec-

tic environment that requires rapid changes), “adaptive

learning must be joined by ‘generative learning’, learning

that enhances our capacity to create”(Senge, 1990,p.14).

Generative learning involves changing mental models

defined as “deeply ingrained assumptions or generaliza-

tions that influence how we understand the world and how

we take action”(Senge, 1990, p. 11), which is one of the

fundamental notionsof DLL.

However, as Lipshitz (2000, p. 468) argues there is

“an evident gap between the frequency and the profound-

ness of references to Argyris & Schön’s work in the

DOI: 10.1111/emre.12615

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any

medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

© 2023 The Authors. European Management Review published by John Wiley & Sons Ltd on behalf of European Academy of Management (EURAM).

European Management Review. 2023;20:741–761. wileyonlinelibrary.com/journal/emre 741

literature”. Indeed, given the widespread attention that

DLL has received in management literature, one would

expect that DLL would have much to contribute to the

body of research on OL, but DLL left a noticeable but

superficial impact on the literature and practice

(Lipshitz, 2000; Witherspoon, 2014).

The superficial impact of DLL is even more surpris-

ing given the recent rise of theory-based learning in man-

agement literature (Ehrig & Schmidt, 2022; Felin &

Zenger, 2009). Scholars advocating this approach suggest

that strategists should design experiments to test their

hypothesis, learn from these experiments and possibly

revise their beliefs about successful courses of action in

the light of the unfolding evidence. Even if this idea of

learning is clearly in line with the theory on DLL, theory-

based learning scholars do not make use of any concept

from Argyris & Schön’s framework.

We suspect that the limited impact of DLL can be

traced back to two features of the original conceptualiza-

tion of DLL: the complexity of its definition (Jaaron &

Backhouse, 2017; Lipshitz, 2000; Mazutis &

Slawinski, 2008;O’connor & Kotze, 2008) and the diffi-

culty in its implementation (Bochman & Kroth, 2010;

Choularton, 2001; Henderson, 1997; Lipshitz, 2000;

Mazutis & Slawinski, 2008; McAvoy & Butler, 2007;

Tsuchiya, 1998; Wong, 2005). The purpose of this study

is to identify whether misconceptions or wrong assump-

tions exist in the management literature when it comes to

define or implement DLL. We attempt to address this

issue by conducting a systematic review of the literature

on DLL, focusing on (a) verifying the consistency of the

studies with the original conceptualization of DLL and

(b) finding the main difficulties in operationalizing DLL.

A thematic analysis suggests that the literature can be

organized along three key categories (a) how to conceptu-

alize DLL, that is, studies that mainly provide theoretical

contributions to the conceptualization of DLL, (b) how

to measure DLL, that is, studies that predominantly

employ methods or tools for the measurement of DLL,

and (c) how to generate DLL, that is, studies that primar-

ily provide models, methodologies, or mechanisms

focused on overcoming the difficulties and obstacles to

produce DLL. From a critical analysis of the literature,

we provide a framework with key insights that seek to

revitalize the conceptualization, measurement, and imple-

mentation of DLL. We argue that clarifying the concept

and operationalization of DLL can be helpful to a variety

of different streams of research on problem solving, OL,

and organizational innovation.

We begin this article by providing the research back-

ground on DLL to clarify its nature and original notions.

We then discuss the motivation for our literature review

and present the method and findings of our review and

finally we present our conclusions and future research

avenues.

THEORETICAL FRAMEWORK

Organizational learning

Learning can be defined in many ways. For the purpose

of this study, learning occurs when new understanding or

insights are connected with new behaviors or actions

(Argyris & Schön, 1996). Argyris & Schön have referred

to learning as new insights or knowledge (Fiol &

Lyles, 1985) that challenges the assumptions about what

is known or done (Di Bella & Nevis, 1998).

Although there have been many reviews of the OL lit-

erature, there does not appear to be a widely accepted

definition of OL (Bontis et al., 2002; Fiol & Lyles, 1985).

Two of the most used definition of OL were provided by

Argyris & Schön (1978) and Fiol & Lyles (1985).

Argyris & Schön defined OL as the process of detecting

and correcting errors. Fiol & Lyles defined OL as the

process of improving actions through better knowledge

and understanding. However, it appears to be some

agreement on the need for a distinction between individ-

ual learning and OL. As Bontis et al. (2002, p. 444) claim

“organization level learning involves embedding individ-

ual and group learning into the non-human aspects of the

organization including systems, structures”. Indeed, orga-

nizations learn when they “encode inferences from his-

tory into routines that guide behavior”(Levitt & March,

1998, p. 319). Organizations do not have brains, but they

have cognitive systems and memories to store the learn-

ing they create (Hedberg, 1981). For Huber (1991, p. 89)

an organization learns “if any of its units acquires knowl-

edge that it recognizes as potentially useful to the

organization”.

Although literature offers diverse theoretical OL

perspectives (Bontis et al., 2002), the theoretical founda-

tions of the current study are based on the “theory of

action”developed by Argyris & Schön (1974,1978).

We focus on this theory of action because it is one of

the most cited theories in the OL field (Fulmer & Keys,

1998; Lichtenstein, 2000; Remedios & Boreham, 2004;

Bochman & Kroth, 2010). A comparison of different

theoretical approaches in the OL domain is beyond the

scope of this paper.

Model I and model II theory-in-use

Argyris & Schön (1974,1978) state that all human action

is based on theories of action. Individuals carry around

their governing variables about how they and others

should behave. According to Argyris & Schön, these gov-

erning variables can be stated in the form of propositions

or causal representations, that is, “if I behave in such and

such a manner, then the following consequences should

occur”(Argyris, 1999, p.179). Since these propositions

742 AUQUI-CACERES and FURLAN