Environmental Innovation and Societal Transitions 18 (2016) 1840

Contents lists available at ScienceDirect

Environmental Innovation and

Societal Transitions
journal homepage: www.elsevier.com/locate/eist

Use of fuzzy cognitive maps to study urban

resilience
and transformation
Marta Olazabal a,b,, Unai Pascual a,b,c
a
University of Cambridge, Department of Land Economy, 16-21 Silver Street, Cambridge CB3 9EP, UK
b
Basque Centre for Climate Change, BC3, Alameda de Urquijo 4, 48008 Bilbao, Spain
c
Ikerbasque, Basque Foundation for Science, Bilbao, Alameda Urquijo 362, 48011 Bilbao, Spain

a r t i c l e i n f o a b s t r a c t

Article history: Transformative change for urban sustainability and resilience calls
Received 6 February 2014 for the use of new governance approaches that take into account
Received in revised form 28 May 2015 the complexity of urban systems and associated stakeholder knowl-
Accepted 22 June 2015
edge and perceptions. This raises the need to explore the cognitive
Available online 15 July 2015
dimension in the management of urban resilience and transforma-
tion. The article presents a Fuzzy Cognitive Mapping approach to
Keywords:
develop plausible policy scenarios that support the decarbonisation
Urban resilience
of the urban energy system of the city of Bilbao, Basque Country.
Transformability
Sustainable transformation Scenario results indicate that a combination of local institutional
Fuzzy Cognitive Mapping and social action may be most conducive for stimulating effective
Bilbao and sustainable transformation of Bilbaos urban energy system.
We address the properties of the resulting cognitive network, with a
focus on the role of the energy systems connectivity which is found
to present conicting potential for resilience and transformation.
2015 Elsevier B.V. All rights reserved.

1. Introduction

With the pursuit of sustainability and resilience in the context of a rapidly changing world, man-
aging the process of transformation is becoming an urgent need. This is especially evident in the case

Corresponding author: Basque Centre for Climate Change, BC3, Alameda de Urquijo 4, 48008 Bilbao, Spain. Tel.: +34
944014690; fax: +34 944054787.
E-mail address: marta.olazabal@bc3research.org (M. Olazabal).

http://dx.doi.org/10.1016/j.eist.2015.06.006
2210-4224/ 2015 Elsevier B.V. All rights reserved.
 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840 19

of urban systems. Currently, cities account for approximately 70% of global CO2 emissions and 70% of
primary energy demand (IEA, 2008). Further, it is expected that the inuence of urban structures and
their management on global energy demand will not diminish in the foreseeable future (Madlener
and Sunak, 2011), thus remaining critical in a context of global climate change and increased resource
scarcity.
There are two main perspectives by which sustainability, resilience and transformation are
collectively approached in current literature: a socialecological perspective and a socio-technical
one. The socialecological perspective, associated with resilience management research, often
describes resilience as the ability to adapt to shocks and reorganise without suffering signicant
changes in structure and identity (Walker et al., 2004). However, it is possible for socialecological
systems to exist in resilient but undesirable states (Walker et al., 2006). Therefore, identifying
the most appropriate and desirable management options for sustainability and resilience in these
systems requires a process where different management approaches can be tested while emphasising
learning, monitoring and continuous knowledge acquisition (Berkes et al., 2003; Folke et al., 2002).
These processes are key for creating opportunities for transformation, i.e. systems deliberate and
desirable change. Within this socialecological perspective, the capacity of a system to undertake
a process of transformation requires two things. First, that those within the system recognise
themselves as being locked in an undesirable (unsustainable) state and second, the recognition that
there is a need to recongure the system by means of new components and processes (Walker et al.,
2004). So, from a socialecological perspective, a core responsibility of resilience management is to
facilitate the deliberate alteration of the fundamental properties of a system thereby undertaking
a process of transformation in order to better cope with emergent conditions (Nelson et al., 2007).
The complementary socio-technical perspective is associated with transition management research.
This perspective is rooted in innovation and technology studies (Rotmans et al., 2000) and views
transitions to sustainability as a long-term process of societal change entailing interactions between
technology, policy, economics and culture (Geels, 2011).
Both the socialecological and socio-technical perspectives attempt to understand complex sys-
tems through emphasising the importance of continuous processes of learning and adjusting (Van der
Brugge and Van Raak, 2007) and by highlighting the need to support innovative means of knowledge
acquisition (see e.g. Beratan, 2007; Nevens et al., 2013). Therefore, emerging from these perspectives,
both resilience and transition management research recognise the importance of participatory pro-
cesses to motivate and engage stakeholders in the process of change. Within these bodies of research,
stakeholders knowledge and experience is seen as a necessary prerequisite for any systems transfor-
mation in order to foster understanding and to develop a shared vision for alternative pathways (see
e.g. Holling, 2001; Loorbach and Rotmans, 2010). This points directly to the need for recognising the
importance of the cognitive dimension, mediated by values and cultural contexts, in analysing drivers
of change towards resilience and transformation management,1 especially in complex and uncertain
decision-making environments.
Human cognitive processes are mechanisms of memory, attention and vision that allow acquir-
ing knowledge and understanding through reasoning and judgement (Adams and Aizawa, 2010a;
Lewandowsky et al., 2007). Cognitive processes are context specic and deeply inuence decision-
making processes (Adams and Aizawa, 2010b; Sieck et al., 2007). Cognition affects leadership, policy
making and governance culture, three aspects that are crucial in the management of any process of
change, including adaptation and transformation (Adger et al., 2012). Increasing visibility of actors
knowledge helps deepen understandings of how systems are structured and how they work, which
is essential for making better policy decisions that avoid undesired social or environmental impacts.
Multiple approaches exist for capturing how actors cognitions affect decision making, including dis-
course analysis (e.g. Q methodology and Delphi techniques), multi-criteria evaluation (e.g. social or
participatory MCA) and mental models (e.g. agent-based or fuzzy modelling). Among them, Fuzzy

1
We use the concept of transition management as generally used in innovation studies and transformation management
as a more general notion that refers to the process of managing change from both a socio-technical and a socialecological
systems perspective.
20 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840

Cognitive Mapping (FCM) is an interesting tool with numerous comparative advantages including the
ability to simplify a complex decision environment while integrating actors different perspectives and
ideas using a semi-quantitative approach. Although not uniquely associated with FCM, an important
advantage of this method is its ability to support the development of policy scenarios. In this paper
we apply a participatory FCM approach in the context of urban low carbon energy transition planning.
We draw from previous work on fuzzy cognitive modelling applied to the resilience management
of socialecological systems (Kok, 2009) and on the potential of fuzzy thinking for addressing com-
plex urban problems (Habib and Shokoohi, 2009) especially those related to climate change and the
environment (e.g. Reckien, 2014).
As mentioned, a focus on urban systems requires understanding resilience and transition dynamics
from a socialecological perspective (Ernstson et al., 2010), as these are inherently coupled with com-
plex environmental systems across scales. In addition, a socio-technical perspective is also necessary as
there are multiple interactions among industry, technology, markets, policies, culture and civil society
in any urban energy system. To date, resilience management has been viewed as a promising applica-
tion of FCM (Kok, 2009) due to its potential for modelling the non-linear dynamics of socialecological
systems emphasised in resilience approaches (see e.g. Folke, 2006), however, the use of FCM for urban
decision-making has not yet been sufciently explored.
To the best of our knowledge, this paper represents the rst time a participatory FCM approach has
been used as a way to explore urban resilience and transformation. For this purpose, we provide an
interpretation of network indices that allows for their utilisation in resilience and transformation man-
agement. A case study of Bilbaos energy system allows us to show how FCM can help to identify and
connect actors perceptions of mechanisms underpinning sustainable urban energy transformation
processes. We also present an innovative approach to use FCM for the development and comparison
of alternative action pathways towards achieving low carbon futures in urban areas.
An overview of the use of FCM for environmental management is provided in the next section. In
Section 3, we discuss how the FCM approach can be utilised to study the resilience and transformability
of urban energy systems based on a scenario analysis. This is followed by the introduction of the Bilbao
case study and a description of the application of FCM there (Section 4). Results are presented and
discussed in Section 5, with special attention focused on policy scenarios. The paper concludes by
reecting on the potential applicability of FCM for resilience and transformation management towards
urban sustainability.

2. FCM for environmental and urban management

Cognitive maps are originally attributed to Tolman (1948), and were introduced in the context of
human systems by Axelrod (1976) to illustrate parts of a system associated with a certain decision.
Cognitive maps can represent an individual or a shared decision environment (Langeld-Smith, 1992).
The notion of fuzziness was introduced into cognitive maps by Kosko (1986) who suggested the use-
fulness of FCMs to those knowledge domains that involve a high degree of uncertainty or so-called
soft knowledge domains, such as organisation theory and political science. Since then, there have
been important advances in this area of research and its applications have expanded into many top-
ics (Glykas, 2010), including environmental management. Table 1 presents an overview of literature
where FCM is applied in the context of environmental management.
An early conceptual study on the potential of FCM in environmental decisions was conducted
by Hobbs et al. (2002), who proposed the use of FCM in ecosystems management. Later, zesmi and
zesmi (2004) compared FCM to other methods for eliciting expert knowledge in ecosystem modelling
and concluded that FCM can accommodate a higher degree of complexity and a higher number of
knowledge sources in very different disciplines.
More recently, Isak et al. (2009) developed a manual for the use of FCM for conservation, Kok
(2009) applied FCM to understanding deforestation in Brazil and Kafetzis et al. (2010) applied it in the
context of two watersheds. Fairweather (2010) used FCM in combination with other semi-quantitative
techniques (i.e. Q methodology) to analyse and compare how dairy farmers think their farms work,
and, Lopolito et al. (2011) used FCM to draw up policy options for the development of bioreneries in
rural areas. Giordano and Vurro (2010) applied the method in analysing alternative risk management
 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840 21

Table 1
Review of environmental management literature on FCM.

Context CSa Focus References

Mb Pc Sd

Natural management Hobbs et al. (2002)

Ecosystem modelling zesmi and zesmi (2004)

Natural conservation Isak et al. (2009)

Deforestation Kok (2009)

Urban systems Habib and Shokoohi (2009)

Water management Kafetzis et al. (2010)

Drought management Giordano and Vurro (2010)

Landscape modelling Wildenberg et al. (2010)

Farming as socialecological system Fairweather (2010)

Environmental industry Lopolito et al. (2011)

Environmental technology Jetter and Schweinfort (2011)

Vulnerability assessment Murungweni et al. (2011)

Economic valuation Kontogianni et al. (2012)

Climate change impacts Reckien et al. (2013) and Reckien (2014)

Farming as socialecological system Vanwindekens et al. (2013)

Environmental assessment of mines Zhang et al. (2013)
a
CS: case study is included.
b
M: the paper is focused on the improvement of the Methodological process.
c
P: the paper stresses the Participatory technique.
d
S: the paper(s) highlight(s) the applicability of FCM for Scenario creation.

strategies regarding drought management in Italy while Wildenberg et al. (2010) gathered the results
of six case studies in different European countries and used FCM from the perspective of landscape
modelling. Kontogianni et al. (2012) have discussed the application of FCM to environmental valuation,
focusing particularly on non-marketed services. Jetter and Schweinfort (2011) tested the potential for
the deployment of PV solar panels by means of FCM. This study and Koks experiment in Brazil (2009)
constitute the rst steps towards an extended application of FCM in complex system scenario planning.
Other recent studies with a signicant level of stakeholder participation include those by Zhang et al.
(2013), who aggregated 57 stakeholders coal-mine maps, Vanwindekens et al. (2013), who carried
out 49 interviews with farmers, Reckien et al. (2013), who used 134 maps of different socio-economic
groups in New Delhi, and Reckien (2014), who gathered 376 maps on the impacts of rainstorms and
heatwaves from 188 participants in different localities across an urban region in India. FCM has also
been used in climate change research such as the studies developed by Reckien et al. (2013) and
Reckien (2014) with the purpose of developing adaptation options, or Murungweni et al. (2011) who
assessed the vulnerability of different types of livelihoods.

3. Modelling resilience and transformation using FCM

The FCM approach is generally used to represent the behaviour of complex systems through
causal reasoning. It is considered a useful tool for setting management objectives, communicat-
ing and learning (van Vliet et al., 2010), especially in the context of scenario planning applications
driven by high uncertainty and complexity (Amer et al., 2013; Kok and van Vliet, 2011). A
FCM integrates accumulated experiences regarding the operation of a complex system. It is the
result of aggregating the knowledge obtained from experts who have their own understanding
about how the system operates and how it behaves under different circumstances (Stylios et al.,
1997).
Among the main advantages of FCM, is its potential to include multiple perspectives and a high
level of integration. One disadvantage is that temporal scales can be weakly incorporated (Kok,
2009; van Vliet et al., 2010). FCMs do not represent systems as physical models do, so the conclu-
sions derived might not concur with existing scientic knowledge (Hobbs et al., 2002; Isak et al.,
22 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840

2009). Instead, FCMs are built by stakeholders with different points of view or perceptions of a phe-
nomenon. This makes FCMs ideal tools when data is scarce (e.g. Reckien, 2014) or subjective aspects
of a problem are relevant (e.g. Zhang et al., 2013). Therefore, FCMs can ideally complement quan-
titative models built on existing scientic knowledge. As stated by zesmi and zesmi, although
FCMs are ideal tools for theory development, hypothesis formation, and data evaluation, [they] are
not substitutes for statistical techniques; they do not provide real-value parameter estimations or
inferential statistical tests (2004, p. 47). However, due to its semi-quantitative nature, FCM exploits
the advantage of qualitative approaches, i.e., capturing complex issues, and quantitative approaches,
i.e., consistency and reliability (Creswell, 2003). The multi-method nature of the approach is high-
lighted by the fact that FCM outputs are quantitative but can only be interpreted qualitatively, i.e.
as a degree of change relative to a baseline (Kok, 2009). The main output of the FCM approach is an
aggregated network of concepts and weighted interconnections that may also be used for scenario
development.
In the context of resilience management, the network perspective used in FCM is helpful to analyse
complex environments (Janssen et al., 2006) and can complement other types of resilience analysis by
exploring the structure of the socialecological system itself. The complexity of operationalising the
concept of resilience has been argued before in the literature regarding social and ecological systems
(Cumming et al., 2005). This complexity also translates to urban systems, especially if one considers
that in urban resilience is two-sided and cannot be seen as a normative positive concept (Chelleri
and Olazabal, 2012; Waters, 2012). Therefore, urban resilience needs to be assessed in current and
alternative future conditions in ways that is coherent with the non-equilibrium view of resilience
where the emphasis is put on dynamics and evolution rather than on maintaining initial conditions
(Pickett et al., 2004).
The network that results from a FCM exercise can be described mainly in terms of its density (D)
and the centrality of its components (Ct). D indicates the general connectivity of the network and
it is the result of dividing the number of existing connections by the number of total potential con-
nections. A larger number of concepts indicate a larger number of potential connections. It is thus
often assumed that a higher density indicates more possibilities for change, as there are more con-
nections in the network. Additionally, if these connections are perceived by the actors of the system,
such actors might turn into catalysts of change (zesmi and zesmi, 2004). Ct not only indicates the
level of connectivity of each concept (Reckien et al., 2013) but also the strength of such connections
(i.e. how much and how strong the concept is connected). Ct is an additive function of the concepts
in-degree (I) and out-degree (O). I indicates the cumulative strength of concepts entering the con-
cept being analysed and O the cumulative strength of this concept on other concepts (zesmi and
zesmi, 2004). Such strength is calculated as an additive function of the weights of the connections
to or from the concept under analysis. This way, the larger the number of connections to or from a
concept, the larger the possibilities are for Ct having a higher value, i.e. the concept being characterised
as having higher connectivity within the network. In other words, a network with high levels of Ct
among its elements, suggests a high density level, i.e. a high level of network connectivity, and vice
versa.
In general, connectivity is often taken as a feature of resilience in the urban resilience literature
(see e.g. Ahern, 2011; Ernstson et al., 2010). As argued by Ahern (2011), a systems connectiv-
ity is generally high in the built environment as it allows the system to continue functioning
in the face of shocks. It therefore correlates positively with increasing resistance, i.e. protect-
ing the urban system against unexpected impacts. Following this idea, an overconnected system
can also lead to undesirable outcomes due to an increase of the rigidity in its control (Holling,
2001). Elaborating on this, here we further argue that, either referring to the system (network)
in general or to specic elements, for connectivity to be benecial in an urban resilience man-
agement context it must also be accompanied by a high degree of redundancy of elements
in the network that provide the same services as those elements likely to be vulnerable to
shocks.
In the context of transformation, network connectivity is not necessarily a desirable property of
the system. If connectivity increases, the number of non-linear feedbacks might also increase. When
planning for transformation, this implies having to deal with a more complex system, since more
 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840 23

variables and connections among variables would need to be considered and the number of possible
futures might increase exponentially. For this reason, building scenarios based on potential policy
options can be helpful for decision-making. However, scenarios cannot be equated with predictions.
Instead, scenarios are suitable policy tools in that they provide alternative and often competing ideas
on how the future may unfold (Jetter and Kok, 2014, p. 11).
The development of scenarios can be based on combining policy measures which control a set of
key elements of the network. Thus, they can be used to optimise the occurrence of positive (desired)
impacts on the system. The scenarios that are built based on the modelled network can therefore be
interpreted as what might happen in the future if alternative sets of policy options are used. This is
done in the context of the collective knowledge that best illustrates the complexity of the system. It
follows that an alternative transition pathway is represented by a combination of policy options that
condition the state of one or more key elements of the network. For modelling purposes, this is done
by setting a xed value on an outgoing inuence of a variable, so that all incoming inuences are also
blocked. Variables can also be removed from the network given the assumption that they have no
signicant inuence over the other elements in the network.
A state vector is used to analyse the dynamic behaviour of the network and to build scenarios
upon the establishment of certain xed conditions. The dynamics of the state vector are calculated by
focusing on the inuence of each factor on the other factors over a number of iterations, or time steps
(k) (normally, 2030 iterations) (Kok, 2009) (see Eq. (1)). The state vector A of the baseline scenario is
calculated by initially setting the state values of all the concepts in the network to one.



(k)  (k)
N
(k+1)
Ai = f Ai + Aj wji (1)

j=
/ i
j=1
(k+1) (k)
where Ai is the value of concept Ci at time step k + 1; Ai is the value of concept Ci at time step k;
(k)
Aj is the value of concept Cj at time step k; and wij is the weight on the inuence from Ci to Cj . The
function f is a threshold function used to normalise the values at each step between 0 and 1 (Groumpos,
2010 describes types of thresholds functions).

4. Low carbon energy pathways in Bilbao

Next, we illustrate how a FCM approach can be used to capture expert knowledge and cognitions
on the part of stakeholders in the urban energy system in the city of Bilbao, Basque Country. This is
part of an exploratory study that specially aims to look at the cognitive dimension of urban resilience
and transformation to a low carbon energy context.

4.1. Case study context

Bilbao is a medium-sized city with a surface area of 41 km2 and 351,629 inhabitants as of 2012.
It is located in the province of Bizkaia, in the Basque Country. The mountainous terrain on which it
stands has contributed to a population density of 8662.95 inhab./km2 (2012). The compactness of
Bilbaos urban form (114.15 dwellings per residential hectare in 2012) is a potent reason for moving
towards a sustainable development plan for the city. Such planning is intrinsically linked with the
post-industrial transformation of the area after its severe industrial decline in the 1980s (similarly to
many UK cities such as Cardiff, Manchester, London or Liverpool).
The economy of Bilbao was traditionally based on steel and shipbuilding until the major indus-
trial crisis of the 1980s, after which the city transformed itself into a service-led city. The large-scale
transformation of its economic structure in the 1990s also included the renovation and revitalisation
of its riverside, in a process that has become a role model for sustainable renovation. Today, Bilbao
has emerged as a potent driver for the economy of the Basque Country. GDP per capita in the Basque
24 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840

Country was 32% above the European regional average in 2010. GDP per capita in Bilbao in 2010 was
about 28,943 Euros, more than 85% of it being generated by the services sector.2
There is a well-established Local Agenda 21, and the local government claims to be committed to
sustainability through initiatives to restore the peri-urban green infrastructure (Green Belt of Bilbao),
and through policies on climate change and sustainable energy, such as the recently approved (though
not yet enacted) Sustainable Energy Action Plan of 2012, drawn up under the Covenant of Mayors
Agreement.3
Regardless of these initiatives, electricity consumption per capita in Bilbao continues to increase
while the size of the population remains approximately constant. The total consumption in Bilbao
increased by 7% from 2003 to 2012 even though consumption by industry decreased by around 18% (BG,
2013), mostly due to the economic recession. There has been a signicant intensication in electricity
consumption by the non-industrial sector (i.e. residential, services and others) with a 9% increase per
capita (BG, 2013). Meanwhile, renewable energy produced in Bilbao only covers 2% of its total energy
demand (BG, 2013), while the rest of the energy (electricity and gas) is supplied by private energy
utilities which have a share in the renewable energy production at the national level.

4.2. Knowledge elicitation process

The FCM used here is designed to represent causal relationships as perceived by stakeholders. This
requires an elicitation process which is generally designed ad-hoc. It can be carried out individually
(see e.g. Reckien, 2014) or in groups (see e.g. Wildenberg et al., 2010) using questionnaires, reviews
of media and written information and/or interviews (Isak et al., 2009; zesmi and zesmi, 2004).
For this exploratory study, the stakeholders individually identied the most important elements and
drew their own maps through face-to-face interaction. This provides them with more freedom and
exibility as they can represent the problem from their own perspectives, avoiding the need to
follow a consensus. All their individual thinking is later merged and aggregated into a single map. Yet,
as noted by Reckien et al. (2013), this individual process is operationally challenging as one needs to
decipher the elements and connections in the process of aggregation.
Fourteen stakeholders were requested to develop individual FCMs regarding the use of energy
in the city of Bilbao. Expertise and representation of diverse backgrounds and social groups were
used as the main criteria to select the stakeholders. These included people from the civil adminis-
tration, NGOs, representatives of the general public, academics and employees of private companies
with diverse technical backgrounds including law, planning, sustainability, social behaviour, building,
energy infrastructures and management. Due to the exploratory nature of this exercise the limited
number of interviews conducted was considered sufcient for representing various local perspectives
(notwithstanding the considerations by other authors e.g. zesmi and zesmi, 2004; Zhang et al.,
2013, about the minimum number of required interviews). The key engaging topic/question used to
stimulate participants was: What do you think inuences the use of energy in Bilbao and what are
its impacts?
As the stakeholders were interviewed, we obtained individual maps containing the concept ele-
ments (Ci ), their interconnections (Ci Cj ) and also the weights of those relationships (wij ). To that end,
each stakeholder was asked to follow a specic step-process to create their individual map (see Isak
et al., 2009; Reckien et al., 2013). First, they listed the elements that might play a role in relation to the
key question about the energy system and they placed their primary element in the middle of a blank
sheet of paper. Then, they placed the remaining elements around the rst one, making the appropriate
connections and indicating whether the relationship was positive or negative. Finally, they weighted
each relationship on a scale from zero to one in 0.1-steps being 0.1 the weight of the weakest relation
and 1 the weight of the strongest relation.

2
The statistics on the Basque Country and the municipality of Bilbao are collected from Eustat (Basque Statistics Ofce),
URL: http://www.eustat.es (last accessed December 18, 2013). The European statistics are collected from the Eurostat Regional
Yearbook 2013, URL: http://epp.eurostat.ec.europa.eu/ (last accessed January 25, 2013).
3
URL: http://www.eumayors.eu/ (last accessed January 25, 2014).
 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840 25

4.3. Post-processing and construction of aggregated network

After the stage of knowledge elicitation and the process of developing individual maps, we pro-
ceeded to post-processing of individual maps and aggregation. This is a time-consuming activity
and requires the involvement of stakeholders to check inconsistencies in their maps and avoid
potential pitfalls (Jetter and Kok, 2014). Each original map, as drafted by the stakeholders, was rst
translated into a single composite matrix (examples of original maps in the original language are
shown in Fig. A1 in Appendix 1; all original digitalized maps are shown in Fig. A2). Aggregation
required a process of adjustment that involved merging similar concepts and renaming ambigu-
ous ones. Stakeholders were contacted in some cases when clarication was needed. In the process
of aggregation, the weights of the connections were added and the entire matrix was normalised.
In this step, we found no contradicting relations, i.e. the same relation opposite signed by differ-
ent stakeholders, and no inconsistencies at this stage that needed to be considered. The exercise
resulted in a matrix of 86 merged concepts from a total of 139 in the original phase. Although
there are alternative ways that cognitive maps can be captured and modelled (Jetter and Kok,
2014), in this study we used FCMapper software based on worksheets in VBA coding (Bachhofer
and Wildenberg, 2009) to model the nal aggregated network. Results are discussed in Section
5.

4.4. Calibration and testing of the aggregated model

Checking the reliability and coherence of the FCM approach strengthens the validity of the
nal network. According to Jetter and Kok (2014) this can be done in three steps: (1) calibra-
tion of the model, (2) testing with complex problems or ideas to investigate and (3) ongoing use
and modication. The identication of missing connections or elements is required to improve
the model. In turn, this process should be transparent to participating stakeholders (Jetter and
Kok, 2014). Again, given the exploratory nature of the study, we only undertook the two rst
steps (calibration and testing) as the model has not yet been utilised for real-world policy prac-
tice.
The calibration of the model (step 1) cannot be understood in the traditional quantitative
way as FCM are semi-quantitative, thus, it cannot be calibrated using quantitative data. Fol-
lowing the recommendations by Jetter and Kok (2014) we have instead tested the outcome of
the model for a selection of well-known cases, as reported by stakeholders during the inter-
views. For this, calibration was undertaken with two contrasting examples: (i) an increase
in the energy price paid by households should lead to a decrease in energy use and pur-
chasing power; and (ii) an increase in the extent of building rehabilitation should lead to a
decrease in energy use and a likely change in lifestyles for instance by improving the quality
of time spent at home which in turn might allow for new habits to emerge. Both examples
were tested successfully. During this process no missing connections or inconsistencies were
found.
Following the suggestion by Jetter and Kok (2014), the testing of the calibrated model (step 2),
has been undertaken through scenario building where three different hypothesis or policy options
appearing to lead to low carbon objectives have been tested. The calculation of the state vector A is
used to build scenarios based on three alternative transition pathways, all of which intend to achieve
a low carbon future in Bilbao. These three scenarios were set based on discussion with the stake-
holders (Scenario A) and from the literature on sustainable urban transitions (Scenario B and Scenario
C).
In Scenario A the focus is on the use of nancial incentives. It is largely determined by the use of
subsidies provided by the local government for activities such as rehabilitating buildings destined to
low income families, under the assumption that the nancial crisis is overcome (Fixed value = 0 for
Economic crisis, see Table 2). This scenario has been previously posed as plausible by stakeholders in
the case study if Bilbao is to become a low carbon city (Olazabal and Pascual, 2013) and repeatedly
emerged as a key potential solution during the interviews to stakeholders. In Scenario B, transforma-
tion is assumed to be driven by bottom up initiatives and the focus is placed on pursuing effective
26 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840

Table 2
Scenarios and xed values of key elements in the network.

Scenario A: optimisation of economic incentives

List of variables inuencing Scenario A Fixed value

Economic crisis 0
Economic incentives 0.8
Energy price (households) 0.7

Scenario B: strong education and awareness campaigns

List of variables inuencing Scenario B Fixed value

Ignorance of the benets of energy efciency measures 0.2

Information availability 0.9
Lack of awareness of the cost of resources for the economy 0.2
Lack of awareness about the impacts of consumption 0.2
Awareness/education 0.8
Social awareness 0.8

Scenario C: stronger local institutional initiatives

List of variables inuencing Scenario C Fixed value

Energy/climate policies/regulations 0.8

Local policy will (initiative) 0.8
No local strategy 0.2

public educational and awareness campaigns. In this scenario a combined strategy to increase infor-
mation for citizens, education, and awareness is implemented with the aim of analysing the effect of
social behavioural change towards a low carbon transition. Lastly, in Scenario C, the transformation
is catalysed by top down policies and the motivation of local governmental organisations. This sce-
nario is focused on strengthening local institutional initiatives so that local decision makers are the
main drivers of change, for example by strengthening the local policy framework favouring energy
efciency and climate action. For all scenarios, xed variables have been set to high (between 0.7 and
0.9) or low (between 0 and 0.2) values, as specied in Table 2.
Eventually, given the large number of potential unintended effects of one-sided strategies which
might result in undesirable or unstable transitions, and given the important role of social movements
in building adaptive capacity (Barthel et al., 2013), a combined scenario was added: Scenario B&C. This
scenario combines Scenario B, representing bottom-up social action, and Scenario C, representing top-
down local governmental action. By adding this scenario we enrich the analysis about how sustainable
urban transitions might be achieved.

4.5. Comparison of scenarios

FCMapper software (Bachhofer and Wildenberg, 2009) provides the scenario results in two ways.
First, it provides the results in a scale from 0 to 1, expressing the state of each variable (c.f. Eq. (1))
through a logistic threshold function (Gray et al., 2014; zesmi and zesmi, 2004; Reckien, 2014).
Second, it translates these results into a scale from 1 to 4 indicating the strength of the change, either
positive or negative.
In order to compare the scenarios, we use a sustainability reference scenario specically developed
for our case study where the network variables are grouped into three sets depending on the desir-
able management objective in terms of sustainability and resilience. This methodological approach
is similar to the one used by Reckien (2014) who also develops an approach to compare scenarios
using results from FCMapper. We classify network variables into positive-type, negative-type or neutral
variables from a sustainability and resilience perspective. Positive-type variables are associated with
a positive impact on sustainability and resilience (e.g. Availability of resources) and their change in the
 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840 27

Fig. 1. Individual maps characteristics: density vs. number of variables.

reference scenario relative to their original value is scaled from 1 (minimum change) to 4 (maximum
positive change). Negative-type variables are associated with a negative impact on sustainability and
resilience (e.g. Energy price-households) and their change in the reference scenario relative to their orig-
inal value is scaled from 1 (minimum change) to 4 (maximum negative change). Neutral variables
are those whose values in the reference scenario remain largely unchanged.

5. Results and discussion

5.1. Key features of the networks in the context of resilience and transformability

Fig. 1 shows the relationship between the main characteristics of the original individual maps
(see Appendix 1) with regard to density (D index) and a number of concepts (hereafter referred
to as variables). We observe that the higher the number of variables mentioned during the inter-
views, the lower the density of the map. This might imply that the more crowded the map gets,
the more difcult it turns for the participant to establish appropriate connections between the vari-
ables.
The network is represented as a visual map in Fig. 2. The aggregated network built after the
post-processing of the individual maps comprises 86 variables and 161 linear connections. Nega-
tive connections (negative wij ) are represented by dotted lines. Positive connections (positive wij ) are
represented by solid lines. The variables are organised into nine groups according to thematic issues
(see Table A1 in Appendix 2): (1) Environment & resources; (2) Innovation & opportunities; (3) Social
behaviour; (4) Local competences; (5) Economy; (6) Lifestyle; (7) Governance; (8) The energy business
(i.e. the corporate energy sector); and (9) Supply & demand.
The network has a density index of 0.022 which is much lower than the density of the original
individual maps because of the higher number of variables included. This means that 2.2% of the
maximum number of connections that could potentially exist in theory between the 86 concepts
are actually made. For illustrative purposes, we have further aggregated the network into the nine
group variables expressed in positive terms, e.g. proactive governance instead of governance alone
(see Fig. 3). Because of the process of aggregation this new network has a density index of 0.7 (nine
variables and 57 connections). We also observe that the centrality (Ct) of the variables is similar (with
the exception of the group variable Life Basics).
28 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840

Fig. 2. Graphic visualisation of the cognitive urban energy map of Bilbao: Solid (dotted) arrows represent positive (negative)
relationships. The bigger the elements in the map, the higher their degree of centrality (Ct).

Interpretation of these gures is challenging. Unless the density values are compared to those of
other networks representing a similar topic (see e.g. livelihoods comparison in Murungweni et al.,
2011), there is no way of knowing whether this is a low or high level of network connectivity for a
network of an urban energy system. In general terms, given the aim of stimulating transformations

Fig. 3. Aggregated network (9 higher-order variables and 57 connections).

 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840 29

Fig. 4. Network indices: out-degree (O), in-degree (I) and centrality (Ct) = O + I. Only elements of the network with Ct > 20% are
shown.

towards a low carbon system, we assume that a higher density is a desirable property of the network.
But of course, having the seeds to stimulate a change does not indicate that the change will lead to
actual desirable impacts. Controlling the agency of change, i.e. governing the mechanisms of transfor-
mation, is also key. Although the 86 concept-aggregated network can be difcult to visualise, it has
more potential in terms of the conclusions that we can obtain from its analysis and the insights that we
can gain from the scenario development. Thus, the original aggregated network with 86 concepts is
used for further analysis. Fig. 4 compares the different levels of the indicators centrality and out- and
in-degrees, for those with a signicant Ct value (see complete list in Table A1 in Appendix 2). For the
sake of comparison, Fig. 4 only shows the indices relative to the maximum centrality, i.e. Ct max 13.25
observed in Energy price (households), which means that from the stakeholders perspective this is the
most strongly connected variable. In general, the variables grouped under energy Supply & demand
(Group 9) are more central. The high Ct of some elements, such as Energy Lobbies, is due to their having
a relatively high out-degree index, i.e. they have a high level of inuence on other elements of the
network. Conversely, other variables of the network, such as Energy Efciency, owe their centrality to
a high in-degree index. Thus, Energy Efciency plays an important role in the energy network of Bilbao
because of its high sensitivity to changes in a large number of other variables in the network.
While we assume that the high density of the network is a positive attribute it is also the case that
the high level of network connectivity, especially regarding some of the variables that are key entry
points for low carbon management (for example energy use in households or energy use in transport),
may make the energy system management more complex, and this may have the unintended effect of
hindering the establishment of smooth transition pathways free from surprises and unintended effects.
Scenario results will later help us to gain deeper understandings of the inuence of connectivity on
resilience and transformation capabilities (see Section 5.2 and 5.3).

5.2. Scenario analysis

The extent of the change in each element of the network has been tested for the three scenarios, as
well as for the combined scenario (see Table A1 in Appendix 2). The proportion of variables that change
30 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840

under these scenarios is relatively high (between 64% and 74%), which suggests a relatively high level
of connectivity of the network and the level of complexity involved in controlling a low carbon urban
energy transition process. A description of the results of each of the scenarios is presented in terms of
the potential of reaching a low carbon energy system.
The results of Scenario A suggest a stimulus for the local economy. In this scenario, energy use
increases as energy prices decrease, mostly in lower income households. This suggests that focusing on
such economic incentives does not sufciently stimulate a change in lifestyles, which means that high-
energy-consumption behaviour patterns are essentially maintained. As shown in other sustainable
urban scenario case studies (Naess and Vogel, 2012) these results reveal that effective sustainability
policies require multiple measures implemented across many sectors and scales, which need to be
not only institutionally but also socially-led. Scenario A shows a perverse effect if the current nan-
cial incentives structure is maintained because, contrary to expectations, building rehabilitation to
enhance energy efciency is expected to decrease. According to various stakeholders, the experience
of Bilbao indicates that as nancial incentives are not targeted at higher income households, the ben-
ets of such incentives (lower energy use, greater comfort) are often not sufciently perceived by the
majority of property owners, at least in the short term.
Scenario B is driven by strong education and awareness campaigns. In general terms, it provides
better results than Scenarios A or C as more variables undergo a desirable sustainable change (see
comparison of scenarios in Fig. 5). In this scenario, the strategy is designed to increase information to
citizens, education, and awareness of the benets of a low carbon transition. The results of Scenario
B suggest that this approach can result in the strengthening of local (community) initiatives and
an improvement in the local economy, especially in terms of attracting businesses to the local area
associated with renewable energy.
Scenario C is primarily focused on giving a key role to local authorities with regard to driving the
low carbon transition, by delivering energy and climate policies and regulations and also by building
an urban strategy towards sustainability and resilience. This scenario provides interesting results in
that environment and resources-related variables and variables related to innovation and generation
of opportunities undergo a positive change. While energy use in the city decreases resulting from
top-down local initiatives, this scenario suggests unintended effects such as an increase in household
energy bills.
In line with van Vliet et al. (2010), some of the results of this analysis might, to some extent, be
considered unexpected, as they lead to conclusions which might be unreasonable based on current
knowledge. For example, the results from Scenarios B and C suggest an increase in the negative-type
variable High-energy consumption lifestyles. This might be seen as illogical given that the two sce-
narios result in a Change in lifestyles/beliefs. Returning to the original aggregated matrix, it can
be seen that the variable High-energy consumption lifestyles is neither directly nor indirectly con-
nected to Change in lifestyles/beliefs. To check the sensitivity of the scenario results regarding this
inconsistency, we rerun the scenarios assuming a link between these two variables and analysed
the potential effects on energy use-related variables. The analysis showed that their values remain
unchanged in respect to original scenarios. This points out to the strong direct link between e.g.
Change in lifestyles/beliefs and Energy use (households) which is already present in the original
unchanged aggregated network. This type of inconsistencies illustrates the need to review and com-
plement these models with other analytical approaches. Ideally the results would need to be presented
back to the stakeholders to identify inconsistencies and to validate the modelling approach consen-
sually.

5.3. Optimising sustainability transitions

The three plus one scenarios are compared against the sustainability reference scenario in Fig. 5,
notwithstanding that establishing a desirable sustainable objective for each variable is not straight-
forward, given that variables are not always verbalised in a positive sense (e.g. No local strategy). The
resulting network contains 53 positive-type variables (e.g. the variable availability of resources) and
27 negative-type variables (e.g., the variable Energy use in households) (see Table A1 in Appendix
2). Six variables are considered neutral, meaning they cannot be related to a particular sustainability
 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840 31

Fig. 5. Comparison of scenarios against the sustainability reference scenario. The different gures represent the positive or
negative change caused by the variables in the network. The X-axis represents strength, which is the relative degree of change
in the variable under the conditions of the new scenario. It ranges from +4 to 4, with +4 being the strongest positive change
and 4 the strongest negative change. The Y-axis represents the number of variables. The graphic shows positive-type variables
(coloured in white and optimally >0) and negative-type variables (coloured in black and optimally <0). The grey bar shows the
number of variables that do not suffer any change under the different scenarios.
32 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840

objective in this case study and their associated value equals zero (see top of Fig. 5). Positive-type
variables are located on the right side of the gure, with more positive-type variables being associated
with an expected greater benecial change. The relative most sustainable, most resilient placement
for negative-type variables would be the left hand side of the gure. The more negative-type variables
located on the left hand side of the gure, the greater the benecial change in terms of a transition to
a sustainable urban energy system.
We observe that the combined Scenario B&C provides better results than either B or C individually
under two criteria: sustainability outcome and agency of change. In this combined scenario none of the
values of positive-type variables decreases under the inuence of the new condition of the system. In
addition, slightly fewer negative-type variables are seen to show an increase in their values. Both these
facts result in a higher sustainability outcome. Also, the number of variables that remain unchanged
signicantly increases, which could be interpreted as the result of better control and agency in the
process of transformation.
In summary, according to collective knowledge gathered through the individual FCMs, the results
of the case study in Bilbao suggest that a combination of top-down and bottom-up actions provides a
higher likelihood for achieving the desired transformation associated with the urban energy system
under study. Factors such as the cost of implementing these policies and their robustness in the long
run fall beyond the scope of this study, although they are critical if any energy transition plan is to be
developed in a city. Further, we nd evidence of the two-sided perspective of resilience. Connectivity
of the network might be a double-edged sword because, on the one hand, this attribute may be desir-
able as it may translate into more seeds of change (zesmi and zesmi, 2004) and could therefore
provide more opportunities for self-organisation after a shock if a process of transformation is under-
taken. On the other hand, when there is a high level of connectivity, there might be more non-linear
feedbacks, thus, change and transformation might be more difcult to control. This undesirability
of a highly connected system is discussed by Holling (2001) who argues that an overconnected sys-
tem can be rigid in its control thus increase the vulnerability to unexpected disturbances. That is,
when the number of direct and indirect drivers increases, the control in the agency of transforma-
tional change may become more complex and the change more difcult to manage. For example in
Scenario A we have observed that, due to the high centrality of key elements of the network, the
urban energy system in Bilbao may be highly resilient to nancial changes i.e. in individual income or
public budgets. But concurring with previous studies (Lopolito et al., 2011), we nd that while eco-
nomic incentives show a high level of connectivity within the network, the scenario analysis suggests
that it can also lead to undesired impacts such as an increase in air pollution (see detailed results
in Table A1). In other words, the energy system is unlikely to undergo a process of transformation
towards sustainability through economic incentives alone. In fact, such an approach may instead lead
to mal-transformation, retaining high energy consumption patterns and thus leading to a deeper
unsustainable lock-in situation (Galarraga and Abadie, 2014). Also, we observe that reinforced posi-
tive impacts are obtained by combining local-institutional and community initiatives that fuel higher
levels of environmental education, awareness and responsibility among citizens (i.e. Scenario B&C).
Positively, in this combined scenario we have observed that there is an increase in the number of
neutral variables (variables that remain unchanged) which might also indicate a weakening in the
connectivity of the system which might be seen as an uptake of a meaningful change towards a low
carbon transition.

6. Conclusions

Processes of sustainable transformation in cities are not merely a problem of technology develop-
ment or deployment; politics, power relations, economics, culture and value systems all inuence
urban transformation (McCormick et al., 2013; Pelling and Manuel-Navarrete, 2011). Higher and
more diverse participation levels from early stages in urban decision-making processes may be key
to gain the knowledge needed to achieve urban resilience and sustainability (Colding and Barthel,
2013). We argue that better understanding of expert knowledge and perceptions on the part of policy
makers and stakeholders, understood as cognitions, can greatly inform resilience and transformation
management. Two important applications of Fuzzy Cognitive Mapping (FCM) as a tool support this
 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840 33

challenge: scenario development (Kok, 2009) and communication and system dynamics visualisation
(van Vliet et al., 2010). Both these uses contribute to the necessary learning process in resilience and
transformation management.
This paper provides an exploratory exercise and contributes to the literature on resilience and
sustainability transitions by using the FCM approach in the particular context of urban decision
making to study resilience and transformability capacities of cities. This is done by means of a
case study of the city of Bilbao, exploring the structure of its urban energy system and develop-
ing low carbon energy transition scenarios. This generates hypotheses of combined policy actions
conducive to the sustainability of the energy system of Bilbao. Accounting for the different knowl-
edge and perceptions of stakeholders and communities concerning the impacts of climate change
and resource use, the study evidences the great complexity, dependence, coupled nature and con-
nectivity of urban adaptive systems. The case study illustrates how FCM may help to draw different
comparable futures regarding resilience and sustainability. For example, we have noted that success-
ful transformation in Bilbao may pivot on a combination of top-down and bottom-up actions to unlock
resilient but unsustainable states. Also, the case study shows that rebound effects of economic incen-
tives should be carefully analysed to prevent undesirable outcomes that might hinder the agency of
change.
From a methodological viewpoint, FCM as a participatory cognitive modelling approach has proven
useful to explore the cognitive dimension of urban resilience and transformation through knowledge
elicitation and scenario building. The FCM approach also allows for discussion of the potential conict-
ing effects of a networks connectivity on resilience and transformability. While a highly connected
system can represent an opportunity for change when a transition is planned, it also uncovers vul-
nerabilities that may lead to collapse in face of unexpected shocks. We posit that the future of FCM
as a method for analysing resilience and transformability should prioritise a focus on: (i) combin-
ing FCM with quantitative data/models; (ii) exploring the applicability of FCM for testing innovative
social approaches to transformation such as visions of transition; and (iii) identifying the advantages
and limitations of the application of FCM to the study of general vs. specic resilience (Folke et al.,
2010), i.e. resilience of the whole system (e.g. resilience of the urban system) vs. resilience of parts
of the system (e.g. resilience of the urban energy system). Further investigation should also focus on
the sensitivity of the network, as this could provide information about critical thresholds or tipping
points.

Acknowledgments

We thank the European Science Foundation funded COST Action Integrated assessment tech-
nologies to support the sustainable development of urban areas (TU0902). We also thank Diana
Reckien for her advice on the use of the FCM method, Martin Wildenberg and Michael Bachhofer
for the FCMapper software support provided in the early stages of this study and Marc Neumann
and Eneko Garmendia for helpful comments on improving the earlier drafts of the manuscript. We
gratefully acknowledge Mary Thompson-Hall for her assistance in editing the nal version of the
manuscript. We also thank the constructive comments of two anonymous reviewers. Finally, this
study would have not seen light without the generous participation of stakeholders of the city of
Bilbao.

Appendix 1. Original individual maps

Appendix 2. Complete results of network and scenario analyses

Table A1 summarises the results of the experiment: network analysis (Section 4.1) and scenario
generation (Section 4.2). The scenario results obtained using FCMapper software show the degree of
change in each variable under the conditions of each of the Scenarios A, B, C and B&C. In FCMapper, the
parameter is scaled from +4 to 4, with +4 being the strongest positive change and 4 the strongest
negative change. The table also shows each of the variable management objectives established in this
experiment (+ when a positive change is desirable and when a negative change is desirable). The
34 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840

Fig. A1. Examples of individual FCMs drawn up during the interviews: (a) Participant #9; (b) Participant #11.

analysis of the network includes the out-degree (O), in-degree (I) and centrality (Ct) of each variable. For
example, air pollution (Id 1), changes strongly in Scenario A until the maximum degree (+4) relative
to its original value. This means that the levels of air pollution will increase strongly if alternative
pathway A is taken. By contrast, air pollution levels signicantly decrease (degree of change: 3) in
Scenarios B, C and in the combined B&C.
M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840 35
36 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840

Fig. A2. Original digitalised maps.

 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840 37

Table A1
Complete list of variables and main results of the FCM scenario and network analyses.

Id Variable Scenariosa MM.O.b Networkc

A B C B&C O I Ct

Group 1: environment & resources

1 Air pollution 4 3 3 3 0.00 4.50 4.50
2 Availability of resources 2 2 4 4 + 3.30 1.70 5.00
3 CC mitigation 3 2 1 2 + 0.00 1.75 1.75
4 Temperate climate 0 0 0 0 + 0.90 0.00 0.90
5 Urban environmental quality 3 3 3 4 + 0.60 1.60 2.20
6 Quality of urban space 3 3 3 4 + 0.00 1.90 1.90
Group 2: innovation & opportunities
7 Building rehabilitation 4 2 4 4 + 1.08 3.40 4.48
8 Business/economic activity generation 3 2 1 2 + 0.00 0.20 0.20
9 Collectivising of goods and services 3 2 2 3 + 0.00 1.00 1.00
11 New development opportunities 0 0 0 0 + 0.30 0.00 0.30
12 No motivation in building sector 4 4 4 4 0.00 4.00 4.00
13 Promotion of self-generation 3 2 2 3 + 0.90 2.20 3.10
14 Publicprivate partnership 3 3 3 3 + 0.80 0.80 1.60
15 R&D investment 1 1 3 3 + 0.80 0.50 1.30
16 Structured social movements 0 0 0 0 + 0.20 0.00 0.20
17 Tech availability and accessibility 0 0 0 0 + 0.80 0.40 1.20
18 Urban compatibility 1 4 3 4 + 0.50 1.00 1.50
Group 3: social behaviour
19 Change in lifestyles/beliefs 3 3 3 3 + 1.10 0.80 1.90
20 Comfort/accessibility (energy use) 1 1 1 1 + 0.70 1.00 1.70
21 Condence in tech 0 0 0 0 + 0.60 0.00 0.60
22 Consumerist leisure activities 1 1 2 2 0.00 0.90 0.90
23 Disbelief in activists 3 3 3 3 0.00 0.80 0.80
24 Education 2 4 + 3.30 1.20 4.50
25 High energy-consuming lifestyles 1 1 1 1 1.20 1.20 2.40
26 Ignorance of the benets of energy efciency measures 4 2 0.00 1.30 1.30
27 Individualism 0 0 0 0 0.40 0.00 0.40
28 Information availability 0 0 + 0.60 0.00 0.60
29 Lack of awareness about the cost of resources for our economy 1 1 0.30 0.30 0.60
30 Lack of awareness about the impacts of consumption 4 2 0.30 0.80 1.10
31 Life quality perceptiond 3 3 3 3 0.00 0.80 0.80
32 Non-consumerist leisure activities 1 1 1 1 + 0.00 0.05 0.05
33 Productive leisure activities 1 1 1 1 + 0.00 0.05 0.05
34 Social awareness 0 0 + 0.70 0.00 0.70
35 Social resignation 0 0 0 0 0.80 0.00 0.80
36 Use of private transport 0 0 0 0 1.50 1.00 2.50
37 Use of public transport 0 0 0 0 + 2.50 0.50 3.00

Group 4: local competencies

38 Business opening times 0 0 0 0 n.a. 1.00 0.00 1.00
39 Complex and slow procedures 0 0 0 0 1.50 0.00 1.50
40 Lack of political class conviction 4 3 3 4 1.50 2.00 3.50
41 Local policy will 3 3 + 3.60 1.30 4.90
42 Mobility strategy 3 3 4 4 + 4.25 3.50 7.75
43 Shared ownership 0 0 0 0 n.a. 0.60 0.00 0.60
44 No local strategy 4 3 2.00 1.00 3.00
45 Non-transcendence of scientic studies 3 2 2 3 1.60 0.50 2.10
46 Votes 0 0 0 0 n.a. 0.70 0.00 0.70

Group 5: economy
47 Competitiveness 2 2 3 3 + 1.00 1.70 2.70
48 Economic crisis 4 3 4 3.90 2.30 6.20
49 Labour productivity 1 1 3 3 + 0.00 0.15 0.15
50 Local private economic interests 1 1 3 3 0.50 1.00 1.50
51 Protable investments 4 4 2 4 + 1.00 2.90 3.90
52 Purchase power 4 4 2 4 + 1.40 2.20 3.60
53 Short-circuit economies 0 0 0 0 + 2.00 0.00 2.00
54 Vulnerability of the system 3 2 3 3 0.00 3.00 3.00
38 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840

Table A1 (Continued)

Id Variable Scenariosa MM.O.b Networkc

A B C B&C O I Ct

Group 6: life basics

55 Culture & knowledge 1 1 1 1 + 0.00 0.40 0.40
56 Food 1 1 2 2 + 0.00 0.20 0.20
57 Human safety & protection 1 1 2 2 + 0.00 0.70 0.70
58 Hygiene 1 1 1 1 + 0.00 0.07 0.07
59 Learning 1 1 1 1 + 0.00 0.03 0.03
60 Leisure & fun 1 1 3 3 + 1.00 0.20 1.20
61 Life & survival 1 1 3 3 + 1.00 0.30 1.30
62 Personal development 1 1 2 2 + 1.00 0.10 1.10

Group 7: governance
63 Economic incentives 3 4 4 + 3.20 3.10 6.30
64 Electric network toll 0 0 0 0 n.a. 0.60 1.00 1.60
65 Energy/climate policies/regulations 3 3 + 5.85 3.25 9.10
66 Environmental responsibility 3 3 3 4 + 2.40 0.70 3.10
67 EU 20/20/20 objective 4 4 4 4 + 0.00 2.90 2.90
68 Law on balance netoe 0 0 0 0 + 0.70 1.00 1.70
69 National strategy on energy sources 0 0 0 0 n.a. 0.80 0.00 0.80
70 Public budget 4 4 2 4 + 0.00 1.00 1.00
71 Regional policy initiative 2 2 4 4 + 1.50 2.10 3.60

Group 8: supply & demand

72 Energy efciency 2 4 4 4 + 1.20 5.40 6.60
73 Energy price (households) 2 3 3 6.38 6.75 13.13
74 Energy price (industry) 2 2 4 4 0.90 1.00 1.90
75 Energy use in households 4 4 4 4 4.85 5.40 10.25
76 Energy use in industry 0 0 0 0 0.50 0.00 0.50
77 Energy use in municipal services 2 2 4 4 0.25 0.50 0.75
78 Energy use in transport 4 4 4 4 2.20 3.70 5.90
79 Energy used in non-material resources 1 1 3 3 0.00 0.25 0.25
80 Green/alternative energy demand 4 4 4 4 + 1.80 1.30 3.10

Group 9: energy business

81 Energy lobbies 0 0 0 0 7.55 0.60 8.15
82 Energy security 1 1 1 1 + 0.80 0.60 1.40
83 Green/alternative energy availability 4 3 1 3 + 2.95 1.00 3.95
84 Networks/infrastructures 1 1 1 1 + 2.30 0.50 2.80
85 Specialised technical energy services 0 0 0 0 + 1.60 0.00 1.60
86 Technical inspections 0 0 0 0 n.a. 0.50 0.00 0.50
a
Shaded cells represent the variables whose value is xed for the generation of each of the scenarios (see Table 2).
b
M.O.: Sustainable Management Objectives: +, a positive change is desirable; , a negative change is desirable; n.a., not
applicable.
c
Network characteristics: O, out-degree; I, in-degree, Ct, centrality.
d
Perceived quality of life is expressed in the negative sense, i.e. perceived worsening in quality of life.
e
This Spanish draft law, with provisions which are widely applied in other European countries, would benet particular
photovoltaic solar energy producers by allowing them to dump the excess onto the electricity infrastructure network and
consume from it during night or low energy production hours.

References

Adams, F., Aizawa, K., 2010a. Cognitive processes. In: The Bounds of Cognition. Wiley-Blackwell, pp. 5775.
Adams, F., Aizawa, K., 2010b. Introduction. In: The Bounds of Cognition. Wiley-Blackwell, pp. 115.
Adger, W.N., Barnett, J., Brown, K., Marshall, N., OBrien, K., 2012. Cultural dimensions of climate change impacts and
adaptation. Nat. Clim. Change 3, 112117.
Ahern, J., 2011. From fail-safe to safe-to-fail: sustainability and resilience in the new urban world. Landsc. Urban Plan. 100,
341343.
Amer, M., Daim, T.U., Jetter, A., 2013. A review of scenario planning. Futures 46, 2340,
http://dx.doi.org/10.1016/j.futures.2012.10.003.
Axelrod, R.M., 1976. Structure of Decision: The Cognitive Maps of Political Elites. Princeton University Press, Princeton, NJ.
Bachhofer, M., Wildenberg, M., 2009. FCmapper Software, http://fcmappers.net (accessed 18.12.13).
 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840 39

Barthel, S., Parker, J., Ernstson, H., 2013. Food and green space in cities: a resilience lens on gardens and urban environmental
movements. Urban Studies, http://dx.doi.org/10.1177/0042098012472744 (in press).
Beratan, K.K., 2007. A cognition-based view of decision processes in complex socialecological systems. Ecol. Soc. 12.
Berkes, F., Colding, J., Folke, C. (Eds.), 2003. Navigating SocialEcological Systems: Building Resilience for Complexity and
Change. Cambridge University Press, Cambridge.
BG, 2013. Udalmap Municipal Information System of the Basque Country, Basque Government,
http://www.ogasun.ejgv.euskadi.net/r51-udalmape/en/contenidos/informacion/udalmap/en udalmap/udalmap.html
(accessed 08.01.14).
Chelleri, L., Olazabal, M., 2012. Findings and nal remarks. In: Chelleri, L., Olazabal, M. (Eds.), Multidisciplinary Perspectives on
Urban Resilience: A Workshop Report. Basque Centre for Climate Change (BC3), Bilbao, Spain.
Colding, J., Barthel, S., 2013. The potential of Urban Green Commons in the resilience building of cities. Ecol. Econ. 86,
156166, http://dx.doi.org/10.1016/j.ecolecon.2012.10.016.
Creswell, J.W., 2003. Research Design: Qualitative, Quantitative, and Mixed Method Approaches. Sage Publications, Thousand
Oaks, CA.
Cumming, G.S., Barnes, G., Perz, S., Schmink, M., Sieving, K.E., Southworth, J., Binford, M., Holt, R.D., Stickler, C., Van Holt, T.,
2005. An exploratory framework for the empirical measurement of resilience. Ecosystems 8, 975987,
http://dx.doi.org/10.1007/s10021-005-0129-z.
Ernstson, H., van der Leeuw, S.E., Redman, C.L., Meffert, D.J., Davis, G., Alfsen, C., Elmqvist, T., 2010. Urban transitions: on urban
resilience and human-dominated ecosystems. AMBIO: J. Hum. Environ. 39, 531545,
http://dx.doi.org/10.1007/s13280-010-0081-9.
Fairweather, J., 2010. Farmer models of socio-ecologic systems: application of causal mapping across multiple locations. Ecol.
Model. 221, 555562, http://dx.doi.org/10.1016/j.ecolmodel.2009.10.026.
Folke, C., 2006. Resilience: the emergence of a perspective for socialecological systems analyses. Glob. Environ. Change
Hum. Policy Dimens. 16, 253267, http://dx.doi.org/10.1016/j.gloenvcha.2006.04.002.
Folke, C., Carpenter, S., Elmqvist, T., Gunderson, L., Holling, C., Walker, B., Bengtsson, J., Berkes, F., Colding, J., Danell, K.,
Falkenmark, M., Gordon, L., Kasperson, R., Kautsky, N., Kinzig, A., Levin, S., Mler, K.-G., Moberg, F., Ohlsson, L., Olsson, P.,
Ostrom, E., Reid, W., Rockstrm, J., Savenije, H., Svedin, U., 2002. In: Science, I.I.C.f. (Ed.), Resilience and Sustainable
Development: Building Adaptive Capacity in a World of Transformations.
http://www.resalliance.org/les/1144440669 resilience and sustainable development.pdf (accessed 24.03.14).
Folke, C., Carpenter, S.R., Walker, B., Scheffer, M., Chapin, T., Rockstrm, J., 2010. Resilience thinking: integrating resilience,
adaptability and transformability. Ecol. Soc. 15, 20.
Galarraga, I., Abadie, L.M., BC3 Working Paper Series 2014-06 2014. The Use of Bonus-Malus Schemes for Promoting
Energy-Efcient Household Appliances: A Case Study for Spain. BC3 Basque Centre for Climate Change, Bilbao.
Geels, F.W., 2011. The multi-level perspective on sustainability transitions: responses to seven criticisms. Environ. Innov. Soc.
Transit. 1, 2440.
Giordano, R., Vurro, M., 2010. Fuzzy Cognitive Map to support conict analysis in drought management. In: Glykas, M. (Ed.),
Fuzzy Cognitive Maps: Advances in Theory, Methodologies, Tools and Applications. Springer, Berlin Heidelberg, pp.
403423.
Glykas, M. (Ed.), 2010. Fuzzy Cognitive Maps: Advances in Theory, Methodologies, Tools and Applications. Springer, Berlin
Heidelberg.
Gray, S.R.J., Gagnon, A.S., Gray, S.A., ODwyer, B., OMahony, C., Muir, D., Devoy, R.J.N., Falaleeva, M., Gault, J., 2014. Are coastal
managers detecting the problem? Assessing stakeholder perception of climate vulnerability using Fuzzy Cognitive
Mapping. Ocean Coast. Manage. 94, 7489, http://dx.doi.org/10.1016/j.ocecoaman.2013.11.008.
Groumpos, P.P., 2010. Fuzzy Cognitive Maps: basic theories and their application to complex systems. In: Glykas, M. (Ed.),
Fuzzy Cognitive Maps: Advances in Theory, Methodologies, Tools and Applications. Springer, Berlin Heidelberg, pp. 1738.
Habib, F., Shokoohi, A., 2009. Classication and resolving urban problems by means of fuzzy approach. World Acad. Sci. Eng.
Technol. 60, 894901.
Hobbs, B.F., Ludsin, S.A., Knight, R.L., Ryan, P.A., Biberhofer, J., Ciborowski, J.J.H., 2002. Fuzzy cognitive mapping as a tool to
dene management objectives for complex ecosystems. Ecol. Appl. 12, 15481565,
http://dx.doi.org/10.1890/1051-0761(2002)012[1548:fcmaat]2.0.co;2.
Holling, C.S., 2001. Understanding the complexity of economic, ecological, and social systems. Ecosystems 4, 390405.
IEA, 2008. World Energy Outlook 2008. International Energy Agency, Paris,
http://www.worldenergyoutlook.org/media/weowebsite/2008-1994/WEO2008.pdf (accessed 11.11.13).
Isak, K.G.Q., Wildenberg, M., Adamescu, C.M., Skov, F., De Blust, G., Internal working paper 2009. Manual for Applying Fuzzy
Cognitive Mapping Experiences from ALTER-Net. ALTER-Net: A Long-Term Biodiversity, Ecosystem and Awareness
Research Network.
Janssen, M.A., Bodin, O., Anderies, J.M., Elmqvist, T., Ernstson, H., McAllister, R.R.J., Olsson, P., Ryan, P., 2006. Toward a network
perspective of the study of resilience in socialecological systems. Ecol. Soc. 11, 115.
Jetter, A., Schweinfort, W., 2011. Building scenarios with Fuzzy Cognitive Maps: an exploratory study of solar energy. Futures
43, 5266.
Jetter, A.J., Kok, K., 2014. Fuzzy Cognitive Maps for futures studies a methodological assessment of concepts and methods.
Futures, http://dx.doi.org/10.1016/j.futures.2014.05.002 (in press).
Kafetzis, A., McRoberts, N., Mouratiadou, I., 2010. Using Fuzzy Cognitive Maps to support the analysis of stakeholders views of
water reosurce use and water quality policy. In: Glykas, M. (Ed.), Fuzzy Cognitive Maps: Advances in Theory,
Methodologies, Tools and Applications. Springer, Berlin Heidelberg, pp. 383402.
Kok, K., 2009. The potential of Fuzzy Cognitive Maps for semi-quantitative scenario development, with an example from
Brazil. Glob. Environ. Change 19, 122133.
Kok, K., van Vliet, M., 2011. Using a participatory scenario development toolbox: added values and impact on quality of
scenarios. J. Water Clim. Change 2, 87105, http://dx.doi.org/10.2166/wcc.2011.032.
40 M. Olazabal, U. Pascual / Environmental Innovation and Societal Transitions 18 (2016) 1840

Kontogianni, A.D., Papageorgiou, E.I., Tourkolias, C., 2012. How do you perceive environmental change? Fuzzy Cognitive
Mapping informing stakeholder analysis for environmental policy making and non-market valuation. Appl. Soft Comput.
12, 37253735, http://dx.doi.org/10.1016/j.asoc.2012.05.003.
Kosko, B., 1986. Fuzzy cognitive maps. Int. J. Man-Mach. Stud. 24, 6575.
Langeld-Smith, K., 1992. Exploring the need for a shared cognitive map. J. Manage. Stud. 29, 349368,
http://dx.doi.org/10.1111/j.1467-6486.1992.tb00669.x.
Lewandowsky, S., Little, D., Kalish, M.L., Durso, F.T., 2007. Knowledge and expertise. In: Nickerson, R.S., Dumais, S.T.,
Lewandowsky, S., Perfect, T.J. (Eds.), Handbook of Applied Cognition. John Wiley & Sons Ltd, pp. 83109.
Loorbach, D., Rotmans, J., 2010. The practice of transition management: examples and lessons from four distinct cases. Futures
42, 237246.
Lopolito, A., Nardone, G., Prosperi, M., Sisto, R., Stasi, A., 2011. Modeling the bio-renery industry in rural areas: a participatory
approach for policy options comparison. Ecol. Econ. 72, 1827, http://dx.doi.org/10.1016/j.ecolecon.2011.09.010.
Madlener, R., Sunak, Y., 2011. Impacts of urbanization on urban structures and energy demand: what can we learn for urban
energy planning and urbanization management? Sustain. Cities Soc. 1, 4553, http://dx.doi.org/10.1016/j.scs.2010.08.006.
McCormick, K., Anderberg, S., Coenen, L., Neij, L., 2013. Advancing sustainable urban transformation. J. Clean. Prod. 50, 111,
http://dx.doi.org/10.1016/j.jclepro.2013.01.003.
Murungweni, C., van Wijk, M.T., Andersson, J.A., Smaling, E.M.A., Giller, K.E., 2011. Application of fuzzy cognitive mapping in
livelihood vulnerability analysis. Ecol. Soc. 16, http://dx.doi.org/10.5751/es-04393-160408.
Naess, P., Vogel, N., 2012. Sustainable urban development and the multi-level transition perspective. Environ. Innov. Soc.
Transit. 4, 3650.
Nelson, D.R., Adger, W.N., Brown, K., 2007. Adaptation to environmental change: contributions of a resilience framework.
Annu. Rev. Environ. Resour. 32, 395419, http://dx.doi.org/10.1146/annurev.energy.32.051807.090348.
Nevens, F., Frantzeskaki, N., Gorissen, L., Loorbach, D., 2013. Urban Transition Labs: co-creating transformative action for
sustainable cities. J. Clean. Prod. 50, 111122, http://dx.doi.org/10.1016/j.jclepro.2012.12.001.
Olazabal, M., Pascual, U., 2013. Identifying Social Determinants of Urban Low Carbon Transitions: The Case of Energy
Transition in Bilbao, Basque Country, BC3 Working Paper Series 2013-11.
Ozesmi, U., Ozesmi, S.L., 2004. Ecological models based on peoples knowledge: a multi-step fuzzy cognitive mapping
approach. Ecol. Model. 176, 4364, http://dx.doi.org/10.1016/j.ecolmodel.2003.10.027.
Pelling, M., Manuel-Navarrete, D., 2011. From resilience to transformation: the adaptive cycle in two Mexican urban centers.
Ecol. Soc. 16.
Pickett, S.T.A., Cadenasso, M.L., Grove, J.M., 2004. Resilient cities: meaning, models, and metaphor for integrating the
ecological, socio-economic, and planning realms. Landsc. Urban Plan. 69, 369384,
http://dx.doi.org/10.1016/j.landurpbplan.2003.10.035.
Reckien, D., 2014. Weather extremes and street life in India implications of Fuzzy Cognitive Mapping as a new tool for
semi-quantitative impact assessment and ranking of adaptation measures. Glob. Environ. Change Hum. Policy Dimens. 26,
113.
Reckien, D., Wildenberg, M., Bachhofer, M., 2013. Subjective realities of climate change: how mental maps of impacts deliver
socially sensible adaptation options. Sustain. Sci. 8, 159172, http://dx.doi.org/10.1007/s11625-012-0179-z.
Rotmans, J., Kemp, R., Asselt, M.V., Geels, F.W., Verbong, G.P.J., Molendijk, K., Notten, P.V., External Report, ICIS Working Paper
N I01-E001 2000. Transitions & Transition Management: The Case for a Low Emission Energy Supply. ICIS, Maastricht,
http://www.tue.nl/en/publication/ep/p/d/ep-uid/147263/ (accessed 11.11.13).
Sieck, W.R., Klein, G., Durso, F.T., 2007. Decision-making. In: Nickerson, R.S., Dumais, S.T., Lewandowsky, S., Perfect, T.J. (Eds.),
Handbook of Applied Cognition. John Wiley & Sons Ltd, pp. 195217.
Stylios, C.D., Georgopoulos, V.C., Groumpos, P.P.C., 1997. The use of fuzzy cognitive maps in modeling systems. In: IEEE
Mediterranean Conference on Control Systems, http://med.ee.nd.edu/MED5-1997/PAPERS/067/067.PDF
Tolman, E.C., 1948. Cognitive maps in rats and men. Psychol. Rev. 55, 189208.
Van der Brugge, R., Van Raak, R., 2007. Facing the adaptive management challenge: insights from transition management.
Ecol. Soc. 12.
van Vliet, M., Kok, K., Veldkamp, T., 2010. Linking stakeholders and modellers in scenario studies: the use of Fuzzy Cognitive
Maps as a communication and learning tool. Futures 42, 114.
Vanwindekens, F.M., Stilmant, D., Baret, P.V., 2013. Development of a broadened cognitive mapping approach for analysing
systems of practices in socialecological systems. Ecol. Model. 250, 352362,
http://dx.doi.org/10.1016/j.ecolmodel.2012.11.023.
Walker, B., Gunderson, L., Kinzig, A., Folke, C., Carpenter, S., Schultz, L., 2006. A handful of heuristics and some propositions for
understanding resilience in socialecological systems. Ecol. Soc. 11.
Walker, B., Holling, C.S., Carpenter, S.R., Kinzig, A., 2004. Resilience, adaptability and transformability in socialecological
systems. Ecol. Soc. 9.
Waters, J.J., 2012. Reconsidering resilience in rapidly urbanising areas. In: Chelleri, L., Olazabal, M. (Eds.), Multidisciplinary
Perspectives on Urban Resilience: A Workshop Report. Basque Centre for Climate Change (BC3), Bilbao, Spain.
Wildenberg, M., Bachhofer, M., Adamescu, M., De Blust, G., Diaz-Delgadod, R., Isak, K., Skov, F., Varjopuro, R.C., 2010. Linking
thoughts to ows fuzzy cognitive mapping as tool for integrated landscape modelling. In: Proceedings of the
International Conference on Integrative Landscape Modelling Symposcience LANDMOD2010, Montpellier, France
http://www.symposcience.org/exl-doc/colloque/ART-00002419.pdf
Zhang, H., Song, J., Su, C., He, M., 2013. Human attitudes in environmental management: Fuzzy Cognitive Maps and policy
option simulations analysis for a coal-mine ecosystem in China. J. Environ. Manage. 115, 227234,
http://dx.doi.org/10.1016/j.jenvman.2012.09.032.