Critical Volcanology? Thinking Holistically About Risk and Uncertainty

Bulletin of Volcanology (2019) 81: 20
https://doi.org/10.1007/s00445-019-1279-8
FORUM CONTRIBUTION
Critical volcanology? Thinking holistically about risk and uncertainty

Amy Donovan 1
Received: 21 August 2018 / Accepted: 19 February 2019 / Published online: 7 March 2019
# International Association of Volcanology & Chemistry of the Earth's Interior 2019
Abstract
This paper tackles two related issues in dealing with so-called Bwicked^ problems: the challenge for scientists wishing to provide
useful policy advice whilst maintaining scientific integrity and the challenge of integrating multiple disciplines across the social
and physical sciences. It focusses particularly on the problems associated with volcanic risks—specifically those problems that
threaten to Bpoliticise^ volcanology via the role of scientists in advising authorities. The Sendai Framework for Disaster Risk
Reduction 2015–2030 calls for a greater role for science in reducing risk. It conceives of science very broadly—incorporating
social, physical, medical, economic, agricultural and behavioural sciences into the policy process. This presents an exciting
opportunity; yet, there are difficulties in approaching the science-policy encounter in the context of risk, especially under high
uncertainty and involving multiple disciplines. The paper reviews literature in volcanic risk assessment, science studies, and
policy studies to suggest that risk requires critical, reflexive, transparent, and discursive management through the availability of
spaces in which academics of all disciplines, policymakers, and stakeholders can openly discuss method, meaning, and inter-
pretation of evidence.
Keywords Volcanic risk . Interdisciplinary volcanology . Philosophy of risk . Risk assessment . Risk and uncertainty
Introduction: the particular challenge initially review the history of volcanological crises to illustrate
of volcanic risk these challenges. It then seeks to engage with wider literature
from political science and philosophy to bring experience
As volcanological science has improved, the prospect of fore- from other domains to the problem, so as to inform the ongo-
casting (and in a few cases, even predicting) volcanic erup- ing debate in volcanology about the role and remit of scientific
tions is being realised—and exciting new monitoring methods advice in decision-making with a rigorous philosophical, pol-
and statistical methods have been developed to render this icy, and empirical evidence base from other sciences.
possible (Aspinall 2006; Aspinall et al. 2003; Johnson et al.
2018; Lindsay et al. 2010; Marzocchi 2008; Marzocchi and
Woo 2009; Neri et al. 2008; Sandri et al. 2004; Sheldrake et al. Volcanic crises: the challenges
2017; Woo 2008). Additional methods and models have been
developed for long-term planning, particularly relevant for The infamous crisis in 1976 on Guadeloupe continues to be a
land use decisions—including hazard and risk maps (Calder key example that demonstrates the challenges of volcano fore-
et al. 2015; Thompson et al. 2015; Tonini et al. 2015). Yet, the casting (Hincks et al. 2014; Komorowski et al. 2015; Tazieff
policy context of volcanic eruptions continues to create chal- 1977). It showed both the potential value of scientific consensus
lenges for scientists, decision-makers, and populations (Aspinall and Cooke 1998) and the deep uncertainties that
(Marzocchi et al. 2012; Papale 2017). This forum paper will abound in volcano forecasting (Hincks et al. 2014) and resulted
in the use of belief-based probabilistic methods: in 1995, when
the Soufrière Hills Volcano awoke in Montserrat, expert elici-
Editorial responsibility: R. Cioni tation was utilised as a means of establishing consensus
(Aspinall and Cooke 1998; Aspinall 2006; Donovan et al.
* Amy Donovan 2012a, d). Elsewhere, from the eruption of Mount St Helens
[email protected]
onwards, the USGS had begun to use event trees as a concep-
1
Department of Geography, University of Cambridge, tual framework within which the likely eruptive scenarios could
Cambridge, UK be established. These have since been used in different ways in
20 Page 2 of 20 Bull Volcanol (2019) 81: 20
numerous crises all over the world (Newhall and Hoblitt 2002; 2008), and diverse field sites with different conditions also
Newhall and Pallister 2015). require adapted approaches—volcanoes have individual
There are substantial geographical variations in how volca- characteristics. The model that works best on one volcano
nic risk is approached, however. In particular, the institutional may work less well on another, because it deals with topogra-
context of scientific advice in volcanic crises is very varied— phy in more appropriate way, for example. The configuration
not only in the variety of roles that scientists are expected to fill, of monitoring networks, similarly, is often dictated by topog-
but also in the availability of expertise and resources for volca- raphy and accessibility. The way that risk (or, more common-
no monitoring and risk management (Loughlin et al. 2015). ly, hazard) is approached also varies considerably: most ob-
Development is a key factor in whether or not countries prepare servatories base their assessments and advice on monitoring
for volcanic eruptions or monitor volcanoes: prioritising volca- data and informal discussion between scientists, not on statis-
no monitoring when people are struggling to find adequate tical methods—though there are drives to enhance this capac-
food, education or healthcare is politically challenging (Tilling ity (Sparks and Aspinall 2004). The most useful format of risk
2008). Volcano monitoring tends to become a concern either or hazard information for authorities also varies—govern-
because of a volcanic crisis or where a country reaches the point ments in some countries want probabilistic assessments, but
in development where it can afford to monitor its volcanoes— others would not know what to do with them and many are still
and even then, monitoring can be controversial, as in the USA dominantly response-focussed in their approach to disaster
when Bobby Jindal criticised the then President Obama for risk in general (e.g. Goitom et al. 2015). Connectivity be-
funding Bsomething called volcano monitoring^ in 2009 (note tween scientists, risk managers, and governments and popu-
1). The politics of particular places can also interfere with the lations also varies (Armijos et al. 2017)—as does the role and
availability of resources and expertise: if volcanology is not attitude of the news media (Harris 2015). All of these factors
funded, no one will study it. Typically, volcanic crises are need- bring localised complexities and nuances to the management
ed before there is significant investment, too—not least because of volcanic risk.
political timescales generally mean that there is little political In this discussion paper, the methods and models that are
capital in investing in volcano monitoring. If a politician is only discussed can essentially be applied in two different scenarios:
in power for 4 years, the probability of an eruption at a partic- volcanic eruptions (referred to as Bcrises^ here) and longer-
ular volcano within that timeframe is usually very low, and so, term (peace-time) planning. This paper is concerned with sci-
the personal-political cost-benefit analysis indicates that there entific advice to governments in volcanic contexts—which
are more socially acceptable policies to invest in. varies between these two scenarios (Fig. 1). While the key
Institutional responsibilities, and funding priorities, are al- questions for each scenario differ, there are a number of fac-
so highly variable in space and time. Scientists who are part of tors that they have in common—particularly a tension be-
a geological survey may be required to advise governments as tween what officials want to know, and what scientific knowl-
part of their work. Sometimes, this role extends to decision- edge is able to give them. While this tension is found across
making, as is the case in Indonesia (Surono et al. 2012): some environmental science, from the IPCC to air pollution to ge-
jurisdictions and political cultures may regard volcanologists netically modified crops (Hastrup 2012; Hinchliffe 2001;
as better qualified to decide on actions because, logically, Levidow 2001; Nowotny 2003; Wynne 2001; Stirling 2012),
volcanologists run the models that define the hazard maps, there are several factors that enhance it in volcanology, includ-
and the hazard maps define the safe zones, and because vol- ing the high uncertainty, the high stakes of a crisis in a popu-
canologists are best placed to interpret monitoring data. Thus, lous area, and the rarity of the situation relative to other envi-
in some contexts, volcanologists are expected to participate in ronmental hazards. Furthermore, volcanic Bcrises^ may last
decision-making. This may be the result of institutional histo- for years with a sustained need for scientific input to
ries, or of the history of crises; it may reflect a desire that decision-making (Armijos et al. 2017; Donovan and
decisions are based solely on scientific evidence. It may also Oppenheimer 2014).
be the result of the use of linked alert level systems—systems
that tie alert levels to a particular decision, and give the task of The practice of advice in volcanology
assigning alerts to scientists. Such systems may even serve to
protect scientists, if they have been devised by civil protection In many countries that monitor volcanoes, the primary modes
or government, because they provide protocols that respond to of decision-making are observatory-based, derived from
particular levels of volcanic activity. They may also be tailored meetings of observatory staff and the decisions of chief
to particular volcanic and/or social contexts (Fearnley and scientists—often in collaboration with civil protection agen-
Beaven 2018). cies (e.g. Donovan and Oppenheimer 2015a; see also Surono
Another geographical factor is the geography of science et al. 2012; Fearnley and Beaven 2018). These may be largely
itself. Different institutions have different approaches to sci- deterministic: they are based on relating physical data about
entific knowledge (Livingstone 2003; Powell 2007; Hulme the Bcause^ to the possible Beffects^. They may also include
Bull Volcanol (2019) 81: 20 Page 3 of 20 20
the communication of a scientific warning to authorities—is

absolutely critical in the management of volcanic risk. It is in
this context that problems may arise and that scientists must
explain a complex and uncertain situation to non-scientists
(Marzocchi et al. 2012). In some cases, alert levels or crisis
hazard maps may be used as boundary objects in this commu-
nication process (Donovan et al. 2018; Fearnley and Beaven
2018). Occasionally, a probabilistic assessment may be done
(Newhall and Hoblitt 2002; Newhall and Pallister 2015;
Wadge et al. 2009). There is no global standard because of
geography—political processes, resources and the roles of
scientists vary between contexts (Donovan and Oppenheimer
2015b), and the characteristics of governments and popula-
tions are highly dynamic and variable. However, there have
been drives from Europe and the USA in particular to move
towards probabilistic assessments of hazard and risk (Aspinall
et al. 2003; Baxter et al. 2008; Neri et al. 2008; Newhall and
Hoblitt 2002; Papale 2017; Sparks and Aspinall 2004).
In general, risk assessments are carried out and then feed
into wider risk management (Renn 2008). In the case of vol-
canology as in other disciplines (Stirling 2007), the depen-
dence on scientific advice (Brisk assessment^) in determining
risk management can effectively conflate the two, with
decision-makers (and the public) then Bblaming^ the science
for their decisions (Donovan et al. 2012a). Once scientists
Fig. 1 The two scenarios that scientists typically encounter. The first communicate a warning, whether publicly or strictly to author-
scenario is crisis-level decision-making. The second is longer-term plan- ities, they lose control of their knowledge. While scientists
ning. Both of these scenarios raise critical questions for scientists, may or may not be involved in decision-making (often they
policymakers, and populations. The key questions for scenario 1 are taken
from Colin Wilson’s helpful summary (Wilson 2017) have to be in order for hazard zones to be established), scien-
tific knowledge travels. This is the key problem with the ar-
gument that Bscientists should only do science^ in volcanic
some level of probabilistic thinking on the part of the crises: it is a false start. The role of the scientists can be lim-
scientists—even very informally, and may involve alert level ited, but the role of the science goes further, as decision-
systems or protocols, some of which are tied to particular makers want to know what the science means, and the public
policy actions (such as evacuation) (Fearnley and Beaven (and media) want to know on what basis their lives are being
2018; Potter et al. 2014; Winson et al. 2014). In a crisis situ- turned upside down (Aspinall and Sparks 2004; Donovan and
ation, scientists must analyse uncertain data and interpret the Oppenheimer 2014). Even if all the scientists say is that the
likely implications for the state of the volcanic system—and volcano is showing signs of unrest, as soon as that is commu-
its possible evolution towards eruption. This may involve seis- nicated beyond the scientific community, it becomes
mic, ground deformation and gas data, alongside other politicised by definition—not because the scientists have done
methods if the observatory is well equipped (Donovan et al. a bad job or overstepped their remit, but because science is
2012c). It may, as in many cases, involve only minimal seis- then interpreted and evaluated by non-scientists (Miller
mic data. This is the first key issue that scientists face: 1986). Because of this, Bscientists only doing science^ is not
interpreting uncertain evidence that may be apparently contra- a viable solution to the problems of volcanic crises—attractive
dictory, may show patterns that are similar or different to past though it is. Scientists also have to be able to engage with
events, may be limited by damaged equipment or poor re- other interpreters of their scientific reasoning—especially as
sources and may also involve disagreement between scien- publics and policymakers become more science-literate
tists. This interpretation may have to be made relatively quick- (Weigold 2001; Davies 2008; Bauer 2009)—and it has been
ly, so that authorities are informed of the challenges. Scientists argued that science communication is an ethical imperative in
must then communicate with local authorities that the volcano some cases (Friedman 2008), particularly as scientists are
is restless and showing signs of activity. This may take the highly trusted (Haynes et al. 2007). Scientists may not be
form of a report that is issued daily or even hourly in a crisis. It making decisions—something outside of their training,
may simply involve a phonecall to civil defence. This step— often—but the knowledge they produce will likely be the
basis (largely) for any political decision. Scientific knowledge The next sections explore these issues in a broader context,
travels in any scientific advisory situation (Bijker et al. 2009; thinking about the production of scientific advice (whether a
Jasanoff and Martello 2004; Levidow and Carr 2007; Owens phone call, a report or a presentation in a meeting) in other
2000; Owens 2015). When there is a great deal of uncertainty contexts that have been studied in detail. It looks at the history
surrounding that knowledge, the potential for misinterpreta- of science advice—something that has really only developed
tion is increased. This applies to probabilities, too— since the mid to late twentieth century in most contexts—to
probability will not prevent the challenges that arise from back up its arguments. It also engages with the philosophy
knowledge that travels—and may compound the problem be- literature in detail, because the provision of scientific advice
cause public and even scientific understanding of probability is not fundamentally what the scientific method was devised
is highly variable (Gigerenzer et al. 2005; Donovan et al. to do—and in reviewing this literature, it argues that too often,
2015). Even in longer-term circumstances (scenario 2, above), the philosophy of science becomes conflated with the philos-
where volcanologists are engaged in preparing hazard maps to ophy of risk in ways that are misleading. In particular, while
denote the most at-risk areas for land-use planning or for the scientists tend to work to reduce the subjectivity and impact of
establishment of evacuation zones, the use of that knowledge the social context on scientific knowledge production, reduc-
by authorities also means that knowledge is travelling. Thus, ing these aspects in risk assessment (note 2) is challenging. In
scientists in other domains that regularly engage with policy such a context, this paper suggests that involving a wider
have called for much greater research into the complexity of disciplinary pool could aid the process. Note that this paper
science-policy encounters (Dessai and Hulme 2004; Hulme contains an appendix that deals with issues of definition—
2008; Lövbrand et al. 2015; Sutherland et al. 2012). both in its use of terminology of risk, and in some key ideas
This paper deals with the issue of travelling knowledge about the philosophy of science and social science.
in the context of volcanic hazard and risk management. It
recognises that there is huge variability globally in terms of Context: subjective advice and the social contract
how, precisely, the roles of volcanologists in crises are
prescribed by governments—and in use of alert level sys- There is a widespread acceptance that Bwe have no better way
tems or similar—but suggests that there are some underly- of producing knowledge about the natural world than by do-
ing themes that require consideration concerning the nature ing science^(Douglas 2009), yet there are still, and perhaps
of volcanic hazard and risk assessment itself. This is not to increasingly, debates about the credibility of scientific knowl-
argue that scientists should therefore get involved in poli- edge, particularly in the context of risk—whether from climate
tics or the making of decisions in a more active way! It is change, nanotechnology or natural hazards (Donovan et al.
simply to suggest that their knowledge is inherently in- 2012a, b, d; Doubleday 2007; Gallopín et al. 2001; Hulme
volved, whether or not they are themselves around a table 2009)—evidenced strikingly in the court cases in Italy follow-
with policymakers—and sometimes, avoiding the misuse ing the 2009 L’Aquila earthquake (Donovan and
of knowledge may require them to sit with decision-makers Oppenheimer 2015b; Hall 2011): in this case, scientists had
in an advisory capacity. Thus, the line inevitably becomes stated that there was little chance of an earthquake, but their
blurred between science and policy—not because of scien- view was misinterpreted and miscommunicated. This mis-
tists, necessarily, but because science travels and is communication took place in a highly charged social, political
questioned throughout the process. In some circumstances, and cultural context that significantly exacerbated the situa-
scientists are involved in decision-making through alert tion. The backlash against the scientists, who did not correct
levels that are tied to actions or through involvement on a the miscommunication of their advice, was striking: it dem-
risk advisory committee that is composed of both scientists onstrates the vulnerability of scientific knowledge once it
and policymakers. In many ways, the latter is an opportu- leaves the scientific community—and that that vulnerability
nity for scientists to explain their science and ensure that it of knowledge can rebound onto the scientists who produced it.
is understood and correctly interpreted in the decision- Transcripts of the meeting of the Commissione Grandi Rischi
making process, without necessarily making any decision on 31 March 2009 show clearly that the scientists had sug-
themselves. This situation requires careful consideration gested a low chance of an earthquake—rather than none at
and input from a broad disciplinary community that can all—and that this was therefore miscommunicated (Benessia
consider the nature of knowledge, of encounters with other and De Marchi 2017; Donovan and Oppenheimer 2015b;
cultures and of politics. These are themes that have been Scolobig et al. 2014). Suggesting a low probability of an event
dealt with in detail in other domains where risk assessment is insufficient, in highly charged contexts, to protect scientists
and management have typically been artificially separated. if the social, cultural and political contexts of the advice are
Indeed, Sheila Jasanoff argues that this separation is a false not carefully, consciously managed.
one because of the ways in which knowledge is produced The science-policy encounter is a fundamentally social en-
(Jasanoff 1993). counter. One of the biggest challenges for scientists involved
in providing advice is that advice is subjective, inevitably: one reclaimed for situated (place-specific) knowledges that do
scientist may not emphasise the same results or ideas as an- not claim to be value-free—but this adds another layer of
other, and so scientific advice is rarely reproducible. It is also confusion about what is really meant by Bobjectivity^ in the
interpretative: it involves individuals and groups interpreting first place. Indeed, the historian of science Lorraine Daston
partial datasets to reach a useful conclusion. Scientific adviso- has identified multiple meanings in the history of science
ry committees or observatory staff meetings overcome some (Daston 1992; Daston 1994). Arguably, then, it would make
of this—they are intersubjective, in that they require numer- more sense to abandon this terminology, and instead focus on
ous experts to agree on a report or piece of advice. However, the different kinds of values (note 5) that can be introduced to
the production of reports and the reception that such reports the scientific process (often unconsciously) (Douglas 2000),
receive are also social activities (Owens 2015). Numerous and on invoking measures to manage them, including integri-
authors have suggested the inclusion of social science in pol- ty, humility and reflexivity (Jasanoff 2003; Jasanoff 2007;
icy advice (Pidgeon and Fischhoff 2011; Victor 2015; Wong- Stirling 2008).
Parodi et al. 2016) because it can provide reflexivity: aware- Recent years have witnessed fierce debate among philoso-
ness and consideration of the social processes that affect the phers about the Bvalue-free ideal^ in science: can science ever
advice. Andy Stirling argued that in advisory contexts, scien- be truly separated from social or individual values? (Betz
tists have to be aware of the plural nature of knowledge when 2013; Douglas 2009; Lacey 2005; Proctor 1991) If the scien-
talking about risk (Stirling 2010): risk, by definition, is some- tific community is genuinely separated from the broader com-
thing we do not know completely. Volcanologists are included munity as a result of its particular practices (Kuhn 1962), then
in this, even if they are only advising about hazards, because the only values that are important are epistemic values (those
as soon as the advice leaves the scientific domain, it is speak- that are associated with doing science—such as the selection
ing implicitly about vulnerable people—and is itself vulnera- of models—and which do not threaten the production of ro-
ble (Cutter 2003). When scientists provide advice to govern- bust knowledge). However, more recently, philosophers of
ments, they are participating in a social contract: they are science have recognised the importance of social and ethical
taking a role in government itself, through their knowledge values within the scientific process—and also the impact that
as it is used by government. social values can have on epistemic values. For example, a
scientist weighing up two models may regard simplicity as
Is risk assessment science? more important than accuracy, and this judgement may be
based on personal experience or even aesthetic preference
The increasing role of science in society has led to extensive (Douglas 2000; Douglas 2009). Similarly, a choice between
critiques of the scientific process, and a growing awareness of two methods is frequently based on epistemic values—for
the social aspects of science (Beck 1992; Latour 2004; example, different methods of expert elicitation have different
Shackley and Wynne 1996), as well as intense investigation positives and negatives (EFSA 2014).
into the role that scientists have in policy processes (Brown The demarcation of science from the social has long been a
2009; Jasanoff 1990; Owens 2015). These include studies of philosophical problem, and in the era of post-normal science
how scientists construct authority—for example, by making (Funtowicz and Ravetz 1992), in which science is needed to
claims about objectivity (note 3) (Aspinall 2010; Aspinall answer socially derived questions, science can no longer be
et al. 2003)—and of how publics test those claims (Jasanoff considered autonomous in selecting the questions it seeks to
2004a, b): this latter is a key issue in volcanology, where answer: governments increasingly require scientists to engage
publics frequently do test knowledge claims because those with social questions such as those related to risk (known as
claims affect their lives (note 4). Objectivity claims are partic- Bmode-2 science^) rather than blue-skies research (Bmode-1
ularly problematic (Jasanoff 2005). Thomas Kuhn argued that science^) (Gibbons et al. 1994). This point is underscored by
‘subjective’ is not the opposite of ‘objective’ (Kuhn 1977), yet Douglas (Douglas 2009), who argues that scientists funda-
there remains a deep mistrust of subjectivity in scientific cul- mentally have a moral obligation to consider the potential
tures (Aspinall 2012; Aspinall et al. 2003). Curtis argues that social and ethical implications of their work in the formulation
BScientists should therefore not be ashamed of subjectivity, of advice based on science. In the field of geohazards and
but we should strive to develop methods to quantify and climate science, this might involve awareness that the choice
sometimes to reduce its effects^ (Curtis 2012). Yet, the au- of a flood model in the production of a hazard map could
thority and indeed value of science is not always necessarily affect the insurance rates or property value of houses in the
decreased by subjectivity—the expert is a subject, valued for map region. If there is considerable difference between
their own personal skills and background, and their ability to models, and high uncertainty, then there is a moral obligation
select appropriate inputs and models depends on that subjec- to be very clear about that in providing advice to government.
tivity as well as being potentially limited by it. Hermansson As Douglas notes, too BHaving clear assessments of uncer-
(2012), for example, suggests that Bobjectivity^ could be tainty is always helpful, but the scientist must still decide that
the assessment is sufficiently accurate, and thus the need for There are many ways in which scientists actively seek to
values is not eliminable^ (p.85): putting numbers on risk and delimit their realm from that of policymakers—a process
uncertainty does not obliterate the need for social values in the known as boundary work (Gieryn 1983). Boundary work is
scientific process. Where the numbers on uncertainty are particularly difficult under conditions of high uncertainty: sci-
themselves uncertain, then the use of numerical estimates is entific knowledge can be demarcated much more effectively
potentially irresponsible and may be unethical (Porter 1996). (though still imperfectly) than uncertain judgement or risk,
To give an example, if scientists estimate that the probability because risk changes in space and time—it travels backwards
of an eruption is 46%, but give uncertainty bounds of 5% to and forwards between the scientists and policymakers and the
87%—and all these are subjective estimates—then population. It is dynamic and multidimensional, depending on
policymakers need to know all of the values (without empha- human and natural systems that are linked in complex ways.
sis on the spuriously accurate 46%), and how they were ob- Boundary work is difficult with risk because risk is not the
tained. It is arguable that in this case, the use of probability is equivalent of science—to assess risk probabilistically, scien-
misleading because of the high range of uncertainties—or that tists have to leave the confines of the science that they are
the probability should be expressed as a range or ball-park comfortable with (Castanos and Lomnitz 2002). This has been
figure, as is done by the UK Meteorological Office, for exam- discussed in detail in the context of seismic hazard assess-
ple (Stephens et al. 2012; Joslyn and Savelli 2010; Morss, ment, for example, as discussed in a recent review paper that
Demuth, and Lazo 2008). expresses some anxieties about belief-based probabilistic
These complex ethical challenges occur in part because the methods because they are beyond traditional seismology
kinds of questions that science is increasingly invited to re- (Mulargia et al. 2017): these methods continue to be contro-
spond to are posed not by scientists but by society (Gibbons versial among scientists because subjective probability does
et al. 1994), either via the impact agenda or because of critical not readily conform to the attributes of Popperian science.
needs—such as in a volcanic crisis. Traditionally, as Gaston Popper himself struggled with belief-based probabilities,
Bachelard noted, opinion and science are opposed to one discussing them at length in Logic of Scientific Discovery
another: imagination, opinion and psychology are Bobstacles^ and its Postscript (Popper 1956 (published 1982); Popper
to scientific endeavour (Bachelard 2002). However, questions 1959 (German original, 1935); Popper 1957). While any prob-
about risk inevitably involve all three of these—risk is an ability that is not 0 or 1 cannot be falsified, belief-based prob-
imagined concept that allows structured opinions about possi- abilities are also not readily testable because they are usually
ble futures to be contained (Hacking 1975; Hacking 1990; single-event probabilities. This does not mean that they are not
Jasanoff 1998; Jasanoff 1999). In spite of efforts to minimise useful, or that they are not technical, or that they are not a valid
the role of the social in risk assessment, the kinds of question approach to risk. They are all of those things, because risk
that are asked in such assessments are almost always itself takes us beyond Popperian science.
Bunscientific^ in the purest sense: they require an opinion Scientific knowledge about risk tends to be created both via
based on scientific ideas, knowledge and experience—and, the scientific method and via the social context in which the
often, a degree of imagination of possible futures knowledge is produced—science and social order are co-
(Funtowicz and Ravetz 1995; Gibbons et al. 1994). produced (Jasanoff 2004a, b). This Bco-production of
Risk assessment therefore complicates the problem of knowledge^ model is widely regarded as more useful than a
demarcation: it is especially difficult to clearly distinguish straightforward linear model where science is clearly demarcat-
science from its social context in the realm of risk assess- ed from policy and is an inert object that feeds into it. Indeed,
ment (Gieryn 1983). However, to fail to retain distinctive- such a model depends on a positivist (pre-Popperian) conception
ness for science threatens its authority—especially if that of science itself (note 7). The linear model is generally more
authority is constructed on things like objectivity claims appealing in theory (Bretton et al. 2015; Marzocchi et al.
(note 6). Indeed, Popper himself argues against claims of 2012), but is simply not evidenced in practice (Donovan and
authority for knowledge (Popper 1963). Furthermore, Oppenheimer 2014; Fischer 2000; Owens 2015), because
there is abundant evidence that claims of authority for knowledge is questioned and contested by its users. Again, this
knowledge do not form the basis by which knowledge is is not to say that scientists and policymakers do not have distinct
judged: both sociologists of science and social psycholo- roles in risk assessment and management (Giordano et al.
gists have shown that trust in science is not only about 2016)—but it is to say that the roles may not be so easily dis-
knowledge, but also about social judgements, such as per- tinguished in practice because the knowledge that they use
ceived motives (Donovan et al. 2018; Eiser et al. 2015; overlaps. Furthermore, the co-production model does not re-
Jasanoff 2005): again, science that is used in risk assess- quire that science is Bsocially constructed^: it is a middle ap-
ment is not necessarily judged in the same ways that the proach that acknowledges both the scientific method and some
scientific community would judge it. It has travelled into social or value-driven processes, consistent with critical realism
the social and political sphere. or pragmatism (see Appendix). In the context of risk assessment,
this is more emphatic because risk is an imagined, possible scientists because it is (i) never complete, (ii) subject to the
future, and Bprobability does not exist^ (De Finetti 1974). whims of the natural system, which may change over time
Thus, risk and science are not logically equivalent. and space, and (iii) not the same as inference—though frequent-
Theoretically, scientific knowledge can approximate a reality ly erroneously equated with it. In the use of subjective proba-
beyond the human (Kant and Guyer 1998). When scientific bilistic methods, for example, evidence cannot always be traced
evidence is applied to a judgement about risk, on the other very easily because the results incorporate inference. Science
hand, the framing of the risk question has an impact on the and scientific hazard and risk assessment involve judgement,
ways in which that evidence is used (Pidgeon et al. 2008). It is and this also raises issues concerning what constitutes evidence
therefore not logical to equate a philosophy of science that (Funtowicz and Ravetz 1995): judgement is difficult to control,
seeks to find out knowledge about the earth system with a but absolutely necessary. There is a level of Bgut feeling^ in-
philosophy of risk that assumes that risk (in an open volved, and this might be affected, for example, by the expert’s
system—note 8) has a Btrue^ and knowable probability, be- perception of those who they know will apply their advice. In
cause risk is a formulation of incomplete knowledge, by def- an ideal situation, the expert effectively communicates the view
inition: if we know that something is going to happen, it is no of the natural system that they believe is ‘objectively’ true
longer a risk—Bwe use probability only in default of (O’Hagan et al. 2006). Even in this case, however, the way in
certainty^ (Wittgenstein 1921). Not only is risk dynamic (it which such information is framed will be mediated through an
changes in time and space in non-linear ways), it is also a expert’s interpretation of the requirements of the decision-mak-
social process (Douglas and Wildavsky 1982). Risk is also er: the results of the elicitation or expert meeting have to be
inherently laden with values because it impacts and depends communicated to non-specialists effectively (Hajer, 2003).
upon people (Douglas 2000). In this respect, risk assessment Expert judgement has been used widely in hazard and risk
can never be completely free of social and ethical values: assessment (O’Hagan et al. 2006) as a form of evidence for
scientists have an obligation to communicate uncertainties evidence-based policy (Aspinall 2010; Aspinall 2006; Cornell
and the limits of probability very clearly because of how their 1968; Lindsay et al. 2010; Pate-Cornell 2002). For example,
knowledge is likely to be used (Fig. 2). some probabilistic assessments use expert elicitation methods
(Aspinall 2006; Bamber and Aspinall 2013; Cornell 1968).
These methods are useful because they allow experts to act as
The limitations of evidence and judgement knowledge-conduits and interpreters, and they can also assess
some of the uncertainty involved in the assessment. They are
An additional issue in any scientific advice is the nature of used to gain formalised quantitative assessments from scientists
expert judgement and evidence. Evidence is a problem for with differing views (Aspinall and Cooke 1998), but they have
Fig. 2 Different types and flows

of knowledge in volcanic crises
and scientific advice. The darker
box shows the domain of
scientists and social scientists—
note that fields deliberately over-
lap slightly to show that none of
these things takes place entirely in
a vacuum, because knowledge
travels and the responsibility to
explain it travels with it, and also
because of the geographical vari-
ation in responsibilities
also been put forward as a means of ‘increasing the objectivity’ Sheffield and Delphi), dependent on context and epistemic
in scientific advice (Aspinall 2012; Donovan et al. 2012b). In values, and involves several separate groups of experts to en-
essence, there is a common theme behind both of these aims: sure that the process has clearly defined limits and controls.
the claim that risk (or probability) is an object (note 9) that can Uncertainty is a fundamental problem in volcanic crises
be accurately represented in risk assessment. (Marzocchi et al. 2012; Loughlin et al. 2015; Fearnley et al.
This claim sits uneasily with the human context and conceit 2018). The primary uncertainty is whether or not volcanic un-
of risk and also with the high levels of uncertainty in a crisis— rest will result in an eruption (Newhall and Hoblitt 2002;
and there is abundant evidence that scientists are not always Newhall and Pallister 2015). There are different ways of
adept at estimating probabilities without training (Donovan conceptualising uncertainty. In the context of a volcano obser-
et al. 2015; Doyle et al. 2011, 2014; Gigerenzer et al. vatory meeting where different datasets are discussed to try to
2007)—probability does not always feature heavily in scientific understand what the volcano is doing, there are uncertainties of
education. While elicitation allows scientists to express how measurement (errors on the datasets themselves), uncertainties
certain they are about the numbers they are supplying and thus of interpretation—which may be due to lack of knowledge, or
does attempt to manage uncertainty to a degree, it is unable to to stochastic processes, model uncertainties, if models are being
take into account the qualitative aspects of uncertainty, and used (e.g. a simple Mogi model (Mogi 1958)), and there may
especially those that are introduced by values, fears or percep- also be wider epistemic uncertainties about the volcanic system
tions in the process (Table 1). These are things that are hard to itself (Woo 2015; Beven et al. 2016; Bretton et al. 2018). On
be conscious of, and may be the result of knowing the group, of top of this, however, there are social forms of uncertainty
institutional practices or of interactions. They might include (Wynne 1992; Stirling 2007; Sword-Daniels et al. 2016).
social or psychological aspects of judgement, anticipation of These may include the uncertainty that arises because a senior
political or public responses to an assessment, personal under- or respected colleague disagrees with you, or uncertainty about
standings of probability and narrow disciplinary expertise the political and societal implications about the advice that is
(Table 1). Understanding and managing these values require being produced—fear of being Bwrong^, anxiety about the
involvement of the social sciences and humanities alongside Bunknown unknowns^, and uncertainty about how people will
the physical sciences and acknowledgement of limitations behave in a crisis—or even whether they will actively oppose
(Morgan 2014; De Marchi 2015). For example, the European the advice (Donovan and Oppenheimer 2014). These social
Food Safety Authority has an extremely thorough guidance uncertainties may be very minor, or they could be
document, including a review of expert elicitation methods substantial—this depends on the situation and on the stakes,
(EFSA 2014). Their proposed method—which is impractically as well as on personalities (some scientists are more prone to
resource-intensive for volcanic contexts—takes into account anxiety and precaution than others). Again, this is not to say
many of the issues with elicitation that have been identified that the final scientific advice is Bsocially constructed^, but
by practitioners, psychologists and sociologists. It also involves rather that there are social elements that are not easily reducible
a choice between the main methods of elicitation (Cooke, and that require reflexivity (Harris 2015; Bretton et al. 2018).
Table 1 Types of uncertainty in volcano advisory science. Epistemic Spiegelhalter 2008; Spiegelhalter and Riesch 2011; Stirling 2007;
uncertainty (arising from lack of knowledge, reducible by knowing more) Sword-Daniels et al. 2016; Wynne 1992)
is italicised. Sources: (Proctor 1991; Proctor and Schiebinger 2008;
Type Description Management
Model errors Can include simplifications, data selection issues, Sensitivity analysis, Monte-Carlo
measurement issues, design issues, favoured models simulations, but some not quantifiable
Eradication of error Data selection in order to minimise error Potentially quantifiable
Relative importance Where a scientist/group cannot agree on the relative Potentially quantifiable depending on nature o
importance of different kinds of evidence f evidence
Interpretative error Errors of interpretation and inference in understanding a dataset Potentially quantifiable
Social uncertainty/indeterminacy Uncertainty about any aspect of the decision-making Not quantifiable—reflexive and intersubjective
process—scientists’ reputations and relationships management
with one another, relative experience, anticipated
reception of advice by population/policymakers
Political uncertainty Concern about the use of advice—e.g. following the L’Aquila trial Not quantifiable—reflexive and intersubjective management
Economic uncertainty Concern about the impact of advice on jobs, livelihoods etc. Potentially quantifiable but not readily
(requires extensive economic data)
Health/wellbeing uncertainty Linked to social uncertainty, but broader—how advice might Not quantifiable—requires reflexivity and careful
affect people if nothing happens etc. communication and social care
Agnotologic uncertainty Uncertainty that arises from culturally induced ignorance—e.g. Reflexive management possible and enhanced
things not studied, or misinterpreted— Bunknown unknowns^ by interdisciplinary approach
Aleatory/stochastic uncertainty Uncertainty due to the inherent randomness of the earth system. Probabilistic
Table 1 shows that aleatory and epistemic uncertainty are not paper (but see Hicks and Few, 2015; Few et al. 2017)—but the
alone in the risk assessment process—social uncertainties (in- complexity of vulnerability and its enhancement by weakened
determinacies or ambiguities in the language of (Stirling 2007; advisory science further emphasises the need for effective collab-
Wynne 1992)) of various kinds also prevail and they are not oration between social and physical sciences.
knowable, but nor are they entirely stochastic. They relate to
knowledge about the social, which requires a different approach
to that typically taken in the physical sciences. This exemplifies Hybrid approaches to policy advice:
the need for much greater engagement with the social sciences incorporating social sciences and humanities
as well as the physical, and complicates interdisciplinary work-
ing: expectations about the nature of knowledge are very differ- Equating the philosophy of risk with the philosophy of science
ent between social and physical scientists (see Appendix). can inhibit interdisciplinary working, and yet prevails in both
Ultimately, it is helpful to consider that, to some degree, all social and physical science approaches. There are significant
knowledge is fundamentally situated—it bears the hallmarks challenges for the broader hazard science community here:
of its production (Haraway 1988), including combinations of disaster risk remains contested between social and physical
social, institutional and epistemic value judgements. sciences, with the social sciences arguing that disasters are
socially constructed, and the physical sciences focussing on
quantitative risk assessment (Gaillard and Mercer 2013;
Dynamic risk: how people think Wisner et al. 2012). This polarisation has parallels within phi-
about volcanoes losophies of knowledge—the BScience Wars^ of the 1990s
were characterised by debate about the authority of
There are often discrepancies in how volcanoes are viewed by science—does it have methods and modes of knowing that
different groups—scientists may be preoccupied with the risk, are more robust or rigorous than other modes (Gould 2000)?
and yet populations may not think of the nearby volcano in this This debate was polarised around realist accounts in which the
way—even after being advised of the risk (Torrent et al. 2017). scientific method is privileged in its ability to discover truths
This is not particularly surprising given the opportunities that about nature, and social constructionist accounts, which argue
volcanoes afford to people in terms of livelihoods and commu- that science is a social activity with no higher access to truth
nity, but also suggests that understanding volcanoes from the (Segerstråle 2000). The problem with these accounts is that
perspective of the people who live on them is also a part of risk they are mutually exclusive from a philosophical perspective.
assessment and management—as has been acknowledged wide- Recent work on interdisciplinarity has therefore emphasised
ly in the literature (Barclay et al. 2008; Gregg et al. 2004; Gregg the need for Bspaces of difference^ (Barry et al. 2008), and
et al. 2008; Kelman and Mather 2008; Paton et al. 2010). People critical reflection on the power relations within scientific
view volcanoes as Bpart of them^ (Tobin et al. 2007), and may knowledge (Lave et al. 2014). These are both important as-
acknowledge that they take risks because the risks are pects of interdisciplinary working. Furthermore, recent work
worthwhile—they view the risks in a wider framework in the social sciences has moved away from social construc-
(Donovan and Oppenheimer 2014; Donovan and Oppenheimer tion and towards relational materialism: an awareness of the
2015b). importance of the material environment and its behaviour and
That risk is dynamic is inherent in the traditional formula in interaction with the social. This presents an opportunity to
disaster studies— it is a combination of hazard and vulnerability, bring social and physical sciences together under a Bcritical
both of which are dynamic and can cascade or interact through realist^ ontology (Bhaskar 2013 (1975)).
time and space in different ways (Pescaroli and Alexander 2015). The extensive work on interdisciplinarity in the past few
But views of risk are also dynamic: at times, people will be more years has demonstrated it continues to face problems in prac-
concerned about eruptive activity and at times more concerned tice (Lave et al. 2014; Lane 2017; MacLeod 2018). This is
about livelihoods or industry or community issues—and the perhaps particularly true in the environmental sciences, be-
timing may depend on volcano behaviour, but it may also depend cause of what Barry et al. (2008) describe as the many ways
on media reports, animal behaviour or other less readily deter- of conceiving of the environment—not to mention the many
mined factors. For example, the attempted relocation of Chaitén disciplines with an interest in theorising the environment and
town after the 2008 eruption failed because some local people humanity’s place within it. There may also be power struggles
wanted to move back to the town whose destruction they had and tacit differences in assumptions between disciplines—
witnessed, not because they rejected or denied the risk, but be- such as the ongoing clash of paradigms between those who
cause of a much more complex view of government, place and focus on analysing hazards as physical objects, and those who
identity (Espinoza et al. 2015; Torrent et al. 2017; Ugarte and conceive of hazards as social constructs within disaster studies
Salgado 2014). A detailed analysis of social vulnerability and (Gaillard and Mercer 2013; Wisner et al. 2012). Social scien-
cultural perspectives on volcanism is beyond the scope of this tists pursuing a transformative agenda have sometimes tended
to denigrate physical science in favour of political interven- approach, and social science, which is most commonly construc-
tions, for example. The problem is that risk is both real and tionist. It also often involves disciplines that focus on the percep-
constructed: there is a Breal^ risk in the sense that the volcano tion, communication and context of risk (Fig. 3). This creates
might erupt, but there are also substantial social factors that discord because the tacit assumptions of one discipline are not
exacerbate that risk and that also feed into its assessment and clear to others. For example, a physical scientist might assume
management (note 10)—and its existence is as a Brisk^ is that social scientists are trying to find out a Btrue^ value for risk
dependent on human recognition of it as such, and on vulner- perception, while the social scientists might not believe such a
able people or assets. In order to manage this complexity, thing exists.
sociologists have repeatedly called for deliberative approaches
to scientific advice, where citizens are consulted so that there
are observers within the scientific process. This is not practi- Approaches to Bwicked^ problems: open
cable in a volcanic crisis, but the involvement of interdisci- spaces
plinary social sciences in risk assessment would solve many of
the problems that have been encountered—such as those at Social scientists and philosophers in other fields have proposed
L’Aquila. This might include the involvement of outreach and several approaches to these complex problems, seeking to deal
education specialists, but might also involve sociologists and with the inevitable value judgements involved in science for risk
human geographers (Lane et al. 2011). assessment (Jasanoff 2003; Wynne 2004; Stirling 2007, 2010;
This suggests that social scientists should be involved along- Lave et al. 2014). Terje Aven and Ortwin Renn advocate a Bsemi-
side physical scientists in policy advice on risk, in general, as has quantitative^ approach to risk itself, which involves the charac-
been suggested by many in other environmental contexts terisation and description of background factors that affect the
(Castree et al. 2014; Hulme 2011). Yet, this is relatively rare in risk assessment and scenario construction (Aven and Renn
many countries, in part because there is not enough encourage- 2009). They argue that this provides a Bbroader uncertainty
ment from governments, and in part because they are not sug- characterization^ and Bscenarios that are basically validated ac-
gested by physical scientists. Sometimes, this is the result of a cording to logical consistency, psychological empathy with the
lack of mutual understanding of philosophies of knowledge. main players involved, congruence with past trends, and narra-
Much of this results from a lack of constructive dialogue— tive plausibility^. In contrast, Cass Sunstein has advocated a
social science for disasters, for example, does not deny the ma- realist approach that attempts to separate value and fact, and that
teriality of the hazard, but argues that it is only hazardous because is largely based on quantitative social psychology (Sunstein
of human behaviour, and particularly the development problems, 2005). In the sociology of scientific knowledge, Sheila Jasanoff
policy decisions and inequalities inherent in vulnerable areas and Brian Wynne have promoted social construction and delib-
(Wisner et al. 2004, 2012). On the physical side, however, the eration as approaches to policymaking with regard to new tech-
tendency to reduce risk to a probability of harm is not compatible nologies (Jasanoff 1996; Wynne and Shackley 1994). Andy
with the more embedded, socially constructed approach, which Stirling has advocated approaches that acknowledge the contin-
argues that risk is much more than a probability (and, indeed, that gency of probability and risk, and that Bkeep it complex^(Stirling
probability itself is a concept that is socially constructed, and 2003; Stirling 2008). There is also the Bpragmatic
emerged in a particular context (Hacking 1975; Hacking constructionist^ approach (Mansilla 2010) in which Ba system
1990)). Risk, in this understanding, requires understanding of of thought in reflective equilibrium^ is imagined, invoking dis-
social, cultural and political factors that are complex and ciplinary insights to deal with interdisciplinary challenges.
interwoven—but that render probability of harm impossible to Isabelle Stengers has advocated a Bslow science^ approach to
calculate without simplification (for wider discussion on the na- contemporary challenges (Stengers 2013). Building on the work
ture of risk, see: Beck 1992; Giddens 1999; Douglas and of science and technology studies, she raises the question of
Wildavsky 1982; Mythen and Walklate 2006; Renn 2008; whether or not the traditional approaches to science, whereby
Jasanoff 1999; Roeser et al. 2012). Probability of a hazard event facts become accepted through their survival of the critiques of
in a particular area is less of an issue, but risk assessment is, Bcompetent colleagues^, are actually appropriate for real-world
because risk must incorporate meaningful analysis of vulnerabil- problems. She suggests that Baccepting messiness^ and learning
ity, and vulnerability cannot be easily calculated beyond damage from it, rather than ignoring it, might be a valuable approach.
to the physical built environment: social vulnerability is depen- This has much in common with the challenges outlined by pro-
dent on demographic factors and policies, among other things. In ponents of Bpost-normal science^ (Funtowicz and Ravetz 1992).
order for social and physical scientists to work together to under- Scholars in the environmental humanities and cultural stud-
stand risk, then, open and transparent discussion of these factors ies have increasingly argued that disasters are strongly cultur-
is necessary. Nevertheless, interdisciplinary working remains a ally mediated. This expands the disciplinary range consider-
challenge because its philosophy of knowledge is unclear: it ably, and draws in historians, philosophers, literary scholars
incorporates both physical science with a dominantly realist and artists to improve communication and depth of dialogue
Fig. 3 Two ways of

conceptualising the arguments of
this discussion paper. Top: the
context of hazard assessment for
geophysical hazards (a simplified,
alternative conception of the
figure in Donovan et al. 2012c)—
note that this figure incorporates
the idea that scientists make
decisions that depend to some
(usually small) extent on their
own social context, experiences
and values; bottom: the complex
disciplinary landscape of risk—
what risk is conceived as varies
between realist and construction-
ist accounts, but also significant
are approaches to how people
perceive risk, and the broader
context of social and cultural
values in which those judgements
take place
between epistemic communities (Button 2016; Krüger et al. (Bhaskar 2013 (1975); Giere 2010 (2006)) in this kind of work-
2015). It also raises the issue of values again: knowledge ing. In a similar vein is post-normal science (Funtowicz and
cultures differ in their values. Mike Hulme has highlighted Ravetz 1992; Funtowicz and Ravetz 1995). These approaches
the importance of culture in how people understand climate, all accept that there is a reality that humanity is engaging with,
both currently and historically (Hulme 2009; Hulme 2010; rather than advocating strong social construction, but also allow
Hulme 2014; Hulme 2016). World views can clash with for a greater awareness of the uncertain, unknowable and un-
Bscience views^, and such a clash requires humility as argued quantifiable aspects of modern problems through the involve-
by Sheila Jasanoff (Jasanoff 2007). Ultimately, the practice of ment of an Bextended peer community^. All of these approaches
inter- and transdisciplinary working requires that the human- focus on philosophies of science that will enable the management
ity of all the participants is recognised and accepted. of real-world problems (those referred to by Science Studies
Ronald Giere and Roy Bhaskar have proposed Bscientific scholars as Bmode 2^, or application-driven science (Gibbons
perspectivism^ and Bcritical realism^ respectively as ways of et al. 1994)). Critical realism (Bhaskar 2013 (1975)) is an increas-
allowing a more nuanced view of the philosophy of science ingly common approach to social scientific research, because it
allows for a reality that has agency but that can only be repre- production of knowledge (Jasanoff 2004a, b) is central in the
sented by human endeavour. It has become increasingly popular working of genuinely interdisciplinary collaboration.
as social sciences have moved away from social construction and Interdisciplinary work requires acknowledgement of the dif-
become interested in materialism and vitalism. Critical realism ferences between realism and constructionism (Fig. 3, lower).
allows for the representation of reality through scientific knowl- Critical realism (or relatedly weak social construction of risk)
edge, whilst acknowledging the need for reflexivity and critical does not remove agency from the natural system. It argues that
thinking in both dealing with how scientific knowledge is pro- how we understand the natural system is not entirely indepen-
duced, and with its uncertainties, and also with the social and dent of human nature and social factors and may actually be
political contexts that entangle volcanological knowledge irre- dependent on the interactions between humans and nature
trievably. Recent work in physical geography has taken these (Braun and Castree 2005; Castree et al. 2014; Castree and
ideas, for example, in the formation of BCritical Physical Braun 2001; Hulme 2009). Disasters are assembled from hu-
Geography^ (CPG) (Lave 2015; Lave et al. 2014; Tadaki et al. man and physical interactions (Donovan 2016). The approach
2015). In introducing a special issue on CPG in 2015, Rebecca to risk that is taken in the assessment of risk from environ-
Lave writes: mental hazards has to acknowledge this interconnection.
We and our field sites are already tangled in political, so- Research in other fields suggests that this means taking an
cial, and economic relations. It is time to acknowledge those approach to risk that acknowledges its social origins whilst
relations and engage them explicitly. Critical physical geogra- not denying the role of the material environment. It also
phy enables us to do just that. (p.573–4). means thinking about how people think about risk—how it
Again, this is not to say that scientists get involved in pol- affects individual narratives and belief systems—and how
itics, but rather that scientists have to be aware of and engage the representation of risk feeds into those beliefs (Demeritt
critically with the social and political contexts, histories and et al. 2010; Demeritt et al. 2013; Irwin 2014; Pidgeon and
consequences of their work. In a risk context—particularly Fischhoff 2011).
perhaps the longer-term planning scenario—this involves The argument in this discussion paper reinforces the view that
careful communication of knowledge and reflexive engage- (i) risk reduction requires inter- and transdisciplinary working;
ment with its production. In the domain of risk, this might (ii) such working requires critical thinking, both about one’s own
involve consideration of the point at which a risk is philosophy of knowledge and that of co-workers. It suggests that
quantifiable—ethically as well as scientifically. the ways that different disciplines approach risk are complemen-
Such arguments do not devalue the scientific method, but tary, but communication between these approaches is essential—
rather show that there is a need to be critical in the framing and and listening is also a part of communication (Pidgeon and
process of scientific risk assessments so that they involve the Fischhoff 2011), because there are tacit assumptions that do
public and acknowledge limitations (Hajer 2003). This re- not carry across disciplinary boundaries. Furthermore, there
quires interaction between scientists, social scientists and phi- are disciplinary differences in how evidence is produced, con-
losophers of science—and also those in the humanities who ceived of and used—differences that are rooted in philosophy.
study the way that texts are read and interpreted (Foley 2018). There are also considerable differences in how uncertainty is
There is a growing call for projects that examine these issues managed or even identified between disciplines—dealing with
from a humanities standpoint, particularly looking at the uncertainty about a medical prognosis on an individual scale is
stories that scientists tell, and the ways in which knowledge different from managing hurricanes (Spiegelhalter 2008, 2017;
becomes accepted. In the context of risk, there is considerable Joslyn and Savelli 2010; Stephens et al. 2012; Beven 2016).
scope for arts-based approaches that use interactive and visual The quantification of the propagation of uncertainty is lim-
methods to communicate risk and uncertainty and that empha- ited if some of the uncertainty is not quantifiable (Brown et al.
sise that communication also involves listening (Pidgeon and 2015; Pereira et al. 2017; Thissen et al. 2017). Much is made of
Fischhoff 2011) (see Fig. 3). scientific uncertainty and the importance of trying to account
for both epistemic and aleatory uncertainties in models
(Aspinall 2006; Woo 2008, 2012; Marzocchi and Woo 2009).
How do we know what we think we know? However, social and political uncertainties are much more per-
vasive and difficult to measure. Combined, these issues require
Figure 3 conceptualises one of the problems that this discus- the involvement of a range of disciplines within and beyond the
sion paper has raised: the calculation of a probability takes natural sciences. This can involve quantification of uncertainty
place in a human framework that belies realism as an approach as far as this is possible—but also an awareness of the non-
to risk—realism in this context has to be critical. However, quantifiable ambiguities and indeterminacies involved. Risk is
that does not negate the usefulness of probabilistic hazard both dynamic and multidimensional: it can be represented only
assessment as a part of a holistic approach. What it does mean in part by a probability. It needs to be both quantified and
is that understanding the social context of science and the co- characterised, and this requires transparency about how the
figures were arrived at, especially in complex systems and the 1. The challenge of volcanic crises increasingly tends to drag
use of belief based methods. It also requires thinking about how scientists beyond Popperian science into subjective prob-
the results will be used—common implicitly but not explicitly ability. However, much scientific advice is still dominant-
in risk assessment (Lave et al. 2014)—and this again brings ly based on Popperian conjecture.
back the focus to inter and transdisciplinary working, with so- 2. The roles, responsibilities and burden on scientists vary
cial scientists and with policymakers, incorporating citizens considerably depending on political and social histories
where possible too. This section can be summarised in three and circumstances. Involvement of scientists in decision-
key principles: framing and dialogue (communication, includ- making is geographically variable in terms of institutional
ing listening); transparency (of method and evidence); critical responsibility, but scientific knowledge (created by scien-
reflection (of power dynamics and positions). These are com- tists) is always involved in decision-making.
plex ideas that require interaction with the humanities, partic- 3. This is the root of many of the challenges with
ularly philosophy and history, to understand the nature of communication—especially in a linear context—
knowledge and its production in different contexts. scientists need to be able to defend the knowledge that
Knowledge itself changes as it moves through social groups. fed into a decision. Arguing that separation of roles will
solve the problem is a red herring, because (incomplete)
knowledge itself permeates all roles, including the public
(who test knowledge claims).
Conclusions: advisory volcano science 4. While having clear roles is important, experience in many
in the ‘Anthropocene’ other fields suggests that public interpretations of warnings
and evacuation orders will involve the testing of knowl-
Cutter et al. (2015) note that disaster incidence can be signifi- edge claims—and probability will not protect against this.
cantly reduced by existing knowledge if it is better deployed, 5. These issues—inductive reasoning, communication, un-
and appear to take a linear approach to Bknowledge transfer^. certain and incomplete knowledges—are dealt with regu-
This forum paper has considered in detail insights from other larly in social science. Social science and even humanities
fields that deal with the use of scientific knowledge in risk- can aid physical scientists in volcanic crises, and enable
related policymaking and practice. It has shown that there are physical scientists to demonstrate a level of transparency.
complex issues in how knowledge itself is understood, and how 6. There are aspects to volcanic risk that are fundamentally
it operates under uncertainty of different kinds in an interdisci- social, too: social vulnerability, cultural understandings
plinary context. The problem of volcanic and other disasters is and politics are critical to the management of risk. Again,
not just about knowledge (social, medical and physical sci- these contextual elements require transdisciplinary input.
ences)—it is about how knowledge is produced and how it 7. Integration of social and physical approaches to risk is
encounters policy and practice, and how it changes in the pro- challenging, but not impossible. It requires a movement
cess (Jasanoff 2005; Haraway 1988). These challenges require beyond Popperian science and social construction to-
interdisciplinary and transdisciplinary approaches in which so- wards critical realism (or weak construction): an approach
cial scientists, physical scientists and philosophers collaborate that recognises the physical dynamics of hazards and the
with policymakers and publics. Not only does this kind of work social dynamics of how we understand them. Such
allow better ‘communication of science’, but it also brings critical volcanology can help protect volcanology from
deeper understanding of what science is doing and what its the political.
limitations are—and fundamentally, can protect the physical sci- 8. Such integration has the potential to significantly enhance
ences from events such as those in L’Aquila by keeping them crisis management and longer-term volcanic risk manage-
transparent and thereby holding them accountable (Jasanoff ment (as is the case in New Zealand). It requires humility
2004a, b; Scolobig et al. 2014; Bretton et al. 2018). Effective and transparency on the part of all participants. It also
working in this respect requires cultural sensitivity and aware- involves social science throughout the process of risk
ness of very diverse disciplinary approaches to knowledge. assessment.
There is much that volcanology can learn from other fields in
this respect (Pearce et al. 2015). The problems of the This article suggests, then, that as a community, volcanol-
Anthropocene—an interconnected, human-physical world— ogy can be significantly enhanced through a broader epistemic
require post-normal, critical approaches that embrace reflexivity flexibility that accepts the diversity in philosophies of knowl-
and acknowledge the embedded, situated nature of knowledge edge and refocuses on providing evidence for transparent and
production and evolution. Creating such approaches requires the effective decision-making. This requires communication be-
availability of open spaces for exchange of ideas, debate and tween scientists, physical and social, to work towards genuine
discussion between different knowledge communities. co-management of risk on active volcanoes that protects both
The argument of this paper is summarised as follows: science and the communities that rely upon it. Technological
approaches to share resources transnationally are increasingly only valid form of knowledge, and developed further in the
utilised in volcanic crises (such as teleconferences) and while twentieth century when the logical positivists developed the
they require careful cross-cultural management, these also criterion of verification. Karl Popper often gets referred to as a
present an opportunity for observatories and civil protection positivist, because, although he criticised logical positivism,
institutions to widen their peer community. his introduction of falsification rather than verification as a
defining characteristic of scientific knowledge did not change
Acknowledgements The author would like to thank Jenni Barclay and an the fundamental approach. This paper will refer to Popper’s
anonymous reviewer for helpful comments that enhanced the manuscript, postpositivist approach as Popperian science. Later philoso-
and Raffaello Cioni and Andy Harris for their careful editorial handling of
this manuscript. phers such as Thomas Kuhn have moved thinking further,
suggesting that science is characterised by paradigm shifts
and that its characteristics should be accuracy, consistency,
breadth of scope, simplicity and fruitfulness.
Appendix There are also ontological issues here. Positivism generally
suggested that we can Bknow^ reality objectively (generally
A brief note on definitions and terminology known as realism). Other philosophers—notably Kant—have
focussed on representing reality, regarding Bthe thing in itself^
BRisk^, Bvulnerability^, Bhazard^ and many of the other terms as ultimately unknowable. This is an important distinction, be-
that are widely used in volcanology and other fields are cause the word Bobjective^ gets used a lot in scientific circles
contested. This paper adopts the definitions of the UNISDR. but has multiple meanings. The Kantian object—something
In particular, it views both hazard and risk as not merely reduc- that represents reality, is knowable. Popper’s approach is also
ible to a probability in practice: a risk assessment requires a realist approach, but shows some awareness of the impor-
quantification of social and physical vulnerability and of capac- tance of probability and conjecture in how we can Bknow^
ity, and even a hazard assessment must take into account the reality (though he is sceptical of belief-based probabilities).
nature of the hazard and its impact and severity. There are While there are positivist and postpositivist approaches to
methods to quantify social vulnerability (Cutter et al. 2003), social knowledge, the problem with the social is that it changes:
but these require too many variables to be applicable in many a rock does not change its composition, but people change their
locations and also ignore key indicators that are not minds all the time. Approaches to social knowledge have tended
measureable (e.g. values and beliefs) but that affect vulnerabil- to be rather broader than those of the natural sciences as a result:
ity significantly. At best, indices may be used for social vulner- knowing Bthe truth^ about the social is more challenging. Social
ability and adaptive capacity, but this paper suggests that quan- science therefore is often interpretative: interpretivism argues
tification should not be the primary goal; concise summation that the social is fundamentally different from the natural and
that can incorporate the qualitative complexities is equally use- requires different techniques in order to understand human be-
ful (and often more explicable to non-experts). In volcanology, haviour and make sense of it. There are different approaches
typical approaches have quantified hazard rather than risk. within this however. Constructivism has tended to suggest that
Hazard is a process with the potential to affect life or property. ideas and interpretations are entirely constructed through social
Risk may be quantified in more reductive ways, such as interactions: people construct their meanings of the world on the
individualised risk per annum (annual risk of death), but this basis of their historical and social perspectives. Constructivism
cannot take into account factors that may make particular indi- (also called constructionism) has been widely used in the social
viduals or groups more vulnerable. Risk, incorporating vulner- sciences, often as the ontological basis for interpretivist re-
ability, is ultimately assembled out of material processes (haz- search. Some sociologists of science have also argued that the
ard), value systems and beliefs, governance systems and deci- physical and life sciences are socially constructed.
sions, political economies and other societal and economic fac- Feminist social science has taken this further by pointing out
tors. It is a highly complex concept. Hazard is a simpler idea, the role of positionality: the social is studied by people, who are
because it refers to the physical phenomenon. Probabilistic as- also social, and so bring their own experiences and assumptions to
sessment of hazards is commonly carried out in volcanology, the study. This is widely accepted in the social sciences: how we
and hazard maps are increasingly probabilistic. interpret other people does depend to some extent on ourselves.
This paper also refers to several philosophical approaches Science studies has taken this further and argued that the natural
to knowledge. Positivism is an approach that suggests that sciences are also affected in this way as scientists make decisions
knowledge about reality must be proven by verification in about which models to use and how to interpret their data.
order to be accepted as fact. It emphasises that reality can be Other ontological approaches have surfaced in the social sci-
known through empirical reasoning and logic. Positivism was ences. These include pragmatism—which focuses on the re-
developed in the eighteenth century, when thinkers such as search problem and then uses all available methods to research
Laplace and Comte argued that the scientific method was the it—and transformative research, which is focussed on political
change and the role of power dynamics. Transformative re-

search is very common in disaster studies, where researchers
are often motivated by the search for political change and social
justice. Figure 4 sums up the general approach that is taken to
Fig. 4 Approaches to knowledge

and how they relate to the choice
of ontological position and
epistemological position.
Ontology typically feeds into
epistemology, which then
influences the selection of
methodology and method
Table 2 Some of the paradigms for thinking about research and method
General perspective Ontology Epistemology Methodology
Positivism/postpositivism Realism-we can know reality Emphasis on quantitative Experimental,mathematical,

measurement quantitative
Constructionism Reality is socially constructed Interpretivism Qualitative
Phenomenology
Hermeneutics
Feminism
Pragmatism Main focus should be on usefulness Should solve problems-depends Mixed methods
not theory on situation Action research
Subjectivism Reality depends on perception Interprevitism Qualitative
Structuralism Humanities-texts,arts
Poststructuralism
Critical realism Realities can be represented but should Critical realism,critical Mixed methods, discourse,analysis,
be interpreted critically/reflexively social science quantitative, qualitative
Transformative Reality depends on political influences Marxism Action research,qualitative
Feminism
research. Note that some of the ontological and epistemological 3

For example, in the kind of statement that Bthis method
terms can really be used in either of these categories depending provides an objective probabilistic approach to volcanic risk
on context: this is summarised in Table 2. a s s e s s m e n t ^, w h e r e Bo b j e c t i v e ^ i s n o t d e f i n e d
philosophically.
4
Jasanoff refers to the public testing of knowledge claims
as Bcivic epistemology^. Epistemology is the study of the
nature of knowledge.
5
Notes from main text Both epistemic values—those associated with the produc-
tion of knowledge, such as valuing that it is robust and defen-
https://www.scientificamerican.com/article/volcano- sible using the scientific method—and non-epistemic values,
monitoring-jindal/ Accessed February 2019. such as personal beliefs—which might include valuing the
2
This discussion paper focusses on risk and hazard assess- lives of local people on the volcano.
6
ment rather than risk management, but would view volcanic i.e. If I claim that my assessment should be taken seriously
hazard and risk assessment as a key part of volcanic risk as scientific because it is Bobjective^, and the population then
management. find out that it was done by expert judgement, the authority of
10
the assessment will crumble because it was not actually Key issues include poverty, lack of mobility and ca-
Bobjective^ as most people would understand the term. If, how- pacity, marginalisation (often of indigenous groups, but
ever, my claim for the distinctiveness of my assessment was that also of particular demographics, such as the elderly), po-
it used established methods for complex uncertain problems, litical corruption, inadequate preparatory policies, inade-
and was observed by independent evaluators, for example, I quate knowledge, poor infrastructure, limited access to
might be more successful in retaining distinctiveness for the education and healthcare …
assessment.
7
Popper acknowledged the conjectured and contested nature
of knowledge. BNeither observation nor reason is an authority.
Intellectual intuition and imagination are most important, but References
they are not reliable: they may show us things very clearly, and
Armijos MT, Phillips J, Wilkinson E, Barclay J, Hicks A, Palacios P,
yet they may mislead us. They are indispensable as the main
Mothes P, Stone J (2017) Adapting to changes in volcanic behav-
sources of our theories; but most of our theories are false anyway. iour: formal and informal interactions for enhanced risk manage-
The most important function of observation and reasoning, and ment at Tungurahua volcano, Ecuador. Glob Environ Chang 45:
even of intuition and imagination, is to help us in the critical 217–226
examination of those bold conjectures which are the means by Aspinall WP (2006) Structured elicitation of expert judgement for prob-
abilistic hazard and risk assessment in volcanic eruptions. In: H.M.
which we probe into the unknown … The more we learn about Mader SC, C. Connor, L. Connor (ed) Statistics in volcanology.
the world, and the deeper our learning, the more conscious, spe- Geological Society of London, London, pp 15–30
cific, and articulate will be our knowledge of what we do not Aspinall W (2010) A route to more tractable expert advice. Nature
know, our knowledge of our ignorance. For this, indeed, is the 463(7279):294–295
main source of our ignorance - the fact that our knowledge can Aspinall W (2012) Comment on Donovan, et al, 2012 "social studies of
volcanology: expert advice and knowledge generation on active
only be finite, while our ignorance must necessarily be infinite.^ volcanoes". Bulletin of Volcanology this issue
Popper, 1963, p.37–8. Aspinall WP, Cooke RM (1998) Expert judgement and the Montserrat
8
Even where the frequency of eruption at a volcano is very volcano eruption. In: Mosleh AB (ed) Proceedings of the 4th inter-
well known and a long-term frequentist probability can be national conference on probabilistic safety assessment and manage-
ment. Springer, New York City, pp 2113–2118
calculated, (i) a probability distribution has to be selected by
Aspinall W, Sparks RSJ (2004) Volcanology and the law. IAVCEI News
someone; (ii) the possibility of unprecedented scenarios 1(4)
remains. Aspinall WP, Woo G, Voight B, Baxter PJ (2003) Evidence-based volca-
9
There is extensive debate in philosophy about what con- nology: application to eruption crises. J Volcanol Geotherm Res
stitutes an Bobject^, and about the nature of Bobjectivity^ and 128:273–285
Aven T, Renn O (2009) The role of quantitative risk assessments for
Bsubjectivity^. An interesting discussion is given in
characterizing risk and uncertainty and delineating appropriate risk
Hermansson (2012) Defending the conception of Bobjective management options, with special emphasis on terrorism risk. Risk
risk^. Risk Analysis: An International Journal 32(1):16–24— Anal 29(4):587–600
Hermansson argues that we should reconceptualise objectivity Bachelard G (2002) The formation of the scientific mind a contribution to
to incorporate values, taking all knowledge as situated. The a psychoanalysis of objective knowledge
Bamber JL, Aspinall W (2013) An expert judgement assessment of future
definition I am using here is Kant’s empirical object, which is sea level rise from the ice sheets. Nat Clim Chang 3(4):424–427
an Bobject in time and space^. In Kantian philosophy, risk Barclay J, Haynes K, Mitchell T, Solana C, Teeuw R, Darnell A,
would be viewed as a Bconcept^ or Bproposition^, not an Crosweller HS, Cole P, Pyle D, Lowe C, Fearnley C, Kelman I
object, because an object has to be completely independent (2008) Framing volcanic risk communication within disaster risk
of perception. Risk is a human construct. It cannot be reduction: finding ways for the social and physical sciences to work
together. Geol Soc Lond, Spec Publ 305(1):163–177
completely independent of perception. This is equally the case Barry A, Born G, Weszkalnys G (2008) Logics of interdisciplinarity.
with probability. To quote John Stuart Mill, BWe must remem- Econ Soc 37(1):20–49
ber that the probability of an event is not a quality of the event Bauer MW (2009) The evolution of public understanding of science—
itself, but a mere name for the degree of ground which we, or discourse and comparative evidence. Sci Technol Soc 14:221–240.
https://doi.org/10.1177/097172180901400202
some one else, have for expecting it^ John Stuart Mill, A
Baxter PJ, Aspinall WP, Neri A, Zuccaro G, Spence RJS, Cioni R, Woo G
System of Logic, Ratiocinative and Inductive [1843], ed. J. (2008) Emergency planning and mitigation at Vesuvius: a new
M. Robson, Vols. 7 and 8 of The Collected Works of John evidence-based approach. J Volcanol Geotherm Res 178(3):454–
Stuart Mill, Toronto/London: University of Toronto Press, 473
1974. See also De Finetti (1974) Theory of Probability. Beck U (1992) Risk society: towards a new modernity. Sage, New Delhi
Wiley, London, Hacking (1975) The emergence of probabili- Benessia A, De Marchi B (2017) When the earth shakes … and science
with it. The management and communication of uncertainty in the
ty. Cambridge University Press, Cambridge, Hacking (1990) L’Aquila earthquake. Futures 91:35–45
The taming of chance. Cambridge University Press, Betz G (2013) In defence of the value free ideal. Eur J Philos Sci 3(2):
Cambridge. See also Appendix 1. 207–220
Beven K (2016) Facets of uncertainty: epistemic uncertainty, non-station- Donovan A (2016) Geopower Reflections on the critical geography of
arity, likelihood, hypothesis testing, and communication. Hydrol Sci disasters. Progress in Human Geography:0309132515627020
J 61:1652–1665 Donovan A, Oppenheimer C (2014) Science, policy and place in volcanic
Beven K, Almeida S, Aspinall W et al (2016) Epistemic uncertainties and disasters: insights from Montserrat. Environ Sci Pol 39:150–161
natural hazard risk assessment–part 2: different natural hazard areas. Donovan A, Oppenheimer C (2015a) At the mercy of the mountain?
Nat Hazards Earth Syst Sci 2016:1–1 Field stations and the culture of volcanology. Environ Plan A
Bhaskar R (2013 (1975)) A realist theory of science. Routledge, 47(1):156–171
Bijker WE, Bal R, Hendriks R (2009) The paradox of scientific authority: Donovan A, Oppenheimer C (2015b) Resilient science: the civic episte-
the role of scientific advice in democracies. MIT press mology of disaster risk reduction. Sci Pub Pol:scv039
Braun B, Castree N (2005) Remaking reality: nature at the millenium. Donovan A, Oppenheimer C, Bravo M (2012a) Contested boundaries:
Routledge delineating the "safe zone" on Montserrat. Appl Geogr 35:508–514
Bretton RJ, Gottsmann J, Aspinall WP, Christie R (2015) Implications of Donovan A, Oppenheimer C, Bravo M (2012b) Reply to comment from
legal scrutiny processes (including the L’Aquila trial and other re- W.P. Aspinall on BSocial studies of volcanology: knowledge gener-
cent court cases) for future volcanic risk governance. J Appl ation and expert advice on active volcanoes^ by Amy Donovan,
Volcanol 4(1):1 Clive Oppenheimer and Michael Bravo [Bull Volcanol (2012) 74:
Bretton RJ, Gottsmann J, Christie R (2018) Hazard communication by 677-689]. Bull Volcanol 74(6):1571–1574
volcanologists: part 2-quality standards for volcanic hazard assess- Donovan A, Oppenheimer C, Bravo M (2012c) Science at the policy
ments. J Appl Volcanol 7:10 interface: volcano-monitoring technologies and volcanic hazard
Brown MB (2009) Science in democracy: expertise, institutions and rep- management. Bull Volcanol 74(5):1005–1022
resentation. MIT Press, Cambridge, Mass Donovan A, Oppenheimer C, Bravo M (2012d) The use of belief-based
Brown C, Brown E, Murray-Rust D, Cojocaru G, Savin C, Rounsevell M probabilistic methods in volcanology: Scientistsâ€™ views and im-
(2015) Analysing uncertainties in climate change impact assessment plications for risk assessments. J Volcanol Geotherm Res 247-248:
across sectors and scenarios. Clim Chang 128:293–306 168–180
Button G (2016) Disaster culture: knowledge and uncertainty in the wake Donovan A, Eiser JR, Sparks RSJ (2015) Expert opinion and probabilis-
of human and environmental catastrophe. Routledge tic volcanic risk assessment. J Risk Res:1–18
Calder E, Wagner K, Ogburn S (2015) Volcanic hazard maps. In: Donovan A, Ayala IA, Eiser J, Sparks R (2018) Risk perception at a
Volcanic hazard maps. Global Volcanic Hazards and Risk. persistently active volcano: warnings and trust at Popocatépetl vol-
Cambridge University Press, Cambridge, pp 335–342 cano in Mexico, 2012–2014. Bull Volcanol 80(5):47
Castanos H, Lomnitz C (2002) PSHA: is it science? Engineering Geology
Doubleday R (2007) Organizing accountability: co-production of
66(3â€B4):315–317
technoscientific and social worlds in a nanoscience laboratory.
Castree N, Braun B (2001) Social nature: theory, practice, and politics.
Area 39(2):166–175
Blackwell Publishers Oxford, Malden, MA
Douglas H (2000) Inductive risk and values in science. Philos Sci 67:
Castree N, Adams WM, Barry J, Brockington D, Büscher B, Corbera E,
559–579
Demeritt D, Duffy R, Felt U, Neves K (2014) Changing the intel-
Douglas H (2009) Science, policy, and the value-free ideal. University of
lectual climate. Nat Clim Chang 4(9):763–768
Pittsburgh Pre
Cornell CA (1968) Engineering Seismic Risk Analysis. Bulletin of the
Douglas M, Wildavsky A (1982) How can we know the risks we face?
Seismological Society of America 58(5):1583-&
Why risk selection is a social Process1. Risk Anal 2(2):49–58
Curtis A (2012) The science of subjectivity. Geology 40(1):95–96
Cutter SL (2003) The vulnerability of science and the science of vulner- Doyle EE, Johnston DM, McClure J, Paton D (2011) The communication
ability. Ann Assoc Am Geogr 93(1):1–12 of uncertain scientific advice during natural hazard events. N Z J
Cutter SL, Boruff BJ, Shirley WL (2003) Social vulnerability to environ- Psychol 40(4):39–50
mental hazards. Social science quarterly, 84(2):242–261 Doyle EE, McClure J, Johnston DM, Paton D (2014) Communicating
Cutter SL, Ismail-Zadeh A, Alcantara-Ayala I, Altan O, Baker DN, likelihoods and probabilities in forecasts of volcanic eruptions. J
Briceno S, Gupta H, Holloway A, Johnston D, McBean GA Volcanol Geotherm Res 272:1–15
(2015) Global risks: Pool knowledge to stem losses from disasters. EFSA (2014) Guidance on expert knowledge elicitation in food and feed
Nature 522:277–279 safety risk assessment. EFSA J 12(6):3734
Daston L (1992) Objectivity and the escape from perspective. Soc Stud Eiser JR, Donovan A, Sparks RSJ (2015) Risk perceptions and trust
Sci 22(4):597–618 following the 2010 and 2011 Icelandic volcanic ash crises. Risk
Daston L (1994) How probabilities came to be objective and subjective. Anal 35(2):332–343
Hist Math 21(3):330–344 Espinoza AE, Espinoza CE, Fuentes AA (2015) Retornando a Chaitén:
Davies SR (2008) Constructing communication: talking to scientists diagnóstico participativo de una comunidad educativa desplazada
about talking to the public. Sci Commun 29:413–434. https://doi. por un desastre socionatural. Magallania (Punta Arenas) 43(3):65–
org/10.1177/1075547008316222 76
De Finetti B (1974) Theory of probability. Wiley, London Fearnley CJ, Beaven S (2018) Volcano alert level systems: managing the
De Marchi B (2015) Risk governance and the integration of different challenges of effective volcanic crisis communication. Bull
types of knowledge. In risk governance (pp. 149-165). Springer, Volcanol
Dordrecht Few R, Armijos MT, Barclay J (2017) Living with Volcan Tungurahua:
Demeritt D, Nobert S, Cloke H, Pappenberger F (2010) Challenges in the dynamics of vulnerability during prolonged volcanic activity.
communicating and using ensembles in operational flood forecast- Geoforum 80:72–81
ing. Meteorol Appl 17(2):209–222 Fischer F (2000) Citizens, experts and the environment: the politics of
Demeritt D, Nobert S, Cloke HL, Pappenberger F (2013) The European local knowledge. Taylor & Francis
flood alert system and the communication, perception, and use of Foley R (2018) The geography of insight: the sciences, the humanities,
ensemble predictions for operational flood risk management. Hydrol how they differ. Oxford University Press, Why They Matter
Process 27(1):147–157 Fearnley CJ, Bird DK, Haynes K, McGuire WJ, Jolly G eds (2018)
Dessai S, Hulme M (2004) Does climate adaptation policy need proba- Observing the Volcano World: Volcano Crisis Communication.
bilities? Clim Pol 4(2):107–128 Springer
Friedman DP (2008) Public outreach: a scientific imperative. J Neurosci Haynes K, Barclay J, Pidgeon N (2007) The issue of trust and its influ-
28:11743–11745 ence on risk communication during a volcanic crisis. Bull Volcanol
Funtowicz SO, Ravetz JR (1992) Three types of risk assessment and the 70:605–621
emergence of post-normal science Hermansson H (2012) Defending the conception of Bobjective risk^. Risk
Funtowicz SO, Ravetz JR (1995) Science for the post normal age. Analysis: An Int J 32(1):16–24
Perspectives on ecological integrity. Springer, In, pp 146–161 Hicks, A. and Few, R., 2015. Trajectories of social vulnerability during
Gaillard JC, Mercer J (2013) From knowledge to action: bridging gaps in the Soufrière Hills volcanic crisis. Journal of applied volcanology,
disaster risk reduction. Prog Hum Geogr 37:93–114 4(1), p.10
Gallopín GC, Funtowicz S, O'Connor M, Ravetz J (2001) Science for the Hinchliffe S (2001) Indeterminacy in-decisions–science, policy and pol-
twenty-first century: from social contract to the scientific core. Int itics in the BSE (bovine spongiform encephalopathy) crisis. Trans
Soc Sci J 53(168):219–229 Inst Br Geogr 26(2):182–204
Gibbons M, Limoges C, Nowotny H, Schwartzman S, Scott P, Trow M Hincks TK, Komorowski J-C, Sparks SR, Aspinall WP (2014)
(1994) The new production of knowledge: the dynamics of science Retrospective analysis of uncertain eruption precursors at La
and research in contemporary societies. Sage, London Soufrière volcano, Guadeloupe, 1975–77: volcanic hazard assess-
Giddens A (1999) 1999. Runaway World, Reith Lectures ment using a Bayesian belief network approach. J Appl Volcanol
Giere RN (2010 (2006)) scientific perspectivism. University of Chicago 3(1):3
Press, Hulme M (2008) Geographical work at the boundaries of climate change.
Gieryn TF (1983) Boundary-work and the demarcation of science from Trans Inst Br Geogr 33(1):5–11
non-science: strains and interests in professional ideologies of sci- Hulme M (2009) Why we disagree about climate change: understanding
entists. Am Sociol Rev 48(6):781–795 controversy, inaction and opportunity. Cambridge University Press,
Gigerenzer G, Hertwig R, Van Den Broek E, Fasolo B, Katsikopoulos Cambridge
KV (2005) BA 30% chance of rain tomorrow^: how does the public Hulme M (2010) Problems with making and governing global kinds of
understand probabilistic weather forecasts? Risk Analysis: An Int J knowledge. Glob Environ Chang 20(4):558–564
25(3):623–629 Hulme M (2011) Meet the humanities. Nat Clim Chang 1(4):177–179
Gigerenzer G, Gaissmaier W, Kurz-milcke E, Schwartz LM, Woloshin S Hulme M (2014) Climate change and virtue: an apologetic. Humanities
(2007) Helping doctors and patients make sense of health statistics. 3(3):299–312
Psychol Sci Public Interest 8(2):53–96 Hulme M (2016) Weathered: cultures of climate. SAGE,
Giordano G, Bretton R, Calder ES, Cas R, Gottsmann J, Lindsay J, Irwin A (2014) Risk, science and public communication. Routledge
Newhall C, Pallister J, Papale P, Rodriguez L (2016) Toward handbook of public communication of science and technology:
IAVCEI guidelines on the roles and responsibilities of scientists 160–172
involved in volcanic hazard evaluation, risk mitigation, and crisis Jasanoff S (1990) The fifth branch: science advisors as policymakers.
response. Bull Volcanol 78:1–3 Harvard University Press, Cambridge, Mass
Goitom B, Oppenheimer C, Hammond JO, Grandin R, Barnie T, Jasanoff S (1993) Bridging the two cultures of risk analysis 1, 2. Risk
Donovan A, Ogubazghi G, Yohannes E, Kibrom G, Kendall J-M Anal 13(2):123–129
(2015) First recorded eruption of Nabro volcano, Eritrea, 2011. Bull
Jasanoff S (1996) Beyond epistemology: relativism and engagement in
Volcanol 77(10):1–21
the politics of science. Soc Stud Sci 26(2):393–418
Gould SJ (2000) Deconstructing the" science wars" by reconstructing an
Jasanoff S (1998) The political science of risk perception. Reliab Eng
old mold. Science 287(5451):253–261
Syst Saf 59:91–99
Gregg CE, Houghton BF, Johnston DM, Paton D, Swanson DA (2004)
Jasanoff S (1999) The songlines of risk. Env Val 8:135–152
The perception of volcanic risk in Kona communities from Mauna
Loa and Hualalai volcanoes, Hawai'i. J Volcanol Geotherm Res 130: Jasanoff S (2003) Technologies of humility: citizen participation in
179–196 governing science. Minerva 41:223–244
Gregg CE, Houghton BF, Paton D, Swanson DA, Lachman R, Bonk WJ Jasanoff S (2004a) Science and citizenship: a new synergy. Sci Public
(2008) Hawaiian cultural influences on support for lava flow hazard Policy 31(2):90–94
mitigation measures during the January 1960 eruption of Kilauea Jasanoff S (2004b) States of knowledge: the co-production of science and
volcano, Kapoho, Hawai[`]i. J Volcanol Geotherm Res 172(3–4): social order. Routledge, Abingdon, In
300–307 Jasanoff S (2005) Designs on nature: science and democracy in Europe
Hacking I (1975) The emergence of probability. Cambridge University and the United States. Princeton University Press, Princeton
Press, Cambridge Jasanoff S (2007) Technologies of humility. Nature 450:33
Hacking I (1990) The taming of chance. Cambridge University Press, Jasanoff S, Martello M (2004) Earthly politics: local and global in envi-
Cambridge ronmental governance. MIT Press, Cambridge, Mass, In
Hajer M (2003) A frame in the fields: policymaking and the reinvention Johnson JB, Watson LM, Palma JL, Dunham EM, Anderson JF (2018)
of politics. In: Hajer M, Wagenaar H (eds) Deliberative policy anal- Forecasting the eruption of an open-vent volcano using resonant
ysis: understanding governance in the network society. Cambridge infrasound tones. Geophys Res Lett 45(5):2213–2220
University Press, Cambridge Joslyn S, Savelli S (2010) Communicating forecast uncertainty: public
Hall SS (2011) Scientists on trial: at fault? Nature News 477(7364):264– perception of weather forecast uncertainty. Meteorol Appl 17:180–
269 195. https://doi.org/10.1002/met.190
Haraway D (1988) Situated knowledges: the science question in femi- Kant I, Guyer P (1998) Critique of pure reason. Cambridge University
nism and the privilege of partial perspective. Fem Stud 14(3):575– Press
599 Kelman I, Mather TA (2008) Living with volcanoes: the sustainable live-
Harris AJ (2015) Forecast communication through the newspaper part 2: lihoods approach for volcano-related opportunities. J Volcanol
perceptions of uncertainty. Bull Volcanol 77(4):30 Geotherm Res 172(3–4):189–198
Hastrup K (2012) Anticipating nature: the productive uncertainty of cli- Komorowski J, Hincks T, Sparks R, Aspinall W, Consortium CAP (2015)
mate models. In the social life of climate change models (pp. 11-39). Improving crisis decision-making at times of uncertain volcanic un-
Routledge rest (Guadeloupe, 1976). Global Volcanic Hazards and Risk 255
Krüger F, Bankoff G, Cannon T, Orlowski B, Schipper ELF (2015) Morss R.E, Demuth JL and Lazo JK (2008) Communicating uncertainty
Cultures and disasters: understanding cultural framings in disaster in weather forecasts: A survey of the US public. Weather and fore-
risk reduction. Routledge casting, 23(5):974–991
Kuhn T (1962) The structure of scientific revolutions. University of Mulargia F, Stark PB, Geller RJ (2017) Why is probabilistic seismic
Chicago Press, Chicago hazard analysis (PSHA) still used? Phys Earth Planet Inter 264:
Kuhn T (1977) Objectivity, value judgement, and theory choice. In: Kuhn 63–75
T (ed) The essential tension: studies in scientific tradition and Mythen G and Walklate S (2006) Beyond the risk society: critical reflec-
change. University of Chicago Press, Chicago, pp 320–329 tions on risk and human security: critical reflections on risk and
Lacey H (2005) Is science value free?: values and scientific understand- human security. McGraw-hill education (UK)
ing. Psychology Press Neri A, Aspinall WP, Cioni R, Bertagnini A, Baxter PJ, Zuccaro G,
Lane SN (2017) Slow science, the geographical expedition, and Critical Andronico D, Barsotti S, Cole PD, Esposti Ongaro T, Hincks TK,
Physical Geography. The Canadian Geographer/Le Géographe Macedonio G, Papale P, Rosi M, Santacroce R, Woo G (2008)
Canadien, 61(1):84–101 Developing an event tree for probabilistic hazard and risk assess-
Lane SN, Odoni N, Landström C, Whatmore SJ, Ward N, Bradley S ment at Vesuvius. J Volcanol Geotherm Res 178(3):397–415
(2011) Doing flood risk science differently: an experiment in radical Newhall C, Hoblitt RP (2002) Constructing event trees for volcanic cri-
scientific method. Trans Inst Br Geogr 36(1):15–36 ses. Bull Volcanol 64:3–20
Latour B (2004) Politics of nature: how to bring the sciences into democ- Newhall CG, Pallister JS (2015) Using multiple data sets to populate
racy. Harvard University Press, Cambridge, Mass probabilistic volcanic event trees. In: Volcanic hazards. Elsevier,
Lave R (2015) Introduction to special issue on critical physical geogra- Risks and Disasters, pp 203–232
phy. In: SAGE publications Sage UK: London, England, Nowotny H (2003) Democratising expertise and socially robust knowl-
Lave R, Wilson MW, Barron ES, Biermann C, Carey MA, Duvall CS, edge. Sci Public Policy 30(3):151–156
Johnson L, Lane KM, McClintock N, Munroe D (2014) O'Hagan A, Buck CE, Daneshkhah A, Eiser R, Garthwaite P (2006)
Intervention: critical physical geography. The Canadian Uncertain judgements: eliciting experts' probabilities. Wiley,
Geographer/Le Géographe canadien 58(1):1–10 London
Levidow L (2001) Precautionary uncertainty: regulating GM crops in Owens S (2000) "engaging the public": information and deliberation in
Europe. Soc Stud Sci 31(6):842–874 environmental policy. Environ Plan A 32:1141–1148
Levidow L, Carr S (2007) Europeanising advisory expertise: the role of Owens S (2015) Knowledge, policy, and expertise: the UK Royal
'independent, objective and transparent' scientific advice in Agri- Commission on environmental pollution 1970–2011. OUP Oxford,
biotech regulation. Environ Plann C: Gov Policy 25:880–895 Papale P (2017) Rational volcanic hazard forecasts and the use of volca-
Lindsay J, Marzocchi W, Jolly G, Constantinescu R, Selva J, Sandri L nic alert levels. J Appl Volcanol 6(1):13
(2010) Towards real-time eruption forecasting in the Auckland vol- Pate-Cornell E (2002) Risk and uncertainty analysis in government safety
canic field: application of BET_EF during the new Zealand National decisions. Risk Anal 22(3):633–646
Disaster Exercise ‘Ruaumoko’. Bull Volcanol 72(2):185–204 Paton D, Sagala S, Okada N, Jang LJ, Burgelt PT, Gregg CE (2010)
Livingstone DN (2003) Putting science in its place: geographies of sci- Making sense of natural hazard mitigation: personal, social and cul-
entific knowledge. University of Chicago Press, Chicago tural influences. Env Haz 9(2):183–196
Loughlin SC, Sparks RSJ, Brown SK, Jenkins SF, Vye-Brown C (2015) Pearce W, Brown B, Nerlich B, Koteyko N (2015) Communicating cli-
Global volcanic hazards and risk. Cambridge University Press mate change: conduits, content, and consensus. Wiley Interdiscip
Lövbrand E, Beck S, Chilvers J, Forsyth T, Hedrén J, Hulme M, Lidskog Rev Clim Chang 6:613–626
R, Vasileiadou E (2015) Who speaks for the future of earth? How Pereira ÂG, Vaz SG, Tognetti S (2017) Uncertainty, assumptions and
critical social science can extend the conversation on the value commitments in the knowledge base of complex environmen-
Anthropocene. Glob Environ Chang 32:211–218 tal problems. Interfaces between Science and Society. Routledge, In,
MacLeod M (2018) What makes interdisciplinarity difficult? Some con- pp 64–81
sequences of domain specificity in interdisciplinary practice. Pescaroli G, Alexander D (2015) A definition of cascading disasters and
Synthese 195:697–720 cascading effects: going beyond the â€ œtoppling dominosâ€ meta-
Mansilla VB (2010) Learning to synthesize: the development of interdis- phor. Planet@ Risk 3(1)
ciplinary understanding. The Oxford Handbook of Pidgeon N, Fischhoff B (2011) The role of social and decision sciences in
Interdisciplinarity. Oxford University Press, Oxford, pp 288–306 communicating uncertain climate risks. Nat Clim Chang 1(1):35–41
Marzocchi W, Woo G (2009) Principles of volcanic risk metrics: theory Pidgeon NF, Lorenzoni I, Poortinga W (2008) Climate change or nuclear
and the case study of Mount Vesuvius and Campi Flegrei, Italy. J power—no thanks! A quantitative study of public perceptions and
Geophys Res 114 risk framing in Britain. Glob Environ Chang 18(1):69–85
Marzocchi W, Sandri L, Selva J (2008) BET_EF: a probabilistic tool for Popper K (1956 (published 1982)) Realism and the aim of science.
long- and short-term eruption forecasting. Bull Volcanol 70(5):623– Routledge, Abingdon
632 Popper KR (1957) Probability magic or knowledge out of ignorance.
Marzocchi W, Newhall C, Woo G (2012) The scientific management of Dialectica 11(3–4):354–374
volcanic crises. J Volcanol Geotherm Res 247-248(0):181–189 Popper K (1959 (German original, 1935)) the logic of scientific discov-
Miller NE (1986) The scientist’s responsibility for public information: a ery. Routledge, Abingdon
guide to effective communication with the media. In: Friedman SM, Popper K (1963) Conjectures and refutations: the growth of scientific
Dunwoody S, Rogers CL (eds) Scientists and journalists: reporting knowledge. Routledge,
science as news. The Free Press, New York, pp 239–253 Porter TM (1996) Trust in numbers: the pursuit of objectivity in science
Mogi K (1958) Relations between the eruptions of various volcanoes and and public life. Princeton University Press
the deformation of the ground surfaces around them. Bull Earthq Potter SH, Jolly GE, Neall VE, Johnston DM, Scott BJ (2014)
Res Inst Univ Tokyo 36:99–134 Communicating the status of volcanic activity: revising New
Morgan MG (2014) Use (and abuse) of expert elicitation in support of Zealand’s volcanic alert level system. J Appl Volcanol 3(1):13
decision making for public policy. Proc Natl Acad Sci 111(20): Powell RC (2007) Geographies of science: histories, localities, practices,
7176–7184 futures. Prog Hum Geogr 31(3):309–329
Proctor R (1991) Value-free science?: purity and power in modern knowl- Sword-Daniels V, Eriksen C, Hudson-Doyle EE, Alaniz R, Adler C,
edge. Harvard University Press Schenk T, Vallance S (2016) Embodied uncertainty: living with
Proctor RN, Schiebinger L (2008) Agnotology: the making and unmak- complexity and natural hazards. Journal of Risk Research:1–18
ing of ignorance. Stanford University Press, Stanford, In Tadaki M, Brierley G, Dickson M, Le Heron R, Salmond J (2015)
Renn O (2008) Risk governance. Earthscan, London Cultivating critical practices in physical geography. Geogr J
Roeser S, Hillerbrand R, Sandin P and Peterson M Eds., (2012) 181(2):160–171
Handbook of risk theory: epistemology, decision theory, ethics, Tazieff H (1977) La Soufriere, volcanology and forecasting. Nature 269:96–97
and social implications of risk (Vol. 1). Springer Science & Thissen W, Kwakkel J, Mens M, van der Sluijs J, Stemberger S,
Business Media Wardekker A, Wildschut D (2017) Dealing with uncertainties in
Sandri L, Marzocchi W, Zaccarelli L (2004) A new perspective in iden- fresh water supply: experiences in the Netherlands. Water Resour
tifying the precursory patterns of eruptions. Bull Volcanol 66:263– Manag 31:703–725
275 Thompson MA, Lindsay JM, Gaillard J-C (2015) The influence of prob-
abilistic volcanic hazard map properties on hazard communication. J
Scolobig A, Mechler R, Komendantova N, Liu W, Schröter D, Patt A
Appl Volcanol 4(1):6
(2014) The co-production of scientific advice and decision making
Tilling RI (2008) The critical role of volcano monitoring in risk reduction.
under uncertainty: lessons from the 2009 L'Aquila earthquake, Italy.
Adv Geosci 14:3–11
Planet@ risk 2(2):71–76
Tobin GA, Whiteford LM, Jones EC, Murphy AD (2007) Chronic haz-
Segerstråle UCO (2000) Beyond the science wars: the missing discourse
ard: weighing risk against the effects of emergency evacuation from
about science and society. SUNY Press
popocatépetl, México. In: Papers and Proceedings of Applied
Shackley S, Wynne B (1996) Representing uncertainty in global climate Geography Conferences. [np]; 1998, p 288
change science and policy: boundary-ordering devices and authori- Tonini R, Sandri L, Thompson MA (2015) PyBetVH: a Python tool for
ty. Sci Technol Hum Values 21(3):275–302 probabilistic volcanic hazard assessment and for generation of
Sheldrake TE, Aspinall W, Odbert H, Wadge G, Sparks R (2017) Bayesian hazard curves and maps. Comput Geosci 79:38–46
Understanding causality and uncertainty in volcanic observations: Torrent JCR, Herrera SR, Bustamante FM (2017) El proyecto nueva
an example of forecasting eruptive activity on Soufrière Hills volca- Chaitén: La asincronía entre Estado, academia y comunidad. AUS
no, Montserrat. J Volcanol Geotherm Res 341:287–300 [Arquitectura/Urbanismo/Sustentabilidad] (19):73-79
Sparks RSJ, Aspinall W (2004) Volcanic activity: Frontiers and chal- Ugarte AM, Salgado M (2014) Sujetos en emergencia: acciones
lenges in forecasting, prediction and risk assessment. In: The State colectivas de resistencia y enfrentamiento del riesgo ante desastres;
of the Planet: Frontiers and challenges in Geophysics. Geophysical el caso de Chaitén, Chile. Revista Invi 29(80):143–168
Monograph 150; IUGG Volume 19, Victor DG (2015) Embed the social sciences in climate policy. Nature
Spiegelhalter DJ (2008) Understanding uncertainty. Ann Fam Med 6(3): 520(April):27–29
196–197 Wadge G, Aspinall W, Barclay J (2009) Risk-based policy support for
Spiegelhalter D (2017) Risk and uncertainty communication. Annu Rev volcanic hazard mitigation. Foreign and Commonwealth Office,
Stat Its Appl 4:31–60 London, In
Spiegelhalter DJ, Riesch H (2011) Don't know, can't know: embracing Weigold MF (2001) Communicating science: a review of the literature.
deeper uncertainties when analysing risks. Philos Trans R Soc A S c i C o m m u n 2 3 : 1 6 4 – 1 9 3 . h t t p s : / / d o i . o r g / 1 0 . 11 7 7 /
Math Phys Eng Sci 369(1956):4730–4750 1075547001023002005
Stengers I (2013) Une autre science Est possible. Manifeste pour un Wilson CJ (2017) Volcanoes: characteristics, tipping points, and those
ralentissement de sciences. Paris, La Découverte pesky unknown unknowns. Elements 13(1):41–46
Stephens EM, Edwards TL, Demeritt D (2012) Communicating probabilistic Winson AE, Costa F, Newhall CG, Woo G (2014) An analysis of the
information from climate model ensembles—lessons from numerical issuance of volcanic alert levels during volcanic crises. J Appl
weather prediction. Wiley Interdiscip Rev Clim Chang 3:409–426 Volcanol 3(1):14
Stirling A (2003) Risk, uncertainty and precaution: some instrumental Wisner B, Blaikie PM, Cannon T, Davis I (2004) At risk: natural hazards,
implications from the social sciences. In: Berkhout F, M. Leach, I. people's vulnerability and disasters. Psychology Press
Scoones (ed) negotiating environmental change: new perspectives Wisner B, Gaillard JC, Kelman I (2012) Handbook of hazards and disas-
from social science. Edward Elgar publishing, Cheltenham ter risk reduction and management. Routledge
Wittgenstein L (1921 (this edition 2001)) Tractatus Logico-
Stirling A (2007) Risk, precaution and science: towards a more construc-
Philosophicus. Routledge, London
tive policy debate. EMBO Rep 8(4):309–315
Wong-Parodi G, Krishnamurti T, Davis A, Schwartz D, Fischhoff B
Stirling A (2008) Opening up and closing down. Sci Technol Hum Values
(2016) A decision science approach for integrating social science
33(2):262–294
in climate and energy solutions. Nature Clim Change 6(6):563–569
Stirling A (2010) Keep it complex. Nature 468(7327):1029–1031 Woo G (2008) Probabilistic criteria for volcano evacuation decisions. Nat
Stirling, A., 2012. Opening up the politics of knowledge and power in Hazards 45:87–97
bioscience. PLoS Biol, 10(1), p.e1001233 Woo G (2012) Calculating catastrophe. Imperial College Press, London
Sunstein CR (2005) Laws of fear: beyond the precautionary principle. Woo G (2015) Cost–benefit analysis in volcanic risk. In: Volcanic haz-
Cambridge University Press ards. Elsevier, Risks and Disasters, pp 289–300
Surono JP, Pallister J, Boichu M, Buongiorno MF, Budisantoso A, Costa Wynne B (1992) Uncertainty and environmental learning: reconceiving
F, Andreastuti S, Prata F, Schneider D, Clarisse L, Humaida H, science and policy in the preventive paradigm. Glob Environ Chang
Sumarti S, Bignami C, Griswold J, Carn S, Oppenheimer C, 2(2):111–127
Lavigne F (2012) The 2010 explosive eruption of Java's Merapi Wynne B (2001) Creating public alienation: expert cultures of risk and
volcano—a ‘100-year’ event. J Volcanol Geotherm Res 241–242: ethics on GMOs. Sci Cult 10(4):445–481
121–135 Wynne B (2004) Risk as globalising Bdemocratic^ discourse? Framing
Sutherland WJ, Bellingan L, Bellingham JR, Blackstock JJ, Bloomfield subjects and citizens. In: Leach M (ed) Science and citizens: glob-
RM, Bravo M, Cadman VM, Cleevely DD, Clements A, Cohen AS alisation and the challenge of engagement. Zed Books, London
(2012) A collaboratively-derived science-policy research agenda. Wynne BE, Shackley S (1994) Environmental models: truth machines of
PLoS One 7(3):e31824 social heuristics? Globe 21:6–8

Critical Volcanology? Thinking Holistically About Risk and Uncertainty

Uploaded by

Copyright:

Available Formats

Critical Volcanology? Thinking Holistically About Risk and Uncertainty

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Critical Volcanology? Thinking Holistically About Risk and Uncertainty

Uploaded by

Copyright:

Available Formats

Bulletin of Volcanology (2019) 81: 20

Critical volcanology? Thinking holistically about risk and uncertainty

the communication of a scientific warning to authorities—is

Fig. 2 Different types and flows

Type Description Management

Fig. 3 Two ways of

change and the role of power dynamics. Transformative re-

Fig. 4 Approaches to knowledge

General perspective Ontology Epistemology Methodology

Positivism/postpositivism Realism-we can know reality Emphasis on quantitative Experimental,mathematical,

research. Note that some of the ontological and epistemological 3

You might also like