N AND P

Transactions of the SDPS: DE

SIG RO

Journal of Integrated Design and Process Science

CE
OR

SS
SOCIETY F

SCIENCE
21 (1), 2017, 43-60 &
DOI 10.3233/jid-2017-0006
http://www.sdpsnet.org S D P S

How Organizational Structure Affects Organizational

Learning
S. Koohborfardhaghighia* J. Altmann b
a
University of Amsterdam, Plantage Muidergracht 12, 1018 TV Amsterdam, The Netherlands
b
Technology Management, Economics, and Policy Program, College of Engineering, Seoul National
University, Gwanak-ro, Gwanak-gu, Seoul 08826 South Korea

Abstract Nowadays, the knowledge has become a principal source of competitive advantage similar to other
valuable resource and strategic assets in an organization. However, without proper management of such a valuable
resource, utilization of knowledge becomes difficult and inefficient. From an organizational perspective, the most
important aspect of knowledge management is to transfer knowledge. In this regards, organizations must adopt
structures that allow them to create and transfer more knowledge. Organizational structure affects the nature of
human interactions and knowledge flow which in its own turn can lead to a competitive advantage in the knowledge
economy. In this article we emphasize the role of an organizational structure as a mechanism for knowledge flow
and organizational learning. Hence we test organizational learning performance under different network structures
which are mainly traditional hierarchical approach, hybrid structures and a fully connected network. Our
experimental results indicate the fact that traditional hierarchical approach has the lowest performance in a learning
organization while a fully connected network provides the best performance. The findings reflect the fact that
organizations need a more flexible hybrid structure to allow mobility and lateral connection between the individuals,
a structure which is between the two extremes, one of traditional hierarchy and the other of full connectivity pattern.

Keywords: Knowledge Management Process, Exploration and Exploitation, Organizational Structure,

Organizational Learning, Hybrid Network Structure

1. Introduction
Researchers and scientists try to understand effective ways of the knowledge management and
knowledge transfer in order to use knowledge as a competitive advantage. Knowledge management in an
organization refers to the process of creating value from the organization's intangible assets (Liebowitz
1999). In other words, it leverages the collective wisdom of the employees by soliciting ideas and
solutions for an organization. As a definition we can say that the knowledge management is the set of
processes that govern the creation, storage, dissemination and use of knowledge.
Knowledge creation involves developing new contents in the implicit and explicit knowledge of the
organization (Pentland 1995). New knowledge can be considered as an essential raw material for
innovation and knowledge creation, both of which are leading to new products and services.
In today's rapidly changing environment, organizations are forced to create knowledge in order to
avoid the obsolescence of existing knowledge and be successful in the innovation process (Sørensen and

* Corresponding author. Email: s.koohborfardhaghighi@uva.nl. Tel: (+32)068-7099831.

1092-0617/$27.50© 201X - Society for Design and Process Science. All rights reserved. Published by IOS Press
44 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning

Stuart 2000). While new knowledge is developed by individuals and institutional network of people that
will be there, organizations play the key role in the management and flow of such new knowledge.
Organizational structures simplify decision-making by depicting a hierarchy of authority to set the
goals and direction of the organization. Managers are tasked with applying these goals to different
departments under their control. Appropriate knowledge moves between individuals in an organization,
organizational departments and an organization and its environments (Barwise and Seligman 1997,
Henczel 2001, Bremer and Cohnitz 2004, Da Silva and Agusti-Cullell 2008). Knowledge only flows
between departments that are connected or related by a set of structural and behavioral rules.
Personal knowledge is not an organizational knowledge; organizational knowledge must be captured
and shared by members of a system. Therefore, in order to become an organization with extensive
knowledge, systems and mechanisms should be used to store the knowledge for the future use (Bajaria
2000). Those steps in knowledge management cycle are called knowledge creation and storage (Alavi and
Leidner 2001).
Distribution and transfer of knowledge is also another important part in the knowledge management
process. Obtained knowledge within the organization should be generalized and available to the others
through social interactions. The goal of knowledge use is to activate the relevant knowledge in order to
create value for the organization, because the organizational knowledge should be embedded and
employed in new products, services and processes. The knowledge itself is not a source of competitive
advantage for the organization, rather organizational power lies in the process how it is used. In
knowledge management cycle we call such processes knowledge transfer and knowledge usage (Alavi
and Leidner 2001).
The power of using knowledge against the competitors is a key success factor at the global age of
information (Yew Wong 2005, Crane, Downes et al. 2008). Employees improve the chances of achieving
organizational goals by utilization of all available resources as well as knowledge. They learn from the
experiences of others who have been more effective in similar or different areas, improve the quality and
speed of problem solving. When an organization becomes a learning organization (Pedler and Aspinwall
1999), knowledge application is necessitated to help organizations to retain correct and valuable
knowledge. From an organizational perspective, the most important aspect of knowledge management is
to transfer knowledge. Therefore, organizations must adopt structures that allow them to create and
transfer more knowledge (Koohborfardhaghighi, Lee et al. 2016).
Organizational structures open up lines of communication and collaboration among its members
differently. In a traditional hierarchy, for example, we are facing different layers of connectivity with
some key individuals at the top with all the information and power. On the other hand, a flat
organizational structure has a very loose communication structure by removing the intermediate layers.
Each organizational structure can cause challenges for communication and collaboration among its
employees.
According to the literature, managers can consider and utilize the organizational structure as a
mechanism for balancing the exploitation (i.e., use of known solutions) and exploration (i.e., the search
for new solutions) (Lenox 2002, Nickerson and Zenger 2002, Argote, McEvily et al. 2003, Benner and
Tushman 2003, Siggelkow and Levinthal 2003). Each organizational structure defines the pattern of
complex relationships within the group differently. It divides the organization into different subgroups
where within each of them a local optimal solution (i.e., with respect to the organizational goal) may
expect to emerge. This way each organization can expect a diverse solution space for the problem at hand.
Furthermore, in order to avoid being trapped in suboptimal solutions achieved by each subgroup,
connections among subgroups need to be added. The existence of such connections among subgroups let
the organization explores more diverse solutions in the space of possibilities, subsequently, achieving
better long term learning performance outcome (Fang, Lee et al. 2010). In this paper, learning
performance is equivalent to achieving the correct knowledge needed at the organizational level.
According to March and Seng, organizational learning is a property that emerges out of the
interactions among the employees within an organization (March 1991, Senge, Kleiner et al. 1999).
Therefore, it can be considered as interpersonal learning among group members and within groups.
 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning 45

Consequently, a learning organization is where employees continually expand their learning capacity to
capture and discover the whole solution together (Senge 2006). Organizational learning is not about
finding out what others already know, it is about solving problems by reflecting, connecting, and testing
until a solution forms part of organizational life (Serrat 2009).
Similarly, in this article we argue that organizational structure affects the nature of human interactions
and knowledge flow which in its own turn can lead to a competitive advantage in the knowledge economy.
We investigate and emphasize the role of an organizational structure as a mechanism for knowledge flow
and organizational learning. In this regard, as an attempt to follow up Fang’s experiments (Fang, Lee et al.
2010), we test organizational learning performance under different network structures which are mainly
fully connected networks, traditional hierarchical networks and hybrid network structures.
Fang in her experiments assumed a full connectivity pattern among each subgroup members (Fang,
Lee et al. 2010). We argue that this assumption is unrealistic due to the fact that in an organization the
activities of the individuals must be coordinated and there must be authorities for controlling, making
decisions and taking responsibilities (hence, we also test a hierarchical network structure). On the other
hand, a pure traditional hierarchical scheme does not show mobility and lateral connection of individuals
in its own turn. So we need a more flexible hybrid structure to allow lateral connection between the
individuals, a structure which is between the two extremes, one of hierarchy and the other of full
connectivity pattern.
The specific purpose of this study is to identify the impact of a more flexible hybrid structure among
individuals on the learning performance outcome within an organization.
To answer our research question, we run a series of simulations to compare the learning dynamics of
organizations that have different subgroup structure. We use the model of March (March 1991) which is
described in chapter 3 of this paper.
The rest of this paper is organized as follows. In section 2, we discuss related works and theoretical
background on the topic. In section 3, we detail the model and its parameters. Experimental setup and
results are presented in section 4 and 5 respectively. Finally, we present our conclusion and discuss the
future work in section 6.

2. Theoretical Background
In the literature organizational structure is in the list of 11 critical success factors of knowledge
management’s successful implementation (Hasanali 2002, Alazmi and Zairi 2003, Yew Wong 2005).
Successful implementation of knowledge management requires a flexible structure with modern style of
control and monitoring. The first step to create a complete system for knowledge transfer in the
organization is to develop team knowledge modules to integrate knowledge resources. Developing team
knowledge modules is based on the idea that in an organizational a task can be represented by a group of
knowledgeable and expertise in that field. However, sometimes for making better decisions the tacit
knowledge provided by other teams of knowledge are necessary. Having such a social experience is an
important element in the organizational structure in the sense that it provides the team members a chance
for knowledge exchange with other team members having the same area of interest or knowledge.
Moreover, it can be considered as a strategy for improving organizational learning performance.

2.1. Organizational learning

With the development of science and technology and expansion of business areas, the business
environment has become a challenging and competitive one. In such an environment, it is natural to see
the changes in the sources of competitive advantage. The greatest competitive advantage in the new
business survival paradigms is learning. Hence, the only way for an organization to overcome the
uncertainty, complexity and dynamics in the business environment is to have competent and efficient
workforces who are considered the organization’s important assets and the foundation of wealth. Thus,
organizations are more successful if they learn faster than their competitors. That is why the learning
46 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning

concept is growing rapidly and organizations employ it as a competitive advantage. In the organizational
learning theory an organization is considered as an adaptive system which is able to sense changes from
the environment and evolve to produce the desired outcome. Therefore, a learning organization actively
create, store, transfer and use the knowledge for its adaptation to the changing environment. The topic of
organizational learning was introduced around 1970 in Senge's most popular book “The Fifth Discipline”
(Senge 2006) which explores his vision of Systems Thinking and Learning Organizations. In the literature
we can find various proposed models for facilitating organizational learning (Argyris and Schön 1978,
March 1991, Bontis, Crossan et al. 2002, Nonaka 2008).
The notion of “tacit” knowledge became popularized by Nonaka (Nonaka 2008). He points to the
valuable subjective insights that are difficult to capture and share because they are hard to formalized and
difficult to communicate to others. He suggests four basic patterns for creating knowledge in any
organization which are mainly from tacit to tacit, from explicit to explicit, from tacit to explicit and from
explicit to tacit knowledge.
When it comes to explore the nature of organizational learning, Argyris and Schön describe what they
call single-loop and double-loop learning approaches (Argyris and Schön 1978). For them learning
involves the detection and correction of error. Within a single-loop learning approach given or chosen
goals, values, and rules are operationalized rather than questioned. Therefore, when something goes
wrong, it is suggested to look for another strategy that will address and work within the governing
variables. On the other hand, a double-loop approach questions the governing variables themselves which
may then lead to an alteration in them. This way we can expect a shift in the way in which strategies and
consequences are framed within an organization.
Crossan et al. consider organizational learning as a dynamic process of strategy renewal which is
happening across three levels of the organization which are mainly individual, group and organizational
(Crossan, Lane et al. 1999, Bontis, Crossan et al. 2002). Their framework consists of four key premises.
First, organizational learning involves a tension between incorporating new learning and using what has
already been learned (i.e., exploration vs exploitation). Second, organizational learning can happen at
different levels of individual, group, and organization. Third, organizational learning at each level is
linked by four broad categories of social and psychological processes which are mainly intuiting,
interpreting, integrating, and institutionalizing (i.e., known as 4Is framework). Forth, cognition and action
influence each other.
A simplification of the 4I framework also is presented by Hullnad (Hulland 1995). The specific focus
of this simplification is on the relationships between the three levels of learning: intuiting and interpreting
processes under the individual level, the integrating processes to inform the group level, and
institutionalizing to inform the organization level. Later, three levels are put against one another to
capture the feed-forward (i.e., how individual learning feeds forward into learning at the group and the
organizational level) and feedback processes (i.e., how the learning at the organization level affects
individual and group learning).
There are significant differences among the presented models. However, adoption of the proper model
needs the right perspective and requirements.
In this article we follow the March’s model in organizational learning (1991) in which maintaining an
appropriate balance between exploration and exploitation is the primary factor in the system survival. The
detail description of the model is presented in Section 3.

2.2. Organizational structure

The organizational structure can be defined as a network of roles with a special arrangement; they
relate to one another based on an organizational hierarchy and act in line with organizational goals. The
organizational structure is important because it is an indicator of various roles, the hierarchy of these roles
and also the distribution of power and authority within an organization. Since organizations have different
needs and characteristics, it is impossible to select a specific organizational structure that suits all of these
diverse needs. The organizational structure is based on several factors including the environment,
 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning 47

technology, the size of the organization and its life span (Ranson, Hinings et al. 1980, Walonick 1993).
Therefore, according to different applications numerous organizational structures such as Divisional,
Functional, Matrix and Project-organizational structure exist in which the overall structure of the
organization is divided into more detailed tasks, to the extent that is reasonable. Accordingly, knowledge
can flow more meaningfully within the organization.
Regardless of the grouping criteria, finally we encounter a meaningful structure that people are located
in it. Within such a structure we observe the collective learning which associates internal as well as
external learning in the organization. One question that may arise here is that why we need a structure. Is
it possible that connectivity pattern to be formed by random processes? Then for sure the answer is no,
because an organization cannot achieve its goal in a truly random way. On the other hand, how about
considering full connectivity patterns among members of an organization as it cannot be random?
Actually this is the point where we argue that it may be a future consideration for an organization but
cannot be a realistic assumption (i.e., limitation in communication capacity due to available time and
resources (Koohborfardhaghighi and Altmann 2014)).
Learning is the combination of Exploration and Exploitation processes. Knowledge exploitation
happens through interpersonal learning and an organizational structure with a special arrangement of
entities within that definitely can affect the possible range of solution space. We address these issues in
more details in the following.
The interest in the classic problem of trade-off between exploration and exploitation in organizational
learning grew in the 1990s, reflecting the importance and impact of this balancing act in an organization
decision making (March 1989, March 1991, March 1994). Similar to the recent research studies (Lewin,
Long et al. 1999, Fleming 2001, Crossan and Berdrow 2003, Siggelkow and Rivkin 2005, Gupta, Smith
et al. 2006, Jansen, Van Den Bosch et al. 2006, Miller, Zhao et al. 2006, Argote and Greve 2007, Kane
and Alavi 2007), Fang et al. (2010) pointed to the classic problem of trade-off between exploration and
exploitation in organizational learning. Individuals inside their model for achieving the organizational
goal had the incentives to explore the environment by themselves or they could achieve that through
social activities which can be regarded as exploitation. They divided the organization into subgroups
based on the idea that this isolation permits subgroups to maintain diversity and to explore more diverse
solution in the space of possibilities. They also assumed that the full connectivity pattern among the
members of the group and learning was one of the consequences of individuals’ interactions. The
following research question was investigated in that model: Does a semi-isolated subgroup structure
which improves the balance of exploration and exploitation, leads to superior long-term learning
performance outcomes? Their research results showed that modest amounts of cross-group linking are
associated with higher equilibrium performance levels.
As we previously mentioned, full connectivity pattern can be considered as a strategy for
communication between the group members and the reason for this argument might be that such a
configuration by reducing the independence of the individuals and increasing the scale of behavior among
them facilitates and speed up collective learning. Here the question rises of are we really going to deviate
from organizational hierarchies to fully networked configuration in real scenarios? Well, although in case
of a fully networked configuration we can agree on increasing the scale of behavior but it does not
necessarily coincide with reality. As it is pointed out in (Senge 2006), the very first thing needed to create
a learning organization is effective leadership which indicates the fact that individuals within an
organization work on different levels and positions. We all know that in an organization the activities of
the individual must be coordinated and there must be authorities for controlling, making decisions and
taking responsibilities. Hence considering a structure with full connectivity pattern similar to a random
structure one is unrealistic. A pure traditional hierarchical scheme also does not show mobility and lateral
connection of individuals. The existence of lateral connection is supported by the fact that two employees
of the same organization who are at the same level of authority may have the tendency to create link with
each other. The existence of such a link may represent the fact that both are as important as each other
and both reliant on the other part for the whole work to be done properly. So we need a more flexible
hybrid structure to allow lateral connection between the individuals, a structure which is between the two
48 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning

extremes, one of Hierarchy and the other of full connectivity pattern. In this article we assigned level wise
relationships among the members of an organization. The executives, directors, managers and employees
in order of rank, fill the levels in the hierarchy respectively. We allowed a small degree of lateral
connection and connectivity among the groups. We followed the same experimental setup in Fang et al.
(2010) and the details are presented in the experimental setup section.

3. Model
March’s model in organizational learning (1991) considers the relation between the exploration of new
possibilities and the exploitation of old certainties in the context of organizations. The model agrees that
both exploration and exploitation are essential for the survival of an organization but finding an
appropriate balance between them is difficult. It is a simple model for the development and diffusion of
organizational knowledge with four key features. The first two features are the external reality (i.e.,
represents organizational goal) and an individual belief set (i.e., represents each individual’s knowledge
with respect to the organizational goal). The second two features cover the social context of
organizational learning which point to the mutual learning of an organization and individuals who are
located in it.
As it is shown in Equation 1, if xj (i.e., jth element of an individual belief set x which has m
dimensions) matches the corresponding element of the external reality, then we can say a score of = 1
has been achieved. Otherwise the individual does not receive any score for this bit of his/her belief set.
The summation of such scores determines the payoff of an individual (i.e., ) within an organization.

Individuals modify their beliefs continuously as a consequence of socialization in to the organization.

And improvement in knowledge comes by imitating the beliefs of superior individuals and their own
discoveries.
Similar to March’s model in organizational learning (1991), our model has three main entities: an
external reality, individuals, and an organization with different structures under study. Our model is
different from March (1991) and Fang et al. (2010) in that an organization is seen as a complex system
wherein individuals through different topologies interact with one another. We model an organization as a
complex system with different structures under study, where learning occurs during the interactions of
individuals which are located in it. During such interactions knowledge exchange happens and new
knowledge would be created by the individuals. In our model, the payoff of an individual (i.e., I )
within an organization is calculate according to Equation 2.

I Gj Gj ⋯ Gj 2

Let xj denote jth element of the bit string x (i.e., jth element of an individual belief set x which has m
dimensions but divided into L independent subsets 1 , 1 2 , … ,
1 ). Then, the linear payoff function can be calculated according to the above formula where Gj=1 if
xj corresponds with reality on that dimension and Gj=0 otherwise. Here, s serves as a tuneable parameter
that can control the difficulty of the search problems, and 1 ≤ s ≤m. In fact we have m-bit string for the
belief set which is partitioned into L independent subsets. Within each subset, there are s bits, whose
performance is coupled. In our formulation we set its value to 5. The higher the s parameter, the more
complex is the problem. After knowledge exchange happens between two individuals, if s continuous bits
of their beliefs matches the exact corresponding s bits of the reality then we say knowledge creation has
 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning 49

happened and they will receive payoff for finding the partial solution to the organizational goal. A
numerical example for this process is presented in Section 4 of this paper.

3.1. Entities
The reality is described with a binary vector having m dimensions, each of which has a value of 1 or
−1. Values are randomly assigned with the probability of 0.5 for each value in each dimension. There are
n individuals in the organization. Each of them holds m beliefs about the corresponding elements of
reality at each time step. Each belief for an individual has a value of 1, 0, or −1. A value of 0 means an
individual is not sure of whether 1 or −1 represents the reality.

3.2. Network structures

A key factor determining the dynamics of our model is its topology of interactions between
individuals. To have a comparative study with the work done by Fang et al. (2010), in addition to
traditional hierarchy and hybrid structure, we also repeat the experiments on a fully connected network.
Some of the structures under our investigation are depicted in Fig. 1.
As it is depicted in Fig. 1, we test different configurations for connectivity patterns of the hybrid
structure. Similar to the research done by Fang et al. (2010), we also consider a rewiring probability in
our model. We consider rewiring probability as a rewiring rate at which the new contacts are made
probabilistically among the individuals within an organization. Our view here is that it is reasonable to
consider the rewiring probability but it should occur at the different levels of the hierarchical structure
namely executives, directors, managers and employees. Each rewire is likely to create a shortcut between
two individuals within a cluster or between different clusters. It shows the tendency for each individual to
be connected to others (i.e., for example an employee within a cluster may have some restrictions to be
bounded to collaborate only with its own cluster members or it may have the chance or tendency to
collaborate with the manager of another cluster). As the rewiring probability increases, the average path
length among the population decreases.
The sensitivity analysis on the organization size in (Fang, Lee et al. 2010) shows no substantive
change in the obtained results. Therefore, we only run our simulation on a network which represents a
small organization with 72 individuals located in 4 clusters.
The network presented in Fig.1 (A) represents a traditional hierarchy within the organizational
structure. There is no mobility and lateral connectivity among the members of the groups in this network.
The network presented in Fig.1 (B) shows a hybrid structure within the individuals of an organization.
We have a low degree of mobility and lateral connectivity among the members of the groups. We also
create a loop of connectivity between the executives. The network presented in Fig.1 (C) follows a similar
trend as Fig.1 (B). The only difference is that we create a loop of connectivity between the directors of 4
clusters. In Fig.1 (D) we create a loop of connectivity between the managers of 4 clusters and similarly in
Fig.1 (E) we create a loop of connectivity between the employees of 4 clusters. We create a loop of
connectivity between the executives, directors, managers and employees of 4 clusters in in Fig.1 (F). The
network presented in Fig.1 (G) shows the network used in Fang et al. (2010) with a very low probability
of rewiring (a loop of interconnection). Fig.1 (H) depicts a network similar to the work done by Fang et al.
(2010) with a higher level of rewiring probability (e = 0.1).

4. Experimental Setup
The experimental settings for our model are as follows: We conduct an agent-based simulation in
Netlogo (Wilensky and Evanston 1999) to test our model. One reason for agent-based simulation is that
from a computational perspective they help parallel implementations. So we have a separate
computational thread for each agent (node) that is responsible for the knowledge exchange. The second
reason is that we need a dynamically changing computing environment to model the real scenario.
50 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning

The presented agent-based model in this paper is a kind of microscale model that simulates the
simultaneous operations and interactions of multiple agents in an attempt to re-create their own
knowledge and predict the correct knowledge needed at the organizational level. Each agent (i.e., an
individual in our model) has its own knowledge which might be different than others. Agents are located
within a network and out of their interactions learning happens. The process of learning is one of
emergence from the lower (micro) level of systems to a higher (macro) level and agent based modelling is
a typical approach which allows researchers to design and test their scenarios.
(A) Traditional hierarchy (B) Hybrid structure with (C) Hybrid structure with
connectivity at executive level connectivity at director level

(D) Hybrid structure with (E) Hybrid structure with (F) Hybrid structure with
connectivity at manager level connectivity at employee level connectivity at all levels

(G) Full connectivity structure with a loop of (H) Full connectivity structure with rewiring
connectivity between the clusters probability at all nodes

Fig. 1. Different network configurations depict variety of connectivity patterns within an

organization. We have 72 individuals located in 4 clusters.
 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning 51

We generate ten suits of experiments (T1, T2….T10) to check the performance of learning
organization under each structure while individuals exchange their knowledge, and we report the average
result over 100 simulations run. GUI of our simulation environment is depicted in Fig. 2.
The graphical user interface of our simulation consists of a two-dimensional field that contains nodes
and their connections. Nodes and links are representative of individuals and their connections within an
organization respectively.
There are 72 individuals in each organization (i.e., n = 72).The number of dimensions in the beliefs is
set to 100 (i.e., m = 100). Reality (i.e., organizational goal) is determined by randomly assigning a value
of 1 or −1 for each of 100 dimensions, whereas each dimension of an individual’s belief set is determined
by assigning a value randomly drawn from 1, 0, or −1. Each organization consists of 4 clusters of
individuals. (i.e., each cluster has 18 individuals within a specific hierarchy). This organizational network
is implemented by first connecting each individual to the other individuals in a pattern explained in Fig. 1.
We then connect the subgroups by creating a loop within the respective level. Similar to the model
presented by Fang et al. (2010), the rewiring probability is set to E = 0.1, while the learning probability is
set to 0.3. Simulation’s parameters are shown in Table 1.

Table 1. Simulation parameters

Simulation Parameters Remarks Parameter Values

Number of individuals in
N 72
the organization
M Dimensions of beliefs 100
Z Size of a subgroup 18
T Simulation runtime 100
C Number of clusters 4
S Degree of complexity 5
Probability of individual
Plearning 0.3
learning from the majority view
E Rewiring Probability 0.1

Fig. 2. GUI of the Model.

52 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning

We also adopt generalized learning model of March (1991) developed by Fang et al. (2010) which is
presented in Equation 2.
A numerical example is presented in Fig. 3. Let us assume s=2, in such a case the utilities of agents
with respect to the Reality vector are as follow:

Fig. 3. A numerical example is presented for better understanding of the process of organizational
learning.

Agent 1: 2*1*1+2*1*1+2*1*1+2*0*0+2*0*0=6
Agent 2: 2*1*0+2*0*1+2*0*1+2*0*0+2*0*0=0
Agent 3: 2*1*0+2*1*1+2*1*1+2*0*0+2*0*1=4
Focal Agent: 2*0*0+2*1*1+2*0*0+2*0*0+2*0*1= 2

Focal Agent decides to learn from agent1 and agent3 because their performance is higher than its own
performance. The focal agent checks the majority view on each of its own bits and updates its belief
vector. Organization’s performance is measured as the average performance across all individuals in the
organization. A perfect numerical calculation of the payoff function is presented in (Fang et al., 2010).
We also follow Fang et al. (2010) to calculate the dissimilarity in beliefs of all individuals in an
organization as follows:

2
3
1
Parameter n shows the number of individuals and m is the number of dimensions in the belief set of
each of them. This dissimilarity index shows the probability that an organization loses its diversity of
belief sets when we change connectivity patterns among the individuals within the organization.

5. Results
For each run of the model, we computed the average payoffs of the population during each period
t = 0, 1, ….,T. We compared the learning performance of hybrid structure with that of traditional
hierarchy and fully connected network over 10 suits of experiments and the average result based upon 100
runs is shown in Fig. 4. As we mentioned before, learning performance is equivalent to achieving the
correct knowledge needed at the organizational level. Therefore, the more knowledge of individuals (i.e.,
beliefs vector in our setup) becomes closer to the organization goal (i.e., reality vector in our setup), the
higher organizational learning will be. Finally, equilibrium occurs when all the individuals has equivalent
knowledge levels.
 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning 53

Our first observation is that the performance of an organization with traditional hierarchical structure
(i.e., Fig. 1(A)) is the lowest. As the result shows, the performance of a fully connected network with
rewiring probability at all nodes (i.e., Fig. 1(H)) is considerably better than the other network structures.
Another interesting observation is that a fully connected network with a loop of connectivity (i.e., Fig.
1(G)) cannot perform better than a hybrid structure with subgroups connectivity at all levels (i.e., Fig.
1(F)). This clarifies our statement about the fact that in addition to having a realistic model of interactions,
with a proper hybrid structure organizational learning can be increased. The obtained result also is in line
with the experiments done by Fang et al. (2010); although, a full connectivity pattern manages to target
larger groups of individuals, but it does not necessarily guarantee the highest performance. In order to
make it more successful we need a certain degree of rewiring probability and in the case that this degree
is not set correctly its performance decreases. We also tested different configurations for connectivity
style of the hybrid structure (i.e., Fig. 1(B-F)). Therefore, according to previous statement and our
observations the connectivity style at employee level produces the lowest performance but still is better
than traditional hierarchical approach. The result also indicates the fact that rewiring probability at
executive, director and manager level produces a nearly similar trend but finally in long run rewiring
probability at manager level produce a better result. Our model showed several properties which we
discuss them in this section.
The first issue is about the meaning of the learning performance. The reported learning performance in
Fang et al. (2010) is 80 while the eventual average learning achieved in our case is 20. The difference in
learning outcomes might be due to the different population sizes of the organization which was 280 in
that model and is 72 in our case. As argued by (Fisher and White 2000), downsizing in a learning
organization may seriously damage the learning capacity of the organization, so our conjecture is that due
to low population size we are not able to reach higher performance. It seems with a low population size
we have lower opportunities for diverse belief combinations. We consider the investigation of such a
dependency as a future extension of our model. The other explanation could be related to the complexity
of knowledge in our model (a large m) which allows lower proportion of correct beliefs than a simple
knowledge.

Fig. 4. Organizational Learning performance over 100 simulations runs. We conducted 10 suits of
experiments for each configuration and the average result is reported in this figure. The x-axis
shows the duration of the simulation, while y-axis is set to the organizational learning performance
(Equation 2).
54 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning

Fig. 5. Diversity of individuals’ belief set over 100 simulations runs. As it is shown with the
traditional hierarchy structure the diversity of the populations beliefs set is the highest while with
the full connectivity structure it is the lowest among the other alternatives. The x-axis shows the
duration of the simulation, while y-axis is set to the diversity of individuals’ belief set (Equation 3).

The larger the number of elements to learn, the longer the time required for recombination of beliefs
which may lead to some level of mastery of a domain. This could be another possibility of our future
work.
We also captured the diversity of belief set with the help of dissimilarity measure and the result is
shown in Fig. 5. This value ranged from [0-1] and shows the probability that an organization loses its
diversity of beliefs when the structure of the network is changed. The result of diversity of belief sets over
time is in line with the result shown in Fig. 4 .When the learning performance is higher the diversity of
belief sets is lower which means that the organization has achieved a high level of eventual knowledge.
The diversity of belief sets for traditional hierarchical structure is higher than the other structures while
this value is the lowest for a fully connected network with rewiring at all levels.

6. Discussions and Conclusions

Nowadays organizations can be seen as socio-economic complex adaptive systems where continuous
processes of change, adaptation, development and learning are visible within their life cycles.
(Koohborfardhaghighi 2017). Organizational learning is about individuals’ learning as agents for the
organization and it can also be seen as a process (Argyris and Schön 1978, Brown and Duguid 1991, Joop
and Andre 1994, Örtenblad 2001). Thus, organizational learning would be processes going on in the
learning organization. A learning organization refers to a particular type of organization; it is simply an
organization that learns. The transfer of knowledge in learning organizations is supposed to go on
between individuals not between individuals and the memory of the organization (Garvin 1985). However,
in order to be valid as organizational learning, individuals learn the “correct” knowledge for the
organization and the knowledge acquired by the individuals is transferred to the organizational memory.
 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning 55

Knowledge management consists of processes that govern the creation, storage, dissemination and use of
such correct knowledge (Koohborfardhaghighi and Altmann 2014b).
Researchers examine organizational learning behaviours along various dimensions and from a variety
of perspectives. For example some researchers and theorists focus on the organization level learning
(Cangelosi and Dill 1965, Duncan 1979, Fiol and Lyles 1985). However, there are different views
regarding the nature of learning at the organization level. Some theorists view the organization as a
collection of individuals (i.e., our view in this paper), while others view it as the systems, structures, and
procedures of the organization. The overall systems perspective opens a space for a more integrative
contribution (Hulland 1995, Bontis, Crossan et al. 2002). In this paper we followed the model of March
in which organizational learning requires organizations to explore and learn new ways, while concurrently
exploiting what has been already learned (March 1991). Managing the tension between exploration and
exploitation is a challenging issue and is a central requirement in a theory of organizational learning.
In practice, we can measure organizational learning using different parameters such as increase in
productivity, increase in financial performance, increase in customer satisfaction, increase in
organizational knowledge base, increase in skill level, etc. However, before selecting each of those
performance measures we should have a clear undersetting of the dimensions of learning organizations.
Serrat in his research suggests concepts that can be used individually or in association to reflect on the
overall learning within an organization (Serrat 2009). Among the suggested concept we can point to
strategic learning, adaptive and generative learning (Koohborfardhaghighi, Lee et al. 2017), key functions,
human nature (Koohborfardhaghighi and Altmann 2014, 2014a, 2017) and organizational context,
learning management systems and work style matrices. A set of competences might need to be developed
to support learning from each dimension’s perspective. There are also other researches that attempts to
determine the extent to which the organization is functioning as a high performance system according to
learning organization theory and principals (Hackman and Oldham 1980, Pasmore 1988, Buckingham and
Coffman 2014).
In this research, we followed the March’s model in organizational learning (1991) in which
maintaining an appropriate balance between exploration and exploitation is the primary factor in the
system survival. The presented model has three main entities: an external reality, individuals, and an
organization with different structures under study. We tested organizational learning performance under
different network structures which are mainly fully connected networks, traditional hierarchical networks
and hybrid network structures. A full connectivity pattern can be considered as a strategy for
communication between the individuals within an organization in the sense that such a configuration by
reducing the independence of the individuals and increasing the scale of behaviour may facilitates and
speed up collective learning. We argued that this assumption is unrealistic due to the fact that in an
organization the activities of the individual must be coordinated and there must be authorities for
controlling, making decisions and taking responsibilities. On the other hand, a pure traditional
hierarchical scheme also does not show mobility and lateral connection of individuals in its own turn. So
we need a more flexible hybrid structure to allow lateral connection between the individuals, a structure
which is between the two extremes, one of hierarchy and the other of full connectivity pattern.
In our model, each individual has its own knowledge which might be different than others. Individuals
are located within a network and out of their interactions learning happens. They attempt to re-create their
own knowledge and predict the correct knowledge needed at the organizational level. Therefore, the
process of learning is one of emergence from the lower (micro) level of systems to a higher (macro) level.
We test different network structures and we observe the learning outcome out of them.
Our experimental results indicated the fact that traditional hierarchical structure has the lowest
performance in a learning organization while a fully connected network provides us the best performance.
Despite the fact that a random or fully connected network heavily affects organizational learning we
argued that considering such structures are not realistic and we need a hybrid structure in between the
extremes to represent the concepts of monitoring and control within an organization. A hybrid structure
with probability of connectivity at all levels produce a better result comparing to full connectivity with a
low degree of rewiring probability.
56 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning

In our future work we are interested to extend the presented model based on Crossan’s 4Is framework.
In this regards, we need to add new parameters to consider other dimensions of organizational learning
such as strategic learning (Koohborfardhaghighi and Altmann 2016a, 2016b), adaptive and generative
learning, and human nature and organizational context.

6.1. Implications
Utilization of network structures and its benefits reside in different domains (Koohborfardhaghighi
2013, 2014c, 2015). In this study, we argue that the work relationships between individuals in an
organization can be utilized to produce positive returns. Despite the creation of new knowledge and the
process of learning at an individual level, organizational structure affects the nature of human interactions
and information flow. The results of our analysis suggest that managers are able to use the organization
structure as a tool for improving the balance between exploration and exploitation and helping the
organization explore a diverse solution space, and subsequently achieving better long term learning
performance outcomes (March 1991, Gupta et al. 2006).
With respect to improving the balance between exploration and exploitation, we should mention that
in socio-economic complex adaptive systems such as organizations, immediate returns from exploitation
cause the system to exhibit a myopic bias. Employees within organizations tend to pursue solutions
similar to already-known solutions because they do not have access to all possible domains of knowledge.
This will push them towards their own prior experiences and utilization of existing routines.
Consequently, it decreases the likelihood that new solutions will be found. Therefore, we can argue that
adaptive processes which are only based on exploitations can become self-destructive in the sense that
they mislead the organization to become trapped in a suboptimal equilibrium. In the literature this is
known as success trap.
This argument is articulated by many researchers (Lenox 2002, Nickerson and Zenger 2002, Benner
and Tushman 2003, Siggelkow and Levinthal 2003, Argote and Greve 2007). However, further research
is needed to understand how organizations can avoid to be stuck in such self-destructive processes. Since
this seems to be more a managerial issue, we need to investigate the role of managers in helping
organizations to better maintain the balance between exploration and exploitation instead of
overemphasizing the already-known solutions.
Finally, while new knowledge is developed by individuals and institutional networks of people,
organizations play the key role in the management and flow of such new information. Every organization
has a hierarchical structure that defines how the organization operates. This includes the arrangements of
the lines of authority, power distance within the organization, how responsibilities are distributed and
how information flows.
By looking at the organizational structure within big enterprises such as Google one can see that
Google’s open communication program delivers substantial managerial features. The “70-20-10 rule” in
Google for example, gives employees the freedom to spend 70 percent of their time on current
assignments, 20 percent on related projects of their choosing, and 10 percent on new projects in any area
they desire. The “70-20-10 rule” within this enterprise represents a managerial guideline, but it also
authorizes the employees to take risks to explore a diverse solution space and achieve more innovation.
Having said that, we need to consider some other cases such as Samsung within which organizational
culture exerts dramatic influence on the organizational structure. That is to say, due to their specific work
culture the level of decentralization is lower and groups are more isolated from each other. This lower
level of decentralization may impact the learning process within subunits of the organization and
providing barriers to the rapid diffusion of ideas across those subunits. In this case such organizations
need proper incentive management to motivate managers and employees to be able to adopt to more
exploration strategies and to conquer more diverse range of solutions.
 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning 57

Acknowledgments
This research was conducted within the project BASMATI (Cloud Brokerage Across Borders for Mobile
Users and Applications), which has received funding from the ICT R&D program of the Korean
MSIP/IITP (R0115-16-0001) and from the European Union’s Horizon 2020 research and innovation
programme under grant agreement no. 723131. The authors are very grateful for the appropriate and
constructive comments and suggestions made by the referees and the Editor.

References
Alavi, M. and D. E. Leidner (2001). Review: Knowledge management and knowledge management
systems: Conceptual foundations and research issues. MIS Quarterly: 107-136.
Alazmi, M. and M. Zairi (2003). Knowledge management critical success factors. Total Quality
Management & Business Excellence 14(2): 199-204.
Argote, L. and H. R. Greve (2007). A behavioral theory of the firm-40 years and counting: Introduction
and impact. Organization Science 18(3): 337-349.
Argote, L., B. McEvily and R. Reagans (2003). Managing knowledge in organizations: An integrative
framework and review of emerging themes. Management science 49(4): 571-582.
Argyris, C. and D. A. Schön (1978). Organizational learning: A theory of action perspective, Addison-
Wesley Reading, MA.
Bajaria, H. J. (2000). Knowledge creation and management: Inseparable twins. Total Quality
Management 11(4-6): 562-573.
Barwise, J. and J. Seligman (1997). Information flow: the logic of distributed systems, Cambridge
University Press.
Benner, M. J. and M. L. Tushman (2003). Exploitation, exploration, and process management: The
productivity dilemma revisited. Academy of management review 28(2): 238-256.
Bontis, N., M. M. Crossan and J. Hulland (2002). Managing an organizational learning system by
aligning stocks and flows. Journal of Management Studies 39(4): 437-469.
Bremer, M. and D. Cohnitz (2004). Information and information flow: an introduction, Walter de Gruyter.
Brown, J. S. and P. Duguid (1991). Organizational learning and communities-of-practice: Toward a
unified view of working, learning, and innovation. Organization science 2(1): 40-57.
Buckingham, M. and C. Coffman (2014). First, break all the rules: What the world's greatest managers do
differently, Simon and Schuster.
Cangelosi, V. E. and W. R. Dill (1965). Organizational learning: Observations toward a theory.
Administrative science quarterly: 175-203.
Crane, N., C. Downes, P. Irish, I. Lachow, M. McCully, E. McDaniel and K. Schulin (2008). Leadership
imperatives of the information age. 5th International Conference on Intellectual Capital, Knowledge
Management and Organizational Learning.
Crossan, M. M. and I. Berdrow (2003). Organizational learning and strategic renewal. Strategic
Management Journal 24(11): 1087-1105.
Crossan, M. M., H. W. Lane and R. E. White (1999). An organizational learning framework: From
intuition to institution. Academy of management review 24(3): 522-537.
Da Silva, F. S. C. and J. Agusti-Cullell (2008). Information flow and knowledge sharing, Elsevier.
Duncan, R. (1979). Organizational learning: Implications for organizational design. Research in
organizational behavior 1: 75-123.
58 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning

Fang, C., J. Lee and M. A. Schilling (2010). Balancing exploration and exploitation through structural
design: The isolation of subgroups and organizational learning. Organization Science 21(3): 625-
642.
Fiol, C. M. and M. A. Lyles (1985). Organizational learning. Academy of management review 10(4):
803-813.
Fisher, S. R. and M. A. White (2000). Downsizing in a learning organization: are there hidden costs?
Academy of Management Review 25(1): 244-251.
Fleming, L. (2001). Recombinant uncertainty in technological search. Management Science 47(1): 117-
132.
Garvin, D. A. (1985). Building a learning organization. Org Dev & Trng, 6E (Iae) 274.
Gupta, A. K., K. G. Smith and C. E. Shalley (2006). The interplay between exploration and exploitation.
Academy of Management Journal 49(4): 693-706.
Hackman, J. R. and G. R. Oldham (1980). Work redesign.
Hasanali, F. (2002). Critical success factors of knowledge management.
Henczel, S. (2001). The information audit: A practical guide, Walter de Gruyter.
Hulland, J. S. (1995). Measuring organizational learning, London,[Ont.]: Western Business School,
University of Western Ontario.
Jansen, J. J., F. A. Van Den Bosch and H. W. Volberda (2006). Exploratory innovation, exploitative
innovation, and performance: Effects of organizational antecedents and environmental moderators.
Management Science 52(11): 1661-1674.
Joop, S. and W. Andre (1994). Becoming a learning organization: beyond the learning curve.
Wakingham: Addison–Weslwy Publishninh Company.
Kane, G. C. and M. Alavi (2007). Information technology and organizational learning: An investigation
of exploration and exploitation processes. Organization Science 18(5): 796-812.
Koohborfardhaghighi, S. (2017). Analyzing Socio-Economic Complex Adaptive Networks: A Hybrid
Approach, Doctoral thesis, Seoul National University, South Korea.
Koohborfardhaghighi, S. and Kim, J. (2013). Using structural information for distributed
recommendation in a social network. Applied intelligence, 38(2), 255-266.
Koohborfardhaghighi, S. and J. Altmann (2014). How placing limitations on the size of personal
networks changes the structural properties of complex networks. Proceedings of the companion
publication of the 23rd international conference on World wide web companion, International World
Wide Web Conferences Steering Committee.
Koohborfardhaghighi, S. and J. Altmann (2014a). How Variability in Individual Patterns of Behavior
Changes the Structural Properties of Networks. Active Media Technology, Springer: 49-60.
Koohborfardhaghighi, S. and J. Altmann (2014b). How structural changes in complex networks impact
organizational learning performance. 6th Intl. W. on Emergent Intelligence on Networked Agents.
Koohborfardhaghighi, S. and Kim, J. (2014c). One Node at One Step Discovery Process as an
Immunization Strategy. Journal of information science and engeeneering, 50(3), 1425-1444.
Koohborfardhaghighi, S. and J. Altmann (2015). A Network Formation Model for Social Object
Networks. In LISS 2014, pp. 615-625. Springer, Berlin, Heidelberg.
Koohborfardhaghighi, S. and J. Altmann (2016a). How network visibility and strategic networking leads
to the emergence of certain network characteristics: a complex adaptive system approach.
Proceedings of the 18th Annual International Conference on Electronic Commerce: e-Commerce in
Smart connected World. Suwon, Republic of Korea, ACM: 1-7.
Koohborfardhaghighi, S. and J. Altmann (2016b). How strategic networking impacts the networking
outcome: a complex adaptive system approach. Proceedings of the 18th Annual International
 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning 59

Conference on Electronic Commerce: e-Commerce in Smart connected World. Suwon, Republic of

Korea, ACM: 1-8.
Koohborfardhaghighi, S., D. B. Lee and J. Kim (2016). A Study on the Connectivity Patterns of
Individuals Within an Informal Communication Network. Advances in Parallel and Distributed
Computing and Ubiquitous Services, Springer: 161-166.
Koohborfardhaghighi, S., D. B. Lee and J. Kim (2017). How different connectivity patterns of individuals
within an organization can speed up organizational learning. Multimedia Tools and Applications
76(17): 17923-17936.
Lenox, M. (2002). Organizational design, information transfer, and the acquisition of rent-producing
resources. Computational & Mathematical Organization Theory 8(2): 113-131.
Lewin, A. Y., C. P. Long and T. N. Carroll (1999). The coevolution of new organizational forms.
Organization Science 10(5): 535-550.
Liebowitz, J. (1999). Building organizational intelligence: A knowledge management primer, CRC press.
March, J. G. (1989). Decisions and organizations, Blackwell.
March, J. G. (1991). Exploration and exploitation in organizational learning. Organization Science 2(1):
71-87.
March, J. G. (1994). Primer on decision making: How decisions happen. New York, NY, Simon and
Schuster.
Miller, K. D., M. Zhao and R. J. Calantone (2006). Adding interpersonal learning and tacit knowledge to
March's exploration-exploitation model. Academy of Management Journal 49(4): 709-722.
Nickerson, J. A. and T. R. Zenger (2002). Being efficiently fickle: A dynamic theory of organizational
choice. Organization Science 13(5): 547-566.
Nonaka, I. (2008). The knowledge-creating company, Harvard Business Review Press.
Örtenblad, A. (2001). On differences between organizational learning and learning organization. The
learning organization 8(3): 125-133.
Pasmore, W. A. (1988). Designing effective organizations: The sociotechnical systems perspective, John
Wiley & Sons Inc.
Pedler, M. and K. Aspinwall (1999). Learning Company. The Experience of Managing, Springer: 141-
154.
Pentland, B. T. (1995). Information systems and organizational learning: the social epistemology of
organizational knowledge systems. Accounting, Management and Information Technologies 5(1): 1-
21.
Ranson, S., B. Hinings and R. Greenwood (1980). The structuring of organizational structures.
Administrative science quarterly: 1-17.
Senge, P., A. Kleiner, C. Roberts, R. Ross, G. Roth, B. Smith and E. C. Guman (1999). The dance of
change: The challenges to sustaining momentum in learning organizations, Wiley Online Library.
Senge, P. M. (2006). The fifth discipline: The art and practice of the learning organization, Broadway
Business.
Serrat, O. (2009). Learning for Change in ADB.
Siggelkow, N. and D. A. Levinthal (2003). Temporarily divide to conquer: Centralized, decentralized, and
reintegrated organizational approaches to exploration and adaptation. Organization Science 14(6):
650-669.
Siggelkow, N. and J. W. Rivkin (2005). Speed and search: Designing organizations for turbulence and
complexity. Organization Science 16(z2): 101-122.
Sørensen, J. B. and T. E. Stuart (2000). Aging, obsolescence, and organizational innovation.
Administrative science quarterly 45(1): 81-112.
60 S. Koohborfardhaghighi and J. Altmann / How Organizational Structure Affects Organizational Learning

Walonick, D. S. (1993). Organizational theory and behavior. Organizational Theory and Behavior.
Wilensky, U. and I. Evanston (1999). NetLogo: Center for connected learning and computer-based
modeling. Northwestern University, Evanston, IL: 49-52.
Yew Wong, K. (2005). Critical success factors for implementing knowledge management in small and
medium enterprises. Industrial Management & Data Systems 105(3): 261-279.

Author Biographies
Dr. Somayeh Koohborfardhaghighi is an Assistant Professor in the Operations Management Section,
Faculty of Economics and Business of the University of Amsterdam. Her research is about Social
Network Analysis, Network Formation Models, and Agent-Based Modelling and Simulation. Her email
address is s.koohborfardhaghighi@uva.nl.
Prof. Dr. Jörn Altmann is a Professor for Technology Management, Economics, and Policy at the
College of Engineering of Seoul National University. Prior to this, he taught computer networks at UC
Berkeley, worked as a Senior Scientist at Hewlett-Packard Labs, and has been a postdoc at EECS and
ICSI of UC Berkeley. Dr. Altmann's research centres on Internet economics with a focus on economic
analysis of Internet services and on integrating economic models into Internet infrastructures. His email
address is jorn.altmann@acm.org.